6
INCUBATOR: Editor: Bob McKercher Triangulation — A Methodological Discussion Martin Oppermann² Griffith University, PMB 50, Gold Coast, Qld, Australia ABSTRACT Triangulation is starting to emerge in tourism research as the ‘new’ method. However, there are many misconceptions of what triangulation is and what it can do. This paper discusses triangulation as a research method in the social sciences. It reflects on the origins of the term and what it currently denotes and what it should connote. It also covers the critiques of triangulation and eventually suggests how it should be used. Copyright # 2000 John Wiley & Sons, Ltd. Keywords: research method; triangulation; multi-method approach. INTRODUCTION W ith the recent (re-)advance of quali- tative methods in research, the tri- angulation method is used by some researchers as a crossing bridge between the pre-eminent quantitative studies and the growing number of qualitative studies (e.g., Bahr et al., 1983; Greene and McClintock, 1985; McClintock and Greene, 1985; Nickel et al., 1995; Stage and Russell, 1985). Also, most textbooks on research methods in the social sciences devote some paragraphs or pages to triangulation (e.g. Bryman, 1988; Hakim, 1987; Layder, 1993; Singleton et al., 1993; ). More recently, tourism researchers also seem to be embracing triangulation (e.g. Seaton, 1997). Unfortunately, the basic concepts that underlie the triangulation method are easily misunder- stood. Often it is interpreted as simply an approach that combines both quantitative and qualitative data collection on the same general subject of interest. Yet without any cross- linkages between and systematic planning of these collection methods, such a methodologi- cal approach is simply a multi-methodological approach with limited cross-validation of the results. This paper will clarify the inherent principles and laws of the triangulation meth- od. In addition, suggestions are made as to how the triangulation method could be ap- plied fruitfully in tourism research. TRIANGULATION In dictionaries, triangulation is usually defined as: (1) division (of an area) into triangles for surveying purposes; (2) measurement and mapping (of an area) by INTERNATIONAL JOURNAL OF TOURISM RESEARCH Int. J. Tourism Res. 2, 141–146 (2000) Copyright # 2000 John Wiley & Sons, Ltd. ² This was the last paper written by Martin Oppermann before his untimely passing late in 1998. He will be sadly missed (Editor).

2000_ Triangulation - A Methodological

Embed Size (px)

Citation preview

Page 1: 2000_ Triangulation - A Methodological

INCUBATOR:Editor: Bob McKercher

Triangulation Ð A MethodologicalDiscussionMartin Oppermann²

Grif®th University, PMB 50, Gold Coast, Qld, Australia

ABSTRACT

Triangulation is starting to emerge in tourismresearch as the `new' method. However, thereare many misconceptions of whattriangulation is and what it can do. Thispaper discusses triangulation as a researchmethod in the social sciences. It re¯ects onthe origins of the term and what it currentlydenotes and what it should connote. It alsocovers the critiques of triangulation andeventually suggests how it should be used.Copyright # 2000 John Wiley & Sons, Ltd.

Keywords: research method; triangulation;multi-method approach.

INTRODUCTION

With the recent (re-)advance of quali-tative methods in research, the tri-angulation method is used by some

researchers as a crossing bridge between thepre-eminent quantitative studies and thegrowing number of qualitative studies (e.g.,Bahr et al., 1983; Greene and McClintock, 1985;McClintock and Greene, 1985; Nickel et al.,

1995; Stage and Russell, 1985). Also, mosttextbooks on research methods in the socialsciences devote some paragraphs or pages totriangulation (e.g. Bryman, 1988; Hakim, 1987;Layder, 1993; Singleton et al., 1993; ). Morerecently, tourism researchers also seem to beembracing triangulation (e.g. Seaton, 1997).Unfortunately, the basic concepts that underliethe triangulation method are easily misunder-stood. Often it is interpreted as simply anapproach that combines both quantitative andqualitative data collection on the same generalsubject of interest. Yet without any cross-linkages between and systematic planning ofthese collection methods, such a methodologi-cal approach is simply a multi-methodologicalapproach with limited cross-validation of theresults. This paper will clarify the inherentprinciples and laws of the triangulation meth-od. In addition, suggestions are made as tohow the triangulation method could be ap-plied fruitfully in tourism research.

