The Debate about Quantitative and Qualitative Research: A Question of Method or Epistemology?

Author(s): Alan Bryman Source: The British Journal of Sociology, Vol. 35, No. 1 (Mar., 1984), pp. 75-92 Published by: Blackwell Publishing on behalf of The London School of Economics and Political Science Stable URL: http://www.jstor.org/stable/590553 Accessed: 27/10/2010 10:41
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=black. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Blackwell Publishing and The London School of Economics and Political Science are collaborating with JSTOR to digitize, preserve and extend access to The British Journal of Sociology.

http://www.jstor.org

Alan Bryman

and The debate about quantitative a qualitativeresearch: questionof method or epistemology?

AB ST RACT

characteristics of The maindimensions the debateaboutthe relative and merits of quantitativeand qualitativemethodology are outlined, emphasizingthe philosophicalissues which underpinmuch of the discussion.A distinctionis drawnbetween epistemological Threeareasare and technicalissues in relationto the controversy. then selected which demonstratea tendency for the debate to oscillate between epistemologicaland technical modes of expression. The question is raised as to whether it is possible to establish a clear symmetry between epistemologicalpositions (e.g. phenomenology, positivism) and associated techniques of social research (e.g. participantobservation,social survey). The conclusion is sceptical about the extent to which a neat correspondencecan currentlybe established. Over the past fifteen years, the debate over the relative virtuesof quantitative and qualitativemethodology has gained considerable impetus. While the exact constitution of the two methodologies varies somewhat from author to author or is defined with varying degrees of specificity, there is substantial agreement about the fundamental antinomies and their practical implications for the conduct of research.One of the difficulties,however,in representing the divergenciesbetween the two methodologies, derives from a tendency for philosophicalissues and technicalissues to be treated issues and occasionallyto be confused.Philosophical simultaneously foundation relate to questionsof epistemology,i.e. the appropriate By for the study of society andits manifestations. contrast,technical or of issuesbespeakthe consideration the superiority appropriateness of methods of researchin relation to one another. Much of the recent methodologicalliteraturesees the latter as derivingfrom the
TheBritishJournalof Sociology VolumeXXXV NumberI 75

76

Alan Bryman

former,i.e. the choice of a particular epistemological baseleadsto a preference for a particularmethod on the grounds of its greater appropriateness given the precedingphilosophicaldeliberations. As we shall see, the two forms of argumentoccasionallybecome confused with each other, and this is particularly when writershave so soughtto articulate relationships the betweenthe two methodologies. In addressing these issues the term 'methodolotgy' whetherdescribed as quantitative qualitative, referto an epistemological or will position; 'method'and 'technique'will be used synonymouslyto referto ways of gatheringdata. As such, 'method' and 'methodology indicate ' different levels of analysis, and since the two terms are often used interchangeablylit is of considerable importanceto distinguishthe relevantrealmsof discourse. In the 1950s and 1960s it was not uncommonto find discussions of the relative advantagesand disadvantages social surveysand of participationobservation.2A solution to many of the discussions seemed to lie in Trow's apparentlysound advice that 'the problem under investigation properlydictates the methodsof investigation'.3 This is a highly seductive solution in that it would appear that whoever argues against it is likely to be implying the absolute superiority of one particulartechnique, a position that requiresa good deal of confidence in one's choice. In more recent years, the nature of the debate seems to have changedsomewhatin that discussions centre upon comparisonsof quantitativeand qualitative methods or methodologies. A prominent feature has been the emergenceof textbooks dealingalmost exclusivelywith qualitative methods,4 along with journals which focus exclusively on data drawn from these methods (notably QualitativeSociology, Urban Life and Anthropologyand EducationQuarterly),as well as special issues of journals with a more general readershipdevoted to discussions of qualitativemethods.S In large part, these expositions comprise a contrast between the two forms of methodology by writers who either are themselvesusersof qualitativemethodology or are rather sympatheticto it. Quantitativemethodologistsseem rarelyto write about the natureof theirresearch activityin contrast to plausiblealternatives. Much of our currentunderstanding about the fundamentals quantitative of methodologyand its epistemological distinctness from qualitative methodology derives from writers within the latter tradition.This is not to suggestthat the depiction of quantitativemethodology is necessarilyinaccurate; indeed many of the expositionsreveala very clear understanding (albeita lack of of sympathy with) its essential characteristics.The distinction between quantitativeand qualitativemethodology has been elaborated not only in sociology, but also in related fields such as evaluation research,6 educational research7 and organiz ational
analysis.8

research and The debateabout quantitative qualitative

77

In some eases writers have ehosen not to use the quantitative/ qualitativedistinetionand have insteadused terms whieh have been used as synonyms. The terms 'positivist' and 'empirieist' often denote the same fundamental approaeh as 'quantitative',while 'naturalistie' field researeh, 'ethnographie', 'interpretivist', and 'eonstruetivist'are sometimesused instead of 'qualitative'.9Whatin apparent the literature what is inereasingly ever the nomenelature, dealing with researehmethodology is a tendeney to talk about the distinetivenessof (and oeeasionallyeompatibilitybetween) quantiteehniques. tative and qualitativemethodologiesas againstpartieular Whethersurveysare 'better' than partieipantobservationseems to have beeome a question that is rarely addressed.Whetherthis is a signifieantshift in emphasisand some of its eonsequeneesare the foei of this paper.
QUANTITATIVE AND QUALlTATIVE METHODOLOGY AS RESEARCH T RAD ITION S

