18
Cornell Hospitality Report Vol. 11, No. 16, August 2011 Unscrambling the Puzzling Matter of Online Consumer Ratings: An Exploratory Analysis by Pradeep Racherla, Ph.D., Daniel Connolly, Ph.D. and Nastasa Christodoulidou, Ph.D.

Unscrambling the Puzzling Matter of Online Consumer ... · Philips Hospitality Americas Chris Proulx, CEO, eCornell & Executive Education Carolyn D. Richmond, Partner, Hospitality

  • Upload
    others

  • View
    1

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Unscrambling the Puzzling Matter of Online Consumer ... · Philips Hospitality Americas Chris Proulx, CEO, eCornell & Executive Education Carolyn D. Richmond, Partner, Hospitality

Cornell Hospitality ReportVol 11 No 16 August 2011

Unscrambling the Puzzling Matter of Online Consumer Ratings An Exploratory Analysis

by Pradeep Racherla PhD Daniel Connolly PhD and Nastasa Christodoulidou PhD

Advisory Board

The Robert A and Jan M Beck Center at Cornell UniversityBack cover photo by permission of The Cornellian and Jeff Wang

Cornell Hospitality Reports Vol 11 No 16 (August 2011)

copy 2011 Cornell University This report may not be reproduced or distributed without the express permission of the publisher

Cornell Hospitality Report is produced for the benefit of the hospitality industry by The Center for Hospitality Research at Cornell University

Rohit Verma Executive DirectorJennifer Macera Associate DirectorGlenn Withiam Director of Publications

Center for Hospitality ResearchCornell University School of Hotel Administration489 Statler HallIthaca NY 14853

Phone 607-255-9780Fax 607-254-2922wwwchrcornelledu

Niklas Andreacuteen Group Vice President Global Hospitality amp Partner Marketing Travelport GDS

Rarsquoanan Ben-Zur Chief Executive Officer French Quarter Holdings Inc

Scott Berman Principal Real Estate Business Advisory Services Industry Leader Hospitality amp Leisure PricewaterhouseCoopers

Raymond Bickson Managing Director and Chief Executive Officer Taj Group of Hotels Resorts and Palaces

Stephen C Brandman Co-Owner Thompson Hotels IncRaj Chandnani Vice President Director of Strategy WATGBenjamin J ldquoPatrickrdquo Denihan Chief Executive Officer

Denihan Hospitality GroupBrian Ferguson Vice President Supply Strategy and Analysis

Expedia North AmericaChuck Floyd Chief Operating OfficerndashNorth America

HyattGregg Gilman Partner Co-Chair Employment Practices

Davis amp Gilbert LLP

Tim Gordon Senior Vice President Hotels pricelinecomSusan Helstab EVP Corporate Marketing

Four Seasons Hotels and ResortsJeffrey A Horwitz Chair Lodging + Gaming and Co-Head

Mergers + Acquisitions ProskauerKevin J Jacobs Senior Vice President Corporate Strategy amp

Treasurer Hilton WorldwideKenneth Kahn PresidentOwner LRP PublicationsKirk Kinsell President of Europe Middle East and Africa

InterContinental Hotels GroupRadhika Kulkarni PhD VP of Advanced Analytics RampD

SAS InstituteGerald Lawless Executive Chairman Jumeirah GroupMark V Lomanno CEO Smith Travel ResearchBetsy MacDonald Managing Director HVS Global Hospitality

ServicesDavid Meltzer Senior Vice President Global Business

Development Sabre Hospitality SolutionsWilliam F Minnock III Senior Vice President Global

Operations Deployment and Program Management Marriott International Inc

Mike Montanari VP Strategic Accounts Sales - Sales Management Schneider Electric North America

Shane OrsquoFlaherty President and CEO Forbes Travel GuideThomas Parham Senior Vice President and General Manager

Philips Hospitality AmericasChris Proulx CEO eCornell amp Executive EducationCarolyn D Richmond Partner Hospitality Practice Fox

Rothschild LLPSusan Robertson CAE EVP of ASAE (501(c)6) amp President of

the ASAE Foundation (501(c)3) ASAE FoundationSteve Russell Chief People Officer Senior VP Human

Resources McDonaldrsquos USAMichele Sarkisian Senior Vice President MaritzJanice L Schnabel Managing Director and Gaming Practice

Leader Marshrsquos Hospitality and Gaming PracticeTrip Schneck Managing Partner District Hospitality PartnersK Vijayaraghavan Chief Executive Sathguru Management

Consultants (P) LtdAdam Weissenberg Vice Chairman and US Tourism

Hospitality amp Leisure Leader Deloitte amp Touche USA LLP

FriendsBerkshire Healthcare bull Center for Advanced Retail Technology bull Cruise Industry News bull DK Shifflet amp Associates bull ehoteliercom bull EyeforTravel bull 4Hotelierscom bull Gerencia de Hoteles amp Restaurantes bull Global Hospitality Resources bull Hospitality Financial and Technological Professionals bull hospitalityInsidecom bull hospitalitynetorg bull Hospitality Technology Magazine bull HotelExecutivecom bull International CHRIE bull International Hotel Conference bull International Society of Hospitality Consultants bull iPerceptions bull JDA Software Group Inc bull JD Power and Associates bull The Lodging Conference bull Lodging Hospitality bull Lodging Magazine bull LRA Worldwide Inc bull Milestone Internet Marketing bull MindFolio bull Mindshare Technologies bull PhoCusWright Inc bull PKF Hospitality Research bull Resort and Recreation Magazine bull The Resort Trades bull RestaurantEdgecom bull Shibata Publishing Co bull Synovate bull The TravelCom Network bull Travel + Hospitality Group bull UniFocus bull USA Today bull WageWatch Inc bull The Wall Street Journal bull WIWIHCOM bull Wyndham Worldwide

Thank you to our generous Corporate Members

PartnersDavis amp Gilbert LLP Deloitte amp Touche USA LLPDenihan Hospitality GroupeCornell amp Executive EducationExpedia Inc Forbes Travel GuideFour Seasons Hotels and Resorts Fox Rothschild LLP French Quarter Holdings Inc HVS Hyatt InterContinental Hotels Group Jumeirah GroupLRP PublicationsMarriott International IncMarshrsquos Hospitality Practice MaritzpricelinecomPricewaterhouseCoopersProskauer Sabre Hospitality SolutionsSathguru Management Consultants (P) Ltd Schneider Electric Thayer Lodging Group Thompson HotelsTravelportWATG

Senior PartnersASAE FoundationCarlson HotelsHilton WorldwideMcDonaldrsquos USAPhilips HospitalitySASSTRTaj Hotels Resorts and Palaces

4 TheCenterforHospitalityResearchbullCornellUniversity

AbouT The AuThors

UnscramblingthePuzzlingMatterofOnlineConsumerRatings

An Exploratory AnalysisbyPradeepRacherlaDanielConnolly

andNatasaChristodoulidou

Pradeep racherla PhD is an assistant professor of marketing at the College of Business West Texas AampM University (pracherlawtamuedu ) He earned a PhD in business administration from the Fox School of

Business Temple University His research interests include consumer generated media healthcare marketing social networks knowledge management and collaborative systems He has written several research papers

and technical reports that have been published or forthcoming in prestigious journals such as Journal of Consumer Behavior Journal of Marketing amp Management Journal of Management Information Systems

Annals of Tourism Research and Cornell Hospitality Quarterly He has presented at numerous conferences both nationally and internationally

Daniel J Connolly PhD is an associate professor of information technology at the Daniels College of Business at the University of Denver with a dual appointment in the School of Hotel Restaurant and Tourism Management and in the Department of Information Technology and Electronic Commerce He also serves as director of undergraduate programs His teaching research and consulting interests focus on the strategic application of information technology and electronic commerce He has written or co-authored numerous publications including articles that have appeared in the Cornell Hotel and Restaurant Administration Quarterly the FIU Hospitality Review Information Technology in Hospitality and Journal of Hospitality and Leisure Marketing His book Technology Strategies for the Hospitality Industry (Pearson Prentice Hall 2005) is the industryrsquos first book on technology strategy

Natasa Christodoulidou PhD is a faculty member in the management and marketing department at California State University where she teaches internet marketing and other related marketing courses at the

undergraduate and MBA level Her research interests are in the areas of hospitality technology travel and tourism electronic commerce hospitality electronic distribution and electronic marketing She is a member of

the Decision Sciences Institute and the Western Decision Sciences Institute She has published more than 20 articles in the past five years in academic and professional journals She presents regularly at academic and

professional conferences around the world such as Eye for Travel FSTEC DSI and WDSI Among her speaking engagements this year are Las Vegas Hawaii France and Singapore

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 5

exeCuTive suMMAry

Thisstudyexploresthepatternsofonlinereviewsofvacationhomesfromacommunity-basedtraveladvisorywebsitewithagoalofunderstandingthebiasesinherentinonlinewordofmouth(WOM)relatedtotourismandhospitalityservicesAnanalysisofnearly3200reviewsfrom

ldquoReviewsitecomrdquo(apseudonym)whichpostsreviewsofvacationrentalpropertiesacrosstheUSAfindsanoverwhelmingpreponderanceof favorable reviewsMore to thepoint relatively few

ldquomoderaterdquoreviewsarepostedandthesecond-highestcategoryisextremelynegativecommentsUsingsemanticprocessingtechniquesontheaggregatereviewtextthestudyidentifiesthenuancedopinionsandconcernsofthetravelerswhowritereviewsNegativereviewstendtobelengthyandargumentativeoftendetailingdisappointmentoverexpectationsnotmetPositivereviewsontheotherhandtendtoberelativelybriefandconfirmtheoverallratingConsumerswhowroteldquohighrdquoreviewsplacedgreaterimportanceonvalueformoneycleanlinessandcomfortthandidthosewhowrotenegativereviewsThosewhowroteldquolowrdquoreviewsplacedtheiremphasisontheserviceprovidedbythepropertystaffandmanagementNegativereviewsweremorelikelytoinvolveahigherpriceaccommodationThisanalysisindicates that the overall numerical ratings typically used in review systems may not be the idealindicator of perceived service quality The results suggest that review sites should develop bettermethodstoaggregatesynthesizeandpublishthereviewcontentsparticularlythenumericalratingsThisandotherreviewsitesshowtheaverageofallthepoint-scaleratingsbutsuchsimplemeansdonottakeintoaccountthebiasesthatareinherentintheratingsystemsInsteadthesitesshouldprovidemore information and heuristics to help the consumers navigate through the clutter and get theinformationtheydesire

6 TheCenterforHospitalityResearchbullCornellUniversity

CorNell hosPiTAliTy rePorT

Giventhatconsumersareincreasinglyrelyingonsearchenginestoacquireinformationabout hotels restaurants and travel purveyors (among many other services)consumer-generated online reviews will inevitably change the structure andaccessibilityofinformationalongwithconsumersrsquoperceptionsofvariousservices

ThepopularityofonlinereviewshasledtoaproliferationofonlinereviewsitesincludingpopularsitessuchasHotelscomTripadvisorcomandYelpcomArecentstudybyForresterIncfoundthatmorethan80percentofwebshoppersreadandconsiderotherconsumersrsquoreviews1SimilarlyCompeteIncfoundthatnearly50percentofpurchasersvisitedamessageboardforumoronlinecommunityforinformationconnectedwiththeironlinetravelpurchasingandoneinthreeofthesebuyerssaidthatconsumer reviews helped with their purchase decision2 Almost half of these consumers said thatconsumersrsquoopinionsactuallycausedthemtochangetheirmindaboutwhattheypurchasedMoreoveramongthosebuyers25percentsaidtheyalsopostedareviewonaconsumerreviewsiteaftermakingtheirpurchase

1ForresterldquoFortyFactsaboutUSOnlineShoppersrdquoForresterInc20062CompeteldquoEmbracingConsumerBuzzCreatesMeasurementChallengesforMarketersrdquoCompeteInc2006

UnscramblingthePuzzlingMatterofOnlineConsumerRatings

An Exploratory Analysis

byPradeepRacherlaDanielConnollyandNatasaChristodoulidou

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 7

Howevertheexistingunderstandingofonlinetourism-relatedreviewsisrelativelyweakeventhoughweknowthatwordofmouth(WOM)ismoreimportantandinfluentialwithinaservicescontextthaninproductmarketingscenariosRecentlysomestudieshaveexploredtourism-andhospital-ity-relatedonlineWOM3Yetresearchhassofaroverlookedcertaininterestingaspectsofonlinereviewsasfollows

Onlinereviewstypicallyuseldquooverallstarratingsrdquoaspri-maryindicatorsofperceivedservicequalitymdashratingsthatarenotrelatedinanywaytotheformalstardesignationsawardedbyForbesorMichelinConsumersrsquomeanstarratingsaretypi-callyaggregatedfromtheratingsgivenbyindividualconsum-ersineachreviewWhatwedonrsquotknowishowconsumersdeterminethestarstheyassigntoaservicewhentheypostareviewonawebsiteThatisdostarratingsaccuratelyreflectconsumersrsquosentimentsregardingvariousattributesofserviceprovidersThisisavitalissuegiventheexperientialandin-tangiblenatureoftravelservicesandthefactthatconsumersusetheoverallratingasanimportantheuristicwhilenarrow-ingdownchoices

FurtherstudieshaveoverlookedtherelationshipbetweenthereviewsandvariousattributesofservicesandserviceprovidersSinceonlinereviewsareaformofWOMvariablessuchasconsumerinvolvementpricingandlengthofstaymayhaveanimpactonthefinalratingandpostingofthereviewprovidedbyconsumersItisimportanttouncovertheserelationshipstobetterunderstandthedynamicsofonlinereviews

FewstudieshaveanalyzedthequalitativeaspectofonlinereviewsOnlinereviewsareessentiallyopen-endedtext-basedconsumer-to-consumer(c2c)communicationsThetextcontentmaycontainnuancedviewsoftheservicesandserviceproviders(fromthereviewwritersrsquopointofview)thatcannotbeexpressedusingcrudenumericalratings4Hencethereisaneedtoanalyzethetextportionofthereviewsandidentifytheissuesthatconsumersaremostlyconcernedwithwhenwritingandpostingonlinerelatedreviews

WeaddresstheabovepointsinthisstudyThefollowingquestionsguidedus

3KHYooandUGretzelldquoWhatMotivatesConsumerstoWriteOnlineTravelReviewsrdquoInformation Technology amp TourismVol10(2008)pp283-295andZXiangandUGretzelldquoRoleofSocialMediainOnlineTravelInformationSearchrdquoTourism Management20094AGhoseandPGIpeirotisldquoEstimatingtheSocio-economicImpactofProductReviewsMiningTextandReviewerCharacteristicsrdquoInformation Systems Research2008

1 Dooverallratingsreflecttheperceivedqualityoftheserviceproviders(asderivedfromtheconsumersrsquoexpressedsentimentinthereviews)Whataretheunderlyingpatternsoftheoverallratingsinonlinereviews

2 Whatarethelinkagesamongthevariousattributesoftheserviceprovidersandcustomerreviewsand

3 WhataretheissuesthatconsumersmostlytalkaboutinthetextportionofonlinereviewsDotheseis-suesdifferfromtheratingsthattheconsumershaveprovidedviathestandardvariablesprovidedbythereviewsites

ResearchSettingandDataCollectionThisstudyisbasedonreviewscollectedfromldquoReviewsitecomrdquoaswehavecalledthecommunity-driventravelsitethatweanalyzedThesiteprovidesreviewsofverifiedrentalpropertiesandorganizesthereputationofvacationpropertiesThepropertylistingwithinthesiteisbasednotontherevenuesharingmodelthatisgenerallypracticedbyotherreviewsitesbutonthereputationtrustandfeedbackfromcustomerswhohavestayedatthevacationrentalpropertiesReviewsitecomisafreeservicethatoperatesonserviceproviderfeesandadvertisingrevenuesUn-likesometravelreviewsitesthefirmrsquosreviewsolicitationstrategyistosendpersonalizedinvitationsonlytothosecustomerswhosestayhasbeenverifiedbythepropertymanagementThesitersquosmanagementbelievesthatthisstrategytoalargeextenteliminatesfakereviews

Thefirmprovideduswith3300reviewsforanalysisAfterinitialdatacleanupabout100reviewswithincom-pleteorunusableinformationwereeliminatedand3197reviewswereusedfortheanalysisThereviewscontainedthefollowingattributesbull Theoverallratingassignedtotheserviceprovider

basedontheaggregationoftheratingsassignedbyindividualreviewers

bull Consumersrsquoratingsonthefollowingsixattributesvaluecleanlinesscomfortservicelocationandcheck-in

bull Thereviewtext(Semanticprocessingtechniqueswereappliedontheaggregatereviewtexttoidentifythenuancedopinionsandconcernsofthetravelerswhowrotereviews)

8 TheCenterforHospitalityResearchbullCornellUniversity

bull Agerangeoftherevieweronan1-4ordinalscale(25-3435-4445-54and55andabove)

bull Averagepriceperdayand

bull Wordcount(usedinthequantitativeanalysis)

PreliminaryHypothesesandResultsTheagegroupsofthereviewwritersareshowninExhibit1Alargepercentageofthecustomerswhoreviewedthepropertiesfallunder44yearsofageThisisconsistentwithpreviousstudiesinthisrealmwhichshowthatgenerationXandGenerationYaremoreactiveintheWOMforums

ReviewRatingDistribution Websitesthathostonlinereviewsgiveconsumerstheoptiontoassignaratingtotheserviceproviderstypicallyonascaleof1to5or1to7Manysitesaggregatethesenumeri-calorstarratingstodetermineanoverall(mean)starratingWeseethisoverallratingasanimportantelementinonlineWOMsinceconsumersoftenusethisratingtonarrowdowntheirconsiderationsetStudiesthathaveusedtheratingastheprimarypredictoroftheserviceprovidersrsquoperceivedqualityorcustomersatisfactiontodeterminetheimpactofonlinereviewsonproductsaleshavearrivedatmixedfind-ings5Howeverthesestudiesarebasedonthefundamentalassumptionthatthenumericalratingsassignedtoservice

5ForexampleseeJChevalierandDMayzlinldquoTheEffectofWordofMouthOnlineOnlineBookReviewsrdquoJournal of Marketing ResearchVol43(2006)pp345-354andWDuanBGuandABWhinstonldquoDoOnlineReviewsMattermdashAnEmpiricalInvestigationofPanelDatardquoDecision Support SystemsVol45(2008)pp1007-1016

providersareanaccurateestimateofthequalityperceivedbythecustomerThispracticefollowsthebasicassumptionofbehavioraltheoriststhatanydatasetwithalargenumberofconsumerresponsestendstofollowanormal(Gaussian)distributionRecentstudieshavequestionedthisassump-tionForinstanceDellaracosandNarayanrsquosstudyonmoviereviewsinYahoocomfoundthatonlytheextremelysatisfiedandextremelydissatisfiedcustomersarelikelytopubliclyexpresstheiropinionsascomparedtocustomerswithmod-erateopinions6SimilarlystudiesusingdatafromTripadvi-sorcomandAmazoncomfoundthatthestarratingsexhibitatruncateddistributioninwhichthemajorityofreviewsarepositive7

TotestthisassumptionweranasimplefrequencychartasshowninExhibit2Inkeepingwithotherstudiesthedistributionisheavilyskewedtowardsthepositiveratings1-pointreviews32(1)2-pointreviews95(3)3-pointreviews458(14)4-pointreviews688(22)and5-pointreviews1912(60)Totestfortherobustnessofthisdistri-butionweappliedtheKolmogorov-Smirnovtestofnormal-ityTheresultsconfirmthenon-normaldistributionofthestarratingsandtheheavyskewtowardspositivereviews

Threeexplanationsarepossibleforsuchadistribution

6CDellarocasandRNarayanldquoAStatisticalMeasureofaPopulationrsquosPropensitytoEngageinPost-purchaseOnlineWord-of-mouthrdquoStatisti-cal ScienceVol21(2006)pp2777ATalwarandRFBoiJurcaUnderstandingUserBehaviorinOnlineFeedbackReportingrdquoinElectronic CommerceSanDiegoCalifornia(2007)pp134-142

Exhibit 1

Distribution of age ranges

Exhibit 2

Distribution of overall ldquostar ratingsrdquo

55 and above 45ndash54

years

35ndash44 years

18ndash34 years

overall ldquostar ratingrdquo

Num

ber o

f rev

iew

s

3295

458 688

1912

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 9

1 AmajorityofthepropertiesthatarelistedonthewebsitehaveoutstandingperceivedqualityandprovidesatisfactoryservicestocustomersWhilethisistheo-reticallypossibleitseemsunlikelywhenappliedtoallthepropertieslistedonthewebsite8

2 ThereviewsdonotconstituteatrulyrepresentativesampleofconsumeropinionBasedonotherstud-iesthisexplanationseemsplausibleConsumerswithmoderateviewsmayseelittleincentivetoreporttheirviewsinanonlineforumandasaconsequenceonlythoseconsumerswhohavepolarizedviewsarelikelytoposttheircomments

3 ThedistributioncouldalsobeanartifactofthefirmrsquosreviewsolicitingstrategyRecentstudiessupportthisideabysuggestingthatcustomersengageinsociallydesirablebehavioreveninimpersonalinteractionswiththeserviceproviders9Inthiscaseitmaybethattheywillprovideareasonablypositiverevieweveniftheexperiencehasnotbeenuptothemark

StarRatingsandPerceivedServiceQualityLetrsquosreturntothefirstquestionweaskedattheoutsetArestarratingsatruerepresentationofserviceprovid-ersrsquoperceivedqualityasreflectedbytheconsumersrsquostatedsentimentsThisbecomesanimportantquestionfortwo

8AKadetldquoRah-RahRatingsOnlinerdquoSmartMoney20079YGreacutegoireandRJFisherldquoCustomerBetrayalandRetaliationWhenYourBestCustomersbecomeYourWorstEnemiesrdquoJournal of the Acad-emy of Marketing ScienceVol36(2008)pp247-261

reasons(a) ratingsaretheprimaryevidencethatconsum-ersconsiderwhileshoppingonlineforinformationand(b)thetruncatednatureofratingdistributionsmeansthatthereviewsavailablemaynottrulyrepresentconsumeropinionortheserviceprovidersrsquoperceivedqualityIfonlythosewithextremeviewspostreviewsthenconsumersshouldalwayslookforserviceprovidersthathavea5-starratingOnthecontraryifthereviewsaccuratelyreflectallconsumersrsquoviewsthenconsumersarebetteroffsearchingforreviewswitha3ratingsincethesetendtobemorebalancedandenablebetterproductanalysisRecentlyHu et aluseddatafromAmazoncomshowthatnumericalscoresdonotreflectbooksrsquotrueperceivedquality10AstheysuggestldquohellipratherthescorereflectsthebalanceofdiverseopinionsInotherwordswhenabookrsquosoverallscoreisaround3itdoesnotsuggestthatconsumersgenerallyagreethatthisisanaver-agebookItrathersuggeststhatroughlyequalnumberofconsumersthinkthatthebookiseitheranoutstandingbookoranabysmalbookrdquo

Totestthisassertionweappliedaregressionequationwithsixpropertyattributes(ievalueformoneycheck-inlocationcleanlinesscomfortandservice)asindependentvariablesandstarratingasthedependentvariableThisanalysisisbasedontheassumptionthatforagivenpropertyifstarratingstrulyrepresenttheperceivedqualityoftheproperties(aswellastheconsumersrsquosentiment)thenthestarratingsandratingsofthesixotherpropertyattributes

10NHuJZhangandPAPavlouOvercomingtheJ-shapedDistribu-tionofProductReviewsrdquoCommunications of the ACMVol52(2009)pp144-147

low reviews t-value high reviews t-value(Constant) 1618 18353 2400 34186vAlue 0254 13726 0378 8767CheCKiN -0083 -3701 -0053 -1192loCATioN 0104 6132 0079 2117CleAN 0144 7574 0112 2808CoMForT 0267 13993 0129 3088serviCe 0098 4209 0107 2401r2 0331 0338F-statistic 5038 24714

Exhibit 3

regression analysis of ldquolowrdquo and ldquohighrdquo reviews

Notes Coefficients (t-statistics reported) Significance is as follows significant at 0001 level significant at 05 level significant at 010 level Dependent variable is overall rating

10 TheCenterforHospitalityResearchbullCornellUniversity

moderateratingsonalltheattributesbutwasstillwillingtoprovideahighoverallstarratingtothepropertyThisexamplesuggeststhatconsumersplacesignificantemphasisontheoverallexperientialvalueprovidedbythepropertiesThisexpectationofvalueismanytimesintangibleanddoesnotnecessarilydependonthefunctioningorratingoftheindividualfeaturesoftheproperty

RelationshipbetweenRatingsandReviewAttributesTounderstandconsumersrsquoword-of-mouthbehavioritisimportanttoconsidervariousfactorssuchasperceivedexpectationsofcostsandbenefitsimportanceofthepur-chase(involvement)personalcharacteristicsandsituationalinfluencesWeconsidertwofactorsthatareknowntoaffecttheintensityandvalence(positiveornegative)ofwordofmouthpriceandwordcount

PricePriceisanimportantfactorthatcontributestopre-purchaseexpectationsTheintensityofexpectations(andtheirfulfillment)partlydrivenbypriceisdirectlypropor-tionaltoconsumersrsquosatisfactionlevelsandWOMpropensityThismaybethereasonthatthetendencytocomplainisdirectlyrelatedtothecostoftheservice11Themoreexpen-siveanitemthegreaterwillbetheperceivedlossesandthegreaterlikelihoodthatdissatisfactionwillresultinacom-plaintSimilarlythelikelihoodofword-of-mouthbeingtrig-geredbythelevelofexpectationndashdisconfirmationincreases

11BeardenWOandTeelJE(1983)Selecteddeterminantsofcon-sumersatisfactionandcomplaintreportsJournal of Marketing Research 2021-28

willshowahighcorrelationFortheanalysiswetookintoconsiderationthefactmostreviewsarepositive(ie4-and5-pointratings)Thereforewedividedthereviewsintotwocategorieshighreviews(starratingabove3n=2597)andlowreviews(ratingsbeloworequalto3n=600)andranseparateregressionanalysesonthesetwogroupsasshowninExhibit3

AninterestingfindingistheweakrelationshipbetweentheattributeratingsandoverallstarratingforeachpropertyTheR-squareoftheregressionequationwithallsixvariablesandtheoverallstarratingsislessthan40percentThatiseveniftheconsumersprovidedhigherratingsonindividualattributestheiroverallratingwasusuallylowandthere-versewasalsotrueThistrendisobservedinboththehighandlowreviewgroups

TheanalysissuggeststhatinhighreviewsconsumersplacedgreaterimportanceonvalueformoneycleanlinessandcomfortThehighertheratingsofaconsumerontheseattributesthehigheristhelikelihoodthattheconsumerwillprovideahigherstarratingCheck-inlocationandservicedidnotprominentlyfigureintheiroverallevalu-ationsofthepropertieseventhoughtheseattributeswereratedhigh

ThelowreviewsshowsimilarresultsinthesensethattheloweraconsumerratesvaluecleanlinessandcomfortthelowerthelikelihoodofapositiveratingInadditioninthecaseoflowreviewsconsumersplacedemphasisontheserviceprovidedbythepropertystaffandmanagement

TheaboveresultisillustratedinoneoftheldquohighrdquoreviewsshowninExhibit4Thereviewergavelowto

star rating Text Commentary

This was our 1st time renting a house vs a condo This house was great If you wanted to spend a little time by yourself you could so many rooms to go to The house was so comfortable like being at home It was our home for the week Thank you Hope to rent it next year if we havent bought one ourselves

ratings for specific Attributes

value Check-in location Clean Comfort service

3 2 3 2 4 3

Exhibit 4

example of a ldquohighrdquo review

Avg rating for Property Attributes

30

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 11

withthecostoftheproduct12Giventhegreaterperceivedriskinservicescustomerswouldseekmoretrustworthyser-viceproviderswithagoodreputationofsatisfyingcustomerneeds13ThisinturnincreasestheexpectationsandleadstomoreextremereactionsfromtheconsumersHenceonecanspeculatethatthehigherthepriceshigherwillbetheaver-agenegativeorpositiveratingsforaservice

Word countThenumberofwordsinareviewmayserveasaproxyfortheamountofinformationinthereviewsHowevertheevidenceonthisaspectofWOMofalltypesisequivocalForinstanceconsumersgenerallyengageinpositiveWOMtoavoidfeelingsofguiltandas-sociationwithbadnewsandtoreducecognitivedissonanceOntheotherhandacustomerwhoishighlydisappointedwillprovideapersuasiveargumenttomakesurethatothercustomersaredissuadedfromrentingthesamepropertyResearchonconventionalwordofmouthhasofferedampleevidencetosuggestthatdissatisfiedcustomersengageintwotothreetimesasmuchWOMassatisfiedcustomers14Webelievethatthislogicextendstoonlinereviewsdespitethefactthatexistingstudiesareequivocalatbest15Thereforeitcanbeassumedthatreviewswitheitherextremelypositiveorextremelynegativeratingswillhavegreaterwordcountasopposedtomoremoderatereviews

TotesttheseassumptionsweappliedANOVAwithpriceandwordcountasdependentvariablesandthereviewratingasfactorsAsillustratedinExhibits5and6theresultsshowthatpriceandwordcountshowasignificantnegativecorrelationwithstarratingThatisreviewswithlowerratingsaretypicallyassociatedwithgreaterwordcountandrepresentpropertiesthatchargehighrates

TheresultonthewordcountsuggeststhatconsumerswhoareunhappywitharentalpropertytendtowritemoretoexpresstheirdissatisfactionAsshowninthisexhibitlowerratingsareassociatedwithhighernumberofwordsinthereviewsandviceversaSimilarlypropertypriceseemstobestronglyassociatedwithnegativewordofmouthsup-portingtheprinciplethathighercostincreasestheconsum-ersrsquoinvolvementinthepurchase

12BeardenWOandMasonJB(1984)AninvestigationofinfluencesonconsumercomplaintreportsAdvancesinConsumerResearch11490-49513WetzerIMZeelenbergMandPietersR(2007)ldquoNevereatinthatrestaurantIdidrdquoExploringwhypeopleengageinnegativeword-of-mouthcommunicationPsychology and Marketing24661-68014SchlossbergH(1991)CustomersatisfactionNotafadbutawayoflifeMarketing News25andWestbrookR(1987)ProductConsump-tion-basedAffectiveResponsesandPost-purchaseProcessesJournal of Marketing Research 24258-27015AlexMSusskindldquoIToldYouSoRestaurantCustomersrsquoofWord-of-MouthCommunicationPatternsrdquoCornell Hotel and Restaurant Adminis-tration QuarterlyVol43No2(April2002)pp75-85

Exhibit 5

Differences in word count across star ratings

Exhibit 6

Differences in average price per day across ratings

Aver

age

pric

e pe

r day

of t

he re

ntal

pro

pert

y

overall ldquostar ratingrdquo

1504

1563

1804

19861987

Aver

age

wor

d co

unt o

f rev

iew

overall ldquostar ratingrdquo

836837

121

15751543

160

140

120

100

80

200

190

180

170

160

150

1 2 3 4 5

1 2 3 4 5

12 TheCenterforHospitalityResearchbullCornellUniversity

thereviewersThesecanbeclassifiedas(a)attributesorfea-turesofthepropertythatfigureprominentlyandcontributetothepositiveornegativeratings(egroomsbeddishesfurnitureorservice)and(b)thepositiveornegativewordsthatrepresentthesentimentofthereviewer(eggreatamazingdirtyandwonderful)Theresultsofthequalitativeanalysisareasfollows

Syntactic featuresTheoveralltoneofthelowreviewsisratherargumentativeThereviewerstendtobeforcefulinpresentingtheirpointofviewTheylayoutagreatamountofreasoningandprovidespecificexamplestodrivehometheirargumentThisisreflectedinthegreaterpercentageofnegativeconnectorwords(34)suchasbut however as ifandeven thoughItisinterestingtonotethatreviewersinthiscategoryspeakdirectlytotheotherconsumersnotablythroughgreateruseofwordssuchasyou(46)andlooktoinfluencetheirdecisionmaking

InthehighreviewsbycontrasttheoveralltonetendstobeenunciativeandthereviewersusethetexttomerelyreinforcetheirnumericalratingsusingfewerwordsonaveragethanthenegativereviewersInthesereviewstheconsumersfocusontheirpersonalexperiencesandappealtotheaudiencethroughemotionsratherthanlogicandreasoningasevidentfromgreateruseofpositiveadjectives

QualitativeAnalysisoftheReviewTextTheaboveanalysisemphasizestheimportanceofthetextcontentinconsumerreviewsWhilenumericalratingsallowthereviewerstoratetheserviceprovideronvariousattributesitisthetextthatprovidesthemwiththeopportu-nitytoarticulatethenuancesoftheiroverallexperienceandconveyusefulinformationaboutaserviceproviderrsquostransac-tionsandservicecapabilitiesThereforesemanticanalysisofthetextcontentprovidesuswithanopportunitytothethirdquestionpresentedattheoutsetofthisreportwhatdoconsumerstalkaboutintheirreviews

Tothisendweanalyzed600eachofrandomlyselectedhighandlowreviewsusingsemanticprocessingtechniquesWeusedcenteringresonanceanalysis(CRA)amodeofnet-work-basedtextanalysisthatrepresentsthecontentoflargesetsoftextsbyidentifyingthemostimportantwordsthatlinkotherwordsinthenetwork16OuranalysisexaminedsyntacticandsemanticfeaturesSyntacticfeaturesincludethepercentageofthewordsthatareopenclass(ienounsverbsadjectives)Semanticfeaturesrepresentthespecial-izedvocabularythatreflectstheoverallopinionexpressedby

16CormanSandDooleyK(2006)CrawdadTextAnalysisSystem12CrawdadTechnologiesLLCChandlerAZ

Prominently Mentioned Features Freq sample Comments

bedroom beds linen sheets 383

Also the same bedroom and bathroom did not have any door separating the bedroom from the bathroom no privacy

Sleeping arrangements were very awkward The listing sheet described 2 full beds We brought our own linens the beds were queen sized Had to purchase sheet

Kitchen Kitchenware Dishes Dishwasher 297

You must not forget to step up when yoursquore walking out of the kitchen into the dining room otherwise yoursquoll trip over the 3-foot step up

No potholders in the kitchen and it didnrsquot seem like there were as many kitchen supplies as before We just had to make do without

beach beaches beachfront 296

The week was the best vacation we have ever had as a family The beach house was a class act

Disappointment 1 It looks like this house is very close to beach but to get there you have to go around through public path

Pool Poolside 218

The pool was too dirty to use We checked in on Thursday and the pool was not cleaned until Saturday

Also it is inconvenient to have to call to get someone to open the pool Didnrsquot understand why Jacuzzi was open and not the pool

bathroom(s) bath bathtub shower 202

The next problem was in another bathroom (main floor)mdashthe wall heater did not work so showering in here was a bit chilly The next problem was in the ldquomasterrdquo bathroom (connected to the bedroom with the King bed) The sliding shower doors need to be totally replaced

Bathtub drains clogged in three second floor bathrooms

Exhibit 7

Top five prominent issues in low reviews

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 13

suchasgreat fantastic andawesome(45)andpersonalpronounssuchasweandI(54)

Semantic featuresExhibits7and8showthemainissuesoraspectsthatareprominentlydiscussedinthelowreviewsTheorderoftheseissuesisbasedonthestatisti-callysignificantfrequencyoftheiroccurrenceinthe1200reviews

WefurtherinvestigatedtheuseofvarioussentimentwordsInterestinglytheprominentissuesdiscussedinthepositivereviewssometimesdonotevenfigureintheratingscalesprovidedtothereviewersOveralltheresultsofthesemanticanalysiscanbesummarizedasfollows

Thepropertyanditsfeaturesaremorecommonlycon-nectedwithextremelypositivesentimentwordssuchasgreat wonderfulandamazinginthehighreviewsthaninthelowreviewsIncomparisonthepropertiesandtheirfea-turesinthelowreviewsaremorelikelytobecharacterizedwithmoderatelypositivewordssuchasniceandokay

Negativereviewsreflecttheextentofexpectationndashdis-confirmationthattheconsumersfaceduringtheirstayintherentalpropertyThesereviewshaveagreaterproportionofwordsexpressingdisappointment(egnotasexpected)Furtherextremelynegativesentimentwords(egdiscom-fortfilthy)andadjectivesaremorelikelytobeassociatedwithfeaturesinlowreviewsthaninhighreviews

Almost40percentoftheconsumerswhowrotehighreviewsexpresstheirwillingnesstoreturntothepropertyasopposedtoonly9percentofthosewritingnegativereviewsFurthermorethan20percentoftheconsumersgivinghighreviewsexpresstheirstrongdesiretorecommendtheprop-ertytotheirfriendsandfamilymembersasopposedtoonly5percentinlowreviews

ReviewofFindingsOurexploratorystudyofconsumerreviewsofUSvacationhomesonReviewsitecomwashighlightedbyoverwhelm-inglypositivereviews

Theoveralldistributionofthereviewswasheav-ilyskewedtowardsthepositiveratingsMorecriticallythisfindingsuggeststhatthereisalackofmoderateandbalancedopinionsWetheorizethatthisiseitherduetopurchasingbias(sincecustomerswentthroughthetroubleofpurchasingandexperiencingthefacilitiestheytendtobepositivelybiasedtowardstheirpurchase)orunder-reportingbias(customersdonothavestrongenoughincentivestotakethetroubleofreportingtheiropinionsandthefewmotivatedindividualswhoprovidetheirreviewsbecomeaself-selectedsample)

Theoverall(mean)starratingtypicallyusedintravelerreviewsitesisprobablynotthemostaccurateindicatorofcustomersrsquoperceptionsofserviceprovidersrsquoqualityWe

Prominently Mentioned Features Freq sample Commentsbeach beaches beachfront 506 Perfect rental property for those who enjoy the activity of the beach and the serenity

of the Sound The house is just down the road from ocean

The large porch offered a beautiful view of the beach and comfortable new rocking chairs

Pool Poolside 218 The outdoor shower dipping pool and screen in porch were an added bonus Perfect house for a low key weekend

House was great for kids Pool was terrific and the playroom was perfectlocation 192 We had a wonderful time My wife loves fishing and the cottage was in a perfect

location to go fishing on the pier

The home is somewhat dated but the location was great We have vacationed in the Chatham area for many summers and this time it was great too

bedroom beds 186 The family suite is huge with a king size and twin bed and still room for a pack-n-play The kitchen and dining room were huge clean and nice

The only comments that I would say to the negative is the front bed room by the front door always seemed damp which seem to attract more bugs

Kitchen Kitchenware Dishes Dishwasher 178 All of the amenities in the house are upscalemdashloved the kitchen and plasma TVs

The kitchen had good basic utensils dishes and cooking pans

Exhibit 7

Top five prominent issues in high reviews

14 TheCenterforHospitalityResearchbullCornellUniversity

foundacorrelationbetweenthestarratingandratingsonotherpropertyattributesoflessthan40percentinmostcases

AhighwordcountinthereviewsiscorrelatedwithalowoverallratingThatisconsumerswhoareextremelydissatisfiedtendtoexpendmoretimeandenergyontheircritiqueofthepropertyitsamenitiesandservicesPriceofthepropertyalsoseemstopositivelycorrelatewithextremenegativereviewsperhapsbecauseahighpriceheightenstravelersrsquoexpectations

Numericalratingsdonotoftencapturetheactualsentimentandthevarietyofdimensionsonwhichtravel-ersevaluatepropertiesBecausenumericalratingsdonotcapturereviewersrsquoopinionsaboutspecificattributestheycanbedeceivingandmaynotrevealthetruequalityofthepropertyortheactualintentionofthereviewerWefoundforinstancethatconsumersvaluelocation(egclosenesstothebeach)andcleanliness(egswimmingpoolareas)morethansuchfeaturesascheck-inandserviceSimilarlyinthenegativereviewsissuessuchaslackofbedroomparapher-naliaorkitchenitemsseemtoberankedhigherinimpor-tancethanissuessuchaslocationandcheck-inThereforeevenwhenapropertyscoreshighonthecertaindimensionstheoverallratingmayremainlowbecauseotherfactorsovershadowthefavorableaspects

WhatItMeansforManagersWeseeimplicationsforonlinereviewsitesandbyexten-sionforproprietorsofvacationhomesandserviceopera-tionsgenerallyWithregardtothereviewsitesifconsumersconsistentlydetectafavorablebiasthereviewsitesmaylosethecriticalaspectsofcredibilityandtrustAlthoughconsumersaregenerallysmartandrationalandcanworkaroundobviousbiasesrecentevidencesuggeststhatcon-sumersaregraduallydefectingfromsomeonlinesitesduetounbalancedreviews17InthisregardsiteadministratorsaswellaspropertymanagersshouldtakestepstoattractthoseconsumerswhoarewillingtoprovideareasonablybalancedanalysisoftheamenitiesandservicesBythiswemeanreviewsthatpresentcomprehensiveandobjectiveanalysisoftheserviceproviderratherthanjustravesorrants

Tothatendsiteoperatorsshouldtakestepstoun-derstandcustomersrsquoreviewwritingbehaviorandtoat-tractmorebalancedreviewsOnewaytodothisistogivepersonalizedinformationtotheconsumerdemonstratingthecontributionmadebyhisorherreviewandthevalueofthereviewtostakeholdersForinstanceLinget alfoundapositivecorrelationbetweenwillingnesstowritereviewsandtheconsumersrsquoperceptionthattheircontributionswouldbeuniqueandhelpfultoothercustomersaswellasperusedby

17HarteveldtHHJohnsonCStarkEandGeldernKV(2009)Us-ingDigitalChannelsToCalmTheAngryTravelerForresterResearch

theserviceproviders18Simplyputthepropensitytowritereviewsincreaseswhentheconsumersfeelthattheirreviewisactuallyhelpingothers(boththeserviceproviderandotherconsumers)

WesuggestthatreviewsitesdevelopbettermethodstoaggregatesynthesizeandpublishspecificreviewcommentsaswellasnumericalratingsFirstmanagersshouldtakeintoconsiderationthelowcorrelationbetweentheoverallratingsandconsumersrsquoassessmentsofvariouspropertyattributes(aswellastheresultsfromthetextanalysis)Theimplicationofsuchaweakcorrelationisthatreviewsitesneedtoexpandonthelistofpossiblevariablessothattheratingsaccuratelyreflectconsumersrsquoneedsandexpectationsregardingserviceproviders

Furthersitemanagersmayhavetorethinkthecurrentpracticeofaveragingalltheratingsforagivenpropertybe-causesuchameanratingfailstotakeintoaccountthebiasesthatareinherentintheratingsystemsAsanexamplethisstudyfindssignificantunder-reportingbiasinthisonlinereviewsiteAcustomerrsquosdecisiontonotpostonlinereviewsmaymeanthelossofimportantinformationthatcanassisttheusersofthesitestomakemorereliableinferencesThemajorityoftodayrsquosfeedbackmechanismsdonotpubliclydisclosethenumberofsilenttransactions(ietransactionsforwhichnofeedbackwaspostedbycustomers)Suchinfor-mationshouldbecomeapartofaserviceproviderrsquosonlineprofileonratingsitesandotherfeedbackmechanismsItisimportanttoprovidemoreinformationandheuristicstohelptheconsumersnavigatethroughtheclutterandgettheinformationtheywantandseek

Itisimportanttoapplymodernsemanticanalysisandsentimentclassificationtechniquestostudythecontentofthereviewsandidentifyvariousdimensionsonwhichcon-sumersevaluatevacationhomesThedetailsprovidedinthereviewtextareinmanywaysabetterreflectionofcustomersatisfactionServiceproviderscanbetterpositionthemselvesusingthedimensionsuncoveredthroughtextanalysisandtargettheneedsandpreferencesoftheircustomers

TheresultsinthisstudyshowthatnegativereviewsaretypicallyassociatedwithagreaterwordcountthanpositivereviewsItisusefultofocusspecificallyonthesenegativereviewsbecausetheycanrevealpatternsofdeficienciesintheservicestandardsanddeliveryThesereviewsshouldbetreatedthesameascustomercommentcardsorletters19Furthercustomerswhowritesuchlengthynegativereviews

18LingKimberlyGerardBeenenPaulLudfordXiaoxingWangKevinChangXinLiDanCosleyDanFrankowskiLorenTerveenAlMRashidandRobertKrautldquoUsingSocialPsychologytoMotivateContributionstoOnlineCommunitiesrdquoJournal of Computer-Mediated Communication10no4(2005)19MilanR(2007)10thingsyoucandoinresponsetotravelerreviewsinHotelmarketingcom

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 15

aremostlikelytospreadnegativeword-of-mouthaboutthepropertyLengthyreviewsareexcellentsourcesofinforma-tionandshouldbeusedasthebasestounderstandthecomplainingbehavioroftheconsumer

WordcountmaybeusedasanadditionalsegmentationvariabletoidentifythoseconsumerswhoaremoreinvolvedandwillingtoprovidebalancedfeedbacktoboththereviewsiteaswellastopropertymanagementVariousotherresearchmethodssuchastelephoneinterviewsorfocusgroupsshouldbeusedtoelicitdetailedfeedbackfromsuchconsumersTheseactionsalsohelpinservicerecoverypost-purchaseengagementandrelationshipbuilding

OutlookWeforeseethreepotentialareasinvestigationthatcanaddvaluetocurrentbestpracticesWestilldonrsquotknowmuchaboutconsumersrsquomotivationtowriteonlinereviews20Recentstudiesidentifyvariousantecedentsofreviewwritingbutsufferfromsamplingbiassincetheysurveyaself-select-edsampleofconsumerswhoalreadyprovidedreviewsAstheresultsinthisstudyindicateunder-reportingbias(lackofmoderateandlowreviews)remainsprevalentinonlinereviewsitesFutureresearchshouldidentifysocio-psycho-logicalfactorsthatincreaseconsumersrsquopropensitytowriteonlinereviews

Futureresearchshouldexploretheimpactofthepres-ence(orlack)ofextremelynegativereviewswithinalargecorpusofpositivereviewsTheevidenceontheimpactofre-

20WangYandFesenmaierDR(2003)AssessingMotivationofCon-tributioninOnlineCommunitiesAnEmpiricalInvestigationofanOn-lineTravelCommunityElectronic Markets1333-45andYooKyungHandUlrikeGretzelldquoWhatMotivatesConsumerstoWriteOnlineTravelReviewsrdquoInformation Technology amp Tourism10No4(2008)283-95

viewvalence(negativeorpositive)isequivocalProponentsofconfirmatorybiassuggestthatconsumerslookforaffir-mativeevidencesupportingaproductchoicealreadymadeIfthatisthecasepositivereviewsaremorelikelytohaveagreatereffectonconsumeractionsOntheotherhandthenotionofnegativitybiassuggeststhatwhenconsumersareneutralnegativereviewstendtobecomemoresalientthanpositivereviews21Inthepresenceoflargenumbersofposi-tivereviewsconsumersspecificallyseeknegativereviewsthattheyfeelwillhelpthemidentifyspecificproblemswithaserviceUnderstandinghowconsumersreconciletheinformationprovidedbybothnegativeandpositivereviewshasimplicationsforthedesignandmanagementofreviewsites

Finallyareviewrsquoswordcounthasimportantimplica-tionsforconsumersrsquotrustinthereviewsWhenconsumersarewillingtoreadandcompareopen-endedcommentsfromotherconsumerstheamountofinformationinareviewcanmatterWOMforhighinvolvementproductsismorepersuasiveandincreasesthedecisionmakerrsquosconfi-dencewhenthemessagesenderprovidesgreaternumberofreasonsfortheiroverallratingLongerreviewsprovidemoreinformationandarelikelytobeperceivedasmorehelpfulandpersuasivethanshorterreviewsThewaysinwhichareviewiswrittenframedandpresentedisaninterestingfactorwhichhasrarelybeeninvestigatedandhasimportantimplicationsparticularlygivenourfindingthattheamountofinformationinareview(wordcount)variessignificantlywiththeratingsofthereviewn

21BaSandPavlouPA(2002)EvidenceoftheEffectofTrustBuildingTechnologyinElectronicMarketsPricePremiumsandBuyerBehaviorMIS Quarterly26243-268

Cornell Hospitality Quarterlyhttpcqxsagepubcom

2011 ReportsVol11No15DesigningaSelf-HealingServiceSystemAnIntegrativeModelbyRobertCFordandMichaelCSturman

Vol11No14ReversingtheGreenBacklashWhyLargeHospitalityCompaniesShouldWelcomeCrediblyGreenCompetitorsbyMichaelGiebelhausenPhDandHaeEunHelenChunPhD

Vol11No13DevelopingaSustainabilityMeasurementFrameworkforHotelsTowardanIndustry-wideReportingStructurebyEricRicaurte

Vol11No12CreatingValueforWomenBusinessTravelersFocusingonEmotionalOutcomesbyJudiBrownellPhD

Vol11No11CustomerLoyaltyANewLookattheBenefitsofImprovingSegmentationEffortswithRewardsProgramsbyClayVoorheesPhDMichaelMcCallPhDandRogerCalantonePhD

Vol11No10CustomerPerceptionsofElectronicFoodOrderingbySherylEKimes

Vol11No92011TravelIndustryBenchmarkingStatusofSeniorDestinationandLodgingMarketingExecutivesbyRohitVermaPhDandKenMcGill

Vol11No8SearchOTAsandOnlineBookingAnExpandedAnalysisoftheBillboardEffectbyChrisAndersonPhD

Vol11No7OnlineMobileandTextFoodOrderingintheUSRestaurantIndustrybySherylEKimesPhDandPhilippFLaqueacute

Vol11No6HotelGuestsrsquoReactionstoGuestRoomSustainabilityInitiativesbyAlexSusskindPhDandRohitVermaPhD

Vol11No5TheImpactofTerrorismandEconomicShocksonUSHotelsbyCathyAEnzRenaacutetaKosovaacuteandMarkLomannoVol11No4ImplementingHumanResourceInnovationsThreeSuccessStoriesfromtheServiceIndustrybyJustinSunandKateWalshPhD

Vol11No3Compendium2011

Vol11No2PositioningaPlaceDevelopingaCompellingDestinationBrandbyRobertJKwortnikPhDandEthanHawkesMBA

Vol11No1TheImpactofHealthInsuranceonEmployeeJobAnxietyWithdrawalBehaviorsandTaskPerformancebySeanWayPhDBillCarrollPhDAlexSusskindPhDandJoeCYLeng

2011 Hospitality ToolsVol2No1TheGameHasChangedANewParadigmforStakeholderEngagementbyMaryBethMcEuen

2011 Industry PerspectivesNo7MegaTips2TwentyTestedTechniquesforIncreasingYourTipsbyMichaelLynn

2011 ProceedingsVol3No5BuildingBrandsintheInternetAgeAnalyticsLoyaltyandCommunicationbyGlennWithiam

Vol3No4BraveNewWorldOnlineHotelDistributionbyGlennWithiam

Vol3No3SocialMediaandtheHospitalityIndustryHoldingtheTigerbytheTailbyGlennWithiam

Vol3No2TheChallengeofHotelandRestaurantSustainabilityFindingProfitinldquoBeingGreenrdquobyGlennWithiam

Vol3No1CautiousOptimismCHRSExaminesHospitalityIndustryTrendsbyGlennWithiam

2010 ReportsVol10No18HowTravelersUseOnlineandSocialMediaChannelstoMakeHotel-choiceDecisionsbyLauraMcCarthyDebraStockandRohitVermaPhD

Vol10No17PublicorPrivateTheHospitalityInvestmentDecisionbyQingzhongMaPhDandAthenaWeiZhangPhD

Vol10No16BestPracticesinSearchEngineMarketingandOptimizationTheCaseoftheStJamesHotelbyGregBodenlcosVictorBogertDanGordonCarterHearneandChrisKAndersonPhD

Vol10No15TheImpactofPrix Fixe MenuPriceFormatsonGuestsrsquoDealPerceptionbyShuoWangandMichaelLynnPhD

Cornell Center for Hospitality Research

Publication Indexwwwchrcornelledu

The Executive Path Hospitality Leadership Through Learning

Complete program information and applications available online

wwwhotelschoolcornelleduexecedPhone + 1 607 255 4919 Email exec_ed_hotelcornelledu

Professionals from around the world are invited to attend 3-day 10-day or online courses at the worldrsquos leading institute for hospitality management education in

Visit our website to apply

Explore develop and apply ideas with global hospitality leaders and

expert Cornell professors

Success

AdvancingBusiness

andPersonal

bull Strategic Leadershipbull Financebull Foodservicebull Human Resources

bull Marketingbull Operationsbull Real Estate

No2RetaliationWhyanIncreaseinClaimsDoesNotMeantheSkyIsFallingbyDavidSherwynJDandGreggGilmanJD

2009 ToolsToolNo12MeasuringtheDiningExperienceTheCaseofVitaNovabyKeshPrasadandFredJDeMiccoPhD

Cornell Center for Hospitality ResearchPublication Index

The Executive Path Hospitality Leadership Through Learning

Complete program information and applications available online

wwwhotelschoolcornelleduexecedPhone + 1 607 255 4919 Email exec_ed_hotelcornelledu

Professionals from around the world are invited to attend 3-day 10-day or online courses at the worldrsquos leading institute for hospitality management education in

Visit our website to apply

Explore develop and apply ideas with global hospitality leaders and

expert Cornell professors

Success

AdvancingBusiness

andPersonal

bull Strategic Leadershipbull Financebull Foodservicebull Human Resources

bull Marketingbull Operationsbull Real Estate

visit our website to apply

wwwchrcornell edu

Page 2: Unscrambling the Puzzling Matter of Online Consumer ... · Philips Hospitality Americas Chris Proulx, CEO, eCornell & Executive Education Carolyn D. Richmond, Partner, Hospitality

Advisory Board

The Robert A and Jan M Beck Center at Cornell UniversityBack cover photo by permission of The Cornellian and Jeff Wang

Cornell Hospitality Reports Vol 11 No 16 (August 2011)

copy 2011 Cornell University This report may not be reproduced or distributed without the express permission of the publisher

Cornell Hospitality Report is produced for the benefit of the hospitality industry by The Center for Hospitality Research at Cornell University

Rohit Verma Executive DirectorJennifer Macera Associate DirectorGlenn Withiam Director of Publications

Center for Hospitality ResearchCornell University School of Hotel Administration489 Statler HallIthaca NY 14853

Phone 607-255-9780Fax 607-254-2922wwwchrcornelledu

Niklas Andreacuteen Group Vice President Global Hospitality amp Partner Marketing Travelport GDS

Rarsquoanan Ben-Zur Chief Executive Officer French Quarter Holdings Inc

Scott Berman Principal Real Estate Business Advisory Services Industry Leader Hospitality amp Leisure PricewaterhouseCoopers

Raymond Bickson Managing Director and Chief Executive Officer Taj Group of Hotels Resorts and Palaces

Stephen C Brandman Co-Owner Thompson Hotels IncRaj Chandnani Vice President Director of Strategy WATGBenjamin J ldquoPatrickrdquo Denihan Chief Executive Officer

Denihan Hospitality GroupBrian Ferguson Vice President Supply Strategy and Analysis

Expedia North AmericaChuck Floyd Chief Operating OfficerndashNorth America

HyattGregg Gilman Partner Co-Chair Employment Practices

Davis amp Gilbert LLP

Tim Gordon Senior Vice President Hotels pricelinecomSusan Helstab EVP Corporate Marketing

Four Seasons Hotels and ResortsJeffrey A Horwitz Chair Lodging + Gaming and Co-Head

Mergers + Acquisitions ProskauerKevin J Jacobs Senior Vice President Corporate Strategy amp

Treasurer Hilton WorldwideKenneth Kahn PresidentOwner LRP PublicationsKirk Kinsell President of Europe Middle East and Africa

InterContinental Hotels GroupRadhika Kulkarni PhD VP of Advanced Analytics RampD

SAS InstituteGerald Lawless Executive Chairman Jumeirah GroupMark V Lomanno CEO Smith Travel ResearchBetsy MacDonald Managing Director HVS Global Hospitality

ServicesDavid Meltzer Senior Vice President Global Business

Development Sabre Hospitality SolutionsWilliam F Minnock III Senior Vice President Global

Operations Deployment and Program Management Marriott International Inc

Mike Montanari VP Strategic Accounts Sales - Sales Management Schneider Electric North America

Shane OrsquoFlaherty President and CEO Forbes Travel GuideThomas Parham Senior Vice President and General Manager

Philips Hospitality AmericasChris Proulx CEO eCornell amp Executive EducationCarolyn D Richmond Partner Hospitality Practice Fox

Rothschild LLPSusan Robertson CAE EVP of ASAE (501(c)6) amp President of

the ASAE Foundation (501(c)3) ASAE FoundationSteve Russell Chief People Officer Senior VP Human

Resources McDonaldrsquos USAMichele Sarkisian Senior Vice President MaritzJanice L Schnabel Managing Director and Gaming Practice

Leader Marshrsquos Hospitality and Gaming PracticeTrip Schneck Managing Partner District Hospitality PartnersK Vijayaraghavan Chief Executive Sathguru Management

Consultants (P) LtdAdam Weissenberg Vice Chairman and US Tourism

Hospitality amp Leisure Leader Deloitte amp Touche USA LLP

FriendsBerkshire Healthcare bull Center for Advanced Retail Technology bull Cruise Industry News bull DK Shifflet amp Associates bull ehoteliercom bull EyeforTravel bull 4Hotelierscom bull Gerencia de Hoteles amp Restaurantes bull Global Hospitality Resources bull Hospitality Financial and Technological Professionals bull hospitalityInsidecom bull hospitalitynetorg bull Hospitality Technology Magazine bull HotelExecutivecom bull International CHRIE bull International Hotel Conference bull International Society of Hospitality Consultants bull iPerceptions bull JDA Software Group Inc bull JD Power and Associates bull The Lodging Conference bull Lodging Hospitality bull Lodging Magazine bull LRA Worldwide Inc bull Milestone Internet Marketing bull MindFolio bull Mindshare Technologies bull PhoCusWright Inc bull PKF Hospitality Research bull Resort and Recreation Magazine bull The Resort Trades bull RestaurantEdgecom bull Shibata Publishing Co bull Synovate bull The TravelCom Network bull Travel + Hospitality Group bull UniFocus bull USA Today bull WageWatch Inc bull The Wall Street Journal bull WIWIHCOM bull Wyndham Worldwide

Thank you to our generous Corporate Members

PartnersDavis amp Gilbert LLP Deloitte amp Touche USA LLPDenihan Hospitality GroupeCornell amp Executive EducationExpedia Inc Forbes Travel GuideFour Seasons Hotels and Resorts Fox Rothschild LLP French Quarter Holdings Inc HVS Hyatt InterContinental Hotels Group Jumeirah GroupLRP PublicationsMarriott International IncMarshrsquos Hospitality Practice MaritzpricelinecomPricewaterhouseCoopersProskauer Sabre Hospitality SolutionsSathguru Management Consultants (P) Ltd Schneider Electric Thayer Lodging Group Thompson HotelsTravelportWATG

Senior PartnersASAE FoundationCarlson HotelsHilton WorldwideMcDonaldrsquos USAPhilips HospitalitySASSTRTaj Hotels Resorts and Palaces

4 TheCenterforHospitalityResearchbullCornellUniversity

AbouT The AuThors

UnscramblingthePuzzlingMatterofOnlineConsumerRatings

An Exploratory AnalysisbyPradeepRacherlaDanielConnolly

andNatasaChristodoulidou

Pradeep racherla PhD is an assistant professor of marketing at the College of Business West Texas AampM University (pracherlawtamuedu ) He earned a PhD in business administration from the Fox School of

Business Temple University His research interests include consumer generated media healthcare marketing social networks knowledge management and collaborative systems He has written several research papers

and technical reports that have been published or forthcoming in prestigious journals such as Journal of Consumer Behavior Journal of Marketing amp Management Journal of Management Information Systems

Annals of Tourism Research and Cornell Hospitality Quarterly He has presented at numerous conferences both nationally and internationally

Daniel J Connolly PhD is an associate professor of information technology at the Daniels College of Business at the University of Denver with a dual appointment in the School of Hotel Restaurant and Tourism Management and in the Department of Information Technology and Electronic Commerce He also serves as director of undergraduate programs His teaching research and consulting interests focus on the strategic application of information technology and electronic commerce He has written or co-authored numerous publications including articles that have appeared in the Cornell Hotel and Restaurant Administration Quarterly the FIU Hospitality Review Information Technology in Hospitality and Journal of Hospitality and Leisure Marketing His book Technology Strategies for the Hospitality Industry (Pearson Prentice Hall 2005) is the industryrsquos first book on technology strategy

Natasa Christodoulidou PhD is a faculty member in the management and marketing department at California State University where she teaches internet marketing and other related marketing courses at the

undergraduate and MBA level Her research interests are in the areas of hospitality technology travel and tourism electronic commerce hospitality electronic distribution and electronic marketing She is a member of

the Decision Sciences Institute and the Western Decision Sciences Institute She has published more than 20 articles in the past five years in academic and professional journals She presents regularly at academic and

professional conferences around the world such as Eye for Travel FSTEC DSI and WDSI Among her speaking engagements this year are Las Vegas Hawaii France and Singapore

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 5

exeCuTive suMMAry

Thisstudyexploresthepatternsofonlinereviewsofvacationhomesfromacommunity-basedtraveladvisorywebsitewithagoalofunderstandingthebiasesinherentinonlinewordofmouth(WOM)relatedtotourismandhospitalityservicesAnanalysisofnearly3200reviewsfrom

ldquoReviewsitecomrdquo(apseudonym)whichpostsreviewsofvacationrentalpropertiesacrosstheUSAfindsanoverwhelmingpreponderanceof favorable reviewsMore to thepoint relatively few

ldquomoderaterdquoreviewsarepostedandthesecond-highestcategoryisextremelynegativecommentsUsingsemanticprocessingtechniquesontheaggregatereviewtextthestudyidentifiesthenuancedopinionsandconcernsofthetravelerswhowritereviewsNegativereviewstendtobelengthyandargumentativeoftendetailingdisappointmentoverexpectationsnotmetPositivereviewsontheotherhandtendtoberelativelybriefandconfirmtheoverallratingConsumerswhowroteldquohighrdquoreviewsplacedgreaterimportanceonvalueformoneycleanlinessandcomfortthandidthosewhowrotenegativereviewsThosewhowroteldquolowrdquoreviewsplacedtheiremphasisontheserviceprovidedbythepropertystaffandmanagementNegativereviewsweremorelikelytoinvolveahigherpriceaccommodationThisanalysisindicates that the overall numerical ratings typically used in review systems may not be the idealindicator of perceived service quality The results suggest that review sites should develop bettermethodstoaggregatesynthesizeandpublishthereviewcontentsparticularlythenumericalratingsThisandotherreviewsitesshowtheaverageofallthepoint-scaleratingsbutsuchsimplemeansdonottakeintoaccountthebiasesthatareinherentintheratingsystemsInsteadthesitesshouldprovidemore information and heuristics to help the consumers navigate through the clutter and get theinformationtheydesire

6 TheCenterforHospitalityResearchbullCornellUniversity

CorNell hosPiTAliTy rePorT

Giventhatconsumersareincreasinglyrelyingonsearchenginestoacquireinformationabout hotels restaurants and travel purveyors (among many other services)consumer-generated online reviews will inevitably change the structure andaccessibilityofinformationalongwithconsumersrsquoperceptionsofvariousservices

ThepopularityofonlinereviewshasledtoaproliferationofonlinereviewsitesincludingpopularsitessuchasHotelscomTripadvisorcomandYelpcomArecentstudybyForresterIncfoundthatmorethan80percentofwebshoppersreadandconsiderotherconsumersrsquoreviews1SimilarlyCompeteIncfoundthatnearly50percentofpurchasersvisitedamessageboardforumoronlinecommunityforinformationconnectedwiththeironlinetravelpurchasingandoneinthreeofthesebuyerssaidthatconsumer reviews helped with their purchase decision2 Almost half of these consumers said thatconsumersrsquoopinionsactuallycausedthemtochangetheirmindaboutwhattheypurchasedMoreoveramongthosebuyers25percentsaidtheyalsopostedareviewonaconsumerreviewsiteaftermakingtheirpurchase

1ForresterldquoFortyFactsaboutUSOnlineShoppersrdquoForresterInc20062CompeteldquoEmbracingConsumerBuzzCreatesMeasurementChallengesforMarketersrdquoCompeteInc2006

UnscramblingthePuzzlingMatterofOnlineConsumerRatings

An Exploratory Analysis

byPradeepRacherlaDanielConnollyandNatasaChristodoulidou

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 7

Howevertheexistingunderstandingofonlinetourism-relatedreviewsisrelativelyweakeventhoughweknowthatwordofmouth(WOM)ismoreimportantandinfluentialwithinaservicescontextthaninproductmarketingscenariosRecentlysomestudieshaveexploredtourism-andhospital-ity-relatedonlineWOM3Yetresearchhassofaroverlookedcertaininterestingaspectsofonlinereviewsasfollows

Onlinereviewstypicallyuseldquooverallstarratingsrdquoaspri-maryindicatorsofperceivedservicequalitymdashratingsthatarenotrelatedinanywaytotheformalstardesignationsawardedbyForbesorMichelinConsumersrsquomeanstarratingsaretypi-callyaggregatedfromtheratingsgivenbyindividualconsum-ersineachreviewWhatwedonrsquotknowishowconsumersdeterminethestarstheyassigntoaservicewhentheypostareviewonawebsiteThatisdostarratingsaccuratelyreflectconsumersrsquosentimentsregardingvariousattributesofserviceprovidersThisisavitalissuegiventheexperientialandin-tangiblenatureoftravelservicesandthefactthatconsumersusetheoverallratingasanimportantheuristicwhilenarrow-ingdownchoices

FurtherstudieshaveoverlookedtherelationshipbetweenthereviewsandvariousattributesofservicesandserviceprovidersSinceonlinereviewsareaformofWOMvariablessuchasconsumerinvolvementpricingandlengthofstaymayhaveanimpactonthefinalratingandpostingofthereviewprovidedbyconsumersItisimportanttouncovertheserelationshipstobetterunderstandthedynamicsofonlinereviews

FewstudieshaveanalyzedthequalitativeaspectofonlinereviewsOnlinereviewsareessentiallyopen-endedtext-basedconsumer-to-consumer(c2c)communicationsThetextcontentmaycontainnuancedviewsoftheservicesandserviceproviders(fromthereviewwritersrsquopointofview)thatcannotbeexpressedusingcrudenumericalratings4Hencethereisaneedtoanalyzethetextportionofthereviewsandidentifytheissuesthatconsumersaremostlyconcernedwithwhenwritingandpostingonlinerelatedreviews

