Building Brand Image Through Event Sponsorship:The Role of Image TransferKevin P. Gwinner and John Eaton

Past sponsorship research has primarily focused on awareness building strategies, and has virtually ignoredbrand image issues. As a result, little guidance is available for firms that seek to use sponsorship opportuni-ties to aid in brand positioning. This study reports the results of an experiment using undergraduate studentsubjects, who assessed the degree to which a sporting event's image was transferred to a brand through eventsponsorship activity. Subjects in the sponsorship pairing treatment were more likely to report similarities onbrand-event personality components than subjects who were not exposed to the event-brand sponsorship link,thus supporting the notion that sponsorship results in image transfer. Further, we found that when event andbrand are matched on either an image or functional basis the transfer process is enhanced. Managementimplications for sporting event sponsorship and future research directions are discussed.

Kevin P. Gwinner (Ph.D., ArizonaState University) is Assistant Professorof Marketing, Kansas State University.John Elaton (M.B A , University ofToledo) is a doctoral candidate inmarketing at Arizona State University.The authors wish to thank StephenNowlis (Arizona State University) forguidance related to experimentaldesign issues and the four anony-mous Journal of Advertisingreviewers and former editor LesCarlson (Clemson University) fortheir helpful comments on earlierversions of this manuscript.

Joumal of Advertising,Volume XXVIII. Number 4Winter 1999

While firms enter into sponsorship arrangements for a variety of reasons,two of the most common are: (1) to increase brand awareness, and (2) toestablish, strengthen, or change brand image (Comwell and Maignan 1998;Crowley 1991; Gwinner 1997; Marshall and Cook 1992; Meenaghan 1991;Meerabeau et al. 1991). Typically, strategies aimed at increasing brandawareness are implemented using a multitude of promotional media andare designed to have tbe sponsoring brand exposed to as many potentialconsumers as possible. Past researcb has examined the effectiveness ofthese awareness building strategies through a variety of methods (e.g., totalevent attendance, exit polls, sales following the event, and number of mediamentions). Regrettably, far less research attention has focused on brandimage issues. Indeed, researchers have suggested that little is understoodabout what makes sponsorship "work," particularly with regard to imageassociation (Javalgi et al. 1994; Lee, Sandier and Shani 1997).

The purpose of this paper is to gain insight into the brand image aspects ofsponsorsbip. Specifically, based on tbeoretical perspectives from the celeb-rity endorsement, schema, and advertising literatures, we propose and testseveral relationships involving the infiuence of sporting event sponsorshipon the sponsoring brand's image.

Image Transfer in Sporting Event SponsorshipsBrand image bas been defined as "perceptions about a brand as reflected

by the brand associations held in memory" (Keller 1993, p. 3). This defini-tion takes an associate memory network view, in tbat brand image is basedupon linkages a consumer holds in his/her memory structure regarding thebrand. These linkages, or in Keller's terminology, "brand associations," aredeveloped from a variety of sources including brand and product categoryexperiences, product attributes, price information, positioning in promo-tional communications, packaging, user imagery (e.g., typical brand users),and usage occasion (Keller 1993). From a tbeoretical position, Keller (1993)suggests tbat brand associations can be infiuenced when a brand becomeslinked with a celebrity through an endorsement or linked with a sportingevent through sponsorsbip activities. In these cases, the pre-existing asso-

48 The Journal ofAdvertisingciations held in consumers' memories regarding a ce-lebrity or sporting event become linked in memory withthe endorsed or sponsoring brand. In essence, the ce-lebrity or event image is transferred to the hrand.

This transfer of associations is consistent withMcCracken's (1989) view ofthe celebrity endorsementprocess. McCracken eschews the "credibility" and "at-tractiveness" models of endorsement used to explainthe persuasive nature of endorsers. Instead he offersa theory of meaning transfer, where "meaning" refersto an overall assessment of what a celebrity "repre-sents" to the consumer. This meaning is built uponan individual's interpretation ofthe celebrity's publicimage as demonstrated in "television, movies, mili-tary, athletics, and other careers" (McCracken 1989,p. 315). According to this theory, the meaning attrib-uted to celebrities moves from the celebrity endorserto the product when the two are paired in an adver-tisement (McCracken 1989). The transfer process iscompleted when a consumer acquires/consumes theproduct, thus transferring the meaning to the user.