Quantitativemethodology is routinely depieted as an approaehto the eonduet of soeial researehwhieh appliesa naturalseienee, and in partieulara positivist, approaehto soeial phenomena.The paratypieallyin the methodologiare of phernalia positivism eharaeterized eal literature as exhibiting a preoeeupationwith operationaldefinitions, objeetivity,replieability,eausality,and the like.10The soeial survey is typieally seen as the preferredinstrument of researeh within this tradition beeause it ean apparentlybe readily adapted to sueh eoneerns. Through questionnaireitems eoneepts ean be operationalized;objeetivity is maintainedby the distanee between observerand observedalong with the possibilityof externaleheeks upon one's questionnaire;replication ean be earried out by employinstg same researehinstrumentin another eontext; and the the problem of causality has been eased by the emergence of path techniquesto which surveysare well analysis and relatedregression as of suited. Research this kind is frequentlydeseribed beingpositisist or empiricist.In attributingto it labels of this kind an essentially epistemologiealpoint is being made, namely that researehof this genre is underpinnedby a distinctive theory of what should pass knowledge.Surveysare seen as instrumentsfor the as warrantable elucidation of researehwhich makes sueh epistemologicalassumptions, though experimentaldesigns and seeondaryanalyses of precollected data are also often recognized as exhibiting the same premises. philosophical underlying QualitativemethodoloXJdiffers in a number of ways. The sine qua non is a commitmentto seeingthe social world fromthe point of view of the actor,a themewhichis rarelyomittedfrommethodological

78

Alan Bryman

writingswithin this tradition.Clearstatementsof this emphasiscan be discernedin a broadrangeof writings.ll Becauseof the commitment to see throughthe eyes of one's subjectsclose involvementis advocated. There is a simultaneousexpressionof preferencefor a contextual understanding that behaviouris to be understoodin so the context of meaningsystems employed by a particular group or society.l2 Qualitative researchis deemedto be much more fluidand flexible than quantitativeresearchin that it emphasizesdiscovering novel or unanticipated findingsandthe possibilityof alteringresearch plans in response to such serendipitousoccurrences.l3This is contrastedsharplywith the quantitative methodologist's research design with its emphasisupon fixed measurements, hypothesis(or hunch) testing, and a much less protractedform of fieldworkinvolvement. The philosophical underpinningsof qtlalitativemethodology are typically attributedto phenomenology, Verstehen5 and symbolic 14 1 interactionism.l6Many of these writersview the phenomenolosgical theme as the most fundamentalone, symbolic interactionismand Verstehen being prominent examples of its basic premises. The contrastwith what is variouslycalledpositivismand a naturalscience approachis ever present among these writers.The point about the phenomenologicalposition is that it takes the actor's perspective as the empiricalpoint of departure. Positivistapproaches taken are to exhibit a tendency for the researcherto view events from the outside and from the point of view of a clusterof empirical concerns which are imposed upon social reality with little referenceto the meaning of the observationsto the subject of investigation.While the possibilitiesof phenomenologically-based researchtraditionare occasionally questioned,l influential writers like Schutz clearly 7 left open the possibility of such a perspectiveby claimingthat it may be necessary to 'abandon the strictly phenomenological method'l8 in order to carry out the study of the social world. His contrast between a naturalscience approachwhich sees people as inert and a phenomenological approachwhich seeks to focus upon the lived experienceof peoplel9 providesa key-noteof this tradition. In order to proceed with researchinto the social world which is informedby epistemolotpcal principles this kind, research of methods are necessary which facilitate an inside view. Unstructuredinterviewingand life histories(the latterto a lesserextent) are frequently mentionedas providing appropriate vehicles,but aboveall participant observationis the most favouredtechnique.'Participant observation' is a ratherbroad term, in that not only does it encapsulatea wide rangeof observational practices,it is also used to denote a fieldwork strategy which includes generalinterviewing,usually of a relatively unstructuredkind, the perusalof documents, and the interviewing of key informants.But it is the ability of the participantobserver to get close to his subjectsand so see the worldfrom theirperspective