WeaddresstheabovepointsinthisstudyThefollowingquestionsguidedus

3KHYooandUGretzelldquoWhatMotivatesConsumerstoWriteOnlineTravelReviewsrdquoInformation Technology amp TourismVol10(2008)pp283-295andZXiangandUGretzelldquoRoleofSocialMediainOnlineTravelInformationSearchrdquoTourism Management20094AGhoseandPGIpeirotisldquoEstimatingtheSocio-economicImpactofProductReviewsMiningTextandReviewerCharacteristicsrdquoInformation Systems Research2008

1 Dooverallratingsreflecttheperceivedqualityoftheserviceproviders(asderivedfromtheconsumersrsquoexpressedsentimentinthereviews)Whataretheunderlyingpatternsoftheoverallratingsinonlinereviews

2 Whatarethelinkagesamongthevariousattributesoftheserviceprovidersandcustomerreviewsand

3 WhataretheissuesthatconsumersmostlytalkaboutinthetextportionofonlinereviewsDotheseis-suesdifferfromtheratingsthattheconsumershaveprovidedviathestandardvariablesprovidedbythereviewsites

ResearchSettingandDataCollectionThisstudyisbasedonreviewscollectedfromldquoReviewsitecomrdquoaswehavecalledthecommunity-driventravelsitethatweanalyzedThesiteprovidesreviewsofverifiedrentalpropertiesandorganizesthereputationofvacationpropertiesThepropertylistingwithinthesiteisbasednotontherevenuesharingmodelthatisgenerallypracticedbyotherreviewsitesbutonthereputationtrustandfeedbackfromcustomerswhohavestayedatthevacationrentalpropertiesReviewsitecomisafreeservicethatoperatesonserviceproviderfeesandadvertisingrevenuesUn-likesometravelreviewsitesthefirmrsquosreviewsolicitationstrategyistosendpersonalizedinvitationsonlytothosecustomerswhosestayhasbeenverifiedbythepropertymanagementThesitersquosmanagementbelievesthatthisstrategytoalargeextenteliminatesfakereviews

Thefirmprovideduswith3300reviewsforanalysisAfterinitialdatacleanupabout100reviewswithincom-pleteorunusableinformationwereeliminatedand3197reviewswereusedfortheanalysisThereviewscontainedthefollowingattributesbull Theoverallratingassignedtotheserviceprovider

basedontheaggregationoftheratingsassignedbyindividualreviewers

bull Consumersrsquoratingsonthefollowingsixattributesvaluecleanlinesscomfortservicelocationandcheck-in

bull Thereviewtext(Semanticprocessingtechniqueswereappliedontheaggregatereviewtexttoidentifythenuancedopinionsandconcernsofthetravelerswhowrotereviews)

8 TheCenterforHospitalityResearchbullCornellUniversity

bull Agerangeoftherevieweronan1-4ordinalscale(25-3435-4445-54and55andabove)

bull Averagepriceperdayand

bull Wordcount(usedinthequantitativeanalysis)

PreliminaryHypothesesandResultsTheagegroupsofthereviewwritersareshowninExhibit1Alargepercentageofthecustomerswhoreviewedthepropertiesfallunder44yearsofageThisisconsistentwithpreviousstudiesinthisrealmwhichshowthatgenerationXandGenerationYaremoreactiveintheWOMforums

ReviewRatingDistribution Websitesthathostonlinereviewsgiveconsumerstheoptiontoassignaratingtotheserviceproviderstypicallyonascaleof1to5or1to7Manysitesaggregatethesenumeri-calorstarratingstodetermineanoverall(mean)starratingWeseethisoverallratingasanimportantelementinonlineWOMsinceconsumersoftenusethisratingtonarrowdowntheirconsiderationsetStudiesthathaveusedtheratingastheprimarypredictoroftheserviceprovidersrsquoperceivedqualityorcustomersatisfactiontodeterminetheimpactofonlinereviewsonproductsaleshavearrivedatmixedfind-ings5Howeverthesestudiesarebasedonthefundamentalassumptionthatthenumericalratingsassignedtoservice

5ForexampleseeJChevalierandDMayzlinldquoTheEffectofWordofMouthOnlineOnlineBookReviewsrdquoJournal of Marketing ResearchVol43(2006)pp345-354andWDuanBGuandABWhinstonldquoDoOnlineReviewsMattermdashAnEmpiricalInvestigationofPanelDatardquoDecision Support SystemsVol45(2008)pp1007-1016

providersareanaccurateestimateofthequalityperceivedbythecustomerThispracticefollowsthebasicassumptionofbehavioraltheoriststhatanydatasetwithalargenumberofconsumerresponsestendstofollowanormal(Gaussian)distributionRecentstudieshavequestionedthisassump-tionForinstanceDellaracosandNarayanrsquosstudyonmoviereviewsinYahoocomfoundthatonlytheextremelysatisfiedandextremelydissatisfiedcustomersarelikelytopubliclyexpresstheiropinionsascomparedtocustomerswithmod-erateopinions6SimilarlystudiesusingdatafromTripadvi-sorcomandAmazoncomfoundthatthestarratingsexhibitatruncateddistributioninwhichthemajorityofreviewsarepositive7

TotestthisassumptionweranasimplefrequencychartasshowninExhibit2Inkeepingwithotherstudiesthedistributionisheavilyskewedtowardsthepositiveratings1-pointreviews32(1)2-pointreviews95(3)3-pointreviews458(14)4-pointreviews688(22)and5-pointreviews1912(60)Totestfortherobustnessofthisdistri-butionweappliedtheKolmogorov-Smirnovtestofnormal-ityTheresultsconfirmthenon-normaldistributionofthestarratingsandtheheavyskewtowardspositivereviews

Threeexplanationsarepossibleforsuchadistribution

6CDellarocasandRNarayanldquoAStatisticalMeasureofaPopulationrsquosPropensitytoEngageinPost-purchaseOnlineWord-of-mouthrdquoStatisti-cal ScienceVol21(2006)pp2777ATalwarandRFBoiJurcaUnderstandingUserBehaviorinOnlineFeedbackReportingrdquoinElectronic CommerceSanDiegoCalifornia(2007)pp134-142

Exhibit 1

Distribution of age ranges

Exhibit 2

Distribution of overall ldquostar ratingsrdquo

55 and above 45ndash54

years

35ndash44 years

18ndash34 years

overall ldquostar ratingrdquo

Num

ber o

f rev

iew

s

3295

458 688

1912

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 9

1 AmajorityofthepropertiesthatarelistedonthewebsitehaveoutstandingperceivedqualityandprovidesatisfactoryservicestocustomersWhilethisistheo-reticallypossibleitseemsunlikelywhenappliedtoallthepropertieslistedonthewebsite8

2 ThereviewsdonotconstituteatrulyrepresentativesampleofconsumeropinionBasedonotherstud-iesthisexplanationseemsplausibleConsumerswithmoderateviewsmayseelittleincentivetoreporttheirviewsinanonlineforumandasaconsequenceonlythoseconsumerswhohavepolarizedviewsarelikelytoposttheircomments

3 ThedistributioncouldalsobeanartifactofthefirmrsquosreviewsolicitingstrategyRecentstudiessupportthisideabysuggestingthatcustomersengageinsociallydesirablebehavioreveninimpersonalinteractionswiththeserviceproviders9Inthiscaseitmaybethattheywillprovideareasonablypositiverevieweveniftheexperiencehasnotbeenuptothemark

StarRatingsandPerceivedServiceQualityLetrsquosreturntothefirstquestionweaskedattheoutsetArestarratingsatruerepresentationofserviceprovid-ersrsquoperceivedqualityasreflectedbytheconsumersrsquostatedsentimentsThisbecomesanimportantquestionfortwo

8AKadetldquoRah-RahRatingsOnlinerdquoSmartMoney20079YGreacutegoireandRJFisherldquoCustomerBetrayalandRetaliationWhenYourBestCustomersbecomeYourWorstEnemiesrdquoJournal of the Acad-emy of Marketing ScienceVol36(2008)pp247-261

reasons(a) ratingsaretheprimaryevidencethatconsum-ersconsiderwhileshoppingonlineforinformationand(b)thetruncatednatureofratingdistributionsmeansthatthereviewsavailablemaynottrulyrepresentconsumeropinionortheserviceprovidersrsquoperceivedqualityIfonlythosewithextremeviewspostreviewsthenconsumersshouldalwayslookforserviceprovidersthathavea5-starratingOnthecontraryifthereviewsaccuratelyreflectallconsumersrsquoviewsthenconsumersarebetteroffsearchingforreviewswitha3ratingsincethesetendtobemorebalancedandenablebetterproductanalysisRecentlyHu et aluseddatafromAmazoncomshowthatnumericalscoresdonotreflectbooksrsquotrueperceivedquality10AstheysuggestldquohellipratherthescorereflectsthebalanceofdiverseopinionsInotherwordswhenabookrsquosoverallscoreisaround3itdoesnotsuggestthatconsumersgenerallyagreethatthisisanaver-agebookItrathersuggeststhatroughlyequalnumberofconsumersthinkthatthebookiseitheranoutstandingbookoranabysmalbookrdquo

Totestthisassertionweappliedaregressionequationwithsixpropertyattributes(ievalueformoneycheck-inlocationcleanlinesscomfortandservice)asindependentvariablesandstarratingasthedependentvariableThisanalysisisbasedontheassumptionthatforagivenpropertyifstarratingstrulyrepresenttheperceivedqualityoftheproperties(aswellastheconsumersrsquosentiment)thenthestarratingsandratingsofthesixotherpropertyattributes

10NHuJZhangandPAPavlouOvercomingtheJ-shapedDistribu-tionofProductReviewsrdquoCommunications of the ACMVol52(2009)pp144-147

low reviews t-value high reviews t-value(Constant) 1618 18353 2400 34186vAlue 0254 13726 0378 8767CheCKiN -0083 -3701 -0053 -1192loCATioN 0104 6132 0079 2117CleAN 0144 7574 0112 2808CoMForT 0267 13993 0129 3088serviCe 0098 4209 0107 2401r2 0331 0338F-statistic 5038 24714

Exhibit 3

regression analysis of ldquolowrdquo and ldquohighrdquo reviews

Notes Coefficients (t-statistics reported) Significance is as follows significant at 0001 level significant at 05 level significant at 010 level Dependent variable is overall rating

10 TheCenterforHospitalityResearchbullCornellUniversity

moderateratingsonalltheattributesbutwasstillwillingtoprovideahighoverallstarratingtothepropertyThisexamplesuggeststhatconsumersplacesignificantemphasisontheoverallexperientialvalueprovidedbythepropertiesThisexpectationofvalueismanytimesintangibleanddoesnotnecessarilydependonthefunctioningorratingoftheindividualfeaturesoftheproperty

RelationshipbetweenRatingsandReviewAttributesTounderstandconsumersrsquoword-of-mouthbehavioritisimportanttoconsidervariousfactorssuchasperceivedexpectationsofcostsandbenefitsimportanceofthepur-chase(involvement)personalcharacteristicsandsituationalinfluencesWeconsidertwofactorsthatareknowntoaffecttheintensityandvalence(positiveornegative)ofwordofmouthpriceandwordcount

PricePriceisanimportantfactorthatcontributestopre-purchaseexpectationsTheintensityofexpectations(andtheirfulfillment)partlydrivenbypriceisdirectlypropor-tionaltoconsumersrsquosatisfactionlevelsandWOMpropensityThismaybethereasonthatthetendencytocomplainisdirectlyrelatedtothecostoftheservice11Themoreexpen-siveanitemthegreaterwillbetheperceivedlossesandthegreaterlikelihoodthatdissatisfactionwillresultinacom-plaintSimilarlythelikelihoodofword-of-mouthbeingtrig-geredbythelevelofexpectationndashdisconfirmationincreases

11BeardenWOandTeelJE(1983)Selecteddeterminantsofcon-sumersatisfactionandcomplaintreportsJournal of Marketing Research 2021-28

willshowahighcorrelationFortheanalysiswetookintoconsiderationthefactmostreviewsarepositive(ie4-and5-pointratings)Thereforewedividedthereviewsintotwocategorieshighreviews(starratingabove3n=2597)andlowreviews(ratingsbeloworequalto3n=600)andranseparateregressionanalysesonthesetwogroupsasshowninExhibit3

AninterestingfindingistheweakrelationshipbetweentheattributeratingsandoverallstarratingforeachpropertyTheR-squareoftheregressionequationwithallsixvariablesandtheoverallstarratingsislessthan40percentThatiseveniftheconsumersprovidedhigherratingsonindividualattributestheiroverallratingwasusuallylowandthere-versewasalsotrueThistrendisobservedinboththehighandlowreviewgroups

TheanalysissuggeststhatinhighreviewsconsumersplacedgreaterimportanceonvalueformoneycleanlinessandcomfortThehighertheratingsofaconsumerontheseattributesthehigheristhelikelihoodthattheconsumerwillprovideahigherstarratingCheck-inlocationandservicedidnotprominentlyfigureintheiroverallevalu-ationsofthepropertieseventhoughtheseattributeswereratedhigh

ThelowreviewsshowsimilarresultsinthesensethattheloweraconsumerratesvaluecleanlinessandcomfortthelowerthelikelihoodofapositiveratingInadditioninthecaseoflowreviewsconsumersplacedemphasisontheserviceprovidedbythepropertystaffandmanagement

TheaboveresultisillustratedinoneoftheldquohighrdquoreviewsshowninExhibit4Thereviewergavelowto

star rating Text Commentary

This was our 1st time renting a house vs a condo This house was great If you wanted to spend a little time by yourself you could so many rooms to go to The house was so comfortable like being at home It was our home for the week Thank you Hope to rent it next year if we havent bought one ourselves

ratings for specific Attributes

value Check-in location Clean Comfort service

3 2 3 2 4 3

Exhibit 4

example of a ldquohighrdquo review

Avg rating for Property Attributes

30

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 11

withthecostoftheproduct12Giventhegreaterperceivedriskinservicescustomerswouldseekmoretrustworthyser-viceproviderswithagoodreputationofsatisfyingcustomerneeds13ThisinturnincreasestheexpectationsandleadstomoreextremereactionsfromtheconsumersHenceonecanspeculatethatthehigherthepriceshigherwillbetheaver-agenegativeorpositiveratingsforaservice

Word countThenumberofwordsinareviewmayserveasaproxyfortheamountofinformationinthereviewsHowevertheevidenceonthisaspectofWOMofalltypesisequivocalForinstanceconsumersgenerallyengageinpositiveWOMtoavoidfeelingsofguiltandas-sociationwithbadnewsandtoreducecognitivedissonanceOntheotherhandacustomerwhoishighlydisappointedwillprovideapersuasiveargumenttomakesurethatothercustomersaredissuadedfromrentingthesamepropertyResearchonconventionalwordofmouthhasofferedampleevidencetosuggestthatdissatisfiedcustomersengageintwotothreetimesasmuchWOMassatisfiedcustomers14Webelievethatthislogicextendstoonlinereviewsdespitethefactthatexistingstudiesareequivocalatbest15Thereforeitcanbeassumedthatreviewswitheitherextremelypositiveorextremelynegativeratingswillhavegreaterwordcountasopposedtomoremoderatereviews

TotesttheseassumptionsweappliedANOVAwithpriceandwordcountasdependentvariablesandthereviewratingasfactorsAsillustratedinExhibits5and6theresultsshowthatpriceandwordcountshowasignificantnegativecorrelationwithstarratingThatisreviewswithlowerratingsaretypicallyassociatedwithgreaterwordcountandrepresentpropertiesthatchargehighrates

TheresultonthewordcountsuggeststhatconsumerswhoareunhappywitharentalpropertytendtowritemoretoexpresstheirdissatisfactionAsshowninthisexhibitlowerratingsareassociatedwithhighernumberofwordsinthereviewsandviceversaSimilarlypropertypriceseemstobestronglyassociatedwithnegativewordofmouthsup-portingtheprinciplethathighercostincreasestheconsum-ersrsquoinvolvementinthepurchase

12BeardenWOandMasonJB(1984)AninvestigationofinfluencesonconsumercomplaintreportsAdvancesinConsumerResearch11490-49513WetzerIMZeelenbergMandPietersR(2007)ldquoNevereatinthatrestaurantIdidrdquoExploringwhypeopleengageinnegativeword-of-mouthcommunicationPsychology and Marketing24661-68014SchlossbergH(1991)CustomersatisfactionNotafadbutawayoflifeMarketing News25andWestbrookR(1987)ProductConsump-tion-basedAffectiveResponsesandPost-purchaseProcessesJournal of Marketing Research 24258-27015AlexMSusskindldquoIToldYouSoRestaurantCustomersrsquoofWord-of-MouthCommunicationPatternsrdquoCornell Hotel and Restaurant Adminis-tration QuarterlyVol43No2(April2002)pp75-85

Exhibit 5

Differences in word count across star ratings

Exhibit 6

Differences in average price per day across ratings

Aver

age

pric

e pe

r day

of t

he re

ntal

pro

pert

y

overall ldquostar ratingrdquo

1504

1563

1804

19861987

Aver

age

wor

d co

unt o

f rev

iew

overall ldquostar ratingrdquo

836837

121

15751543

160

140

120

100

80

200

190

180

170

160

150

1 2 3 4 5

1 2 3 4 5

12 TheCenterforHospitalityResearchbullCornellUniversity

thereviewersThesecanbeclassifiedas(a)attributesorfea-turesofthepropertythatfigureprominentlyandcontributetothepositiveornegativeratings(egroomsbeddishesfurnitureorservice)and(b)thepositiveornegativewordsthatrepresentthesentimentofthereviewer(eggreatamazingdirtyandwonderful)Theresultsofthequalitativeanalysisareasfollows

Syntactic featuresTheoveralltoneofthelowreviewsisratherargumentativeThereviewerstendtobeforcefulinpresentingtheirpointofviewTheylayoutagreatamountofreasoningandprovidespecificexamplestodrivehometheirargumentThisisreflectedinthegreaterpercentageofnegativeconnectorwords(34)suchasbut however as ifandeven thoughItisinterestingtonotethatreviewersinthiscategoryspeakdirectlytotheotherconsumersnotablythroughgreateruseofwordssuchasyou(46)andlooktoinfluencetheirdecisionmaking

InthehighreviewsbycontrasttheoveralltonetendstobeenunciativeandthereviewersusethetexttomerelyreinforcetheirnumericalratingsusingfewerwordsonaveragethanthenegativereviewersInthesereviewstheconsumersfocusontheirpersonalexperiencesandappealtotheaudiencethroughemotionsratherthanlogicandreasoningasevidentfromgreateruseofpositiveadjectives

QualitativeAnalysisoftheReviewTextTheaboveanalysisemphasizestheimportanceofthetextcontentinconsumerreviewsWhilenumericalratingsallowthereviewerstoratetheserviceprovideronvariousattributesitisthetextthatprovidesthemwiththeopportu-nitytoarticulatethenuancesoftheiroverallexperienceandconveyusefulinformationaboutaserviceproviderrsquostransac-tionsandservicecapabilitiesThereforesemanticanalysisofthetextcontentprovidesuswithanopportunitytothethirdquestionpresentedattheoutsetofthisreportwhatdoconsumerstalkaboutintheirreviews

Tothisendweanalyzed600eachofrandomlyselectedhighandlowreviewsusingsemanticprocessingtechniquesWeusedcenteringresonanceanalysis(CRA)amodeofnet-work-basedtextanalysisthatrepresentsthecontentoflargesetsoftextsbyidentifyingthemostimportantwordsthatlinkotherwordsinthenetwork16OuranalysisexaminedsyntacticandsemanticfeaturesSyntacticfeaturesincludethepercentageofthewordsthatareopenclass(ienounsverbsadjectives)Semanticfeaturesrepresentthespecial-izedvocabularythatreflectstheoverallopinionexpressedby

16CormanSandDooleyK(2006)CrawdadTextAnalysisSystem12CrawdadTechnologiesLLCChandlerAZ

Prominently Mentioned Features Freq sample Comments

bedroom beds linen sheets 383

Also the same bedroom and bathroom did not have any door separating the bedroom from the bathroom no privacy

Sleeping arrangements were very awkward The listing sheet described 2 full beds We brought our own linens the beds were queen sized Had to purchase sheet

Kitchen Kitchenware Dishes Dishwasher 297

You must not forget to step up when yoursquore walking out of the kitchen into the dining room otherwise yoursquoll trip over the 3-foot step up

No potholders in the kitchen and it didnrsquot seem like there were as many kitchen supplies as before We just had to make do without

beach beaches beachfront 296

The week was the best vacation we have ever had as a family The beach house was a class act

Disappointment 1 It looks like this house is very close to beach but to get there you have to go around through public path

Pool Poolside 218

The pool was too dirty to use We checked in on Thursday and the pool was not cleaned until Saturday

Also it is inconvenient to have to call to get someone to open the pool Didnrsquot understand why Jacuzzi was open and not the pool

bathroom(s) bath bathtub shower 202

The next problem was in another bathroom (main floor)mdashthe wall heater did not work so showering in here was a bit chilly The next problem was in the ldquomasterrdquo bathroom (connected to the bedroom with the King bed) The sliding shower doors need to be totally replaced

Bathtub drains clogged in three second floor bathrooms

Exhibit 7

Top five prominent issues in low reviews

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 13

suchasgreat fantastic andawesome(45)andpersonalpronounssuchasweandI(54)

Semantic featuresExhibits7and8showthemainissuesoraspectsthatareprominentlydiscussedinthelowreviewsTheorderoftheseissuesisbasedonthestatisti-callysignificantfrequencyoftheiroccurrenceinthe1200reviews

WefurtherinvestigatedtheuseofvarioussentimentwordsInterestinglytheprominentissuesdiscussedinthepositivereviewssometimesdonotevenfigureintheratingscalesprovidedtothereviewersOveralltheresultsofthesemanticanalysiscanbesummarizedasfollows

Thepropertyanditsfeaturesaremorecommonlycon-nectedwithextremelypositivesentimentwordssuchasgreat wonderfulandamazinginthehighreviewsthaninthelowreviewsIncomparisonthepropertiesandtheirfea-turesinthelowreviewsaremorelikelytobecharacterizedwithmoderatelypositivewordssuchasniceandokay

Negativereviewsreflecttheextentofexpectationndashdis-confirmationthattheconsumersfaceduringtheirstayintherentalpropertyThesereviewshaveagreaterproportionofwordsexpressingdisappointment(egnotasexpected)Furtherextremelynegativesentimentwords(egdiscom-fortfilthy)andadjectivesaremorelikelytobeassociatedwithfeaturesinlowreviewsthaninhighreviews

Almost40percentoftheconsumerswhowrotehighreviewsexpresstheirwillingnesstoreturntothepropertyasopposedtoonly9percentofthosewritingnegativereviewsFurthermorethan20percentoftheconsumersgivinghighreviewsexpresstheirstrongdesiretorecommendtheprop-ertytotheirfriendsandfamilymembersasopposedtoonly5percentinlowreviews

ReviewofFindingsOurexploratorystudyofconsumerreviewsofUSvacationhomesonReviewsitecomwashighlightedbyoverwhelm-inglypositivereviews

Theoveralldistributionofthereviewswasheav-ilyskewedtowardsthepositiveratingsMorecriticallythisfindingsuggeststhatthereisalackofmoderateandbalancedopinionsWetheorizethatthisiseitherduetopurchasingbias(sincecustomerswentthroughthetroubleofpurchasingandexperiencingthefacilitiestheytendtobepositivelybiasedtowardstheirpurchase)orunder-reportingbias(customersdonothavestrongenoughincentivestotakethetroubleofreportingtheiropinionsandthefewmotivatedindividualswhoprovidetheirreviewsbecomeaself-selectedsample)

Theoverall(mean)starratingtypicallyusedintravelerreviewsitesisprobablynotthemostaccurateindicatorofcustomersrsquoperceptionsofserviceprovidersrsquoqualityWe

Prominently Mentioned Features Freq sample Commentsbeach beaches beachfront 506 Perfect rental property for those who enjoy the activity of the beach and the serenity

of the Sound The house is just down the road from ocean

The large porch offered a beautiful view of the beach and comfortable new rocking chairs

Pool Poolside 218 The outdoor shower dipping pool and screen in porch were an added bonus Perfect house for a low key weekend

House was great for kids Pool was terrific and the playroom was perfectlocation 192 We had a wonderful time My wife loves fishing and the cottage was in a perfect

location to go fishing on the pier

The home is somewhat dated but the location was great We have vacationed in the Chatham area for many summers and this time it was great too

bedroom beds 186 The family suite is huge with a king size and twin bed and still room for a pack-n-play The kitchen and dining room were huge clean and nice

The only comments that I would say to the negative is the front bed room by the front door always seemed damp which seem to attract more bugs

Kitchen Kitchenware Dishes Dishwasher 178 All of the amenities in the house are upscalemdashloved the kitchen and plasma TVs

The kitchen had good basic utensils dishes and cooking pans

Exhibit 7

Top five prominent issues in high reviews

14 TheCenterforHospitalityResearchbullCornellUniversity

foundacorrelationbetweenthestarratingandratingsonotherpropertyattributesoflessthan40percentinmostcases

AhighwordcountinthereviewsiscorrelatedwithalowoverallratingThatisconsumerswhoareextremelydissatisfiedtendtoexpendmoretimeandenergyontheircritiqueofthepropertyitsamenitiesandservicesPriceofthepropertyalsoseemstopositivelycorrelatewithextremenegativereviewsperhapsbecauseahighpriceheightenstravelersrsquoexpectations

Numericalratingsdonotoftencapturetheactualsentimentandthevarietyofdimensionsonwhichtravel-ersevaluatepropertiesBecausenumericalratingsdonotcapturereviewersrsquoopinionsaboutspecificattributestheycanbedeceivingandmaynotrevealthetruequalityofthepropertyortheactualintentionofthereviewerWefoundforinstancethatconsumersvaluelocation(egclosenesstothebeach)andcleanliness(egswimmingpoolareas)morethansuchfeaturesascheck-inandserviceSimilarlyinthenegativereviewsissuessuchaslackofbedroomparapher-naliaorkitchenitemsseemtoberankedhigherinimpor-tancethanissuessuchaslocationandcheck-inThereforeevenwhenapropertyscoreshighonthecertaindimensionstheoverallratingmayremainlowbecauseotherfactorsovershadowthefavorableaspects

WhatItMeansforManagersWeseeimplicationsforonlinereviewsitesandbyexten-sionforproprietorsofvacationhomesandserviceopera-tionsgenerallyWithregardtothereviewsitesifconsumersconsistentlydetectafavorablebiasthereviewsitesmaylosethecriticalaspectsofcredibilityandtrustAlthoughconsumersaregenerallysmartandrationalandcanworkaroundobviousbiasesrecentevidencesuggeststhatcon-sumersaregraduallydefectingfromsomeonlinesitesduetounbalancedreviews17InthisregardsiteadministratorsaswellaspropertymanagersshouldtakestepstoattractthoseconsumerswhoarewillingtoprovideareasonablybalancedanalysisoftheamenitiesandservicesBythiswemeanreviewsthatpresentcomprehensiveandobjectiveanalysisoftheserviceproviderratherthanjustravesorrants

Tothatendsiteoperatorsshouldtakestepstoun-derstandcustomersrsquoreviewwritingbehaviorandtoat-tractmorebalancedreviewsOnewaytodothisistogivepersonalizedinformationtotheconsumerdemonstratingthecontributionmadebyhisorherreviewandthevalueofthereviewtostakeholdersForinstanceLinget alfoundapositivecorrelationbetweenwillingnesstowritereviewsandtheconsumersrsquoperceptionthattheircontributionswouldbeuniqueandhelpfultoothercustomersaswellasperusedby

17HarteveldtHHJohnsonCStarkEandGeldernKV(2009)Us-ingDigitalChannelsToCalmTheAngryTravelerForresterResearch

theserviceproviders18Simplyputthepropensitytowritereviewsincreaseswhentheconsumersfeelthattheirreviewisactuallyhelpingothers(boththeserviceproviderandotherconsumers)

WesuggestthatreviewsitesdevelopbettermethodstoaggregatesynthesizeandpublishspecificreviewcommentsaswellasnumericalratingsFirstmanagersshouldtakeintoconsiderationthelowcorrelationbetweentheoverallratingsandconsumersrsquoassessmentsofvariouspropertyattributes(aswellastheresultsfromthetextanalysis)Theimplicationofsuchaweakcorrelationisthatreviewsitesneedtoexpandonthelistofpossiblevariablessothattheratingsaccuratelyreflectconsumersrsquoneedsandexpectationsregardingserviceproviders

Furthersitemanagersmayhavetorethinkthecurrentpracticeofaveragingalltheratingsforagivenpropertybe-causesuchameanratingfailstotakeintoaccountthebiasesthatareinherentintheratingsystemsAsanexamplethisstudyfindssignificantunder-reportingbiasinthisonlinereviewsiteAcustomerrsquosdecisiontonotpostonlinereviewsmaymeanthelossofimportantinformationthatcanassisttheusersofthesitestomakemorereliableinferencesThemajorityoftodayrsquosfeedbackmechanismsdonotpubliclydisclosethenumberofsilenttransactions(ietransactionsforwhichnofeedbackwaspostedbycustomers)Suchinfor-mationshouldbecomeapartofaserviceproviderrsquosonlineprofileonratingsitesandotherfeedbackmechanismsItisimportanttoprovidemoreinformationandheuristicstohelptheconsumersnavigatethroughtheclutterandgettheinformationtheywantandseek

Itisimportanttoapplymodernsemanticanalysisandsentimentclassificationtechniquestostudythecontentofthereviewsandidentifyvariousdimensionsonwhichcon-sumersevaluatevacationhomesThedetailsprovidedinthereviewtextareinmanywaysabetterreflectionofcustomersatisfactionServiceproviderscanbetterpositionthemselvesusingthedimensionsuncoveredthroughtextanalysisandtargettheneedsandpreferencesoftheircustomers

TheresultsinthisstudyshowthatnegativereviewsaretypicallyassociatedwithagreaterwordcountthanpositivereviewsItisusefultofocusspecificallyonthesenegativereviewsbecausetheycanrevealpatternsofdeficienciesintheservicestandardsanddeliveryThesereviewsshouldbetreatedthesameascustomercommentcardsorletters19Furthercustomerswhowritesuchlengthynegativereviews

18LingKimberlyGerardBeenenPaulLudfordXiaoxingWangKevinChangXinLiDanCosleyDanFrankowskiLorenTerveenAlMRashidandRobertKrautldquoUsingSocialPsychologytoMotivateContributionstoOnlineCommunitiesrdquoJournal of Computer-Mediated Communication10no4(2005)19MilanR(2007)10thingsyoucandoinresponsetotravelerreviewsinHotelmarketingcom

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 15

aremostlikelytospreadnegativeword-of-mouthaboutthepropertyLengthyreviewsareexcellentsourcesofinforma-tionandshouldbeusedasthebasestounderstandthecomplainingbehavioroftheconsumer

WordcountmaybeusedasanadditionalsegmentationvariabletoidentifythoseconsumerswhoaremoreinvolvedandwillingtoprovidebalancedfeedbacktoboththereviewsiteaswellastopropertymanagementVariousotherresearchmethodssuchastelephoneinterviewsorfocusgroupsshouldbeusedtoelicitdetailedfeedbackfromsuchconsumersTheseactionsalsohelpinservicerecoverypost-purchaseengagementandrelationshipbuilding

OutlookWeforeseethreepotentialareasinvestigationthatcanaddvaluetocurrentbestpracticesWestilldonrsquotknowmuchaboutconsumersrsquomotivationtowriteonlinereviews20Recentstudiesidentifyvariousantecedentsofreviewwritingbutsufferfromsamplingbiassincetheysurveyaself-select-edsampleofconsumerswhoalreadyprovidedreviewsAstheresultsinthisstudyindicateunder-reportingbias(lackofmoderateandlowreviews)remainsprevalentinonlinereviewsitesFutureresearchshouldidentifysocio-psycho-logicalfactorsthatincreaseconsumersrsquopropensitytowriteonlinereviews

Futureresearchshouldexploretheimpactofthepres-ence(orlack)ofextremelynegativereviewswithinalargecorpusofpositivereviewsTheevidenceontheimpactofre-

20WangYandFesenmaierDR(2003)AssessingMotivationofCon-tributioninOnlineCommunitiesAnEmpiricalInvestigationofanOn-lineTravelCommunityElectronic Markets1333-45andYooKyungHandUlrikeGretzelldquoWhatMotivatesConsumerstoWriteOnlineTravelReviewsrdquoInformation Technology amp Tourism10No4(2008)283-95

viewvalence(negativeorpositive)isequivocalProponentsofconfirmatorybiassuggestthatconsumerslookforaffir-mativeevidencesupportingaproductchoicealreadymadeIfthatisthecasepositivereviewsaremorelikelytohaveagreatereffectonconsumeractionsOntheotherhandthenotionofnegativitybiassuggeststhatwhenconsumersareneutralnegativereviewstendtobecomemoresalientthanpositivereviews21Inthepresenceoflargenumbersofposi-tivereviewsconsumersspecificallyseeknegativereviewsthattheyfeelwillhelpthemidentifyspecificproblemswithaserviceUnderstandinghowconsumersreconciletheinformationprovidedbybothnegativeandpositivereviewshasimplicationsforthedesignandmanagementofreviewsites

Finallyareviewrsquoswordcounthasimportantimplica-tionsforconsumersrsquotrustinthereviewsWhenconsumersarewillingtoreadandcompareopen-endedcommentsfromotherconsumerstheamountofinformationinareviewcanmatterWOMforhighinvolvementproductsismorepersuasiveandincreasesthedecisionmakerrsquosconfi-dencewhenthemessagesenderprovidesgreaternumberofreasonsfortheiroverallratingLongerreviewsprovidemoreinformationandarelikelytobeperceivedasmorehelpfulandpersuasivethanshorterreviewsThewaysinwhichareviewiswrittenframedandpresentedisaninterestingfactorwhichhasrarelybeeninvestigatedandhasimportantimplicationsparticularlygivenourfindingthattheamountofinformationinareview(wordcount)variessignificantlywiththeratingsofthereviewn

21BaSandPavlouPA(2002)EvidenceoftheEffectofTrustBuildingTechnologyinElectronicMarketsPricePremiumsandBuyerBehaviorMIS Quarterly26243-268

Cornell Hospitality Quarterlyhttpcqxsagepubcom

2011 ReportsVol11No15DesigningaSelf-HealingServiceSystemAnIntegrativeModelbyRobertCFordandMichaelCSturman

Vol11No14ReversingtheGreenBacklashWhyLargeHospitalityCompaniesShouldWelcomeCrediblyGreenCompetitorsbyMichaelGiebelhausenPhDandHaeEunHelenChunPhD

Vol11No13DevelopingaSustainabilityMeasurementFrameworkforHotelsTowardanIndustry-wideReportingStructurebyEricRicaurte

Vol11No12CreatingValueforWomenBusinessTravelersFocusingonEmotionalOutcomesbyJudiBrownellPhD

Vol11No11CustomerLoyaltyANewLookattheBenefitsofImprovingSegmentationEffortswithRewardsProgramsbyClayVoorheesPhDMichaelMcCallPhDandRogerCalantonePhD

Vol11No10CustomerPerceptionsofElectronicFoodOrderingbySherylEKimes

Vol11No92011TravelIndustryBenchmarkingStatusofSeniorDestinationandLodgingMarketingExecutivesbyRohitVermaPhDandKenMcGill

Vol11No8SearchOTAsandOnlineBookingAnExpandedAnalysisoftheBillboardEffectbyChrisAndersonPhD

Vol11No7OnlineMobileandTextFoodOrderingintheUSRestaurantIndustrybySherylEKimesPhDandPhilippFLaqueacute

Vol11No6HotelGuestsrsquoReactionstoGuestRoomSustainabilityInitiativesbyAlexSusskindPhDandRohitVermaPhD

Vol11No5TheImpactofTerrorismandEconomicShocksonUSHotelsbyCathyAEnzRenaacutetaKosovaacuteandMarkLomannoVol11No4ImplementingHumanResourceInnovationsThreeSuccessStoriesfromtheServiceIndustrybyJustinSunandKateWalshPhD

Vol11No3Compendium2011

Vol11No2PositioningaPlaceDevelopingaCompellingDestinationBrandbyRobertJKwortnikPhDandEthanHawkesMBA

Vol11No1TheImpactofHealthInsuranceonEmployeeJobAnxietyWithdrawalBehaviorsandTaskPerformancebySeanWayPhDBillCarrollPhDAlexSusskindPhDandJoeCYLeng

2011 Hospitality ToolsVol2No1TheGameHasChangedANewParadigmforStakeholderEngagementbyMaryBethMcEuen

2011 Industry PerspectivesNo7MegaTips2TwentyTestedTechniquesforIncreasingYourTipsbyMichaelLynn

2011 ProceedingsVol3No5BuildingBrandsintheInternetAgeAnalyticsLoyaltyandCommunicationbyGlennWithiam

Vol3No4BraveNewWorldOnlineHotelDistributionbyGlennWithiam

Vol3No3SocialMediaandtheHospitalityIndustryHoldingtheTigerbytheTailbyGlennWithiam

Vol3No2TheChallengeofHotelandRestaurantSustainabilityFindingProfitinldquoBeingGreenrdquobyGlennWithiam

Vol3No1CautiousOptimismCHRSExaminesHospitalityIndustryTrendsbyGlennWithiam

2010 ReportsVol10No18HowTravelersUseOnlineandSocialMediaChannelstoMakeHotel-choiceDecisionsbyLauraMcCarthyDebraStockandRohitVermaPhD

Vol10No17PublicorPrivateTheHospitalityInvestmentDecisionbyQingzhongMaPhDandAthenaWeiZhangPhD

Vol10No16BestPracticesinSearchEngineMarketingandOptimizationTheCaseoftheStJamesHotelbyGregBodenlcosVictorBogertDanGordonCarterHearneandChrisKAndersonPhD

Vol10No15TheImpactofPrix Fixe MenuPriceFormatsonGuestsrsquoDealPerceptionbyShuoWangandMichaelLynnPhD

Cornell Center for Hospitality Research

Publication Indexwwwchrcornelledu

The Executive Path Hospitality Leadership Through Learning

Complete program information and applications available online

wwwhotelschoolcornelleduexecedPhone + 1 607 255 4919 Email exec_ed_hotelcornelledu

Professionals from around the world are invited to attend 3-day 10-day or online courses at the worldrsquos leading institute for hospitality management education in

Visit our website to apply

Explore develop and apply ideas with global hospitality leaders and

expert Cornell professors

Success

AdvancingBusiness

andPersonal

bull Strategic Leadershipbull Financebull Foodservicebull Human Resources

bull Marketingbull Operationsbull Real Estate

No2RetaliationWhyanIncreaseinClaimsDoesNotMeantheSkyIsFallingbyDavidSherwynJDandGreggGilmanJD

2009 ToolsToolNo12MeasuringtheDiningExperienceTheCaseofVitaNovabyKeshPrasadandFredJDeMiccoPhD

Cornell Center for Hospitality ResearchPublication Index

The Executive Path Hospitality Leadership Through Learning

Complete program information and applications available online

wwwhotelschoolcornelleduexecedPhone + 1 607 255 4919 Email exec_ed_hotelcornelledu

Professionals from around the world are invited to attend 3-day 10-day or online courses at the worldrsquos leading institute for hospitality management education in

Visit our website to apply

Explore develop and apply ideas with global hospitality leaders and

expert Cornell professors

Success

AdvancingBusiness

andPersonal

bull Strategic Leadershipbull Financebull Foodservicebull Human Resources

bull Marketingbull Operationsbull Real Estate

visit our website to apply

wwwchrcornell edu

Page 3: Unscrambling the Puzzling Matter of Online Consumer ... · Philips Hospitality Americas Chris Proulx, CEO, eCornell & Executive Education Carolyn D. Richmond, Partner, Hospitality

FriendsBerkshire Healthcare bull Center for Advanced Retail Technology bull Cruise Industry News bull DK Shifflet amp Associates bull ehoteliercom bull EyeforTravel bull 4Hotelierscom bull Gerencia de Hoteles amp Restaurantes bull Global Hospitality Resources bull Hospitality Financial and Technological Professionals bull hospitalityInsidecom bull hospitalitynetorg bull Hospitality Technology Magazine bull HotelExecutivecom bull International CHRIE bull International Hotel Conference bull International Society of Hospitality Consultants bull iPerceptions bull JDA Software Group Inc bull JD Power and Associates bull The Lodging Conference bull Lodging Hospitality bull Lodging Magazine bull LRA Worldwide Inc bull Milestone Internet Marketing bull MindFolio bull Mindshare Technologies bull PhoCusWright Inc bull PKF Hospitality Research bull Resort and Recreation Magazine bull The Resort Trades bull RestaurantEdgecom bull Shibata Publishing Co bull Synovate bull The TravelCom Network bull Travel + Hospitality Group bull UniFocus bull USA Today bull WageWatch Inc bull The Wall Street Journal bull WIWIHCOM bull Wyndham Worldwide

Thank you to our generous Corporate Members

PartnersDavis amp Gilbert LLP Deloitte amp Touche USA LLPDenihan Hospitality GroupeCornell amp Executive EducationExpedia Inc Forbes Travel GuideFour Seasons Hotels and Resorts Fox Rothschild LLP French Quarter Holdings Inc HVS Hyatt InterContinental Hotels Group Jumeirah GroupLRP PublicationsMarriott International IncMarshrsquos Hospitality Practice MaritzpricelinecomPricewaterhouseCoopersProskauer Sabre Hospitality SolutionsSathguru Management Consultants (P) Ltd Schneider Electric Thayer Lodging Group Thompson HotelsTravelportWATG

Senior PartnersASAE FoundationCarlson HotelsHilton WorldwideMcDonaldrsquos USAPhilips HospitalitySASSTRTaj Hotels Resorts and Palaces

4 TheCenterforHospitalityResearchbullCornellUniversity

AbouT The AuThors

UnscramblingthePuzzlingMatterofOnlineConsumerRatings

An Exploratory AnalysisbyPradeepRacherlaDanielConnolly

andNatasaChristodoulidou

Pradeep racherla PhD is an assistant professor of marketing at the College of Business West Texas AampM University (pracherlawtamuedu ) He earned a PhD in business administration from the Fox School of

Business Temple University His research interests include consumer generated media healthcare marketing social networks knowledge management and collaborative systems He has written several research papers

and technical reports that have been published or forthcoming in prestigious journals such as Journal of Consumer Behavior Journal of Marketing amp Management Journal of Management Information Systems

Annals of Tourism Research and Cornell Hospitality Quarterly He has presented at numerous conferences both nationally and internationally

Daniel J Connolly PhD is an associate professor of information technology at the Daniels College of Business at the University of Denver with a dual appointment in the School of Hotel Restaurant and Tourism Management and in the Department of Information Technology and Electronic Commerce He also serves as director of undergraduate programs His teaching research and consulting interests focus on the strategic application of information technology and electronic commerce He has written or co-authored numerous publications including articles that have appeared in the Cornell Hotel and Restaurant Administration Quarterly the FIU Hospitality Review Information Technology in Hospitality and Journal of Hospitality and Leisure Marketing His book Technology Strategies for the Hospitality Industry (Pearson Prentice Hall 2005) is the industryrsquos first book on technology strategy

Natasa Christodoulidou PhD is a faculty member in the management and marketing department at California State University where she teaches internet marketing and other related marketing courses at the

undergraduate and MBA level Her research interests are in the areas of hospitality technology travel and tourism electronic commerce hospitality electronic distribution and electronic marketing She is a member of

the Decision Sciences Institute and the Western Decision Sciences Institute She has published more than 20 articles in the past five years in academic and professional journals She presents regularly at academic and

professional conferences around the world such as Eye for Travel FSTEC DSI and WDSI Among her speaking engagements this year are Las Vegas Hawaii France and Singapore

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 5

exeCuTive suMMAry

Thisstudyexploresthepatternsofonlinereviewsofvacationhomesfromacommunity-basedtraveladvisorywebsitewithagoalofunderstandingthebiasesinherentinonlinewordofmouth(WOM)relatedtotourismandhospitalityservicesAnanalysisofnearly3200reviewsfrom

ldquoReviewsitecomrdquo(apseudonym)whichpostsreviewsofvacationrentalpropertiesacrosstheUSAfindsanoverwhelmingpreponderanceof favorable reviewsMore to thepoint relatively few

ldquomoderaterdquoreviewsarepostedandthesecond-highestcategoryisextremelynegativecommentsUsingsemanticprocessingtechniquesontheaggregatereviewtextthestudyidentifiesthenuancedopinionsandconcernsofthetravelerswhowritereviewsNegativereviewstendtobelengthyandargumentativeoftendetailingdisappointmentoverexpectationsnotmetPositivereviewsontheotherhandtendtoberelativelybriefandconfirmtheoverallratingConsumerswhowroteldquohighrdquoreviewsplacedgreaterimportanceonvalueformoneycleanlinessandcomfortthandidthosewhowrotenegativereviewsThosewhowroteldquolowrdquoreviewsplacedtheiremphasisontheserviceprovidedbythepropertystaffandmanagementNegativereviewsweremorelikelytoinvolveahigherpriceaccommodationThisanalysisindicates that the overall numerical ratings typically used in review systems may not be the idealindicator of perceived service quality The results suggest that review sites should develop bettermethodstoaggregatesynthesizeandpublishthereviewcontentsparticularlythenumericalratingsThisandotherreviewsitesshowtheaverageofallthepoint-scaleratingsbutsuchsimplemeansdonottakeintoaccountthebiasesthatareinherentintheratingsystemsInsteadthesitesshouldprovidemore information and heuristics to help the consumers navigate through the clutter and get theinformationtheydesire

6 TheCenterforHospitalityResearchbullCornellUniversity

CorNell hosPiTAliTy rePorT

Giventhatconsumersareincreasinglyrelyingonsearchenginestoacquireinformationabout hotels restaurants and travel purveyors (among many other services)consumer-generated online reviews will inevitably change the structure andaccessibilityofinformationalongwithconsumersrsquoperceptionsofvariousservices

ThepopularityofonlinereviewshasledtoaproliferationofonlinereviewsitesincludingpopularsitessuchasHotelscomTripadvisorcomandYelpcomArecentstudybyForresterIncfoundthatmorethan80percentofwebshoppersreadandconsiderotherconsumersrsquoreviews1SimilarlyCompeteIncfoundthatnearly50percentofpurchasersvisitedamessageboardforumoronlinecommunityforinformationconnectedwiththeironlinetravelpurchasingandoneinthreeofthesebuyerssaidthatconsumer reviews helped with their purchase decision2 Almost half of these consumers said thatconsumersrsquoopinionsactuallycausedthemtochangetheirmindaboutwhattheypurchasedMoreoveramongthosebuyers25percentsaidtheyalsopostedareviewonaconsumerreviewsiteaftermakingtheirpurchase

1ForresterldquoFortyFactsaboutUSOnlineShoppersrdquoForresterInc20062CompeteldquoEmbracingConsumerBuzzCreatesMeasurementChallengesforMarketersrdquoCompeteInc2006

UnscramblingthePuzzlingMatterofOnlineConsumerRatings

An Exploratory Analysis

byPradeepRacherlaDanielConnollyandNatasaChristodoulidou

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 7

Howevertheexistingunderstandingofonlinetourism-relatedreviewsisrelativelyweakeventhoughweknowthatwordofmouth(WOM)ismoreimportantandinfluentialwithinaservicescontextthaninproductmarketingscenariosRecentlysomestudieshaveexploredtourism-andhospital-ity-relatedonlineWOM3Yetresearchhassofaroverlookedcertaininterestingaspectsofonlinereviewsasfollows

Onlinereviewstypicallyuseldquooverallstarratingsrdquoaspri-maryindicatorsofperceivedservicequalitymdashratingsthatarenotrelatedinanywaytotheformalstardesignationsawardedbyForbesorMichelinConsumersrsquomeanstarratingsaretypi-callyaggregatedfromtheratingsgivenbyindividualconsum-ersineachreviewWhatwedonrsquotknowishowconsumersdeterminethestarstheyassigntoaservicewhentheypostareviewonawebsiteThatisdostarratingsaccuratelyreflectconsumersrsquosentimentsregardingvariousattributesofserviceprovidersThisisavitalissuegiventheexperientialandin-tangiblenatureoftravelservicesandthefactthatconsumersusetheoverallratingasanimportantheuristicwhilenarrow-ingdownchoices

FurtherstudieshaveoverlookedtherelationshipbetweenthereviewsandvariousattributesofservicesandserviceprovidersSinceonlinereviewsareaformofWOMvariablessuchasconsumerinvolvementpricingandlengthofstaymayhaveanimpactonthefinalratingandpostingofthereviewprovidedbyconsumersItisimportanttouncovertheserelationshipstobetterunderstandthedynamicsofonlinereviews

FewstudieshaveanalyzedthequalitativeaspectofonlinereviewsOnlinereviewsareessentiallyopen-endedtext-basedconsumer-to-consumer(c2c)communicationsThetextcontentmaycontainnuancedviewsoftheservicesandserviceproviders(fromthereviewwritersrsquopointofview)thatcannotbeexpressedusingcrudenumericalratings4Hencethereisaneedtoanalyzethetextportionofthereviewsandidentifytheissuesthatconsumersaremostlyconcernedwithwhenwritingandpostingonlinerelatedreviews

WeaddresstheabovepointsinthisstudyThefollowingquestionsguidedus

3KHYooandUGretzelldquoWhatMotivatesConsumerstoWriteOnlineTravelReviewsrdquoInformation Technology amp TourismVol10(2008)pp283-295andZXiangandUGretzelldquoRoleofSocialMediainOnlineTravelInformationSearchrdquoTourism Management20094AGhoseandPGIpeirotisldquoEstimatingtheSocio-economicImpactofProductReviewsMiningTextandReviewerCharacteristicsrdquoInformation Systems Research2008

1 Dooverallratingsreflecttheperceivedqualityoftheserviceproviders(asderivedfromtheconsumersrsquoexpressedsentimentinthereviews)Whataretheunderlyingpatternsoftheoverallratingsinonlinereviews

2 Whatarethelinkagesamongthevariousattributesoftheserviceprovidersandcustomerreviewsand

3 WhataretheissuesthatconsumersmostlytalkaboutinthetextportionofonlinereviewsDotheseis-suesdifferfromtheratingsthattheconsumershaveprovidedviathestandardvariablesprovidedbythereviewsites

ResearchSettingandDataCollectionThisstudyisbasedonreviewscollectedfromldquoReviewsitecomrdquoaswehavecalledthecommunity-driventravelsitethatweanalyzedThesiteprovidesreviewsofverifiedrentalpropertiesandorganizesthereputationofvacationpropertiesThepropertylistingwithinthesiteisbasednotontherevenuesharingmodelthatisgenerallypracticedbyotherreviewsitesbutonthereputationtrustandfeedbackfromcustomerswhohavestayedatthevacationrentalpropertiesReviewsitecomisafreeservicethatoperatesonserviceproviderfeesandadvertisingrevenuesUn-likesometravelreviewsitesthefirmrsquosreviewsolicitationstrategyistosendpersonalizedinvitationsonlytothosecustomerswhosestayhasbeenverifiedbythepropertymanagementThesitersquosmanagementbelievesthatthisstrategytoalargeextenteliminatesfakereviews

Thefirmprovideduswith3300reviewsforanalysisAfterinitialdatacleanupabout100reviewswithincom-pleteorunusableinformationwereeliminatedand3197reviewswereusedfortheanalysisThereviewscontainedthefollowingattributesbull Theoverallratingassignedtotheserviceprovider

basedontheaggregationoftheratingsassignedbyindividualreviewers

bull Consumersrsquoratingsonthefollowingsixattributesvaluecleanlinesscomfortservicelocationandcheck-in

bull Thereviewtext(Semanticprocessingtechniqueswereappliedontheaggregatereviewtexttoidentifythenuancedopinionsandconcernsofthetravelerswhowrotereviews)

8 TheCenterforHospitalityResearchbullCornellUniversity

bull Agerangeoftherevieweronan1-4ordinalscale(25-3435-4445-54and55andabove)

bull Averagepriceperdayand

bull Wordcount(usedinthequantitativeanalysis)

PreliminaryHypothesesandResultsTheagegroupsofthereviewwritersareshowninExhibit1Alargepercentageofthecustomerswhoreviewedthepropertiesfallunder44yearsofageThisisconsistentwithpreviousstudiesinthisrealmwhichshowthatgenerationXandGenerationYaremoreactiveintheWOMforums

ReviewRatingDistribution Websitesthathostonlinereviewsgiveconsumerstheoptiontoassignaratingtotheserviceproviderstypicallyonascaleof1to5or1to7Manysitesaggregatethesenumeri-calorstarratingstodetermineanoverall(mean)starratingWeseethisoverallratingasanimportantelementinonlineWOMsinceconsumersoftenusethisratingtonarrowdowntheirconsiderationsetStudiesthathaveusedtheratingastheprimarypredictoroftheserviceprovidersrsquoperceivedqualityorcustomersatisfactiontodeterminetheimpactofonlinereviewsonproductsaleshavearrivedatmixedfind-ings5Howeverthesestudiesarebasedonthefundamentalassumptionthatthenumericalratingsassignedtoservice

5ForexampleseeJChevalierandDMayzlinldquoTheEffectofWordofMouthOnlineOnlineBookReviewsrdquoJournal of Marketing ResearchVol43(2006)pp345-354andWDuanBGuandABWhinstonldquoDoOnlineReviewsMattermdashAnEmpiricalInvestigationofPanelDatardquoDecision Support SystemsVol45(2008)pp1007-1016

providersareanaccurateestimateofthequalityperceivedbythecustomerThispracticefollowsthebasicassumptionofbehavioraltheoriststhatanydatasetwithalargenumberofconsumerresponsestendstofollowanormal(Gaussian)distributionRecentstudieshavequestionedthisassump-tionForinstanceDellaracosandNarayanrsquosstudyonmoviereviewsinYahoocomfoundthatonlytheextremelysatisfiedandextremelydissatisfiedcustomersarelikelytopubliclyexpresstheiropinionsascomparedtocustomerswithmod-erateopinions6SimilarlystudiesusingdatafromTripadvi-sorcomandAmazoncomfoundthatthestarratingsexhibitatruncateddistributioninwhichthemajorityofreviewsarepositive7

TotestthisassumptionweranasimplefrequencychartasshowninExhibit2Inkeepingwithotherstudiesthedistributionisheavilyskewedtowardsthepositiveratings1-pointreviews32(1)2-pointreviews95(3)3-pointreviews458(14)4-pointreviews688(22)and5-pointreviews1912(60)Totestfortherobustnessofthisdistri-butionweappliedtheKolmogorov-Smirnovtestofnormal-ityTheresultsconfirmthenon-normaldistributionofthestarratingsandtheheavyskewtowardspositivereviews

Threeexplanationsarepossibleforsuchadistribution

6CDellarocasandRNarayanldquoAStatisticalMeasureofaPopulationrsquosPropensitytoEngageinPost-purchaseOnlineWord-of-mouthrdquoStatisti-cal ScienceVol21(2006)pp2777ATalwarandRFBoiJurcaUnderstandingUserBehaviorinOnlineFeedbackReportingrdquoinElectronic CommerceSanDiegoCalifornia(2007)pp134-142

Exhibit 1

Distribution of age ranges

Exhibit 2

Distribution of overall ldquostar ratingsrdquo

55 and above 45ndash54

years

35ndash44 years

18ndash34 years

overall ldquostar ratingrdquo

Num

ber o

f rev

iew

s

3295

458 688

1912

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 9

1 AmajorityofthepropertiesthatarelistedonthewebsitehaveoutstandingperceivedqualityandprovidesatisfactoryservicestocustomersWhilethisistheo-reticallypossibleitseemsunlikelywhenappliedtoallthepropertieslistedonthewebsite8

2 ThereviewsdonotconstituteatrulyrepresentativesampleofconsumeropinionBasedonotherstud-iesthisexplanationseemsplausibleConsumerswithmoderateviewsmayseelittleincentivetoreporttheirviewsinanonlineforumandasaconsequenceonlythoseconsumerswhohavepolarizedviewsarelikelytoposttheircomments

3 ThedistributioncouldalsobeanartifactofthefirmrsquosreviewsolicitingstrategyRecentstudiessupportthisideabysuggestingthatcustomersengageinsociallydesirablebehavioreveninimpersonalinteractionswiththeserviceproviders9Inthiscaseitmaybethattheywillprovideareasonablypositiverevieweveniftheexperiencehasnotbeenuptothemark

StarRatingsandPerceivedServiceQualityLetrsquosreturntothefirstquestionweaskedattheoutsetArestarratingsatruerepresentationofserviceprovid-ersrsquoperceivedqualityasreflectedbytheconsumersrsquostatedsentimentsThisbecomesanimportantquestionfortwo

8AKadetldquoRah-RahRatingsOnlinerdquoSmartMoney20079YGreacutegoireandRJFisherldquoCustomerBetrayalandRetaliationWhenYourBestCustomersbecomeYourWorstEnemiesrdquoJournal of the Acad-emy of Marketing ScienceVol36(2008)pp247-261

reasons(a) ratingsaretheprimaryevidencethatconsum-ersconsiderwhileshoppingonlineforinformationand(b)thetruncatednatureofratingdistributionsmeansthatthereviewsavailablemaynottrulyrepresentconsumeropinionortheserviceprovidersrsquoperceivedqualityIfonlythosewithextremeviewspostreviewsthenconsumersshouldalwayslookforserviceprovidersthathavea5-starratingOnthecontraryifthereviewsaccuratelyreflectallconsumersrsquoviewsthenconsumersarebetteroffsearchingforreviewswitha3ratingsincethesetendtobemorebalancedandenablebetterproductanalysisRecentlyHu et aluseddatafromAmazoncomshowthatnumericalscoresdonotreflectbooksrsquotrueperceivedquality10AstheysuggestldquohellipratherthescorereflectsthebalanceofdiverseopinionsInotherwordswhenabookrsquosoverallscoreisaround3itdoesnotsuggestthatconsumersgenerallyagreethatthisisanaver-agebookItrathersuggeststhatroughlyequalnumberofconsumersthinkthatthebookiseitheranoutstandingbookoranabysmalbookrdquo

Totestthisassertionweappliedaregressionequationwithsixpropertyattributes(ievalueformoneycheck-inlocationcleanlinesscomfortandservice)asindependentvariablesandstarratingasthedependentvariableThisanalysisisbasedontheassumptionthatforagivenpropertyifstarratingstrulyrepresenttheperceivedqualityoftheproperties(aswellastheconsumersrsquosentiment)thenthestarratingsandratingsofthesixotherpropertyattributes

10NHuJZhangandPAPavlouOvercomingtheJ-shapedDistribu-tionofProductReviewsrdquoCommunications of the ACMVol52(2009)pp144-147

low reviews t-value high reviews t-value(Constant) 1618 18353 2400 34186vAlue 0254 13726 0378 8767CheCKiN -0083 -3701 -0053 -1192loCATioN 0104 6132 0079 2117CleAN 0144 7574 0112 2808CoMForT 0267 13993 0129 3088serviCe 0098 4209 0107 2401r2 0331 0338F-statistic 5038 24714

Exhibit 3

regression analysis of ldquolowrdquo and ldquohighrdquo reviews

Notes Coefficients (t-statistics reported) Significance is as follows significant at 0001 level significant at 05 level significant at 010 level Dependent variable is overall rating

10 TheCenterforHospitalityResearchbullCornellUniversity

moderateratingsonalltheattributesbutwasstillwillingtoprovideahighoverallstarratingtothepropertyThisexamplesuggeststhatconsumersplacesignificantemphasisontheoverallexperientialvalueprovidedbythepropertiesThisexpectationofvalueismanytimesintangibleanddoesnotnecessarilydependonthefunctioningorratingoftheindividualfeaturesoftheproperty

RelationshipbetweenRatingsandReviewAttributesTounderstandconsumersrsquoword-of-mouthbehavioritisimportanttoconsidervariousfactorssuchasperceivedexpectationsofcostsandbenefitsimportanceofthepur-chase(involvement)personalcharacteristicsandsituationalinfluencesWeconsidertwofactorsthatareknowntoaffecttheintensityandvalence(positiveornegative)ofwordofmouthpriceandwordcount

PricePriceisanimportantfactorthatcontributestopre-purchaseexpectationsTheintensityofexpectations(andtheirfulfillment)partlydrivenbypriceisdirectlypropor-tionaltoconsumersrsquosatisfactionlevelsandWOMpropensityThismaybethereasonthatthetendencytocomplainisdirectlyrelatedtothecostoftheservice11Themoreexpen-siveanitemthegreaterwillbetheperceivedlossesandthegreaterlikelihoodthatdissatisfactionwillresultinacom-plaintSimilarlythelikelihoodofword-of-mouthbeingtrig-geredbythelevelofexpectationndashdisconfirmationincreases

11BeardenWOandTeelJE(1983)Selecteddeterminantsofcon-sumersatisfactionandcomplaintreportsJournal of Marketing Research 2021-28

willshowahighcorrelationFortheanalysiswetookintoconsiderationthefactmostreviewsarepositive(ie4-and5-pointratings)Thereforewedividedthereviewsintotwocategorieshighreviews(starratingabove3n=2597)andlowreviews(ratingsbeloworequalto3n=600)andranseparateregressionanalysesonthesetwogroupsasshowninExhibit3

AninterestingfindingistheweakrelationshipbetweentheattributeratingsandoverallstarratingforeachpropertyTheR-squareoftheregressionequationwithallsixvariablesandtheoverallstarratingsislessthan40percentThatiseveniftheconsumersprovidedhigherratingsonindividualattributestheiroverallratingwasusuallylowandthere-versewasalsotrueThistrendisobservedinboththehighandlowreviewgroups

TheanalysissuggeststhatinhighreviewsconsumersplacedgreaterimportanceonvalueformoneycleanlinessandcomfortThehighertheratingsofaconsumerontheseattributesthehigheristhelikelihoodthattheconsumerwillprovideahigherstarratingCheck-inlocationandservicedidnotprominentlyfigureintheiroverallevalu-ationsofthepropertieseventhoughtheseattributeswereratedhigh

ThelowreviewsshowsimilarresultsinthesensethattheloweraconsumerratesvaluecleanlinessandcomfortthelowerthelikelihoodofapositiveratingInadditioninthecaseoflowreviewsconsumersplacedemphasisontheserviceprovidedbythepropertystaffandmanagement

TheaboveresultisillustratedinoneoftheldquohighrdquoreviewsshowninExhibit4Thereviewergavelowto

star rating Text Commentary

This was our 1st time renting a house vs a condo This house was great If you wanted to spend a little time by yourself you could so many rooms to go to The house was so comfortable like being at home It was our home for the week Thank you Hope to rent it next year if we havent bought one ourselves

ratings for specific Attributes

value Check-in location Clean Comfort service

3 2 3 2 4 3

Exhibit 4

example of a ldquohighrdquo review

Avg rating for Property Attributes

30

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 11

withthecostoftheproduct12Giventhegreaterperceivedriskinservicescustomerswouldseekmoretrustworthyser-viceproviderswithagoodreputationofsatisfyingcustomerneeds13ThisinturnincreasestheexpectationsandleadstomoreextremereactionsfromtheconsumersHenceonecanspeculatethatthehigherthepriceshigherwillbetheaver-agenegativeorpositiveratingsforaservice

Word countThenumberofwordsinareviewmayserveasaproxyfortheamountofinformationinthereviewsHowevertheevidenceonthisaspectofWOMofalltypesisequivocalForinstanceconsumersgenerallyengageinpositiveWOMtoavoidfeelingsofguiltandas-sociationwithbadnewsandtoreducecognitivedissonanceOntheotherhandacustomerwhoishighlydisappointedwillprovideapersuasiveargumenttomakesurethatothercustomersaredissuadedfromrentingthesamepropertyResearchonconventionalwordofmouthhasofferedampleevidencetosuggestthatdissatisfiedcustomersengageintwotothreetimesasmuchWOMassatisfiedcustomers14Webelievethatthislogicextendstoonlinereviewsdespitethefactthatexistingstudiesareequivocalatbest15Thereforeitcanbeassumedthatreviewswitheitherextremelypositiveorextremelynegativeratingswillhavegreaterwordcountasopposedtomoremoderatereviews

TotesttheseassumptionsweappliedANOVAwithpriceandwordcountasdependentvariablesandthereviewratingasfactorsAsillustratedinExhibits5and6theresultsshowthatpriceandwordcountshowasignificantnegativecorrelationwithstarratingThatisreviewswithlowerratingsaretypicallyassociatedwithgreaterwordcountandrepresentpropertiesthatchargehighrates

TheresultonthewordcountsuggeststhatconsumerswhoareunhappywitharentalpropertytendtowritemoretoexpresstheirdissatisfactionAsshowninthisexhibitlowerratingsareassociatedwithhighernumberofwordsinthereviewsandviceversaSimilarlypropertypriceseemstobestronglyassociatedwithnegativewordofmouthsup-portingtheprinciplethathighercostincreasestheconsum-ersrsquoinvolvementinthepurchase

12BeardenWOandMasonJB(1984)AninvestigationofinfluencesonconsumercomplaintreportsAdvancesinConsumerResearch11490-49513WetzerIMZeelenbergMandPietersR(2007)ldquoNevereatinthatrestaurantIdidrdquoExploringwhypeopleengageinnegativeword-of-mouthcommunicationPsychology and Marketing24661-68014SchlossbergH(1991)CustomersatisfactionNotafadbutawayoflifeMarketing News25andWestbrookR(1987)ProductConsump-tion-basedAffectiveResponsesandPost-purchaseProcessesJournal of Marketing Research 24258-27015AlexMSusskindldquoIToldYouSoRestaurantCustomersrsquoofWord-of-MouthCommunicationPatternsrdquoCornell Hotel and Restaurant Adminis-tration QuarterlyVol43No2(April2002)pp75-85

Exhibit 5

Differences in word count across star ratings

Exhibit 6

Differences in average price per day across ratings

Aver

age

pric

e pe

r day

of t

he re

ntal

pro

pert

y

overall ldquostar ratingrdquo

1504

1563

1804

19861987

Aver

age

wor

d co

unt o

f rev

iew

overall ldquostar ratingrdquo

836837

121

15751543

160

140

120

100

80

200

190

180

170

160

150

1 2 3 4 5

1 2 3 4 5

12 TheCenterforHospitalityResearchbullCornellUniversity

thereviewersThesecanbeclassifiedas(a)attributesorfea-turesofthepropertythatfigureprominentlyandcontributetothepositiveornegativeratings(egroomsbeddishesfurnitureorservice)and(b)thepositiveornegativewordsthatrepresentthesentimentofthereviewer(eggreatamazingdirtyandwonderful)Theresultsofthequalitativeanalysisareasfollows