With regard to implications for brand image, sport-ing event8 and celebrity endorsers are similar on twolevels. First, consumers can associate both sportingevents and celebrities with particular meanings. Whilecelebrities derive their meanings from consumer per-ceptions of their various public activities (e.g., movies,athletics, politics, etc.), the meanings associated withsporting events are derived from the type of event, theevent's characteristics (e.g., professional status, venue,size, etc.), and individual consumer factors such as one'spast experiences with the event (Gwinner 1997). Sec-ond, events may act in a manner analogous to celebrityendorsers in the transfer of image to sponsoring brands.That is, just as consumers associate a celebrity's "mean-ings" with the brand they endorse, consumers may alsoassociate a sporting event's "memings" with a sponsor-ing brand. Obviously, if this image transfer process isoccurring, then brand managers considering sponsor-ship arrangements should not only consider exposureissues (i.e., brand awareness) but should also take intoaccount the congruence between a sporting event's im-age and the image/positioning goals for their brands.While it is possible that the directionality ofthe imagetransfer may move from brand to event rather thanevent to brand, this is less likely to occur when theevent has a strong established image relative to thesponsoring brand. Further, since the primary focus ofthe spectator is typically on the activities ofthe sport-ing event rather than on the sponsors, the event's im-age is likely to be more salient in their mind, suggest-ing the image transfer process would move from eventto brand. This discussion leads to the first hypothesis:

HI: A sporting event's image will transfer toa sponsoring brand's image when they arelinked through sponsorship.

The Impact of Event and Brand Match-up on Image Transfer

Many scholars have examined or commented onthe importance of matching the characteristics ofspokespersons with the characteristics of the prod-ucts they endorse (see Lynch and Schuler [1994] foran excellent review of this literature). Generally, thesetypes of studies have found that a match betweenendorser and brand leads to a variety of positive out-comes for firms including enhanced spokesperson ex-pertise/credibility, a more positive attitude towardthe ad, a more positive attitude toward the brand andhigher brand recall.

Kahle and Homer (1985) were among the first toempirically examine and propose a "match-up" hy-pothesis in the context of celebrity endorsement. Theseauthors argued and found support for the notion thatadvertising effectiveness is increased when the im-age of the celebrity converges with the image of theendorsed product. Specifically, they found physicallyattractive celebrity endorsers of a beauty enhancingproduct (i.e., endorser-brand match) to have a posi-tive influence on consumer's brand attitudes, pur-chase intentions, brand recall, and recall of adver-tisement arguments. Building on Kahle and Homer,many spokesperson-brand congruence studies havebeen conducted by manipulating some physical at-tribute of the spokesperson to be in-congruence orout-of-congruence with a given product. For example,in one experiment. Lynch and Schuler (1994) ma-nipulated muscularity ofthe spokesperson to be in orout of congruence with products that either helped toproduce muscularity (e.g., exercise equipment) or prod-ucts perceived to be masculine in nature (e.g., carbatteries). In another study, Kamins (1990) manipu-lated spokesperson attractiveness to be in or out ofcongruence with an attractiveness related product.The Kamins study found that when product andspokesperson physical characteristics were congru-ent, then spokesperson believability/knowledge wasincreased, while Lynch and Schuler (1994) found thatcongruency led to high perceived spokesperson knowl-edge. One notable exception to the manipulation ofspokesperson physical characteristics is Misra andBeatty (1990) who examined image congruence in aholistic manner akin to McCracken's notion of mean-ing. In their study, Misra and Beatty matched spokes-person characteristics with product characteristics.

Winter 1999 49

For example, in their pretest of celebrities, ClintEastwood was associated with the characteristics of"tough" and "rugged." In the congruent condition thiscelebrity was paired with a flctitious brand of jeans(Unitough jeans), while in the incongruent conditionhe was paired with a fictitious board game calledFunnybone. Results of this study indicated that thecongruence condition resulted in higher recall andmore favorable brand attitudes than incongruent orneutral congruency pairings. Kamins and Gupta(1994) also manipulated spokesperson-product con-gruence in terms of image. They found increased con-gruence resulted in perceptions of higher believabil-ity and attractiveness ofthe spokesperson and a morefavorable product attitude.