Thedebateabout quantitative qualitative and research

79

that is its chief attraction.In so doing qualitative researchers produce data which they often call 'rich'20by which is meant data with a great deal of depth. Survey data are typically seen as deficient in this respect for they provide superficial evidence on the social world, winkling out the causal relationshipsbetween arbitrarily chosen variableswhich have little or no meaningto those individuals whose social worlds they aremeantto represent. Blumer's2lcritique of 'variableanalysis'still stands as one of the most incisiveattacks on such research, and is widely accepted within the qualitative tradition. What is clear from the variousdiscussionsabout these two methodologies is that they are being explicated at an epistemological level and an attempt is then made to establisha link betweenit and a technicallevel, i.e. the practiceof socialresearch. The epistemological nature of the discussionis occasionally reinforcedby recourseto the term 'paradigm'-usually in a Kuhniansense-to denote the two traditions.22 so far as paradigms meant to be incommenIn are surable, then it is even clearerthat two divergentepistemological bases are being expounded.In the context of this kind of discussion the question of techniquesof investigation no longerwhetherA is is 'better' than B, but is A the appropriatetechnique in terms of a particular of epistemological set premises Proponents qualitative X? of methodology justify their preference for participant observation by reference to its ability to meet a prior set of epistemological requirements, which have been summarised brieflyabove.The social surveyis seen as relevantto a differentintellectualtradition,i.e. one informed by the preoccupationsof a natural science approach. As Johnson has argued,the revivalof interest in participantobservation and field research 'is related to the abstract intellectual debates in a very fundamentalway'.23 This implies that the surge of interest in phenomenologicalideas, along with a resurgenceof interest in symbolic interactionism, to an increasein participant led observationand associatedresearchtechniques.It may also be the case that for some social scientists,a disillusionment with the spread of quantificationin researchled to a flirtationwith methods which had often been seen as impressionistic, unscientific,and the spread or of phenomenologicalwriting provided a ready-madejustification for theirresearch. The apparentlinking of more abstractphilosophicalissues with questions of researchpractice appearsa more sophisticatedway of treatingthe comparability differentmethodsof investigation of than a direct juxtaposition in terms of relative superiority.It is also apparent that the notion of the 'appropriateness' a particular of method is different.In Trow'swidely quoted observation (see above) it is the problemthat determinesthe techniqueto be employed. It is not preciselyclear what this means,but the notion of a problem

80

Bryman Alan

which deliberations not does seem to includethe more philosophical years. In this latter milieu it is not been in operationin recent have technique the much a problemthat determines use of a particular so position. The to a philosophical a but prior intellectualcommitment is problem then presumablyformulatedwithin the context of these This suggestion also makes some sense in terms of commitments. most of whom of individualbiographies many socialresearchers, the techniqueor tradition. research particular seem to be weddedto a do hiatus which opens up traversethe epistemological researchers Few traditions. the between research One peculiarityof the variouswritingswhich have spawnedthese and is debates the fact that it is the terms'quantitative' 'qualitative' or referencepoints for the intellectual whichare used as symbols of Yet undercurrents. the question of the presence or absence of superficialmanifestation the underlying data quantitative is but a issues. Indeed, neither directly signifiesthe clusters epistemological are of commitmentsfor which they are presVmedto stand. There who would deny the validity of at least fewhard-nosedpositivists employ while many participantobservers somequalitativematerial; evidence in their research,albeit of a a modicum of quantitative use kind, or alternatively a varietyof 'quasi-statistical' rudimentary of Whilesuch considerations degreeof terms,as Gans24calls them. not be forgotten, it is slightly puzzling that should quantification the it is this particulardimension of the debate that is taken as focus. terminological
TECHNIQUEAND EPISTEMOLOGY

distincIt has been observedabove that the quantitative/qualitative tion has become one whichin largepart derivesfrom epistemological be issues and that questions of researchtechnique are taken to to be a quite systematicallyrelated to these issues. This seems different form of argument from that which takes place when writers distinguishbetween methods or techniques.Trow's dictum to a that problems determine methods is essentially a reference only suggestsnot technical rather than an epistemologicalissue. It that one techniquecan never be inherentlysuperiorto its supposed in alternatives,but also that a techniqueis likely to be more useful to have sought some contexts than others. Others, like Zelditch2s by systematizesuch considerations delineatingthe linkagesbetween nexus objects and techniques.An example of the object-technique that: and Lininger by Warwick can be discernedin the suggestion and The samplesurveyis an appropriate useful meansof gathering when the goalsof the research informationunderthreeconditions;

and rssearch Thedevbato about quant-itative qualitative

81

call for quantitativedata, when the informationsought is reasonably specific and familiar to the respondents, and when the prior knowledgeof particular researcher himself has considerable problems and the range of responses likely to emerge. All of these conditions are met in the areasof researchthat have been voting, the traditionalstrongholdsof the survey-- publicopinionb attitudesand beliefs,and economicbehaviour. when the Participantobservationis usually more appropriate study requires an examination of complex social relationships or intricate patterns of interaction; ... when the investigator desires first-hand behavioural information on certain social processes,such as leadershipand influence in a smallgroup;when a majorgoal of the study is to constructa qualitativecontextual picture of a certain situation or flow of events; and when it is necessary to infer latent value patterns or belief systems from such behaviour as ceremonialpostures, gestures, dances, facial expressions subtleintlectionsof the voice.26 or Such arguments 'technical'in that they simplyseek to demarcate are methods of those substantiveissues or domainsin which particular Thereis a myriadof or investigationare appropriate inappropriate. technical reasonswhy participantobservationis preferableto social surveysin such a sense or vice versa.The final lines of Gans'classic tell studyor the Levittowners the readerthat 'Themail questionnaires and interviewsprovidedmore systematicallycollected data and are thus more scientific in one sense, althoughless so in another, for they can only reportwhat people say they do and feel, andnot erhat a researcher seen them say, do and feel'.27 In other words, the has gap between word and deed maE give participantobservationa the technical edge over a survey, particularlywrhen possibilityof a disjuncture may be problematic. In another classic participant observer study, Whyte28 notes that a questionnaireto delineate is the distributionof the attitudesof racketeers not a feasibleunderof taking.Considerations these kinds are boundup with researchers' judgments about technical viability and are quite distinct from philosophicaldebateswllich arguefor the superiorityof a particular epistemological bedrock from which considerations of method then emerge. in The more recent mode of discussingmethods of investigation knowledgebases occasionallyloses sight of its terms of appropriate level or mode of position by vacillatingbetween an epistemological discussionon the one hand and a technicalone on the other. This reveals itself in three main areas each of which forms the subject of the subsequentsections. (i) Techniqueand Sensitivity One of the argumentsthat is often