Syntactic featuresTheoveralltoneofthelowreviewsisratherargumentativeThereviewerstendtobeforcefulinpresentingtheirpointofviewTheylayoutagreatamountofreasoningandprovidespecificexamplestodrivehometheirargumentThisisreflectedinthegreaterpercentageofnegativeconnectorwords(34)suchasbut however as ifandeven thoughItisinterestingtonotethatreviewersinthiscategoryspeakdirectlytotheotherconsumersnotablythroughgreateruseofwordssuchasyou(46)andlooktoinfluencetheirdecisionmaking

InthehighreviewsbycontrasttheoveralltonetendstobeenunciativeandthereviewersusethetexttomerelyreinforcetheirnumericalratingsusingfewerwordsonaveragethanthenegativereviewersInthesereviewstheconsumersfocusontheirpersonalexperiencesandappealtotheaudiencethroughemotionsratherthanlogicandreasoningasevidentfromgreateruseofpositiveadjectives

QualitativeAnalysisoftheReviewTextTheaboveanalysisemphasizestheimportanceofthetextcontentinconsumerreviewsWhilenumericalratingsallowthereviewerstoratetheserviceprovideronvariousattributesitisthetextthatprovidesthemwiththeopportu-nitytoarticulatethenuancesoftheiroverallexperienceandconveyusefulinformationaboutaserviceproviderrsquostransac-tionsandservicecapabilitiesThereforesemanticanalysisofthetextcontentprovidesuswithanopportunitytothethirdquestionpresentedattheoutsetofthisreportwhatdoconsumerstalkaboutintheirreviews

Tothisendweanalyzed600eachofrandomlyselectedhighandlowreviewsusingsemanticprocessingtechniquesWeusedcenteringresonanceanalysis(CRA)amodeofnet-work-basedtextanalysisthatrepresentsthecontentoflargesetsoftextsbyidentifyingthemostimportantwordsthatlinkotherwordsinthenetwork16OuranalysisexaminedsyntacticandsemanticfeaturesSyntacticfeaturesincludethepercentageofthewordsthatareopenclass(ienounsverbsadjectives)Semanticfeaturesrepresentthespecial-izedvocabularythatreflectstheoverallopinionexpressedby

16CormanSandDooleyK(2006)CrawdadTextAnalysisSystem12CrawdadTechnologiesLLCChandlerAZ

Prominently Mentioned Features Freq sample Comments

bedroom beds linen sheets 383

Also the same bedroom and bathroom did not have any door separating the bedroom from the bathroom no privacy

Sleeping arrangements were very awkward The listing sheet described 2 full beds We brought our own linens the beds were queen sized Had to purchase sheet

Kitchen Kitchenware Dishes Dishwasher 297

You must not forget to step up when yoursquore walking out of the kitchen into the dining room otherwise yoursquoll trip over the 3-foot step up

No potholders in the kitchen and it didnrsquot seem like there were as many kitchen supplies as before We just had to make do without

beach beaches beachfront 296

The week was the best vacation we have ever had as a family The beach house was a class act

Disappointment 1 It looks like this house is very close to beach but to get there you have to go around through public path

Pool Poolside 218

The pool was too dirty to use We checked in on Thursday and the pool was not cleaned until Saturday

Also it is inconvenient to have to call to get someone to open the pool Didnrsquot understand why Jacuzzi was open and not the pool

bathroom(s) bath bathtub shower 202

The next problem was in another bathroom (main floor)mdashthe wall heater did not work so showering in here was a bit chilly The next problem was in the ldquomasterrdquo bathroom (connected to the bedroom with the King bed) The sliding shower doors need to be totally replaced

Bathtub drains clogged in three second floor bathrooms

Exhibit 7

Top five prominent issues in low reviews

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 13

suchasgreat fantastic andawesome(45)andpersonalpronounssuchasweandI(54)

Semantic featuresExhibits7and8showthemainissuesoraspectsthatareprominentlydiscussedinthelowreviewsTheorderoftheseissuesisbasedonthestatisti-callysignificantfrequencyoftheiroccurrenceinthe1200reviews

WefurtherinvestigatedtheuseofvarioussentimentwordsInterestinglytheprominentissuesdiscussedinthepositivereviewssometimesdonotevenfigureintheratingscalesprovidedtothereviewersOveralltheresultsofthesemanticanalysiscanbesummarizedasfollows

Thepropertyanditsfeaturesaremorecommonlycon-nectedwithextremelypositivesentimentwordssuchasgreat wonderfulandamazinginthehighreviewsthaninthelowreviewsIncomparisonthepropertiesandtheirfea-turesinthelowreviewsaremorelikelytobecharacterizedwithmoderatelypositivewordssuchasniceandokay

Negativereviewsreflecttheextentofexpectationndashdis-confirmationthattheconsumersfaceduringtheirstayintherentalpropertyThesereviewshaveagreaterproportionofwordsexpressingdisappointment(egnotasexpected)Furtherextremelynegativesentimentwords(egdiscom-fortfilthy)andadjectivesaremorelikelytobeassociatedwithfeaturesinlowreviewsthaninhighreviews

Almost40percentoftheconsumerswhowrotehighreviewsexpresstheirwillingnesstoreturntothepropertyasopposedtoonly9percentofthosewritingnegativereviewsFurthermorethan20percentoftheconsumersgivinghighreviewsexpresstheirstrongdesiretorecommendtheprop-ertytotheirfriendsandfamilymembersasopposedtoonly5percentinlowreviews

ReviewofFindingsOurexploratorystudyofconsumerreviewsofUSvacationhomesonReviewsitecomwashighlightedbyoverwhelm-inglypositivereviews

Theoveralldistributionofthereviewswasheav-ilyskewedtowardsthepositiveratingsMorecriticallythisfindingsuggeststhatthereisalackofmoderateandbalancedopinionsWetheorizethatthisiseitherduetopurchasingbias(sincecustomerswentthroughthetroubleofpurchasingandexperiencingthefacilitiestheytendtobepositivelybiasedtowardstheirpurchase)orunder-reportingbias(customersdonothavestrongenoughincentivestotakethetroubleofreportingtheiropinionsandthefewmotivatedindividualswhoprovidetheirreviewsbecomeaself-selectedsample)

Theoverall(mean)starratingtypicallyusedintravelerreviewsitesisprobablynotthemostaccurateindicatorofcustomersrsquoperceptionsofserviceprovidersrsquoqualityWe

Prominently Mentioned Features Freq sample Commentsbeach beaches beachfront 506 Perfect rental property for those who enjoy the activity of the beach and the serenity

of the Sound The house is just down the road from ocean

The large porch offered a beautiful view of the beach and comfortable new rocking chairs

Pool Poolside 218 The outdoor shower dipping pool and screen in porch were an added bonus Perfect house for a low key weekend

House was great for kids Pool was terrific and the playroom was perfectlocation 192 We had a wonderful time My wife loves fishing and the cottage was in a perfect

location to go fishing on the pier

The home is somewhat dated but the location was great We have vacationed in the Chatham area for many summers and this time it was great too

bedroom beds 186 The family suite is huge with a king size and twin bed and still room for a pack-n-play The kitchen and dining room were huge clean and nice

The only comments that I would say to the negative is the front bed room by the front door always seemed damp which seem to attract more bugs

Kitchen Kitchenware Dishes Dishwasher 178 All of the amenities in the house are upscalemdashloved the kitchen and plasma TVs

The kitchen had good basic utensils dishes and cooking pans

Exhibit 7

Top five prominent issues in high reviews

14 TheCenterforHospitalityResearchbullCornellUniversity

foundacorrelationbetweenthestarratingandratingsonotherpropertyattributesoflessthan40percentinmostcases

AhighwordcountinthereviewsiscorrelatedwithalowoverallratingThatisconsumerswhoareextremelydissatisfiedtendtoexpendmoretimeandenergyontheircritiqueofthepropertyitsamenitiesandservicesPriceofthepropertyalsoseemstopositivelycorrelatewithextremenegativereviewsperhapsbecauseahighpriceheightenstravelersrsquoexpectations

Numericalratingsdonotoftencapturetheactualsentimentandthevarietyofdimensionsonwhichtravel-ersevaluatepropertiesBecausenumericalratingsdonotcapturereviewersrsquoopinionsaboutspecificattributestheycanbedeceivingandmaynotrevealthetruequalityofthepropertyortheactualintentionofthereviewerWefoundforinstancethatconsumersvaluelocation(egclosenesstothebeach)andcleanliness(egswimmingpoolareas)morethansuchfeaturesascheck-inandserviceSimilarlyinthenegativereviewsissuessuchaslackofbedroomparapher-naliaorkitchenitemsseemtoberankedhigherinimpor-tancethanissuessuchaslocationandcheck-inThereforeevenwhenapropertyscoreshighonthecertaindimensionstheoverallratingmayremainlowbecauseotherfactorsovershadowthefavorableaspects

WhatItMeansforManagersWeseeimplicationsforonlinereviewsitesandbyexten-sionforproprietorsofvacationhomesandserviceopera-tionsgenerallyWithregardtothereviewsitesifconsumersconsistentlydetectafavorablebiasthereviewsitesmaylosethecriticalaspectsofcredibilityandtrustAlthoughconsumersaregenerallysmartandrationalandcanworkaroundobviousbiasesrecentevidencesuggeststhatcon-sumersaregraduallydefectingfromsomeonlinesitesduetounbalancedreviews17InthisregardsiteadministratorsaswellaspropertymanagersshouldtakestepstoattractthoseconsumerswhoarewillingtoprovideareasonablybalancedanalysisoftheamenitiesandservicesBythiswemeanreviewsthatpresentcomprehensiveandobjectiveanalysisoftheserviceproviderratherthanjustravesorrants

Tothatendsiteoperatorsshouldtakestepstoun-derstandcustomersrsquoreviewwritingbehaviorandtoat-tractmorebalancedreviewsOnewaytodothisistogivepersonalizedinformationtotheconsumerdemonstratingthecontributionmadebyhisorherreviewandthevalueofthereviewtostakeholdersForinstanceLinget alfoundapositivecorrelationbetweenwillingnesstowritereviewsandtheconsumersrsquoperceptionthattheircontributionswouldbeuniqueandhelpfultoothercustomersaswellasperusedby

17HarteveldtHHJohnsonCStarkEandGeldernKV(2009)Us-ingDigitalChannelsToCalmTheAngryTravelerForresterResearch

theserviceproviders18Simplyputthepropensitytowritereviewsincreaseswhentheconsumersfeelthattheirreviewisactuallyhelpingothers(boththeserviceproviderandotherconsumers)

WesuggestthatreviewsitesdevelopbettermethodstoaggregatesynthesizeandpublishspecificreviewcommentsaswellasnumericalratingsFirstmanagersshouldtakeintoconsiderationthelowcorrelationbetweentheoverallratingsandconsumersrsquoassessmentsofvariouspropertyattributes(aswellastheresultsfromthetextanalysis)Theimplicationofsuchaweakcorrelationisthatreviewsitesneedtoexpandonthelistofpossiblevariablessothattheratingsaccuratelyreflectconsumersrsquoneedsandexpectationsregardingserviceproviders

Furthersitemanagersmayhavetorethinkthecurrentpracticeofaveragingalltheratingsforagivenpropertybe-causesuchameanratingfailstotakeintoaccountthebiasesthatareinherentintheratingsystemsAsanexamplethisstudyfindssignificantunder-reportingbiasinthisonlinereviewsiteAcustomerrsquosdecisiontonotpostonlinereviewsmaymeanthelossofimportantinformationthatcanassisttheusersofthesitestomakemorereliableinferencesThemajorityoftodayrsquosfeedbackmechanismsdonotpubliclydisclosethenumberofsilenttransactions(ietransactionsforwhichnofeedbackwaspostedbycustomers)Suchinfor-mationshouldbecomeapartofaserviceproviderrsquosonlineprofileonratingsitesandotherfeedbackmechanismsItisimportanttoprovidemoreinformationandheuristicstohelptheconsumersnavigatethroughtheclutterandgettheinformationtheywantandseek

Itisimportanttoapplymodernsemanticanalysisandsentimentclassificationtechniquestostudythecontentofthereviewsandidentifyvariousdimensionsonwhichcon-sumersevaluatevacationhomesThedetailsprovidedinthereviewtextareinmanywaysabetterreflectionofcustomersatisfactionServiceproviderscanbetterpositionthemselvesusingthedimensionsuncoveredthroughtextanalysisandtargettheneedsandpreferencesoftheircustomers

TheresultsinthisstudyshowthatnegativereviewsaretypicallyassociatedwithagreaterwordcountthanpositivereviewsItisusefultofocusspecificallyonthesenegativereviewsbecausetheycanrevealpatternsofdeficienciesintheservicestandardsanddeliveryThesereviewsshouldbetreatedthesameascustomercommentcardsorletters19Furthercustomerswhowritesuchlengthynegativereviews

18LingKimberlyGerardBeenenPaulLudfordXiaoxingWangKevinChangXinLiDanCosleyDanFrankowskiLorenTerveenAlMRashidandRobertKrautldquoUsingSocialPsychologytoMotivateContributionstoOnlineCommunitiesrdquoJournal of Computer-Mediated Communication10no4(2005)19MilanR(2007)10thingsyoucandoinresponsetotravelerreviewsinHotelmarketingcom

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 15

aremostlikelytospreadnegativeword-of-mouthaboutthepropertyLengthyreviewsareexcellentsourcesofinforma-tionandshouldbeusedasthebasestounderstandthecomplainingbehavioroftheconsumer

WordcountmaybeusedasanadditionalsegmentationvariabletoidentifythoseconsumerswhoaremoreinvolvedandwillingtoprovidebalancedfeedbacktoboththereviewsiteaswellastopropertymanagementVariousotherresearchmethodssuchastelephoneinterviewsorfocusgroupsshouldbeusedtoelicitdetailedfeedbackfromsuchconsumersTheseactionsalsohelpinservicerecoverypost-purchaseengagementandrelationshipbuilding

OutlookWeforeseethreepotentialareasinvestigationthatcanaddvaluetocurrentbestpracticesWestilldonrsquotknowmuchaboutconsumersrsquomotivationtowriteonlinereviews20Recentstudiesidentifyvariousantecedentsofreviewwritingbutsufferfromsamplingbiassincetheysurveyaself-select-edsampleofconsumerswhoalreadyprovidedreviewsAstheresultsinthisstudyindicateunder-reportingbias(lackofmoderateandlowreviews)remainsprevalentinonlinereviewsitesFutureresearchshouldidentifysocio-psycho-logicalfactorsthatincreaseconsumersrsquopropensitytowriteonlinereviews

Futureresearchshouldexploretheimpactofthepres-ence(orlack)ofextremelynegativereviewswithinalargecorpusofpositivereviewsTheevidenceontheimpactofre-

20WangYandFesenmaierDR(2003)AssessingMotivationofCon-tributioninOnlineCommunitiesAnEmpiricalInvestigationofanOn-lineTravelCommunityElectronic Markets1333-45andYooKyungHandUlrikeGretzelldquoWhatMotivatesConsumerstoWriteOnlineTravelReviewsrdquoInformation Technology amp Tourism10No4(2008)283-95

viewvalence(negativeorpositive)isequivocalProponentsofconfirmatorybiassuggestthatconsumerslookforaffir-mativeevidencesupportingaproductchoicealreadymadeIfthatisthecasepositivereviewsaremorelikelytohaveagreatereffectonconsumeractionsOntheotherhandthenotionofnegativitybiassuggeststhatwhenconsumersareneutralnegativereviewstendtobecomemoresalientthanpositivereviews21Inthepresenceoflargenumbersofposi-tivereviewsconsumersspecificallyseeknegativereviewsthattheyfeelwillhelpthemidentifyspecificproblemswithaserviceUnderstandinghowconsumersreconciletheinformationprovidedbybothnegativeandpositivereviewshasimplicationsforthedesignandmanagementofreviewsites

Finallyareviewrsquoswordcounthasimportantimplica-tionsforconsumersrsquotrustinthereviewsWhenconsumersarewillingtoreadandcompareopen-endedcommentsfromotherconsumerstheamountofinformationinareviewcanmatterWOMforhighinvolvementproductsismorepersuasiveandincreasesthedecisionmakerrsquosconfi-dencewhenthemessagesenderprovidesgreaternumberofreasonsfortheiroverallratingLongerreviewsprovidemoreinformationandarelikelytobeperceivedasmorehelpfulandpersuasivethanshorterreviewsThewaysinwhichareviewiswrittenframedandpresentedisaninterestingfactorwhichhasrarelybeeninvestigatedandhasimportantimplicationsparticularlygivenourfindingthattheamountofinformationinareview(wordcount)variessignificantlywiththeratingsofthereviewn

21BaSandPavlouPA(2002)EvidenceoftheEffectofTrustBuildingTechnologyinElectronicMarketsPricePremiumsandBuyerBehaviorMIS Quarterly26243-268

Cornell Hospitality Quarterlyhttpcqxsagepubcom

2011 ReportsVol11No15DesigningaSelf-HealingServiceSystemAnIntegrativeModelbyRobertCFordandMichaelCSturman

Vol11No14ReversingtheGreenBacklashWhyLargeHospitalityCompaniesShouldWelcomeCrediblyGreenCompetitorsbyMichaelGiebelhausenPhDandHaeEunHelenChunPhD

Vol11No13DevelopingaSustainabilityMeasurementFrameworkforHotelsTowardanIndustry-wideReportingStructurebyEricRicaurte

Vol11No12CreatingValueforWomenBusinessTravelersFocusingonEmotionalOutcomesbyJudiBrownellPhD

Vol11No11CustomerLoyaltyANewLookattheBenefitsofImprovingSegmentationEffortswithRewardsProgramsbyClayVoorheesPhDMichaelMcCallPhDandRogerCalantonePhD

Vol11No10CustomerPerceptionsofElectronicFoodOrderingbySherylEKimes

Vol11No92011TravelIndustryBenchmarkingStatusofSeniorDestinationandLodgingMarketingExecutivesbyRohitVermaPhDandKenMcGill

Vol11No8SearchOTAsandOnlineBookingAnExpandedAnalysisoftheBillboardEffectbyChrisAndersonPhD

Vol11No7OnlineMobileandTextFoodOrderingintheUSRestaurantIndustrybySherylEKimesPhDandPhilippFLaqueacute

Vol11No6HotelGuestsrsquoReactionstoGuestRoomSustainabilityInitiativesbyAlexSusskindPhDandRohitVermaPhD

Vol11No5TheImpactofTerrorismandEconomicShocksonUSHotelsbyCathyAEnzRenaacutetaKosovaacuteandMarkLomannoVol11No4ImplementingHumanResourceInnovationsThreeSuccessStoriesfromtheServiceIndustrybyJustinSunandKateWalshPhD

Vol11No3Compendium2011

Vol11No2PositioningaPlaceDevelopingaCompellingDestinationBrandbyRobertJKwortnikPhDandEthanHawkesMBA

Vol11No1TheImpactofHealthInsuranceonEmployeeJobAnxietyWithdrawalBehaviorsandTaskPerformancebySeanWayPhDBillCarrollPhDAlexSusskindPhDandJoeCYLeng

2011 Hospitality ToolsVol2No1TheGameHasChangedANewParadigmforStakeholderEngagementbyMaryBethMcEuen

2011 Industry PerspectivesNo7MegaTips2TwentyTestedTechniquesforIncreasingYourTipsbyMichaelLynn

2011 ProceedingsVol3No5BuildingBrandsintheInternetAgeAnalyticsLoyaltyandCommunicationbyGlennWithiam

Vol3No4BraveNewWorldOnlineHotelDistributionbyGlennWithiam

Vol3No3SocialMediaandtheHospitalityIndustryHoldingtheTigerbytheTailbyGlennWithiam

Vol3No2TheChallengeofHotelandRestaurantSustainabilityFindingProfitinldquoBeingGreenrdquobyGlennWithiam

Vol3No1CautiousOptimismCHRSExaminesHospitalityIndustryTrendsbyGlennWithiam

2010 ReportsVol10No18HowTravelersUseOnlineandSocialMediaChannelstoMakeHotel-choiceDecisionsbyLauraMcCarthyDebraStockandRohitVermaPhD

Vol10No17PublicorPrivateTheHospitalityInvestmentDecisionbyQingzhongMaPhDandAthenaWeiZhangPhD

Vol10No16BestPracticesinSearchEngineMarketingandOptimizationTheCaseoftheStJamesHotelbyGregBodenlcosVictorBogertDanGordonCarterHearneandChrisKAndersonPhD

Vol10No15TheImpactofPrix Fixe MenuPriceFormatsonGuestsrsquoDealPerceptionbyShuoWangandMichaelLynnPhD

Cornell Center for Hospitality Research

Publication Indexwwwchrcornelledu

The Executive Path Hospitality Leadership Through Learning

Complete program information and applications available online

wwwhotelschoolcornelleduexecedPhone + 1 607 255 4919 Email exec_ed_hotelcornelledu

Professionals from around the world are invited to attend 3-day 10-day or online courses at the worldrsquos leading institute for hospitality management education in

Visit our website to apply

Explore develop and apply ideas with global hospitality leaders and

expert Cornell professors

Success

AdvancingBusiness

andPersonal

bull Strategic Leadershipbull Financebull Foodservicebull Human Resources

bull Marketingbull Operationsbull Real Estate

No2RetaliationWhyanIncreaseinClaimsDoesNotMeantheSkyIsFallingbyDavidSherwynJDandGreggGilmanJD

2009 ToolsToolNo12MeasuringtheDiningExperienceTheCaseofVitaNovabyKeshPrasadandFredJDeMiccoPhD

Cornell Center for Hospitality ResearchPublication Index

The Executive Path Hospitality Leadership Through Learning

Complete program information and applications available online

wwwhotelschoolcornelleduexecedPhone + 1 607 255 4919 Email exec_ed_hotelcornelledu

Professionals from around the world are invited to attend 3-day 10-day or online courses at the worldrsquos leading institute for hospitality management education in

Visit our website to apply

Explore develop and apply ideas with global hospitality leaders and

expert Cornell professors

Success

AdvancingBusiness

andPersonal

bull Strategic Leadershipbull Financebull Foodservicebull Human Resources

bull Marketingbull Operationsbull Real Estate

visit our website to apply

wwwchrcornell edu

Page 4: Unscrambling the Puzzling Matter of Online Consumer ... · Philips Hospitality Americas Chris Proulx, CEO, eCornell & Executive Education Carolyn D. Richmond, Partner, Hospitality

4 TheCenterforHospitalityResearchbullCornellUniversity

AbouT The AuThors

UnscramblingthePuzzlingMatterofOnlineConsumerRatings

An Exploratory AnalysisbyPradeepRacherlaDanielConnolly

andNatasaChristodoulidou

Pradeep racherla PhD is an assistant professor of marketing at the College of Business West Texas AampM University (pracherlawtamuedu ) He earned a PhD in business administration from the Fox School of

Business Temple University His research interests include consumer generated media healthcare marketing social networks knowledge management and collaborative systems He has written several research papers

and technical reports that have been published or forthcoming in prestigious journals such as Journal of Consumer Behavior Journal of Marketing amp Management Journal of Management Information Systems

Annals of Tourism Research and Cornell Hospitality Quarterly He has presented at numerous conferences both nationally and internationally

Daniel J Connolly PhD is an associate professor of information technology at the Daniels College of Business at the University of Denver with a dual appointment in the School of Hotel Restaurant and Tourism Management and in the Department of Information Technology and Electronic Commerce He also serves as director of undergraduate programs His teaching research and consulting interests focus on the strategic application of information technology and electronic commerce He has written or co-authored numerous publications including articles that have appeared in the Cornell Hotel and Restaurant Administration Quarterly the FIU Hospitality Review Information Technology in Hospitality and Journal of Hospitality and Leisure Marketing His book Technology Strategies for the Hospitality Industry (Pearson Prentice Hall 2005) is the industryrsquos first book on technology strategy

Natasa Christodoulidou PhD is a faculty member in the management and marketing department at California State University where she teaches internet marketing and other related marketing courses at the

undergraduate and MBA level Her research interests are in the areas of hospitality technology travel and tourism electronic commerce hospitality electronic distribution and electronic marketing She is a member of

the Decision Sciences Institute and the Western Decision Sciences Institute She has published more than 20 articles in the past five years in academic and professional journals She presents regularly at academic and

professional conferences around the world such as Eye for Travel FSTEC DSI and WDSI Among her speaking engagements this year are Las Vegas Hawaii France and Singapore

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 5

exeCuTive suMMAry

Thisstudyexploresthepatternsofonlinereviewsofvacationhomesfromacommunity-basedtraveladvisorywebsitewithagoalofunderstandingthebiasesinherentinonlinewordofmouth(WOM)relatedtotourismandhospitalityservicesAnanalysisofnearly3200reviewsfrom

ldquoReviewsitecomrdquo(apseudonym)whichpostsreviewsofvacationrentalpropertiesacrosstheUSAfindsanoverwhelmingpreponderanceof favorable reviewsMore to thepoint relatively few

ldquomoderaterdquoreviewsarepostedandthesecond-highestcategoryisextremelynegativecommentsUsingsemanticprocessingtechniquesontheaggregatereviewtextthestudyidentifiesthenuancedopinionsandconcernsofthetravelerswhowritereviewsNegativereviewstendtobelengthyandargumentativeoftendetailingdisappointmentoverexpectationsnotmetPositivereviewsontheotherhandtendtoberelativelybriefandconfirmtheoverallratingConsumerswhowroteldquohighrdquoreviewsplacedgreaterimportanceonvalueformoneycleanlinessandcomfortthandidthosewhowrotenegativereviewsThosewhowroteldquolowrdquoreviewsplacedtheiremphasisontheserviceprovidedbythepropertystaffandmanagementNegativereviewsweremorelikelytoinvolveahigherpriceaccommodationThisanalysisindicates that the overall numerical ratings typically used in review systems may not be the idealindicator of perceived service quality The results suggest that review sites should develop bettermethodstoaggregatesynthesizeandpublishthereviewcontentsparticularlythenumericalratingsThisandotherreviewsitesshowtheaverageofallthepoint-scaleratingsbutsuchsimplemeansdonottakeintoaccountthebiasesthatareinherentintheratingsystemsInsteadthesitesshouldprovidemore information and heuristics to help the consumers navigate through the clutter and get theinformationtheydesire

6 TheCenterforHospitalityResearchbullCornellUniversity

CorNell hosPiTAliTy rePorT

Giventhatconsumersareincreasinglyrelyingonsearchenginestoacquireinformationabout hotels restaurants and travel purveyors (among many other services)consumer-generated online reviews will inevitably change the structure andaccessibilityofinformationalongwithconsumersrsquoperceptionsofvariousservices

ThepopularityofonlinereviewshasledtoaproliferationofonlinereviewsitesincludingpopularsitessuchasHotelscomTripadvisorcomandYelpcomArecentstudybyForresterIncfoundthatmorethan80percentofwebshoppersreadandconsiderotherconsumersrsquoreviews1SimilarlyCompeteIncfoundthatnearly50percentofpurchasersvisitedamessageboardforumoronlinecommunityforinformationconnectedwiththeironlinetravelpurchasingandoneinthreeofthesebuyerssaidthatconsumer reviews helped with their purchase decision2 Almost half of these consumers said thatconsumersrsquoopinionsactuallycausedthemtochangetheirmindaboutwhattheypurchasedMoreoveramongthosebuyers25percentsaidtheyalsopostedareviewonaconsumerreviewsiteaftermakingtheirpurchase

1ForresterldquoFortyFactsaboutUSOnlineShoppersrdquoForresterInc20062CompeteldquoEmbracingConsumerBuzzCreatesMeasurementChallengesforMarketersrdquoCompeteInc2006

UnscramblingthePuzzlingMatterofOnlineConsumerRatings

An Exploratory Analysis

byPradeepRacherlaDanielConnollyandNatasaChristodoulidou

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 7

Howevertheexistingunderstandingofonlinetourism-relatedreviewsisrelativelyweakeventhoughweknowthatwordofmouth(WOM)ismoreimportantandinfluentialwithinaservicescontextthaninproductmarketingscenariosRecentlysomestudieshaveexploredtourism-andhospital-ity-relatedonlineWOM3Yetresearchhassofaroverlookedcertaininterestingaspectsofonlinereviewsasfollows

Onlinereviewstypicallyuseldquooverallstarratingsrdquoaspri-maryindicatorsofperceivedservicequalitymdashratingsthatarenotrelatedinanywaytotheformalstardesignationsawardedbyForbesorMichelinConsumersrsquomeanstarratingsaretypi-callyaggregatedfromtheratingsgivenbyindividualconsum-ersineachreviewWhatwedonrsquotknowishowconsumersdeterminethestarstheyassigntoaservicewhentheypostareviewonawebsiteThatisdostarratingsaccuratelyreflectconsumersrsquosentimentsregardingvariousattributesofserviceprovidersThisisavitalissuegiventheexperientialandin-tangiblenatureoftravelservicesandthefactthatconsumersusetheoverallratingasanimportantheuristicwhilenarrow-ingdownchoices

FurtherstudieshaveoverlookedtherelationshipbetweenthereviewsandvariousattributesofservicesandserviceprovidersSinceonlinereviewsareaformofWOMvariablessuchasconsumerinvolvementpricingandlengthofstaymayhaveanimpactonthefinalratingandpostingofthereviewprovidedbyconsumersItisimportanttouncovertheserelationshipstobetterunderstandthedynamicsofonlinereviews

FewstudieshaveanalyzedthequalitativeaspectofonlinereviewsOnlinereviewsareessentiallyopen-endedtext-basedconsumer-to-consumer(c2c)communicationsThetextcontentmaycontainnuancedviewsoftheservicesandserviceproviders(fromthereviewwritersrsquopointofview)thatcannotbeexpressedusingcrudenumericalratings4Hencethereisaneedtoanalyzethetextportionofthereviewsandidentifytheissuesthatconsumersaremostlyconcernedwithwhenwritingandpostingonlinerelatedreviews

WeaddresstheabovepointsinthisstudyThefollowingquestionsguidedus

3KHYooandUGretzelldquoWhatMotivatesConsumerstoWriteOnlineTravelReviewsrdquoInformation Technology amp TourismVol10(2008)pp283-295andZXiangandUGretzelldquoRoleofSocialMediainOnlineTravelInformationSearchrdquoTourism Management20094AGhoseandPGIpeirotisldquoEstimatingtheSocio-economicImpactofProductReviewsMiningTextandReviewerCharacteristicsrdquoInformation Systems Research2008

1 Dooverallratingsreflecttheperceivedqualityoftheserviceproviders(asderivedfromtheconsumersrsquoexpressedsentimentinthereviews)Whataretheunderlyingpatternsoftheoverallratingsinonlinereviews

2 Whatarethelinkagesamongthevariousattributesoftheserviceprovidersandcustomerreviewsand

3 WhataretheissuesthatconsumersmostlytalkaboutinthetextportionofonlinereviewsDotheseis-suesdifferfromtheratingsthattheconsumershaveprovidedviathestandardvariablesprovidedbythereviewsites

ResearchSettingandDataCollectionThisstudyisbasedonreviewscollectedfromldquoReviewsitecomrdquoaswehavecalledthecommunity-driventravelsitethatweanalyzedThesiteprovidesreviewsofverifiedrentalpropertiesandorganizesthereputationofvacationpropertiesThepropertylistingwithinthesiteisbasednotontherevenuesharingmodelthatisgenerallypracticedbyotherreviewsitesbutonthereputationtrustandfeedbackfromcustomerswhohavestayedatthevacationrentalpropertiesReviewsitecomisafreeservicethatoperatesonserviceproviderfeesandadvertisingrevenuesUn-likesometravelreviewsitesthefirmrsquosreviewsolicitationstrategyistosendpersonalizedinvitationsonlytothosecustomerswhosestayhasbeenverifiedbythepropertymanagementThesitersquosmanagementbelievesthatthisstrategytoalargeextenteliminatesfakereviews

Thefirmprovideduswith3300reviewsforanalysisAfterinitialdatacleanupabout100reviewswithincom-pleteorunusableinformationwereeliminatedand3197reviewswereusedfortheanalysisThereviewscontainedthefollowingattributesbull Theoverallratingassignedtotheserviceprovider

basedontheaggregationoftheratingsassignedbyindividualreviewers

bull Consumersrsquoratingsonthefollowingsixattributesvaluecleanlinesscomfortservicelocationandcheck-in

bull Thereviewtext(Semanticprocessingtechniqueswereappliedontheaggregatereviewtexttoidentifythenuancedopinionsandconcernsofthetravelerswhowrotereviews)

8 TheCenterforHospitalityResearchbullCornellUniversity

bull Agerangeoftherevieweronan1-4ordinalscale(25-3435-4445-54and55andabove)

bull Averagepriceperdayand

bull Wordcount(usedinthequantitativeanalysis)

PreliminaryHypothesesandResultsTheagegroupsofthereviewwritersareshowninExhibit1Alargepercentageofthecustomerswhoreviewedthepropertiesfallunder44yearsofageThisisconsistentwithpreviousstudiesinthisrealmwhichshowthatgenerationXandGenerationYaremoreactiveintheWOMforums

ReviewRatingDistribution Websitesthathostonlinereviewsgiveconsumerstheoptiontoassignaratingtotheserviceproviderstypicallyonascaleof1to5or1to7Manysitesaggregatethesenumeri-calorstarratingstodetermineanoverall(mean)starratingWeseethisoverallratingasanimportantelementinonlineWOMsinceconsumersoftenusethisratingtonarrowdowntheirconsiderationsetStudiesthathaveusedtheratingastheprimarypredictoroftheserviceprovidersrsquoperceivedqualityorcustomersatisfactiontodeterminetheimpactofonlinereviewsonproductsaleshavearrivedatmixedfind-ings5Howeverthesestudiesarebasedonthefundamentalassumptionthatthenumericalratingsassignedtoservice

5ForexampleseeJChevalierandDMayzlinldquoTheEffectofWordofMouthOnlineOnlineBookReviewsrdquoJournal of Marketing ResearchVol43(2006)pp345-354andWDuanBGuandABWhinstonldquoDoOnlineReviewsMattermdashAnEmpiricalInvestigationofPanelDatardquoDecision Support SystemsVol45(2008)pp1007-1016

providersareanaccurateestimateofthequalityperceivedbythecustomerThispracticefollowsthebasicassumptionofbehavioraltheoriststhatanydatasetwithalargenumberofconsumerresponsestendstofollowanormal(Gaussian)distributionRecentstudieshavequestionedthisassump-tionForinstanceDellaracosandNarayanrsquosstudyonmoviereviewsinYahoocomfoundthatonlytheextremelysatisfiedandextremelydissatisfiedcustomersarelikelytopubliclyexpresstheiropinionsascomparedtocustomerswithmod-erateopinions6SimilarlystudiesusingdatafromTripadvi-sorcomandAmazoncomfoundthatthestarratingsexhibitatruncateddistributioninwhichthemajorityofreviewsarepositive7

TotestthisassumptionweranasimplefrequencychartasshowninExhibit2Inkeepingwithotherstudiesthedistributionisheavilyskewedtowardsthepositiveratings1-pointreviews32(1)2-pointreviews95(3)3-pointreviews458(14)4-pointreviews688(22)and5-pointreviews1912(60)Totestfortherobustnessofthisdistri-butionweappliedtheKolmogorov-Smirnovtestofnormal-ityTheresultsconfirmthenon-normaldistributionofthestarratingsandtheheavyskewtowardspositivereviews

Threeexplanationsarepossibleforsuchadistribution

6CDellarocasandRNarayanldquoAStatisticalMeasureofaPopulationrsquosPropensitytoEngageinPost-purchaseOnlineWord-of-mouthrdquoStatisti-cal ScienceVol21(2006)pp2777ATalwarandRFBoiJurcaUnderstandingUserBehaviorinOnlineFeedbackReportingrdquoinElectronic CommerceSanDiegoCalifornia(2007)pp134-142

Exhibit 1

Distribution of age ranges

Exhibit 2

Distribution of overall ldquostar ratingsrdquo

55 and above 45ndash54

years

35ndash44 years

18ndash34 years

overall ldquostar ratingrdquo

Num

ber o

f rev

iew

s

3295

458 688

1912

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 9

1 AmajorityofthepropertiesthatarelistedonthewebsitehaveoutstandingperceivedqualityandprovidesatisfactoryservicestocustomersWhilethisistheo-reticallypossibleitseemsunlikelywhenappliedtoallthepropertieslistedonthewebsite8

2 ThereviewsdonotconstituteatrulyrepresentativesampleofconsumeropinionBasedonotherstud-iesthisexplanationseemsplausibleConsumerswithmoderateviewsmayseelittleincentivetoreporttheirviewsinanonlineforumandasaconsequenceonlythoseconsumerswhohavepolarizedviewsarelikelytoposttheircomments

3 ThedistributioncouldalsobeanartifactofthefirmrsquosreviewsolicitingstrategyRecentstudiessupportthisideabysuggestingthatcustomersengageinsociallydesirablebehavioreveninimpersonalinteractionswiththeserviceproviders9Inthiscaseitmaybethattheywillprovideareasonablypositiverevieweveniftheexperiencehasnotbeenuptothemark

StarRatingsandPerceivedServiceQualityLetrsquosreturntothefirstquestionweaskedattheoutsetArestarratingsatruerepresentationofserviceprovid-ersrsquoperceivedqualityasreflectedbytheconsumersrsquostatedsentimentsThisbecomesanimportantquestionfortwo

8AKadetldquoRah-RahRatingsOnlinerdquoSmartMoney20079YGreacutegoireandRJFisherldquoCustomerBetrayalandRetaliationWhenYourBestCustomersbecomeYourWorstEnemiesrdquoJournal of the Acad-emy of Marketing ScienceVol36(2008)pp247-261

reasons(a) ratingsaretheprimaryevidencethatconsum-ersconsiderwhileshoppingonlineforinformationand(b)thetruncatednatureofratingdistributionsmeansthatthereviewsavailablemaynottrulyrepresentconsumeropinionortheserviceprovidersrsquoperceivedqualityIfonlythosewithextremeviewspostreviewsthenconsumersshouldalwayslookforserviceprovidersthathavea5-starratingOnthecontraryifthereviewsaccuratelyreflectallconsumersrsquoviewsthenconsumersarebetteroffsearchingforreviewswitha3ratingsincethesetendtobemorebalancedandenablebetterproductanalysisRecentlyHu et aluseddatafromAmazoncomshowthatnumericalscoresdonotreflectbooksrsquotrueperceivedquality10AstheysuggestldquohellipratherthescorereflectsthebalanceofdiverseopinionsInotherwordswhenabookrsquosoverallscoreisaround3itdoesnotsuggestthatconsumersgenerallyagreethatthisisanaver-agebookItrathersuggeststhatroughlyequalnumberofconsumersthinkthatthebookiseitheranoutstandingbookoranabysmalbookrdquo

Totestthisassertionweappliedaregressionequationwithsixpropertyattributes(ievalueformoneycheck-inlocationcleanlinesscomfortandservice)asindependentvariablesandstarratingasthedependentvariableThisanalysisisbasedontheassumptionthatforagivenpropertyifstarratingstrulyrepresenttheperceivedqualityoftheproperties(aswellastheconsumersrsquosentiment)thenthestarratingsandratingsofthesixotherpropertyattributes

10NHuJZhangandPAPavlouOvercomingtheJ-shapedDistribu-tionofProductReviewsrdquoCommunications of the ACMVol52(2009)pp144-147

low reviews t-value high reviews t-value(Constant) 1618 18353 2400 34186vAlue 0254 13726 0378 8767CheCKiN -0083 -3701 -0053 -1192loCATioN 0104 6132 0079 2117CleAN 0144 7574 0112 2808CoMForT 0267 13993 0129 3088serviCe 0098 4209 0107 2401r2 0331 0338F-statistic 5038 24714

Exhibit 3

regression analysis of ldquolowrdquo and ldquohighrdquo reviews

Notes Coefficients (t-statistics reported) Significance is as follows significant at 0001 level significant at 05 level significant at 010 level Dependent variable is overall rating

10 TheCenterforHospitalityResearchbullCornellUniversity

moderateratingsonalltheattributesbutwasstillwillingtoprovideahighoverallstarratingtothepropertyThisexamplesuggeststhatconsumersplacesignificantemphasisontheoverallexperientialvalueprovidedbythepropertiesThisexpectationofvalueismanytimesintangibleanddoesnotnecessarilydependonthefunctioningorratingoftheindividualfeaturesoftheproperty

RelationshipbetweenRatingsandReviewAttributesTounderstandconsumersrsquoword-of-mouthbehavioritisimportanttoconsidervariousfactorssuchasperceivedexpectationsofcostsandbenefitsimportanceofthepur-chase(involvement)personalcharacteristicsandsituationalinfluencesWeconsidertwofactorsthatareknowntoaffecttheintensityandvalence(positiveornegative)ofwordofmouthpriceandwordcount

PricePriceisanimportantfactorthatcontributestopre-purchaseexpectationsTheintensityofexpectations(andtheirfulfillment)partlydrivenbypriceisdirectlypropor-tionaltoconsumersrsquosatisfactionlevelsandWOMpropensityThismaybethereasonthatthetendencytocomplainisdirectlyrelatedtothecostoftheservice11Themoreexpen-siveanitemthegreaterwillbetheperceivedlossesandthegreaterlikelihoodthatdissatisfactionwillresultinacom-plaintSimilarlythelikelihoodofword-of-mouthbeingtrig-geredbythelevelofexpectationndashdisconfirmationincreases

11BeardenWOandTeelJE(1983)Selecteddeterminantsofcon-sumersatisfactionandcomplaintreportsJournal of Marketing Research 2021-28

willshowahighcorrelationFortheanalysiswetookintoconsiderationthefactmostreviewsarepositive(ie4-and5-pointratings)Thereforewedividedthereviewsintotwocategorieshighreviews(starratingabove3n=2597)andlowreviews(ratingsbeloworequalto3n=600)andranseparateregressionanalysesonthesetwogroupsasshowninExhibit3

AninterestingfindingistheweakrelationshipbetweentheattributeratingsandoverallstarratingforeachpropertyTheR-squareoftheregressionequationwithallsixvariablesandtheoverallstarratingsislessthan40percentThatiseveniftheconsumersprovidedhigherratingsonindividualattributestheiroverallratingwasusuallylowandthere-versewasalsotrueThistrendisobservedinboththehighandlowreviewgroups

TheanalysissuggeststhatinhighreviewsconsumersplacedgreaterimportanceonvalueformoneycleanlinessandcomfortThehighertheratingsofaconsumerontheseattributesthehigheristhelikelihoodthattheconsumerwillprovideahigherstarratingCheck-inlocationandservicedidnotprominentlyfigureintheiroverallevalu-ationsofthepropertieseventhoughtheseattributeswereratedhigh

ThelowreviewsshowsimilarresultsinthesensethattheloweraconsumerratesvaluecleanlinessandcomfortthelowerthelikelihoodofapositiveratingInadditioninthecaseoflowreviewsconsumersplacedemphasisontheserviceprovidedbythepropertystaffandmanagement

TheaboveresultisillustratedinoneoftheldquohighrdquoreviewsshowninExhibit4Thereviewergavelowto

star rating Text Commentary

This was our 1st time renting a house vs a condo This house was great If you wanted to spend a little time by yourself you could so many rooms to go to The house was so comfortable like being at home It was our home for the week Thank you Hope to rent it next year if we havent bought one ourselves

ratings for specific Attributes

value Check-in location Clean Comfort service

3 2 3 2 4 3

Exhibit 4

example of a ldquohighrdquo review

Avg rating for Property Attributes

30

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 11

withthecostoftheproduct12Giventhegreaterperceivedriskinservicescustomerswouldseekmoretrustworthyser-viceproviderswithagoodreputationofsatisfyingcustomerneeds13ThisinturnincreasestheexpectationsandleadstomoreextremereactionsfromtheconsumersHenceonecanspeculatethatthehigherthepriceshigherwillbetheaver-agenegativeorpositiveratingsforaservice

Word countThenumberofwordsinareviewmayserveasaproxyfortheamountofinformationinthereviewsHowevertheevidenceonthisaspectofWOMofalltypesisequivocalForinstanceconsumersgenerallyengageinpositiveWOMtoavoidfeelingsofguiltandas-sociationwithbadnewsandtoreducecognitivedissonanceOntheotherhandacustomerwhoishighlydisappointedwillprovideapersuasiveargumenttomakesurethatothercustomersaredissuadedfromrentingthesamepropertyResearchonconventionalwordofmouthhasofferedampleevidencetosuggestthatdissatisfiedcustomersengageintwotothreetimesasmuchWOMassatisfiedcustomers14Webelievethatthislogicextendstoonlinereviewsdespitethefactthatexistingstudiesareequivocalatbest15Thereforeitcanbeassumedthatreviewswitheitherextremelypositiveorextremelynegativeratingswillhavegreaterwordcountasopposedtomoremoderatereviews

TotesttheseassumptionsweappliedANOVAwithpriceandwordcountasdependentvariablesandthereviewratingasfactorsAsillustratedinExhibits5and6theresultsshowthatpriceandwordcountshowasignificantnegativecorrelationwithstarratingThatisreviewswithlowerratingsaretypicallyassociatedwithgreaterwordcountandrepresentpropertiesthatchargehighrates

TheresultonthewordcountsuggeststhatconsumerswhoareunhappywitharentalpropertytendtowritemoretoexpresstheirdissatisfactionAsshowninthisexhibitlowerratingsareassociatedwithhighernumberofwordsinthereviewsandviceversaSimilarlypropertypriceseemstobestronglyassociatedwithnegativewordofmouthsup-portingtheprinciplethathighercostincreasestheconsum-ersrsquoinvolvementinthepurchase

12BeardenWOandMasonJB(1984)AninvestigationofinfluencesonconsumercomplaintreportsAdvancesinConsumerResearch11490-49513WetzerIMZeelenbergMandPietersR(2007)ldquoNevereatinthatrestaurantIdidrdquoExploringwhypeopleengageinnegativeword-of-mouthcommunicationPsychology and Marketing24661-68014SchlossbergH(1991)CustomersatisfactionNotafadbutawayoflifeMarketing News25andWestbrookR(1987)ProductConsump-tion-basedAffectiveResponsesandPost-purchaseProcessesJournal of Marketing Research 24258-27015AlexMSusskindldquoIToldYouSoRestaurantCustomersrsquoofWord-of-MouthCommunicationPatternsrdquoCornell Hotel and Restaurant Adminis-tration QuarterlyVol43No2(April2002)pp75-85

Exhibit 5

Differences in word count across star ratings

Exhibit 6

Differences in average price per day across ratings

Aver

age

pric

e pe

r day

of t

he re

ntal

pro

pert

y

overall ldquostar ratingrdquo

1504

1563

1804

19861987

Aver

age

wor

d co

unt o

f rev

iew

overall ldquostar ratingrdquo

836837

121

15751543

160

140

120

100

80

200

190

180

170

160

150

1 2 3 4 5

1 2 3 4 5

12 TheCenterforHospitalityResearchbullCornellUniversity

thereviewersThesecanbeclassifiedas(a)attributesorfea-turesofthepropertythatfigureprominentlyandcontributetothepositiveornegativeratings(egroomsbeddishesfurnitureorservice)and(b)thepositiveornegativewordsthatrepresentthesentimentofthereviewer(eggreatamazingdirtyandwonderful)Theresultsofthequalitativeanalysisareasfollows

Syntactic featuresTheoveralltoneofthelowreviewsisratherargumentativeThereviewerstendtobeforcefulinpresentingtheirpointofviewTheylayoutagreatamountofreasoningandprovidespecificexamplestodrivehometheirargumentThisisreflectedinthegreaterpercentageofnegativeconnectorwords(34)suchasbut however as ifandeven thoughItisinterestingtonotethatreviewersinthiscategoryspeakdirectlytotheotherconsumersnotablythroughgreateruseofwordssuchasyou(46)andlooktoinfluencetheirdecisionmaking

InthehighreviewsbycontrasttheoveralltonetendstobeenunciativeandthereviewersusethetexttomerelyreinforcetheirnumericalratingsusingfewerwordsonaveragethanthenegativereviewersInthesereviewstheconsumersfocusontheirpersonalexperiencesandappealtotheaudiencethroughemotionsratherthanlogicandreasoningasevidentfromgreateruseofpositiveadjectives

QualitativeAnalysisoftheReviewTextTheaboveanalysisemphasizestheimportanceofthetextcontentinconsumerreviewsWhilenumericalratingsallowthereviewerstoratetheserviceprovideronvariousattributesitisthetextthatprovidesthemwiththeopportu-nitytoarticulatethenuancesoftheiroverallexperienceandconveyusefulinformationaboutaserviceproviderrsquostransac-tionsandservicecapabilitiesThereforesemanticanalysisofthetextcontentprovidesuswithanopportunitytothethirdquestionpresentedattheoutsetofthisreportwhatdoconsumerstalkaboutintheirreviews

Tothisendweanalyzed600eachofrandomlyselectedhighandlowreviewsusingsemanticprocessingtechniquesWeusedcenteringresonanceanalysis(CRA)amodeofnet-work-basedtextanalysisthatrepresentsthecontentoflargesetsoftextsbyidentifyingthemostimportantwordsthatlinkotherwordsinthenetwork16OuranalysisexaminedsyntacticandsemanticfeaturesSyntacticfeaturesincludethepercentageofthewordsthatareopenclass(ienounsverbsadjectives)Semanticfeaturesrepresentthespecial-izedvocabularythatreflectstheoverallopinionexpressedby

16CormanSandDooleyK(2006)CrawdadTextAnalysisSystem12CrawdadTechnologiesLLCChandlerAZ

Prominently Mentioned Features Freq sample Comments

bedroom beds linen sheets 383

Also the same bedroom and bathroom did not have any door separating the bedroom from the bathroom no privacy

Sleeping arrangements were very awkward The listing sheet described 2 full beds We brought our own linens the beds were queen sized Had to purchase sheet

Kitchen Kitchenware Dishes Dishwasher 297

You must not forget to step up when yoursquore walking out of the kitchen into the dining room otherwise yoursquoll trip over the 3-foot step up

No potholders in the kitchen and it didnrsquot seem like there were as many kitchen supplies as before We just had to make do without

beach beaches beachfront 296

The week was the best vacation we have ever had as a family The beach house was a class act

Disappointment 1 It looks like this house is very close to beach but to get there you have to go around through public path

Pool Poolside 218

The pool was too dirty to use We checked in on Thursday and the pool was not cleaned until Saturday

Also it is inconvenient to have to call to get someone to open the pool Didnrsquot understand why Jacuzzi was open and not the pool

bathroom(s) bath bathtub shower 202

The next problem was in another bathroom (main floor)mdashthe wall heater did not work so showering in here was a bit chilly The next problem was in the ldquomasterrdquo bathroom (connected to the bedroom with the King bed) The sliding shower doors need to be totally replaced

Bathtub drains clogged in three second floor bathrooms

Exhibit 7

Top five prominent issues in low reviews

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 13

suchasgreat fantastic andawesome(45)andpersonalpronounssuchasweandI(54)

Semantic featuresExhibits7and8showthemainissuesoraspectsthatareprominentlydiscussedinthelowreviewsTheorderoftheseissuesisbasedonthestatisti-callysignificantfrequencyoftheiroccurrenceinthe1200reviews

WefurtherinvestigatedtheuseofvarioussentimentwordsInterestinglytheprominentissuesdiscussedinthepositivereviewssometimesdonotevenfigureintheratingscalesprovidedtothereviewersOveralltheresultsofthesemanticanalysiscanbesummarizedasfollows

Thepropertyanditsfeaturesaremorecommonlycon-nectedwithextremelypositivesentimentwordssuchasgreat wonderfulandamazinginthehighreviewsthaninthelowreviewsIncomparisonthepropertiesandtheirfea-turesinthelowreviewsaremorelikelytobecharacterizedwithmoderatelypositivewordssuchasniceandokay

Negativereviewsreflecttheextentofexpectationndashdis-confirmationthattheconsumersfaceduringtheirstayintherentalpropertyThesereviewshaveagreaterproportionofwordsexpressingdisappointment(egnotasexpected)Furtherextremelynegativesentimentwords(egdiscom-fortfilthy)andadjectivesaremorelikelytobeassociatedwithfeaturesinlowreviewsthaninhighreviews

Almost40percentoftheconsumerswhowrotehighreviewsexpresstheirwillingnesstoreturntothepropertyasopposedtoonly9percentofthosewritingnegativereviewsFurthermorethan20percentoftheconsumersgivinghighreviewsexpresstheirstrongdesiretorecommendtheprop-ertytotheirfriendsandfamilymembersasopposedtoonly5percentinlowreviews

ReviewofFindingsOurexploratorystudyofconsumerreviewsofUSvacationhomesonReviewsitecomwashighlightedbyoverwhelm-inglypositivereviews

Theoveralldistributionofthereviewswasheav-ilyskewedtowardsthepositiveratingsMorecriticallythisfindingsuggeststhatthereisalackofmoderateandbalancedopinionsWetheorizethatthisiseitherduetopurchasingbias(sincecustomerswentthroughthetroubleofpurchasingandexperiencingthefacilitiestheytendtobepositivelybiasedtowardstheirpurchase)orunder-reportingbias(customersdonothavestrongenoughincentivestotakethetroubleofreportingtheiropinionsandthefewmotivatedindividualswhoprovidetheirreviewsbecomeaself-selectedsample)

Theoverall(mean)starratingtypicallyusedintravelerreviewsitesisprobablynotthemostaccurateindicatorofcustomersrsquoperceptionsofserviceprovidersrsquoqualityWe

Prominently Mentioned Features Freq sample Commentsbeach beaches beachfront 506 Perfect rental property for those who enjoy the activity of the beach and the serenity

of the Sound The house is just down the road from ocean

The large porch offered a beautiful view of the beach and comfortable new rocking chairs

Pool Poolside 218 The outdoor shower dipping pool and screen in porch were an added bonus Perfect house for a low key weekend

House was great for kids Pool was terrific and the playroom was perfectlocation 192 We had a wonderful time My wife loves fishing and the cottage was in a perfect

location to go fishing on the pier

The home is somewhat dated but the location was great We have vacationed in the Chatham area for many summers and this time it was great too

bedroom beds 186 The family suite is huge with a king size and twin bed and still room for a pack-n-play The kitchen and dining room were huge clean and nice

The only comments that I would say to the negative is the front bed room by the front door always seemed damp which seem to attract more bugs

Kitchen Kitchenware Dishes Dishwasher 178 All of the amenities in the house are upscalemdashloved the kitchen and plasma TVs

The kitchen had good basic utensils dishes and cooking pans

Exhibit 7

Top five prominent issues in high reviews

14 TheCenterforHospitalityResearchbullCornellUniversity

foundacorrelationbetweenthestarratingandratingsonotherpropertyattributesoflessthan40percentinmostcases

AhighwordcountinthereviewsiscorrelatedwithalowoverallratingThatisconsumerswhoareextremelydissatisfiedtendtoexpendmoretimeandenergyontheircritiqueofthepropertyitsamenitiesandservicesPriceofthepropertyalsoseemstopositivelycorrelatewithextremenegativereviewsperhapsbecauseahighpriceheightenstravelersrsquoexpectations

Numericalratingsdonotoftencapturetheactualsentimentandthevarietyofdimensionsonwhichtravel-ersevaluatepropertiesBecausenumericalratingsdonotcapturereviewersrsquoopinionsaboutspecificattributestheycanbedeceivingandmaynotrevealthetruequalityofthepropertyortheactualintentionofthereviewerWefoundforinstancethatconsumersvaluelocation(egclosenesstothebeach)andcleanliness(egswimmingpoolareas)morethansuchfeaturesascheck-inandserviceSimilarlyinthenegativereviewsissuessuchaslackofbedroomparapher-naliaorkitchenitemsseemtoberankedhigherinimpor-tancethanissuessuchaslocationandcheck-inThereforeevenwhenapropertyscoreshighonthecertaindimensionstheoverallratingmayremainlowbecauseotherfactorsovershadowthefavorableaspects

WhatItMeansforManagersWeseeimplicationsforonlinereviewsitesandbyexten-sionforproprietorsofvacationhomesandserviceopera-tionsgenerallyWithregardtothereviewsitesifconsumersconsistentlydetectafavorablebiasthereviewsitesmaylosethecriticalaspectsofcredibilityandtrustAlthoughconsumersaregenerallysmartandrationalandcanworkaroundobviousbiasesrecentevidencesuggeststhatcon-sumersaregraduallydefectingfromsomeonlinesitesduetounbalancedreviews17InthisregardsiteadministratorsaswellaspropertymanagersshouldtakestepstoattractthoseconsumerswhoarewillingtoprovideareasonablybalancedanalysisoftheamenitiesandservicesBythiswemeanreviewsthatpresentcomprehensiveandobjectiveanalysisoftheserviceproviderratherthanjustravesorrants

Tothatendsiteoperatorsshouldtakestepstoun-derstandcustomersrsquoreviewwritingbehaviorandtoat-tractmorebalancedreviewsOnewaytodothisistogivepersonalizedinformationtotheconsumerdemonstratingthecontributionmadebyhisorherreviewandthevalueofthereviewtostakeholdersForinstanceLinget alfoundapositivecorrelationbetweenwillingnesstowritereviewsandtheconsumersrsquoperceptionthattheircontributionswouldbeuniqueandhelpfultoothercustomersaswellasperusedby

17HarteveldtHHJohnsonCStarkEandGeldernKV(2009)Us-ingDigitalChannelsToCalmTheAngryTravelerForresterResearch

theserviceproviders18Simplyputthepropensitytowritereviewsincreaseswhentheconsumersfeelthattheirreviewisactuallyhelpingothers(boththeserviceproviderandotherconsumers)

WesuggestthatreviewsitesdevelopbettermethodstoaggregatesynthesizeandpublishspecificreviewcommentsaswellasnumericalratingsFirstmanagersshouldtakeintoconsiderationthelowcorrelationbetweentheoverallratingsandconsumersrsquoassessmentsofvariouspropertyattributes(aswellastheresultsfromthetextanalysis)Theimplicationofsuchaweakcorrelationisthatreviewsitesneedtoexpandonthelistofpossiblevariablessothattheratingsaccuratelyreflectconsumersrsquoneedsandexpectationsregardingserviceproviders

Furthersitemanagersmayhavetorethinkthecurrentpracticeofaveragingalltheratingsforagivenpropertybe-causesuchameanratingfailstotakeintoaccountthebiasesthatareinherentintheratingsystemsAsanexamplethisstudyfindssignificantunder-reportingbiasinthisonlinereviewsiteAcustomerrsquosdecisiontonotpostonlinereviewsmaymeanthelossofimportantinformationthatcanassisttheusersofthesitestomakemorereliableinferencesThemajorityoftodayrsquosfeedbackmechanismsdonotpubliclydisclosethenumberofsilenttransactions(ietransactionsforwhichnofeedbackwaspostedbycustomers)Suchinfor-mationshouldbecomeapartofaserviceproviderrsquosonlineprofileonratingsitesandotherfeedbackmechanismsItisimportanttoprovidemoreinformationandheuristicstohelptheconsumersnavigatethroughtheclutterandgettheinformationtheywantandseek

Itisimportanttoapplymodernsemanticanalysisandsentimentclassificationtechniquestostudythecontentofthereviewsandidentifyvariousdimensionsonwhichcon-sumersevaluatevacationhomesThedetailsprovidedinthereviewtextareinmanywaysabetterreflectionofcustomersatisfactionServiceproviderscanbetterpositionthemselvesusingthedimensionsuncoveredthroughtextanalysisandtargettheneedsandpreferencesoftheircustomers

TheresultsinthisstudyshowthatnegativereviewsaretypicallyassociatedwithagreaterwordcountthanpositivereviewsItisusefultofocusspecificallyonthesenegativereviewsbecausetheycanrevealpatternsofdeficienciesintheservicestandardsanddeliveryThesereviewsshouldbetreatedthesameascustomercommentcardsorletters19Furthercustomerswhowritesuchlengthynegativereviews

18LingKimberlyGerardBeenenPaulLudfordXiaoxingWangKevinChangXinLiDanCosleyDanFrankowskiLorenTerveenAlMRashidandRobertKrautldquoUsingSocialPsychologytoMotivateContributionstoOnlineCommunitiesrdquoJournal of Computer-Mediated Communication10no4(2005)19MilanR(2007)10thingsyoucandoinresponsetotravelerreviewsinHotelmarketingcom

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 15

aremostlikelytospreadnegativeword-of-mouthaboutthepropertyLengthyreviewsareexcellentsourcesofinforma-tionandshouldbeusedasthebasestounderstandthecomplainingbehavioroftheconsumer

WordcountmaybeusedasanadditionalsegmentationvariabletoidentifythoseconsumerswhoaremoreinvolvedandwillingtoprovidebalancedfeedbacktoboththereviewsiteaswellastopropertymanagementVariousotherresearchmethodssuchastelephoneinterviewsorfocusgroupsshouldbeusedtoelicitdetailedfeedbackfromsuchconsumersTheseactionsalsohelpinservicerecoverypost-purchaseengagementandrelationshipbuilding

OutlookWeforeseethreepotentialareasinvestigationthatcanaddvaluetocurrentbestpracticesWestilldonrsquotknowmuchaboutconsumersrsquomotivationtowriteonlinereviews20Recentstudiesidentifyvariousantecedentsofreviewwritingbutsufferfromsamplingbiassincetheysurveyaself-select-edsampleofconsumerswhoalreadyprovidedreviewsAstheresultsinthisstudyindicateunder-reportingbias(lackofmoderateandlowreviews)remainsprevalentinonlinereviewsitesFutureresearchshouldidentifysocio-psycho-logicalfactorsthatincreaseconsumersrsquopropensitytowriteonlinereviews

Futureresearchshouldexploretheimpactofthepres-ence(orlack)ofextremelynegativereviewswithinalargecorpusofpositivereviewsTheevidenceontheimpactofre-

20WangYandFesenmaierDR(2003)AssessingMotivationofCon-tributioninOnlineCommunitiesAnEmpiricalInvestigationofanOn-lineTravelCommunityElectronic Markets1333-45andYooKyungHandUlrikeGretzelldquoWhatMotivatesConsumerstoWriteOnlineTravelReviewsrdquoInformation Technology amp Tourism10No4(2008)283-95

viewvalence(negativeorpositive)isequivocalProponentsofconfirmatorybiassuggestthatconsumerslookforaffir-mativeevidencesupportingaproductchoicealreadymadeIfthatisthecasepositivereviewsaremorelikelytohaveagreatereffectonconsumeractionsOntheotherhandthenotionofnegativitybiassuggeststhatwhenconsumersareneutralnegativereviewstendtobecomemoresalientthanpositivereviews21Inthepresenceoflargenumbersofposi-tivereviewsconsumersspecificallyseeknegativereviewsthattheyfeelwillhelpthemidentifyspecificproblemswithaserviceUnderstandinghowconsumersreconciletheinformationprovidedbybothnegativeandpositivereviewshasimplicationsforthedesignandmanagementofreviewsites

Finallyareviewrsquoswordcounthasimportantimplica-tionsforconsumersrsquotrustinthereviewsWhenconsumersarewillingtoreadandcompareopen-endedcommentsfromotherconsumerstheamountofinformationinareviewcanmatterWOMforhighinvolvementproductsismorepersuasiveandincreasesthedecisionmakerrsquosconfi-dencewhenthemessagesenderprovidesgreaternumberofreasonsfortheiroverallratingLongerreviewsprovidemoreinformationandarelikelytobeperceivedasmorehelpfulandpersuasivethanshorterreviewsThewaysinwhichareviewiswrittenframedandpresentedisaninterestingfactorwhichhasrarelybeeninvestigatedandhasimportantimplicationsparticularlygivenourfindingthattheamountofinformationinareview(wordcount)variessignificantlywiththeratingsofthereviewn

21BaSandPavlouPA(2002)EvidenceoftheEffectofTrustBuildingTechnologyinElectronicMarketsPricePremiumsandBuyerBehaviorMIS Quarterly26243-268

Cornell Hospitality Quarterlyhttpcqxsagepubcom

2011 ReportsVol11No15DesigningaSelf-HealingServiceSystemAnIntegrativeModelbyRobertCFordandMichaelCSturman

Vol11No14ReversingtheGreenBacklashWhyLargeHospitalityCompaniesShouldWelcomeCrediblyGreenCompetitorsbyMichaelGiebelhausenPhDandHaeEunHelenChunPhD

Vol11No13DevelopingaSustainabilityMeasurementFrameworkforHotelsTowardanIndustry-wideReportingStructurebyEricRicaurte

Vol11No12CreatingValueforWomenBusinessTravelersFocusingonEmotionalOutcomesbyJudiBrownellPhD

Vol11No11CustomerLoyaltyANewLookattheBenefitsofImprovingSegmentationEffortswithRewardsProgramsbyClayVoorheesPhDMichaelMcCallPhDandRogerCalantonePhD

Vol11No10CustomerPerceptionsofElectronicFoodOrderingbySherylEKimes

Vol11No92011TravelIndustryBenchmarkingStatusofSeniorDestinationandLodgingMarketingExecutivesbyRohitVermaPhDandKenMcGill

Vol11No8SearchOTAsandOnlineBookingAnExpandedAnalysisoftheBillboardEffectbyChrisAndersonPhD

Vol11No7OnlineMobileandTextFoodOrderingintheUSRestaurantIndustrybySherylEKimesPhDandPhilippFLaqueacute

Vol11No6HotelGuestsrsquoReactionstoGuestRoomSustainabilityInitiativesbyAlexSusskindPhDandRohitVermaPhD

Vol11No5TheImpactofTerrorismandEconomicShocksonUSHotelsbyCathyAEnzRenaacutetaKosovaacuteandMarkLomannoVol11No4ImplementingHumanResourceInnovationsThreeSuccessStoriesfromtheServiceIndustrybyJustinSunandKateWalshPhD

Vol11No3Compendium2011

Vol11No2PositioningaPlaceDevelopingaCompellingDestinationBrandbyRobertJKwortnikPhDandEthanHawkesMBA

Vol11No1TheImpactofHealthInsuranceonEmployeeJobAnxietyWithdrawalBehaviorsandTaskPerformancebySeanWayPhDBillCarrollPhDAlexSusskindPhDandJoeCYLeng