So why does this match-up hypothesis seem to be ineffect? One of the more compelling arguments pro-posed is based on schema theory (Lynch and Schuler1994; Misra and Beatty 1990). A schema is a cogni-tive structure that represents knowledge about a typeof stimulus, for example, a person, event, or object(Bartlett 1932; Lord and Foti 1986). Schema theory isbased on research which found that memory is not averbatim account of past experiences, but rather ablend of both specific memories as well as generalabstractions about tj^es of people, activities, and ob-jects (Bartlett 1932; Rumelhart and Ortony 1977).Schema represents a mechanism to allow individualsto function in a complex environment. That is, in-stead of having to recall from memory what behav-iors are appropriate in a specific situation (e.g., board-ing a United Airlines flight) or what evaluations havebeen made of some specific person (e.g.. Dr. Bemhard)or specific object (Campbell's soup), one is able tosimply recall knowledge related to the general type ofsituation (airline boarding), person (heart surgeon),or object (soup).

With regard to the use of schema theory in support ofthe match-up hypothesis, Misra and Beatty (1990) foundevidence of a "filtering model," which suggested spokes-person characteristics that are incongruent with brandschema characteristics will be 'Tiltered out" and notencoded as well as congruent information. They arguedthat the better recall demonstrated by subjects in theircongruent condition is a result of better or more effec-tive encoding of information. Further, they proposed,but did not test, that this encoded information associ-ated with a congruent celebrity spokesperson schemawould become integrated with the product's schema. Ifthis were to hold true, then one would expect theschemas of celebrities and the schemas ofthe productsthey endorse to become more similar, assuming con-gruence of some salient characteristics.

In an event sponsorship context, McDaniel (1999)has explored an aspect ofthe match-up hypothesis bymatching event and brand in terms of involvement.He found that subjects rated attitude toward the adsignificantly more positively when a highly involvingproduct (e.g., an automobile) was paired with a highlyinvolving sporting event (e.g., the Olympics) thanwhen the product was paired with a low involvementsporting event (e.g., PBA Bowling). In his study, event-product involvement match was not found to have aneffect on attitude toward the brand or purchase in-tention. While providing insight into one match updimension, McDaniel's study did not attempt to matchthe sporting event and the sponsoring brand on at-tributes related to their respective "meanings" in themanner that McCracken (1989) discussed or thatMisra and Beatty (1990) explored.

This begs the question, "on what basis might wejudge a sporting event to be similar or dissimilar to aproduct?" McDonald (1991) discusses the importanceof product relevance to the sponsored event, suggest-ing that it might occur directly or indirectly. Thedirect method occurs when the sponsoring firm's prod-ucts are (or could be) used in the event. Indirectly,relevance can be achieved if some aspect of thesponsor's image corresponds with the event. Gwinner(1997) has used the terms "functional based" and"image based similarity" to refer to the potential con-gruence between events and the brands/companiesthat act as sponsors. Consistent with McDonald (1991),Gwinner (1997) has suggested that functional basedsimilarity can occur when the sponsored brand "isactually used by the participants during the event;.."(p. 152). Examples of this type of similarity wouldinclude Seiko being an official timer at the U.S. OpenTennis Championships or Gatorade sponsoring theIronman Triathlon. In both cases, functional similar-ity is present because of the use of the sponsoringbrand in the event. Image based similarity has beendescribed as occurring when the "image of the eventis related to the image ofthe brand..." (Gwinner 1997,p. 152). For example, the Master's Golf Tournamentand Cadillac Automobiles may be similar in terms ofa prestige image. Drawing on the earlier schematheory discussion, it can be argued that congruentevent-brand information in the form of either func-tional or image based similarity will lead to enhancedimage transfer. Thus, extending the notion of thematch-up hypothesis found in the celebrity endorsercontext, we offer the following two-part hypothesis:

H2a: Similarity between brand and event willinfluence the image transfer such thatthe image transfer will be stronger for

60 The Journal of Advertisingbrands having functional-based similar-ity with tbe event they are sponsoringthan wben brands have no similaritywith the event.

H2b: Similarity between brand and event willinfluence the image transfer such thattbe image transfer will be stronger forbrands having image-based similaritywith the event they are sponsoring thanwhen brands have no similarity withthe event.