TRIANGULATION

In dictionaries, triangulation is usually de®nedas:

(1) division (of an area) into triangles forsurveying purposes;

(2) measurement and mapping (of an area) by

INTERNATIONAL JOURNAL OF TOURISM RESEARCHInt. J. Tourism Res. 2, 141±146 (2000)

Copyright # 2000 John Wiley & Sons, Ltd.

² This was the last paper written by Martin Oppermannbefore his untimely passing late in 1998. He will be sadlymissed (Editor).

Page 2: 2000_ Triangulation - A Methodological

the use of triangles with a known baselength and base angles (Thompson, 1995).

Essentially what these meanings refer to istriangulation's original origin as coming fromthe area of surveying, where use was made ofthe fact that if one knows three relatedmeasures in a triangle (e.g. two angles and adistance) one can calculate the other distancesand angles. In the nineteenth century inparticular, precise measurements of large partsof Europe were made using ®rst large triangleswhich were then systematically subdivided toarrive at a very dense grid of survey points.In today's modern technology, triangulation

is also used extensively, usually without theactual user realising it. Handheld GlobalPositioning Systems (GPS) automatically cal-culate the current position of the GPS usingexisting knowledge of the locations of satellitesat any given time. Expensive rental cars inEurope have senders integrated that allowtracing and position establishment of thesecars in the event they are stolen. Modern car-based street map computers also include a GPSsystem that allows the calculation of the exactposition of the car within a metre and conse-quently allows the user to see, even on adetailed city street map, where he or she iscurrently located. Advanced computer sys-tems will even calculate and show the nearestpath to the destination location. Anotherexample where GPS ®nds application is inback-country hiking, where some hikers arealready equipped with these, allowing them tokeep better track of their position.The crucial component of triangulation,

besides the requirement that each measure iscorrect, is that those measures need to beinterrelated, meaning they all have to relate tothe same triangle in question. Although thisseems obvious, its implications for the transferof the term triangulation to other areas outsidesurveying are important. It also should benoted that the measurement and the wholeconcept are based within one paradigm and donot rely on different theories, methods, etc. Oras Blaikie (1991, p. 118) phrased it, `allmeasurement is of the same kind and is basedon common ontology and epistemology'.In principle, surveyors distinguish between

intersection and resection triangulation. In the

former, one is interested in a point in thedistance that might be inaccessible or whichone wants to know. This would be the case inthe above stolen rental car example. Throughmeasuring the signal of such a car from twodifferent locations (angle between signal andthe respective other station), one can deter-mine the third angle and the distances fromeach location and, therefore, the exact locationof the stolen rental car. Resection triangulationrefers to identifying the point one currently isat by measuring angles between at least threeother points whose exact locations are known.For example, a handheld GPS does that.If one transfers these to research, intersec-

tion could be seen as analysing something`unknown', for example, forecasting the num-ber of tourist arrivals in the year 2010. To thisend, several research methods and/or datacould be used to arrive at a best estimate.Resection would mean that we currently knowwhere we are but would like to know how wearrived there.

TRIANGULATION AS A RESEARCHAPPROACH

Webb et al. (1966) were among the ®rst tointroduce the term triangulation into the socialscience discipline as a research approach.`Social scientists have borrowed the termtriangulation ¼ to help describe how the useof multiple approaches to a research questioncan enable the researcher to ªzero inº on theanswers or information sought' (Singleton etal., 1993, p. 391). Denzin (1978) distinguishedbetween a number of different `triangulation'approaches, namely methodological triangula-tion, data triangulation, investigator triangula-tion and multiple triangulation.Methodological triangulation refers to using

more than one research method in measuringthe same object of interest, for example, usingparticipant observation as well as question-naires. Data triangulation refers to using thesame approach for different sets of data inorder to verify or falsify generalisable trendsdetected in one data set. For example, ifpleasure travel propensity is closely relatedto age in one country, one could analyse othercountries' data in order to check if similartrends can be observed elsewhere and if they

Copyright # 2000 John Wiley & Sons, Ltd. Int. J. Tourism Res. 2, 141±146 (2000)

142 M. Oppermann

Page 3: 2000_ Triangulation - A Methodological

are consistent. Investigator triangulation refersto making use of different investigators with adifferent background. For example, femaleinterviewers might obtain a very different setof responses from male tourists than maleinterviewers. Multiple triangulation would in-clude using at least two of the other triangula-tion methods in combination.