2011 Hospitality ToolsVol2No1TheGameHasChangedANewParadigmforStakeholderEngagementbyMaryBethMcEuen

2011 Industry PerspectivesNo7MegaTips2TwentyTestedTechniquesforIncreasingYourTipsbyMichaelLynn

2011 ProceedingsVol3No5BuildingBrandsintheInternetAgeAnalyticsLoyaltyandCommunicationbyGlennWithiam

Vol3No4BraveNewWorldOnlineHotelDistributionbyGlennWithiam

Vol3No3SocialMediaandtheHospitalityIndustryHoldingtheTigerbytheTailbyGlennWithiam

Vol3No2TheChallengeofHotelandRestaurantSustainabilityFindingProfitinldquoBeingGreenrdquobyGlennWithiam

Vol3No1CautiousOptimismCHRSExaminesHospitalityIndustryTrendsbyGlennWithiam

2010 ReportsVol10No18HowTravelersUseOnlineandSocialMediaChannelstoMakeHotel-choiceDecisionsbyLauraMcCarthyDebraStockandRohitVermaPhD

Vol10No17PublicorPrivateTheHospitalityInvestmentDecisionbyQingzhongMaPhDandAthenaWeiZhangPhD

Vol10No16BestPracticesinSearchEngineMarketingandOptimizationTheCaseoftheStJamesHotelbyGregBodenlcosVictorBogertDanGordonCarterHearneandChrisKAndersonPhD

Vol10No15TheImpactofPrix Fixe MenuPriceFormatsonGuestsrsquoDealPerceptionbyShuoWangandMichaelLynnPhD

Cornell Center for Hospitality Research

Publication Indexwwwchrcornelledu

The Executive Path Hospitality Leadership Through Learning

Complete program information and applications available online

wwwhotelschoolcornelleduexecedPhone + 1 607 255 4919 Email exec_ed_hotelcornelledu

Professionals from around the world are invited to attend 3-day 10-day or online courses at the worldrsquos leading institute for hospitality management education in

Visit our website to apply

Explore develop and apply ideas with global hospitality leaders and

expert Cornell professors

Success

AdvancingBusiness

andPersonal

bull Strategic Leadershipbull Financebull Foodservicebull Human Resources

bull Marketingbull Operationsbull Real Estate

No2RetaliationWhyanIncreaseinClaimsDoesNotMeantheSkyIsFallingbyDavidSherwynJDandGreggGilmanJD

2009 ToolsToolNo12MeasuringtheDiningExperienceTheCaseofVitaNovabyKeshPrasadandFredJDeMiccoPhD

Cornell Center for Hospitality ResearchPublication Index

The Executive Path Hospitality Leadership Through Learning

Complete program information and applications available online

wwwhotelschoolcornelleduexecedPhone + 1 607 255 4919 Email exec_ed_hotelcornelledu

Professionals from around the world are invited to attend 3-day 10-day or online courses at the worldrsquos leading institute for hospitality management education in

Visit our website to apply

Explore develop and apply ideas with global hospitality leaders and

expert Cornell professors

Success

AdvancingBusiness

andPersonal

bull Strategic Leadershipbull Financebull Foodservicebull Human Resources

bull Marketingbull Operationsbull Real Estate

visit our website to apply

wwwchrcornell edu

Page 5: Unscrambling the Puzzling Matter of Online Consumer ... · Philips Hospitality Americas Chris Proulx, CEO, eCornell & Executive Education Carolyn D. Richmond, Partner, Hospitality

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 5

exeCuTive suMMAry

Thisstudyexploresthepatternsofonlinereviewsofvacationhomesfromacommunity-basedtraveladvisorywebsitewithagoalofunderstandingthebiasesinherentinonlinewordofmouth(WOM)relatedtotourismandhospitalityservicesAnanalysisofnearly3200reviewsfrom

ldquoReviewsitecomrdquo(apseudonym)whichpostsreviewsofvacationrentalpropertiesacrosstheUSAfindsanoverwhelmingpreponderanceof favorable reviewsMore to thepoint relatively few

ldquomoderaterdquoreviewsarepostedandthesecond-highestcategoryisextremelynegativecommentsUsingsemanticprocessingtechniquesontheaggregatereviewtextthestudyidentifiesthenuancedopinionsandconcernsofthetravelerswhowritereviewsNegativereviewstendtobelengthyandargumentativeoftendetailingdisappointmentoverexpectationsnotmetPositivereviewsontheotherhandtendtoberelativelybriefandconfirmtheoverallratingConsumerswhowroteldquohighrdquoreviewsplacedgreaterimportanceonvalueformoneycleanlinessandcomfortthandidthosewhowrotenegativereviewsThosewhowroteldquolowrdquoreviewsplacedtheiremphasisontheserviceprovidedbythepropertystaffandmanagementNegativereviewsweremorelikelytoinvolveahigherpriceaccommodationThisanalysisindicates that the overall numerical ratings typically used in review systems may not be the idealindicator of perceived service quality The results suggest that review sites should develop bettermethodstoaggregatesynthesizeandpublishthereviewcontentsparticularlythenumericalratingsThisandotherreviewsitesshowtheaverageofallthepoint-scaleratingsbutsuchsimplemeansdonottakeintoaccountthebiasesthatareinherentintheratingsystemsInsteadthesitesshouldprovidemore information and heuristics to help the consumers navigate through the clutter and get theinformationtheydesire

6 TheCenterforHospitalityResearchbullCornellUniversity

CorNell hosPiTAliTy rePorT

Giventhatconsumersareincreasinglyrelyingonsearchenginestoacquireinformationabout hotels restaurants and travel purveyors (among many other services)consumer-generated online reviews will inevitably change the structure andaccessibilityofinformationalongwithconsumersrsquoperceptionsofvariousservices

ThepopularityofonlinereviewshasledtoaproliferationofonlinereviewsitesincludingpopularsitessuchasHotelscomTripadvisorcomandYelpcomArecentstudybyForresterIncfoundthatmorethan80percentofwebshoppersreadandconsiderotherconsumersrsquoreviews1SimilarlyCompeteIncfoundthatnearly50percentofpurchasersvisitedamessageboardforumoronlinecommunityforinformationconnectedwiththeironlinetravelpurchasingandoneinthreeofthesebuyerssaidthatconsumer reviews helped with their purchase decision2 Almost half of these consumers said thatconsumersrsquoopinionsactuallycausedthemtochangetheirmindaboutwhattheypurchasedMoreoveramongthosebuyers25percentsaidtheyalsopostedareviewonaconsumerreviewsiteaftermakingtheirpurchase

1ForresterldquoFortyFactsaboutUSOnlineShoppersrdquoForresterInc20062CompeteldquoEmbracingConsumerBuzzCreatesMeasurementChallengesforMarketersrdquoCompeteInc2006

UnscramblingthePuzzlingMatterofOnlineConsumerRatings

An Exploratory Analysis

byPradeepRacherlaDanielConnollyandNatasaChristodoulidou

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 7

Howevertheexistingunderstandingofonlinetourism-relatedreviewsisrelativelyweakeventhoughweknowthatwordofmouth(WOM)ismoreimportantandinfluentialwithinaservicescontextthaninproductmarketingscenariosRecentlysomestudieshaveexploredtourism-andhospital-ity-relatedonlineWOM3Yetresearchhassofaroverlookedcertaininterestingaspectsofonlinereviewsasfollows

Onlinereviewstypicallyuseldquooverallstarratingsrdquoaspri-maryindicatorsofperceivedservicequalitymdashratingsthatarenotrelatedinanywaytotheformalstardesignationsawardedbyForbesorMichelinConsumersrsquomeanstarratingsaretypi-callyaggregatedfromtheratingsgivenbyindividualconsum-ersineachreviewWhatwedonrsquotknowishowconsumersdeterminethestarstheyassigntoaservicewhentheypostareviewonawebsiteThatisdostarratingsaccuratelyreflectconsumersrsquosentimentsregardingvariousattributesofserviceprovidersThisisavitalissuegiventheexperientialandin-tangiblenatureoftravelservicesandthefactthatconsumersusetheoverallratingasanimportantheuristicwhilenarrow-ingdownchoices

FurtherstudieshaveoverlookedtherelationshipbetweenthereviewsandvariousattributesofservicesandserviceprovidersSinceonlinereviewsareaformofWOMvariablessuchasconsumerinvolvementpricingandlengthofstaymayhaveanimpactonthefinalratingandpostingofthereviewprovidedbyconsumersItisimportanttouncovertheserelationshipstobetterunderstandthedynamicsofonlinereviews

FewstudieshaveanalyzedthequalitativeaspectofonlinereviewsOnlinereviewsareessentiallyopen-endedtext-basedconsumer-to-consumer(c2c)communicationsThetextcontentmaycontainnuancedviewsoftheservicesandserviceproviders(fromthereviewwritersrsquopointofview)thatcannotbeexpressedusingcrudenumericalratings4Hencethereisaneedtoanalyzethetextportionofthereviewsandidentifytheissuesthatconsumersaremostlyconcernedwithwhenwritingandpostingonlinerelatedreviews

WeaddresstheabovepointsinthisstudyThefollowingquestionsguidedus

3KHYooandUGretzelldquoWhatMotivatesConsumerstoWriteOnlineTravelReviewsrdquoInformation Technology amp TourismVol10(2008)pp283-295andZXiangandUGretzelldquoRoleofSocialMediainOnlineTravelInformationSearchrdquoTourism Management20094AGhoseandPGIpeirotisldquoEstimatingtheSocio-economicImpactofProductReviewsMiningTextandReviewerCharacteristicsrdquoInformation Systems Research2008

1 Dooverallratingsreflecttheperceivedqualityoftheserviceproviders(asderivedfromtheconsumersrsquoexpressedsentimentinthereviews)Whataretheunderlyingpatternsoftheoverallratingsinonlinereviews

2 Whatarethelinkagesamongthevariousattributesoftheserviceprovidersandcustomerreviewsand

3 WhataretheissuesthatconsumersmostlytalkaboutinthetextportionofonlinereviewsDotheseis-suesdifferfromtheratingsthattheconsumershaveprovidedviathestandardvariablesprovidedbythereviewsites

ResearchSettingandDataCollectionThisstudyisbasedonreviewscollectedfromldquoReviewsitecomrdquoaswehavecalledthecommunity-driventravelsitethatweanalyzedThesiteprovidesreviewsofverifiedrentalpropertiesandorganizesthereputationofvacationpropertiesThepropertylistingwithinthesiteisbasednotontherevenuesharingmodelthatisgenerallypracticedbyotherreviewsitesbutonthereputationtrustandfeedbackfromcustomerswhohavestayedatthevacationrentalpropertiesReviewsitecomisafreeservicethatoperatesonserviceproviderfeesandadvertisingrevenuesUn-likesometravelreviewsitesthefirmrsquosreviewsolicitationstrategyistosendpersonalizedinvitationsonlytothosecustomerswhosestayhasbeenverifiedbythepropertymanagementThesitersquosmanagementbelievesthatthisstrategytoalargeextenteliminatesfakereviews

Thefirmprovideduswith3300reviewsforanalysisAfterinitialdatacleanupabout100reviewswithincom-pleteorunusableinformationwereeliminatedand3197reviewswereusedfortheanalysisThereviewscontainedthefollowingattributesbull Theoverallratingassignedtotheserviceprovider

basedontheaggregationoftheratingsassignedbyindividualreviewers

bull Consumersrsquoratingsonthefollowingsixattributesvaluecleanlinesscomfortservicelocationandcheck-in

bull Thereviewtext(Semanticprocessingtechniqueswereappliedontheaggregatereviewtexttoidentifythenuancedopinionsandconcernsofthetravelerswhowrotereviews)

8 TheCenterforHospitalityResearchbullCornellUniversity

bull Agerangeoftherevieweronan1-4ordinalscale(25-3435-4445-54and55andabove)

bull Averagepriceperdayand

bull Wordcount(usedinthequantitativeanalysis)

PreliminaryHypothesesandResultsTheagegroupsofthereviewwritersareshowninExhibit1Alargepercentageofthecustomerswhoreviewedthepropertiesfallunder44yearsofageThisisconsistentwithpreviousstudiesinthisrealmwhichshowthatgenerationXandGenerationYaremoreactiveintheWOMforums

ReviewRatingDistribution Websitesthathostonlinereviewsgiveconsumerstheoptiontoassignaratingtotheserviceproviderstypicallyonascaleof1to5or1to7Manysitesaggregatethesenumeri-calorstarratingstodetermineanoverall(mean)starratingWeseethisoverallratingasanimportantelementinonlineWOMsinceconsumersoftenusethisratingtonarrowdowntheirconsiderationsetStudiesthathaveusedtheratingastheprimarypredictoroftheserviceprovidersrsquoperceivedqualityorcustomersatisfactiontodeterminetheimpactofonlinereviewsonproductsaleshavearrivedatmixedfind-ings5Howeverthesestudiesarebasedonthefundamentalassumptionthatthenumericalratingsassignedtoservice

5ForexampleseeJChevalierandDMayzlinldquoTheEffectofWordofMouthOnlineOnlineBookReviewsrdquoJournal of Marketing ResearchVol43(2006)pp345-354andWDuanBGuandABWhinstonldquoDoOnlineReviewsMattermdashAnEmpiricalInvestigationofPanelDatardquoDecision Support SystemsVol45(2008)pp1007-1016

providersareanaccurateestimateofthequalityperceivedbythecustomerThispracticefollowsthebasicassumptionofbehavioraltheoriststhatanydatasetwithalargenumberofconsumerresponsestendstofollowanormal(Gaussian)distributionRecentstudieshavequestionedthisassump-tionForinstanceDellaracosandNarayanrsquosstudyonmoviereviewsinYahoocomfoundthatonlytheextremelysatisfiedandextremelydissatisfiedcustomersarelikelytopubliclyexpresstheiropinionsascomparedtocustomerswithmod-erateopinions6SimilarlystudiesusingdatafromTripadvi-sorcomandAmazoncomfoundthatthestarratingsexhibitatruncateddistributioninwhichthemajorityofreviewsarepositive7

TotestthisassumptionweranasimplefrequencychartasshowninExhibit2Inkeepingwithotherstudiesthedistributionisheavilyskewedtowardsthepositiveratings1-pointreviews32(1)2-pointreviews95(3)3-pointreviews458(14)4-pointreviews688(22)and5-pointreviews1912(60)Totestfortherobustnessofthisdistri-butionweappliedtheKolmogorov-Smirnovtestofnormal-ityTheresultsconfirmthenon-normaldistributionofthestarratingsandtheheavyskewtowardspositivereviews

Threeexplanationsarepossibleforsuchadistribution

6CDellarocasandRNarayanldquoAStatisticalMeasureofaPopulationrsquosPropensitytoEngageinPost-purchaseOnlineWord-of-mouthrdquoStatisti-cal ScienceVol21(2006)pp2777ATalwarandRFBoiJurcaUnderstandingUserBehaviorinOnlineFeedbackReportingrdquoinElectronic CommerceSanDiegoCalifornia(2007)pp134-142

Exhibit 1

Distribution of age ranges

Exhibit 2

Distribution of overall ldquostar ratingsrdquo

55 and above 45ndash54

years

35ndash44 years

18ndash34 years

overall ldquostar ratingrdquo

Num

ber o

f rev

iew

s

3295

458 688

1912

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 9

1 AmajorityofthepropertiesthatarelistedonthewebsitehaveoutstandingperceivedqualityandprovidesatisfactoryservicestocustomersWhilethisistheo-reticallypossibleitseemsunlikelywhenappliedtoallthepropertieslistedonthewebsite8

2 ThereviewsdonotconstituteatrulyrepresentativesampleofconsumeropinionBasedonotherstud-iesthisexplanationseemsplausibleConsumerswithmoderateviewsmayseelittleincentivetoreporttheirviewsinanonlineforumandasaconsequenceonlythoseconsumerswhohavepolarizedviewsarelikelytoposttheircomments

3 ThedistributioncouldalsobeanartifactofthefirmrsquosreviewsolicitingstrategyRecentstudiessupportthisideabysuggestingthatcustomersengageinsociallydesirablebehavioreveninimpersonalinteractionswiththeserviceproviders9Inthiscaseitmaybethattheywillprovideareasonablypositiverevieweveniftheexperiencehasnotbeenuptothemark

StarRatingsandPerceivedServiceQualityLetrsquosreturntothefirstquestionweaskedattheoutsetArestarratingsatruerepresentationofserviceprovid-ersrsquoperceivedqualityasreflectedbytheconsumersrsquostatedsentimentsThisbecomesanimportantquestionfortwo

8AKadetldquoRah-RahRatingsOnlinerdquoSmartMoney20079YGreacutegoireandRJFisherldquoCustomerBetrayalandRetaliationWhenYourBestCustomersbecomeYourWorstEnemiesrdquoJournal of the Acad-emy of Marketing ScienceVol36(2008)pp247-261

reasons(a) ratingsaretheprimaryevidencethatconsum-ersconsiderwhileshoppingonlineforinformationand(b)thetruncatednatureofratingdistributionsmeansthatthereviewsavailablemaynottrulyrepresentconsumeropinionortheserviceprovidersrsquoperceivedqualityIfonlythosewithextremeviewspostreviewsthenconsumersshouldalwayslookforserviceprovidersthathavea5-starratingOnthecontraryifthereviewsaccuratelyreflectallconsumersrsquoviewsthenconsumersarebetteroffsearchingforreviewswitha3ratingsincethesetendtobemorebalancedandenablebetterproductanalysisRecentlyHu et aluseddatafromAmazoncomshowthatnumericalscoresdonotreflectbooksrsquotrueperceivedquality10AstheysuggestldquohellipratherthescorereflectsthebalanceofdiverseopinionsInotherwordswhenabookrsquosoverallscoreisaround3itdoesnotsuggestthatconsumersgenerallyagreethatthisisanaver-agebookItrathersuggeststhatroughlyequalnumberofconsumersthinkthatthebookiseitheranoutstandingbookoranabysmalbookrdquo

Totestthisassertionweappliedaregressionequationwithsixpropertyattributes(ievalueformoneycheck-inlocationcleanlinesscomfortandservice)asindependentvariablesandstarratingasthedependentvariableThisanalysisisbasedontheassumptionthatforagivenpropertyifstarratingstrulyrepresenttheperceivedqualityoftheproperties(aswellastheconsumersrsquosentiment)thenthestarratingsandratingsofthesixotherpropertyattributes

10NHuJZhangandPAPavlouOvercomingtheJ-shapedDistribu-tionofProductReviewsrdquoCommunications of the ACMVol52(2009)pp144-147

low reviews t-value high reviews t-value(Constant) 1618 18353 2400 34186vAlue 0254 13726 0378 8767CheCKiN -0083 -3701 -0053 -1192loCATioN 0104 6132 0079 2117CleAN 0144 7574 0112 2808CoMForT 0267 13993 0129 3088serviCe 0098 4209 0107 2401r2 0331 0338F-statistic 5038 24714

Exhibit 3

regression analysis of ldquolowrdquo and ldquohighrdquo reviews

Notes Coefficients (t-statistics reported) Significance is as follows significant at 0001 level significant at 05 level significant at 010 level Dependent variable is overall rating

10 TheCenterforHospitalityResearchbullCornellUniversity

moderateratingsonalltheattributesbutwasstillwillingtoprovideahighoverallstarratingtothepropertyThisexamplesuggeststhatconsumersplacesignificantemphasisontheoverallexperientialvalueprovidedbythepropertiesThisexpectationofvalueismanytimesintangibleanddoesnotnecessarilydependonthefunctioningorratingoftheindividualfeaturesoftheproperty

RelationshipbetweenRatingsandReviewAttributesTounderstandconsumersrsquoword-of-mouthbehavioritisimportanttoconsidervariousfactorssuchasperceivedexpectationsofcostsandbenefitsimportanceofthepur-chase(involvement)personalcharacteristicsandsituationalinfluencesWeconsidertwofactorsthatareknowntoaffecttheintensityandvalence(positiveornegative)ofwordofmouthpriceandwordcount

PricePriceisanimportantfactorthatcontributestopre-purchaseexpectationsTheintensityofexpectations(andtheirfulfillment)partlydrivenbypriceisdirectlypropor-tionaltoconsumersrsquosatisfactionlevelsandWOMpropensityThismaybethereasonthatthetendencytocomplainisdirectlyrelatedtothecostoftheservice11Themoreexpen-siveanitemthegreaterwillbetheperceivedlossesandthegreaterlikelihoodthatdissatisfactionwillresultinacom-plaintSimilarlythelikelihoodofword-of-mouthbeingtrig-geredbythelevelofexpectationndashdisconfirmationincreases

11BeardenWOandTeelJE(1983)Selecteddeterminantsofcon-sumersatisfactionandcomplaintreportsJournal of Marketing Research 2021-28

willshowahighcorrelationFortheanalysiswetookintoconsiderationthefactmostreviewsarepositive(ie4-and5-pointratings)Thereforewedividedthereviewsintotwocategorieshighreviews(starratingabove3n=2597)andlowreviews(ratingsbeloworequalto3n=600)andranseparateregressionanalysesonthesetwogroupsasshowninExhibit3

AninterestingfindingistheweakrelationshipbetweentheattributeratingsandoverallstarratingforeachpropertyTheR-squareoftheregressionequationwithallsixvariablesandtheoverallstarratingsislessthan40percentThatiseveniftheconsumersprovidedhigherratingsonindividualattributestheiroverallratingwasusuallylowandthere-versewasalsotrueThistrendisobservedinboththehighandlowreviewgroups

TheanalysissuggeststhatinhighreviewsconsumersplacedgreaterimportanceonvalueformoneycleanlinessandcomfortThehighertheratingsofaconsumerontheseattributesthehigheristhelikelihoodthattheconsumerwillprovideahigherstarratingCheck-inlocationandservicedidnotprominentlyfigureintheiroverallevalu-ationsofthepropertieseventhoughtheseattributeswereratedhigh

ThelowreviewsshowsimilarresultsinthesensethattheloweraconsumerratesvaluecleanlinessandcomfortthelowerthelikelihoodofapositiveratingInadditioninthecaseoflowreviewsconsumersplacedemphasisontheserviceprovidedbythepropertystaffandmanagement

TheaboveresultisillustratedinoneoftheldquohighrdquoreviewsshowninExhibit4Thereviewergavelowto

star rating Text Commentary

This was our 1st time renting a house vs a condo This house was great If you wanted to spend a little time by yourself you could so many rooms to go to The house was so comfortable like being at home It was our home for the week Thank you Hope to rent it next year if we havent bought one ourselves

ratings for specific Attributes

value Check-in location Clean Comfort service

3 2 3 2 4 3

Exhibit 4

example of a ldquohighrdquo review

Avg rating for Property Attributes

30

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 11

withthecostoftheproduct12Giventhegreaterperceivedriskinservicescustomerswouldseekmoretrustworthyser-viceproviderswithagoodreputationofsatisfyingcustomerneeds13ThisinturnincreasestheexpectationsandleadstomoreextremereactionsfromtheconsumersHenceonecanspeculatethatthehigherthepriceshigherwillbetheaver-agenegativeorpositiveratingsforaservice

Word countThenumberofwordsinareviewmayserveasaproxyfortheamountofinformationinthereviewsHowevertheevidenceonthisaspectofWOMofalltypesisequivocalForinstanceconsumersgenerallyengageinpositiveWOMtoavoidfeelingsofguiltandas-sociationwithbadnewsandtoreducecognitivedissonanceOntheotherhandacustomerwhoishighlydisappointedwillprovideapersuasiveargumenttomakesurethatothercustomersaredissuadedfromrentingthesamepropertyResearchonconventionalwordofmouthhasofferedampleevidencetosuggestthatdissatisfiedcustomersengageintwotothreetimesasmuchWOMassatisfiedcustomers14Webelievethatthislogicextendstoonlinereviewsdespitethefactthatexistingstudiesareequivocalatbest15Thereforeitcanbeassumedthatreviewswitheitherextremelypositiveorextremelynegativeratingswillhavegreaterwordcountasopposedtomoremoderatereviews

TotesttheseassumptionsweappliedANOVAwithpriceandwordcountasdependentvariablesandthereviewratingasfactorsAsillustratedinExhibits5and6theresultsshowthatpriceandwordcountshowasignificantnegativecorrelationwithstarratingThatisreviewswithlowerratingsaretypicallyassociatedwithgreaterwordcountandrepresentpropertiesthatchargehighrates

TheresultonthewordcountsuggeststhatconsumerswhoareunhappywitharentalpropertytendtowritemoretoexpresstheirdissatisfactionAsshowninthisexhibitlowerratingsareassociatedwithhighernumberofwordsinthereviewsandviceversaSimilarlypropertypriceseemstobestronglyassociatedwithnegativewordofmouthsup-portingtheprinciplethathighercostincreasestheconsum-ersrsquoinvolvementinthepurchase

12BeardenWOandMasonJB(1984)AninvestigationofinfluencesonconsumercomplaintreportsAdvancesinConsumerResearch11490-49513WetzerIMZeelenbergMandPietersR(2007)ldquoNevereatinthatrestaurantIdidrdquoExploringwhypeopleengageinnegativeword-of-mouthcommunicationPsychology and Marketing24661-68014SchlossbergH(1991)CustomersatisfactionNotafadbutawayoflifeMarketing News25andWestbrookR(1987)ProductConsump-tion-basedAffectiveResponsesandPost-purchaseProcessesJournal of Marketing Research 24258-27015AlexMSusskindldquoIToldYouSoRestaurantCustomersrsquoofWord-of-MouthCommunicationPatternsrdquoCornell Hotel and Restaurant Adminis-tration QuarterlyVol43No2(April2002)pp75-85

Exhibit 5

Differences in word count across star ratings

Exhibit 6

Differences in average price per day across ratings

Aver

age

pric

e pe

r day

of t

he re

ntal

pro

pert

y

overall ldquostar ratingrdquo

1504

1563

1804

19861987

Aver

age

wor

d co

unt o

f rev

iew

overall ldquostar ratingrdquo

836837

121

15751543

160

140

120

100

80

200

190

180

170

160

150

1 2 3 4 5

1 2 3 4 5

12 TheCenterforHospitalityResearchbullCornellUniversity

thereviewersThesecanbeclassifiedas(a)attributesorfea-turesofthepropertythatfigureprominentlyandcontributetothepositiveornegativeratings(egroomsbeddishesfurnitureorservice)and(b)thepositiveornegativewordsthatrepresentthesentimentofthereviewer(eggreatamazingdirtyandwonderful)Theresultsofthequalitativeanalysisareasfollows

Syntactic featuresTheoveralltoneofthelowreviewsisratherargumentativeThereviewerstendtobeforcefulinpresentingtheirpointofviewTheylayoutagreatamountofreasoningandprovidespecificexamplestodrivehometheirargumentThisisreflectedinthegreaterpercentageofnegativeconnectorwords(34)suchasbut however as ifandeven thoughItisinterestingtonotethatreviewersinthiscategoryspeakdirectlytotheotherconsumersnotablythroughgreateruseofwordssuchasyou(46)andlooktoinfluencetheirdecisionmaking

InthehighreviewsbycontrasttheoveralltonetendstobeenunciativeandthereviewersusethetexttomerelyreinforcetheirnumericalratingsusingfewerwordsonaveragethanthenegativereviewersInthesereviewstheconsumersfocusontheirpersonalexperiencesandappealtotheaudiencethroughemotionsratherthanlogicandreasoningasevidentfromgreateruseofpositiveadjectives

QualitativeAnalysisoftheReviewTextTheaboveanalysisemphasizestheimportanceofthetextcontentinconsumerreviewsWhilenumericalratingsallowthereviewerstoratetheserviceprovideronvariousattributesitisthetextthatprovidesthemwiththeopportu-nitytoarticulatethenuancesoftheiroverallexperienceandconveyusefulinformationaboutaserviceproviderrsquostransac-tionsandservicecapabilitiesThereforesemanticanalysisofthetextcontentprovidesuswithanopportunitytothethirdquestionpresentedattheoutsetofthisreportwhatdoconsumerstalkaboutintheirreviews

Tothisendweanalyzed600eachofrandomlyselectedhighandlowreviewsusingsemanticprocessingtechniquesWeusedcenteringresonanceanalysis(CRA)amodeofnet-work-basedtextanalysisthatrepresentsthecontentoflargesetsoftextsbyidentifyingthemostimportantwordsthatlinkotherwordsinthenetwork16OuranalysisexaminedsyntacticandsemanticfeaturesSyntacticfeaturesincludethepercentageofthewordsthatareopenclass(ienounsverbsadjectives)Semanticfeaturesrepresentthespecial-izedvocabularythatreflectstheoverallopinionexpressedby

16CormanSandDooleyK(2006)CrawdadTextAnalysisSystem12CrawdadTechnologiesLLCChandlerAZ

Prominently Mentioned Features Freq sample Comments

bedroom beds linen sheets 383

Also the same bedroom and bathroom did not have any door separating the bedroom from the bathroom no privacy

Sleeping arrangements were very awkward The listing sheet described 2 full beds We brought our own linens the beds were queen sized Had to purchase sheet

Kitchen Kitchenware Dishes Dishwasher 297

You must not forget to step up when yoursquore walking out of the kitchen into the dining room otherwise yoursquoll trip over the 3-foot step up

No potholders in the kitchen and it didnrsquot seem like there were as many kitchen supplies as before We just had to make do without

beach beaches beachfront 296

The week was the best vacation we have ever had as a family The beach house was a class act

Disappointment 1 It looks like this house is very close to beach but to get there you have to go around through public path

Pool Poolside 218

The pool was too dirty to use We checked in on Thursday and the pool was not cleaned until Saturday

Also it is inconvenient to have to call to get someone to open the pool Didnrsquot understand why Jacuzzi was open and not the pool

bathroom(s) bath bathtub shower 202

The next problem was in another bathroom (main floor)mdashthe wall heater did not work so showering in here was a bit chilly The next problem was in the ldquomasterrdquo bathroom (connected to the bedroom with the King bed) The sliding shower doors need to be totally replaced

Bathtub drains clogged in three second floor bathrooms

Exhibit 7

Top five prominent issues in low reviews

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 13

suchasgreat fantastic andawesome(45)andpersonalpronounssuchasweandI(54)

Semantic featuresExhibits7and8showthemainissuesoraspectsthatareprominentlydiscussedinthelowreviewsTheorderoftheseissuesisbasedonthestatisti-callysignificantfrequencyoftheiroccurrenceinthe1200reviews

WefurtherinvestigatedtheuseofvarioussentimentwordsInterestinglytheprominentissuesdiscussedinthepositivereviewssometimesdonotevenfigureintheratingscalesprovidedtothereviewersOveralltheresultsofthesemanticanalysiscanbesummarizedasfollows

Thepropertyanditsfeaturesaremorecommonlycon-nectedwithextremelypositivesentimentwordssuchasgreat wonderfulandamazinginthehighreviewsthaninthelowreviewsIncomparisonthepropertiesandtheirfea-turesinthelowreviewsaremorelikelytobecharacterizedwithmoderatelypositivewordssuchasniceandokay

Negativereviewsreflecttheextentofexpectationndashdis-confirmationthattheconsumersfaceduringtheirstayintherentalpropertyThesereviewshaveagreaterproportionofwordsexpressingdisappointment(egnotasexpected)Furtherextremelynegativesentimentwords(egdiscom-fortfilthy)andadjectivesaremorelikelytobeassociatedwithfeaturesinlowreviewsthaninhighreviews

Almost40percentoftheconsumerswhowrotehighreviewsexpresstheirwillingnesstoreturntothepropertyasopposedtoonly9percentofthosewritingnegativereviewsFurthermorethan20percentoftheconsumersgivinghighreviewsexpresstheirstrongdesiretorecommendtheprop-ertytotheirfriendsandfamilymembersasopposedtoonly5percentinlowreviews

ReviewofFindingsOurexploratorystudyofconsumerreviewsofUSvacationhomesonReviewsitecomwashighlightedbyoverwhelm-inglypositivereviews

Theoveralldistributionofthereviewswasheav-ilyskewedtowardsthepositiveratingsMorecriticallythisfindingsuggeststhatthereisalackofmoderateandbalancedopinionsWetheorizethatthisiseitherduetopurchasingbias(sincecustomerswentthroughthetroubleofpurchasingandexperiencingthefacilitiestheytendtobepositivelybiasedtowardstheirpurchase)orunder-reportingbias(customersdonothavestrongenoughincentivestotakethetroubleofreportingtheiropinionsandthefewmotivatedindividualswhoprovidetheirreviewsbecomeaself-selectedsample)

Theoverall(mean)starratingtypicallyusedintravelerreviewsitesisprobablynotthemostaccurateindicatorofcustomersrsquoperceptionsofserviceprovidersrsquoqualityWe

Prominently Mentioned Features Freq sample Commentsbeach beaches beachfront 506 Perfect rental property for those who enjoy the activity of the beach and the serenity

of the Sound The house is just down the road from ocean

The large porch offered a beautiful view of the beach and comfortable new rocking chairs

Pool Poolside 218 The outdoor shower dipping pool and screen in porch were an added bonus Perfect house for a low key weekend

House was great for kids Pool was terrific and the playroom was perfectlocation 192 We had a wonderful time My wife loves fishing and the cottage was in a perfect

location to go fishing on the pier

The home is somewhat dated but the location was great We have vacationed in the Chatham area for many summers and this time it was great too

bedroom beds 186 The family suite is huge with a king size and twin bed and still room for a pack-n-play The kitchen and dining room were huge clean and nice

The only comments that I would say to the negative is the front bed room by the front door always seemed damp which seem to attract more bugs

Kitchen Kitchenware Dishes Dishwasher 178 All of the amenities in the house are upscalemdashloved the kitchen and plasma TVs

The kitchen had good basic utensils dishes and cooking pans

Exhibit 7

Top five prominent issues in high reviews

14 TheCenterforHospitalityResearchbullCornellUniversity

foundacorrelationbetweenthestarratingandratingsonotherpropertyattributesoflessthan40percentinmostcases

AhighwordcountinthereviewsiscorrelatedwithalowoverallratingThatisconsumerswhoareextremelydissatisfiedtendtoexpendmoretimeandenergyontheircritiqueofthepropertyitsamenitiesandservicesPriceofthepropertyalsoseemstopositivelycorrelatewithextremenegativereviewsperhapsbecauseahighpriceheightenstravelersrsquoexpectations

Numericalratingsdonotoftencapturetheactualsentimentandthevarietyofdimensionsonwhichtravel-ersevaluatepropertiesBecausenumericalratingsdonotcapturereviewersrsquoopinionsaboutspecificattributestheycanbedeceivingandmaynotrevealthetruequalityofthepropertyortheactualintentionofthereviewerWefoundforinstancethatconsumersvaluelocation(egclosenesstothebeach)andcleanliness(egswimmingpoolareas)morethansuchfeaturesascheck-inandserviceSimilarlyinthenegativereviewsissuessuchaslackofbedroomparapher-naliaorkitchenitemsseemtoberankedhigherinimpor-tancethanissuessuchaslocationandcheck-inThereforeevenwhenapropertyscoreshighonthecertaindimensionstheoverallratingmayremainlowbecauseotherfactorsovershadowthefavorableaspects

WhatItMeansforManagersWeseeimplicationsforonlinereviewsitesandbyexten-sionforproprietorsofvacationhomesandserviceopera-tionsgenerallyWithregardtothereviewsitesifconsumersconsistentlydetectafavorablebiasthereviewsitesmaylosethecriticalaspectsofcredibilityandtrustAlthoughconsumersaregenerallysmartandrationalandcanworkaroundobviousbiasesrecentevidencesuggeststhatcon-sumersaregraduallydefectingfromsomeonlinesitesduetounbalancedreviews17InthisregardsiteadministratorsaswellaspropertymanagersshouldtakestepstoattractthoseconsumerswhoarewillingtoprovideareasonablybalancedanalysisoftheamenitiesandservicesBythiswemeanreviewsthatpresentcomprehensiveandobjectiveanalysisoftheserviceproviderratherthanjustravesorrants

Tothatendsiteoperatorsshouldtakestepstoun-derstandcustomersrsquoreviewwritingbehaviorandtoat-tractmorebalancedreviewsOnewaytodothisistogivepersonalizedinformationtotheconsumerdemonstratingthecontributionmadebyhisorherreviewandthevalueofthereviewtostakeholdersForinstanceLinget alfoundapositivecorrelationbetweenwillingnesstowritereviewsandtheconsumersrsquoperceptionthattheircontributionswouldbeuniqueandhelpfultoothercustomersaswellasperusedby

17HarteveldtHHJohnsonCStarkEandGeldernKV(2009)Us-ingDigitalChannelsToCalmTheAngryTravelerForresterResearch

theserviceproviders18Simplyputthepropensitytowritereviewsincreaseswhentheconsumersfeelthattheirreviewisactuallyhelpingothers(boththeserviceproviderandotherconsumers)

WesuggestthatreviewsitesdevelopbettermethodstoaggregatesynthesizeandpublishspecificreviewcommentsaswellasnumericalratingsFirstmanagersshouldtakeintoconsiderationthelowcorrelationbetweentheoverallratingsandconsumersrsquoassessmentsofvariouspropertyattributes(aswellastheresultsfromthetextanalysis)Theimplicationofsuchaweakcorrelationisthatreviewsitesneedtoexpandonthelistofpossiblevariablessothattheratingsaccuratelyreflectconsumersrsquoneedsandexpectationsregardingserviceproviders

Furthersitemanagersmayhavetorethinkthecurrentpracticeofaveragingalltheratingsforagivenpropertybe-causesuchameanratingfailstotakeintoaccountthebiasesthatareinherentintheratingsystemsAsanexamplethisstudyfindssignificantunder-reportingbiasinthisonlinereviewsiteAcustomerrsquosdecisiontonotpostonlinereviewsmaymeanthelossofimportantinformationthatcanassisttheusersofthesitestomakemorereliableinferencesThemajorityoftodayrsquosfeedbackmechanismsdonotpubliclydisclosethenumberofsilenttransactions(ietransactionsforwhichnofeedbackwaspostedbycustomers)Suchinfor-mationshouldbecomeapartofaserviceproviderrsquosonlineprofileonratingsitesandotherfeedbackmechanismsItisimportanttoprovidemoreinformationandheuristicstohelptheconsumersnavigatethroughtheclutterandgettheinformationtheywantandseek

Itisimportanttoapplymodernsemanticanalysisandsentimentclassificationtechniquestostudythecontentofthereviewsandidentifyvariousdimensionsonwhichcon-sumersevaluatevacationhomesThedetailsprovidedinthereviewtextareinmanywaysabetterreflectionofcustomersatisfactionServiceproviderscanbetterpositionthemselvesusingthedimensionsuncoveredthroughtextanalysisandtargettheneedsandpreferencesoftheircustomers

TheresultsinthisstudyshowthatnegativereviewsaretypicallyassociatedwithagreaterwordcountthanpositivereviewsItisusefultofocusspecificallyonthesenegativereviewsbecausetheycanrevealpatternsofdeficienciesintheservicestandardsanddeliveryThesereviewsshouldbetreatedthesameascustomercommentcardsorletters19Furthercustomerswhowritesuchlengthynegativereviews

18LingKimberlyGerardBeenenPaulLudfordXiaoxingWangKevinChangXinLiDanCosleyDanFrankowskiLorenTerveenAlMRashidandRobertKrautldquoUsingSocialPsychologytoMotivateContributionstoOnlineCommunitiesrdquoJournal of Computer-Mediated Communication10no4(2005)19MilanR(2007)10thingsyoucandoinresponsetotravelerreviewsinHotelmarketingcom

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 15

aremostlikelytospreadnegativeword-of-mouthaboutthepropertyLengthyreviewsareexcellentsourcesofinforma-tionandshouldbeusedasthebasestounderstandthecomplainingbehavioroftheconsumer

WordcountmaybeusedasanadditionalsegmentationvariabletoidentifythoseconsumerswhoaremoreinvolvedandwillingtoprovidebalancedfeedbacktoboththereviewsiteaswellastopropertymanagementVariousotherresearchmethodssuchastelephoneinterviewsorfocusgroupsshouldbeusedtoelicitdetailedfeedbackfromsuchconsumersTheseactionsalsohelpinservicerecoverypost-purchaseengagementandrelationshipbuilding

OutlookWeforeseethreepotentialareasinvestigationthatcanaddvaluetocurrentbestpracticesWestilldonrsquotknowmuchaboutconsumersrsquomotivationtowriteonlinereviews20Recentstudiesidentifyvariousantecedentsofreviewwritingbutsufferfromsamplingbiassincetheysurveyaself-select-edsampleofconsumerswhoalreadyprovidedreviewsAstheresultsinthisstudyindicateunder-reportingbias(lackofmoderateandlowreviews)remainsprevalentinonlinereviewsitesFutureresearchshouldidentifysocio-psycho-logicalfactorsthatincreaseconsumersrsquopropensitytowriteonlinereviews

Futureresearchshouldexploretheimpactofthepres-ence(orlack)ofextremelynegativereviewswithinalargecorpusofpositivereviewsTheevidenceontheimpactofre-

20WangYandFesenmaierDR(2003)AssessingMotivationofCon-tributioninOnlineCommunitiesAnEmpiricalInvestigationofanOn-lineTravelCommunityElectronic Markets1333-45andYooKyungHandUlrikeGretzelldquoWhatMotivatesConsumerstoWriteOnlineTravelReviewsrdquoInformation Technology amp Tourism10No4(2008)283-95

viewvalence(negativeorpositive)isequivocalProponentsofconfirmatorybiassuggestthatconsumerslookforaffir-mativeevidencesupportingaproductchoicealreadymadeIfthatisthecasepositivereviewsaremorelikelytohaveagreatereffectonconsumeractionsOntheotherhandthenotionofnegativitybiassuggeststhatwhenconsumersareneutralnegativereviewstendtobecomemoresalientthanpositivereviews21Inthepresenceoflargenumbersofposi-tivereviewsconsumersspecificallyseeknegativereviewsthattheyfeelwillhelpthemidentifyspecificproblemswithaserviceUnderstandinghowconsumersreconciletheinformationprovidedbybothnegativeandpositivereviewshasimplicationsforthedesignandmanagementofreviewsites

Finallyareviewrsquoswordcounthasimportantimplica-tionsforconsumersrsquotrustinthereviewsWhenconsumersarewillingtoreadandcompareopen-endedcommentsfromotherconsumerstheamountofinformationinareviewcanmatterWOMforhighinvolvementproductsismorepersuasiveandincreasesthedecisionmakerrsquosconfi-dencewhenthemessagesenderprovidesgreaternumberofreasonsfortheiroverallratingLongerreviewsprovidemoreinformationandarelikelytobeperceivedasmorehelpfulandpersuasivethanshorterreviewsThewaysinwhichareviewiswrittenframedandpresentedisaninterestingfactorwhichhasrarelybeeninvestigatedandhasimportantimplicationsparticularlygivenourfindingthattheamountofinformationinareview(wordcount)variessignificantlywiththeratingsofthereviewn

21BaSandPavlouPA(2002)EvidenceoftheEffectofTrustBuildingTechnologyinElectronicMarketsPricePremiumsandBuyerBehaviorMIS Quarterly26243-268

Cornell Hospitality Quarterlyhttpcqxsagepubcom

2011 ReportsVol11No15DesigningaSelf-HealingServiceSystemAnIntegrativeModelbyRobertCFordandMichaelCSturman

Vol11No14ReversingtheGreenBacklashWhyLargeHospitalityCompaniesShouldWelcomeCrediblyGreenCompetitorsbyMichaelGiebelhausenPhDandHaeEunHelenChunPhD

Vol11No13DevelopingaSustainabilityMeasurementFrameworkforHotelsTowardanIndustry-wideReportingStructurebyEricRicaurte

Vol11No12CreatingValueforWomenBusinessTravelersFocusingonEmotionalOutcomesbyJudiBrownellPhD

Vol11No11CustomerLoyaltyANewLookattheBenefitsofImprovingSegmentationEffortswithRewardsProgramsbyClayVoorheesPhDMichaelMcCallPhDandRogerCalantonePhD

Vol11No10CustomerPerceptionsofElectronicFoodOrderingbySherylEKimes

Vol11No92011TravelIndustryBenchmarkingStatusofSeniorDestinationandLodgingMarketingExecutivesbyRohitVermaPhDandKenMcGill

Vol11No8SearchOTAsandOnlineBookingAnExpandedAnalysisoftheBillboardEffectbyChrisAndersonPhD

Vol11No7OnlineMobileandTextFoodOrderingintheUSRestaurantIndustrybySherylEKimesPhDandPhilippFLaqueacute

Vol11No6HotelGuestsrsquoReactionstoGuestRoomSustainabilityInitiativesbyAlexSusskindPhDandRohitVermaPhD

Vol11No5TheImpactofTerrorismandEconomicShocksonUSHotelsbyCathyAEnzRenaacutetaKosovaacuteandMarkLomannoVol11No4ImplementingHumanResourceInnovationsThreeSuccessStoriesfromtheServiceIndustrybyJustinSunandKateWalshPhD

Vol11No3Compendium2011

Vol11No2PositioningaPlaceDevelopingaCompellingDestinationBrandbyRobertJKwortnikPhDandEthanHawkesMBA

Vol11No1TheImpactofHealthInsuranceonEmployeeJobAnxietyWithdrawalBehaviorsandTaskPerformancebySeanWayPhDBillCarrollPhDAlexSusskindPhDandJoeCYLeng

2011 Hospitality ToolsVol2No1TheGameHasChangedANewParadigmforStakeholderEngagementbyMaryBethMcEuen

2011 Industry PerspectivesNo7MegaTips2TwentyTestedTechniquesforIncreasingYourTipsbyMichaelLynn

2011 ProceedingsVol3No5BuildingBrandsintheInternetAgeAnalyticsLoyaltyandCommunicationbyGlennWithiam

Vol3No4BraveNewWorldOnlineHotelDistributionbyGlennWithiam

Vol3No3SocialMediaandtheHospitalityIndustryHoldingtheTigerbytheTailbyGlennWithiam

Vol3No2TheChallengeofHotelandRestaurantSustainabilityFindingProfitinldquoBeingGreenrdquobyGlennWithiam

Vol3No1CautiousOptimismCHRSExaminesHospitalityIndustryTrendsbyGlennWithiam

2010 ReportsVol10No18HowTravelersUseOnlineandSocialMediaChannelstoMakeHotel-choiceDecisionsbyLauraMcCarthyDebraStockandRohitVermaPhD

Vol10No17PublicorPrivateTheHospitalityInvestmentDecisionbyQingzhongMaPhDandAthenaWeiZhangPhD

Vol10No16BestPracticesinSearchEngineMarketingandOptimizationTheCaseoftheStJamesHotelbyGregBodenlcosVictorBogertDanGordonCarterHearneandChrisKAndersonPhD

Vol10No15TheImpactofPrix Fixe MenuPriceFormatsonGuestsrsquoDealPerceptionbyShuoWangandMichaelLynnPhD

Cornell Center for Hospitality Research

Publication Indexwwwchrcornelledu

The Executive Path Hospitality Leadership Through Learning

Complete program information and applications available online

wwwhotelschoolcornelleduexecedPhone + 1 607 255 4919 Email exec_ed_hotelcornelledu

Professionals from around the world are invited to attend 3-day 10-day or online courses at the worldrsquos leading institute for hospitality management education in

Visit our website to apply

Explore develop and apply ideas with global hospitality leaders and

expert Cornell professors

Success

AdvancingBusiness

andPersonal

bull Strategic Leadershipbull Financebull Foodservicebull Human Resources

bull Marketingbull Operationsbull Real Estate

No2RetaliationWhyanIncreaseinClaimsDoesNotMeantheSkyIsFallingbyDavidSherwynJDandGreggGilmanJD

2009 ToolsToolNo12MeasuringtheDiningExperienceTheCaseofVitaNovabyKeshPrasadandFredJDeMiccoPhD

Cornell Center for Hospitality ResearchPublication Index

The Executive Path Hospitality Leadership Through Learning

Complete program information and applications available online

wwwhotelschoolcornelleduexecedPhone + 1 607 255 4919 Email exec_ed_hotelcornelledu

Professionals from around the world are invited to attend 3-day 10-day or online courses at the worldrsquos leading institute for hospitality management education in

Visit our website to apply

Explore develop and apply ideas with global hospitality leaders and

expert Cornell professors

Success

AdvancingBusiness

andPersonal

bull Strategic Leadershipbull Financebull Foodservicebull Human Resources

bull Marketingbull Operationsbull Real Estate

visit our website to apply

wwwchrcornell edu

Page 6: Unscrambling the Puzzling Matter of Online Consumer ... · Philips Hospitality Americas Chris Proulx, CEO, eCornell & Executive Education Carolyn D. Richmond, Partner, Hospitality

6 TheCenterforHospitalityResearchbullCornellUniversity

CorNell hosPiTAliTy rePorT

Giventhatconsumersareincreasinglyrelyingonsearchenginestoacquireinformationabout hotels restaurants and travel purveyors (among many other services)consumer-generated online reviews will inevitably change the structure andaccessibilityofinformationalongwithconsumersrsquoperceptionsofvariousservices

ThepopularityofonlinereviewshasledtoaproliferationofonlinereviewsitesincludingpopularsitessuchasHotelscomTripadvisorcomandYelpcomArecentstudybyForresterIncfoundthatmorethan80percentofwebshoppersreadandconsiderotherconsumersrsquoreviews1SimilarlyCompeteIncfoundthatnearly50percentofpurchasersvisitedamessageboardforumoronlinecommunityforinformationconnectedwiththeironlinetravelpurchasingandoneinthreeofthesebuyerssaidthatconsumer reviews helped with their purchase decision2 Almost half of these consumers said thatconsumersrsquoopinionsactuallycausedthemtochangetheirmindaboutwhattheypurchasedMoreoveramongthosebuyers25percentsaidtheyalsopostedareviewonaconsumerreviewsiteaftermakingtheirpurchase

1ForresterldquoFortyFactsaboutUSOnlineShoppersrdquoForresterInc20062CompeteldquoEmbracingConsumerBuzzCreatesMeasurementChallengesforMarketersrdquoCompeteInc2006

UnscramblingthePuzzlingMatterofOnlineConsumerRatings

An Exploratory Analysis

byPradeepRacherlaDanielConnollyandNatasaChristodoulidou

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 7

Howevertheexistingunderstandingofonlinetourism-relatedreviewsisrelativelyweakeventhoughweknowthatwordofmouth(WOM)ismoreimportantandinfluentialwithinaservicescontextthaninproductmarketingscenariosRecentlysomestudieshaveexploredtourism-andhospital-ity-relatedonlineWOM3Yetresearchhassofaroverlookedcertaininterestingaspectsofonlinereviewsasfollows

Onlinereviewstypicallyuseldquooverallstarratingsrdquoaspri-maryindicatorsofperceivedservicequalitymdashratingsthatarenotrelatedinanywaytotheformalstardesignationsawardedbyForbesorMichelinConsumersrsquomeanstarratingsaretypi-callyaggregatedfromtheratingsgivenbyindividualconsum-ersineachreviewWhatwedonrsquotknowishowconsumersdeterminethestarstheyassigntoaservicewhentheypostareviewonawebsiteThatisdostarratingsaccuratelyreflectconsumersrsquosentimentsregardingvariousattributesofserviceprovidersThisisavitalissuegiventheexperientialandin-tangiblenatureoftravelservicesandthefactthatconsumersusetheoverallratingasanimportantheuristicwhilenarrow-ingdownchoices

FurtherstudieshaveoverlookedtherelationshipbetweenthereviewsandvariousattributesofservicesandserviceprovidersSinceonlinereviewsareaformofWOMvariablessuchasconsumerinvolvementpricingandlengthofstaymayhaveanimpactonthefinalratingandpostingofthereviewprovidedbyconsumersItisimportanttouncovertheserelationshipstobetterunderstandthedynamicsofonlinereviews

FewstudieshaveanalyzedthequalitativeaspectofonlinereviewsOnlinereviewsareessentiallyopen-endedtext-basedconsumer-to-consumer(c2c)communicationsThetextcontentmaycontainnuancedviewsoftheservicesandserviceproviders(fromthereviewwritersrsquopointofview)thatcannotbeexpressedusingcrudenumericalratings4Hencethereisaneedtoanalyzethetextportionofthereviewsandidentifytheissuesthatconsumersaremostlyconcernedwithwhenwritingandpostingonlinerelatedreviews

WeaddresstheabovepointsinthisstudyThefollowingquestionsguidedus

3KHYooandUGretzelldquoWhatMotivatesConsumerstoWriteOnlineTravelReviewsrdquoInformation Technology amp TourismVol10(2008)pp283-295andZXiangandUGretzelldquoRoleofSocialMediainOnlineTravelInformationSearchrdquoTourism Management20094AGhoseandPGIpeirotisldquoEstimatingtheSocio-economicImpactofProductReviewsMiningTextandReviewerCharacteristicsrdquoInformation Systems Research2008

1 Dooverallratingsreflecttheperceivedqualityoftheserviceproviders(asderivedfromtheconsumersrsquoexpressedsentimentinthereviews)Whataretheunderlyingpatternsoftheoverallratingsinonlinereviews

2 Whatarethelinkagesamongthevariousattributesoftheserviceprovidersandcustomerreviewsand

3 WhataretheissuesthatconsumersmostlytalkaboutinthetextportionofonlinereviewsDotheseis-suesdifferfromtheratingsthattheconsumershaveprovidedviathestandardvariablesprovidedbythereviewsites

ResearchSettingandDataCollectionThisstudyisbasedonreviewscollectedfromldquoReviewsitecomrdquoaswehavecalledthecommunity-driventravelsitethatweanalyzedThesiteprovidesreviewsofverifiedrentalpropertiesandorganizesthereputationofvacationpropertiesThepropertylistingwithinthesiteisbasednotontherevenuesharingmodelthatisgenerallypracticedbyotherreviewsitesbutonthereputationtrustandfeedbackfromcustomerswhohavestayedatthevacationrentalpropertiesReviewsitecomisafreeservicethatoperatesonserviceproviderfeesandadvertisingrevenuesUn-likesometravelreviewsitesthefirmrsquosreviewsolicitationstrategyistosendpersonalizedinvitationsonlytothosecustomerswhosestayhasbeenverifiedbythepropertymanagementThesitersquosmanagementbelievesthatthisstrategytoalargeextenteliminatesfakereviews

Thefirmprovideduswith3300reviewsforanalysisAfterinitialdatacleanupabout100reviewswithincom-pleteorunusableinformationwereeliminatedand3197reviewswereusedfortheanalysisThereviewscontainedthefollowingattributesbull Theoverallratingassignedtotheserviceprovider

basedontheaggregationoftheratingsassignedbyindividualreviewers

bull Consumersrsquoratingsonthefollowingsixattributesvaluecleanlinesscomfortservicelocationandcheck-in

bull Thereviewtext(Semanticprocessingtechniqueswereappliedontheaggregatereviewtexttoidentifythenuancedopinionsandconcernsofthetravelerswhowrotereviews)

8 TheCenterforHospitalityResearchbullCornellUniversity

bull Agerangeoftherevieweronan1-4ordinalscale(25-3435-4445-54and55andabove)

bull Averagepriceperdayand

bull Wordcount(usedinthequantitativeanalysis)

PreliminaryHypothesesandResultsTheagegroupsofthereviewwritersareshowninExhibit1Alargepercentageofthecustomerswhoreviewedthepropertiesfallunder44yearsofageThisisconsistentwithpreviousstudiesinthisrealmwhichshowthatgenerationXandGenerationYaremoreactiveintheWOMforums

ReviewRatingDistribution Websitesthathostonlinereviewsgiveconsumerstheoptiontoassignaratingtotheserviceproviderstypicallyonascaleof1to5or1to7Manysitesaggregatethesenumeri-calorstarratingstodetermineanoverall(mean)starratingWeseethisoverallratingasanimportantelementinonlineWOMsinceconsumersoftenusethisratingtonarrowdowntheirconsiderationsetStudiesthathaveusedtheratingastheprimarypredictoroftheserviceprovidersrsquoperceivedqualityorcustomersatisfactiontodeterminetheimpactofonlinereviewsonproductsaleshavearrivedatmixedfind-ings5Howeverthesestudiesarebasedonthefundamentalassumptionthatthenumericalratingsassignedtoservice

5ForexampleseeJChevalierandDMayzlinldquoTheEffectofWordofMouthOnlineOnlineBookReviewsrdquoJournal of Marketing ResearchVol43(2006)pp345-354andWDuanBGuandABWhinstonldquoDoOnlineReviewsMattermdashAnEmpiricalInvestigationofPanelDatardquoDecision Support SystemsVol45(2008)pp1007-1016

providersareanaccurateestimateofthequalityperceivedbythecustomerThispracticefollowsthebasicassumptionofbehavioraltheoriststhatanydatasetwithalargenumberofconsumerresponsestendstofollowanormal(Gaussian)distributionRecentstudieshavequestionedthisassump-tionForinstanceDellaracosandNarayanrsquosstudyonmoviereviewsinYahoocomfoundthatonlytheextremelysatisfiedandextremelydissatisfiedcustomersarelikelytopubliclyexpresstheiropinionsascomparedtocustomerswithmod-erateopinions6SimilarlystudiesusingdatafromTripadvi-sorcomandAmazoncomfoundthatthestarratingsexhibitatruncateddistributioninwhichthemajorityofreviewsarepositive7

TotestthisassumptionweranasimplefrequencychartasshowninExhibit2Inkeepingwithotherstudiesthedistributionisheavilyskewedtowardsthepositiveratings1-pointreviews32(1)2-pointreviews95(3)3-pointreviews458(14)4-pointreviews688(22)and5-pointreviews1912(60)Totestfortherobustnessofthisdistri-butionweappliedtheKolmogorov-Smirnovtestofnormal-ityTheresultsconfirmthenon-normaldistributionofthestarratingsandtheheavyskewtowardspositivereviews

Threeexplanationsarepossibleforsuchadistribution

6CDellarocasandRNarayanldquoAStatisticalMeasureofaPopulationrsquosPropensitytoEngageinPost-purchaseOnlineWord-of-mouthrdquoStatisti-cal ScienceVol21(2006)pp2777ATalwarandRFBoiJurcaUnderstandingUserBehaviorinOnlineFeedbackReportingrdquoinElectronic CommerceSanDiegoCalifornia(2007)pp134-142

Exhibit 1

Distribution of age ranges

Exhibit 2

Distribution of overall ldquostar ratingsrdquo

55 and above 45ndash54

years

35ndash44 years

18ndash34 years

overall ldquostar ratingrdquo

Num

ber o

f rev

iew

s

3295

458 688

1912

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 9

1 AmajorityofthepropertiesthatarelistedonthewebsitehaveoutstandingperceivedqualityandprovidesatisfactoryservicestocustomersWhilethisistheo-reticallypossibleitseemsunlikelywhenappliedtoallthepropertieslistedonthewebsite8

2 ThereviewsdonotconstituteatrulyrepresentativesampleofconsumeropinionBasedonotherstud-iesthisexplanationseemsplausibleConsumerswithmoderateviewsmayseelittleincentivetoreporttheirviewsinanonlineforumandasaconsequenceonlythoseconsumerswhohavepolarizedviewsarelikelytoposttheircomments

3 ThedistributioncouldalsobeanartifactofthefirmrsquosreviewsolicitingstrategyRecentstudiessupportthisideabysuggestingthatcustomersengageinsociallydesirablebehavioreveninimpersonalinteractionswiththeserviceproviders9Inthiscaseitmaybethattheywillprovideareasonablypositiverevieweveniftheexperiencehasnotbeenuptothemark

StarRatingsandPerceivedServiceQualityLetrsquosreturntothefirstquestionweaskedattheoutsetArestarratingsatruerepresentationofserviceprovid-ersrsquoperceivedqualityasreflectedbytheconsumersrsquostatedsentimentsThisbecomesanimportantquestionfortwo

8AKadetldquoRah-RahRatingsOnlinerdquoSmartMoney20079YGreacutegoireandRJFisherldquoCustomerBetrayalandRetaliationWhenYourBestCustomersbecomeYourWorstEnemiesrdquoJournal of the Acad-emy of Marketing ScienceVol36(2008)pp247-261

reasons(a) ratingsaretheprimaryevidencethatconsum-ersconsiderwhileshoppingonlineforinformationand(b)thetruncatednatureofratingdistributionsmeansthatthereviewsavailablemaynottrulyrepresentconsumeropinionortheserviceprovidersrsquoperceivedqualityIfonlythosewithextremeviewspostreviewsthenconsumersshouldalwayslookforserviceprovidersthathavea5-starratingOnthecontraryifthereviewsaccuratelyreflectallconsumersrsquoviewsthenconsumersarebetteroffsearchingforreviewswitha3ratingsincethesetendtobemorebalancedandenablebetterproductanalysisRecentlyHu et aluseddatafromAmazoncomshowthatnumericalscoresdonotreflectbooksrsquotrueperceivedquality10AstheysuggestldquohellipratherthescorereflectsthebalanceofdiverseopinionsInotherwordswhenabookrsquosoverallscoreisaround3itdoesnotsuggestthatconsumersgenerallyagreethatthisisanaver-agebookItrathersuggeststhatroughlyequalnumberofconsumersthinkthatthebookiseitheranoutstandingbookoranabysmalbookrdquo

Totestthisassertionweappliedaregressionequationwithsixpropertyattributes(ievalueformoneycheck-inlocationcleanlinesscomfortandservice)asindependentvariablesandstarratingasthedependentvariableThisanalysisisbasedontheassumptionthatforagivenpropertyifstarratingstrulyrepresenttheperceivedqualityoftheproperties(aswellastheconsumersrsquosentiment)thenthestarratingsandratingsofthesixotherpropertyattributes

10NHuJZhangandPAPavlouOvercomingtheJ-shapedDistribu-tionofProductReviewsrdquoCommunications of the ACMVol52(2009)pp144-147

low reviews t-value high reviews t-value(Constant) 1618 18353 2400 34186vAlue 0254 13726 0378 8767CheCKiN -0083 -3701 -0053 -1192loCATioN 0104 6132 0079 2117CleAN 0144 7574 0112 2808CoMForT 0267 13993 0129 3088serviCe 0098 4209 0107 2401r2 0331 0338F-statistic 5038 24714

Exhibit 3

regression analysis of ldquolowrdquo and ldquohighrdquo reviews

Notes Coefficients (t-statistics reported) Significance is as follows significant at 0001 level significant at 05 level significant at 010 level Dependent variable is overall rating

10 TheCenterforHospitalityResearchbullCornellUniversity

moderateratingsonalltheattributesbutwasstillwillingtoprovideahighoverallstarratingtothepropertyThisexamplesuggeststhatconsumersplacesignificantemphasisontheoverallexperientialvalueprovidedbythepropertiesThisexpectationofvalueismanytimesintangibleanddoesnotnecessarilydependonthefunctioningorratingoftheindividualfeaturesoftheproperty

RelationshipbetweenRatingsandReviewAttributesTounderstandconsumersrsquoword-of-mouthbehavioritisimportanttoconsidervariousfactorssuchasperceivedexpectationsofcostsandbenefitsimportanceofthepur-chase(involvement)personalcharacteristicsandsituationalinfluencesWeconsidertwofactorsthatareknowntoaffecttheintensityandvalence(positiveornegative)ofwordofmouthpriceandwordcount

PricePriceisanimportantfactorthatcontributestopre-purchaseexpectationsTheintensityofexpectations(andtheirfulfillment)partlydrivenbypriceisdirectlypropor-tionaltoconsumersrsquosatisfactionlevelsandWOMpropensityThismaybethereasonthatthetendencytocomplainisdirectlyrelatedtothecostoftheservice11Themoreexpen-siveanitemthegreaterwillbetheperceivedlossesandthegreaterlikelihoodthatdissatisfactionwillresultinacom-plaintSimilarlythelikelihoodofword-of-mouthbeingtrig-geredbythelevelofexpectationndashdisconfirmationincreases

11BeardenWOandTeelJE(1983)Selecteddeterminantsofcon-sumersatisfactionandcomplaintreportsJournal of Marketing Research 2021-28

willshowahighcorrelationFortheanalysiswetookintoconsiderationthefactmostreviewsarepositive(ie4-and5-pointratings)Thereforewedividedthereviewsintotwocategorieshighreviews(starratingabove3n=2597)andlowreviews(ratingsbeloworequalto3n=600)andranseparateregressionanalysesonthesetwogroupsasshowninExhibit3

AninterestingfindingistheweakrelationshipbetweentheattributeratingsandoverallstarratingforeachpropertyTheR-squareoftheregressionequationwithallsixvariablesandtheoverallstarratingsislessthan40percentThatiseveniftheconsumersprovidedhigherratingsonindividualattributestheiroverallratingwasusuallylowandthere-versewasalsotrueThistrendisobservedinboththehighandlowreviewgroups

TheanalysissuggeststhatinhighreviewsconsumersplacedgreaterimportanceonvalueformoneycleanlinessandcomfortThehighertheratingsofaconsumerontheseattributesthehigheristhelikelihoodthattheconsumerwillprovideahigherstarratingCheck-inlocationandservicedidnotprominentlyfigureintheiroverallevalu-ationsofthepropertieseventhoughtheseattributeswereratedhigh

ThelowreviewsshowsimilarresultsinthesensethattheloweraconsumerratesvaluecleanlinessandcomfortthelowerthelikelihoodofapositiveratingInadditioninthecaseoflowreviewsconsumersplacedemphasisontheserviceprovidedbythepropertystaffandmanagement

TheaboveresultisillustratedinoneoftheldquohighrdquoreviewsshowninExhibit4Thereviewergavelowto

star rating Text Commentary

This was our 1st time renting a house vs a condo This house was great If you wanted to spend a little time by yourself you could so many rooms to go to The house was so comfortable like being at home It was our home for the week Thank you Hope to rent it next year if we havent bought one ourselves

ratings for specific Attributes

value Check-in location Clean Comfort service

3 2 3 2 4 3

Exhibit 4

example of a ldquohighrdquo review

Avg rating for Property Attributes

30

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 11

withthecostoftheproduct12Giventhegreaterperceivedriskinservicescustomerswouldseekmoretrustworthyser-viceproviderswithagoodreputationofsatisfyingcustomerneeds13ThisinturnincreasestheexpectationsandleadstomoreextremereactionsfromtheconsumersHenceonecanspeculatethatthehigherthepriceshigherwillbetheaver-agenegativeorpositiveratingsforaservice