MethodOverview

To test the hypotheses an experiment was conductedusing one between groups factor (sponsorsbip: [yessponsorship tnd no sponsorship]) and one repeatedmeasures factor (level of event-brand similarity: [im-age based, functional based, and no similarity]). Wbilethere may be many ways to examine image transfer,our examination uses brand and event personality.Brand personality can be regarded as "the set of hu-man characteristics associated with a brand..." (Aaker1997, p. 347). Brand personality has been describedas an important aspect of brand image tbat is im-pacted by one's expectations ofthe type of person whowould use a particular product - user imagery - andin which situations a product might be used - usageimagery (Aaker 1997; Keller 1993; Plummer 1985).Both user and usage imagery can be communicatedin an event sponsorship context. The brand personal-ity concept is important because it serves as a mecha-nism upon which producers can differentiate theirgoods and services. This becomes especially criticalwhen other potentially differentiating features areperceived by consumers as equal across competingbrands (Plummer 1985).

Prete8t8The first pretest sought to find appropriate pair-

ings of sporting events and sponsoring products torepresent each of the three types of similarity (func-tional based, image based, and no similarity) for usein the main experiment. As illustrated in Table 1,eacb potential sporting event was paired with threedifferent brands. Gwinner (1997) suggests tbat cer-tain event characteristics (e.g., size, history, venue,etc.) will influence an event's image. Using this as aguide, we selected seven sporting events based ontheir national visibility and rich, long histories. This

was done because subjects in the main study neededto have some prior image of the event in order toincrease our confidence in the image transfer mea-sure. That is, they needed to have an image of theevent in order to have it transfer to the brand.

We developed scale items to measure functionaland image based similarity based on the definitionprovided by Gwinner (1997). Functional based simi-larity was measured using three items assessed onseven-point strongly disagree/strongly agree scales.The three items were: (1) "It is likely that (partici-pants) in the (event name) use (brand name) duringtbe (event name)," (2) "Wben I watch the (event name),I often see (brand name) being (used)," and (3) "(Brandname) is not a product that (participants) in tbe (eventname) would consider (using)." The third item wasreverse coded. The parenthetical "participants" labelin these questions was replaced by the appropriateparticipant title, depending upon the specific event(e.g., player, rider, driver, etc.). Cronbacb's alpha forthis scale is .89, thus demonstrating good reliability.

Image based similarity was also assessed withseven-point strongly disagree/strongly agree anchoredscales using the following three measures: (1) "The(event name) and (brand name) have a similar im-age," (2) "The ideas I associate with (brand name) arerelated to the ideas I associate with the (event name),"and (3) "My image of the (event name) is very differ-ent from the image I have of (brand name)." Thethird iniage based similarity measure was reversecoded in the analysis. Cronbach's alpha for this scaleis .90. The best "No Similarity" pairing wasoperationalized as the event-product score receivingtbe lowest score on a summed scale consisting of allsix items listed above.

One hundred and thirty-five undergraduate stu-dents enrolled in a marketing management courseresponded to the similarity pretest survey. In orderto reduce respondent fatigue, each subject was ran-domly assigned to respond to questions regardingthree of the seven events. This resulted in a useablesample of between 41 and 50 subjects per event.

As indicated in Table 1, for the image similaritycondition, the pairing of the U.S. Open Golf Champi-onship and Acura Automobiles was found to be thebest pairing. For functional similarity, the pretestshowed that the Indianapolis 500 Auto Race andGoodyear Tires was the best pairing. Finally, the bestillustration of tbe no similarity condition was WorldCup Soccer and Camel Cigarettes. These pairingswere assessed after removing those subjects whoscored below the scale mid-point on a brand familiar-ity question in order to increase validity.

Winter 1999

Similarity Type

Image Based

Functional based

No similarity

Tabie 1Pretest 1 Pairings of Sporting Events and Sponsoring Products

Event Alternatives

Professional Beach Volleyball

KentucKy Derby

U.S. Open Golf Championship

indianapoiis 500 Auto Race

NCAA basketball tournament

Worid Cup Soccer

Rose Bowl

Product Pairing Alternatives

Cuervo Gold tequilaDorito's tortilla chipsNintendo video gamesAmerican ExpressOldsmobile automobilesJohn Hancock insurance

) Acura automobiiesSony camcordersMichelob beerGoodyear tiresPennzoil motor oilShell gasolineChampion brand uniformsReebok shoesPowerade sports drink