WHY TRIANGULATE?

The primary reason of triangulation is therecognition that data-set or investigator surveybias can be introduced by using only oneresearch method. As Blaikie (1991, p. 115)observed, `the common theme in discussionsof triangulation has been the desire to over-come problems of bias and validity. It has beenargued that the de®ciencies of any one methodcan be overcome by combining methods andthus capitalizing on their individual strengths'.A number of potential biases can be identi-

®ed, namely methodological/instrument bias,data bias, and investigator bias. By using onlyone method, for example, interviewing withclosed questions, the data are limited toresponses to the given questions and especiallythe categories provided. Other, possibly moreimportant, categories not included will not bedetected and, therefore, the results will bebiased towards the preconceived categoriesprovided. In addition and depending on theissue at hand, many respondents might not beable to recall correctly what they did or howmuch they spent or, in other cases, might notwant to disclose information that they feel issensitive or politically incorrect. Anothermethod such as observation might provide amore accurate picture.Data bias is introduced when the survey

population has unusual characteristics that arenot representative of the wider population. Forexample, if one bases a general theory of travelbehaviour on the one by, for example, SouthKorean travellers, then one fails to recognizethat this theory may not be applicable to otherpopulations in the world. Thus, it is alwaysimportant to cross-validate results derivedwith one population set with others. Theinvestigator bias was already alluded to above;different investigators might arrive at differentresults simply for reasons of their gender

(especially in face-to-face interviewing), back-ground (oriental versus occidental, socio-eco-nomic status), training (geographer versuspsychologists versus architect), etc. This needsto be recognised. Thus, the use of differentinvestigators (e.g. multi-disciplinary team)will help identify and recognise such a bias.

THE IMPORTANCE OF THE THIRDMEASURE

An early reference was made to triangulationas something different or outside the strictsurveying area by Holmes (1860; after OxfordDictionary, Simpson and Warner, 1989): `Asagacious person who has triangulated a race,that is taken three or more observations fromthe several standing-places of three differentgenerations'. Quite interesting in this earlyreference is that:

(1) there are more than two observations;(2) the observations are not made from

`similar' viewpoints.

This particular example of triangulation isactually not covered in the four differenttriangulation methods suggested by Denzin(1978; ) because it uses the same method(observation), the same data (the race), thesame investigator and is also no combinationof any of these two. In fact, it appears closer tothe surveyor meaning of triangulation as itrequires the repeated observation of the samepoint from different angles, not unlike inter-section.To what extent the original reference should

be constructed as including a time dimensionis questionable. The inclusion of `three differ-ent generations' could simply be interpreted asmaking reference to the spatial separation ofgenerations and thus asking the triangulator toreally choose different spatial areas rather thanjust three different observation points fromwithin one area, which could introduce anobservation bias (e.g. repeated observation ofthe same thing not realising that a bias isoccurring in all observations). However, it alsocould be interpreted that by standing in areaswith people of differing backgrounds (differ-ent generations), the observer will undoubt-edly be in¯uenced by the discussionsoverheard and might thus make a different

Copyright # 2000 John Wiley & Sons, Ltd. Int. J. Tourism Res. 2, 141±146 (2000)

Triangulation 143

Page 4: 2000_ Triangulation - A Methodological

judgement of the race not because of the angleof observation but because of the physicalsurroundings of the observation point. In asense, it will make a difference if one observeslocal people in a slum from the comfort zone ofan air-conditioned bus or if one observes themby wandering through. Thus, by changing thephysical and social surroundings of the ob-servation point, it can be interpreted that someform of investigator triangulation is beingintroduced.Yet, arguably more important is the use of