82 proferredin supportof qualitative methodologyis techniquesaremoresensitive to the complexitiesof that its associated than quantitative socialphenomena methods which enigmatic quality. The quest tend to ride roughshodover their indicators(and abstractcausal for directly observablequantitative relationships imposedupon an amongthem) which are unsuspectingsocial reality underlying neither capturesthe phenomena in their full understanding their contextual complexity nor facilitates an of significance. involvement, Prolongedand close however, provides empirical cerns. leverageupon such conThis form of reasoning revealsitself in two drawn from an comparisons, both ineach of the educationalcontext, between research traditionsarticulatedin this conducted end an articlehighly of paper. Light,29at the supportiveof the qualitative the Colemanreport30which found that the schools agenda,attacks are poor predictorsof children achievement.The researchwas a attend standard quantitativeresearchwhich piece of causal attemptedto sift out relevant variableswhich were expressedas operational an underlying definitionsof conceptualbase. Lightcontrasts this study with a recentstudyfrom England3l. . . systematically in schools and came observed students to very different moreholistic data it conclusions.With richer, found that schoolsmadean encein the proportionof enormous studentswho passednational differgot arrestedfor exams or delinquency. . . Whilethe output data, they also investigators collected went into the schools socialprocesseslay behind the successesand to find out what trast.In contrast to the failuresof the conwhichtried to analyse a wastefully expensive ColemanReport, trainingprogramme by isolating variables from the whole, the Britishstudy examinedthe a few and discoveredkey dimensionsof whole educationalprograms only systematic that observation overtime could discover.32 The example second derivesfrom a monograph which part of a written by Patton33 forms series Group on Evaluationwhose producedby the North Dakota Study 'particularly since, in work has been describedby Mishleras useful addition philosophical and methodological to outliningsome of the crucial differencesbetween and phenomenological approaches,they also specify the positivist doing methods for phenomenologicalresearch'.34In the Patton quantitative report in question, outlines and qualitative paradigms methodologyas opposing and their philosophical aims underpinnings. a chapterwhich to explicate the In characteristics the two of draws on a study heavily by Shapiro3swhich methodologies,he Follow Throughprogramme in schools. Such soughtto evaluatethe andout the open projectsaim to widen educational processto child-teacher and child environment enhancethe development of children.complexity interactions The of the psychologicalas well as among processes within

Alan Bryman

Thedebateabout quantitative qualitative and research

83

children, and their pedagogical ramifications, given considerable are attention. In line with manyother studiesof educational innovation, when statisticalcomparisons test scoreswere carriedout between on children undergoingthe programme and those not, there were no discernibledifferences.Shapiroalso carriedout a more qualitative investigationbased upon the observationof childrenin classrooms. These studies, by contrast, found 'the quality of relationshipbetween teacherand childrenand amongthe children,the varietyand interest of the curriculum, the generalatmosphereof the classand room were notably different'.36 Pattonarguesthat Shapiro's analysis demonstrateshow 'quantitative methodologicalproceduresdetermined the results'.37 Similarly Light observed in his ColemanRutter comparisonthat the former's'methoddeterminedwhat was
measured.38

In spite of the fact that both writersseem convinced that they understandwell the implicationsof these comparisons, their implications are less clearin the context of the issues being addressedin this paper.Both writershave a point. One cannothelp but be uneasy when studies emerge with discrepant results which seem to be attributableto the methods employed. Even more so when it is rememberedthat the practical issues and irnplications stake in at these examples are ones of great mapnitude.But two problems remain. First, how is one to 'know' which is the 'correct'analysis? Both Light and Shapiro(and Patton) opt for the qualitative studies, presumablybecause the closer involvementof researchers such in studies yield 'richer', more complete data. But these are rather subjectiverules of inferenceand it may be that there are occasions when the close involvementof the researcherobscures a different range or level of phenomena.Second, what does all this have to do with the clash between positivismand phenomenologywhich is of considerableinterest to both Light and Patton? If it is true that educational innovations do make a difference and that qualitative researchbetter equips the researcherfor such inferences, then an important methodologicalpoint is being establishedat a technical rather than an epistemologicallevel. All that is being said is that, as Patton seems to observe,39the researchtechnique must fit the problemat hand. Why,then, all this talk of the divergent philosophical bases of the two methodologies?They are quite redundantto the question of the suitability of one techniqueas againstanother in terms of solving a researchproblem. If the researchproblem is one which directly emanates from a particularepistemological position then the question of the appropriateness a research of technique is significant, for the technique must properly reflect the epistemologicalframeworkin which the researchis embedded. If the problem is one such as those mentioned here (e.g. does a particulareducationalinnovationresult in a numberof anticipated