Dell computersClorox bleachCamei cigarettes

Irish Spring soapRenuzit air freshenersLevi's jeans

51

Mean Similarity (std dev)9.99.7ao

10.39.28.8

12.110.59.4

19.217.014.116.916.015.5

Functional6.9 (4.8)6.4 (4.5)4.1 (2.0)Functional

5.3 (3.7)4.7 (2.5)4.1 (2.3)

(5.8)(4.7)(5.3)(5.0)(5.1)(4.0)(5.5)(5.1)(5.2)(2.6)(3.6)(4.8)(3.1)(4.4)(4.3)

image7.8 (4.2)8.4 (5.1)4.8 (3.1)

image7.3 (5.1)5.2 (3.6)7.5 (5.1)

Note: Bold faced event-product combinations were rated as the best representatives of their respective similarity type category (based onhigh mean value for image and functional simiiarity pairings and low mean value for no simiiarity pairing) and were subsequently used inthe main study.

A second pretest was conducted to create a set ofpersonality-oriented adjectives that could be used todescribe image dimensions of each ofthe three eventsselected from the first pretest. Twenty at^ectives weregenerated by the authors for each ofthe three sportingevents. In order to increase the saliency of the task,only adjectives that could potentially describe the par-ticular event were included (Graeff 1996). Eighty-oneundergraduate students enrolled in a marketing courseused a seven-point scale to rate the 60 adjectives ontheir usefulness in describing each of the three events(20 adjectives per event). The ten adjectives rated asmost useful in describing the event were selected foruse in the experiment and are listed in Table 2.

Experimental ProcedureThree hundred and sixty undergraduate business

students participated in the experiment for extra

credit. Because of the use of sporting events withlong, rich histories, we believe a student sample willbe familiar with the events and, therefore, the resultswill be generalizable to a larger population. Subjectswere randomly assigned to one of the two sponsor-ship treatments. Randomization assures that the im-pact of individuals* prior product schemas will notbias the results in any given treatment. In each con-dition, subjects assessed event-brand combinationsrepresenting all three similarity conditions (i.e., im-age based, functional, no similarity). Cell sizes wereevenly distributed, resulting in a uniform 180 sub-jects per cell assignment.

Independent VariablesSponsorship condition. This condition consisted of

two levels, one in which the event and brand werepaired in a sponsor relationship ("yes sponsorship")

52 The JourncU ofAdvertisingTable 2

Adjectives Used in iMeasurement of Image Transfer from Event to BrandU.S. Open Golf Championship

(Image Based Similarity)CalmMatureLeisurelyCleanFormalCivilizedAccuratePressureOrderlySlow

Indianapolis 500 Auto Race(Functional Based Similarity)

FastDangerousExcitingAggressiveMasculineWildHistoricTacticalStrategicMonotonous

World Cup Soccer(No Similarity)ActiveMulti-culturalEnergeticExcitingForeignThrillingYouthfulTacticalToughBrazen

and one in which they were not ("no sponsorship"). Inthe "yes sponsorship" level subjects viewed three blackand white photographs, each depicting one ofthe sport-ing events. Each picture was designed to resemble amagazine advertisement for the event and the actual logoofthe sponsoring brand was superimposed on the photo-graph. In addition, ad copy typical of a sponsorship ar-rangement was included on the photo (e.g., "Camel ciga-rettes is proud to sponsor World Cup Soccer W).

The cover page also added to the sponsorship ma-nipulation through the survey title, 'Troduct Spon-sorship of Sporting Events Survey" and through thesurvey instructions which referred to the sponsorshipties in the photographs ("Yes Sponsorship" instruc-tions: "We are conducting this study to better under-stand how students feel about corporate sponsorshipof sporting events"). For the "no sponsorship" level nophotos were included; the survey was titled simply,"Image Survey," and mention ofthe sponsorship wasnot included in the instructions ("No Sponsorship"instructions: "We are conducting this study to betterunderstand the images students have regarding dif-ferent events and products").

Similarity condition. As described in pretest 1, thesimilarity condition was manipulated by pairing sport-ing events and sponsoring brands to create three simi-lsirity levels: image similarity (the U.S. Open GolfChampionship and Acura Automobile), functionalsimilarity (the Indianapolis 500 Auto Race andGoodyear Tires), and no similarity (World Cup Soccerand Camel Cigarettes).