`at least three' measures, whereby the inclu-sion of `at least' indicates that the three is notderived from the `tri' of triangulation. Goingback to the surveying origins, measuring atleast three reference points is standard proce-dure. Whereas two reference points, as anintersection, allow exact calculation of thepoint of interest (through bisecting the twolines), it does not allow for an accuracy test andif, for example, an instrument or investigatorerror occurs, a wrong point would be calcu-lated. Thus, by using a thirdmeasurement, anyerrors of the ®rst two can be detected and,more in general, the accuracy of the ®rst twomeasurements be calculated. Hence, the moremeasurements that are being taken the moreaccurate will be the ®nal calculation.If one transfers this notion to triangulation

as a social science method, the more differentthe instruments/methodologies, data sets andinvestigators used in analysing a speci®cproblem, the greater will be the con®dence inthe ®nal result, provided all (sub-) results aremore or less congruent. Hence, a true triangu-lation should not consist of solely two differentresearch methods, data sets, etc. but at least ofthree.

CRITIQUE OF TRIANGULATION INSOCIAL RESEARCH

As early as 1983, critiques of the triangulationmethod, at least as it was proposed by Denzin(1978) and his followers, already called forabandonment of the concept, or at least itsterminology (e.g., Miller, 1983; Fielding andFielding, 1986; Blaikie, 1991; Flick, 1992;McFee, 1992). Miller (1983, p. 77), in discussingthe future of triangulation, suggested: `Theprincipal use of any metaphor is to illuminate

meaning; when it fails to do this, it should beabandoned. In qualitative research, the con-cept of triangulation should be allowed to diepeacefully'. More recently, Blaikie (1991, p.131) was possibly even harsher when hesuggested `a need:

(1) for a moratorium on the use of the conceptof triangulation in social research;

(2) to identify appropriate and inappropriatecombinations of methods and data sources,in light of the incommensurability ofontological and epistemological assump-tions of methodological perspectives; and,

(3) to develop suitable new labels for theseappropriate combinations.'

Fielding and Fielding (1986, p. 33) suggestedthat `theoretical triangulation does not neces-sarily reduce bias, nor does methodologicaltriangulation necessarily increase validity.Theories are generally the product of quitedifferent traditions so when they are com-bined, one might get a fuller picture, but not amore ªobjectiveº one'. `We should combinetheories and methods carefully and purpose-fully with the intention of adding breadth anddepth to our analysis but not for the purpose ofpursuing ªobjectiveº truth' (Fielding andFielding, 1986, p. 33).Thus, the major thrust of the critique at

triangulation in social sciences has been at itsuse in the form of a multi-method and/ormulti-theory approach and not at its use asdata and/or investigator triangulation. It isimportant here to recognise this difference.This author agrees with the critiques for-warded on the use of the term triangulationas a multi-method approach.Multi-method approach should be referred

to as multi-method approach as it is virtuallyimpossible to obtain the `truth', but thestrength is in the addition and breadth ofinsight into a particular issue. The termtriangulation, if people insist on using it,should be used only for data triangulationand possibly investigator triangulation and itis in data triangulation that this author sees itsbest application and its truest resemblance tothe origins of the term triangulation, becauseall measurements would be of the same kindand based on the same ontology and episte-mology (Blaikie, 1991). Much too seldom are

Copyright # 2000 John Wiley & Sons, Ltd. Int. J. Tourism Res. 2, 141±146 (2000)

144 M. Oppermann

Page 5: 2000_ Triangulation - A Methodological

studies replicated with the same methodologyon different data sets in order to ®nd simila-rities and or differences.

CONCLUSION

This elaboration has shown that one should beaware that the term triangulation in socialresearch is somewhat questionable and mayfalsely relay a sense of scienti®cness andexactness. Triangulation is not simply thecrossing bridge between quantitative andqualitative research methods but it should beseen as a principle behind any research thatattempts to constitute scienti®c advance. Tri-angulation is more about veri®cation of resultsand, in the process, identifying and eliminat-ing methodological shortcomings, data orinvestigator bias.A multi-method approach allows research-

ers to be more con®dent about their results. Italso may help in uncovering a deviant or off-quadrant dimension of a phenomenon (Jick,1983). Divergent results can lead to newinsights and consequently enriched explana-tions of a research issue. The multi-dimen-sional perspective will provide new insightsbehind the respective walls of individualmethodological or data approaches. For exam-ple, Oppermann (1995) showed that by lookingat three time-dimensions simultaneously(period, age and cohort) rather than the typi-cal single one, a whole new picture emerges.Whereas analysis along each individual timedimension also provided statistically signi®-cant results, it did not reveal the actualunderlying trends.However, a note of caution is required.