84

Alan Bryman

benefits?),then the issue of the epistemological statusof techniques would seem to servelittle purpose.Indeed, it should be noted that the preferencefor the qualitativestudy in both of the cases cited above seem to be basedupon technicalratherthan epistemological criteria. Other examples of discussionsof this kind exist in the methodological literature in the social sciences. In terms of the questions being addressedhere, they serve as an example of a tendencyfor epistemological technicalissuesto becomeconfused. and
(ii) Qualitative Research as Preparation This next theme is a long-

standing in the literatureon research one methodology.Its fundamental pointis thatbecauseof the unstructured natureof most qualitative researchwith its associatedlack of specifiedhypotheses,except in a very loose sense, qualitativeresearchis inherently exploratory. As a result of this emphasis,the qualitativeresearcher embarkson a voyage of discoveryratherthan one of verification,so that his or her researchis likely to stimulatenew leads and avenuesof research that the quantitativeresearcher unlikely to hit upon, but which is may be used as a basisfor furtherresearch. Suchresearch follow will up the leads suggested qualitative by research will seek to confirm and or reject them using the more rigorousframeworkassociatedwith a naturalscienceapproach, quantitative i.e. methodology. A concordatof this kindbetweenthe two methodologies clearly is attractiveto those engagedin quantitative research. providesthem It with a continuous supply of leads, hunches, or hypotheses which they can confirm, reject, or qualify, while simultaneously retaininsg their methodological ascendancy over qualitative research. Since this position takes the view that evidencemust passa particular type of test prior to its acceptance,qualitativeresearchmerelyprovides fodder for quantitativeresearchers so occupies a lower rung on and the epistemologicalladder. However,researchers the qualitative in mould often accept this position too. Gans in his study o f the West End, refers to his researchas a 'reconnaissance-an initial explorationof a communityto providean overview'and then points out that: 'Many of the hypothesesreportedhere can eventuallybe tested againstthe resultsof moresystematicsocialscienceresearch'.40 This view of qualitativeresearchas a preparationfor quantitative researchis one which can be noted in a varietyof contexts, though there are those who object too, albeitoften on technicalratherthan epistemological grounds.4l Commentslike those of Gans which view qualitativeresearchers as providersof ideas are ones which operate at a technical level, i.e. they are talkingabout relationships betweenresearch techniques and their associateddata. One might anticipate,however, that the more recent writingon methodologywhichemphasises epistemological distinctions would be less likely to exhibit a preparedness to

research and The debateabout quantitative qualitative

85

process.Lofland accepta rathersecondaryrole in the overallresearch in a book which seeks to distinFish the two methodologiesand to observes:'Quarltitative underpinnings delineatetheir epistemological studies serveprimarilyto firmup and modify knowledgefirst gained in a fundamentallyqualitativefashion'.42Similarobservationscan be found in Everedand Louis's specificationof the philosophical positions of 'inquiry from the inside' and 'inquiry from the outside',43 in the introductionby Shaffir et al. to their collection of reminiscencesby qualitativeresearchers(though they do observe and in Faraday that such research is not always preliminary),44 and Plummer'ssuggestionsabout the use of life histories in the explorationof sexualbehaviour.45 The interesting feature about this perspective derives from the distinct impressionthat can be gleanedfrom the recentmethodological literature that quantitative and qualitative methodology are within a qualitativetradition epistemologicallydistinct. Researchers have increasinglysought to present their work as an alternative modus operandifor the conduct of social research.The suggestion is research somethingwhichis priorto morerigorous, that qualitative hypothesis testing researchseems to belie this point. I his is so, differencesbetween the first becauseby diminishing epistemological the two approachesit accepts by implication the notion of verification of unstructured research, thereby in part accepting the positivist frameworkin which quantitativemethodologyis deemed to be embedded.Second, in affirminga view of qualitativeresearch as something likely to be in need of confirmationit belittles the per significanceof qualitativeresearch se, and is indicativeof a lack of confidencein its associatedaccount of a theory of knowledgefor the social sciences. In these ways, evidence is found for a second issues drift out of alignareain which technicaland epistemological ment. While there may be technical reasons why social research might usefully be built upon a modus operandiin which qualitative researchprovidesinsights and hunches for empiricalconfirmation, seem to indicate the philosophicalaccounts of the two approaches in very fundamentaldivergences orientationbetween the two methto odologies. The suggestionthat one is or may be preparatory the framework. otherplacesboth withinthe sameepistemological Met-hods The third area in which technical and (iiz) Combining epistemologicalissues become confused is the suggestionthat both quantitative and qualitativeresearchare best thought of as compof lementaryand should thereforebe mixed in research many kinds. This emphasis has coincided-with the growing attention focused in upon 'triangulation'46 social reseaJch.\Shile this termis occasionally taken to refer to a broad approach in which are combined 'multiple observers,theoreticalperspectives,and methodologies',47