Dependent MeasuresAdjective based image transfer measure. As stated

above, this study examines brand/event personality

as a specific aspect of image. Recall that in pretesttwo, three groups of ten adjectives were selected asbeing useful in describing each ofthe three events. Inthe experiment each of the 10 adjectives were ratedas to how well it described the particular event (l=verywell; 7=not at all) and then, separately, subjects re-sponded as to how well the same 10 adjectives de-scribed the brand. If an image transfer is occurring,one would expect the image of the event and theimage of the brand to be more similar in the "yessponsorship" condition, as the event's image would be"transferring" to the brand. Accordingly, we calcu-lated a measure of congruence/similarity by takingthe sum of the absolute differences between the cor-responding adjectives in the event and the brand rat-ings. For example, if a subject rated the adjective"mature" as a "2" for the event and a "5" for thebrand, then the absolute difference for those corre-sponding adjectives would be "3." We summed theten absolute difference scores for each event-brandpair to create a congruence index. Smaller numbersin the index indicate greater congruence (i.e., lessdifference between event and brand). Hypothesis onewould predict that this measure will be significantlysmaller for those subjects in the "yes sponsorship"level than in the "no sponsorship" levelindicatingmore similarity.

Holistic image transfer measure. Although the dif-ference score method discussed above has been usedby most studies examining self-image congruence,Sirgy et al. (1997) have argued for a more parsimoni-ous measure of congruence. These authors suggestthat a method which directly measures congruencyusing the respondent's own image dimensions andemploying a holistic evaluation is more appropriatefor examining image congruence between a brand

Winter 1999 63

and one's own image. We extended this advice byadapting the Sirgy et al. (1997) measure to studyevent-brand congruence. Specifically, we offered thefollowing instructions (adapted from Sirgy et al. 1997)and asked subjects to rate the consistency betweenthe event image and the brand image:

Take a moment to think about the (sporting eventname). Think about the various images and expe-riences one would encounter when they attendedor watched this event. Imagine this event in yourmind and then describe the event using severaladjectives such as: exciting, traditional, young,conservative, sexy, or whatever adjectives you thinkdescribe the image of this sporting event."

Subsequent to this mental imagery task, consistencywas scored on a seven point scale (l=Strongly Agree,7=Strongly Disagree) keyed to the following ques-tion: "My image of the (sporting event name) is con-sistent with my image of (brand name)."

ResultsHj^othesis One was analyzed using a one-way

MANOVA, between groups design. With sponsorshiptreatment (yes, no) as the between groups factor andthe three image congruence scores (based on summedabsolute differences) as the criterion, a significantmultivariate effect was found for sponsorship treat-ment (Wilk's lambda=.89, F [3,324]=12.33; p< .0001).The sample means are displayed in Table 3. Tuke/sHSD test reveals that subjects who were exposed tothe sponsorship arrangement had significantly lowerdifference scores (i.e., higher image congruence) thanthose subjects not exposed to the sponsorship tie forboth the functional similarity (Indianapolis 500 andGoodyear tires) and image similarity (U.S. Open GolfChampionship and Acura automobiles) event-brandcombinations. There was not a significant differencebetween subjects in the "yes sponsorship" level and the"no sponsorship" level with regard to the no similarityevent-brand combination (World Cup soccer and Camelcigarettes). Therefore, two ofthe three "absolute differ-ence" congruence measures support Hypothesis One.

Hypothesis One was also tested using the holisticimage congruence measures adopted from Sirgy et al.(1997). Again, using a one-way MANOVA, betweengroups design, we found a significant multivariateeffect (Wilk's lambda=.92, F [3,3531=10.48; p< .0001)for sponsorship treatment. As with the adjective basedcongruence measure, the Tukey HSD test showedsignificantly higher congruence in the "yes sponsor-ship" treatment level than in the "no sponsorship"treatment level for both the functional similarity and

image similarity event-brand combinations. Althougha significant difTerence does exist for the no similar-ity event-brand pairing, the means are the oppositeofthe hypothesized direction. That is, there is greatercongruence between World Cup Soccer and Camelcigarettes in the "no sponsorship" condition. Thus,Hypothesis One is supported by two of the three ho-listic congruence measures. This unexpected patternof results in the no similarity pairing is considered inmore detail in the discussion section.