Word countThenumberofwordsinareviewmayserveasaproxyfortheamountofinformationinthereviewsHowevertheevidenceonthisaspectofWOMofalltypesisequivocalForinstanceconsumersgenerallyengageinpositiveWOMtoavoidfeelingsofguiltandas-sociationwithbadnewsandtoreducecognitivedissonanceOntheotherhandacustomerwhoishighlydisappointedwillprovideapersuasiveargumenttomakesurethatothercustomersaredissuadedfromrentingthesamepropertyResearchonconventionalwordofmouthhasofferedampleevidencetosuggestthatdissatisfiedcustomersengageintwotothreetimesasmuchWOMassatisfiedcustomers14Webelievethatthislogicextendstoonlinereviewsdespitethefactthatexistingstudiesareequivocalatbest15Thereforeitcanbeassumedthatreviewswitheitherextremelypositiveorextremelynegativeratingswillhavegreaterwordcountasopposedtomoremoderatereviews

TotesttheseassumptionsweappliedANOVAwithpriceandwordcountasdependentvariablesandthereviewratingasfactorsAsillustratedinExhibits5and6theresultsshowthatpriceandwordcountshowasignificantnegativecorrelationwithstarratingThatisreviewswithlowerratingsaretypicallyassociatedwithgreaterwordcountandrepresentpropertiesthatchargehighrates

TheresultonthewordcountsuggeststhatconsumerswhoareunhappywitharentalpropertytendtowritemoretoexpresstheirdissatisfactionAsshowninthisexhibitlowerratingsareassociatedwithhighernumberofwordsinthereviewsandviceversaSimilarlypropertypriceseemstobestronglyassociatedwithnegativewordofmouthsup-portingtheprinciplethathighercostincreasestheconsum-ersrsquoinvolvementinthepurchase

12BeardenWOandMasonJB(1984)AninvestigationofinfluencesonconsumercomplaintreportsAdvancesinConsumerResearch11490-49513WetzerIMZeelenbergMandPietersR(2007)ldquoNevereatinthatrestaurantIdidrdquoExploringwhypeopleengageinnegativeword-of-mouthcommunicationPsychology and Marketing24661-68014SchlossbergH(1991)CustomersatisfactionNotafadbutawayoflifeMarketing News25andWestbrookR(1987)ProductConsump-tion-basedAffectiveResponsesandPost-purchaseProcessesJournal of Marketing Research 24258-27015AlexMSusskindldquoIToldYouSoRestaurantCustomersrsquoofWord-of-MouthCommunicationPatternsrdquoCornell Hotel and Restaurant Adminis-tration QuarterlyVol43No2(April2002)pp75-85

Exhibit 5

Differences in word count across star ratings

Exhibit 6

Differences in average price per day across ratings

Aver

age

pric

e pe

r day

of t

he re

ntal

pro

pert

y

overall ldquostar ratingrdquo

1504

1563

1804

19861987

Aver

age

wor

d co

unt o

f rev

iew

overall ldquostar ratingrdquo

836837

121

15751543

160

140

120

100

80

200

190

180

170

160

150

1 2 3 4 5

1 2 3 4 5

12 TheCenterforHospitalityResearchbullCornellUniversity

thereviewersThesecanbeclassifiedas(a)attributesorfea-turesofthepropertythatfigureprominentlyandcontributetothepositiveornegativeratings(egroomsbeddishesfurnitureorservice)and(b)thepositiveornegativewordsthatrepresentthesentimentofthereviewer(eggreatamazingdirtyandwonderful)Theresultsofthequalitativeanalysisareasfollows

Syntactic featuresTheoveralltoneofthelowreviewsisratherargumentativeThereviewerstendtobeforcefulinpresentingtheirpointofviewTheylayoutagreatamountofreasoningandprovidespecificexamplestodrivehometheirargumentThisisreflectedinthegreaterpercentageofnegativeconnectorwords(34)suchasbut however as ifandeven thoughItisinterestingtonotethatreviewersinthiscategoryspeakdirectlytotheotherconsumersnotablythroughgreateruseofwordssuchasyou(46)andlooktoinfluencetheirdecisionmaking

InthehighreviewsbycontrasttheoveralltonetendstobeenunciativeandthereviewersusethetexttomerelyreinforcetheirnumericalratingsusingfewerwordsonaveragethanthenegativereviewersInthesereviewstheconsumersfocusontheirpersonalexperiencesandappealtotheaudiencethroughemotionsratherthanlogicandreasoningasevidentfromgreateruseofpositiveadjectives

QualitativeAnalysisoftheReviewTextTheaboveanalysisemphasizestheimportanceofthetextcontentinconsumerreviewsWhilenumericalratingsallowthereviewerstoratetheserviceprovideronvariousattributesitisthetextthatprovidesthemwiththeopportu-nitytoarticulatethenuancesoftheiroverallexperienceandconveyusefulinformationaboutaserviceproviderrsquostransac-tionsandservicecapabilitiesThereforesemanticanalysisofthetextcontentprovidesuswithanopportunitytothethirdquestionpresentedattheoutsetofthisreportwhatdoconsumerstalkaboutintheirreviews

Tothisendweanalyzed600eachofrandomlyselectedhighandlowreviewsusingsemanticprocessingtechniquesWeusedcenteringresonanceanalysis(CRA)amodeofnet-work-basedtextanalysisthatrepresentsthecontentoflargesetsoftextsbyidentifyingthemostimportantwordsthatlinkotherwordsinthenetwork16OuranalysisexaminedsyntacticandsemanticfeaturesSyntacticfeaturesincludethepercentageofthewordsthatareopenclass(ienounsverbsadjectives)Semanticfeaturesrepresentthespecial-izedvocabularythatreflectstheoverallopinionexpressedby

16CormanSandDooleyK(2006)CrawdadTextAnalysisSystem12CrawdadTechnologiesLLCChandlerAZ

Prominently Mentioned Features Freq sample Comments

bedroom beds linen sheets 383

Also the same bedroom and bathroom did not have any door separating the bedroom from the bathroom no privacy

Sleeping arrangements were very awkward The listing sheet described 2 full beds We brought our own linens the beds were queen sized Had to purchase sheet

Kitchen Kitchenware Dishes Dishwasher 297

You must not forget to step up when yoursquore walking out of the kitchen into the dining room otherwise yoursquoll trip over the 3-foot step up

No potholders in the kitchen and it didnrsquot seem like there were as many kitchen supplies as before We just had to make do without

beach beaches beachfront 296

The week was the best vacation we have ever had as a family The beach house was a class act

Disappointment 1 It looks like this house is very close to beach but to get there you have to go around through public path

Pool Poolside 218

The pool was too dirty to use We checked in on Thursday and the pool was not cleaned until Saturday

Also it is inconvenient to have to call to get someone to open the pool Didnrsquot understand why Jacuzzi was open and not the pool

bathroom(s) bath bathtub shower 202

The next problem was in another bathroom (main floor)mdashthe wall heater did not work so showering in here was a bit chilly The next problem was in the ldquomasterrdquo bathroom (connected to the bedroom with the King bed) The sliding shower doors need to be totally replaced

Bathtub drains clogged in three second floor bathrooms

Exhibit 7

Top five prominent issues in low reviews

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 13

suchasgreat fantastic andawesome(45)andpersonalpronounssuchasweandI(54)

Semantic featuresExhibits7and8showthemainissuesoraspectsthatareprominentlydiscussedinthelowreviewsTheorderoftheseissuesisbasedonthestatisti-callysignificantfrequencyoftheiroccurrenceinthe1200reviews

WefurtherinvestigatedtheuseofvarioussentimentwordsInterestinglytheprominentissuesdiscussedinthepositivereviewssometimesdonotevenfigureintheratingscalesprovidedtothereviewersOveralltheresultsofthesemanticanalysiscanbesummarizedasfollows

Thepropertyanditsfeaturesaremorecommonlycon-nectedwithextremelypositivesentimentwordssuchasgreat wonderfulandamazinginthehighreviewsthaninthelowreviewsIncomparisonthepropertiesandtheirfea-turesinthelowreviewsaremorelikelytobecharacterizedwithmoderatelypositivewordssuchasniceandokay

Negativereviewsreflecttheextentofexpectationndashdis-confirmationthattheconsumersfaceduringtheirstayintherentalpropertyThesereviewshaveagreaterproportionofwordsexpressingdisappointment(egnotasexpected)Furtherextremelynegativesentimentwords(egdiscom-fortfilthy)andadjectivesaremorelikelytobeassociatedwithfeaturesinlowreviewsthaninhighreviews

Almost40percentoftheconsumerswhowrotehighreviewsexpresstheirwillingnesstoreturntothepropertyasopposedtoonly9percentofthosewritingnegativereviewsFurthermorethan20percentoftheconsumersgivinghighreviewsexpresstheirstrongdesiretorecommendtheprop-ertytotheirfriendsandfamilymembersasopposedtoonly5percentinlowreviews

ReviewofFindingsOurexploratorystudyofconsumerreviewsofUSvacationhomesonReviewsitecomwashighlightedbyoverwhelm-inglypositivereviews

Theoveralldistributionofthereviewswasheav-ilyskewedtowardsthepositiveratingsMorecriticallythisfindingsuggeststhatthereisalackofmoderateandbalancedopinionsWetheorizethatthisiseitherduetopurchasingbias(sincecustomerswentthroughthetroubleofpurchasingandexperiencingthefacilitiestheytendtobepositivelybiasedtowardstheirpurchase)orunder-reportingbias(customersdonothavestrongenoughincentivestotakethetroubleofreportingtheiropinionsandthefewmotivatedindividualswhoprovidetheirreviewsbecomeaself-selectedsample)

Theoverall(mean)starratingtypicallyusedintravelerreviewsitesisprobablynotthemostaccurateindicatorofcustomersrsquoperceptionsofserviceprovidersrsquoqualityWe

Prominently Mentioned Features Freq sample Commentsbeach beaches beachfront 506 Perfect rental property for those who enjoy the activity of the beach and the serenity

of the Sound The house is just down the road from ocean

The large porch offered a beautiful view of the beach and comfortable new rocking chairs

Pool Poolside 218 The outdoor shower dipping pool and screen in porch were an added bonus Perfect house for a low key weekend

House was great for kids Pool was terrific and the playroom was perfectlocation 192 We had a wonderful time My wife loves fishing and the cottage was in a perfect

location to go fishing on the pier

The home is somewhat dated but the location was great We have vacationed in the Chatham area for many summers and this time it was great too

bedroom beds 186 The family suite is huge with a king size and twin bed and still room for a pack-n-play The kitchen and dining room were huge clean and nice

The only comments that I would say to the negative is the front bed room by the front door always seemed damp which seem to attract more bugs

Kitchen Kitchenware Dishes Dishwasher 178 All of the amenities in the house are upscalemdashloved the kitchen and plasma TVs

The kitchen had good basic utensils dishes and cooking pans

Exhibit 7

Top five prominent issues in high reviews

14 TheCenterforHospitalityResearchbullCornellUniversity

foundacorrelationbetweenthestarratingandratingsonotherpropertyattributesoflessthan40percentinmostcases

AhighwordcountinthereviewsiscorrelatedwithalowoverallratingThatisconsumerswhoareextremelydissatisfiedtendtoexpendmoretimeandenergyontheircritiqueofthepropertyitsamenitiesandservicesPriceofthepropertyalsoseemstopositivelycorrelatewithextremenegativereviewsperhapsbecauseahighpriceheightenstravelersrsquoexpectations

Numericalratingsdonotoftencapturetheactualsentimentandthevarietyofdimensionsonwhichtravel-ersevaluatepropertiesBecausenumericalratingsdonotcapturereviewersrsquoopinionsaboutspecificattributestheycanbedeceivingandmaynotrevealthetruequalityofthepropertyortheactualintentionofthereviewerWefoundforinstancethatconsumersvaluelocation(egclosenesstothebeach)andcleanliness(egswimmingpoolareas)morethansuchfeaturesascheck-inandserviceSimilarlyinthenegativereviewsissuessuchaslackofbedroomparapher-naliaorkitchenitemsseemtoberankedhigherinimpor-tancethanissuessuchaslocationandcheck-inThereforeevenwhenapropertyscoreshighonthecertaindimensionstheoverallratingmayremainlowbecauseotherfactorsovershadowthefavorableaspects

WhatItMeansforManagersWeseeimplicationsforonlinereviewsitesandbyexten-sionforproprietorsofvacationhomesandserviceopera-tionsgenerallyWithregardtothereviewsitesifconsumersconsistentlydetectafavorablebiasthereviewsitesmaylosethecriticalaspectsofcredibilityandtrustAlthoughconsumersaregenerallysmartandrationalandcanworkaroundobviousbiasesrecentevidencesuggeststhatcon-sumersaregraduallydefectingfromsomeonlinesitesduetounbalancedreviews17InthisregardsiteadministratorsaswellaspropertymanagersshouldtakestepstoattractthoseconsumerswhoarewillingtoprovideareasonablybalancedanalysisoftheamenitiesandservicesBythiswemeanreviewsthatpresentcomprehensiveandobjectiveanalysisoftheserviceproviderratherthanjustravesorrants

Tothatendsiteoperatorsshouldtakestepstoun-derstandcustomersrsquoreviewwritingbehaviorandtoat-tractmorebalancedreviewsOnewaytodothisistogivepersonalizedinformationtotheconsumerdemonstratingthecontributionmadebyhisorherreviewandthevalueofthereviewtostakeholdersForinstanceLinget alfoundapositivecorrelationbetweenwillingnesstowritereviewsandtheconsumersrsquoperceptionthattheircontributionswouldbeuniqueandhelpfultoothercustomersaswellasperusedby

17HarteveldtHHJohnsonCStarkEandGeldernKV(2009)Us-ingDigitalChannelsToCalmTheAngryTravelerForresterResearch

theserviceproviders18Simplyputthepropensitytowritereviewsincreaseswhentheconsumersfeelthattheirreviewisactuallyhelpingothers(boththeserviceproviderandotherconsumers)

WesuggestthatreviewsitesdevelopbettermethodstoaggregatesynthesizeandpublishspecificreviewcommentsaswellasnumericalratingsFirstmanagersshouldtakeintoconsiderationthelowcorrelationbetweentheoverallratingsandconsumersrsquoassessmentsofvariouspropertyattributes(aswellastheresultsfromthetextanalysis)Theimplicationofsuchaweakcorrelationisthatreviewsitesneedtoexpandonthelistofpossiblevariablessothattheratingsaccuratelyreflectconsumersrsquoneedsandexpectationsregardingserviceproviders

Furthersitemanagersmayhavetorethinkthecurrentpracticeofaveragingalltheratingsforagivenpropertybe-causesuchameanratingfailstotakeintoaccountthebiasesthatareinherentintheratingsystemsAsanexamplethisstudyfindssignificantunder-reportingbiasinthisonlinereviewsiteAcustomerrsquosdecisiontonotpostonlinereviewsmaymeanthelossofimportantinformationthatcanassisttheusersofthesitestomakemorereliableinferencesThemajorityoftodayrsquosfeedbackmechanismsdonotpubliclydisclosethenumberofsilenttransactions(ietransactionsforwhichnofeedbackwaspostedbycustomers)Suchinfor-mationshouldbecomeapartofaserviceproviderrsquosonlineprofileonratingsitesandotherfeedbackmechanismsItisimportanttoprovidemoreinformationandheuristicstohelptheconsumersnavigatethroughtheclutterandgettheinformationtheywantandseek

Itisimportanttoapplymodernsemanticanalysisandsentimentclassificationtechniquestostudythecontentofthereviewsandidentifyvariousdimensionsonwhichcon-sumersevaluatevacationhomesThedetailsprovidedinthereviewtextareinmanywaysabetterreflectionofcustomersatisfactionServiceproviderscanbetterpositionthemselvesusingthedimensionsuncoveredthroughtextanalysisandtargettheneedsandpreferencesoftheircustomers

TheresultsinthisstudyshowthatnegativereviewsaretypicallyassociatedwithagreaterwordcountthanpositivereviewsItisusefultofocusspecificallyonthesenegativereviewsbecausetheycanrevealpatternsofdeficienciesintheservicestandardsanddeliveryThesereviewsshouldbetreatedthesameascustomercommentcardsorletters19Furthercustomerswhowritesuchlengthynegativereviews

18LingKimberlyGerardBeenenPaulLudfordXiaoxingWangKevinChangXinLiDanCosleyDanFrankowskiLorenTerveenAlMRashidandRobertKrautldquoUsingSocialPsychologytoMotivateContributionstoOnlineCommunitiesrdquoJournal of Computer-Mediated Communication10no4(2005)19MilanR(2007)10thingsyoucandoinresponsetotravelerreviewsinHotelmarketingcom

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 15

aremostlikelytospreadnegativeword-of-mouthaboutthepropertyLengthyreviewsareexcellentsourcesofinforma-tionandshouldbeusedasthebasestounderstandthecomplainingbehavioroftheconsumer

WordcountmaybeusedasanadditionalsegmentationvariabletoidentifythoseconsumerswhoaremoreinvolvedandwillingtoprovidebalancedfeedbacktoboththereviewsiteaswellastopropertymanagementVariousotherresearchmethodssuchastelephoneinterviewsorfocusgroupsshouldbeusedtoelicitdetailedfeedbackfromsuchconsumersTheseactionsalsohelpinservicerecoverypost-purchaseengagementandrelationshipbuilding

OutlookWeforeseethreepotentialareasinvestigationthatcanaddvaluetocurrentbestpracticesWestilldonrsquotknowmuchaboutconsumersrsquomotivationtowriteonlinereviews20Recentstudiesidentifyvariousantecedentsofreviewwritingbutsufferfromsamplingbiassincetheysurveyaself-select-edsampleofconsumerswhoalreadyprovidedreviewsAstheresultsinthisstudyindicateunder-reportingbias(lackofmoderateandlowreviews)remainsprevalentinonlinereviewsitesFutureresearchshouldidentifysocio-psycho-logicalfactorsthatincreaseconsumersrsquopropensitytowriteonlinereviews

Futureresearchshouldexploretheimpactofthepres-ence(orlack)ofextremelynegativereviewswithinalargecorpusofpositivereviewsTheevidenceontheimpactofre-

20WangYandFesenmaierDR(2003)AssessingMotivationofCon-tributioninOnlineCommunitiesAnEmpiricalInvestigationofanOn-lineTravelCommunityElectronic Markets1333-45andYooKyungHandUlrikeGretzelldquoWhatMotivatesConsumerstoWriteOnlineTravelReviewsrdquoInformation Technology amp Tourism10No4(2008)283-95

viewvalence(negativeorpositive)isequivocalProponentsofconfirmatorybiassuggestthatconsumerslookforaffir-mativeevidencesupportingaproductchoicealreadymadeIfthatisthecasepositivereviewsaremorelikelytohaveagreatereffectonconsumeractionsOntheotherhandthenotionofnegativitybiassuggeststhatwhenconsumersareneutralnegativereviewstendtobecomemoresalientthanpositivereviews21Inthepresenceoflargenumbersofposi-tivereviewsconsumersspecificallyseeknegativereviewsthattheyfeelwillhelpthemidentifyspecificproblemswithaserviceUnderstandinghowconsumersreconciletheinformationprovidedbybothnegativeandpositivereviewshasimplicationsforthedesignandmanagementofreviewsites

Finallyareviewrsquoswordcounthasimportantimplica-tionsforconsumersrsquotrustinthereviewsWhenconsumersarewillingtoreadandcompareopen-endedcommentsfromotherconsumerstheamountofinformationinareviewcanmatterWOMforhighinvolvementproductsismorepersuasiveandincreasesthedecisionmakerrsquosconfi-dencewhenthemessagesenderprovidesgreaternumberofreasonsfortheiroverallratingLongerreviewsprovidemoreinformationandarelikelytobeperceivedasmorehelpfulandpersuasivethanshorterreviewsThewaysinwhichareviewiswrittenframedandpresentedisaninterestingfactorwhichhasrarelybeeninvestigatedandhasimportantimplicationsparticularlygivenourfindingthattheamountofinformationinareview(wordcount)variessignificantlywiththeratingsofthereviewn

21BaSandPavlouPA(2002)EvidenceoftheEffectofTrustBuildingTechnologyinElectronicMarketsPricePremiumsandBuyerBehaviorMIS Quarterly26243-268

Cornell Hospitality Quarterlyhttpcqxsagepubcom

2011 ReportsVol11No15DesigningaSelf-HealingServiceSystemAnIntegrativeModelbyRobertCFordandMichaelCSturman

Vol11No14ReversingtheGreenBacklashWhyLargeHospitalityCompaniesShouldWelcomeCrediblyGreenCompetitorsbyMichaelGiebelhausenPhDandHaeEunHelenChunPhD

Vol11No13DevelopingaSustainabilityMeasurementFrameworkforHotelsTowardanIndustry-wideReportingStructurebyEricRicaurte

Vol11No12CreatingValueforWomenBusinessTravelersFocusingonEmotionalOutcomesbyJudiBrownellPhD

Vol11No11CustomerLoyaltyANewLookattheBenefitsofImprovingSegmentationEffortswithRewardsProgramsbyClayVoorheesPhDMichaelMcCallPhDandRogerCalantonePhD

Vol11No10CustomerPerceptionsofElectronicFoodOrderingbySherylEKimes

Vol11No92011TravelIndustryBenchmarkingStatusofSeniorDestinationandLodgingMarketingExecutivesbyRohitVermaPhDandKenMcGill

Vol11No8SearchOTAsandOnlineBookingAnExpandedAnalysisoftheBillboardEffectbyChrisAndersonPhD

Vol11No7OnlineMobileandTextFoodOrderingintheUSRestaurantIndustrybySherylEKimesPhDandPhilippFLaqueacute

Vol11No6HotelGuestsrsquoReactionstoGuestRoomSustainabilityInitiativesbyAlexSusskindPhDandRohitVermaPhD

Vol11No5TheImpactofTerrorismandEconomicShocksonUSHotelsbyCathyAEnzRenaacutetaKosovaacuteandMarkLomannoVol11No4ImplementingHumanResourceInnovationsThreeSuccessStoriesfromtheServiceIndustrybyJustinSunandKateWalshPhD

Vol11No3Compendium2011

Vol11No2PositioningaPlaceDevelopingaCompellingDestinationBrandbyRobertJKwortnikPhDandEthanHawkesMBA

Vol11No1TheImpactofHealthInsuranceonEmployeeJobAnxietyWithdrawalBehaviorsandTaskPerformancebySeanWayPhDBillCarrollPhDAlexSusskindPhDandJoeCYLeng

2011 Hospitality ToolsVol2No1TheGameHasChangedANewParadigmforStakeholderEngagementbyMaryBethMcEuen

2011 Industry PerspectivesNo7MegaTips2TwentyTestedTechniquesforIncreasingYourTipsbyMichaelLynn

2011 ProceedingsVol3No5BuildingBrandsintheInternetAgeAnalyticsLoyaltyandCommunicationbyGlennWithiam

Vol3No4BraveNewWorldOnlineHotelDistributionbyGlennWithiam

Vol3No3SocialMediaandtheHospitalityIndustryHoldingtheTigerbytheTailbyGlennWithiam

Vol3No2TheChallengeofHotelandRestaurantSustainabilityFindingProfitinldquoBeingGreenrdquobyGlennWithiam

Vol3No1CautiousOptimismCHRSExaminesHospitalityIndustryTrendsbyGlennWithiam

2010 ReportsVol10No18HowTravelersUseOnlineandSocialMediaChannelstoMakeHotel-choiceDecisionsbyLauraMcCarthyDebraStockandRohitVermaPhD

Vol10No17PublicorPrivateTheHospitalityInvestmentDecisionbyQingzhongMaPhDandAthenaWeiZhangPhD

Vol10No16BestPracticesinSearchEngineMarketingandOptimizationTheCaseoftheStJamesHotelbyGregBodenlcosVictorBogertDanGordonCarterHearneandChrisKAndersonPhD

Vol10No15TheImpactofPrix Fixe MenuPriceFormatsonGuestsrsquoDealPerceptionbyShuoWangandMichaelLynnPhD

Cornell Center for Hospitality Research

Publication Indexwwwchrcornelledu

The Executive Path Hospitality Leadership Through Learning

Complete program information and applications available online

wwwhotelschoolcornelleduexecedPhone + 1 607 255 4919 Email exec_ed_hotelcornelledu

Professionals from around the world are invited to attend 3-day 10-day or online courses at the worldrsquos leading institute for hospitality management education in

Visit our website to apply

Explore develop and apply ideas with global hospitality leaders and

expert Cornell professors

Success

AdvancingBusiness

andPersonal

bull Strategic Leadershipbull Financebull Foodservicebull Human Resources

bull Marketingbull Operationsbull Real Estate

No2RetaliationWhyanIncreaseinClaimsDoesNotMeantheSkyIsFallingbyDavidSherwynJDandGreggGilmanJD

2009 ToolsToolNo12MeasuringtheDiningExperienceTheCaseofVitaNovabyKeshPrasadandFredJDeMiccoPhD

Cornell Center for Hospitality ResearchPublication Index

The Executive Path Hospitality Leadership Through Learning

Complete program information and applications available online

wwwhotelschoolcornelleduexecedPhone + 1 607 255 4919 Email exec_ed_hotelcornelledu

Professionals from around the world are invited to attend 3-day 10-day or online courses at the worldrsquos leading institute for hospitality management education in

Visit our website to apply

Explore develop and apply ideas with global hospitality leaders and

expert Cornell professors

Success

AdvancingBusiness

andPersonal

bull Strategic Leadershipbull Financebull Foodservicebull Human Resources

bull Marketingbull Operationsbull Real Estate

visit our website to apply

wwwchrcornell edu

Page 7: Unscrambling the Puzzling Matter of Online Consumer ... · Philips Hospitality Americas Chris Proulx, CEO, eCornell & Executive Education Carolyn D. Richmond, Partner, Hospitality

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 7

Howevertheexistingunderstandingofonlinetourism-relatedreviewsisrelativelyweakeventhoughweknowthatwordofmouth(WOM)ismoreimportantandinfluentialwithinaservicescontextthaninproductmarketingscenariosRecentlysomestudieshaveexploredtourism-andhospital-ity-relatedonlineWOM3Yetresearchhassofaroverlookedcertaininterestingaspectsofonlinereviewsasfollows

Onlinereviewstypicallyuseldquooverallstarratingsrdquoaspri-maryindicatorsofperceivedservicequalitymdashratingsthatarenotrelatedinanywaytotheformalstardesignationsawardedbyForbesorMichelinConsumersrsquomeanstarratingsaretypi-callyaggregatedfromtheratingsgivenbyindividualconsum-ersineachreviewWhatwedonrsquotknowishowconsumersdeterminethestarstheyassigntoaservicewhentheypostareviewonawebsiteThatisdostarratingsaccuratelyreflectconsumersrsquosentimentsregardingvariousattributesofserviceprovidersThisisavitalissuegiventheexperientialandin-tangiblenatureoftravelservicesandthefactthatconsumersusetheoverallratingasanimportantheuristicwhilenarrow-ingdownchoices

FurtherstudieshaveoverlookedtherelationshipbetweenthereviewsandvariousattributesofservicesandserviceprovidersSinceonlinereviewsareaformofWOMvariablessuchasconsumerinvolvementpricingandlengthofstaymayhaveanimpactonthefinalratingandpostingofthereviewprovidedbyconsumersItisimportanttouncovertheserelationshipstobetterunderstandthedynamicsofonlinereviews

FewstudieshaveanalyzedthequalitativeaspectofonlinereviewsOnlinereviewsareessentiallyopen-endedtext-basedconsumer-to-consumer(c2c)communicationsThetextcontentmaycontainnuancedviewsoftheservicesandserviceproviders(fromthereviewwritersrsquopointofview)thatcannotbeexpressedusingcrudenumericalratings4Hencethereisaneedtoanalyzethetextportionofthereviewsandidentifytheissuesthatconsumersaremostlyconcernedwithwhenwritingandpostingonlinerelatedreviews

WeaddresstheabovepointsinthisstudyThefollowingquestionsguidedus

3KHYooandUGretzelldquoWhatMotivatesConsumerstoWriteOnlineTravelReviewsrdquoInformation Technology amp TourismVol10(2008)pp283-295andZXiangandUGretzelldquoRoleofSocialMediainOnlineTravelInformationSearchrdquoTourism Management20094AGhoseandPGIpeirotisldquoEstimatingtheSocio-economicImpactofProductReviewsMiningTextandReviewerCharacteristicsrdquoInformation Systems Research2008

1 Dooverallratingsreflecttheperceivedqualityoftheserviceproviders(asderivedfromtheconsumersrsquoexpressedsentimentinthereviews)Whataretheunderlyingpatternsoftheoverallratingsinonlinereviews

2 Whatarethelinkagesamongthevariousattributesoftheserviceprovidersandcustomerreviewsand

3 WhataretheissuesthatconsumersmostlytalkaboutinthetextportionofonlinereviewsDotheseis-suesdifferfromtheratingsthattheconsumershaveprovidedviathestandardvariablesprovidedbythereviewsites

ResearchSettingandDataCollectionThisstudyisbasedonreviewscollectedfromldquoReviewsitecomrdquoaswehavecalledthecommunity-driventravelsitethatweanalyzedThesiteprovidesreviewsofverifiedrentalpropertiesandorganizesthereputationofvacationpropertiesThepropertylistingwithinthesiteisbasednotontherevenuesharingmodelthatisgenerallypracticedbyotherreviewsitesbutonthereputationtrustandfeedbackfromcustomerswhohavestayedatthevacationrentalpropertiesReviewsitecomisafreeservicethatoperatesonserviceproviderfeesandadvertisingrevenuesUn-likesometravelreviewsitesthefirmrsquosreviewsolicitationstrategyistosendpersonalizedinvitationsonlytothosecustomerswhosestayhasbeenverifiedbythepropertymanagementThesitersquosmanagementbelievesthatthisstrategytoalargeextenteliminatesfakereviews

Thefirmprovideduswith3300reviewsforanalysisAfterinitialdatacleanupabout100reviewswithincom-pleteorunusableinformationwereeliminatedand3197reviewswereusedfortheanalysisThereviewscontainedthefollowingattributesbull Theoverallratingassignedtotheserviceprovider

basedontheaggregationoftheratingsassignedbyindividualreviewers

bull Consumersrsquoratingsonthefollowingsixattributesvaluecleanlinesscomfortservicelocationandcheck-in

bull Thereviewtext(Semanticprocessingtechniqueswereappliedontheaggregatereviewtexttoidentifythenuancedopinionsandconcernsofthetravelerswhowrotereviews)

8 TheCenterforHospitalityResearchbullCornellUniversity

bull Agerangeoftherevieweronan1-4ordinalscale(25-3435-4445-54and55andabove)

bull Averagepriceperdayand

bull Wordcount(usedinthequantitativeanalysis)

PreliminaryHypothesesandResultsTheagegroupsofthereviewwritersareshowninExhibit1Alargepercentageofthecustomerswhoreviewedthepropertiesfallunder44yearsofageThisisconsistentwithpreviousstudiesinthisrealmwhichshowthatgenerationXandGenerationYaremoreactiveintheWOMforums

ReviewRatingDistribution Websitesthathostonlinereviewsgiveconsumerstheoptiontoassignaratingtotheserviceproviderstypicallyonascaleof1to5or1to7Manysitesaggregatethesenumeri-calorstarratingstodetermineanoverall(mean)starratingWeseethisoverallratingasanimportantelementinonlineWOMsinceconsumersoftenusethisratingtonarrowdowntheirconsiderationsetStudiesthathaveusedtheratingastheprimarypredictoroftheserviceprovidersrsquoperceivedqualityorcustomersatisfactiontodeterminetheimpactofonlinereviewsonproductsaleshavearrivedatmixedfind-ings5Howeverthesestudiesarebasedonthefundamentalassumptionthatthenumericalratingsassignedtoservice

5ForexampleseeJChevalierandDMayzlinldquoTheEffectofWordofMouthOnlineOnlineBookReviewsrdquoJournal of Marketing ResearchVol43(2006)pp345-354andWDuanBGuandABWhinstonldquoDoOnlineReviewsMattermdashAnEmpiricalInvestigationofPanelDatardquoDecision Support SystemsVol45(2008)pp1007-1016

providersareanaccurateestimateofthequalityperceivedbythecustomerThispracticefollowsthebasicassumptionofbehavioraltheoriststhatanydatasetwithalargenumberofconsumerresponsestendstofollowanormal(Gaussian)distributionRecentstudieshavequestionedthisassump-tionForinstanceDellaracosandNarayanrsquosstudyonmoviereviewsinYahoocomfoundthatonlytheextremelysatisfiedandextremelydissatisfiedcustomersarelikelytopubliclyexpresstheiropinionsascomparedtocustomerswithmod-erateopinions6SimilarlystudiesusingdatafromTripadvi-sorcomandAmazoncomfoundthatthestarratingsexhibitatruncateddistributioninwhichthemajorityofreviewsarepositive7

TotestthisassumptionweranasimplefrequencychartasshowninExhibit2Inkeepingwithotherstudiesthedistributionisheavilyskewedtowardsthepositiveratings1-pointreviews32(1)2-pointreviews95(3)3-pointreviews458(14)4-pointreviews688(22)and5-pointreviews1912(60)Totestfortherobustnessofthisdistri-butionweappliedtheKolmogorov-Smirnovtestofnormal-ityTheresultsconfirmthenon-normaldistributionofthestarratingsandtheheavyskewtowardspositivereviews

Threeexplanationsarepossibleforsuchadistribution

6CDellarocasandRNarayanldquoAStatisticalMeasureofaPopulationrsquosPropensitytoEngageinPost-purchaseOnlineWord-of-mouthrdquoStatisti-cal ScienceVol21(2006)pp2777ATalwarandRFBoiJurcaUnderstandingUserBehaviorinOnlineFeedbackReportingrdquoinElectronic CommerceSanDiegoCalifornia(2007)pp134-142

Exhibit 1

Distribution of age ranges

Exhibit 2

Distribution of overall ldquostar ratingsrdquo

55 and above 45ndash54

years

35ndash44 years

18ndash34 years

overall ldquostar ratingrdquo

Num

ber o

f rev

iew

s

3295

458 688

1912

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 9

1 AmajorityofthepropertiesthatarelistedonthewebsitehaveoutstandingperceivedqualityandprovidesatisfactoryservicestocustomersWhilethisistheo-reticallypossibleitseemsunlikelywhenappliedtoallthepropertieslistedonthewebsite8

2 ThereviewsdonotconstituteatrulyrepresentativesampleofconsumeropinionBasedonotherstud-iesthisexplanationseemsplausibleConsumerswithmoderateviewsmayseelittleincentivetoreporttheirviewsinanonlineforumandasaconsequenceonlythoseconsumerswhohavepolarizedviewsarelikelytoposttheircomments

3 ThedistributioncouldalsobeanartifactofthefirmrsquosreviewsolicitingstrategyRecentstudiessupportthisideabysuggestingthatcustomersengageinsociallydesirablebehavioreveninimpersonalinteractionswiththeserviceproviders9Inthiscaseitmaybethattheywillprovideareasonablypositiverevieweveniftheexperiencehasnotbeenuptothemark

StarRatingsandPerceivedServiceQualityLetrsquosreturntothefirstquestionweaskedattheoutsetArestarratingsatruerepresentationofserviceprovid-ersrsquoperceivedqualityasreflectedbytheconsumersrsquostatedsentimentsThisbecomesanimportantquestionfortwo

8AKadetldquoRah-RahRatingsOnlinerdquoSmartMoney20079YGreacutegoireandRJFisherldquoCustomerBetrayalandRetaliationWhenYourBestCustomersbecomeYourWorstEnemiesrdquoJournal of the Acad-emy of Marketing ScienceVol36(2008)pp247-261

reasons(a) ratingsaretheprimaryevidencethatconsum-ersconsiderwhileshoppingonlineforinformationand(b)thetruncatednatureofratingdistributionsmeansthatthereviewsavailablemaynottrulyrepresentconsumeropinionortheserviceprovidersrsquoperceivedqualityIfonlythosewithextremeviewspostreviewsthenconsumersshouldalwayslookforserviceprovidersthathavea5-starratingOnthecontraryifthereviewsaccuratelyreflectallconsumersrsquoviewsthenconsumersarebetteroffsearchingforreviewswitha3ratingsincethesetendtobemorebalancedandenablebetterproductanalysisRecentlyHu et aluseddatafromAmazoncomshowthatnumericalscoresdonotreflectbooksrsquotrueperceivedquality10AstheysuggestldquohellipratherthescorereflectsthebalanceofdiverseopinionsInotherwordswhenabookrsquosoverallscoreisaround3itdoesnotsuggestthatconsumersgenerallyagreethatthisisanaver-agebookItrathersuggeststhatroughlyequalnumberofconsumersthinkthatthebookiseitheranoutstandingbookoranabysmalbookrdquo

Totestthisassertionweappliedaregressionequationwithsixpropertyattributes(ievalueformoneycheck-inlocationcleanlinesscomfortandservice)asindependentvariablesandstarratingasthedependentvariableThisanalysisisbasedontheassumptionthatforagivenpropertyifstarratingstrulyrepresenttheperceivedqualityoftheproperties(aswellastheconsumersrsquosentiment)thenthestarratingsandratingsofthesixotherpropertyattributes

10NHuJZhangandPAPavlouOvercomingtheJ-shapedDistribu-tionofProductReviewsrdquoCommunications of the ACMVol52(2009)pp144-147

low reviews t-value high reviews t-value(Constant) 1618 18353 2400 34186vAlue 0254 13726 0378 8767CheCKiN -0083 -3701 -0053 -1192loCATioN 0104 6132 0079 2117CleAN 0144 7574 0112 2808CoMForT 0267 13993 0129 3088serviCe 0098 4209 0107 2401r2 0331 0338F-statistic 5038 24714

Exhibit 3

regression analysis of ldquolowrdquo and ldquohighrdquo reviews

Notes Coefficients (t-statistics reported) Significance is as follows significant at 0001 level significant at 05 level significant at 010 level Dependent variable is overall rating

10 TheCenterforHospitalityResearchbullCornellUniversity

moderateratingsonalltheattributesbutwasstillwillingtoprovideahighoverallstarratingtothepropertyThisexamplesuggeststhatconsumersplacesignificantemphasisontheoverallexperientialvalueprovidedbythepropertiesThisexpectationofvalueismanytimesintangibleanddoesnotnecessarilydependonthefunctioningorratingoftheindividualfeaturesoftheproperty

RelationshipbetweenRatingsandReviewAttributesTounderstandconsumersrsquoword-of-mouthbehavioritisimportanttoconsidervariousfactorssuchasperceivedexpectationsofcostsandbenefitsimportanceofthepur-chase(involvement)personalcharacteristicsandsituationalinfluencesWeconsidertwofactorsthatareknowntoaffecttheintensityandvalence(positiveornegative)ofwordofmouthpriceandwordcount

PricePriceisanimportantfactorthatcontributestopre-purchaseexpectationsTheintensityofexpectations(andtheirfulfillment)partlydrivenbypriceisdirectlypropor-tionaltoconsumersrsquosatisfactionlevelsandWOMpropensityThismaybethereasonthatthetendencytocomplainisdirectlyrelatedtothecostoftheservice11Themoreexpen-siveanitemthegreaterwillbetheperceivedlossesandthegreaterlikelihoodthatdissatisfactionwillresultinacom-plaintSimilarlythelikelihoodofword-of-mouthbeingtrig-geredbythelevelofexpectationndashdisconfirmationincreases

11BeardenWOandTeelJE(1983)Selecteddeterminantsofcon-sumersatisfactionandcomplaintreportsJournal of Marketing Research 2021-28

willshowahighcorrelationFortheanalysiswetookintoconsiderationthefactmostreviewsarepositive(ie4-and5-pointratings)Thereforewedividedthereviewsintotwocategorieshighreviews(starratingabove3n=2597)andlowreviews(ratingsbeloworequalto3n=600)andranseparateregressionanalysesonthesetwogroupsasshowninExhibit3

AninterestingfindingistheweakrelationshipbetweentheattributeratingsandoverallstarratingforeachpropertyTheR-squareoftheregressionequationwithallsixvariablesandtheoverallstarratingsislessthan40percentThatiseveniftheconsumersprovidedhigherratingsonindividualattributestheiroverallratingwasusuallylowandthere-versewasalsotrueThistrendisobservedinboththehighandlowreviewgroups

TheanalysissuggeststhatinhighreviewsconsumersplacedgreaterimportanceonvalueformoneycleanlinessandcomfortThehighertheratingsofaconsumerontheseattributesthehigheristhelikelihoodthattheconsumerwillprovideahigherstarratingCheck-inlocationandservicedidnotprominentlyfigureintheiroverallevalu-ationsofthepropertieseventhoughtheseattributeswereratedhigh

ThelowreviewsshowsimilarresultsinthesensethattheloweraconsumerratesvaluecleanlinessandcomfortthelowerthelikelihoodofapositiveratingInadditioninthecaseoflowreviewsconsumersplacedemphasisontheserviceprovidedbythepropertystaffandmanagement

TheaboveresultisillustratedinoneoftheldquohighrdquoreviewsshowninExhibit4Thereviewergavelowto

star rating Text Commentary

This was our 1st time renting a house vs a condo This house was great If you wanted to spend a little time by yourself you could so many rooms to go to The house was so comfortable like being at home It was our home for the week Thank you Hope to rent it next year if we havent bought one ourselves

ratings for specific Attributes

value Check-in location Clean Comfort service

3 2 3 2 4 3

Exhibit 4

example of a ldquohighrdquo review

Avg rating for Property Attributes

30

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 11

withthecostoftheproduct12Giventhegreaterperceivedriskinservicescustomerswouldseekmoretrustworthyser-viceproviderswithagoodreputationofsatisfyingcustomerneeds13ThisinturnincreasestheexpectationsandleadstomoreextremereactionsfromtheconsumersHenceonecanspeculatethatthehigherthepriceshigherwillbetheaver-agenegativeorpositiveratingsforaservice

Word countThenumberofwordsinareviewmayserveasaproxyfortheamountofinformationinthereviewsHowevertheevidenceonthisaspectofWOMofalltypesisequivocalForinstanceconsumersgenerallyengageinpositiveWOMtoavoidfeelingsofguiltandas-sociationwithbadnewsandtoreducecognitivedissonanceOntheotherhandacustomerwhoishighlydisappointedwillprovideapersuasiveargumenttomakesurethatothercustomersaredissuadedfromrentingthesamepropertyResearchonconventionalwordofmouthhasofferedampleevidencetosuggestthatdissatisfiedcustomersengageintwotothreetimesasmuchWOMassatisfiedcustomers14Webelievethatthislogicextendstoonlinereviewsdespitethefactthatexistingstudiesareequivocalatbest15Thereforeitcanbeassumedthatreviewswitheitherextremelypositiveorextremelynegativeratingswillhavegreaterwordcountasopposedtomoremoderatereviews

TotesttheseassumptionsweappliedANOVAwithpriceandwordcountasdependentvariablesandthereviewratingasfactorsAsillustratedinExhibits5and6theresultsshowthatpriceandwordcountshowasignificantnegativecorrelationwithstarratingThatisreviewswithlowerratingsaretypicallyassociatedwithgreaterwordcountandrepresentpropertiesthatchargehighrates

TheresultonthewordcountsuggeststhatconsumerswhoareunhappywitharentalpropertytendtowritemoretoexpresstheirdissatisfactionAsshowninthisexhibitlowerratingsareassociatedwithhighernumberofwordsinthereviewsandviceversaSimilarlypropertypriceseemstobestronglyassociatedwithnegativewordofmouthsup-portingtheprinciplethathighercostincreasestheconsum-ersrsquoinvolvementinthepurchase

12BeardenWOandMasonJB(1984)AninvestigationofinfluencesonconsumercomplaintreportsAdvancesinConsumerResearch11490-49513WetzerIMZeelenbergMandPietersR(2007)ldquoNevereatinthatrestaurantIdidrdquoExploringwhypeopleengageinnegativeword-of-mouthcommunicationPsychology and Marketing24661-68014SchlossbergH(1991)CustomersatisfactionNotafadbutawayoflifeMarketing News25andWestbrookR(1987)ProductConsump-tion-basedAffectiveResponsesandPost-purchaseProcessesJournal of Marketing Research 24258-27015AlexMSusskindldquoIToldYouSoRestaurantCustomersrsquoofWord-of-MouthCommunicationPatternsrdquoCornell Hotel and Restaurant Adminis-tration QuarterlyVol43No2(April2002)pp75-85

Exhibit 5

Differences in word count across star ratings

Exhibit 6

Differences in average price per day across ratings

Aver

age

pric

e pe

r day

of t

he re

ntal

pro

pert

y

overall ldquostar ratingrdquo

1504

1563

1804

19861987

Aver

age

wor

d co

unt o

f rev

iew

overall ldquostar ratingrdquo

836837

121

15751543

160

140

120

100

80

200

190

180

170

160

150

1 2 3 4 5

1 2 3 4 5

12 TheCenterforHospitalityResearchbullCornellUniversity

thereviewersThesecanbeclassifiedas(a)attributesorfea-turesofthepropertythatfigureprominentlyandcontributetothepositiveornegativeratings(egroomsbeddishesfurnitureorservice)and(b)thepositiveornegativewordsthatrepresentthesentimentofthereviewer(eggreatamazingdirtyandwonderful)Theresultsofthequalitativeanalysisareasfollows

Syntactic featuresTheoveralltoneofthelowreviewsisratherargumentativeThereviewerstendtobeforcefulinpresentingtheirpointofviewTheylayoutagreatamountofreasoningandprovidespecificexamplestodrivehometheirargumentThisisreflectedinthegreaterpercentageofnegativeconnectorwords(34)suchasbut however as ifandeven thoughItisinterestingtonotethatreviewersinthiscategoryspeakdirectlytotheotherconsumersnotablythroughgreateruseofwordssuchasyou(46)andlooktoinfluencetheirdecisionmaking

InthehighreviewsbycontrasttheoveralltonetendstobeenunciativeandthereviewersusethetexttomerelyreinforcetheirnumericalratingsusingfewerwordsonaveragethanthenegativereviewersInthesereviewstheconsumersfocusontheirpersonalexperiencesandappealtotheaudiencethroughemotionsratherthanlogicandreasoningasevidentfromgreateruseofpositiveadjectives

QualitativeAnalysisoftheReviewTextTheaboveanalysisemphasizestheimportanceofthetextcontentinconsumerreviewsWhilenumericalratingsallowthereviewerstoratetheserviceprovideronvariousattributesitisthetextthatprovidesthemwiththeopportu-nitytoarticulatethenuancesoftheiroverallexperienceandconveyusefulinformationaboutaserviceproviderrsquostransac-tionsandservicecapabilitiesThereforesemanticanalysisofthetextcontentprovidesuswithanopportunitytothethirdquestionpresentedattheoutsetofthisreportwhatdoconsumerstalkaboutintheirreviews

Tothisendweanalyzed600eachofrandomlyselectedhighandlowreviewsusingsemanticprocessingtechniquesWeusedcenteringresonanceanalysis(CRA)amodeofnet-work-basedtextanalysisthatrepresentsthecontentoflargesetsoftextsbyidentifyingthemostimportantwordsthatlinkotherwordsinthenetwork16OuranalysisexaminedsyntacticandsemanticfeaturesSyntacticfeaturesincludethepercentageofthewordsthatareopenclass(ienounsverbsadjectives)Semanticfeaturesrepresentthespecial-izedvocabularythatreflectstheoverallopinionexpressedby

16CormanSandDooleyK(2006)CrawdadTextAnalysisSystem12CrawdadTechnologiesLLCChandlerAZ

Prominently Mentioned Features Freq sample Comments

bedroom beds linen sheets 383

Also the same bedroom and bathroom did not have any door separating the bedroom from the bathroom no privacy

Sleeping arrangements were very awkward The listing sheet described 2 full beds We brought our own linens the beds were queen sized Had to purchase sheet

Kitchen Kitchenware Dishes Dishwasher 297

You must not forget to step up when yoursquore walking out of the kitchen into the dining room otherwise yoursquoll trip over the 3-foot step up

No potholders in the kitchen and it didnrsquot seem like there were as many kitchen supplies as before We just had to make do without

beach beaches beachfront 296

The week was the best vacation we have ever had as a family The beach house was a class act

Disappointment 1 It looks like this house is very close to beach but to get there you have to go around through public path

Pool Poolside 218

The pool was too dirty to use We checked in on Thursday and the pool was not cleaned until Saturday

Also it is inconvenient to have to call to get someone to open the pool Didnrsquot understand why Jacuzzi was open and not the pool

bathroom(s) bath bathtub shower 202

The next problem was in another bathroom (main floor)mdashthe wall heater did not work so showering in here was a bit chilly The next problem was in the ldquomasterrdquo bathroom (connected to the bedroom with the King bed) The sliding shower doors need to be totally replaced

Bathtub drains clogged in three second floor bathrooms

Exhibit 7

Top five prominent issues in low reviews

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 13

suchasgreat fantastic andawesome(45)andpersonalpronounssuchasweandI(54)

Semantic featuresExhibits7and8showthemainissuesoraspectsthatareprominentlydiscussedinthelowreviewsTheorderoftheseissuesisbasedonthestatisti-callysignificantfrequencyoftheiroccurrenceinthe1200reviews

WefurtherinvestigatedtheuseofvarioussentimentwordsInterestinglytheprominentissuesdiscussedinthepositivereviewssometimesdonotevenfigureintheratingscalesprovidedtothereviewersOveralltheresultsofthesemanticanalysiscanbesummarizedasfollows

Thepropertyanditsfeaturesaremorecommonlycon-nectedwithextremelypositivesentimentwordssuchasgreat wonderfulandamazinginthehighreviewsthaninthelowreviewsIncomparisonthepropertiesandtheirfea-turesinthelowreviewsaremorelikelytobecharacterizedwithmoderatelypositivewordssuchasniceandokay

Negativereviewsreflecttheextentofexpectationndashdis-confirmationthattheconsumersfaceduringtheirstayintherentalpropertyThesereviewshaveagreaterproportionofwordsexpressingdisappointment(egnotasexpected)Furtherextremelynegativesentimentwords(egdiscom-fortfilthy)andadjectivesaremorelikelytobeassociatedwithfeaturesinlowreviewsthaninhighreviews

Almost40percentoftheconsumerswhowrotehighreviewsexpresstheirwillingnesstoreturntothepropertyasopposedtoonly9percentofthosewritingnegativereviewsFurthermorethan20percentoftheconsumersgivinghighreviewsexpresstheirstrongdesiretorecommendtheprop-ertytotheirfriendsandfamilymembersasopposedtoonly5percentinlowreviews

ReviewofFindingsOurexploratorystudyofconsumerreviewsofUSvacationhomesonReviewsitecomwashighlightedbyoverwhelm-inglypositivereviews

Theoveralldistributionofthereviewswasheav-ilyskewedtowardsthepositiveratingsMorecriticallythisfindingsuggeststhatthereisalackofmoderateandbalancedopinionsWetheorizethatthisiseitherduetopurchasingbias(sincecustomerswentthroughthetroubleofpurchasingandexperiencingthefacilitiestheytendtobepositivelybiasedtowardstheirpurchase)orunder-reportingbias(customersdonothavestrongenoughincentivestotakethetroubleofreportingtheiropinionsandthefewmotivatedindividualswhoprovidetheirreviewsbecomeaself-selectedsample)

Theoverall(mean)starratingtypicallyusedintravelerreviewsitesisprobablynotthemostaccurateindicatorofcustomersrsquoperceptionsofserviceprovidersrsquoqualityWe

Prominently Mentioned Features Freq sample Commentsbeach beaches beachfront 506 Perfect rental property for those who enjoy the activity of the beach and the serenity

of the Sound The house is just down the road from ocean

The large porch offered a beautiful view of the beach and comfortable new rocking chairs

Pool Poolside 218 The outdoor shower dipping pool and screen in porch were an added bonus Perfect house for a low key weekend

House was great for kids Pool was terrific and the playroom was perfectlocation 192 We had a wonderful time My wife loves fishing and the cottage was in a perfect

location to go fishing on the pier

The home is somewhat dated but the location was great We have vacationed in the Chatham area for many summers and this time it was great too

bedroom beds 186 The family suite is huge with a king size and twin bed and still room for a pack-n-play The kitchen and dining room were huge clean and nice

The only comments that I would say to the negative is the front bed room by the front door always seemed damp which seem to attract more bugs

Kitchen Kitchenware Dishes Dishwasher 178 All of the amenities in the house are upscalemdashloved the kitchen and plasma TVs

The kitchen had good basic utensils dishes and cooking pans

Exhibit 7

Top five prominent issues in high reviews

14 TheCenterforHospitalityResearchbullCornellUniversity

foundacorrelationbetweenthestarratingandratingsonotherpropertyattributesoflessthan40percentinmostcases

AhighwordcountinthereviewsiscorrelatedwithalowoverallratingThatisconsumerswhoareextremelydissatisfiedtendtoexpendmoretimeandenergyontheircritiqueofthepropertyitsamenitiesandservicesPriceofthepropertyalsoseemstopositivelycorrelatewithextremenegativereviewsperhapsbecauseahighpriceheightenstravelersrsquoexpectations

Numericalratingsdonotoftencapturetheactualsentimentandthevarietyofdimensionsonwhichtravel-ersevaluatepropertiesBecausenumericalratingsdonotcapturereviewersrsquoopinionsaboutspecificattributestheycanbedeceivingandmaynotrevealthetruequalityofthepropertyortheactualintentionofthereviewerWefoundforinstancethatconsumersvaluelocation(egclosenesstothebeach)andcleanliness(egswimmingpoolareas)morethansuchfeaturesascheck-inandserviceSimilarlyinthenegativereviewsissuessuchaslackofbedroomparapher-naliaorkitchenitemsseemtoberankedhigherinimpor-tancethanissuessuchaslocationandcheck-inThereforeevenwhenapropertyscoreshighonthecertaindimensionstheoverallratingmayremainlowbecauseotherfactorsovershadowthefavorableaspects

WhatItMeansforManagersWeseeimplicationsforonlinereviewsitesandbyexten-sionforproprietorsofvacationhomesandserviceopera-tionsgenerallyWithregardtothereviewsitesifconsumersconsistentlydetectafavorablebiasthereviewsitesmaylosethecriticalaspectsofcredibilityandtrustAlthoughconsumersaregenerallysmartandrationalandcanworkaroundobviousbiasesrecentevidencesuggeststhatcon-sumersaregraduallydefectingfromsomeonlinesitesduetounbalancedreviews17InthisregardsiteadministratorsaswellaspropertymanagersshouldtakestepstoattractthoseconsumerswhoarewillingtoprovideareasonablybalancedanalysisoftheamenitiesandservicesBythiswemeanreviewsthatpresentcomprehensiveandobjectiveanalysisoftheserviceproviderratherthanjustravesorrants

Tothatendsiteoperatorsshouldtakestepstoun-derstandcustomersrsquoreviewwritingbehaviorandtoat-tractmorebalancedreviewsOnewaytodothisistogivepersonalizedinformationtotheconsumerdemonstratingthecontributionmadebyhisorherreviewandthevalueofthereviewtostakeholdersForinstanceLinget alfoundapositivecorrelationbetweenwillingnesstowritereviewsandtheconsumersrsquoperceptionthattheircontributionswouldbeuniqueandhelpfultoothercustomersaswellasperusedby

17HarteveldtHHJohnsonCStarkEandGeldernKV(2009)Us-ingDigitalChannelsToCalmTheAngryTravelerForresterResearch

theserviceproviders18Simplyputthepropensitytowritereviewsincreaseswhentheconsumersfeelthattheirreviewisactuallyhelpingothers(boththeserviceproviderandotherconsumers)

WesuggestthatreviewsitesdevelopbettermethodstoaggregatesynthesizeandpublishspecificreviewcommentsaswellasnumericalratingsFirstmanagersshouldtakeintoconsiderationthelowcorrelationbetweentheoverallratingsandconsumersrsquoassessmentsofvariouspropertyattributes(aswellastheresultsfromthetextanalysis)Theimplicationofsuchaweakcorrelationisthatreviewsitesneedtoexpandonthelistofpossiblevariablessothattheratingsaccuratelyreflectconsumersrsquoneedsandexpectationsregardingserviceproviders

Furthersitemanagersmayhavetorethinkthecurrentpracticeofaveragingalltheratingsforagivenpropertybe-causesuchameanratingfailstotakeintoaccountthebiasesthatareinherentintheratingsystemsAsanexamplethisstudyfindssignificantunder-reportingbiasinthisonlinereviewsiteAcustomerrsquosdecisiontonotpostonlinereviewsmaymeanthelossofimportantinformationthatcanassisttheusersofthesitestomakemorereliableinferencesThemajorityoftodayrsquosfeedbackmechanismsdonotpubliclydisclosethenumberofsilenttransactions(ietransactionsforwhichnofeedbackwaspostedbycustomers)Suchinfor-mationshouldbecomeapartofaserviceproviderrsquosonlineprofileonratingsitesandotherfeedbackmechanismsItisimportanttoprovidemoreinformationandheuristicstohelptheconsumersnavigatethroughtheclutterandgettheinformationtheywantandseek

Itisimportanttoapplymodernsemanticanalysisandsentimentclassificationtechniquestostudythecontentofthereviewsandidentifyvariousdimensionsonwhichcon-sumersevaluatevacationhomesThedetailsprovidedinthereviewtextareinmanywaysabetterreflectionofcustomersatisfactionServiceproviderscanbetterpositionthemselvesusingthedimensionsuncoveredthroughtextanalysisandtargettheneedsandpreferencesoftheircustomers

TheresultsinthisstudyshowthatnegativereviewsaretypicallyassociatedwithagreaterwordcountthanpositivereviewsItisusefultofocusspecificallyonthesenegativereviewsbecausetheycanrevealpatternsofdeficienciesintheservicestandardsanddeliveryThesereviewsshouldbetreatedthesameascustomercommentcardsorletters19Furthercustomerswhowritesuchlengthynegativereviews

18LingKimberlyGerardBeenenPaulLudfordXiaoxingWangKevinChangXinLiDanCosleyDanFrankowskiLorenTerveenAlMRashidandRobertKrautldquoUsingSocialPsychologytoMotivateContributionstoOnlineCommunitiesrdquoJournal of Computer-Mediated Communication10no4(2005)19MilanR(2007)10thingsyoucandoinresponsetotravelerreviewsinHotelmarketingcom

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 15

aremostlikelytospreadnegativeword-of-mouthaboutthepropertyLengthyreviewsareexcellentsourcesofinforma-tionandshouldbeusedasthebasestounderstandthecomplainingbehavioroftheconsumer

WordcountmaybeusedasanadditionalsegmentationvariabletoidentifythoseconsumerswhoaremoreinvolvedandwillingtoprovidebalancedfeedbacktoboththereviewsiteaswellastopropertymanagementVariousotherresearchmethodssuchastelephoneinterviewsorfocusgroupsshouldbeusedtoelicitdetailedfeedbackfromsuchconsumersTheseactionsalsohelpinservicerecoverypost-purchaseengagementandrelationshipbuilding

OutlookWeforeseethreepotentialareasinvestigationthatcanaddvaluetocurrentbestpracticesWestilldonrsquotknowmuchaboutconsumersrsquomotivationtowriteonlinereviews20Recentstudiesidentifyvariousantecedentsofreviewwritingbutsufferfromsamplingbiassincetheysurveyaself-select-edsampleofconsumerswhoalreadyprovidedreviewsAstheresultsinthisstudyindicateunder-reportingbias(lackofmoderateandlowreviews)remainsprevalentinonlinereviewsitesFutureresearchshouldidentifysocio-psycho-logicalfactorsthatincreaseconsumersrsquopropensitytowriteonlinereviews

Futureresearchshouldexploretheimpactofthepres-ence(orlack)ofextremelynegativereviewswithinalargecorpusofpositivereviewsTheevidenceontheimpactofre-

20WangYandFesenmaierDR(2003)AssessingMotivationofCon-tributioninOnlineCommunitiesAnEmpiricalInvestigationofanOn-lineTravelCommunityElectronic Markets1333-45andYooKyungHandUlrikeGretzelldquoWhatMotivatesConsumerstoWriteOnlineTravelReviewsrdquoInformation Technology amp Tourism10No4(2008)283-95

viewvalence(negativeorpositive)isequivocalProponentsofconfirmatorybiassuggestthatconsumerslookforaffir-mativeevidencesupportingaproductchoicealreadymadeIfthatisthecasepositivereviewsaremorelikelytohaveagreatereffectonconsumeractionsOntheotherhandthenotionofnegativitybiassuggeststhatwhenconsumersareneutralnegativereviewstendtobecomemoresalientthanpositivereviews21Inthepresenceoflargenumbersofposi-tivereviewsconsumersspecificallyseeknegativereviewsthattheyfeelwillhelpthemidentifyspecificproblemswithaserviceUnderstandinghowconsumersreconciletheinformationprovidedbybothnegativeandpositivereviewshasimplicationsforthedesignandmanagementofreviewsites

Finallyareviewrsquoswordcounthasimportantimplica-tionsforconsumersrsquotrustinthereviewsWhenconsumersarewillingtoreadandcompareopen-endedcommentsfromotherconsumerstheamountofinformationinareviewcanmatterWOMforhighinvolvementproductsismorepersuasiveandincreasesthedecisionmakerrsquosconfi-dencewhenthemessagesenderprovidesgreaternumberofreasonsfortheiroverallratingLongerreviewsprovidemoreinformationandarelikelytobeperceivedasmorehelpfulandpersuasivethanshorterreviewsThewaysinwhichareviewiswrittenframedandpresentedisaninterestingfactorwhichhasrarelybeeninvestigatedandhasimportantimplicationsparticularlygivenourfindingthattheamountofinformationinareview(wordcount)variessignificantlywiththeratingsofthereviewn

21BaSandPavlouPA(2002)EvidenceoftheEffectofTrustBuildingTechnologyinElectronicMarketsPricePremiumsandBuyerBehaviorMIS Quarterly26243-268

Cornell Hospitality Quarterlyhttpcqxsagepubcom

2011 ReportsVol11No15DesigningaSelf-HealingServiceSystemAnIntegrativeModelbyRobertCFordandMichaelCSturman

Vol11No14ReversingtheGreenBacklashWhyLargeHospitalityCompaniesShouldWelcomeCrediblyGreenCompetitorsbyMichaelGiebelhausenPhDandHaeEunHelenChunPhD

Vol11No13DevelopingaSustainabilityMeasurementFrameworkforHotelsTowardanIndustry-wideReportingStructurebyEricRicaurte

Vol11No12CreatingValueforWomenBusinessTravelersFocusingonEmotionalOutcomesbyJudiBrownellPhD

Vol11No11CustomerLoyaltyANewLookattheBenefitsofImprovingSegmentationEffortswithRewardsProgramsbyClayVoorheesPhDMichaelMcCallPhDandRogerCalantonePhD

Vol11No10CustomerPerceptionsofElectronicFoodOrderingbySherylEKimes

Vol11No92011TravelIndustryBenchmarkingStatusofSeniorDestinationandLodgingMarketingExecutivesbyRohitVermaPhDandKenMcGill

Vol11No8SearchOTAsandOnlineBookingAnExpandedAnalysisoftheBillboardEffectbyChrisAndersonPhD

Vol11No7OnlineMobileandTextFoodOrderingintheUSRestaurantIndustrybySherylEKimesPhDandPhilippFLaqueacute

Vol11No6HotelGuestsrsquoReactionstoGuestRoomSustainabilityInitiativesbyAlexSusskindPhDandRohitVermaPhD

Vol11No5TheImpactofTerrorismandEconomicShocksonUSHotelsbyCathyAEnzRenaacutetaKosovaacuteandMarkLomannoVol11No4ImplementingHumanResourceInnovationsThreeSuccessStoriesfromtheServiceIndustrybyJustinSunandKateWalshPhD

Vol11No3Compendium2011

Vol11No2PositioningaPlaceDevelopingaCompellingDestinationBrandbyRobertJKwortnikPhDandEthanHawkesMBA

Vol11No1TheImpactofHealthInsuranceonEmployeeJobAnxietyWithdrawalBehaviorsandTaskPerformancebySeanWayPhDBillCarrollPhDAlexSusskindPhDandJoeCYLeng

2011 Hospitality ToolsVol2No1TheGameHasChangedANewParadigmforStakeholderEngagementbyMaryBethMcEuen

2011 Industry PerspectivesNo7MegaTips2TwentyTestedTechniquesforIncreasingYourTipsbyMichaelLynn

2011 ProceedingsVol3No5BuildingBrandsintheInternetAgeAnalyticsLoyaltyandCommunicationbyGlennWithiam

Vol3No4BraveNewWorldOnlineHotelDistributionbyGlennWithiam

Vol3No3SocialMediaandtheHospitalityIndustryHoldingtheTigerbytheTailbyGlennWithiam

Vol3No2TheChallengeofHotelandRestaurantSustainabilityFindingProfitinldquoBeingGreenrdquobyGlennWithiam

Vol3No1CautiousOptimismCHRSExaminesHospitalityIndustryTrendsbyGlennWithiam

2010 ReportsVol10No18HowTravelersUseOnlineandSocialMediaChannelstoMakeHotel-choiceDecisionsbyLauraMcCarthyDebraStockandRohitVermaPhD

Vol10No17PublicorPrivateTheHospitalityInvestmentDecisionbyQingzhongMaPhDandAthenaWeiZhangPhD

Vol10No16BestPracticesinSearchEngineMarketingandOptimizationTheCaseoftheStJamesHotelbyGregBodenlcosVictorBogertDanGordonCarterHearneandChrisKAndersonPhD

Vol10No15TheImpactofPrix Fixe MenuPriceFormatsonGuestsrsquoDealPerceptionbyShuoWangandMichaelLynnPhD

Cornell Center for Hospitality Research

Publication Indexwwwchrcornelledu

The Executive Path Hospitality Leadership Through Learning

Complete program information and applications available online

wwwhotelschoolcornelleduexecedPhone + 1 607 255 4919 Email exec_ed_hotelcornelledu

Professionals from around the world are invited to attend 3-day 10-day or online courses at the worldrsquos leading institute for hospitality management education in

Visit our website to apply

Explore develop and apply ideas with global hospitality leaders and

expert Cornell professors

Success

AdvancingBusiness

andPersonal

bull Strategic Leadershipbull Financebull Foodservicebull Human Resources

bull Marketingbull Operationsbull Real Estate

No2RetaliationWhyanIncreaseinClaimsDoesNotMeantheSkyIsFallingbyDavidSherwynJDandGreggGilmanJD