86

Alan Bryman

it generallydenotesa reference a combinationof research to methods. The point about the advocacyof combinedstrategies that it seems is to exudegoodsense.Whyshouldtherenot be attemptsby researchers to capitalizeupon the strengths differenttechniquesand combine of them in overall researchprojects? Such a view seems to lack the methodologicalparochialism that is at risk when writersextol the virtues of a particular method, while directly or inferentially denigrating alternatives. the The difficulty with this thesis, in the context of the present discussion,is that the argumentfor triangulated strategiesis essentially a technicalone. It impliesthat a better overallview of reality is achievedwhen, say, a social surveyis linkedto some unstructured questioningor participantobservation.It also enablesthe researcher to check the possibleeccentricities a particular of techniquein order to discernwhether any inherent bias is present.Thus Whyte48has expressed his irritation with the tendency for the two types of researchto polarizeand has expressedhis preferencefor combining the two strategies,thereby enjoying the fruits of each. He demonstrates the utility of employing an integratedstrategy by drawing upon his researchinto Peruvianvillagesin which both surveysand anthropological techniqueswere employed.Anomalousresultswhich were derivedfrom surveydatawerecross-checked againstqualitative evidence,and out of this processit was possibleto makea substantial theoreticaladvance.So Whyteasserts:'Mystrategycalls for a weaving back and forth among methods through the various stages of research.'49 Recommendations this kind arebasedupon technical of considerations, namely that a superiorpiece of researchwill emerge if techniquesare combined. This contention may well be true, but the debates about quantitativemethodology are, as has repeatedly been observed, epistemologicalin nature. In spite of this, many writers who address and often acknowledgethe distinctivenessof the two methodologiesin philosophicalterms, make pleas for the mixingof the two. Douglas, whose work is generally located outside the positivist stream and who is often accreditedas one of those who has influenced the outlining of the philosophical bedrock of qualitative methodology5tells us that: Since all researchmethods have costs and benefits,and since they differ greatly in their particularcosts and benefits, a researcher generally finds it best to use some combinationor mixture of
methods.Sl

Similarindicationscan be discernedin Agar'sintroductorybook on ethnography,S2 Wilsonin his exposition of the 'qualitative-phenomenologicalhypothesis',S3 Rist when explicatingeducationalresearch paradigms,S4 and many others. Other researcherslike Sieber and

Thedebateabout quantitative qualitative and research

87

Trendss point to the cumulativeadvantagesthat accrued to their researchby combining both quantitativeand qualitativemethods, though the latter author is somewhat suspiciousabout the extent to which a neat, additivedovetailingis as easy as some writersmake it sound. In slight contrast,Myerssought to developa methodological strategy 'somewhere between ethnographicand conventional surveymethods'.S6 James in the context of her reflectionsupon the ethnographicstudy of drug use, views such researchas 'fillingout the gaps' and 'putting meat on the bones' of quantitativeanalyses of these phenomena.S Finally, van Maananin the preface to a 7 special number of the Adminzstrative Science Quarterlyvery succinctly outlines the epistemological issues at stake, but then asserts that 'qualitative quantitative not mutuallyexclusive'.S8 and are Many of these authors explicate the philosophicalassumptions upon which the two methodologiesare supposed to be grounded, while others seem to take them as 'givens'in their discussions,and then move to a considerationof the possibility of mixing them in pieces of research.The difficulty is that at a technicallevel methods may be commensurable Whyte and others have sought to show, as but at an epistemologicallevel quantitativeand qualitativemethodologies are written about as though their knowledge bases are quite incommensurable. Consider statementsuch as: a When we speak of 'quantitative'or 'qualitative'methodologies, we are, in the final analyses speakingof an interrelatedset of assumptions about the social world which are philosophical, ideological, and epistemological. They encompass more than simplydatagathering techniques.S9 The recent methodologicalliteratureis replete with views such as this which make clear statementsabout issues which are more than technical ones. Yet, as with the two previoussections, one finds researchers oscillatingbetween epistemological and technicalmodes of expressionand levels of analysis.In pointing to the virtuesof a triangulatedstrategyvirtuallyall of the writerscited here, many of whom have done a great deal to reinforcethe philosophicalissues, move in the directionof a technicallevel. Thisis not surprising since positivismand phenomenology,to take the two majorphilosophical strands,are far apartin termsof what they view as the properstance to be takenin relationto the social world,what is to passas warrantable knowledge,and the way in which knowledgeis accumulated. As such, the possibility of a reconciliationindeed seems remote. In the context of a particular study a researcher may perceiveareas in which a useful contributionmight be made by both quantitative and qualitativemethods, but it cannot be derivedfrom this that the epistemologicalissues signified by the debate between quantitative and qualitative methodologiesarezpsofacto reconciled.