Multi-method approaches, especially where itinvolves multiple qualitative methods, may bedif®cult to replicate. `Replicating a mixed-methods package, including idiosyncratictechniques, is a nearly impossible task andnot likely to become a popular exercise' (Jick,1983, p. 146).Finally, it is time for tourism researchers to

take on `new' challenges, namely systemati-cally trying to add knowledge to the ®eldrather than continue engaging in producingmore and more case studies of limited addi-tional scienti®c value. Why are researchers soafraid of testing other peoples' work in

different settings, of using the same methodswith other data, and of strategically combiningseveral methods to reach their goal. Is the sheerpressure of `publish or perish' too much? Doesit force researchers into doing the `quick anddirty' in order to add quantitatively to theirCV? Tourism is strategically placed at theinterface of so many disciplines that inherentlytourism is an interdisciplinary ®eld. Thisshould stimulate interdisciplinary approachesusing multiple methods as well as usingdifferent data sets and investigators in thequest for `truth'.

REFERENCES

Bahr HM, Caplow T, Chadwick BA. 1983. Middle-town III: problems of replication, longitudinalmeasurement, and triangulation.Annual Review ofSociology 9: 243±264.

Blaikie NWH. 1991. A critique of the use oftriangulation in social research. Quality andQuantity 25: 115±136.

Bryman A. 1988. Quantity and Quality in SocialResearch. Unwin: London.

Denzin N. 1978. The Research Act. Aldine: Chicago.Fielding NG, Fielding JL. 1986. Linking Data. Sage:Beverley Hills.

Flick U. 1992. Triangulation revisited: strategy ofvalidation or alternative? Journal for the Theory ofSocial Behaviour 22(2): 175±197.

Greene J, McClintock C. 1985. Triangulation inevaluation: design and analysis issues. EvaluationReview 9: 523±545.

Hakim C. 1987. Research Design. Allen & Unwin:London.

Jick TD. 1983. Mixing qualitative and quantitativemethods: triangulation in action. In QualitativeMethodology, Van Maanen J (ed.) Sage: NewburyPark; 135±148.

Layder D. 1993. New Strategies in Social Research.Polity Press: Cambridge.

McClintock C, Greene J. 1985. Triangulation inpractice. Evaluation and Program Planning 8: 351±357.

McFee G. 1992. Triangulation in research: twoconfusions. Educational Research 34(3): 215±219.

Miller SI. 1983. Mapping, metaphors and meaning:a note on the case of triangulation in research.Sociologia Internationalis 219: 69±79.

Nickel B, Berger M, Schmidt P, Plies K. 1995.Qualitative sampling in a multi-method survey.Quality and Quantity 29: 233±240.

Oppermann M. 1995. Multitemporal perspective of

Copyright # 2000 John Wiley & Sons, Ltd. Int. J. Tourism Res. 2, 141±146 (2000)

Triangulation 145

Page 6: 2000_ Triangulation - A Methodological

changing travel patterns. Journal of Travel andTourism Marketing 4(3): 101±109.

Seaton AV. 1997. Unobtrusive observational mea-sures as a quality extension of visitor surveys atfestivals and events: mass observation revisited.Journal of Travel Research 35(4): 25±30.

Simpson JA, Waner ESC (eds). 1989. The OxfordEnglish Dictionary, 2nd edn. Clarendon Press:Oxford.

Singleton RA, Straits BC, Straits MM. 1993. Ap-

proaches to Social Research. Oxford UniversityPress: New York.

Stage FK, Russell RV. 1992. Using method triangu-lation in college student research. Journal of CollegeStudent Development 22: 485±491.

Thompson D, (ed.). 1995. The Concise OxfordDictionary, 9th edn. Clarendon Press: Oxford.

Webb EJ, Campbell DT, Schwartz RD, Sechcrest L.1966.Unobtrusive Measures: Nonreactive Research inthe Social Sciences. Rand McNally: Chicago.

Copyright # 2000 John Wiley & Sons, Ltd. Int. J. Tourism Res. 2, 141±146 (2000)

146 M. Oppermann