2009 ToolsToolNo12MeasuringtheDiningExperienceTheCaseofVitaNovabyKeshPrasadandFredJDeMiccoPhD

Cornell Center for Hospitality ResearchPublication Index

The Executive Path Hospitality Leadership Through Learning

Complete program information and applications available online

wwwhotelschoolcornelleduexecedPhone + 1 607 255 4919 Email exec_ed_hotelcornelledu

Professionals from around the world are invited to attend 3-day 10-day or online courses at the worldrsquos leading institute for hospitality management education in

Visit our website to apply

Explore develop and apply ideas with global hospitality leaders and

expert Cornell professors

Success

AdvancingBusiness

andPersonal

bull Strategic Leadershipbull Financebull Foodservicebull Human Resources

bull Marketingbull Operationsbull Real Estate

visit our website to apply

wwwchrcornell edu

Page 8: Unscrambling the Puzzling Matter of Online Consumer ... · Philips Hospitality Americas Chris Proulx, CEO, eCornell & Executive Education Carolyn D. Richmond, Partner, Hospitality

8 TheCenterforHospitalityResearchbullCornellUniversity

bull Agerangeoftherevieweronan1-4ordinalscale(25-3435-4445-54and55andabove)

bull Averagepriceperdayand

bull Wordcount(usedinthequantitativeanalysis)

PreliminaryHypothesesandResultsTheagegroupsofthereviewwritersareshowninExhibit1Alargepercentageofthecustomerswhoreviewedthepropertiesfallunder44yearsofageThisisconsistentwithpreviousstudiesinthisrealmwhichshowthatgenerationXandGenerationYaremoreactiveintheWOMforums

ReviewRatingDistribution Websitesthathostonlinereviewsgiveconsumerstheoptiontoassignaratingtotheserviceproviderstypicallyonascaleof1to5or1to7Manysitesaggregatethesenumeri-calorstarratingstodetermineanoverall(mean)starratingWeseethisoverallratingasanimportantelementinonlineWOMsinceconsumersoftenusethisratingtonarrowdowntheirconsiderationsetStudiesthathaveusedtheratingastheprimarypredictoroftheserviceprovidersrsquoperceivedqualityorcustomersatisfactiontodeterminetheimpactofonlinereviewsonproductsaleshavearrivedatmixedfind-ings5Howeverthesestudiesarebasedonthefundamentalassumptionthatthenumericalratingsassignedtoservice

5ForexampleseeJChevalierandDMayzlinldquoTheEffectofWordofMouthOnlineOnlineBookReviewsrdquoJournal of Marketing ResearchVol43(2006)pp345-354andWDuanBGuandABWhinstonldquoDoOnlineReviewsMattermdashAnEmpiricalInvestigationofPanelDatardquoDecision Support SystemsVol45(2008)pp1007-1016

providersareanaccurateestimateofthequalityperceivedbythecustomerThispracticefollowsthebasicassumptionofbehavioraltheoriststhatanydatasetwithalargenumberofconsumerresponsestendstofollowanormal(Gaussian)distributionRecentstudieshavequestionedthisassump-tionForinstanceDellaracosandNarayanrsquosstudyonmoviereviewsinYahoocomfoundthatonlytheextremelysatisfiedandextremelydissatisfiedcustomersarelikelytopubliclyexpresstheiropinionsascomparedtocustomerswithmod-erateopinions6SimilarlystudiesusingdatafromTripadvi-sorcomandAmazoncomfoundthatthestarratingsexhibitatruncateddistributioninwhichthemajorityofreviewsarepositive7

TotestthisassumptionweranasimplefrequencychartasshowninExhibit2Inkeepingwithotherstudiesthedistributionisheavilyskewedtowardsthepositiveratings1-pointreviews32(1)2-pointreviews95(3)3-pointreviews458(14)4-pointreviews688(22)and5-pointreviews1912(60)Totestfortherobustnessofthisdistri-butionweappliedtheKolmogorov-Smirnovtestofnormal-ityTheresultsconfirmthenon-normaldistributionofthestarratingsandtheheavyskewtowardspositivereviews

Threeexplanationsarepossibleforsuchadistribution

6CDellarocasandRNarayanldquoAStatisticalMeasureofaPopulationrsquosPropensitytoEngageinPost-purchaseOnlineWord-of-mouthrdquoStatisti-cal ScienceVol21(2006)pp2777ATalwarandRFBoiJurcaUnderstandingUserBehaviorinOnlineFeedbackReportingrdquoinElectronic CommerceSanDiegoCalifornia(2007)pp134-142

Exhibit 1

Distribution of age ranges

Exhibit 2

Distribution of overall ldquostar ratingsrdquo

55 and above 45ndash54

years

35ndash44 years

18ndash34 years

overall ldquostar ratingrdquo

Num

ber o

f rev

iew

s

3295

458 688

1912

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 9

1 AmajorityofthepropertiesthatarelistedonthewebsitehaveoutstandingperceivedqualityandprovidesatisfactoryservicestocustomersWhilethisistheo-reticallypossibleitseemsunlikelywhenappliedtoallthepropertieslistedonthewebsite8

2 ThereviewsdonotconstituteatrulyrepresentativesampleofconsumeropinionBasedonotherstud-iesthisexplanationseemsplausibleConsumerswithmoderateviewsmayseelittleincentivetoreporttheirviewsinanonlineforumandasaconsequenceonlythoseconsumerswhohavepolarizedviewsarelikelytoposttheircomments

3 ThedistributioncouldalsobeanartifactofthefirmrsquosreviewsolicitingstrategyRecentstudiessupportthisideabysuggestingthatcustomersengageinsociallydesirablebehavioreveninimpersonalinteractionswiththeserviceproviders9Inthiscaseitmaybethattheywillprovideareasonablypositiverevieweveniftheexperiencehasnotbeenuptothemark

StarRatingsandPerceivedServiceQualityLetrsquosreturntothefirstquestionweaskedattheoutsetArestarratingsatruerepresentationofserviceprovid-ersrsquoperceivedqualityasreflectedbytheconsumersrsquostatedsentimentsThisbecomesanimportantquestionfortwo

8AKadetldquoRah-RahRatingsOnlinerdquoSmartMoney20079YGreacutegoireandRJFisherldquoCustomerBetrayalandRetaliationWhenYourBestCustomersbecomeYourWorstEnemiesrdquoJournal of the Acad-emy of Marketing ScienceVol36(2008)pp247-261

reasons(a) ratingsaretheprimaryevidencethatconsum-ersconsiderwhileshoppingonlineforinformationand(b)thetruncatednatureofratingdistributionsmeansthatthereviewsavailablemaynottrulyrepresentconsumeropinionortheserviceprovidersrsquoperceivedqualityIfonlythosewithextremeviewspostreviewsthenconsumersshouldalwayslookforserviceprovidersthathavea5-starratingOnthecontraryifthereviewsaccuratelyreflectallconsumersrsquoviewsthenconsumersarebetteroffsearchingforreviewswitha3ratingsincethesetendtobemorebalancedandenablebetterproductanalysisRecentlyHu et aluseddatafromAmazoncomshowthatnumericalscoresdonotreflectbooksrsquotrueperceivedquality10AstheysuggestldquohellipratherthescorereflectsthebalanceofdiverseopinionsInotherwordswhenabookrsquosoverallscoreisaround3itdoesnotsuggestthatconsumersgenerallyagreethatthisisanaver-agebookItrathersuggeststhatroughlyequalnumberofconsumersthinkthatthebookiseitheranoutstandingbookoranabysmalbookrdquo

Totestthisassertionweappliedaregressionequationwithsixpropertyattributes(ievalueformoneycheck-inlocationcleanlinesscomfortandservice)asindependentvariablesandstarratingasthedependentvariableThisanalysisisbasedontheassumptionthatforagivenpropertyifstarratingstrulyrepresenttheperceivedqualityoftheproperties(aswellastheconsumersrsquosentiment)thenthestarratingsandratingsofthesixotherpropertyattributes

10NHuJZhangandPAPavlouOvercomingtheJ-shapedDistribu-tionofProductReviewsrdquoCommunications of the ACMVol52(2009)pp144-147

low reviews t-value high reviews t-value(Constant) 1618 18353 2400 34186vAlue 0254 13726 0378 8767CheCKiN -0083 -3701 -0053 -1192loCATioN 0104 6132 0079 2117CleAN 0144 7574 0112 2808CoMForT 0267 13993 0129 3088serviCe 0098 4209 0107 2401r2 0331 0338F-statistic 5038 24714

Exhibit 3

regression analysis of ldquolowrdquo and ldquohighrdquo reviews

Notes Coefficients (t-statistics reported) Significance is as follows significant at 0001 level significant at 05 level significant at 010 level Dependent variable is overall rating

10 TheCenterforHospitalityResearchbullCornellUniversity

moderateratingsonalltheattributesbutwasstillwillingtoprovideahighoverallstarratingtothepropertyThisexamplesuggeststhatconsumersplacesignificantemphasisontheoverallexperientialvalueprovidedbythepropertiesThisexpectationofvalueismanytimesintangibleanddoesnotnecessarilydependonthefunctioningorratingoftheindividualfeaturesoftheproperty

RelationshipbetweenRatingsandReviewAttributesTounderstandconsumersrsquoword-of-mouthbehavioritisimportanttoconsidervariousfactorssuchasperceivedexpectationsofcostsandbenefitsimportanceofthepur-chase(involvement)personalcharacteristicsandsituationalinfluencesWeconsidertwofactorsthatareknowntoaffecttheintensityandvalence(positiveornegative)ofwordofmouthpriceandwordcount

PricePriceisanimportantfactorthatcontributestopre-purchaseexpectationsTheintensityofexpectations(andtheirfulfillment)partlydrivenbypriceisdirectlypropor-tionaltoconsumersrsquosatisfactionlevelsandWOMpropensityThismaybethereasonthatthetendencytocomplainisdirectlyrelatedtothecostoftheservice11Themoreexpen-siveanitemthegreaterwillbetheperceivedlossesandthegreaterlikelihoodthatdissatisfactionwillresultinacom-plaintSimilarlythelikelihoodofword-of-mouthbeingtrig-geredbythelevelofexpectationndashdisconfirmationincreases

11BeardenWOandTeelJE(1983)Selecteddeterminantsofcon-sumersatisfactionandcomplaintreportsJournal of Marketing Research 2021-28

willshowahighcorrelationFortheanalysiswetookintoconsiderationthefactmostreviewsarepositive(ie4-and5-pointratings)Thereforewedividedthereviewsintotwocategorieshighreviews(starratingabove3n=2597)andlowreviews(ratingsbeloworequalto3n=600)andranseparateregressionanalysesonthesetwogroupsasshowninExhibit3

AninterestingfindingistheweakrelationshipbetweentheattributeratingsandoverallstarratingforeachpropertyTheR-squareoftheregressionequationwithallsixvariablesandtheoverallstarratingsislessthan40percentThatiseveniftheconsumersprovidedhigherratingsonindividualattributestheiroverallratingwasusuallylowandthere-versewasalsotrueThistrendisobservedinboththehighandlowreviewgroups

TheanalysissuggeststhatinhighreviewsconsumersplacedgreaterimportanceonvalueformoneycleanlinessandcomfortThehighertheratingsofaconsumerontheseattributesthehigheristhelikelihoodthattheconsumerwillprovideahigherstarratingCheck-inlocationandservicedidnotprominentlyfigureintheiroverallevalu-ationsofthepropertieseventhoughtheseattributeswereratedhigh

ThelowreviewsshowsimilarresultsinthesensethattheloweraconsumerratesvaluecleanlinessandcomfortthelowerthelikelihoodofapositiveratingInadditioninthecaseoflowreviewsconsumersplacedemphasisontheserviceprovidedbythepropertystaffandmanagement

TheaboveresultisillustratedinoneoftheldquohighrdquoreviewsshowninExhibit4Thereviewergavelowto

star rating Text Commentary

This was our 1st time renting a house vs a condo This house was great If you wanted to spend a little time by yourself you could so many rooms to go to The house was so comfortable like being at home It was our home for the week Thank you Hope to rent it next year if we havent bought one ourselves

ratings for specific Attributes

value Check-in location Clean Comfort service

3 2 3 2 4 3

Exhibit 4

example of a ldquohighrdquo review

Avg rating for Property Attributes

30

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 11

withthecostoftheproduct12Giventhegreaterperceivedriskinservicescustomerswouldseekmoretrustworthyser-viceproviderswithagoodreputationofsatisfyingcustomerneeds13ThisinturnincreasestheexpectationsandleadstomoreextremereactionsfromtheconsumersHenceonecanspeculatethatthehigherthepriceshigherwillbetheaver-agenegativeorpositiveratingsforaservice

Word countThenumberofwordsinareviewmayserveasaproxyfortheamountofinformationinthereviewsHowevertheevidenceonthisaspectofWOMofalltypesisequivocalForinstanceconsumersgenerallyengageinpositiveWOMtoavoidfeelingsofguiltandas-sociationwithbadnewsandtoreducecognitivedissonanceOntheotherhandacustomerwhoishighlydisappointedwillprovideapersuasiveargumenttomakesurethatothercustomersaredissuadedfromrentingthesamepropertyResearchonconventionalwordofmouthhasofferedampleevidencetosuggestthatdissatisfiedcustomersengageintwotothreetimesasmuchWOMassatisfiedcustomers14Webelievethatthislogicextendstoonlinereviewsdespitethefactthatexistingstudiesareequivocalatbest15Thereforeitcanbeassumedthatreviewswitheitherextremelypositiveorextremelynegativeratingswillhavegreaterwordcountasopposedtomoremoderatereviews

TotesttheseassumptionsweappliedANOVAwithpriceandwordcountasdependentvariablesandthereviewratingasfactorsAsillustratedinExhibits5and6theresultsshowthatpriceandwordcountshowasignificantnegativecorrelationwithstarratingThatisreviewswithlowerratingsaretypicallyassociatedwithgreaterwordcountandrepresentpropertiesthatchargehighrates

TheresultonthewordcountsuggeststhatconsumerswhoareunhappywitharentalpropertytendtowritemoretoexpresstheirdissatisfactionAsshowninthisexhibitlowerratingsareassociatedwithhighernumberofwordsinthereviewsandviceversaSimilarlypropertypriceseemstobestronglyassociatedwithnegativewordofmouthsup-portingtheprinciplethathighercostincreasestheconsum-ersrsquoinvolvementinthepurchase

12BeardenWOandMasonJB(1984)AninvestigationofinfluencesonconsumercomplaintreportsAdvancesinConsumerResearch11490-49513WetzerIMZeelenbergMandPietersR(2007)ldquoNevereatinthatrestaurantIdidrdquoExploringwhypeopleengageinnegativeword-of-mouthcommunicationPsychology and Marketing24661-68014SchlossbergH(1991)CustomersatisfactionNotafadbutawayoflifeMarketing News25andWestbrookR(1987)ProductConsump-tion-basedAffectiveResponsesandPost-purchaseProcessesJournal of Marketing Research 24258-27015AlexMSusskindldquoIToldYouSoRestaurantCustomersrsquoofWord-of-MouthCommunicationPatternsrdquoCornell Hotel and Restaurant Adminis-tration QuarterlyVol43No2(April2002)pp75-85

Exhibit 5

Differences in word count across star ratings

Exhibit 6

Differences in average price per day across ratings

Aver

age

pric

e pe

r day

of t

he re

ntal

pro

pert

y

overall ldquostar ratingrdquo

1504

1563

1804

19861987

Aver

age

wor

d co

unt o

f rev

iew

overall ldquostar ratingrdquo

836837

121

15751543

160

140

120

100

80

200

190

180

170

160

150

1 2 3 4 5

1 2 3 4 5

12 TheCenterforHospitalityResearchbullCornellUniversity

thereviewersThesecanbeclassifiedas(a)attributesorfea-turesofthepropertythatfigureprominentlyandcontributetothepositiveornegativeratings(egroomsbeddishesfurnitureorservice)and(b)thepositiveornegativewordsthatrepresentthesentimentofthereviewer(eggreatamazingdirtyandwonderful)Theresultsofthequalitativeanalysisareasfollows

Syntactic featuresTheoveralltoneofthelowreviewsisratherargumentativeThereviewerstendtobeforcefulinpresentingtheirpointofviewTheylayoutagreatamountofreasoningandprovidespecificexamplestodrivehometheirargumentThisisreflectedinthegreaterpercentageofnegativeconnectorwords(34)suchasbut however as ifandeven thoughItisinterestingtonotethatreviewersinthiscategoryspeakdirectlytotheotherconsumersnotablythroughgreateruseofwordssuchasyou(46)andlooktoinfluencetheirdecisionmaking

InthehighreviewsbycontrasttheoveralltonetendstobeenunciativeandthereviewersusethetexttomerelyreinforcetheirnumericalratingsusingfewerwordsonaveragethanthenegativereviewersInthesereviewstheconsumersfocusontheirpersonalexperiencesandappealtotheaudiencethroughemotionsratherthanlogicandreasoningasevidentfromgreateruseofpositiveadjectives

QualitativeAnalysisoftheReviewTextTheaboveanalysisemphasizestheimportanceofthetextcontentinconsumerreviewsWhilenumericalratingsallowthereviewerstoratetheserviceprovideronvariousattributesitisthetextthatprovidesthemwiththeopportu-nitytoarticulatethenuancesoftheiroverallexperienceandconveyusefulinformationaboutaserviceproviderrsquostransac-tionsandservicecapabilitiesThereforesemanticanalysisofthetextcontentprovidesuswithanopportunitytothethirdquestionpresentedattheoutsetofthisreportwhatdoconsumerstalkaboutintheirreviews

Tothisendweanalyzed600eachofrandomlyselectedhighandlowreviewsusingsemanticprocessingtechniquesWeusedcenteringresonanceanalysis(CRA)amodeofnet-work-basedtextanalysisthatrepresentsthecontentoflargesetsoftextsbyidentifyingthemostimportantwordsthatlinkotherwordsinthenetwork16OuranalysisexaminedsyntacticandsemanticfeaturesSyntacticfeaturesincludethepercentageofthewordsthatareopenclass(ienounsverbsadjectives)Semanticfeaturesrepresentthespecial-izedvocabularythatreflectstheoverallopinionexpressedby

16CormanSandDooleyK(2006)CrawdadTextAnalysisSystem12CrawdadTechnologiesLLCChandlerAZ

Prominently Mentioned Features Freq sample Comments

bedroom beds linen sheets 383

Also the same bedroom and bathroom did not have any door separating the bedroom from the bathroom no privacy

Sleeping arrangements were very awkward The listing sheet described 2 full beds We brought our own linens the beds were queen sized Had to purchase sheet

Kitchen Kitchenware Dishes Dishwasher 297

You must not forget to step up when yoursquore walking out of the kitchen into the dining room otherwise yoursquoll trip over the 3-foot step up

No potholders in the kitchen and it didnrsquot seem like there were as many kitchen supplies as before We just had to make do without

beach beaches beachfront 296

The week was the best vacation we have ever had as a family The beach house was a class act

Disappointment 1 It looks like this house is very close to beach but to get there you have to go around through public path

Pool Poolside 218

The pool was too dirty to use We checked in on Thursday and the pool was not cleaned until Saturday

Also it is inconvenient to have to call to get someone to open the pool Didnrsquot understand why Jacuzzi was open and not the pool

bathroom(s) bath bathtub shower 202

The next problem was in another bathroom (main floor)mdashthe wall heater did not work so showering in here was a bit chilly The next problem was in the ldquomasterrdquo bathroom (connected to the bedroom with the King bed) The sliding shower doors need to be totally replaced

Bathtub drains clogged in three second floor bathrooms

Exhibit 7

Top five prominent issues in low reviews

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 13

suchasgreat fantastic andawesome(45)andpersonalpronounssuchasweandI(54)

Semantic featuresExhibits7and8showthemainissuesoraspectsthatareprominentlydiscussedinthelowreviewsTheorderoftheseissuesisbasedonthestatisti-callysignificantfrequencyoftheiroccurrenceinthe1200reviews

WefurtherinvestigatedtheuseofvarioussentimentwordsInterestinglytheprominentissuesdiscussedinthepositivereviewssometimesdonotevenfigureintheratingscalesprovidedtothereviewersOveralltheresultsofthesemanticanalysiscanbesummarizedasfollows

Thepropertyanditsfeaturesaremorecommonlycon-nectedwithextremelypositivesentimentwordssuchasgreat wonderfulandamazinginthehighreviewsthaninthelowreviewsIncomparisonthepropertiesandtheirfea-turesinthelowreviewsaremorelikelytobecharacterizedwithmoderatelypositivewordssuchasniceandokay

Negativereviewsreflecttheextentofexpectationndashdis-confirmationthattheconsumersfaceduringtheirstayintherentalpropertyThesereviewshaveagreaterproportionofwordsexpressingdisappointment(egnotasexpected)Furtherextremelynegativesentimentwords(egdiscom-fortfilthy)andadjectivesaremorelikelytobeassociatedwithfeaturesinlowreviewsthaninhighreviews

Almost40percentoftheconsumerswhowrotehighreviewsexpresstheirwillingnesstoreturntothepropertyasopposedtoonly9percentofthosewritingnegativereviewsFurthermorethan20percentoftheconsumersgivinghighreviewsexpresstheirstrongdesiretorecommendtheprop-ertytotheirfriendsandfamilymembersasopposedtoonly5percentinlowreviews

ReviewofFindingsOurexploratorystudyofconsumerreviewsofUSvacationhomesonReviewsitecomwashighlightedbyoverwhelm-inglypositivereviews

Theoveralldistributionofthereviewswasheav-ilyskewedtowardsthepositiveratingsMorecriticallythisfindingsuggeststhatthereisalackofmoderateandbalancedopinionsWetheorizethatthisiseitherduetopurchasingbias(sincecustomerswentthroughthetroubleofpurchasingandexperiencingthefacilitiestheytendtobepositivelybiasedtowardstheirpurchase)orunder-reportingbias(customersdonothavestrongenoughincentivestotakethetroubleofreportingtheiropinionsandthefewmotivatedindividualswhoprovidetheirreviewsbecomeaself-selectedsample)

Theoverall(mean)starratingtypicallyusedintravelerreviewsitesisprobablynotthemostaccurateindicatorofcustomersrsquoperceptionsofserviceprovidersrsquoqualityWe

Prominently Mentioned Features Freq sample Commentsbeach beaches beachfront 506 Perfect rental property for those who enjoy the activity of the beach and the serenity

of the Sound The house is just down the road from ocean

The large porch offered a beautiful view of the beach and comfortable new rocking chairs

Pool Poolside 218 The outdoor shower dipping pool and screen in porch were an added bonus Perfect house for a low key weekend

House was great for kids Pool was terrific and the playroom was perfectlocation 192 We had a wonderful time My wife loves fishing and the cottage was in a perfect

location to go fishing on the pier

The home is somewhat dated but the location was great We have vacationed in the Chatham area for many summers and this time it was great too

bedroom beds 186 The family suite is huge with a king size and twin bed and still room for a pack-n-play The kitchen and dining room were huge clean and nice

The only comments that I would say to the negative is the front bed room by the front door always seemed damp which seem to attract more bugs

Kitchen Kitchenware Dishes Dishwasher 178 All of the amenities in the house are upscalemdashloved the kitchen and plasma TVs

The kitchen had good basic utensils dishes and cooking pans

Exhibit 7

Top five prominent issues in high reviews

14 TheCenterforHospitalityResearchbullCornellUniversity

foundacorrelationbetweenthestarratingandratingsonotherpropertyattributesoflessthan40percentinmostcases

AhighwordcountinthereviewsiscorrelatedwithalowoverallratingThatisconsumerswhoareextremelydissatisfiedtendtoexpendmoretimeandenergyontheircritiqueofthepropertyitsamenitiesandservicesPriceofthepropertyalsoseemstopositivelycorrelatewithextremenegativereviewsperhapsbecauseahighpriceheightenstravelersrsquoexpectations

Numericalratingsdonotoftencapturetheactualsentimentandthevarietyofdimensionsonwhichtravel-ersevaluatepropertiesBecausenumericalratingsdonotcapturereviewersrsquoopinionsaboutspecificattributestheycanbedeceivingandmaynotrevealthetruequalityofthepropertyortheactualintentionofthereviewerWefoundforinstancethatconsumersvaluelocation(egclosenesstothebeach)andcleanliness(egswimmingpoolareas)morethansuchfeaturesascheck-inandserviceSimilarlyinthenegativereviewsissuessuchaslackofbedroomparapher-naliaorkitchenitemsseemtoberankedhigherinimpor-tancethanissuessuchaslocationandcheck-inThereforeevenwhenapropertyscoreshighonthecertaindimensionstheoverallratingmayremainlowbecauseotherfactorsovershadowthefavorableaspects

WhatItMeansforManagersWeseeimplicationsforonlinereviewsitesandbyexten-sionforproprietorsofvacationhomesandserviceopera-tionsgenerallyWithregardtothereviewsitesifconsumersconsistentlydetectafavorablebiasthereviewsitesmaylosethecriticalaspectsofcredibilityandtrustAlthoughconsumersaregenerallysmartandrationalandcanworkaroundobviousbiasesrecentevidencesuggeststhatcon-sumersaregraduallydefectingfromsomeonlinesitesduetounbalancedreviews17InthisregardsiteadministratorsaswellaspropertymanagersshouldtakestepstoattractthoseconsumerswhoarewillingtoprovideareasonablybalancedanalysisoftheamenitiesandservicesBythiswemeanreviewsthatpresentcomprehensiveandobjectiveanalysisoftheserviceproviderratherthanjustravesorrants

Tothatendsiteoperatorsshouldtakestepstoun-derstandcustomersrsquoreviewwritingbehaviorandtoat-tractmorebalancedreviewsOnewaytodothisistogivepersonalizedinformationtotheconsumerdemonstratingthecontributionmadebyhisorherreviewandthevalueofthereviewtostakeholdersForinstanceLinget alfoundapositivecorrelationbetweenwillingnesstowritereviewsandtheconsumersrsquoperceptionthattheircontributionswouldbeuniqueandhelpfultoothercustomersaswellasperusedby

17HarteveldtHHJohnsonCStarkEandGeldernKV(2009)Us-ingDigitalChannelsToCalmTheAngryTravelerForresterResearch

theserviceproviders18Simplyputthepropensitytowritereviewsincreaseswhentheconsumersfeelthattheirreviewisactuallyhelpingothers(boththeserviceproviderandotherconsumers)

WesuggestthatreviewsitesdevelopbettermethodstoaggregatesynthesizeandpublishspecificreviewcommentsaswellasnumericalratingsFirstmanagersshouldtakeintoconsiderationthelowcorrelationbetweentheoverallratingsandconsumersrsquoassessmentsofvariouspropertyattributes(aswellastheresultsfromthetextanalysis)Theimplicationofsuchaweakcorrelationisthatreviewsitesneedtoexpandonthelistofpossiblevariablessothattheratingsaccuratelyreflectconsumersrsquoneedsandexpectationsregardingserviceproviders

Furthersitemanagersmayhavetorethinkthecurrentpracticeofaveragingalltheratingsforagivenpropertybe-causesuchameanratingfailstotakeintoaccountthebiasesthatareinherentintheratingsystemsAsanexamplethisstudyfindssignificantunder-reportingbiasinthisonlinereviewsiteAcustomerrsquosdecisiontonotpostonlinereviewsmaymeanthelossofimportantinformationthatcanassisttheusersofthesitestomakemorereliableinferencesThemajorityoftodayrsquosfeedbackmechanismsdonotpubliclydisclosethenumberofsilenttransactions(ietransactionsforwhichnofeedbackwaspostedbycustomers)Suchinfor-mationshouldbecomeapartofaserviceproviderrsquosonlineprofileonratingsitesandotherfeedbackmechanismsItisimportanttoprovidemoreinformationandheuristicstohelptheconsumersnavigatethroughtheclutterandgettheinformationtheywantandseek

Itisimportanttoapplymodernsemanticanalysisandsentimentclassificationtechniquestostudythecontentofthereviewsandidentifyvariousdimensionsonwhichcon-sumersevaluatevacationhomesThedetailsprovidedinthereviewtextareinmanywaysabetterreflectionofcustomersatisfactionServiceproviderscanbetterpositionthemselvesusingthedimensionsuncoveredthroughtextanalysisandtargettheneedsandpreferencesoftheircustomers

TheresultsinthisstudyshowthatnegativereviewsaretypicallyassociatedwithagreaterwordcountthanpositivereviewsItisusefultofocusspecificallyonthesenegativereviewsbecausetheycanrevealpatternsofdeficienciesintheservicestandardsanddeliveryThesereviewsshouldbetreatedthesameascustomercommentcardsorletters19Furthercustomerswhowritesuchlengthynegativereviews

18LingKimberlyGerardBeenenPaulLudfordXiaoxingWangKevinChangXinLiDanCosleyDanFrankowskiLorenTerveenAlMRashidandRobertKrautldquoUsingSocialPsychologytoMotivateContributionstoOnlineCommunitiesrdquoJournal of Computer-Mediated Communication10no4(2005)19MilanR(2007)10thingsyoucandoinresponsetotravelerreviewsinHotelmarketingcom

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 15

aremostlikelytospreadnegativeword-of-mouthaboutthepropertyLengthyreviewsareexcellentsourcesofinforma-tionandshouldbeusedasthebasestounderstandthecomplainingbehavioroftheconsumer

WordcountmaybeusedasanadditionalsegmentationvariabletoidentifythoseconsumerswhoaremoreinvolvedandwillingtoprovidebalancedfeedbacktoboththereviewsiteaswellastopropertymanagementVariousotherresearchmethodssuchastelephoneinterviewsorfocusgroupsshouldbeusedtoelicitdetailedfeedbackfromsuchconsumersTheseactionsalsohelpinservicerecoverypost-purchaseengagementandrelationshipbuilding

OutlookWeforeseethreepotentialareasinvestigationthatcanaddvaluetocurrentbestpracticesWestilldonrsquotknowmuchaboutconsumersrsquomotivationtowriteonlinereviews20Recentstudiesidentifyvariousantecedentsofreviewwritingbutsufferfromsamplingbiassincetheysurveyaself-select-edsampleofconsumerswhoalreadyprovidedreviewsAstheresultsinthisstudyindicateunder-reportingbias(lackofmoderateandlowreviews)remainsprevalentinonlinereviewsitesFutureresearchshouldidentifysocio-psycho-logicalfactorsthatincreaseconsumersrsquopropensitytowriteonlinereviews

Futureresearchshouldexploretheimpactofthepres-ence(orlack)ofextremelynegativereviewswithinalargecorpusofpositivereviewsTheevidenceontheimpactofre-

20WangYandFesenmaierDR(2003)AssessingMotivationofCon-tributioninOnlineCommunitiesAnEmpiricalInvestigationofanOn-lineTravelCommunityElectronic Markets1333-45andYooKyungHandUlrikeGretzelldquoWhatMotivatesConsumerstoWriteOnlineTravelReviewsrdquoInformation Technology amp Tourism10No4(2008)283-95

viewvalence(negativeorpositive)isequivocalProponentsofconfirmatorybiassuggestthatconsumerslookforaffir-mativeevidencesupportingaproductchoicealreadymadeIfthatisthecasepositivereviewsaremorelikelytohaveagreatereffectonconsumeractionsOntheotherhandthenotionofnegativitybiassuggeststhatwhenconsumersareneutralnegativereviewstendtobecomemoresalientthanpositivereviews21Inthepresenceoflargenumbersofposi-tivereviewsconsumersspecificallyseeknegativereviewsthattheyfeelwillhelpthemidentifyspecificproblemswithaserviceUnderstandinghowconsumersreconciletheinformationprovidedbybothnegativeandpositivereviewshasimplicationsforthedesignandmanagementofreviewsites

Finallyareviewrsquoswordcounthasimportantimplica-tionsforconsumersrsquotrustinthereviewsWhenconsumersarewillingtoreadandcompareopen-endedcommentsfromotherconsumerstheamountofinformationinareviewcanmatterWOMforhighinvolvementproductsismorepersuasiveandincreasesthedecisionmakerrsquosconfi-dencewhenthemessagesenderprovidesgreaternumberofreasonsfortheiroverallratingLongerreviewsprovidemoreinformationandarelikelytobeperceivedasmorehelpfulandpersuasivethanshorterreviewsThewaysinwhichareviewiswrittenframedandpresentedisaninterestingfactorwhichhasrarelybeeninvestigatedandhasimportantimplicationsparticularlygivenourfindingthattheamountofinformationinareview(wordcount)variessignificantlywiththeratingsofthereviewn

21BaSandPavlouPA(2002)EvidenceoftheEffectofTrustBuildingTechnologyinElectronicMarketsPricePremiumsandBuyerBehaviorMIS Quarterly26243-268

Cornell Hospitality Quarterlyhttpcqxsagepubcom

2011 ReportsVol11No15DesigningaSelf-HealingServiceSystemAnIntegrativeModelbyRobertCFordandMichaelCSturman

Vol11No14ReversingtheGreenBacklashWhyLargeHospitalityCompaniesShouldWelcomeCrediblyGreenCompetitorsbyMichaelGiebelhausenPhDandHaeEunHelenChunPhD

Vol11No13DevelopingaSustainabilityMeasurementFrameworkforHotelsTowardanIndustry-wideReportingStructurebyEricRicaurte

Vol11No12CreatingValueforWomenBusinessTravelersFocusingonEmotionalOutcomesbyJudiBrownellPhD

Vol11No11CustomerLoyaltyANewLookattheBenefitsofImprovingSegmentationEffortswithRewardsProgramsbyClayVoorheesPhDMichaelMcCallPhDandRogerCalantonePhD

Vol11No10CustomerPerceptionsofElectronicFoodOrderingbySherylEKimes

Vol11No92011TravelIndustryBenchmarkingStatusofSeniorDestinationandLodgingMarketingExecutivesbyRohitVermaPhDandKenMcGill

Vol11No8SearchOTAsandOnlineBookingAnExpandedAnalysisoftheBillboardEffectbyChrisAndersonPhD

Vol11No7OnlineMobileandTextFoodOrderingintheUSRestaurantIndustrybySherylEKimesPhDandPhilippFLaqueacute

Vol11No6HotelGuestsrsquoReactionstoGuestRoomSustainabilityInitiativesbyAlexSusskindPhDandRohitVermaPhD

Vol11No5TheImpactofTerrorismandEconomicShocksonUSHotelsbyCathyAEnzRenaacutetaKosovaacuteandMarkLomannoVol11No4ImplementingHumanResourceInnovationsThreeSuccessStoriesfromtheServiceIndustrybyJustinSunandKateWalshPhD

Vol11No3Compendium2011

Vol11No2PositioningaPlaceDevelopingaCompellingDestinationBrandbyRobertJKwortnikPhDandEthanHawkesMBA

Vol11No1TheImpactofHealthInsuranceonEmployeeJobAnxietyWithdrawalBehaviorsandTaskPerformancebySeanWayPhDBillCarrollPhDAlexSusskindPhDandJoeCYLeng

2011 Hospitality ToolsVol2No1TheGameHasChangedANewParadigmforStakeholderEngagementbyMaryBethMcEuen

2011 Industry PerspectivesNo7MegaTips2TwentyTestedTechniquesforIncreasingYourTipsbyMichaelLynn

2011 ProceedingsVol3No5BuildingBrandsintheInternetAgeAnalyticsLoyaltyandCommunicationbyGlennWithiam

Vol3No4BraveNewWorldOnlineHotelDistributionbyGlennWithiam

Vol3No3SocialMediaandtheHospitalityIndustryHoldingtheTigerbytheTailbyGlennWithiam

Vol3No2TheChallengeofHotelandRestaurantSustainabilityFindingProfitinldquoBeingGreenrdquobyGlennWithiam

Vol3No1CautiousOptimismCHRSExaminesHospitalityIndustryTrendsbyGlennWithiam

2010 ReportsVol10No18HowTravelersUseOnlineandSocialMediaChannelstoMakeHotel-choiceDecisionsbyLauraMcCarthyDebraStockandRohitVermaPhD

Vol10No17PublicorPrivateTheHospitalityInvestmentDecisionbyQingzhongMaPhDandAthenaWeiZhangPhD

Vol10No16BestPracticesinSearchEngineMarketingandOptimizationTheCaseoftheStJamesHotelbyGregBodenlcosVictorBogertDanGordonCarterHearneandChrisKAndersonPhD

Vol10No15TheImpactofPrix Fixe MenuPriceFormatsonGuestsrsquoDealPerceptionbyShuoWangandMichaelLynnPhD

Cornell Center for Hospitality Research

Publication Indexwwwchrcornelledu

The Executive Path Hospitality Leadership Through Learning

Complete program information and applications available online

wwwhotelschoolcornelleduexecedPhone + 1 607 255 4919 Email exec_ed_hotelcornelledu

Professionals from around the world are invited to attend 3-day 10-day or online courses at the worldrsquos leading institute for hospitality management education in

Visit our website to apply

Explore develop and apply ideas with global hospitality leaders and

expert Cornell professors

Success

AdvancingBusiness

andPersonal

bull Strategic Leadershipbull Financebull Foodservicebull Human Resources

bull Marketingbull Operationsbull Real Estate

No2RetaliationWhyanIncreaseinClaimsDoesNotMeantheSkyIsFallingbyDavidSherwynJDandGreggGilmanJD

2009 ToolsToolNo12MeasuringtheDiningExperienceTheCaseofVitaNovabyKeshPrasadandFredJDeMiccoPhD

Cornell Center for Hospitality ResearchPublication Index

The Executive Path Hospitality Leadership Through Learning

Complete program information and applications available online

wwwhotelschoolcornelleduexecedPhone + 1 607 255 4919 Email exec_ed_hotelcornelledu

Professionals from around the world are invited to attend 3-day 10-day or online courses at the worldrsquos leading institute for hospitality management education in

Visit our website to apply

Explore develop and apply ideas with global hospitality leaders and

expert Cornell professors

Success

AdvancingBusiness

andPersonal

bull Strategic Leadershipbull Financebull Foodservicebull Human Resources

bull Marketingbull Operationsbull Real Estate

visit our website to apply

wwwchrcornell edu

Page 9: Unscrambling the Puzzling Matter of Online Consumer ... · Philips Hospitality Americas Chris Proulx, CEO, eCornell & Executive Education Carolyn D. Richmond, Partner, Hospitality

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 9

1 AmajorityofthepropertiesthatarelistedonthewebsitehaveoutstandingperceivedqualityandprovidesatisfactoryservicestocustomersWhilethisistheo-reticallypossibleitseemsunlikelywhenappliedtoallthepropertieslistedonthewebsite8

2 ThereviewsdonotconstituteatrulyrepresentativesampleofconsumeropinionBasedonotherstud-iesthisexplanationseemsplausibleConsumerswithmoderateviewsmayseelittleincentivetoreporttheirviewsinanonlineforumandasaconsequenceonlythoseconsumerswhohavepolarizedviewsarelikelytoposttheircomments

3 ThedistributioncouldalsobeanartifactofthefirmrsquosreviewsolicitingstrategyRecentstudiessupportthisideabysuggestingthatcustomersengageinsociallydesirablebehavioreveninimpersonalinteractionswiththeserviceproviders9Inthiscaseitmaybethattheywillprovideareasonablypositiverevieweveniftheexperiencehasnotbeenuptothemark

StarRatingsandPerceivedServiceQualityLetrsquosreturntothefirstquestionweaskedattheoutsetArestarratingsatruerepresentationofserviceprovid-ersrsquoperceivedqualityasreflectedbytheconsumersrsquostatedsentimentsThisbecomesanimportantquestionfortwo

8AKadetldquoRah-RahRatingsOnlinerdquoSmartMoney20079YGreacutegoireandRJFisherldquoCustomerBetrayalandRetaliationWhenYourBestCustomersbecomeYourWorstEnemiesrdquoJournal of the Acad-emy of Marketing ScienceVol36(2008)pp247-261

reasons(a) ratingsaretheprimaryevidencethatconsum-ersconsiderwhileshoppingonlineforinformationand(b)thetruncatednatureofratingdistributionsmeansthatthereviewsavailablemaynottrulyrepresentconsumeropinionortheserviceprovidersrsquoperceivedqualityIfonlythosewithextremeviewspostreviewsthenconsumersshouldalwayslookforserviceprovidersthathavea5-starratingOnthecontraryifthereviewsaccuratelyreflectallconsumersrsquoviewsthenconsumersarebetteroffsearchingforreviewswitha3ratingsincethesetendtobemorebalancedandenablebetterproductanalysisRecentlyHu et aluseddatafromAmazoncomshowthatnumericalscoresdonotreflectbooksrsquotrueperceivedquality10AstheysuggestldquohellipratherthescorereflectsthebalanceofdiverseopinionsInotherwordswhenabookrsquosoverallscoreisaround3itdoesnotsuggestthatconsumersgenerallyagreethatthisisanaver-agebookItrathersuggeststhatroughlyequalnumberofconsumersthinkthatthebookiseitheranoutstandingbookoranabysmalbookrdquo

Totestthisassertionweappliedaregressionequationwithsixpropertyattributes(ievalueformoneycheck-inlocationcleanlinesscomfortandservice)asindependentvariablesandstarratingasthedependentvariableThisanalysisisbasedontheassumptionthatforagivenpropertyifstarratingstrulyrepresenttheperceivedqualityoftheproperties(aswellastheconsumersrsquosentiment)thenthestarratingsandratingsofthesixotherpropertyattributes

10NHuJZhangandPAPavlouOvercomingtheJ-shapedDistribu-tionofProductReviewsrdquoCommunications of the ACMVol52(2009)pp144-147

low reviews t-value high reviews t-value(Constant) 1618 18353 2400 34186vAlue 0254 13726 0378 8767CheCKiN -0083 -3701 -0053 -1192loCATioN 0104 6132 0079 2117CleAN 0144 7574 0112 2808CoMForT 0267 13993 0129 3088serviCe 0098 4209 0107 2401r2 0331 0338F-statistic 5038 24714

Exhibit 3

regression analysis of ldquolowrdquo and ldquohighrdquo reviews

Notes Coefficients (t-statistics reported) Significance is as follows significant at 0001 level significant at 05 level significant at 010 level Dependent variable is overall rating

10 TheCenterforHospitalityResearchbullCornellUniversity

moderateratingsonalltheattributesbutwasstillwillingtoprovideahighoverallstarratingtothepropertyThisexamplesuggeststhatconsumersplacesignificantemphasisontheoverallexperientialvalueprovidedbythepropertiesThisexpectationofvalueismanytimesintangibleanddoesnotnecessarilydependonthefunctioningorratingoftheindividualfeaturesoftheproperty

RelationshipbetweenRatingsandReviewAttributesTounderstandconsumersrsquoword-of-mouthbehavioritisimportanttoconsidervariousfactorssuchasperceivedexpectationsofcostsandbenefitsimportanceofthepur-chase(involvement)personalcharacteristicsandsituationalinfluencesWeconsidertwofactorsthatareknowntoaffecttheintensityandvalence(positiveornegative)ofwordofmouthpriceandwordcount

PricePriceisanimportantfactorthatcontributestopre-purchaseexpectationsTheintensityofexpectations(andtheirfulfillment)partlydrivenbypriceisdirectlypropor-tionaltoconsumersrsquosatisfactionlevelsandWOMpropensityThismaybethereasonthatthetendencytocomplainisdirectlyrelatedtothecostoftheservice11Themoreexpen-siveanitemthegreaterwillbetheperceivedlossesandthegreaterlikelihoodthatdissatisfactionwillresultinacom-plaintSimilarlythelikelihoodofword-of-mouthbeingtrig-geredbythelevelofexpectationndashdisconfirmationincreases

11BeardenWOandTeelJE(1983)Selecteddeterminantsofcon-sumersatisfactionandcomplaintreportsJournal of Marketing Research 2021-28

willshowahighcorrelationFortheanalysiswetookintoconsiderationthefactmostreviewsarepositive(ie4-and5-pointratings)Thereforewedividedthereviewsintotwocategorieshighreviews(starratingabove3n=2597)andlowreviews(ratingsbeloworequalto3n=600)andranseparateregressionanalysesonthesetwogroupsasshowninExhibit3

AninterestingfindingistheweakrelationshipbetweentheattributeratingsandoverallstarratingforeachpropertyTheR-squareoftheregressionequationwithallsixvariablesandtheoverallstarratingsislessthan40percentThatiseveniftheconsumersprovidedhigherratingsonindividualattributestheiroverallratingwasusuallylowandthere-versewasalsotrueThistrendisobservedinboththehighandlowreviewgroups

TheanalysissuggeststhatinhighreviewsconsumersplacedgreaterimportanceonvalueformoneycleanlinessandcomfortThehighertheratingsofaconsumerontheseattributesthehigheristhelikelihoodthattheconsumerwillprovideahigherstarratingCheck-inlocationandservicedidnotprominentlyfigureintheiroverallevalu-ationsofthepropertieseventhoughtheseattributeswereratedhigh

ThelowreviewsshowsimilarresultsinthesensethattheloweraconsumerratesvaluecleanlinessandcomfortthelowerthelikelihoodofapositiveratingInadditioninthecaseoflowreviewsconsumersplacedemphasisontheserviceprovidedbythepropertystaffandmanagement

TheaboveresultisillustratedinoneoftheldquohighrdquoreviewsshowninExhibit4Thereviewergavelowto

star rating Text Commentary

This was our 1st time renting a house vs a condo This house was great If you wanted to spend a little time by yourself you could so many rooms to go to The house was so comfortable like being at home It was our home for the week Thank you Hope to rent it next year if we havent bought one ourselves

ratings for specific Attributes

value Check-in location Clean Comfort service

3 2 3 2 4 3

Exhibit 4

example of a ldquohighrdquo review

Avg rating for Property Attributes

30

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 11

withthecostoftheproduct12Giventhegreaterperceivedriskinservicescustomerswouldseekmoretrustworthyser-viceproviderswithagoodreputationofsatisfyingcustomerneeds13ThisinturnincreasestheexpectationsandleadstomoreextremereactionsfromtheconsumersHenceonecanspeculatethatthehigherthepriceshigherwillbetheaver-agenegativeorpositiveratingsforaservice

Word countThenumberofwordsinareviewmayserveasaproxyfortheamountofinformationinthereviewsHowevertheevidenceonthisaspectofWOMofalltypesisequivocalForinstanceconsumersgenerallyengageinpositiveWOMtoavoidfeelingsofguiltandas-sociationwithbadnewsandtoreducecognitivedissonanceOntheotherhandacustomerwhoishighlydisappointedwillprovideapersuasiveargumenttomakesurethatothercustomersaredissuadedfromrentingthesamepropertyResearchonconventionalwordofmouthhasofferedampleevidencetosuggestthatdissatisfiedcustomersengageintwotothreetimesasmuchWOMassatisfiedcustomers14Webelievethatthislogicextendstoonlinereviewsdespitethefactthatexistingstudiesareequivocalatbest15Thereforeitcanbeassumedthatreviewswitheitherextremelypositiveorextremelynegativeratingswillhavegreaterwordcountasopposedtomoremoderatereviews

TotesttheseassumptionsweappliedANOVAwithpriceandwordcountasdependentvariablesandthereviewratingasfactorsAsillustratedinExhibits5and6theresultsshowthatpriceandwordcountshowasignificantnegativecorrelationwithstarratingThatisreviewswithlowerratingsaretypicallyassociatedwithgreaterwordcountandrepresentpropertiesthatchargehighrates

TheresultonthewordcountsuggeststhatconsumerswhoareunhappywitharentalpropertytendtowritemoretoexpresstheirdissatisfactionAsshowninthisexhibitlowerratingsareassociatedwithhighernumberofwordsinthereviewsandviceversaSimilarlypropertypriceseemstobestronglyassociatedwithnegativewordofmouthsup-portingtheprinciplethathighercostincreasestheconsum-ersrsquoinvolvementinthepurchase

12BeardenWOandMasonJB(1984)AninvestigationofinfluencesonconsumercomplaintreportsAdvancesinConsumerResearch11490-49513WetzerIMZeelenbergMandPietersR(2007)ldquoNevereatinthatrestaurantIdidrdquoExploringwhypeopleengageinnegativeword-of-mouthcommunicationPsychology and Marketing24661-68014SchlossbergH(1991)CustomersatisfactionNotafadbutawayoflifeMarketing News25andWestbrookR(1987)ProductConsump-tion-basedAffectiveResponsesandPost-purchaseProcessesJournal of Marketing Research 24258-27015AlexMSusskindldquoIToldYouSoRestaurantCustomersrsquoofWord-of-MouthCommunicationPatternsrdquoCornell Hotel and Restaurant Adminis-tration QuarterlyVol43No2(April2002)pp75-85

Exhibit 5

Differences in word count across star ratings

Exhibit 6

Differences in average price per day across ratings

Aver

age

pric

e pe

r day

of t

he re

ntal

pro

pert

y

overall ldquostar ratingrdquo

1504

1563

1804

19861987

Aver

age

wor

d co

unt o

f rev

iew

overall ldquostar ratingrdquo

836837

121

15751543

160

140

120

100

80

200

190

180

170

160

150

1 2 3 4 5

1 2 3 4 5

12 TheCenterforHospitalityResearchbullCornellUniversity

thereviewersThesecanbeclassifiedas(a)attributesorfea-turesofthepropertythatfigureprominentlyandcontributetothepositiveornegativeratings(egroomsbeddishesfurnitureorservice)and(b)thepositiveornegativewordsthatrepresentthesentimentofthereviewer(eggreatamazingdirtyandwonderful)Theresultsofthequalitativeanalysisareasfollows

Syntactic featuresTheoveralltoneofthelowreviewsisratherargumentativeThereviewerstendtobeforcefulinpresentingtheirpointofviewTheylayoutagreatamountofreasoningandprovidespecificexamplestodrivehometheirargumentThisisreflectedinthegreaterpercentageofnegativeconnectorwords(34)suchasbut however as ifandeven thoughItisinterestingtonotethatreviewersinthiscategoryspeakdirectlytotheotherconsumersnotablythroughgreateruseofwordssuchasyou(46)andlooktoinfluencetheirdecisionmaking

InthehighreviewsbycontrasttheoveralltonetendstobeenunciativeandthereviewersusethetexttomerelyreinforcetheirnumericalratingsusingfewerwordsonaveragethanthenegativereviewersInthesereviewstheconsumersfocusontheirpersonalexperiencesandappealtotheaudiencethroughemotionsratherthanlogicandreasoningasevidentfromgreateruseofpositiveadjectives

QualitativeAnalysisoftheReviewTextTheaboveanalysisemphasizestheimportanceofthetextcontentinconsumerreviewsWhilenumericalratingsallowthereviewerstoratetheserviceprovideronvariousattributesitisthetextthatprovidesthemwiththeopportu-nitytoarticulatethenuancesoftheiroverallexperienceandconveyusefulinformationaboutaserviceproviderrsquostransac-tionsandservicecapabilitiesThereforesemanticanalysisofthetextcontentprovidesuswithanopportunitytothethirdquestionpresentedattheoutsetofthisreportwhatdoconsumerstalkaboutintheirreviews

Tothisendweanalyzed600eachofrandomlyselectedhighandlowreviewsusingsemanticprocessingtechniquesWeusedcenteringresonanceanalysis(CRA)amodeofnet-work-basedtextanalysisthatrepresentsthecontentoflargesetsoftextsbyidentifyingthemostimportantwordsthatlinkotherwordsinthenetwork16OuranalysisexaminedsyntacticandsemanticfeaturesSyntacticfeaturesincludethepercentageofthewordsthatareopenclass(ienounsverbsadjectives)Semanticfeaturesrepresentthespecial-izedvocabularythatreflectstheoverallopinionexpressedby

16CormanSandDooleyK(2006)CrawdadTextAnalysisSystem12CrawdadTechnologiesLLCChandlerAZ

Prominently Mentioned Features Freq sample Comments

bedroom beds linen sheets 383

Also the same bedroom and bathroom did not have any door separating the bedroom from the bathroom no privacy

Sleeping arrangements were very awkward The listing sheet described 2 full beds We brought our own linens the beds were queen sized Had to purchase sheet

Kitchen Kitchenware Dishes Dishwasher 297

You must not forget to step up when yoursquore walking out of the kitchen into the dining room otherwise yoursquoll trip over the 3-foot step up

No potholders in the kitchen and it didnrsquot seem like there were as many kitchen supplies as before We just had to make do without

beach beaches beachfront 296

The week was the best vacation we have ever had as a family The beach house was a class act

Disappointment 1 It looks like this house is very close to beach but to get there you have to go around through public path

Pool Poolside 218

The pool was too dirty to use We checked in on Thursday and the pool was not cleaned until Saturday

Also it is inconvenient to have to call to get someone to open the pool Didnrsquot understand why Jacuzzi was open and not the pool

bathroom(s) bath bathtub shower 202

The next problem was in another bathroom (main floor)mdashthe wall heater did not work so showering in here was a bit chilly The next problem was in the ldquomasterrdquo bathroom (connected to the bedroom with the King bed) The sliding shower doors need to be totally replaced

Bathtub drains clogged in three second floor bathrooms

Exhibit 7

Top five prominent issues in low reviews

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 13

suchasgreat fantastic andawesome(45)andpersonalpronounssuchasweandI(54)

Semantic featuresExhibits7and8showthemainissuesoraspectsthatareprominentlydiscussedinthelowreviewsTheorderoftheseissuesisbasedonthestatisti-callysignificantfrequencyoftheiroccurrenceinthe1200reviews

WefurtherinvestigatedtheuseofvarioussentimentwordsInterestinglytheprominentissuesdiscussedinthepositivereviewssometimesdonotevenfigureintheratingscalesprovidedtothereviewersOveralltheresultsofthesemanticanalysiscanbesummarizedasfollows

Thepropertyanditsfeaturesaremorecommonlycon-nectedwithextremelypositivesentimentwordssuchasgreat wonderfulandamazinginthehighreviewsthaninthelowreviewsIncomparisonthepropertiesandtheirfea-turesinthelowreviewsaremorelikelytobecharacterizedwithmoderatelypositivewordssuchasniceandokay

Negativereviewsreflecttheextentofexpectationndashdis-confirmationthattheconsumersfaceduringtheirstayintherentalpropertyThesereviewshaveagreaterproportionofwordsexpressingdisappointment(egnotasexpected)Furtherextremelynegativesentimentwords(egdiscom-fortfilthy)andadjectivesaremorelikelytobeassociatedwithfeaturesinlowreviewsthaninhighreviews

Almost40percentoftheconsumerswhowrotehighreviewsexpresstheirwillingnesstoreturntothepropertyasopposedtoonly9percentofthosewritingnegativereviewsFurthermorethan20percentoftheconsumersgivinghighreviewsexpresstheirstrongdesiretorecommendtheprop-ertytotheirfriendsandfamilymembersasopposedtoonly5percentinlowreviews

ReviewofFindingsOurexploratorystudyofconsumerreviewsofUSvacationhomesonReviewsitecomwashighlightedbyoverwhelm-inglypositivereviews

Theoveralldistributionofthereviewswasheav-ilyskewedtowardsthepositiveratingsMorecriticallythisfindingsuggeststhatthereisalackofmoderateandbalancedopinionsWetheorizethatthisiseitherduetopurchasingbias(sincecustomerswentthroughthetroubleofpurchasingandexperiencingthefacilitiestheytendtobepositivelybiasedtowardstheirpurchase)orunder-reportingbias(customersdonothavestrongenoughincentivestotakethetroubleofreportingtheiropinionsandthefewmotivatedindividualswhoprovidetheirreviewsbecomeaself-selectedsample)

Theoverall(mean)starratingtypicallyusedintravelerreviewsitesisprobablynotthemostaccurateindicatorofcustomersrsquoperceptionsofserviceprovidersrsquoqualityWe

Prominently Mentioned Features Freq sample Commentsbeach beaches beachfront 506 Perfect rental property for those who enjoy the activity of the beach and the serenity

of the Sound The house is just down the road from ocean

The large porch offered a beautiful view of the beach and comfortable new rocking chairs

Pool Poolside 218 The outdoor shower dipping pool and screen in porch were an added bonus Perfect house for a low key weekend

House was great for kids Pool was terrific and the playroom was perfectlocation 192 We had a wonderful time My wife loves fishing and the cottage was in a perfect

location to go fishing on the pier

The home is somewhat dated but the location was great We have vacationed in the Chatham area for many summers and this time it was great too

bedroom beds 186 The family suite is huge with a king size and twin bed and still room for a pack-n-play The kitchen and dining room were huge clean and nice

The only comments that I would say to the negative is the front bed room by the front door always seemed damp which seem to attract more bugs

Kitchen Kitchenware Dishes Dishwasher 178 All of the amenities in the house are upscalemdashloved the kitchen and plasma TVs

The kitchen had good basic utensils dishes and cooking pans

Exhibit 7

Top five prominent issues in high reviews

14 TheCenterforHospitalityResearchbullCornellUniversity

foundacorrelationbetweenthestarratingandratingsonotherpropertyattributesoflessthan40percentinmostcases

AhighwordcountinthereviewsiscorrelatedwithalowoverallratingThatisconsumerswhoareextremelydissatisfiedtendtoexpendmoretimeandenergyontheircritiqueofthepropertyitsamenitiesandservicesPriceofthepropertyalsoseemstopositivelycorrelatewithextremenegativereviewsperhapsbecauseahighpriceheightenstravelersrsquoexpectations

Numericalratingsdonotoftencapturetheactualsentimentandthevarietyofdimensionsonwhichtravel-ersevaluatepropertiesBecausenumericalratingsdonotcapturereviewersrsquoopinionsaboutspecificattributestheycanbedeceivingandmaynotrevealthetruequalityofthepropertyortheactualintentionofthereviewerWefoundforinstancethatconsumersvaluelocation(egclosenesstothebeach)andcleanliness(egswimmingpoolareas)morethansuchfeaturesascheck-inandserviceSimilarlyinthenegativereviewsissuessuchaslackofbedroomparapher-naliaorkitchenitemsseemtoberankedhigherinimpor-tancethanissuessuchaslocationandcheck-inThereforeevenwhenapropertyscoreshighonthecertaindimensionstheoverallratingmayremainlowbecauseotherfactorsovershadowthefavorableaspects

WhatItMeansforManagersWeseeimplicationsforonlinereviewsitesandbyexten-sionforproprietorsofvacationhomesandserviceopera-tionsgenerallyWithregardtothereviewsitesifconsumersconsistentlydetectafavorablebiasthereviewsitesmaylosethecriticalaspectsofcredibilityandtrustAlthoughconsumersaregenerallysmartandrationalandcanworkaroundobviousbiasesrecentevidencesuggeststhatcon-sumersaregraduallydefectingfromsomeonlinesitesduetounbalancedreviews17InthisregardsiteadministratorsaswellaspropertymanagersshouldtakestepstoattractthoseconsumerswhoarewillingtoprovideareasonablybalancedanalysisoftheamenitiesandservicesBythiswemeanreviewsthatpresentcomprehensiveandobjectiveanalysisoftheserviceproviderratherthanjustravesorrants

Tothatendsiteoperatorsshouldtakestepstoun-derstandcustomersrsquoreviewwritingbehaviorandtoat-tractmorebalancedreviewsOnewaytodothisistogivepersonalizedinformationtotheconsumerdemonstratingthecontributionmadebyhisorherreviewandthevalueofthereviewtostakeholdersForinstanceLinget alfoundapositivecorrelationbetweenwillingnesstowritereviewsandtheconsumersrsquoperceptionthattheircontributionswouldbeuniqueandhelpfultoothercustomersaswellasperusedby

17HarteveldtHHJohnsonCStarkEandGeldernKV(2009)Us-ingDigitalChannelsToCalmTheAngryTravelerForresterResearch

theserviceproviders18Simplyputthepropensitytowritereviewsincreaseswhentheconsumersfeelthattheirreviewisactuallyhelpingothers(boththeserviceproviderandotherconsumers)

WesuggestthatreviewsitesdevelopbettermethodstoaggregatesynthesizeandpublishspecificreviewcommentsaswellasnumericalratingsFirstmanagersshouldtakeintoconsiderationthelowcorrelationbetweentheoverallratingsandconsumersrsquoassessmentsofvariouspropertyattributes(aswellastheresultsfromthetextanalysis)Theimplicationofsuchaweakcorrelationisthatreviewsitesneedtoexpandonthelistofpossiblevariablessothattheratingsaccuratelyreflectconsumersrsquoneedsandexpectationsregardingserviceproviders

Furthersitemanagersmayhavetorethinkthecurrentpracticeofaveragingalltheratingsforagivenpropertybe-causesuchameanratingfailstotakeintoaccountthebiasesthatareinherentintheratingsystemsAsanexamplethisstudyfindssignificantunder-reportingbiasinthisonlinereviewsiteAcustomerrsquosdecisiontonotpostonlinereviewsmaymeanthelossofimportantinformationthatcanassisttheusersofthesitestomakemorereliableinferencesThemajorityoftodayrsquosfeedbackmechanismsdonotpubliclydisclosethenumberofsilenttransactions(ietransactionsforwhichnofeedbackwaspostedbycustomers)Suchinfor-mationshouldbecomeapartofaserviceproviderrsquosonlineprofileonratingsitesandotherfeedbackmechanismsItisimportanttoprovidemoreinformationandheuristicstohelptheconsumersnavigatethroughtheclutterandgettheinformationtheywantandseek

Itisimportanttoapplymodernsemanticanalysisandsentimentclassificationtechniquestostudythecontentofthereviewsandidentifyvariousdimensionsonwhichcon-sumersevaluatevacationhomesThedetailsprovidedinthereviewtextareinmanywaysabetterreflectionofcustomersatisfactionServiceproviderscanbetterpositionthemselvesusingthedimensionsuncoveredthroughtextanalysisandtargettheneedsandpreferencesoftheircustomers

TheresultsinthisstudyshowthatnegativereviewsaretypicallyassociatedwithagreaterwordcountthanpositivereviewsItisusefultofocusspecificallyonthesenegativereviewsbecausetheycanrevealpatternsofdeficienciesintheservicestandardsanddeliveryThesereviewsshouldbetreatedthesameascustomercommentcardsorletters19Furthercustomerswhowritesuchlengthynegativereviews

18LingKimberlyGerardBeenenPaulLudfordXiaoxingWangKevinChangXinLiDanCosleyDanFrankowskiLorenTerveenAlMRashidandRobertKrautldquoUsingSocialPsychologytoMotivateContributionstoOnlineCommunitiesrdquoJournal of Computer-Mediated Communication10no4(2005)19MilanR(2007)10thingsyoucandoinresponsetotravelerreviewsinHotelmarketingcom

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 15

aremostlikelytospreadnegativeword-of-mouthaboutthepropertyLengthyreviewsareexcellentsourcesofinforma-tionandshouldbeusedasthebasestounderstandthecomplainingbehavioroftheconsumer

WordcountmaybeusedasanadditionalsegmentationvariabletoidentifythoseconsumerswhoaremoreinvolvedandwillingtoprovidebalancedfeedbacktoboththereviewsiteaswellastopropertymanagementVariousotherresearchmethodssuchastelephoneinterviewsorfocusgroupsshouldbeusedtoelicitdetailedfeedbackfromsuchconsumersTheseactionsalsohelpinservicerecoverypost-purchaseengagementandrelationshipbuilding

OutlookWeforeseethreepotentialareasinvestigationthatcanaddvaluetocurrentbestpracticesWestilldonrsquotknowmuchaboutconsumersrsquomotivationtowriteonlinereviews20Recentstudiesidentifyvariousantecedentsofreviewwritingbutsufferfromsamplingbiassincetheysurveyaself-select-edsampleofconsumerswhoalreadyprovidedreviewsAstheresultsinthisstudyindicateunder-reportingbias(lackofmoderateandlowreviews)remainsprevalentinonlinereviewsitesFutureresearchshouldidentifysocio-psycho-logicalfactorsthatincreaseconsumersrsquopropensitytowriteonlinereviews

Futureresearchshouldexploretheimpactofthepres-ence(orlack)ofextremelynegativereviewswithinalargecorpusofpositivereviewsTheevidenceontheimpactofre-

20WangYandFesenmaierDR(2003)AssessingMotivationofCon-tributioninOnlineCommunitiesAnEmpiricalInvestigationofanOn-lineTravelCommunityElectronic Markets1333-45andYooKyungHandUlrikeGretzelldquoWhatMotivatesConsumerstoWriteOnlineTravelReviewsrdquoInformation Technology amp Tourism10No4(2008)283-95

viewvalence(negativeorpositive)isequivocalProponentsofconfirmatorybiassuggestthatconsumerslookforaffir-mativeevidencesupportingaproductchoicealreadymadeIfthatisthecasepositivereviewsaremorelikelytohaveagreatereffectonconsumeractionsOntheotherhandthenotionofnegativitybiassuggeststhatwhenconsumersareneutralnegativereviewstendtobecomemoresalientthanpositivereviews21Inthepresenceoflargenumbersofposi-tivereviewsconsumersspecificallyseeknegativereviewsthattheyfeelwillhelpthemidentifyspecificproblemswithaserviceUnderstandinghowconsumersreconciletheinformationprovidedbybothnegativeandpositivereviewshasimplicationsforthedesignandmanagementofreviewsites

Finallyareviewrsquoswordcounthasimportantimplica-tionsforconsumersrsquotrustinthereviewsWhenconsumersarewillingtoreadandcompareopen-endedcommentsfromotherconsumerstheamountofinformationinareviewcanmatterWOMforhighinvolvementproductsismorepersuasiveandincreasesthedecisionmakerrsquosconfi-dencewhenthemessagesenderprovidesgreaternumberofreasonsfortheiroverallratingLongerreviewsprovidemoreinformationandarelikelytobeperceivedasmorehelpfulandpersuasivethanshorterreviewsThewaysinwhichareviewiswrittenframedandpresentedisaninterestingfactorwhichhasrarelybeeninvestigatedandhasimportantimplicationsparticularlygivenourfindingthattheamountofinformationinareview(wordcount)variessignificantlywiththeratingsofthereviewn

21BaSandPavlouPA(2002)EvidenceoftheEffectofTrustBuildingTechnologyinElectronicMarketsPricePremiumsandBuyerBehaviorMIS Quarterly26243-268

Cornell Hospitality Quarterlyhttpcqxsagepubcom

2011 ReportsVol11No15DesigningaSelf-HealingServiceSystemAnIntegrativeModelbyRobertCFordandMichaelCSturman

Vol11No14ReversingtheGreenBacklashWhyLargeHospitalityCompaniesShouldWelcomeCrediblyGreenCompetitorsbyMichaelGiebelhausenPhDandHaeEunHelenChunPhD

Vol11No13DevelopingaSustainabilityMeasurementFrameworkforHotelsTowardanIndustry-wideReportingStructurebyEricRicaurte

Vol11No12CreatingValueforWomenBusinessTravelersFocusingonEmotionalOutcomesbyJudiBrownellPhD