88

Alan Bryman

It may be that at the technical level the quantitative/qualitative distinction is a rather artificial one. The arguments for trian;ulatin; research techniques suggest this in part. But even research which relies almost exclusively upon one mode rather than the other often contains elements of both. Survey researchers often punctuate their research reports with brief transcripts of the verbalizations of their respondents. While the use of these transcripts is often to illustrate a quantitatively established point and thereby relieve the reader from the tedium of a large number of tables, their use is often to give some sense of how respondents view a particular cluster of issues. Indeed, some survey researchers seem to exhibit a commitment to the epistemology of qualitative research, in particular its emphasis upon seeing through the respondents' eyes, yet use the technical paraphernalia of the survey.60 In contrast, qualitative researchers frequently make quasi-quantitative assertions, such as 'many', 'frequently', or 'some of the time'.61 While the establishment of such implicit frequencies is far removed from the rigorous statistical techniques often associated with quantitative methodology, they contribute to a blurring of the lines between the two styles. At the epistemological level, the distinction is less obviously artificial since the underlying tenets relate to fundamentally different views about the nature of the social sciences, which have resisted reconciliation for a very long time.62 However, a great deal of research which is apparently either quantitative or qualitative in orientation is conducted with little, if any, recourse to such philosophical debates. At the technical level, researchers seek to achieve a degree of congruence between a research problem and a technique, or cluster of techniques, to answer the issue at hand. Consequently, while the quantitative/qualitative distinction may be a useful device for distinguishing types of technique as an organizing principle in the context of text-books about research methods, its use as an account of research practice is not without problems. The suggestion that participant observers are carrying out research which is outside the positivist mainstream often seems highly farfetched when their research monographs are examined closely. In other words, while the apparent debate between quantitative and qualitative methodology may have some meaning at the epistemological level, e.g. in terms of causal adequacy as against adequacy at the level of meaning, in the context of research practice there is no direct link between these precepts and particular techniques, since research typically comprises both elements. This is also a clear inference that can be gleaned in the writings of the advocate of methodological 'triangulation'. Indeed, there may be a case for saying that techniques are neutral in respect of epistemological issues and debates.

Thedebateabout quantitative qualitative and research

DISCUSSION

89

This paper has distinguished between technical and epistemolc)gical levels of discussion in the literature dealing with the quantitative/ qualitative distinction. Ihree areas have been pinpointed in which the levels of discussion become unclear, fundamentally because the writers concerned often shuttle uneasily between epistemological and technical spheres of discourse. The idea that there is a link between methodology qua epistemology, on the one hand, and technical issues relating to research method, on the other hand, is a conxention that has increased in prominence in the last ten years or so. The basic problem with this line of discussion, a difficulty erhich may lie behind some of the inconsistencies which some writers have exhibited, is precisely that it is a conxJention.There is no necessary 1:1 relationship between methodology and technique in the practice of social research. Snizek63 has shown, drawing upon an analysis of journal articles, that research techniques cannot be directly extrapolated from a knowledge of a researcher's epistemological assumptions. Similarly Marsh64who has also sought to distinguish philosophical issues from technieal ones, has questioned whether the survey technique is inherently positivistic. One might equally question whether participant observation is inherently phenomenological, for it is difficult to discern in the writings of generations of social scientists using this technique such as \Shyte, Gans or Skolnick,65 as uTell some of the as more recent writers, a deep preoccupation with philosophical matters. Rather, they exhibited a concern for achieving a piece of research in a manner that was most appropriate to the topic at hand. Indeed, much of this work could easily be regarded as positiist, or a ariant of it, and some writers66 have located participant observation in the same epistemological space as the social survey. \\lhile they may hane had a preference for a particular style of research, this preference ^ras more likely to be a personal one, often deriving from their training. It may be that this is not a good reason for choosing a particular research method, but it does suggest that attempting to relate questions of method to philosophical debates in the manner of many recent authors fail to supply a sufficiently accurate account of the research process. It may also be that we are witnessing the classic confusion of 'is' and 'ought', namely that mans writers feel that the choice of method should be taken in the light of an appreciation of philosophical contexts, but this is not what they appear to be saying. Their argument seems to be that quantitative and qualitative methodology (and their Xarious synonyms) are or exhibit distinctive epistemologies and that particular methods of research are appropriate to each. The argument of this article is that, while these are highly stimulating suggestions, they need to be subjected

9o

Alan Bryman

to considerableinvestigationbefore they can be consideredaxioms of research the socialsciences. in AlanBryman Department SocialSciences of Loughborough University of Technology
N OTE S

1 J. Buchler, The Conceptof Method, New York, ColumbiaUniversityPress, 1961, p. 126. 2 For example, H. S. Becker and B. Geer, 'Participantobservationand interviewing: a comparison', Human Organization, no. 3, vol. 16, Fall, 1957, pp. 28-32; M. Zelditch, 'Some methodological problems of field studies', American Journal of Sociology, vol. 67, no. 5, March 1962, pp. 566-76. 3 M. Trow, 'Comment on participant observation and interviewing', Human Organization,vol. 16, no. 3, Fall 1957, p.33. 4 Prominent examples are: W.J. Filstead, Qualitative Methodology: FirsthandInvolvement with the Social World, Chicago, Markham, 1970; J. Lofland, Analyzing Social Settings, Belmont, Wadsworth,1971; R. Bogdan and S. J. Taylor, Introduction to Qualitative Research Methods, New York, Wiley, 1976; J. D. Douglas, InvestigativeSocial Research,Beverley Hills, Sage, 1976; H. Schwartz and J. Jacobs, Qualitative Sociology: A Method to the Madness, New York, Free Press, 1979; W. B. Shaffir, R. A. Stebbins, and A. Turowetz, Fieldwork Experience: Qualitative Approaches to Social Research, New York, St Martin'sPress, 1981. 5 In particular, Administrative Science Quarterly, vol. 24, no 4, December 1979 and Sociological Review, vol. 27, no. 4, November 1979. 6 M. Q. Patton, Alternative Evaluation ResearchParadigms,N. Dakota, University of North Dakota Press, 1975.