Hypotheses 2a and 2b state that image transferwill be stronger in sponsor relationships when thereis either functional (H2a) or image (H2b) based simi-larity than when there is no similarity. Accordingly,these hypotheses were tested using only data fromthose subjects exposed to the sponsorship relation-ship (data in the top row of Table 3). Since eachsubject responded to all three event-brand sponsor-ships these hypotheses were analyzed using a one-way analysis of variance, repeated measures design.The test reveals a significant effect for sponsorshipsimilarity using the summed absolute difference scoresas the dependent congruence measures, F(2,328)=231.85; p

64

"Yes" Sponsorship^"No" Sponsorship^

Table 3Summary of Image Congruence Means and

ImageSimilarity

14.218.6

Adjective Measure^FunctionalSimilarity

20.322.5

WoSimilarity

34.331.9

The Journal of Advertising

Image Transfer Strength

ImageSimilarity

3.34.2

Holistic Measured

FunctionalSimilarity

2.73.3

NoSimilarity

6.45.9

^Sum ofthe absolute differences between the ten corresponding adjectives in each event-brand pairing, iower numbers indicate greater con-gruence/image transfer.

^Singie item measure of congruence based on respondent's own adjectives, iower numbers indicate greater congruence/image transfer.^Means are significantly different between sponsorship treatment ieveis (p

Winter 1999 55

awareness) are the overriding goal, recent researchhas begun to document the importance of image re-lated ohjectives to corporate sponsors (Ahratt, Claytonand Pitt 1987; Hoek, Gendall and Sanders 1993; Irwinand Sutton 1994). Indeed, in their framework forevaluating the attractiveness of sponsorship opportu-nities, Irwin and Asimakopoulos (1992) describe im-age association as one ofthe six primary sport spon-sorship objectives. The point is that if image transferis of concem, then event selection should be madewith degree of similarity in mind. To enhance thestrength of image transfer in cases of event sponsor-ship, it seems plausible that the marketing managermay wish to alter the communication regarding theproduct prior to the event, to be more congruent (oneither a functional or image basis) with the image ofthe event. For example, a flurry of advertising linkingthe attributes ofthe event with the advertised product.Our findings indicate that if the match between theevent and product can be made stronger, then the re-sulting image transfer will be more pronounced.

Limitations and Directions forFuture Research

The results of this study should be considered inlight of several constraints. First, due to concernsover our subjects' available time, we conducted thisexperiment using one event-brand pairing per simi-larity tjTJe and a single exposure to the sponsorshipstimuli. Other studies may wish to develop researchdesigns allowing for a more robust treatment of thesimilarity condition utilizing multiple sponsorshipties. Although some experimental control may be lost,a field experiment at an actual sporting event wouldprovide a context in which multiple brands coupledwith the dynamic environment of a realistic eventcould be studied. Further, a field study would be ableto incorporate other elements that are difficult tocreate in a "lab" setting (e.g., the influence of otherspectators and the "secondary" nature ofthe sponsor-ship association versus the "primary" concern of theevent itself). Second, our use of a student sampleshould be expanded to include non-student respon-dents. For example, future studies might explore howdifferent "types" of fans experience the image trans-fer process. Given the increasing amount of researchon sports-fan team identification (e.g.. Fisher andWakefield 1998; Mael and Ashforth 1992; Wann andBranscombe 1995; Wann and Dolan 1994), it wouldbe interesting to test for differences on sponsorshipissues (e.g., image transfer, sponsor recall, sponsorpatronage, etc.) between high identified and low iden-

tified sports fans. A third issue is our use of differentsets of instructions in the "yes" and "no" sponsorshipconditions, as well as the use of photos only in the"yes" sponsorship condition. Given our research de-sign, we are unable to assess the influence, if any,that these differences between conditions may havehad on the dependent variables above and beyondestablishing the event-brand linkage. Finally, we ex-amined image congruence afler the exposure to spon-sorship stimuli. Thus, we assumedrather thantestedthe direction of the image transfer. Accord-ingly, it is possible that the image transfer occursfrom the brand to the event rather than from theevent to the brand. Examining the direction of theimage flow and under what conditions it might beswitched represents a fruitful future research direc-tion. The use of known versus unknown brands mightbe instrumental in understanding these effects (Tripp,Jensen and Carlson 1994).