Vol11No11CustomerLoyaltyANewLookattheBenefitsofImprovingSegmentationEffortswithRewardsProgramsbyClayVoorheesPhDMichaelMcCallPhDandRogerCalantonePhD

Vol11No10CustomerPerceptionsofElectronicFoodOrderingbySherylEKimes

Vol11No92011TravelIndustryBenchmarkingStatusofSeniorDestinationandLodgingMarketingExecutivesbyRohitVermaPhDandKenMcGill

Vol11No8SearchOTAsandOnlineBookingAnExpandedAnalysisoftheBillboardEffectbyChrisAndersonPhD

Vol11No7OnlineMobileandTextFoodOrderingintheUSRestaurantIndustrybySherylEKimesPhDandPhilippFLaqueacute

Vol11No6HotelGuestsrsquoReactionstoGuestRoomSustainabilityInitiativesbyAlexSusskindPhDandRohitVermaPhD

Vol11No5TheImpactofTerrorismandEconomicShocksonUSHotelsbyCathyAEnzRenaacutetaKosovaacuteandMarkLomannoVol11No4ImplementingHumanResourceInnovationsThreeSuccessStoriesfromtheServiceIndustrybyJustinSunandKateWalshPhD

Vol11No3Compendium2011

Vol11No2PositioningaPlaceDevelopingaCompellingDestinationBrandbyRobertJKwortnikPhDandEthanHawkesMBA

Vol11No1TheImpactofHealthInsuranceonEmployeeJobAnxietyWithdrawalBehaviorsandTaskPerformancebySeanWayPhDBillCarrollPhDAlexSusskindPhDandJoeCYLeng

2011 Hospitality ToolsVol2No1TheGameHasChangedANewParadigmforStakeholderEngagementbyMaryBethMcEuen

2011 Industry PerspectivesNo7MegaTips2TwentyTestedTechniquesforIncreasingYourTipsbyMichaelLynn

2011 ProceedingsVol3No5BuildingBrandsintheInternetAgeAnalyticsLoyaltyandCommunicationbyGlennWithiam

Vol3No4BraveNewWorldOnlineHotelDistributionbyGlennWithiam

Vol3No3SocialMediaandtheHospitalityIndustryHoldingtheTigerbytheTailbyGlennWithiam

Vol3No2TheChallengeofHotelandRestaurantSustainabilityFindingProfitinldquoBeingGreenrdquobyGlennWithiam

Vol3No1CautiousOptimismCHRSExaminesHospitalityIndustryTrendsbyGlennWithiam

2010 ReportsVol10No18HowTravelersUseOnlineandSocialMediaChannelstoMakeHotel-choiceDecisionsbyLauraMcCarthyDebraStockandRohitVermaPhD

Vol10No17PublicorPrivateTheHospitalityInvestmentDecisionbyQingzhongMaPhDandAthenaWeiZhangPhD

Vol10No16BestPracticesinSearchEngineMarketingandOptimizationTheCaseoftheStJamesHotelbyGregBodenlcosVictorBogertDanGordonCarterHearneandChrisKAndersonPhD

Vol10No15TheImpactofPrix Fixe MenuPriceFormatsonGuestsrsquoDealPerceptionbyShuoWangandMichaelLynnPhD

Cornell Center for Hospitality Research

Publication Indexwwwchrcornelledu

The Executive Path Hospitality Leadership Through Learning

Complete program information and applications available online

wwwhotelschoolcornelleduexecedPhone + 1 607 255 4919 Email exec_ed_hotelcornelledu

Professionals from around the world are invited to attend 3-day 10-day or online courses at the worldrsquos leading institute for hospitality management education in

Visit our website to apply

Explore develop and apply ideas with global hospitality leaders and

expert Cornell professors

Success

AdvancingBusiness

andPersonal

bull Strategic Leadershipbull Financebull Foodservicebull Human Resources

bull Marketingbull Operationsbull Real Estate

No2RetaliationWhyanIncreaseinClaimsDoesNotMeantheSkyIsFallingbyDavidSherwynJDandGreggGilmanJD

2009 ToolsToolNo12MeasuringtheDiningExperienceTheCaseofVitaNovabyKeshPrasadandFredJDeMiccoPhD

Cornell Center for Hospitality ResearchPublication Index

The Executive Path Hospitality Leadership Through Learning

Complete program information and applications available online

wwwhotelschoolcornelleduexecedPhone + 1 607 255 4919 Email exec_ed_hotelcornelledu

Professionals from around the world are invited to attend 3-day 10-day or online courses at the worldrsquos leading institute for hospitality management education in

Visit our website to apply

Explore develop and apply ideas with global hospitality leaders and

expert Cornell professors

Success

AdvancingBusiness

andPersonal

bull Strategic Leadershipbull Financebull Foodservicebull Human Resources

bull Marketingbull Operationsbull Real Estate

visit our website to apply

wwwchrcornell edu

Page 10: Unscrambling the Puzzling Matter of Online Consumer ... · Philips Hospitality Americas Chris Proulx, CEO, eCornell & Executive Education Carolyn D. Richmond, Partner, Hospitality

10 TheCenterforHospitalityResearchbullCornellUniversity

moderateratingsonalltheattributesbutwasstillwillingtoprovideahighoverallstarratingtothepropertyThisexamplesuggeststhatconsumersplacesignificantemphasisontheoverallexperientialvalueprovidedbythepropertiesThisexpectationofvalueismanytimesintangibleanddoesnotnecessarilydependonthefunctioningorratingoftheindividualfeaturesoftheproperty

RelationshipbetweenRatingsandReviewAttributesTounderstandconsumersrsquoword-of-mouthbehavioritisimportanttoconsidervariousfactorssuchasperceivedexpectationsofcostsandbenefitsimportanceofthepur-chase(involvement)personalcharacteristicsandsituationalinfluencesWeconsidertwofactorsthatareknowntoaffecttheintensityandvalence(positiveornegative)ofwordofmouthpriceandwordcount

PricePriceisanimportantfactorthatcontributestopre-purchaseexpectationsTheintensityofexpectations(andtheirfulfillment)partlydrivenbypriceisdirectlypropor-tionaltoconsumersrsquosatisfactionlevelsandWOMpropensityThismaybethereasonthatthetendencytocomplainisdirectlyrelatedtothecostoftheservice11Themoreexpen-siveanitemthegreaterwillbetheperceivedlossesandthegreaterlikelihoodthatdissatisfactionwillresultinacom-plaintSimilarlythelikelihoodofword-of-mouthbeingtrig-geredbythelevelofexpectationndashdisconfirmationincreases

11BeardenWOandTeelJE(1983)Selecteddeterminantsofcon-sumersatisfactionandcomplaintreportsJournal of Marketing Research 2021-28

willshowahighcorrelationFortheanalysiswetookintoconsiderationthefactmostreviewsarepositive(ie4-and5-pointratings)Thereforewedividedthereviewsintotwocategorieshighreviews(starratingabove3n=2597)andlowreviews(ratingsbeloworequalto3n=600)andranseparateregressionanalysesonthesetwogroupsasshowninExhibit3

AninterestingfindingistheweakrelationshipbetweentheattributeratingsandoverallstarratingforeachpropertyTheR-squareoftheregressionequationwithallsixvariablesandtheoverallstarratingsislessthan40percentThatiseveniftheconsumersprovidedhigherratingsonindividualattributestheiroverallratingwasusuallylowandthere-versewasalsotrueThistrendisobservedinboththehighandlowreviewgroups

TheanalysissuggeststhatinhighreviewsconsumersplacedgreaterimportanceonvalueformoneycleanlinessandcomfortThehighertheratingsofaconsumerontheseattributesthehigheristhelikelihoodthattheconsumerwillprovideahigherstarratingCheck-inlocationandservicedidnotprominentlyfigureintheiroverallevalu-ationsofthepropertieseventhoughtheseattributeswereratedhigh

ThelowreviewsshowsimilarresultsinthesensethattheloweraconsumerratesvaluecleanlinessandcomfortthelowerthelikelihoodofapositiveratingInadditioninthecaseoflowreviewsconsumersplacedemphasisontheserviceprovidedbythepropertystaffandmanagement

TheaboveresultisillustratedinoneoftheldquohighrdquoreviewsshowninExhibit4Thereviewergavelowto

star rating Text Commentary

This was our 1st time renting a house vs a condo This house was great If you wanted to spend a little time by yourself you could so many rooms to go to The house was so comfortable like being at home It was our home for the week Thank you Hope to rent it next year if we havent bought one ourselves

ratings for specific Attributes

value Check-in location Clean Comfort service

3 2 3 2 4 3

Exhibit 4

example of a ldquohighrdquo review

Avg rating for Property Attributes

30

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 11

withthecostoftheproduct12Giventhegreaterperceivedriskinservicescustomerswouldseekmoretrustworthyser-viceproviderswithagoodreputationofsatisfyingcustomerneeds13ThisinturnincreasestheexpectationsandleadstomoreextremereactionsfromtheconsumersHenceonecanspeculatethatthehigherthepriceshigherwillbetheaver-agenegativeorpositiveratingsforaservice

Word countThenumberofwordsinareviewmayserveasaproxyfortheamountofinformationinthereviewsHowevertheevidenceonthisaspectofWOMofalltypesisequivocalForinstanceconsumersgenerallyengageinpositiveWOMtoavoidfeelingsofguiltandas-sociationwithbadnewsandtoreducecognitivedissonanceOntheotherhandacustomerwhoishighlydisappointedwillprovideapersuasiveargumenttomakesurethatothercustomersaredissuadedfromrentingthesamepropertyResearchonconventionalwordofmouthhasofferedampleevidencetosuggestthatdissatisfiedcustomersengageintwotothreetimesasmuchWOMassatisfiedcustomers14Webelievethatthislogicextendstoonlinereviewsdespitethefactthatexistingstudiesareequivocalatbest15Thereforeitcanbeassumedthatreviewswitheitherextremelypositiveorextremelynegativeratingswillhavegreaterwordcountasopposedtomoremoderatereviews

TotesttheseassumptionsweappliedANOVAwithpriceandwordcountasdependentvariablesandthereviewratingasfactorsAsillustratedinExhibits5and6theresultsshowthatpriceandwordcountshowasignificantnegativecorrelationwithstarratingThatisreviewswithlowerratingsaretypicallyassociatedwithgreaterwordcountandrepresentpropertiesthatchargehighrates

TheresultonthewordcountsuggeststhatconsumerswhoareunhappywitharentalpropertytendtowritemoretoexpresstheirdissatisfactionAsshowninthisexhibitlowerratingsareassociatedwithhighernumberofwordsinthereviewsandviceversaSimilarlypropertypriceseemstobestronglyassociatedwithnegativewordofmouthsup-portingtheprinciplethathighercostincreasestheconsum-ersrsquoinvolvementinthepurchase

12BeardenWOandMasonJB(1984)AninvestigationofinfluencesonconsumercomplaintreportsAdvancesinConsumerResearch11490-49513WetzerIMZeelenbergMandPietersR(2007)ldquoNevereatinthatrestaurantIdidrdquoExploringwhypeopleengageinnegativeword-of-mouthcommunicationPsychology and Marketing24661-68014SchlossbergH(1991)CustomersatisfactionNotafadbutawayoflifeMarketing News25andWestbrookR(1987)ProductConsump-tion-basedAffectiveResponsesandPost-purchaseProcessesJournal of Marketing Research 24258-27015AlexMSusskindldquoIToldYouSoRestaurantCustomersrsquoofWord-of-MouthCommunicationPatternsrdquoCornell Hotel and Restaurant Adminis-tration QuarterlyVol43No2(April2002)pp75-85

Exhibit 5

Differences in word count across star ratings

Exhibit 6

Differences in average price per day across ratings

Aver

age

pric

e pe

r day

of t

he re

ntal

pro

pert

y

overall ldquostar ratingrdquo

1504

1563

1804

19861987

Aver

age

wor

d co

unt o

f rev

iew

overall ldquostar ratingrdquo

836837

121

15751543

160

140

120

100

80

200

190

180

170

160

150

1 2 3 4 5

1 2 3 4 5

12 TheCenterforHospitalityResearchbullCornellUniversity

thereviewersThesecanbeclassifiedas(a)attributesorfea-turesofthepropertythatfigureprominentlyandcontributetothepositiveornegativeratings(egroomsbeddishesfurnitureorservice)and(b)thepositiveornegativewordsthatrepresentthesentimentofthereviewer(eggreatamazingdirtyandwonderful)Theresultsofthequalitativeanalysisareasfollows

Syntactic featuresTheoveralltoneofthelowreviewsisratherargumentativeThereviewerstendtobeforcefulinpresentingtheirpointofviewTheylayoutagreatamountofreasoningandprovidespecificexamplestodrivehometheirargumentThisisreflectedinthegreaterpercentageofnegativeconnectorwords(34)suchasbut however as ifandeven thoughItisinterestingtonotethatreviewersinthiscategoryspeakdirectlytotheotherconsumersnotablythroughgreateruseofwordssuchasyou(46)andlooktoinfluencetheirdecisionmaking

InthehighreviewsbycontrasttheoveralltonetendstobeenunciativeandthereviewersusethetexttomerelyreinforcetheirnumericalratingsusingfewerwordsonaveragethanthenegativereviewersInthesereviewstheconsumersfocusontheirpersonalexperiencesandappealtotheaudiencethroughemotionsratherthanlogicandreasoningasevidentfromgreateruseofpositiveadjectives

QualitativeAnalysisoftheReviewTextTheaboveanalysisemphasizestheimportanceofthetextcontentinconsumerreviewsWhilenumericalratingsallowthereviewerstoratetheserviceprovideronvariousattributesitisthetextthatprovidesthemwiththeopportu-nitytoarticulatethenuancesoftheiroverallexperienceandconveyusefulinformationaboutaserviceproviderrsquostransac-tionsandservicecapabilitiesThereforesemanticanalysisofthetextcontentprovidesuswithanopportunitytothethirdquestionpresentedattheoutsetofthisreportwhatdoconsumerstalkaboutintheirreviews

Tothisendweanalyzed600eachofrandomlyselectedhighandlowreviewsusingsemanticprocessingtechniquesWeusedcenteringresonanceanalysis(CRA)amodeofnet-work-basedtextanalysisthatrepresentsthecontentoflargesetsoftextsbyidentifyingthemostimportantwordsthatlinkotherwordsinthenetwork16OuranalysisexaminedsyntacticandsemanticfeaturesSyntacticfeaturesincludethepercentageofthewordsthatareopenclass(ienounsverbsadjectives)Semanticfeaturesrepresentthespecial-izedvocabularythatreflectstheoverallopinionexpressedby

16CormanSandDooleyK(2006)CrawdadTextAnalysisSystem12CrawdadTechnologiesLLCChandlerAZ

Prominently Mentioned Features Freq sample Comments

bedroom beds linen sheets 383

Also the same bedroom and bathroom did not have any door separating the bedroom from the bathroom no privacy

Sleeping arrangements were very awkward The listing sheet described 2 full beds We brought our own linens the beds were queen sized Had to purchase sheet

Kitchen Kitchenware Dishes Dishwasher 297

You must not forget to step up when yoursquore walking out of the kitchen into the dining room otherwise yoursquoll trip over the 3-foot step up

No potholders in the kitchen and it didnrsquot seem like there were as many kitchen supplies as before We just had to make do without

beach beaches beachfront 296

The week was the best vacation we have ever had as a family The beach house was a class act

Disappointment 1 It looks like this house is very close to beach but to get there you have to go around through public path

Pool Poolside 218

The pool was too dirty to use We checked in on Thursday and the pool was not cleaned until Saturday

Also it is inconvenient to have to call to get someone to open the pool Didnrsquot understand why Jacuzzi was open and not the pool

bathroom(s) bath bathtub shower 202

The next problem was in another bathroom (main floor)mdashthe wall heater did not work so showering in here was a bit chilly The next problem was in the ldquomasterrdquo bathroom (connected to the bedroom with the King bed) The sliding shower doors need to be totally replaced

Bathtub drains clogged in three second floor bathrooms

Exhibit 7

Top five prominent issues in low reviews

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 13

suchasgreat fantastic andawesome(45)andpersonalpronounssuchasweandI(54)

Semantic featuresExhibits7and8showthemainissuesoraspectsthatareprominentlydiscussedinthelowreviewsTheorderoftheseissuesisbasedonthestatisti-callysignificantfrequencyoftheiroccurrenceinthe1200reviews

WefurtherinvestigatedtheuseofvarioussentimentwordsInterestinglytheprominentissuesdiscussedinthepositivereviewssometimesdonotevenfigureintheratingscalesprovidedtothereviewersOveralltheresultsofthesemanticanalysiscanbesummarizedasfollows

Thepropertyanditsfeaturesaremorecommonlycon-nectedwithextremelypositivesentimentwordssuchasgreat wonderfulandamazinginthehighreviewsthaninthelowreviewsIncomparisonthepropertiesandtheirfea-turesinthelowreviewsaremorelikelytobecharacterizedwithmoderatelypositivewordssuchasniceandokay

Negativereviewsreflecttheextentofexpectationndashdis-confirmationthattheconsumersfaceduringtheirstayintherentalpropertyThesereviewshaveagreaterproportionofwordsexpressingdisappointment(egnotasexpected)Furtherextremelynegativesentimentwords(egdiscom-fortfilthy)andadjectivesaremorelikelytobeassociatedwithfeaturesinlowreviewsthaninhighreviews

Almost40percentoftheconsumerswhowrotehighreviewsexpresstheirwillingnesstoreturntothepropertyasopposedtoonly9percentofthosewritingnegativereviewsFurthermorethan20percentoftheconsumersgivinghighreviewsexpresstheirstrongdesiretorecommendtheprop-ertytotheirfriendsandfamilymembersasopposedtoonly5percentinlowreviews

ReviewofFindingsOurexploratorystudyofconsumerreviewsofUSvacationhomesonReviewsitecomwashighlightedbyoverwhelm-inglypositivereviews

Theoveralldistributionofthereviewswasheav-ilyskewedtowardsthepositiveratingsMorecriticallythisfindingsuggeststhatthereisalackofmoderateandbalancedopinionsWetheorizethatthisiseitherduetopurchasingbias(sincecustomerswentthroughthetroubleofpurchasingandexperiencingthefacilitiestheytendtobepositivelybiasedtowardstheirpurchase)orunder-reportingbias(customersdonothavestrongenoughincentivestotakethetroubleofreportingtheiropinionsandthefewmotivatedindividualswhoprovidetheirreviewsbecomeaself-selectedsample)

Theoverall(mean)starratingtypicallyusedintravelerreviewsitesisprobablynotthemostaccurateindicatorofcustomersrsquoperceptionsofserviceprovidersrsquoqualityWe

Prominently Mentioned Features Freq sample Commentsbeach beaches beachfront 506 Perfect rental property for those who enjoy the activity of the beach and the serenity

of the Sound The house is just down the road from ocean

The large porch offered a beautiful view of the beach and comfortable new rocking chairs

Pool Poolside 218 The outdoor shower dipping pool and screen in porch were an added bonus Perfect house for a low key weekend

House was great for kids Pool was terrific and the playroom was perfectlocation 192 We had a wonderful time My wife loves fishing and the cottage was in a perfect

location to go fishing on the pier

The home is somewhat dated but the location was great We have vacationed in the Chatham area for many summers and this time it was great too

bedroom beds 186 The family suite is huge with a king size and twin bed and still room for a pack-n-play The kitchen and dining room were huge clean and nice

The only comments that I would say to the negative is the front bed room by the front door always seemed damp which seem to attract more bugs

Kitchen Kitchenware Dishes Dishwasher 178 All of the amenities in the house are upscalemdashloved the kitchen and plasma TVs

The kitchen had good basic utensils dishes and cooking pans

Exhibit 7

Top five prominent issues in high reviews

14 TheCenterforHospitalityResearchbullCornellUniversity

foundacorrelationbetweenthestarratingandratingsonotherpropertyattributesoflessthan40percentinmostcases

AhighwordcountinthereviewsiscorrelatedwithalowoverallratingThatisconsumerswhoareextremelydissatisfiedtendtoexpendmoretimeandenergyontheircritiqueofthepropertyitsamenitiesandservicesPriceofthepropertyalsoseemstopositivelycorrelatewithextremenegativereviewsperhapsbecauseahighpriceheightenstravelersrsquoexpectations

Numericalratingsdonotoftencapturetheactualsentimentandthevarietyofdimensionsonwhichtravel-ersevaluatepropertiesBecausenumericalratingsdonotcapturereviewersrsquoopinionsaboutspecificattributestheycanbedeceivingandmaynotrevealthetruequalityofthepropertyortheactualintentionofthereviewerWefoundforinstancethatconsumersvaluelocation(egclosenesstothebeach)andcleanliness(egswimmingpoolareas)morethansuchfeaturesascheck-inandserviceSimilarlyinthenegativereviewsissuessuchaslackofbedroomparapher-naliaorkitchenitemsseemtoberankedhigherinimpor-tancethanissuessuchaslocationandcheck-inThereforeevenwhenapropertyscoreshighonthecertaindimensionstheoverallratingmayremainlowbecauseotherfactorsovershadowthefavorableaspects

WhatItMeansforManagersWeseeimplicationsforonlinereviewsitesandbyexten-sionforproprietorsofvacationhomesandserviceopera-tionsgenerallyWithregardtothereviewsitesifconsumersconsistentlydetectafavorablebiasthereviewsitesmaylosethecriticalaspectsofcredibilityandtrustAlthoughconsumersaregenerallysmartandrationalandcanworkaroundobviousbiasesrecentevidencesuggeststhatcon-sumersaregraduallydefectingfromsomeonlinesitesduetounbalancedreviews17InthisregardsiteadministratorsaswellaspropertymanagersshouldtakestepstoattractthoseconsumerswhoarewillingtoprovideareasonablybalancedanalysisoftheamenitiesandservicesBythiswemeanreviewsthatpresentcomprehensiveandobjectiveanalysisoftheserviceproviderratherthanjustravesorrants

Tothatendsiteoperatorsshouldtakestepstoun-derstandcustomersrsquoreviewwritingbehaviorandtoat-tractmorebalancedreviewsOnewaytodothisistogivepersonalizedinformationtotheconsumerdemonstratingthecontributionmadebyhisorherreviewandthevalueofthereviewtostakeholdersForinstanceLinget alfoundapositivecorrelationbetweenwillingnesstowritereviewsandtheconsumersrsquoperceptionthattheircontributionswouldbeuniqueandhelpfultoothercustomersaswellasperusedby

17HarteveldtHHJohnsonCStarkEandGeldernKV(2009)Us-ingDigitalChannelsToCalmTheAngryTravelerForresterResearch

theserviceproviders18Simplyputthepropensitytowritereviewsincreaseswhentheconsumersfeelthattheirreviewisactuallyhelpingothers(boththeserviceproviderandotherconsumers)

WesuggestthatreviewsitesdevelopbettermethodstoaggregatesynthesizeandpublishspecificreviewcommentsaswellasnumericalratingsFirstmanagersshouldtakeintoconsiderationthelowcorrelationbetweentheoverallratingsandconsumersrsquoassessmentsofvariouspropertyattributes(aswellastheresultsfromthetextanalysis)Theimplicationofsuchaweakcorrelationisthatreviewsitesneedtoexpandonthelistofpossiblevariablessothattheratingsaccuratelyreflectconsumersrsquoneedsandexpectationsregardingserviceproviders

Furthersitemanagersmayhavetorethinkthecurrentpracticeofaveragingalltheratingsforagivenpropertybe-causesuchameanratingfailstotakeintoaccountthebiasesthatareinherentintheratingsystemsAsanexamplethisstudyfindssignificantunder-reportingbiasinthisonlinereviewsiteAcustomerrsquosdecisiontonotpostonlinereviewsmaymeanthelossofimportantinformationthatcanassisttheusersofthesitestomakemorereliableinferencesThemajorityoftodayrsquosfeedbackmechanismsdonotpubliclydisclosethenumberofsilenttransactions(ietransactionsforwhichnofeedbackwaspostedbycustomers)Suchinfor-mationshouldbecomeapartofaserviceproviderrsquosonlineprofileonratingsitesandotherfeedbackmechanismsItisimportanttoprovidemoreinformationandheuristicstohelptheconsumersnavigatethroughtheclutterandgettheinformationtheywantandseek

Itisimportanttoapplymodernsemanticanalysisandsentimentclassificationtechniquestostudythecontentofthereviewsandidentifyvariousdimensionsonwhichcon-sumersevaluatevacationhomesThedetailsprovidedinthereviewtextareinmanywaysabetterreflectionofcustomersatisfactionServiceproviderscanbetterpositionthemselvesusingthedimensionsuncoveredthroughtextanalysisandtargettheneedsandpreferencesoftheircustomers

TheresultsinthisstudyshowthatnegativereviewsaretypicallyassociatedwithagreaterwordcountthanpositivereviewsItisusefultofocusspecificallyonthesenegativereviewsbecausetheycanrevealpatternsofdeficienciesintheservicestandardsanddeliveryThesereviewsshouldbetreatedthesameascustomercommentcardsorletters19Furthercustomerswhowritesuchlengthynegativereviews

18LingKimberlyGerardBeenenPaulLudfordXiaoxingWangKevinChangXinLiDanCosleyDanFrankowskiLorenTerveenAlMRashidandRobertKrautldquoUsingSocialPsychologytoMotivateContributionstoOnlineCommunitiesrdquoJournal of Computer-Mediated Communication10no4(2005)19MilanR(2007)10thingsyoucandoinresponsetotravelerreviewsinHotelmarketingcom

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 15

aremostlikelytospreadnegativeword-of-mouthaboutthepropertyLengthyreviewsareexcellentsourcesofinforma-tionandshouldbeusedasthebasestounderstandthecomplainingbehavioroftheconsumer

WordcountmaybeusedasanadditionalsegmentationvariabletoidentifythoseconsumerswhoaremoreinvolvedandwillingtoprovidebalancedfeedbacktoboththereviewsiteaswellastopropertymanagementVariousotherresearchmethodssuchastelephoneinterviewsorfocusgroupsshouldbeusedtoelicitdetailedfeedbackfromsuchconsumersTheseactionsalsohelpinservicerecoverypost-purchaseengagementandrelationshipbuilding

OutlookWeforeseethreepotentialareasinvestigationthatcanaddvaluetocurrentbestpracticesWestilldonrsquotknowmuchaboutconsumersrsquomotivationtowriteonlinereviews20Recentstudiesidentifyvariousantecedentsofreviewwritingbutsufferfromsamplingbiassincetheysurveyaself-select-edsampleofconsumerswhoalreadyprovidedreviewsAstheresultsinthisstudyindicateunder-reportingbias(lackofmoderateandlowreviews)remainsprevalentinonlinereviewsitesFutureresearchshouldidentifysocio-psycho-logicalfactorsthatincreaseconsumersrsquopropensitytowriteonlinereviews

Futureresearchshouldexploretheimpactofthepres-ence(orlack)ofextremelynegativereviewswithinalargecorpusofpositivereviewsTheevidenceontheimpactofre-

20WangYandFesenmaierDR(2003)AssessingMotivationofCon-tributioninOnlineCommunitiesAnEmpiricalInvestigationofanOn-lineTravelCommunityElectronic Markets1333-45andYooKyungHandUlrikeGretzelldquoWhatMotivatesConsumerstoWriteOnlineTravelReviewsrdquoInformation Technology amp Tourism10No4(2008)283-95

viewvalence(negativeorpositive)isequivocalProponentsofconfirmatorybiassuggestthatconsumerslookforaffir-mativeevidencesupportingaproductchoicealreadymadeIfthatisthecasepositivereviewsaremorelikelytohaveagreatereffectonconsumeractionsOntheotherhandthenotionofnegativitybiassuggeststhatwhenconsumersareneutralnegativereviewstendtobecomemoresalientthanpositivereviews21Inthepresenceoflargenumbersofposi-tivereviewsconsumersspecificallyseeknegativereviewsthattheyfeelwillhelpthemidentifyspecificproblemswithaserviceUnderstandinghowconsumersreconciletheinformationprovidedbybothnegativeandpositivereviewshasimplicationsforthedesignandmanagementofreviewsites

Finallyareviewrsquoswordcounthasimportantimplica-tionsforconsumersrsquotrustinthereviewsWhenconsumersarewillingtoreadandcompareopen-endedcommentsfromotherconsumerstheamountofinformationinareviewcanmatterWOMforhighinvolvementproductsismorepersuasiveandincreasesthedecisionmakerrsquosconfi-dencewhenthemessagesenderprovidesgreaternumberofreasonsfortheiroverallratingLongerreviewsprovidemoreinformationandarelikelytobeperceivedasmorehelpfulandpersuasivethanshorterreviewsThewaysinwhichareviewiswrittenframedandpresentedisaninterestingfactorwhichhasrarelybeeninvestigatedandhasimportantimplicationsparticularlygivenourfindingthattheamountofinformationinareview(wordcount)variessignificantlywiththeratingsofthereviewn

21BaSandPavlouPA(2002)EvidenceoftheEffectofTrustBuildingTechnologyinElectronicMarketsPricePremiumsandBuyerBehaviorMIS Quarterly26243-268

Cornell Hospitality Quarterlyhttpcqxsagepubcom

2011 ReportsVol11No15DesigningaSelf-HealingServiceSystemAnIntegrativeModelbyRobertCFordandMichaelCSturman

Vol11No14ReversingtheGreenBacklashWhyLargeHospitalityCompaniesShouldWelcomeCrediblyGreenCompetitorsbyMichaelGiebelhausenPhDandHaeEunHelenChunPhD

Vol11No13DevelopingaSustainabilityMeasurementFrameworkforHotelsTowardanIndustry-wideReportingStructurebyEricRicaurte

Vol11No12CreatingValueforWomenBusinessTravelersFocusingonEmotionalOutcomesbyJudiBrownellPhD

Vol11No11CustomerLoyaltyANewLookattheBenefitsofImprovingSegmentationEffortswithRewardsProgramsbyClayVoorheesPhDMichaelMcCallPhDandRogerCalantonePhD

Vol11No10CustomerPerceptionsofElectronicFoodOrderingbySherylEKimes

Vol11No92011TravelIndustryBenchmarkingStatusofSeniorDestinationandLodgingMarketingExecutivesbyRohitVermaPhDandKenMcGill

Vol11No8SearchOTAsandOnlineBookingAnExpandedAnalysisoftheBillboardEffectbyChrisAndersonPhD

Vol11No7OnlineMobileandTextFoodOrderingintheUSRestaurantIndustrybySherylEKimesPhDandPhilippFLaqueacute

Vol11No6HotelGuestsrsquoReactionstoGuestRoomSustainabilityInitiativesbyAlexSusskindPhDandRohitVermaPhD

Vol11No5TheImpactofTerrorismandEconomicShocksonUSHotelsbyCathyAEnzRenaacutetaKosovaacuteandMarkLomannoVol11No4ImplementingHumanResourceInnovationsThreeSuccessStoriesfromtheServiceIndustrybyJustinSunandKateWalshPhD

Vol11No3Compendium2011

Vol11No2PositioningaPlaceDevelopingaCompellingDestinationBrandbyRobertJKwortnikPhDandEthanHawkesMBA

Vol11No1TheImpactofHealthInsuranceonEmployeeJobAnxietyWithdrawalBehaviorsandTaskPerformancebySeanWayPhDBillCarrollPhDAlexSusskindPhDandJoeCYLeng

2011 Hospitality ToolsVol2No1TheGameHasChangedANewParadigmforStakeholderEngagementbyMaryBethMcEuen

2011 Industry PerspectivesNo7MegaTips2TwentyTestedTechniquesforIncreasingYourTipsbyMichaelLynn

2011 ProceedingsVol3No5BuildingBrandsintheInternetAgeAnalyticsLoyaltyandCommunicationbyGlennWithiam

Vol3No4BraveNewWorldOnlineHotelDistributionbyGlennWithiam

Vol3No3SocialMediaandtheHospitalityIndustryHoldingtheTigerbytheTailbyGlennWithiam

Vol3No2TheChallengeofHotelandRestaurantSustainabilityFindingProfitinldquoBeingGreenrdquobyGlennWithiam

Vol3No1CautiousOptimismCHRSExaminesHospitalityIndustryTrendsbyGlennWithiam

2010 ReportsVol10No18HowTravelersUseOnlineandSocialMediaChannelstoMakeHotel-choiceDecisionsbyLauraMcCarthyDebraStockandRohitVermaPhD

Vol10No17PublicorPrivateTheHospitalityInvestmentDecisionbyQingzhongMaPhDandAthenaWeiZhangPhD

Vol10No16BestPracticesinSearchEngineMarketingandOptimizationTheCaseoftheStJamesHotelbyGregBodenlcosVictorBogertDanGordonCarterHearneandChrisKAndersonPhD

Vol10No15TheImpactofPrix Fixe MenuPriceFormatsonGuestsrsquoDealPerceptionbyShuoWangandMichaelLynnPhD

Cornell Center for Hospitality Research

Publication Indexwwwchrcornelledu

The Executive Path Hospitality Leadership Through Learning

Complete program information and applications available online

wwwhotelschoolcornelleduexecedPhone + 1 607 255 4919 Email exec_ed_hotelcornelledu

Professionals from around the world are invited to attend 3-day 10-day or online courses at the worldrsquos leading institute for hospitality management education in

Visit our website to apply

Explore develop and apply ideas with global hospitality leaders and

expert Cornell professors

Success

AdvancingBusiness

andPersonal

bull Strategic Leadershipbull Financebull Foodservicebull Human Resources

bull Marketingbull Operationsbull Real Estate

No2RetaliationWhyanIncreaseinClaimsDoesNotMeantheSkyIsFallingbyDavidSherwynJDandGreggGilmanJD

2009 ToolsToolNo12MeasuringtheDiningExperienceTheCaseofVitaNovabyKeshPrasadandFredJDeMiccoPhD

Cornell Center for Hospitality ResearchPublication Index

The Executive Path Hospitality Leadership Through Learning

Complete program information and applications available online

wwwhotelschoolcornelleduexecedPhone + 1 607 255 4919 Email exec_ed_hotelcornelledu

Professionals from around the world are invited to attend 3-day 10-day or online courses at the worldrsquos leading institute for hospitality management education in

Visit our website to apply

Explore develop and apply ideas with global hospitality leaders and

expert Cornell professors

Success

AdvancingBusiness

andPersonal

bull Strategic Leadershipbull Financebull Foodservicebull Human Resources

bull Marketingbull Operationsbull Real Estate

visit our website to apply

wwwchrcornell edu

Page 11: Unscrambling the Puzzling Matter of Online Consumer ... · Philips Hospitality Americas Chris Proulx, CEO, eCornell & Executive Education Carolyn D. Richmond, Partner, Hospitality

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 11

withthecostoftheproduct12Giventhegreaterperceivedriskinservicescustomerswouldseekmoretrustworthyser-viceproviderswithagoodreputationofsatisfyingcustomerneeds13ThisinturnincreasestheexpectationsandleadstomoreextremereactionsfromtheconsumersHenceonecanspeculatethatthehigherthepriceshigherwillbetheaver-agenegativeorpositiveratingsforaservice

Word countThenumberofwordsinareviewmayserveasaproxyfortheamountofinformationinthereviewsHowevertheevidenceonthisaspectofWOMofalltypesisequivocalForinstanceconsumersgenerallyengageinpositiveWOMtoavoidfeelingsofguiltandas-sociationwithbadnewsandtoreducecognitivedissonanceOntheotherhandacustomerwhoishighlydisappointedwillprovideapersuasiveargumenttomakesurethatothercustomersaredissuadedfromrentingthesamepropertyResearchonconventionalwordofmouthhasofferedampleevidencetosuggestthatdissatisfiedcustomersengageintwotothreetimesasmuchWOMassatisfiedcustomers14Webelievethatthislogicextendstoonlinereviewsdespitethefactthatexistingstudiesareequivocalatbest15Thereforeitcanbeassumedthatreviewswitheitherextremelypositiveorextremelynegativeratingswillhavegreaterwordcountasopposedtomoremoderatereviews

TotesttheseassumptionsweappliedANOVAwithpriceandwordcountasdependentvariablesandthereviewratingasfactorsAsillustratedinExhibits5and6theresultsshowthatpriceandwordcountshowasignificantnegativecorrelationwithstarratingThatisreviewswithlowerratingsaretypicallyassociatedwithgreaterwordcountandrepresentpropertiesthatchargehighrates

TheresultonthewordcountsuggeststhatconsumerswhoareunhappywitharentalpropertytendtowritemoretoexpresstheirdissatisfactionAsshowninthisexhibitlowerratingsareassociatedwithhighernumberofwordsinthereviewsandviceversaSimilarlypropertypriceseemstobestronglyassociatedwithnegativewordofmouthsup-portingtheprinciplethathighercostincreasestheconsum-ersrsquoinvolvementinthepurchase

12BeardenWOandMasonJB(1984)AninvestigationofinfluencesonconsumercomplaintreportsAdvancesinConsumerResearch11490-49513WetzerIMZeelenbergMandPietersR(2007)ldquoNevereatinthatrestaurantIdidrdquoExploringwhypeopleengageinnegativeword-of-mouthcommunicationPsychology and Marketing24661-68014SchlossbergH(1991)CustomersatisfactionNotafadbutawayoflifeMarketing News25andWestbrookR(1987)ProductConsump-tion-basedAffectiveResponsesandPost-purchaseProcessesJournal of Marketing Research 24258-27015AlexMSusskindldquoIToldYouSoRestaurantCustomersrsquoofWord-of-MouthCommunicationPatternsrdquoCornell Hotel and Restaurant Adminis-tration QuarterlyVol43No2(April2002)pp75-85

Exhibit 5

Differences in word count across star ratings

Exhibit 6

Differences in average price per day across ratings

Aver

age

pric

e pe

r day

of t

he re

ntal

pro

pert

y

overall ldquostar ratingrdquo

1504

1563

1804

19861987

Aver

age

wor

d co

unt o

f rev

iew

overall ldquostar ratingrdquo

836837

121

15751543

160

140

120

100

80

200

190

180

170

160

150

1 2 3 4 5

1 2 3 4 5

12 TheCenterforHospitalityResearchbullCornellUniversity

thereviewersThesecanbeclassifiedas(a)attributesorfea-turesofthepropertythatfigureprominentlyandcontributetothepositiveornegativeratings(egroomsbeddishesfurnitureorservice)and(b)thepositiveornegativewordsthatrepresentthesentimentofthereviewer(eggreatamazingdirtyandwonderful)Theresultsofthequalitativeanalysisareasfollows

Syntactic featuresTheoveralltoneofthelowreviewsisratherargumentativeThereviewerstendtobeforcefulinpresentingtheirpointofviewTheylayoutagreatamountofreasoningandprovidespecificexamplestodrivehometheirargumentThisisreflectedinthegreaterpercentageofnegativeconnectorwords(34)suchasbut however as ifandeven thoughItisinterestingtonotethatreviewersinthiscategoryspeakdirectlytotheotherconsumersnotablythroughgreateruseofwordssuchasyou(46)andlooktoinfluencetheirdecisionmaking

InthehighreviewsbycontrasttheoveralltonetendstobeenunciativeandthereviewersusethetexttomerelyreinforcetheirnumericalratingsusingfewerwordsonaveragethanthenegativereviewersInthesereviewstheconsumersfocusontheirpersonalexperiencesandappealtotheaudiencethroughemotionsratherthanlogicandreasoningasevidentfromgreateruseofpositiveadjectives

QualitativeAnalysisoftheReviewTextTheaboveanalysisemphasizestheimportanceofthetextcontentinconsumerreviewsWhilenumericalratingsallowthereviewerstoratetheserviceprovideronvariousattributesitisthetextthatprovidesthemwiththeopportu-nitytoarticulatethenuancesoftheiroverallexperienceandconveyusefulinformationaboutaserviceproviderrsquostransac-tionsandservicecapabilitiesThereforesemanticanalysisofthetextcontentprovidesuswithanopportunitytothethirdquestionpresentedattheoutsetofthisreportwhatdoconsumerstalkaboutintheirreviews

Tothisendweanalyzed600eachofrandomlyselectedhighandlowreviewsusingsemanticprocessingtechniquesWeusedcenteringresonanceanalysis(CRA)amodeofnet-work-basedtextanalysisthatrepresentsthecontentoflargesetsoftextsbyidentifyingthemostimportantwordsthatlinkotherwordsinthenetwork16OuranalysisexaminedsyntacticandsemanticfeaturesSyntacticfeaturesincludethepercentageofthewordsthatareopenclass(ienounsverbsadjectives)Semanticfeaturesrepresentthespecial-izedvocabularythatreflectstheoverallopinionexpressedby

16CormanSandDooleyK(2006)CrawdadTextAnalysisSystem12CrawdadTechnologiesLLCChandlerAZ

Prominently Mentioned Features Freq sample Comments

bedroom beds linen sheets 383

Also the same bedroom and bathroom did not have any door separating the bedroom from the bathroom no privacy

Sleeping arrangements were very awkward The listing sheet described 2 full beds We brought our own linens the beds were queen sized Had to purchase sheet

Kitchen Kitchenware Dishes Dishwasher 297

You must not forget to step up when yoursquore walking out of the kitchen into the dining room otherwise yoursquoll trip over the 3-foot step up

No potholders in the kitchen and it didnrsquot seem like there were as many kitchen supplies as before We just had to make do without

beach beaches beachfront 296

The week was the best vacation we have ever had as a family The beach house was a class act

Disappointment 1 It looks like this house is very close to beach but to get there you have to go around through public path

Pool Poolside 218

The pool was too dirty to use We checked in on Thursday and the pool was not cleaned until Saturday

Also it is inconvenient to have to call to get someone to open the pool Didnrsquot understand why Jacuzzi was open and not the pool

bathroom(s) bath bathtub shower 202

The next problem was in another bathroom (main floor)mdashthe wall heater did not work so showering in here was a bit chilly The next problem was in the ldquomasterrdquo bathroom (connected to the bedroom with the King bed) The sliding shower doors need to be totally replaced

Bathtub drains clogged in three second floor bathrooms

Exhibit 7

Top five prominent issues in low reviews

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 13

suchasgreat fantastic andawesome(45)andpersonalpronounssuchasweandI(54)

Semantic featuresExhibits7and8showthemainissuesoraspectsthatareprominentlydiscussedinthelowreviewsTheorderoftheseissuesisbasedonthestatisti-callysignificantfrequencyoftheiroccurrenceinthe1200reviews

WefurtherinvestigatedtheuseofvarioussentimentwordsInterestinglytheprominentissuesdiscussedinthepositivereviewssometimesdonotevenfigureintheratingscalesprovidedtothereviewersOveralltheresultsofthesemanticanalysiscanbesummarizedasfollows

Thepropertyanditsfeaturesaremorecommonlycon-nectedwithextremelypositivesentimentwordssuchasgreat wonderfulandamazinginthehighreviewsthaninthelowreviewsIncomparisonthepropertiesandtheirfea-turesinthelowreviewsaremorelikelytobecharacterizedwithmoderatelypositivewordssuchasniceandokay

Negativereviewsreflecttheextentofexpectationndashdis-confirmationthattheconsumersfaceduringtheirstayintherentalpropertyThesereviewshaveagreaterproportionofwordsexpressingdisappointment(egnotasexpected)Furtherextremelynegativesentimentwords(egdiscom-fortfilthy)andadjectivesaremorelikelytobeassociatedwithfeaturesinlowreviewsthaninhighreviews

Almost40percentoftheconsumerswhowrotehighreviewsexpresstheirwillingnesstoreturntothepropertyasopposedtoonly9percentofthosewritingnegativereviewsFurthermorethan20percentoftheconsumersgivinghighreviewsexpresstheirstrongdesiretorecommendtheprop-ertytotheirfriendsandfamilymembersasopposedtoonly5percentinlowreviews

ReviewofFindingsOurexploratorystudyofconsumerreviewsofUSvacationhomesonReviewsitecomwashighlightedbyoverwhelm-inglypositivereviews

Theoveralldistributionofthereviewswasheav-ilyskewedtowardsthepositiveratingsMorecriticallythisfindingsuggeststhatthereisalackofmoderateandbalancedopinionsWetheorizethatthisiseitherduetopurchasingbias(sincecustomerswentthroughthetroubleofpurchasingandexperiencingthefacilitiestheytendtobepositivelybiasedtowardstheirpurchase)orunder-reportingbias(customersdonothavestrongenoughincentivestotakethetroubleofreportingtheiropinionsandthefewmotivatedindividualswhoprovidetheirreviewsbecomeaself-selectedsample)

Theoverall(mean)starratingtypicallyusedintravelerreviewsitesisprobablynotthemostaccurateindicatorofcustomersrsquoperceptionsofserviceprovidersrsquoqualityWe

Prominently Mentioned Features Freq sample Commentsbeach beaches beachfront 506 Perfect rental property for those who enjoy the activity of the beach and the serenity

of the Sound The house is just down the road from ocean

The large porch offered a beautiful view of the beach and comfortable new rocking chairs

Pool Poolside 218 The outdoor shower dipping pool and screen in porch were an added bonus Perfect house for a low key weekend

House was great for kids Pool was terrific and the playroom was perfectlocation 192 We had a wonderful time My wife loves fishing and the cottage was in a perfect

location to go fishing on the pier

The home is somewhat dated but the location was great We have vacationed in the Chatham area for many summers and this time it was great too

bedroom beds 186 The family suite is huge with a king size and twin bed and still room for a pack-n-play The kitchen and dining room were huge clean and nice

The only comments that I would say to the negative is the front bed room by the front door always seemed damp which seem to attract more bugs

Kitchen Kitchenware Dishes Dishwasher 178 All of the amenities in the house are upscalemdashloved the kitchen and plasma TVs

The kitchen had good basic utensils dishes and cooking pans

Exhibit 7

Top five prominent issues in high reviews

14 TheCenterforHospitalityResearchbullCornellUniversity

foundacorrelationbetweenthestarratingandratingsonotherpropertyattributesoflessthan40percentinmostcases

AhighwordcountinthereviewsiscorrelatedwithalowoverallratingThatisconsumerswhoareextremelydissatisfiedtendtoexpendmoretimeandenergyontheircritiqueofthepropertyitsamenitiesandservicesPriceofthepropertyalsoseemstopositivelycorrelatewithextremenegativereviewsperhapsbecauseahighpriceheightenstravelersrsquoexpectations

Numericalratingsdonotoftencapturetheactualsentimentandthevarietyofdimensionsonwhichtravel-ersevaluatepropertiesBecausenumericalratingsdonotcapturereviewersrsquoopinionsaboutspecificattributestheycanbedeceivingandmaynotrevealthetruequalityofthepropertyortheactualintentionofthereviewerWefoundforinstancethatconsumersvaluelocation(egclosenesstothebeach)andcleanliness(egswimmingpoolareas)morethansuchfeaturesascheck-inandserviceSimilarlyinthenegativereviewsissuessuchaslackofbedroomparapher-naliaorkitchenitemsseemtoberankedhigherinimpor-tancethanissuessuchaslocationandcheck-inThereforeevenwhenapropertyscoreshighonthecertaindimensionstheoverallratingmayremainlowbecauseotherfactorsovershadowthefavorableaspects

WhatItMeansforManagersWeseeimplicationsforonlinereviewsitesandbyexten-sionforproprietorsofvacationhomesandserviceopera-tionsgenerallyWithregardtothereviewsitesifconsumersconsistentlydetectafavorablebiasthereviewsitesmaylosethecriticalaspectsofcredibilityandtrustAlthoughconsumersaregenerallysmartandrationalandcanworkaroundobviousbiasesrecentevidencesuggeststhatcon-sumersaregraduallydefectingfromsomeonlinesitesduetounbalancedreviews17InthisregardsiteadministratorsaswellaspropertymanagersshouldtakestepstoattractthoseconsumerswhoarewillingtoprovideareasonablybalancedanalysisoftheamenitiesandservicesBythiswemeanreviewsthatpresentcomprehensiveandobjectiveanalysisoftheserviceproviderratherthanjustravesorrants

Tothatendsiteoperatorsshouldtakestepstoun-derstandcustomersrsquoreviewwritingbehaviorandtoat-tractmorebalancedreviewsOnewaytodothisistogivepersonalizedinformationtotheconsumerdemonstratingthecontributionmadebyhisorherreviewandthevalueofthereviewtostakeholdersForinstanceLinget alfoundapositivecorrelationbetweenwillingnesstowritereviewsandtheconsumersrsquoperceptionthattheircontributionswouldbeuniqueandhelpfultoothercustomersaswellasperusedby

17HarteveldtHHJohnsonCStarkEandGeldernKV(2009)Us-ingDigitalChannelsToCalmTheAngryTravelerForresterResearch

theserviceproviders18Simplyputthepropensitytowritereviewsincreaseswhentheconsumersfeelthattheirreviewisactuallyhelpingothers(boththeserviceproviderandotherconsumers)

WesuggestthatreviewsitesdevelopbettermethodstoaggregatesynthesizeandpublishspecificreviewcommentsaswellasnumericalratingsFirstmanagersshouldtakeintoconsiderationthelowcorrelationbetweentheoverallratingsandconsumersrsquoassessmentsofvariouspropertyattributes(aswellastheresultsfromthetextanalysis)Theimplicationofsuchaweakcorrelationisthatreviewsitesneedtoexpandonthelistofpossiblevariablessothattheratingsaccuratelyreflectconsumersrsquoneedsandexpectationsregardingserviceproviders

Furthersitemanagersmayhavetorethinkthecurrentpracticeofaveragingalltheratingsforagivenpropertybe-causesuchameanratingfailstotakeintoaccountthebiasesthatareinherentintheratingsystemsAsanexamplethisstudyfindssignificantunder-reportingbiasinthisonlinereviewsiteAcustomerrsquosdecisiontonotpostonlinereviewsmaymeanthelossofimportantinformationthatcanassisttheusersofthesitestomakemorereliableinferencesThemajorityoftodayrsquosfeedbackmechanismsdonotpubliclydisclosethenumberofsilenttransactions(ietransactionsforwhichnofeedbackwaspostedbycustomers)Suchinfor-mationshouldbecomeapartofaserviceproviderrsquosonlineprofileonratingsitesandotherfeedbackmechanismsItisimportanttoprovidemoreinformationandheuristicstohelptheconsumersnavigatethroughtheclutterandgettheinformationtheywantandseek

Itisimportanttoapplymodernsemanticanalysisandsentimentclassificationtechniquestostudythecontentofthereviewsandidentifyvariousdimensionsonwhichcon-sumersevaluatevacationhomesThedetailsprovidedinthereviewtextareinmanywaysabetterreflectionofcustomersatisfactionServiceproviderscanbetterpositionthemselvesusingthedimensionsuncoveredthroughtextanalysisandtargettheneedsandpreferencesoftheircustomers

TheresultsinthisstudyshowthatnegativereviewsaretypicallyassociatedwithagreaterwordcountthanpositivereviewsItisusefultofocusspecificallyonthesenegativereviewsbecausetheycanrevealpatternsofdeficienciesintheservicestandardsanddeliveryThesereviewsshouldbetreatedthesameascustomercommentcardsorletters19Furthercustomerswhowritesuchlengthynegativereviews

18LingKimberlyGerardBeenenPaulLudfordXiaoxingWangKevinChangXinLiDanCosleyDanFrankowskiLorenTerveenAlMRashidandRobertKrautldquoUsingSocialPsychologytoMotivateContributionstoOnlineCommunitiesrdquoJournal of Computer-Mediated Communication10no4(2005)19MilanR(2007)10thingsyoucandoinresponsetotravelerreviewsinHotelmarketingcom

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 15

aremostlikelytospreadnegativeword-of-mouthaboutthepropertyLengthyreviewsareexcellentsourcesofinforma-tionandshouldbeusedasthebasestounderstandthecomplainingbehavioroftheconsumer

WordcountmaybeusedasanadditionalsegmentationvariabletoidentifythoseconsumerswhoaremoreinvolvedandwillingtoprovidebalancedfeedbacktoboththereviewsiteaswellastopropertymanagementVariousotherresearchmethodssuchastelephoneinterviewsorfocusgroupsshouldbeusedtoelicitdetailedfeedbackfromsuchconsumersTheseactionsalsohelpinservicerecoverypost-purchaseengagementandrelationshipbuilding

OutlookWeforeseethreepotentialareasinvestigationthatcanaddvaluetocurrentbestpracticesWestilldonrsquotknowmuchaboutconsumersrsquomotivationtowriteonlinereviews20Recentstudiesidentifyvariousantecedentsofreviewwritingbutsufferfromsamplingbiassincetheysurveyaself-select-edsampleofconsumerswhoalreadyprovidedreviewsAstheresultsinthisstudyindicateunder-reportingbias(lackofmoderateandlowreviews)remainsprevalentinonlinereviewsitesFutureresearchshouldidentifysocio-psycho-logicalfactorsthatincreaseconsumersrsquopropensitytowriteonlinereviews

Futureresearchshouldexploretheimpactofthepres-ence(orlack)ofextremelynegativereviewswithinalargecorpusofpositivereviewsTheevidenceontheimpactofre-

20WangYandFesenmaierDR(2003)AssessingMotivationofCon-tributioninOnlineCommunitiesAnEmpiricalInvestigationofanOn-lineTravelCommunityElectronic Markets1333-45andYooKyungHandUlrikeGretzelldquoWhatMotivatesConsumerstoWriteOnlineTravelReviewsrdquoInformation Technology amp Tourism10No4(2008)283-95

viewvalence(negativeorpositive)isequivocalProponentsofconfirmatorybiassuggestthatconsumerslookforaffir-mativeevidencesupportingaproductchoicealreadymadeIfthatisthecasepositivereviewsaremorelikelytohaveagreatereffectonconsumeractionsOntheotherhandthenotionofnegativitybiassuggeststhatwhenconsumersareneutralnegativereviewstendtobecomemoresalientthanpositivereviews21Inthepresenceoflargenumbersofposi-tivereviewsconsumersspecificallyseeknegativereviewsthattheyfeelwillhelpthemidentifyspecificproblemswithaserviceUnderstandinghowconsumersreconciletheinformationprovidedbybothnegativeandpositivereviewshasimplicationsforthedesignandmanagementofreviewsites

Finallyareviewrsquoswordcounthasimportantimplica-tionsforconsumersrsquotrustinthereviewsWhenconsumersarewillingtoreadandcompareopen-endedcommentsfromotherconsumerstheamountofinformationinareviewcanmatterWOMforhighinvolvementproductsismorepersuasiveandincreasesthedecisionmakerrsquosconfi-dencewhenthemessagesenderprovidesgreaternumberofreasonsfortheiroverallratingLongerreviewsprovidemoreinformationandarelikelytobeperceivedasmorehelpfulandpersuasivethanshorterreviewsThewaysinwhichareviewiswrittenframedandpresentedisaninterestingfactorwhichhasrarelybeeninvestigatedandhasimportantimplicationsparticularlygivenourfindingthattheamountofinformationinareview(wordcount)variessignificantlywiththeratingsofthereviewn

21BaSandPavlouPA(2002)EvidenceoftheEffectofTrustBuildingTechnologyinElectronicMarketsPricePremiumsandBuyerBehaviorMIS Quarterly26243-268

Cornell Hospitality Quarterlyhttpcqxsagepubcom

2011 ReportsVol11No15DesigningaSelf-HealingServiceSystemAnIntegrativeModelbyRobertCFordandMichaelCSturman

Vol11No14ReversingtheGreenBacklashWhyLargeHospitalityCompaniesShouldWelcomeCrediblyGreenCompetitorsbyMichaelGiebelhausenPhDandHaeEunHelenChunPhD

Vol11No13DevelopingaSustainabilityMeasurementFrameworkforHotelsTowardanIndustry-wideReportingStructurebyEricRicaurte

Vol11No12CreatingValueforWomenBusinessTravelersFocusingonEmotionalOutcomesbyJudiBrownellPhD

Vol11No11CustomerLoyaltyANewLookattheBenefitsofImprovingSegmentationEffortswithRewardsProgramsbyClayVoorheesPhDMichaelMcCallPhDandRogerCalantonePhD

Vol11No10CustomerPerceptionsofElectronicFoodOrderingbySherylEKimes

Vol11No92011TravelIndustryBenchmarkingStatusofSeniorDestinationandLodgingMarketingExecutivesbyRohitVermaPhDandKenMcGill

Vol11No8SearchOTAsandOnlineBookingAnExpandedAnalysisoftheBillboardEffectbyChrisAndersonPhD

Vol11No7OnlineMobileandTextFoodOrderingintheUSRestaurantIndustrybySherylEKimesPhDandPhilippFLaqueacute

Vol11No6HotelGuestsrsquoReactionstoGuestRoomSustainabilityInitiativesbyAlexSusskindPhDandRohitVermaPhD

Vol11No5TheImpactofTerrorismandEconomicShocksonUSHotelsbyCathyAEnzRenaacutetaKosovaacuteandMarkLomannoVol11No4ImplementingHumanResourceInnovationsThreeSuccessStoriesfromtheServiceIndustrybyJustinSunandKateWalshPhD

Vol11No3Compendium2011

Vol11No2PositioningaPlaceDevelopingaCompellingDestinationBrandbyRobertJKwortnikPhDandEthanHawkesMBA

Vol11No1TheImpactofHealthInsuranceonEmployeeJobAnxietyWithdrawalBehaviorsandTaskPerformancebySeanWayPhDBillCarrollPhDAlexSusskindPhDandJoeCYLeng

2011 Hospitality ToolsVol2No1TheGameHasChangedANewParadigmforStakeholderEngagementbyMaryBethMcEuen

2011 Industry PerspectivesNo7MegaTips2TwentyTestedTechniquesforIncreasingYourTipsbyMichaelLynn

2011 ProceedingsVol3No5BuildingBrandsintheInternetAgeAnalyticsLoyaltyandCommunicationbyGlennWithiam

Vol3No4BraveNewWorldOnlineHotelDistributionbyGlennWithiam

Vol3No3SocialMediaandtheHospitalityIndustryHoldingtheTigerbytheTailbyGlennWithiam

Vol3No2TheChallengeofHotelandRestaurantSustainabilityFindingProfitinldquoBeingGreenrdquobyGlennWithiam

Vol3No1CautiousOptimismCHRSExaminesHospitalityIndustryTrendsbyGlennWithiam

2010 ReportsVol10No18HowTravelersUseOnlineandSocialMediaChannelstoMakeHotel-choiceDecisionsbyLauraMcCarthyDebraStockandRohitVermaPhD

Vol10No17PublicorPrivateTheHospitalityInvestmentDecisionbyQingzhongMaPhDandAthenaWeiZhangPhD

Vol10No16BestPracticesinSearchEngineMarketingandOptimizationTheCaseoftheStJamesHotelbyGregBodenlcosVictorBogertDanGordonCarterHearneandChrisKAndersonPhD

Vol10No15TheImpactofPrix Fixe MenuPriceFormatsonGuestsrsquoDealPerceptionbyShuoWangandMichaelLynnPhD

Cornell Center for Hospitality Research

Publication Indexwwwchrcornelledu

The Executive Path Hospitality Leadership Through Learning

Complete program information and applications available online

wwwhotelschoolcornelleduexecedPhone + 1 607 255 4919 Email exec_ed_hotelcornelledu

Professionals from around the world are invited to attend 3-day 10-day or online courses at the worldrsquos leading institute for hospitality management education in

Visit our website to apply

Explore develop and apply ideas with global hospitality leaders and

expert Cornell professors

Success

AdvancingBusiness

andPersonal

bull Strategic Leadershipbull Financebull Foodservicebull Human Resources

bull Marketingbull Operationsbull Real Estate

No2RetaliationWhyanIncreaseinClaimsDoesNotMeantheSkyIsFallingbyDavidSherwynJDandGreggGilmanJD

2009 ToolsToolNo12MeasuringtheDiningExperienceTheCaseofVitaNovabyKeshPrasadandFredJDeMiccoPhD

Cornell Center for Hospitality ResearchPublication Index

The Executive Path Hospitality Leadership Through Learning

Complete program information and applications available online

wwwhotelschoolcornelleduexecedPhone + 1 607 255 4919 Email exec_ed_hotelcornelledu

Professionals from around the world are invited to attend 3-day 10-day or online courses at the worldrsquos leading institute for hospitality management education in

Visit our website to apply

Explore develop and apply ideas with global hospitality leaders and

expert Cornell professors

Success

AdvancingBusiness

andPersonal

bull Strategic Leadershipbull Financebull Foodservicebull Human Resources

bull Marketingbull Operationsbull Real Estate

visit our website to apply

wwwchrcornell edu

Page 12: Unscrambling the Puzzling Matter of Online Consumer ... · Philips Hospitality Americas Chris Proulx, CEO, eCornell & Executive Education Carolyn D. Richmond, Partner, Hospitality

12 TheCenterforHospitalityResearchbullCornellUniversity

thereviewersThesecanbeclassifiedas(a)attributesorfea-turesofthepropertythatfigureprominentlyandcontributetothepositiveornegativeratings(egroomsbeddishesfurnitureorservice)and(b)thepositiveornegativewordsthatrepresentthesentimentofthereviewer(eggreatamazingdirtyandwonderful)Theresultsofthequalitativeanalysisareasfollows

Syntactic featuresTheoveralltoneofthelowreviewsisratherargumentativeThereviewerstendtobeforcefulinpresentingtheirpointofviewTheylayoutagreatamountofreasoningandprovidespecificexamplestodrivehometheirargumentThisisreflectedinthegreaterpercentageofnegativeconnectorwords(34)suchasbut however as ifandeven thoughItisinterestingtonotethatreviewersinthiscategoryspeakdirectlytotheotherconsumersnotablythroughgreateruseofwordssuchasyou(46)andlooktoinfluencetheirdecisionmaking

InthehighreviewsbycontrasttheoveralltonetendstobeenunciativeandthereviewersusethetexttomerelyreinforcetheirnumericalratingsusingfewerwordsonaveragethanthenegativereviewersInthesereviewstheconsumersfocusontheirpersonalexperiencesandappealtotheaudiencethroughemotionsratherthanlogicandreasoningasevidentfromgreateruseofpositiveadjectives

QualitativeAnalysisoftheReviewTextTheaboveanalysisemphasizestheimportanceofthetextcontentinconsumerreviewsWhilenumericalratingsallowthereviewerstoratetheserviceprovideronvariousattributesitisthetextthatprovidesthemwiththeopportu-nitytoarticulatethenuancesoftheiroverallexperienceandconveyusefulinformationaboutaserviceproviderrsquostransac-tionsandservicecapabilitiesThereforesemanticanalysisofthetextcontentprovidesuswithanopportunitytothethirdquestionpresentedattheoutsetofthisreportwhatdoconsumerstalkaboutintheirreviews

Tothisendweanalyzed600eachofrandomlyselectedhighandlowreviewsusingsemanticprocessingtechniquesWeusedcenteringresonanceanalysis(CRA)amodeofnet-work-basedtextanalysisthatrepresentsthecontentoflargesetsoftextsbyidentifyingthemostimportantwordsthatlinkotherwordsinthenetwork16OuranalysisexaminedsyntacticandsemanticfeaturesSyntacticfeaturesincludethepercentageofthewordsthatareopenclass(ienounsverbsadjectives)Semanticfeaturesrepresentthespecial-izedvocabularythatreflectstheoverallopinionexpressedby

16CormanSandDooleyK(2006)CrawdadTextAnalysisSystem12CrawdadTechnologiesLLCChandlerAZ

Prominently Mentioned Features Freq sample Comments

bedroom beds linen sheets 383

Also the same bedroom and bathroom did not have any door separating the bedroom from the bathroom no privacy

Sleeping arrangements were very awkward The listing sheet described 2 full beds We brought our own linens the beds were queen sized Had to purchase sheet

Kitchen Kitchenware Dishes Dishwasher 297

You must not forget to step up when yoursquore walking out of the kitchen into the dining room otherwise yoursquoll trip over the 3-foot step up

No potholders in the kitchen and it didnrsquot seem like there were as many kitchen supplies as before We just had to make do without

beach beaches beachfront 296

The week was the best vacation we have ever had as a family The beach house was a class act

Disappointment 1 It looks like this house is very close to beach but to get there you have to go around through public path

Pool Poolside 218

The pool was too dirty to use We checked in on Thursday and the pool was not cleaned until Saturday

Also it is inconvenient to have to call to get someone to open the pool Didnrsquot understand why Jacuzzi was open and not the pool

bathroom(s) bath bathtub shower 202

The next problem was in another bathroom (main floor)mdashthe wall heater did not work so showering in here was a bit chilly The next problem was in the ldquomasterrdquo bathroom (connected to the bedroom with the King bed) The sliding shower doors need to be totally replaced

Bathtub drains clogged in three second floor bathrooms

Exhibit 7

Top five prominent issues in low reviews

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 13

suchasgreat fantastic andawesome(45)andpersonalpronounssuchasweandI(54)

Semantic featuresExhibits7and8showthemainissuesoraspectsthatareprominentlydiscussedinthelowreviewsTheorderoftheseissuesisbasedonthestatisti-callysignificantfrequencyoftheiroccurrenceinthe1200reviews

WefurtherinvestigatedtheuseofvarioussentimentwordsInterestinglytheprominentissuesdiscussedinthepositivereviewssometimesdonotevenfigureintheratingscalesprovidedtothereviewersOveralltheresultsofthesemanticanalysiscanbesummarizedasfollows

Thepropertyanditsfeaturesaremorecommonlycon-nectedwithextremelypositivesentimentwordssuchasgreat wonderfulandamazinginthehighreviewsthaninthelowreviewsIncomparisonthepropertiesandtheirfea-turesinthelowreviewsaremorelikelytobecharacterizedwithmoderatelypositivewordssuchasniceandokay

Negativereviewsreflecttheextentofexpectationndashdis-confirmationthattheconsumersfaceduringtheirstayintherentalpropertyThesereviewshaveagreaterproportionofwordsexpressingdisappointment(egnotasexpected)Furtherextremelynegativesentimentwords(egdiscom-fortfilthy)andadjectivesaremorelikelytobeassociatedwithfeaturesinlowreviewsthaninhighreviews

Almost40percentoftheconsumerswhowrotehighreviewsexpresstheirwillingnesstoreturntothepropertyasopposedtoonly9percentofthosewritingnegativereviewsFurthermorethan20percentoftheconsumersgivinghighreviewsexpresstheirstrongdesiretorecommendtheprop-ertytotheirfriendsandfamilymembersasopposedtoonly5percentinlowreviews

ReviewofFindingsOurexploratorystudyofconsumerreviewsofUSvacationhomesonReviewsitecomwashighlightedbyoverwhelm-inglypositivereviews

Theoveralldistributionofthereviewswasheav-ilyskewedtowardsthepositiveratingsMorecriticallythisfindingsuggeststhatthereisalackofmoderateandbalancedopinionsWetheorizethatthisiseitherduetopurchasingbias(sincecustomerswentthroughthetroubleofpurchasingandexperiencingthefacilitiestheytendtobepositivelybiasedtowardstheirpurchase)orunder-reportingbias(customersdonothavestrongenoughincentivestotakethetroubleofreportingtheiropinionsandthefewmotivatedindividualswhoprovidetheirreviewsbecomeaself-selectedsample)

Theoverall(mean)starratingtypicallyusedintravelerreviewsitesisprobablynotthemostaccurateindicatorofcustomersrsquoperceptionsofserviceprovidersrsquoqualityWe

Prominently Mentioned Features Freq sample Commentsbeach beaches beachfront 506 Perfect rental property for those who enjoy the activity of the beach and the serenity

of the Sound The house is just down the road from ocean

The large porch offered a beautiful view of the beach and comfortable new rocking chairs

Pool Poolside 218 The outdoor shower dipping pool and screen in porch were an added bonus Perfect house for a low key weekend

House was great for kids Pool was terrific and the playroom was perfectlocation 192 We had a wonderful time My wife loves fishing and the cottage was in a perfect

location to go fishing on the pier

The home is somewhat dated but the location was great We have vacationed in the Chatham area for many summers and this time it was great too

bedroom beds 186 The family suite is huge with a king size and twin bed and still room for a pack-n-play The kitchen and dining room were huge clean and nice

The only comments that I would say to the negative is the front bed room by the front door always seemed damp which seem to attract more bugs

Kitchen Kitchenware Dishes Dishwasher 178 All of the amenities in the house are upscalemdashloved the kitchen and plasma TVs

The kitchen had good basic utensils dishes and cooking pans

Exhibit 7

Top five prominent issues in high reviews

14 TheCenterforHospitalityResearchbullCornellUniversity

foundacorrelationbetweenthestarratingandratingsonotherpropertyattributesoflessthan40percentinmostcases

AhighwordcountinthereviewsiscorrelatedwithalowoverallratingThatisconsumerswhoareextremelydissatisfiedtendtoexpendmoretimeandenergyontheircritiqueofthepropertyitsamenitiesandservicesPriceofthepropertyalsoseemstopositivelycorrelatewithextremenegativereviewsperhapsbecauseahighpriceheightenstravelersrsquoexpectations

Numericalratingsdonotoftencapturetheactualsentimentandthevarietyofdimensionsonwhichtravel-ersevaluatepropertiesBecausenumericalratingsdonotcapturereviewersrsquoopinionsaboutspecificattributestheycanbedeceivingandmaynotrevealthetruequalityofthepropertyortheactualintentionofthereviewerWefoundforinstancethatconsumersvaluelocation(egclosenesstothebeach)andcleanliness(egswimmingpoolareas)morethansuchfeaturesascheck-inandserviceSimilarlyinthenegativereviewsissuessuchaslackofbedroomparapher-naliaorkitchenitemsseemtoberankedhigherinimpor-tancethanissuessuchaslocationandcheck-inThereforeevenwhenapropertyscoreshighonthecertaindimensionstheoverallratingmayremainlowbecauseotherfactorsovershadowthefavorableaspects

WhatItMeansforManagersWeseeimplicationsforonlinereviewsitesandbyexten-sionforproprietorsofvacationhomesandserviceopera-tionsgenerallyWithregardtothereviewsitesifconsumersconsistentlydetectafavorablebiasthereviewsitesmaylosethecriticalaspectsofcredibilityandtrustAlthoughconsumersaregenerallysmartandrationalandcanworkaroundobviousbiasesrecentevidencesuggeststhatcon-sumersaregraduallydefectingfromsomeonlinesitesduetounbalancedreviews17InthisregardsiteadministratorsaswellaspropertymanagersshouldtakestepstoattractthoseconsumerswhoarewillingtoprovideareasonablybalancedanalysisoftheamenitiesandservicesBythiswemeanreviewsthatpresentcomprehensiveandobjectiveanalysisoftheserviceproviderratherthanjustravesorrants

Tothatendsiteoperatorsshouldtakestepstoun-derstandcustomersrsquoreviewwritingbehaviorandtoat-tractmorebalancedreviewsOnewaytodothisistogivepersonalizedinformationtotheconsumerdemonstratingthecontributionmadebyhisorherreviewandthevalueofthereviewtostakeholdersForinstanceLinget alfoundapositivecorrelationbetweenwillingnesstowritereviewsandtheconsumersrsquoperceptionthattheircontributionswouldbeuniqueandhelpfultoothercustomersaswellasperusedby

17HarteveldtHHJohnsonCStarkEandGeldernKV(2009)Us-ingDigitalChannelsToCalmTheAngryTravelerForresterResearch

theserviceproviders18Simplyputthepropensitytowritereviewsincreaseswhentheconsumersfeelthattheirreviewisactuallyhelpingothers(boththeserviceproviderandotherconsumers)

WesuggestthatreviewsitesdevelopbettermethodstoaggregatesynthesizeandpublishspecificreviewcommentsaswellasnumericalratingsFirstmanagersshouldtakeintoconsiderationthelowcorrelationbetweentheoverallratingsandconsumersrsquoassessmentsofvariouspropertyattributes(aswellastheresultsfromthetextanalysis)Theimplicationofsuchaweakcorrelationisthatreviewsitesneedtoexpandonthelistofpossiblevariablessothattheratingsaccuratelyreflectconsumersrsquoneedsandexpectationsregardingserviceproviders

Furthersitemanagersmayhavetorethinkthecurrentpracticeofaveragingalltheratingsforagivenpropertybe-causesuchameanratingfailstotakeintoaccountthebiasesthatareinherentintheratingsystemsAsanexamplethisstudyfindssignificantunder-reportingbiasinthisonlinereviewsiteAcustomerrsquosdecisiontonotpostonlinereviewsmaymeanthelossofimportantinformationthatcanassisttheusersofthesitestomakemorereliableinferencesThemajorityoftodayrsquosfeedbackmechanismsdonotpubliclydisclosethenumberofsilenttransactions(ietransactionsforwhichnofeedbackwaspostedbycustomers)Suchinfor-mationshouldbecomeapartofaserviceproviderrsquosonlineprofileonratingsitesandotherfeedbackmechanismsItisimportanttoprovidemoreinformationandheuristicstohelptheconsumersnavigatethroughtheclutterandgettheinformationtheywantandseek

Itisimportanttoapplymodernsemanticanalysisandsentimentclassificationtechniquestostudythecontentofthereviewsandidentifyvariousdimensionsonwhichcon-sumersevaluatevacationhomesThedetailsprovidedinthereviewtextareinmanywaysabetterreflectionofcustomersatisfactionServiceproviderscanbetterpositionthemselvesusingthedimensionsuncoveredthroughtextanalysisandtargettheneedsandpreferencesoftheircustomers

TheresultsinthisstudyshowthatnegativereviewsaretypicallyassociatedwithagreaterwordcountthanpositivereviewsItisusefultofocusspecificallyonthesenegativereviewsbecausetheycanrevealpatternsofdeficienciesintheservicestandardsanddeliveryThesereviewsshouldbetreatedthesameascustomercommentcardsorletters19Furthercustomerswhowritesuchlengthynegativereviews

18LingKimberlyGerardBeenenPaulLudfordXiaoxingWangKevinChangXinLiDanCosleyDanFrankowskiLorenTerveenAlMRashidandRobertKrautldquoUsingSocialPsychologytoMotivateContributionstoOnlineCommunitiesrdquoJournal of Computer-Mediated Communication10no4(2005)19MilanR(2007)10thingsyoucandoinresponsetotravelerreviewsinHotelmarketingcom

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 15

aremostlikelytospreadnegativeword-of-mouthaboutthepropertyLengthyreviewsareexcellentsourcesofinforma-tionandshouldbeusedasthebasestounderstandthecomplainingbehavioroftheconsumer

WordcountmaybeusedasanadditionalsegmentationvariabletoidentifythoseconsumerswhoaremoreinvolvedandwillingtoprovidebalancedfeedbacktoboththereviewsiteaswellastopropertymanagementVariousotherresearchmethodssuchastelephoneinterviewsorfocusgroupsshouldbeusedtoelicitdetailedfeedbackfromsuchconsumersTheseactionsalsohelpinservicerecoverypost-purchaseengagementandrelationshipbuilding

OutlookWeforeseethreepotentialareasinvestigationthatcanaddvaluetocurrentbestpracticesWestilldonrsquotknowmuchaboutconsumersrsquomotivationtowriteonlinereviews20Recentstudiesidentifyvariousantecedentsofreviewwritingbutsufferfromsamplingbiassincetheysurveyaself-select-edsampleofconsumerswhoalreadyprovidedreviewsAstheresultsinthisstudyindicateunder-reportingbias(lackofmoderateandlowreviews)remainsprevalentinonlinereviewsitesFutureresearchshouldidentifysocio-psycho-logicalfactorsthatincreaseconsumersrsquopropensitytowriteonlinereviews

Futureresearchshouldexploretheimpactofthepres-ence(orlack)ofextremelynegativereviewswithinalargecorpusofpositivereviewsTheevidenceontheimpactofre-

20WangYandFesenmaierDR(2003)AssessingMotivationofCon-tributioninOnlineCommunitiesAnEmpiricalInvestigationofanOn-lineTravelCommunityElectronic Markets1333-45andYooKyungHandUlrikeGretzelldquoWhatMotivatesConsumerstoWriteOnlineTravelReviewsrdquoInformation Technology amp Tourism10No4(2008)283-95

viewvalence(negativeorpositive)isequivocalProponentsofconfirmatorybiassuggestthatconsumerslookforaffir-mativeevidencesupportingaproductchoicealreadymadeIfthatisthecasepositivereviewsaremorelikelytohaveagreatereffectonconsumeractionsOntheotherhandthenotionofnegativitybiassuggeststhatwhenconsumersareneutralnegativereviewstendtobecomemoresalientthanpositivereviews21Inthepresenceoflargenumbersofposi-tivereviewsconsumersspecificallyseeknegativereviewsthattheyfeelwillhelpthemidentifyspecificproblemswithaserviceUnderstandinghowconsumersreconciletheinformationprovidedbybothnegativeandpositivereviewshasimplicationsforthedesignandmanagementofreviewsites

Finallyareviewrsquoswordcounthasimportantimplica-tionsforconsumersrsquotrustinthereviewsWhenconsumersarewillingtoreadandcompareopen-endedcommentsfromotherconsumerstheamountofinformationinareviewcanmatterWOMforhighinvolvementproductsismorepersuasiveandincreasesthedecisionmakerrsquosconfi-dencewhenthemessagesenderprovidesgreaternumberofreasonsfortheiroverallratingLongerreviewsprovidemoreinformationandarelikelytobeperceivedasmorehelpfulandpersuasivethanshorterreviewsThewaysinwhichareviewiswrittenframedandpresentedisaninterestingfactorwhichhasrarelybeeninvestigatedandhasimportantimplicationsparticularlygivenourfindingthattheamountofinformationinareview(wordcount)variessignificantlywiththeratingsofthereviewn

21BaSandPavlouPA(2002)EvidenceoftheEffectofTrustBuildingTechnologyinElectronicMarketsPricePremiumsandBuyerBehaviorMIS Quarterly26243-268

Cornell Hospitality Quarterlyhttpcqxsagepubcom

2011 ReportsVol11No15DesigningaSelf-HealingServiceSystemAnIntegrativeModelbyRobertCFordandMichaelCSturman

Vol11No14ReversingtheGreenBacklashWhyLargeHospitalityCompaniesShouldWelcomeCrediblyGreenCompetitorsbyMichaelGiebelhausenPhDandHaeEunHelenChunPhD

Vol11No13DevelopingaSustainabilityMeasurementFrameworkforHotelsTowardanIndustry-wideReportingStructurebyEricRicaurte

Vol11No12CreatingValueforWomenBusinessTravelersFocusingonEmotionalOutcomesbyJudiBrownellPhD

Vol11No11CustomerLoyaltyANewLookattheBenefitsofImprovingSegmentationEffortswithRewardsProgramsbyClayVoorheesPhDMichaelMcCallPhDandRogerCalantonePhD

Vol11No10CustomerPerceptionsofElectronicFoodOrderingbySherylEKimes

Vol11No92011TravelIndustryBenchmarkingStatusofSeniorDestinationandLodgingMarketingExecutivesbyRohitVermaPhDandKenMcGill

Vol11No8SearchOTAsandOnlineBookingAnExpandedAnalysisoftheBillboardEffectbyChrisAndersonPhD

Vol11No7OnlineMobileandTextFoodOrderingintheUSRestaurantIndustrybySherylEKimesPhDandPhilippFLaqueacute

Vol11No6HotelGuestsrsquoReactionstoGuestRoomSustainabilityInitiativesbyAlexSusskindPhDandRohitVermaPhD

Vol11No5TheImpactofTerrorismandEconomicShocksonUSHotelsbyCathyAEnzRenaacutetaKosovaacuteandMarkLomannoVol11No4ImplementingHumanResourceInnovationsThreeSuccessStoriesfromtheServiceIndustrybyJustinSunandKateWalshPhD

Vol11No3Compendium2011

Vol11No2PositioningaPlaceDevelopingaCompellingDestinationBrandbyRobertJKwortnikPhDandEthanHawkesMBA

Vol11No1TheImpactofHealthInsuranceonEmployeeJobAnxietyWithdrawalBehaviorsandTaskPerformancebySeanWayPhDBillCarrollPhDAlexSusskindPhDandJoeCYLeng

2011 Hospitality ToolsVol2No1TheGameHasChangedANewParadigmforStakeholderEngagementbyMaryBethMcEuen

2011 Industry PerspectivesNo7MegaTips2TwentyTestedTechniquesforIncreasingYourTipsbyMichaelLynn

2011 ProceedingsVol3No5BuildingBrandsintheInternetAgeAnalyticsLoyaltyandCommunicationbyGlennWithiam

Vol3No4BraveNewWorldOnlineHotelDistributionbyGlennWithiam

Vol3No3SocialMediaandtheHospitalityIndustryHoldingtheTigerbytheTailbyGlennWithiam

Vol3No2TheChallengeofHotelandRestaurantSustainabilityFindingProfitinldquoBeingGreenrdquobyGlennWithiam

Vol3No1CautiousOptimismCHRSExaminesHospitalityIndustryTrendsbyGlennWithiam

2010 ReportsVol10No18HowTravelersUseOnlineandSocialMediaChannelstoMakeHotel-choiceDecisionsbyLauraMcCarthyDebraStockandRohitVermaPhD

Vol10No17PublicorPrivateTheHospitalityInvestmentDecisionbyQingzhongMaPhDandAthenaWeiZhangPhD

Vol10No16BestPracticesinSearchEngineMarketingandOptimizationTheCaseoftheStJamesHotelbyGregBodenlcosVictorBogertDanGordonCarterHearneandChrisKAndersonPhD

Vol10No15TheImpactofPrix Fixe MenuPriceFormatsonGuestsrsquoDealPerceptionbyShuoWangandMichaelLynnPhD

Cornell Center for Hospitality Research

Publication Indexwwwchrcornelledu

The Executive Path Hospitality Leadership Through Learning

Complete program information and applications available online

wwwhotelschoolcornelleduexecedPhone + 1 607 255 4919 Email exec_ed_hotelcornelledu

Professionals from around the world are invited to attend 3-day 10-day or online courses at the worldrsquos leading institute for hospitality management education in

Visit our website to apply

Explore develop and apply ideas with global hospitality leaders and

expert Cornell professors

Success

AdvancingBusiness

andPersonal

bull Strategic Leadershipbull Financebull Foodservicebull Human Resources

bull Marketingbull Operationsbull Real Estate

No2RetaliationWhyanIncreaseinClaimsDoesNotMeantheSkyIsFallingbyDavidSherwynJDandGreggGilmanJD

2009 ToolsToolNo12MeasuringtheDiningExperienceTheCaseofVitaNovabyKeshPrasadandFredJDeMiccoPhD

Cornell Center for Hospitality ResearchPublication Index

The Executive Path Hospitality Leadership Through Learning

Complete program information and applications available online

wwwhotelschoolcornelleduexecedPhone + 1 607 255 4919 Email exec_ed_hotelcornelledu

Professionals from around the world are invited to attend 3-day 10-day or online courses at the worldrsquos leading institute for hospitality management education in

Visit our website to apply

Explore develop and apply ideas with global hospitality leaders and

expert Cornell professors

Success

AdvancingBusiness

andPersonal

bull Strategic Leadershipbull Financebull Foodservicebull Human Resources

bull Marketingbull Operationsbull Real Estate

visit our website to apply

wwwchrcornell edu

Page 13: Unscrambling the Puzzling Matter of Online Consumer ... · Philips Hospitality Americas Chris Proulx, CEO, eCornell & Executive Education Carolyn D. Richmond, Partner, Hospitality

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 13

suchasgreat fantastic andawesome(45)andpersonalpronounssuchasweandI(54)

Semantic featuresExhibits7and8showthemainissuesoraspectsthatareprominentlydiscussedinthelowreviewsTheorderoftheseissuesisbasedonthestatisti-callysignificantfrequencyoftheiroccurrenceinthe1200reviews

WefurtherinvestigatedtheuseofvarioussentimentwordsInterestinglytheprominentissuesdiscussedinthepositivereviewssometimesdonotevenfigureintheratingscalesprovidedtothereviewersOveralltheresultsofthesemanticanalysiscanbesummarizedasfollows

Thepropertyanditsfeaturesaremorecommonlycon-nectedwithextremelypositivesentimentwordssuchasgreat wonderfulandamazinginthehighreviewsthaninthelowreviewsIncomparisonthepropertiesandtheirfea-turesinthelowreviewsaremorelikelytobecharacterizedwithmoderatelypositivewordssuchasniceandokay

Negativereviewsreflecttheextentofexpectationndashdis-confirmationthattheconsumersfaceduringtheirstayintherentalpropertyThesereviewshaveagreaterproportionofwordsexpressingdisappointment(egnotasexpected)Furtherextremelynegativesentimentwords(egdiscom-fortfilthy)andadjectivesaremorelikelytobeassociatedwithfeaturesinlowreviewsthaninhighreviews

Almost40percentoftheconsumerswhowrotehighreviewsexpresstheirwillingnesstoreturntothepropertyasopposedtoonly9percentofthosewritingnegativereviewsFurthermorethan20percentoftheconsumersgivinghighreviewsexpresstheirstrongdesiretorecommendtheprop-ertytotheirfriendsandfamilymembersasopposedtoonly5percentinlowreviews

ReviewofFindingsOurexploratorystudyofconsumerreviewsofUSvacationhomesonReviewsitecomwashighlightedbyoverwhelm-inglypositivereviews

Theoveralldistributionofthereviewswasheav-ilyskewedtowardsthepositiveratingsMorecriticallythisfindingsuggeststhatthereisalackofmoderateandbalancedopinionsWetheorizethatthisiseitherduetopurchasingbias(sincecustomerswentthroughthetroubleofpurchasingandexperiencingthefacilitiestheytendtobepositivelybiasedtowardstheirpurchase)orunder-reportingbias(customersdonothavestrongenoughincentivestotakethetroubleofreportingtheiropinionsandthefewmotivatedindividualswhoprovidetheirreviewsbecomeaself-selectedsample)

Theoverall(mean)starratingtypicallyusedintravelerreviewsitesisprobablynotthemostaccurateindicatorofcustomersrsquoperceptionsofserviceprovidersrsquoqualityWe

Prominently Mentioned Features Freq sample Commentsbeach beaches beachfront 506 Perfect rental property for those who enjoy the activity of the beach and the serenity

of the Sound The house is just down the road from ocean

The large porch offered a beautiful view of the beach and comfortable new rocking chairs

Pool Poolside 218 The outdoor shower dipping pool and screen in porch were an added bonus Perfect house for a low key weekend

House was great for kids Pool was terrific and the playroom was perfectlocation 192 We had a wonderful time My wife loves fishing and the cottage was in a perfect

location to go fishing on the pier

The home is somewhat dated but the location was great We have vacationed in the Chatham area for many summers and this time it was great too

bedroom beds 186 The family suite is huge with a king size and twin bed and still room for a pack-n-play The kitchen and dining room were huge clean and nice

The only comments that I would say to the negative is the front bed room by the front door always seemed damp which seem to attract more bugs

Kitchen Kitchenware Dishes Dishwasher 178 All of the amenities in the house are upscalemdashloved the kitchen and plasma TVs

The kitchen had good basic utensils dishes and cooking pans

Exhibit 7

Top five prominent issues in high reviews

14 TheCenterforHospitalityResearchbullCornellUniversity

foundacorrelationbetweenthestarratingandratingsonotherpropertyattributesoflessthan40percentinmostcases

AhighwordcountinthereviewsiscorrelatedwithalowoverallratingThatisconsumerswhoareextremelydissatisfiedtendtoexpendmoretimeandenergyontheircritiqueofthepropertyitsamenitiesandservicesPriceofthepropertyalsoseemstopositivelycorrelatewithextremenegativereviewsperhapsbecauseahighpriceheightenstravelersrsquoexpectations

Numericalratingsdonotoftencapturetheactualsentimentandthevarietyofdimensionsonwhichtravel-ersevaluatepropertiesBecausenumericalratingsdonotcapturereviewersrsquoopinionsaboutspecificattributestheycanbedeceivingandmaynotrevealthetruequalityofthepropertyortheactualintentionofthereviewerWefoundforinstancethatconsumersvaluelocation(egclosenesstothebeach)andcleanliness(egswimmingpoolareas)morethansuchfeaturesascheck-inandserviceSimilarlyinthenegativereviewsissuessuchaslackofbedroomparapher-naliaorkitchenitemsseemtoberankedhigherinimpor-tancethanissuessuchaslocationandcheck-inThereforeevenwhenapropertyscoreshighonthecertaindimensionstheoverallratingmayremainlowbecauseotherfactorsovershadowthefavorableaspects

WhatItMeansforManagersWeseeimplicationsforonlinereviewsitesandbyexten-sionforproprietorsofvacationhomesandserviceopera-tionsgenerallyWithregardtothereviewsitesifconsumersconsistentlydetectafavorablebiasthereviewsitesmaylosethecriticalaspectsofcredibilityandtrustAlthoughconsumersaregenerallysmartandrationalandcanworkaroundobviousbiasesrecentevidencesuggeststhatcon-sumersaregraduallydefectingfromsomeonlinesitesduetounbalancedreviews17InthisregardsiteadministratorsaswellaspropertymanagersshouldtakestepstoattractthoseconsumerswhoarewillingtoprovideareasonablybalancedanalysisoftheamenitiesandservicesBythiswemeanreviewsthatpresentcomprehensiveandobjectiveanalysisoftheserviceproviderratherthanjustravesorrants

Tothatendsiteoperatorsshouldtakestepstoun-derstandcustomersrsquoreviewwritingbehaviorandtoat-tractmorebalancedreviewsOnewaytodothisistogivepersonalizedinformationtotheconsumerdemonstratingthecontributionmadebyhisorherreviewandthevalueofthereviewtostakeholdersForinstanceLinget alfoundapositivecorrelationbetweenwillingnesstowritereviewsandtheconsumersrsquoperceptionthattheircontributionswouldbeuniqueandhelpfultoothercustomersaswellasperusedby

17HarteveldtHHJohnsonCStarkEandGeldernKV(2009)Us-ingDigitalChannelsToCalmTheAngryTravelerForresterResearch

theserviceproviders18Simplyputthepropensitytowritereviewsincreaseswhentheconsumersfeelthattheirreviewisactuallyhelpingothers(boththeserviceproviderandotherconsumers)

WesuggestthatreviewsitesdevelopbettermethodstoaggregatesynthesizeandpublishspecificreviewcommentsaswellasnumericalratingsFirstmanagersshouldtakeintoconsiderationthelowcorrelationbetweentheoverallratingsandconsumersrsquoassessmentsofvariouspropertyattributes(aswellastheresultsfromthetextanalysis)Theimplicationofsuchaweakcorrelationisthatreviewsitesneedtoexpandonthelistofpossiblevariablessothattheratingsaccuratelyreflectconsumersrsquoneedsandexpectationsregardingserviceproviders

Furthersitemanagersmayhavetorethinkthecurrentpracticeofaveragingalltheratingsforagivenpropertybe-causesuchameanratingfailstotakeintoaccountthebiasesthatareinherentintheratingsystemsAsanexamplethisstudyfindssignificantunder-reportingbiasinthisonlinereviewsiteAcustomerrsquosdecisiontonotpostonlinereviewsmaymeanthelossofimportantinformationthatcanassisttheusersofthesitestomakemorereliableinferencesThemajorityoftodayrsquosfeedbackmechanismsdonotpubliclydisclosethenumberofsilenttransactions(ietransactionsforwhichnofeedbackwaspostedbycustomers)Suchinfor-mationshouldbecomeapartofaserviceproviderrsquosonlineprofileonratingsitesandotherfeedbackmechanismsItisimportanttoprovidemoreinformationandheuristicstohelptheconsumersnavigatethroughtheclutterandgettheinformationtheywantandseek

Itisimportanttoapplymodernsemanticanalysisandsentimentclassificationtechniquestostudythecontentofthereviewsandidentifyvariousdimensionsonwhichcon-sumersevaluatevacationhomesThedetailsprovidedinthereviewtextareinmanywaysabetterreflectionofcustomersatisfactionServiceproviderscanbetterpositionthemselvesusingthedimensionsuncoveredthroughtextanalysisandtargettheneedsandpreferencesoftheircustomers

TheresultsinthisstudyshowthatnegativereviewsaretypicallyassociatedwithagreaterwordcountthanpositivereviewsItisusefultofocusspecificallyonthesenegativereviewsbecausetheycanrevealpatternsofdeficienciesintheservicestandardsanddeliveryThesereviewsshouldbetreatedthesameascustomercommentcardsorletters19Furthercustomerswhowritesuchlengthynegativereviews

18LingKimberlyGerardBeenenPaulLudfordXiaoxingWangKevinChangXinLiDanCosleyDanFrankowskiLorenTerveenAlMRashidandRobertKrautldquoUsingSocialPsychologytoMotivateContributionstoOnlineCommunitiesrdquoJournal of Computer-Mediated Communication10no4(2005)19MilanR(2007)10thingsyoucandoinresponsetotravelerreviewsinHotelmarketingcom

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 15

aremostlikelytospreadnegativeword-of-mouthaboutthepropertyLengthyreviewsareexcellentsourcesofinforma-tionandshouldbeusedasthebasestounderstandthecomplainingbehavioroftheconsumer

WordcountmaybeusedasanadditionalsegmentationvariabletoidentifythoseconsumerswhoaremoreinvolvedandwillingtoprovidebalancedfeedbacktoboththereviewsiteaswellastopropertymanagementVariousotherresearchmethodssuchastelephoneinterviewsorfocusgroupsshouldbeusedtoelicitdetailedfeedbackfromsuchconsumersTheseactionsalsohelpinservicerecoverypost-purchaseengagementandrelationshipbuilding

OutlookWeforeseethreepotentialareasinvestigationthatcanaddvaluetocurrentbestpracticesWestilldonrsquotknowmuchaboutconsumersrsquomotivationtowriteonlinereviews20Recentstudiesidentifyvariousantecedentsofreviewwritingbutsufferfromsamplingbiassincetheysurveyaself-select-edsampleofconsumerswhoalreadyprovidedreviewsAstheresultsinthisstudyindicateunder-reportingbias(lackofmoderateandlowreviews)remainsprevalentinonlinereviewsitesFutureresearchshouldidentifysocio-psycho-logicalfactorsthatincreaseconsumersrsquopropensitytowriteonlinereviews

Futureresearchshouldexploretheimpactofthepres-ence(orlack)ofextremelynegativereviewswithinalargecorpusofpositivereviewsTheevidenceontheimpactofre-

20WangYandFesenmaierDR(2003)AssessingMotivationofCon-tributioninOnlineCommunitiesAnEmpiricalInvestigationofanOn-lineTravelCommunityElectronic Markets1333-45andYooKyungHandUlrikeGretzelldquoWhatMotivatesConsumerstoWriteOnlineTravelReviewsrdquoInformation Technology amp Tourism10No4(2008)283-95

viewvalence(negativeorpositive)isequivocalProponentsofconfirmatorybiassuggestthatconsumerslookforaffir-mativeevidencesupportingaproductchoicealreadymadeIfthatisthecasepositivereviewsaremorelikelytohaveagreatereffectonconsumeractionsOntheotherhandthenotionofnegativitybiassuggeststhatwhenconsumersareneutralnegativereviewstendtobecomemoresalientthanpositivereviews21Inthepresenceoflargenumbersofposi-tivereviewsconsumersspecificallyseeknegativereviewsthattheyfeelwillhelpthemidentifyspecificproblemswithaserviceUnderstandinghowconsumersreconciletheinformationprovidedbybothnegativeandpositivereviewshasimplicationsforthedesignandmanagementofreviewsites

Finallyareviewrsquoswordcounthasimportantimplica-tionsforconsumersrsquotrustinthereviewsWhenconsumersarewillingtoreadandcompareopen-endedcommentsfromotherconsumerstheamountofinformationinareviewcanmatterWOMforhighinvolvementproductsismorepersuasiveandincreasesthedecisionmakerrsquosconfi-dencewhenthemessagesenderprovidesgreaternumberofreasonsfortheiroverallratingLongerreviewsprovidemoreinformationandarelikelytobeperceivedasmorehelpfulandpersuasivethanshorterreviewsThewaysinwhichareviewiswrittenframedandpresentedisaninterestingfactorwhichhasrarelybeeninvestigatedandhasimportantimplicationsparticularlygivenourfindingthattheamountofinformationinareview(wordcount)variessignificantlywiththeratingsofthereviewn

21BaSandPavlouPA(2002)EvidenceoftheEffectofTrustBuildingTechnologyinElectronicMarketsPricePremiumsandBuyerBehaviorMIS Quarterly26243-268

Cornell Hospitality Quarterlyhttpcqxsagepubcom

2011 ReportsVol11No15DesigningaSelf-HealingServiceSystemAnIntegrativeModelbyRobertCFordandMichaelCSturman

Vol11No14ReversingtheGreenBacklashWhyLargeHospitalityCompaniesShouldWelcomeCrediblyGreenCompetitorsbyMichaelGiebelhausenPhDandHaeEunHelenChunPhD

Vol11No13DevelopingaSustainabilityMeasurementFrameworkforHotelsTowardanIndustry-wideReportingStructurebyEricRicaurte

Vol11No12CreatingValueforWomenBusinessTravelersFocusingonEmotionalOutcomesbyJudiBrownellPhD

Vol11No11CustomerLoyaltyANewLookattheBenefitsofImprovingSegmentationEffortswithRewardsProgramsbyClayVoorheesPhDMichaelMcCallPhDandRogerCalantonePhD

Vol11No10CustomerPerceptionsofElectronicFoodOrderingbySherylEKimes

Vol11No92011TravelIndustryBenchmarkingStatusofSeniorDestinationandLodgingMarketingExecutivesbyRohitVermaPhDandKenMcGill

Vol11No8SearchOTAsandOnlineBookingAnExpandedAnalysisoftheBillboardEffectbyChrisAndersonPhD

Vol11No7OnlineMobileandTextFoodOrderingintheUSRestaurantIndustrybySherylEKimesPhDandPhilippFLaqueacute

Vol11No6HotelGuestsrsquoReactionstoGuestRoomSustainabilityInitiativesbyAlexSusskindPhDandRohitVermaPhD

Vol11No5TheImpactofTerrorismandEconomicShocksonUSHotelsbyCathyAEnzRenaacutetaKosovaacuteandMarkLomannoVol11No4ImplementingHumanResourceInnovationsThreeSuccessStoriesfromtheServiceIndustrybyJustinSunandKateWalshPhD

Vol11No3Compendium2011

Vol11No2PositioningaPlaceDevelopingaCompellingDestinationBrandbyRobertJKwortnikPhDandEthanHawkesMBA

Vol11No1TheImpactofHealthInsuranceonEmployeeJobAnxietyWithdrawalBehaviorsandTaskPerformancebySeanWayPhDBillCarrollPhDAlexSusskindPhDandJoeCYLeng

2011 Hospitality ToolsVol2No1TheGameHasChangedANewParadigmforStakeholderEngagementbyMaryBethMcEuen

2011 Industry PerspectivesNo7MegaTips2TwentyTestedTechniquesforIncreasingYourTipsbyMichaelLynn

2011 ProceedingsVol3No5BuildingBrandsintheInternetAgeAnalyticsLoyaltyandCommunicationbyGlennWithiam

Vol3No4BraveNewWorldOnlineHotelDistributionbyGlennWithiam

Vol3No3SocialMediaandtheHospitalityIndustryHoldingtheTigerbytheTailbyGlennWithiam

Vol3No2TheChallengeofHotelandRestaurantSustainabilityFindingProfitinldquoBeingGreenrdquobyGlennWithiam

Vol3No1CautiousOptimismCHRSExaminesHospitalityIndustryTrendsbyGlennWithiam

2010 ReportsVol10No18HowTravelersUseOnlineandSocialMediaChannelstoMakeHotel-choiceDecisionsbyLauraMcCarthyDebraStockandRohitVermaPhD

Vol10No17PublicorPrivateTheHospitalityInvestmentDecisionbyQingzhongMaPhDandAthenaWeiZhangPhD

Vol10No16BestPracticesinSearchEngineMarketingandOptimizationTheCaseoftheStJamesHotelbyGregBodenlcosVictorBogertDanGordonCarterHearneandChrisKAndersonPhD

Vol10No15TheImpactofPrix Fixe MenuPriceFormatsonGuestsrsquoDealPerceptionbyShuoWangandMichaelLynnPhD

Cornell Center for Hospitality Research

Publication Indexwwwchrcornelledu

The Executive Path Hospitality Leadership Through Learning

Complete program information and applications available online

wwwhotelschoolcornelleduexecedPhone + 1 607 255 4919 Email exec_ed_hotelcornelledu

Professionals from around the world are invited to attend 3-day 10-day or online courses at the worldrsquos leading institute for hospitality management education in

Visit our website to apply

Explore develop and apply ideas with global hospitality leaders and

expert Cornell professors

Success

AdvancingBusiness

andPersonal

bull Strategic Leadershipbull Financebull Foodservicebull Human Resources

bull Marketingbull Operationsbull Real Estate

No2RetaliationWhyanIncreaseinClaimsDoesNotMeantheSkyIsFallingbyDavidSherwynJDandGreggGilmanJD

2009 ToolsToolNo12MeasuringtheDiningExperienceTheCaseofVitaNovabyKeshPrasadandFredJDeMiccoPhD

Cornell Center for Hospitality ResearchPublication Index

The Executive Path Hospitality Leadership Through Learning

Complete program information and applications available online

wwwhotelschoolcornelleduexecedPhone + 1 607 255 4919 Email exec_ed_hotelcornelledu

Professionals from around the world are invited to attend 3-day 10-day or online courses at the worldrsquos leading institute for hospitality management education in

Visit our website to apply

Explore develop and apply ideas with global hospitality leaders and

expert Cornell professors

Success

AdvancingBusiness

andPersonal

bull Strategic Leadershipbull Financebull Foodservicebull Human Resources

bull Marketingbull Operationsbull Real Estate

visit our website to apply

wwwchrcornell edu

Page 14: Unscrambling the Puzzling Matter of Online Consumer ... · Philips Hospitality Americas Chris Proulx, CEO, eCornell & Executive Education Carolyn D. Richmond, Partner, Hospitality

14 TheCenterforHospitalityResearchbullCornellUniversity

foundacorrelationbetweenthestarratingandratingsonotherpropertyattributesoflessthan40percentinmostcases

AhighwordcountinthereviewsiscorrelatedwithalowoverallratingThatisconsumerswhoareextremelydissatisfiedtendtoexpendmoretimeandenergyontheircritiqueofthepropertyitsamenitiesandservicesPriceofthepropertyalsoseemstopositivelycorrelatewithextremenegativereviewsperhapsbecauseahighpriceheightenstravelersrsquoexpectations

Numericalratingsdonotoftencapturetheactualsentimentandthevarietyofdimensionsonwhichtravel-ersevaluatepropertiesBecausenumericalratingsdonotcapturereviewersrsquoopinionsaboutspecificattributestheycanbedeceivingandmaynotrevealthetruequalityofthepropertyortheactualintentionofthereviewerWefoundforinstancethatconsumersvaluelocation(egclosenesstothebeach)andcleanliness(egswimmingpoolareas)morethansuchfeaturesascheck-inandserviceSimilarlyinthenegativereviewsissuessuchaslackofbedroomparapher-naliaorkitchenitemsseemtoberankedhigherinimpor-tancethanissuessuchaslocationandcheck-inThereforeevenwhenapropertyscoreshighonthecertaindimensionstheoverallratingmayremainlowbecauseotherfactorsovershadowthefavorableaspects

WhatItMeansforManagersWeseeimplicationsforonlinereviewsitesandbyexten-sionforproprietorsofvacationhomesandserviceopera-tionsgenerallyWithregardtothereviewsitesifconsumersconsistentlydetectafavorablebiasthereviewsitesmaylosethecriticalaspectsofcredibilityandtrustAlthoughconsumersaregenerallysmartandrationalandcanworkaroundobviousbiasesrecentevidencesuggeststhatcon-sumersaregraduallydefectingfromsomeonlinesitesduetounbalancedreviews17InthisregardsiteadministratorsaswellaspropertymanagersshouldtakestepstoattractthoseconsumerswhoarewillingtoprovideareasonablybalancedanalysisoftheamenitiesandservicesBythiswemeanreviewsthatpresentcomprehensiveandobjectiveanalysisoftheserviceproviderratherthanjustravesorrants

Tothatendsiteoperatorsshouldtakestepstoun-derstandcustomersrsquoreviewwritingbehaviorandtoat-tractmorebalancedreviewsOnewaytodothisistogivepersonalizedinformationtotheconsumerdemonstratingthecontributionmadebyhisorherreviewandthevalueofthereviewtostakeholdersForinstanceLinget alfoundapositivecorrelationbetweenwillingnesstowritereviewsandtheconsumersrsquoperceptionthattheircontributionswouldbeuniqueandhelpfultoothercustomersaswellasperusedby

17HarteveldtHHJohnsonCStarkEandGeldernKV(2009)Us-ingDigitalChannelsToCalmTheAngryTravelerForresterResearch

theserviceproviders18Simplyputthepropensitytowritereviewsincreaseswhentheconsumersfeelthattheirreviewisactuallyhelpingothers(boththeserviceproviderandotherconsumers)

WesuggestthatreviewsitesdevelopbettermethodstoaggregatesynthesizeandpublishspecificreviewcommentsaswellasnumericalratingsFirstmanagersshouldtakeintoconsiderationthelowcorrelationbetweentheoverallratingsandconsumersrsquoassessmentsofvariouspropertyattributes(aswellastheresultsfromthetextanalysis)Theimplicationofsuchaweakcorrelationisthatreviewsitesneedtoexpandonthelistofpossiblevariablessothattheratingsaccuratelyreflectconsumersrsquoneedsandexpectationsregardingserviceproviders

Furthersitemanagersmayhavetorethinkthecurrentpracticeofaveragingalltheratingsforagivenpropertybe-causesuchameanratingfailstotakeintoaccountthebiasesthatareinherentintheratingsystemsAsanexamplethisstudyfindssignificantunder-reportingbiasinthisonlinereviewsiteAcustomerrsquosdecisiontonotpostonlinereviewsmaymeanthelossofimportantinformationthatcanassisttheusersofthesitestomakemorereliableinferencesThemajorityoftodayrsquosfeedbackmechanismsdonotpubliclydisclosethenumberofsilenttransactions(ietransactionsforwhichnofeedbackwaspostedbycustomers)Suchinfor-mationshouldbecomeapartofaserviceproviderrsquosonlineprofileonratingsitesandotherfeedbackmechanismsItisimportanttoprovidemoreinformationandheuristicstohelptheconsumersnavigatethroughtheclutterandgettheinformationtheywantandseek

Itisimportanttoapplymodernsemanticanalysisandsentimentclassificationtechniquestostudythecontentofthereviewsandidentifyvariousdimensionsonwhichcon-sumersevaluatevacationhomesThedetailsprovidedinthereviewtextareinmanywaysabetterreflectionofcustomersatisfactionServiceproviderscanbetterpositionthemselvesusingthedimensionsuncoveredthroughtextanalysisandtargettheneedsandpreferencesoftheircustomers

TheresultsinthisstudyshowthatnegativereviewsaretypicallyassociatedwithagreaterwordcountthanpositivereviewsItisusefultofocusspecificallyonthesenegativereviewsbecausetheycanrevealpatternsofdeficienciesintheservicestandardsanddeliveryThesereviewsshouldbetreatedthesameascustomercommentcardsorletters19Furthercustomerswhowritesuchlengthynegativereviews

18LingKimberlyGerardBeenenPaulLudfordXiaoxingWangKevinChangXinLiDanCosleyDanFrankowskiLorenTerveenAlMRashidandRobertKrautldquoUsingSocialPsychologytoMotivateContributionstoOnlineCommunitiesrdquoJournal of Computer-Mediated Communication10no4(2005)19MilanR(2007)10thingsyoucandoinresponsetotravelerreviewsinHotelmarketingcom

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 15

aremostlikelytospreadnegativeword-of-mouthaboutthepropertyLengthyreviewsareexcellentsourcesofinforma-tionandshouldbeusedasthebasestounderstandthecomplainingbehavioroftheconsumer

WordcountmaybeusedasanadditionalsegmentationvariabletoidentifythoseconsumerswhoaremoreinvolvedandwillingtoprovidebalancedfeedbacktoboththereviewsiteaswellastopropertymanagementVariousotherresearchmethodssuchastelephoneinterviewsorfocusgroupsshouldbeusedtoelicitdetailedfeedbackfromsuchconsumersTheseactionsalsohelpinservicerecoverypost-purchaseengagementandrelationshipbuilding

OutlookWeforeseethreepotentialareasinvestigationthatcanaddvaluetocurrentbestpracticesWestilldonrsquotknowmuchaboutconsumersrsquomotivationtowriteonlinereviews20Recentstudiesidentifyvariousantecedentsofreviewwritingbutsufferfromsamplingbiassincetheysurveyaself-select-edsampleofconsumerswhoalreadyprovidedreviewsAstheresultsinthisstudyindicateunder-reportingbias(lackofmoderateandlowreviews)remainsprevalentinonlinereviewsitesFutureresearchshouldidentifysocio-psycho-logicalfactorsthatincreaseconsumersrsquopropensitytowriteonlinereviews

Futureresearchshouldexploretheimpactofthepres-ence(orlack)ofextremelynegativereviewswithinalargecorpusofpositivereviewsTheevidenceontheimpactofre-

20WangYandFesenmaierDR(2003)AssessingMotivationofCon-tributioninOnlineCommunitiesAnEmpiricalInvestigationofanOn-lineTravelCommunityElectronic Markets1333-45andYooKyungHandUlrikeGretzelldquoWhatMotivatesConsumerstoWriteOnlineTravelReviewsrdquoInformation Technology amp Tourism10No4(2008)283-95

viewvalence(negativeorpositive)isequivocalProponentsofconfirmatorybiassuggestthatconsumerslookforaffir-mativeevidencesupportingaproductchoicealreadymadeIfthatisthecasepositivereviewsaremorelikelytohaveagreatereffectonconsumeractionsOntheotherhandthenotionofnegativitybiassuggeststhatwhenconsumersareneutralnegativereviewstendtobecomemoresalientthanpositivereviews21Inthepresenceoflargenumbersofposi-tivereviewsconsumersspecificallyseeknegativereviewsthattheyfeelwillhelpthemidentifyspecificproblemswithaserviceUnderstandinghowconsumersreconciletheinformationprovidedbybothnegativeandpositivereviewshasimplicationsforthedesignandmanagementofreviewsites

Finallyareviewrsquoswordcounthasimportantimplica-tionsforconsumersrsquotrustinthereviewsWhenconsumersarewillingtoreadandcompareopen-endedcommentsfromotherconsumerstheamountofinformationinareviewcanmatterWOMforhighinvolvementproductsismorepersuasiveandincreasesthedecisionmakerrsquosconfi-dencewhenthemessagesenderprovidesgreaternumberofreasonsfortheiroverallratingLongerreviewsprovidemoreinformationandarelikelytobeperceivedasmorehelpfulandpersuasivethanshorterreviewsThewaysinwhichareviewiswrittenframedandpresentedisaninterestingfactorwhichhasrarelybeeninvestigatedandhasimportantimplicationsparticularlygivenourfindingthattheamountofinformationinareview(wordcount)variessignificantlywiththeratingsofthereviewn

21BaSandPavlouPA(2002)EvidenceoftheEffectofTrustBuildingTechnologyinElectronicMarketsPricePremiumsandBuyerBehaviorMIS Quarterly26243-268

Cornell Hospitality Quarterlyhttpcqxsagepubcom

2011 ReportsVol11No15DesigningaSelf-HealingServiceSystemAnIntegrativeModelbyRobertCFordandMichaelCSturman

Vol11No14ReversingtheGreenBacklashWhyLargeHospitalityCompaniesShouldWelcomeCrediblyGreenCompetitorsbyMichaelGiebelhausenPhDandHaeEunHelenChunPhD

Vol11No13DevelopingaSustainabilityMeasurementFrameworkforHotelsTowardanIndustry-wideReportingStructurebyEricRicaurte

Vol11No12CreatingValueforWomenBusinessTravelersFocusingonEmotionalOutcomesbyJudiBrownellPhD

Vol11No11CustomerLoyaltyANewLookattheBenefitsofImprovingSegmentationEffortswithRewardsProgramsbyClayVoorheesPhDMichaelMcCallPhDandRogerCalantonePhD

Vol11No10CustomerPerceptionsofElectronicFoodOrderingbySherylEKimes

Vol11No92011TravelIndustryBenchmarkingStatusofSeniorDestinationandLodgingMarketingExecutivesbyRohitVermaPhDandKenMcGill

Vol11No8SearchOTAsandOnlineBookingAnExpandedAnalysisoftheBillboardEffectbyChrisAndersonPhD

Vol11No7OnlineMobileandTextFoodOrderingintheUSRestaurantIndustrybySherylEKimesPhDandPhilippFLaqueacute

Vol11No6HotelGuestsrsquoReactionstoGuestRoomSustainabilityInitiativesbyAlexSusskindPhDandRohitVermaPhD

Vol11No5TheImpactofTerrorismandEconomicShocksonUSHotelsbyCathyAEnzRenaacutetaKosovaacuteandMarkLomannoVol11No4ImplementingHumanResourceInnovationsThreeSuccessStoriesfromtheServiceIndustrybyJustinSunandKateWalshPhD

Vol11No3Compendium2011

Vol11No2PositioningaPlaceDevelopingaCompellingDestinationBrandbyRobertJKwortnikPhDandEthanHawkesMBA

Vol11No1TheImpactofHealthInsuranceonEmployeeJobAnxietyWithdrawalBehaviorsandTaskPerformancebySeanWayPhDBillCarrollPhDAlexSusskindPhDandJoeCYLeng

2011 Hospitality ToolsVol2No1TheGameHasChangedANewParadigmforStakeholderEngagementbyMaryBethMcEuen

2011 Industry PerspectivesNo7MegaTips2TwentyTestedTechniquesforIncreasingYourTipsbyMichaelLynn

2011 ProceedingsVol3No5BuildingBrandsintheInternetAgeAnalyticsLoyaltyandCommunicationbyGlennWithiam

Vol3No4BraveNewWorldOnlineHotelDistributionbyGlennWithiam

Vol3No3SocialMediaandtheHospitalityIndustryHoldingtheTigerbytheTailbyGlennWithiam

Vol3No2TheChallengeofHotelandRestaurantSustainabilityFindingProfitinldquoBeingGreenrdquobyGlennWithiam

Vol3No1CautiousOptimismCHRSExaminesHospitalityIndustryTrendsbyGlennWithiam

2010 ReportsVol10No18HowTravelersUseOnlineandSocialMediaChannelstoMakeHotel-choiceDecisionsbyLauraMcCarthyDebraStockandRohitVermaPhD

Vol10No17PublicorPrivateTheHospitalityInvestmentDecisionbyQingzhongMaPhDandAthenaWeiZhangPhD

Vol10No16BestPracticesinSearchEngineMarketingandOptimizationTheCaseoftheStJamesHotelbyGregBodenlcosVictorBogertDanGordonCarterHearneandChrisKAndersonPhD

Vol10No15TheImpactofPrix Fixe MenuPriceFormatsonGuestsrsquoDealPerceptionbyShuoWangandMichaelLynnPhD

Cornell Center for Hospitality Research

Publication Indexwwwchrcornelledu

The Executive Path Hospitality Leadership Through Learning

Complete program information and applications available online

wwwhotelschoolcornelleduexecedPhone + 1 607 255 4919 Email exec_ed_hotelcornelledu

Professionals from around the world are invited to attend 3-day 10-day or online courses at the worldrsquos leading institute for hospitality management education in

Visit our website to apply

Explore develop and apply ideas with global hospitality leaders and

expert Cornell professors

Success

AdvancingBusiness

andPersonal

bull Strategic Leadershipbull Financebull Foodservicebull Human Resources

bull Marketingbull Operationsbull Real Estate

No2RetaliationWhyanIncreaseinClaimsDoesNotMeantheSkyIsFallingbyDavidSherwynJDandGreggGilmanJD

2009 ToolsToolNo12MeasuringtheDiningExperienceTheCaseofVitaNovabyKeshPrasadandFredJDeMiccoPhD

Cornell Center for Hospitality ResearchPublication Index

The Executive Path Hospitality Leadership Through Learning

Complete program information and applications available online

wwwhotelschoolcornelleduexecedPhone + 1 607 255 4919 Email exec_ed_hotelcornelledu

Professionals from around the world are invited to attend 3-day 10-day or online courses at the worldrsquos leading institute for hospitality management education in

Visit our website to apply

Explore develop and apply ideas with global hospitality leaders and

expert Cornell professors

Success

AdvancingBusiness

andPersonal

bull Strategic Leadershipbull Financebull Foodservicebull Human Resources

bull Marketingbull Operationsbull Real Estate

visit our website to apply

wwwchrcornell edu

Page 15: Unscrambling the Puzzling Matter of Online Consumer ... · Philips Hospitality Americas Chris Proulx, CEO, eCornell & Executive Education Carolyn D. Richmond, Partner, Hospitality

CornellHospitalityReportbullAugust2011bullwwwchrcornelledu 15

aremostlikelytospreadnegativeword-of-mouthaboutthepropertyLengthyreviewsareexcellentsourcesofinforma-tionandshouldbeusedasthebasestounderstandthecomplainingbehavioroftheconsumer

WordcountmaybeusedasanadditionalsegmentationvariabletoidentifythoseconsumerswhoaremoreinvolvedandwillingtoprovidebalancedfeedbacktoboththereviewsiteaswellastopropertymanagementVariousotherresearchmethodssuchastelephoneinterviewsorfocusgroupsshouldbeusedtoelicitdetailedfeedbackfromsuchconsumersTheseactionsalsohelpinservicerecoverypost-purchaseengagementandrelationshipbuilding

OutlookWeforeseethreepotentialareasinvestigationthatcanaddvaluetocurrentbestpracticesWestilldonrsquotknowmuchaboutconsumersrsquomotivationtowriteonlinereviews20Recentstudiesidentifyvariousantecedentsofreviewwritingbutsufferfromsamplingbiassincetheysurveyaself-select-edsampleofconsumerswhoalreadyprovidedreviewsAstheresultsinthisstudyindicateunder-reportingbias(lackofmoderateandlowreviews)remainsprevalentinonlinereviewsitesFutureresearchshouldidentifysocio-psycho-logicalfactorsthatincreaseconsumersrsquopropensitytowriteonlinereviews

Futureresearchshouldexploretheimpactofthepres-ence(orlack)ofextremelynegativereviewswithinalargecorpusofpositivereviewsTheevidenceontheimpactofre-

20WangYandFesenmaierDR(2003)AssessingMotivationofCon-tributioninOnlineCommunitiesAnEmpiricalInvestigationofanOn-lineTravelCommunityElectronic Markets1333-45andYooKyungHandUlrikeGretzelldquoWhatMotivatesConsumerstoWriteOnlineTravelReviewsrdquoInformation Technology amp Tourism10No4(2008)283-95

viewvalence(negativeorpositive)isequivocalProponentsofconfirmatorybiassuggestthatconsumerslookforaffir-mativeevidencesupportingaproductchoicealreadymadeIfthatisthecasepositivereviewsaremorelikelytohaveagreatereffectonconsumeractionsOntheotherhandthenotionofnegativitybiassuggeststhatwhenconsumersareneutralnegativereviewstendtobecomemoresalientthanpositivereviews21Inthepresenceoflargenumbersofposi-tivereviewsconsumersspecificallyseeknegativereviewsthattheyfeelwillhelpthemidentifyspecificproblemswithaserviceUnderstandinghowconsumersreconciletheinformationprovidedbybothnegativeandpositivereviewshasimplicationsforthedesignandmanagementofreviewsites

Finallyareviewrsquoswordcounthasimportantimplica-tionsforconsumersrsquotrustinthereviewsWhenconsumersarewillingtoreadandcompareopen-endedcommentsfromotherconsumerstheamountofinformationinareviewcanmatterWOMforhighinvolvementproductsismorepersuasiveandincreasesthedecisionmakerrsquosconfi-dencewhenthemessagesenderprovidesgreaternumberofreasonsfortheiroverallratingLongerreviewsprovidemoreinformationandarelikelytobeperceivedasmorehelpfulandpersuasivethanshorterreviewsThewaysinwhichareviewiswrittenframedandpresentedisaninterestingfactorwhichhasrarelybeeninvestigatedandhasimportantimplicationsparticularlygivenourfindingthattheamountofinformationinareview(wordcount)variessignificantlywiththeratingsofthereviewn

21BaSandPavlouPA(2002)EvidenceoftheEffectofTrustBuildingTechnologyinElectronicMarketsPricePremiumsandBuyerBehaviorMIS Quarterly26243-268

Cornell Hospitality Quarterlyhttpcqxsagepubcom

2011 ReportsVol11No15DesigningaSelf-HealingServiceSystemAnIntegrativeModelbyRobertCFordandMichaelCSturman

Vol11No14ReversingtheGreenBacklashWhyLargeHospitalityCompaniesShouldWelcomeCrediblyGreenCompetitorsbyMichaelGiebelhausenPhDandHaeEunHelenChunPhD

Vol11No13DevelopingaSustainabilityMeasurementFrameworkforHotelsTowardanIndustry-wideReportingStructurebyEricRicaurte

Vol11No12CreatingValueforWomenBusinessTravelersFocusingonEmotionalOutcomesbyJudiBrownellPhD

Vol11No11CustomerLoyaltyANewLookattheBenefitsofImprovingSegmentationEffortswithRewardsProgramsbyClayVoorheesPhDMichaelMcCallPhDandRogerCalantonePhD

Vol11No10CustomerPerceptionsofElectronicFoodOrderingbySherylEKimes

Vol11No92011TravelIndustryBenchmarkingStatusofSeniorDestinationandLodgingMarketingExecutivesbyRohitVermaPhDandKenMcGill

Vol11No8SearchOTAsandOnlineBookingAnExpandedAnalysisoftheBillboardEffectbyChrisAndersonPhD

Vol11No7OnlineMobileandTextFoodOrderingintheUSRestaurantIndustrybySherylEKimesPhDandPhilippFLaqueacute

Vol11No6HotelGuestsrsquoReactionstoGuestRoomSustainabilityInitiativesbyAlexSusskindPhDandRohitVermaPhD

Vol11No5TheImpactofTerrorismandEconomicShocksonUSHotelsbyCathyAEnzRenaacutetaKosovaacuteandMarkLomannoVol11No4ImplementingHumanResourceInnovationsThreeSuccessStoriesfromtheServiceIndustrybyJustinSunandKateWalshPhD

Vol11No3Compendium2011

Vol11No2PositioningaPlaceDevelopingaCompellingDestinationBrandbyRobertJKwortnikPhDandEthanHawkesMBA

Vol11No1TheImpactofHealthInsuranceonEmployeeJobAnxietyWithdrawalBehaviorsandTaskPerformancebySeanWayPhDBillCarrollPhDAlexSusskindPhDandJoeCYLeng

2011 Hospitality ToolsVol2No1TheGameHasChangedANewParadigmforStakeholderEngagementbyMaryBethMcEuen

2011 Industry PerspectivesNo7MegaTips2TwentyTestedTechniquesforIncreasingYourTipsbyMichaelLynn

2011 ProceedingsVol3No5BuildingBrandsintheInternetAgeAnalyticsLoyaltyandCommunicationbyGlennWithiam

Vol3No4BraveNewWorldOnlineHotelDistributionbyGlennWithiam

Vol3No3SocialMediaandtheHospitalityIndustryHoldingtheTigerbytheTailbyGlennWithiam

Vol3No2TheChallengeofHotelandRestaurantSustainabilityFindingProfitinldquoBeingGreenrdquobyGlennWithiam

Vol3No1CautiousOptimismCHRSExaminesHospitalityIndustryTrendsbyGlennWithiam

2010 ReportsVol10No18HowTravelersUseOnlineandSocialMediaChannelstoMakeHotel-choiceDecisionsbyLauraMcCarthyDebraStockandRohitVermaPhD

Vol10No17PublicorPrivateTheHospitalityInvestmentDecisionbyQingzhongMaPhDandAthenaWeiZhangPhD

Vol10No16BestPracticesinSearchEngineMarketingandOptimizationTheCaseoftheStJamesHotelbyGregBodenlcosVictorBogertDanGordonCarterHearneandChrisKAndersonPhD

Vol10No15TheImpactofPrix Fixe MenuPriceFormatsonGuestsrsquoDealPerceptionbyShuoWangandMichaelLynnPhD

Cornell Center for Hospitality Research

Publication Indexwwwchrcornelledu

The Executive Path Hospitality Leadership Through Learning

Complete program information and applications available online

wwwhotelschoolcornelleduexecedPhone + 1 607 255 4919 Email exec_ed_hotelcornelledu

Professionals from around the world are invited to attend 3-day 10-day or online courses at the worldrsquos leading institute for hospitality management education in

Visit our website to apply

Explore develop and apply ideas with global hospitality leaders and

expert Cornell professors

Success

AdvancingBusiness

andPersonal

bull Strategic Leadershipbull Financebull Foodservicebull Human Resources

bull Marketingbull Operationsbull Real Estate

No2RetaliationWhyanIncreaseinClaimsDoesNotMeantheSkyIsFallingbyDavidSherwynJDandGreggGilmanJD

2009 ToolsToolNo12MeasuringtheDiningExperienceTheCaseofVitaNovabyKeshPrasadandFredJDeMiccoPhD

Cornell Center for Hospitality ResearchPublication Index

The Executive Path Hospitality Leadership Through Learning

Complete program information and applications available online

wwwhotelschoolcornelleduexecedPhone + 1 607 255 4919 Email exec_ed_hotelcornelledu

Professionals from around the world are invited to attend 3-day 10-day or online courses at the worldrsquos leading institute for hospitality management education in

Visit our website to apply

Explore develop and apply ideas with global hospitality leaders and

expert Cornell professors

Success

AdvancingBusiness

andPersonal

bull Strategic Leadershipbull Financebull Foodservicebull Human Resources

bull Marketingbull Operationsbull Real Estate

visit our website to apply

wwwchrcornell edu

Page 16: Unscrambling the Puzzling Matter of Online Consumer ... · Philips Hospitality Americas Chris Proulx, CEO, eCornell & Executive Education Carolyn D. Richmond, Partner, Hospitality

Cornell Hospitality Quarterlyhttpcqxsagepubcom

2011 ReportsVol11No15DesigningaSelf-HealingServiceSystemAnIntegrativeModelbyRobertCFordandMichaelCSturman

Vol11No14ReversingtheGreenBacklashWhyLargeHospitalityCompaniesShouldWelcomeCrediblyGreenCompetitorsbyMichaelGiebelhausenPhDandHaeEunHelenChunPhD

Vol11No13DevelopingaSustainabilityMeasurementFrameworkforHotelsTowardanIndustry-wideReportingStructurebyEricRicaurte

Vol11No12CreatingValueforWomenBusinessTravelersFocusingonEmotionalOutcomesbyJudiBrownellPhD

Vol11No11CustomerLoyaltyANewLookattheBenefitsofImprovingSegmentationEffortswithRewardsProgramsbyClayVoorheesPhDMichaelMcCallPhDandRogerCalantonePhD

Vol11No10CustomerPerceptionsofElectronicFoodOrderingbySherylEKimes

Vol11No92011TravelIndustryBenchmarkingStatusofSeniorDestinationandLodgingMarketingExecutivesbyRohitVermaPhDandKenMcGill

Vol11No8SearchOTAsandOnlineBookingAnExpandedAnalysisoftheBillboardEffectbyChrisAndersonPhD

Vol11No7OnlineMobileandTextFoodOrderingintheUSRestaurantIndustrybySherylEKimesPhDandPhilippFLaqueacute

Vol11No6HotelGuestsrsquoReactionstoGuestRoomSustainabilityInitiativesbyAlexSusskindPhDandRohitVermaPhD

Vol11No5TheImpactofTerrorismandEconomicShocksonUSHotelsbyCathyAEnzRenaacutetaKosovaacuteandMarkLomannoVol11No4ImplementingHumanResourceInnovationsThreeSuccessStoriesfromtheServiceIndustrybyJustinSunandKateWalshPhD

Vol11No3Compendium2011

Vol11No2PositioningaPlaceDevelopingaCompellingDestinationBrandbyRobertJKwortnikPhDandEthanHawkesMBA

Vol11No1TheImpactofHealthInsuranceonEmployeeJobAnxietyWithdrawalBehaviorsandTaskPerformancebySeanWayPhDBillCarrollPhDAlexSusskindPhDandJoeCYLeng

2011 Hospitality ToolsVol2No1TheGameHasChangedANewParadigmforStakeholderEngagementbyMaryBethMcEuen

2011 Industry PerspectivesNo7MegaTips2TwentyTestedTechniquesforIncreasingYourTipsbyMichaelLynn

2011 ProceedingsVol3No5BuildingBrandsintheInternetAgeAnalyticsLoyaltyandCommunicationbyGlennWithiam

Vol3No4BraveNewWorldOnlineHotelDistributionbyGlennWithiam

Vol3No3SocialMediaandtheHospitalityIndustryHoldingtheTigerbytheTailbyGlennWithiam

Vol3No2TheChallengeofHotelandRestaurantSustainabilityFindingProfitinldquoBeingGreenrdquobyGlennWithiam

Vol3No1CautiousOptimismCHRSExaminesHospitalityIndustryTrendsbyGlennWithiam

2010 ReportsVol10No18HowTravelersUseOnlineandSocialMediaChannelstoMakeHotel-choiceDecisionsbyLauraMcCarthyDebraStockandRohitVermaPhD

Vol10No17PublicorPrivateTheHospitalityInvestmentDecisionbyQingzhongMaPhDandAthenaWeiZhangPhD

Vol10No16BestPracticesinSearchEngineMarketingandOptimizationTheCaseoftheStJamesHotelbyGregBodenlcosVictorBogertDanGordonCarterHearneandChrisKAndersonPhD

Vol10No15TheImpactofPrix Fixe MenuPriceFormatsonGuestsrsquoDealPerceptionbyShuoWangandMichaelLynnPhD

Cornell Center for Hospitality Research

Publication Indexwwwchrcornelledu

The Executive Path Hospitality Leadership Through Learning

Complete program information and applications available online

wwwhotelschoolcornelleduexecedPhone + 1 607 255 4919 Email exec_ed_hotelcornelledu

Professionals from around the world are invited to attend 3-day 10-day or online courses at the worldrsquos leading institute for hospitality management education in

Visit our website to apply

Explore develop and apply ideas with global hospitality leaders and

expert Cornell professors

Success

AdvancingBusiness

andPersonal

bull Strategic Leadershipbull Financebull Foodservicebull Human Resources

bull Marketingbull Operationsbull Real Estate

No2RetaliationWhyanIncreaseinClaimsDoesNotMeantheSkyIsFallingbyDavidSherwynJDandGreggGilmanJD

2009 ToolsToolNo12MeasuringtheDiningExperienceTheCaseofVitaNovabyKeshPrasadandFredJDeMiccoPhD

Cornell Center for Hospitality ResearchPublication Index

The Executive Path Hospitality Leadership Through Learning

Complete program information and applications available online

wwwhotelschoolcornelleduexecedPhone + 1 607 255 4919 Email exec_ed_hotelcornelledu

Professionals from around the world are invited to attend 3-day 10-day or online courses at the worldrsquos leading institute for hospitality management education in

Visit our website to apply

Explore develop and apply ideas with global hospitality leaders and

expert Cornell professors

Success

AdvancingBusiness

andPersonal

bull Strategic Leadershipbull Financebull Foodservicebull Human Resources

bull Marketingbull Operationsbull Real Estate

visit our website to apply

wwwchrcornell edu

Page 17: Unscrambling the Puzzling Matter of Online Consumer ... · Philips Hospitality Americas Chris Proulx, CEO, eCornell & Executive Education Carolyn D. Richmond, Partner, Hospitality

No2RetaliationWhyanIncreaseinClaimsDoesNotMeantheSkyIsFallingbyDavidSherwynJDandGreggGilmanJD

2009 ToolsToolNo12MeasuringtheDiningExperienceTheCaseofVitaNovabyKeshPrasadandFredJDeMiccoPhD

Cornell Center for Hospitality ResearchPublication Index

The Executive Path Hospitality Leadership Through Learning

Complete program information and applications available online

wwwhotelschoolcornelleduexecedPhone + 1 607 255 4919 Email exec_ed_hotelcornelledu

Professionals from around the world are invited to attend 3-day 10-day or online courses at the worldrsquos leading institute for hospitality management education in

Visit our website to apply

Explore develop and apply ideas with global hospitality leaders and

expert Cornell professors

Success

AdvancingBusiness

andPersonal

bull Strategic Leadershipbull Financebull Foodservicebull Human Resources

bull Marketingbull Operationsbull Real Estate

visit our website to apply

wwwchrcornell edu

Page 18: Unscrambling the Puzzling Matter of Online Consumer ... · Philips Hospitality Americas Chris Proulx, CEO, eCornell & Executive Education Carolyn D. Richmond, Partner, Hospitality

wwwchrcornell edu