Another interesting issue surrounding image trans-fer and related to degree of similarity is that of con-flicting images among multiple sponsorships. It isnot uncommon for firms to engage in multiple spon-sorship relationships over the course of any givenyear (Farrelly, Quester and Burton 1997). It may beinsightful to consider the effect to a consumer's brandschema when a brand sponsors multiple events withconflicting images. Would the image transfer cancelout? Would the most recent image have the largerimpact? Here schema theory may prove useful forproviding insight. As discussed, discrepant informa-tion is less likely to lead to schema change because itis often discounted by the individual. However, a brandassociated with multiple events, each having a differ-ent and discrepant meaning, might lend credence tothe discrepant information, or at least call into ques-tion the existing schema. Alternatively, social adap-tation theory would suggest that discrepant eventassociations would, presumably, not be effectivesources of information and, therefore, would not fa-cilitate adaptation to environmental conditions (Kahleand Homer 1985). In contrast, it may be possible for afirm to positively leverage the image transfer by spon-soring multiple events with consistent images. Ofcourse, the sponsorship of two (or more) events withconflicting images only becomes an issue when a giventargeted consumer group is aware of both sponsor-ship ties. Still, within a single event it may be possibleto have conflicting similarity types (i.e., functional andimage based). For example, an interesting future re-search question could address how image transfer mi^tbe affected when an event-product pairing is high infunctional similarity, but low in image similarity (and

The Journeil ofAdvertisingvice versa). Additionally, will image transfer be en-hanced when both types of similarity are present (andconsistent) in the sponsorship arrangement?

A related issue is the impact of multiple sponsors ofa given event. Gwinner (1997) suggests that the trans-fer of an event's image to a sponsoring brand will bemoderated by the exclusiveness of the sponsorshiparrangement as measured by the number of othersponsors and the level of the sponsorship. He sug-gests transfer will be less in instances of multiplesponsors and lower sponsorship levels (e.g., title spon-sor versus perimeter fence signage). This is consis-tent with recent research in the celebrity endorserliterature which has found that the number of prod-ucts endorsed by a celebrity is negatively related toendorser credibility, likability and attitude towardthe ad (Tripp, Jensen and Carlson 1994). As such,studies might explore how image transfer and brandattitudes are impacted by the number of sponsorsand the level of sponsorship. Further, although ourresults are supportive of the image transfer hypoth-esis, they do not provide insight into the enduringnature of this phenomenon. Additional studies areneeded to assess the long term influence of eventsponsorship as it relates to image transfer.

ReferenceAaker, Jennifer L. (1997), Ilimensions of Brand Personality,"

Winter 1999 57

Plummer, Joseph T. (1986), "How Personality Makes a Differ-ence," c/oumoi of Advertising Research, 24 (6), 27-31.

Rumelhart, David E. and Andrew Ortony (1977), The Representationof Knowledge in Memory," in Schooling and the Acquisition ofKnowledge, Ridiard C. Anderson, Rand J. Spiro, and William E.Montague, eda, Hillsdale, NJ: Lawrence &U>aum Associates.

Sherif, Muzafer and Carl I. Hovland (1961), Social Judgement:Aaaimilation and Contrast Effects in Communication and At-titude Change, New Haven: Yale University Press.

Sirgy, M. Joseph, Dhniv Grewal, Tamara F. Mangleburg, Jae-okPark, Kye-Sung Chon, CB. Claihome, J.S. Johar and HaroldBerkman (1997), "Assessing the Predictive Validity of TwoMethods of Measuring Self-Image Congruence," Journal ofthe Academy of Marketing Science, 25 (3), 229-241.

Tripp, Carolyn, Thomas D. Jensen and Les Carlson (1994), "TheEffects of Multiple Product Endorsements by Celebrities on(Consumers' Attitudes and Intentions," Joumal of ConsumerResearch, 20 (4), 535-547.

Wann, Daniel L. and Thomas J. Dolan (1994), "Attributions ofHighly Identified Sports Spectators," The Joumal of SocialPsychology, 134 (6), 783-792.

and Nyla R. Branscombe (1995), "Influence of Iden-tification with a Sports Team On Objective Knowledge andSubjective Beliefs," International Joumal of Sports Psychol-ogy, 26 (Oct./Dec.), 551-567.