7 R. C. Rist, 'Overview- on the relations among educational research paradigms:from disdain to detente', Anthropology and Education Quarterly, vol. 8, 1977, pp. 42-9 and A. J. Magoon, 'Constructivist approaches in educational research', Review of Educational Research, vol. 47, no. 4, Fall 1977, pp. 651-93. 8 R. Evered and M. R. Louis, 'Alternativeperspectivesin the organizational sciences: "Inquiry from the inside" and "Inquiry from the outside"', Academy of ManagementReview, vol. 6, no. 3, 1981, pp. 585-95. 9 For example, F. W. Lutz and M. A. Ramsay, 'The use of anthropological field methods in education', Educational Researcher,vol. 3, November 1974, pp.5-9; Magoon,op. cit.; Douglas op. cit.; S. Wilson, 'The use of ethnographic techniques in educational research', Review of Educational Research, vol. 47, no. 1, Winter1977, pp. 245-65. 10 Such emphases are typically stressed because they represent the manifestations of positivism in sociology. More philosophical treatments tend to stresspositivism'sfundamental tenets, e.g. L. Kolakowski, Positivist Philosophy, Harmondsworth, Penguin, 1972, ch. 1, or A. Giddens 'Introduction' in Positivism and Sociology, London, Heinemann, 1974, pp. 1-22. 11 Examples are legion, with the following being a short list from a wide field: Wilson, op. cit., pp. 249, 259; Magoon,op. cit., p. 652; Lofland, op. cit., pp. 4, 7; Filstead, op. cit., pp. 6-7; Douglas, op. cit., pp. 24, 190-1; R. M. Emerson, 'Observational

research and Thedebateabout quantitative qualitative

Field Work', Annual Review of Sociology, vol. 7, 1981, pp. 351-7; M. H. Agar, 17zeProfessional Stranger,New York, Academic Press, 1980, p. 194; t J . S. Spradley,Participan Observation, New York, Holt, Rinehart and Winston,1980, p. 194. 12 E. G. Mishler, 'Meaningin context: is there any other kind?Harvard Educational Review, vol. 49, no. 1, February 1979, pp. 2, 8; J. van Maanen, 'Reclaiming qualitative research for organizational research: a preface', Administrative Science 01arterly, vol. 24, no. 4, December 1979, p. 520; P. Halfpenny, 'The analysis of qualitative data', Sociological Review, vol. 27, no. 4, November 1979, p. 803. 13 B. Glaser and A. L. Strauss, The Discovery of Grounded Theory, Chicago, Aldine, 1967; W. B. Shaffir, R. A. Stebbins, and A. Turowetz, Fieldwork Experience, New York, St. Martin'sPress, 1981, pp. 6, 7, 24; P. Rock, The Making of Symbolic Interactionism, London, Macmillan, 1979, pp. 183, 207; Lofland, op. cit., p. 76; Bogdan and Taylor, op. cit., p. 80; J. Irwin, 'Participant observation on criminals', in Douglas, Research on Deviance, New York, RandomHouse, pp. 130-1. 14 S. T. Bruyn, The Human Perspective in Sociology, New Jersey, Prentice Hall; C. A. B. Warren,Identity and Community in the Gay World, New York, Wiley, p. 10; Wilson, op. cit., pp. 245-249; Mishler, op. cit., 10-11; Rock, op. cit., pp. 192, 195; Bogdan and Taylor, op. cit., pp. 2,5. 15 Patton, op. cit., p. 19; Filstead, op. cit., p. 4; Warren,op. cit., p. 160; Bogdan and Taylor, op. cit., p. 14; A. Faraday and K. Plummer, 'Doing life histories', Sociologzcal Review, vol. 27, no. 4, November1979, p. 776. 16 Spradley, op. cit., pp. 8-9; Shaffir et al., op. cit., p. 112; Rock, op. cit., p. 178; Irwin, op. cit., p . 131; Bogdanand Taylor, op. cit., p. 13. 17 J.L. Heap and P.A. Roth, 'On phenomenological sociology', American Sociological Review, vol.

91

38, no. 3, June 1973, pp. 354-67. 18 A. Schutz, ThePhenomenology of the Social World,London, Heinemann,p. 31. 19 A. Schutz, Collected Papers Volume 1, The Hague, Martinus Nijhof,1967, p. 34. 20 Emerson, op. cit., pp. 315, 360; Agar, op. cit., p. 11; Lofland, op. cit., p. 76; Evered and Louis, op. cit. 21 H. Blumer,'Sociologicalanalysis and "the variable"', American Sociological Review, vol. 21, no. 6, December 1956, pp. 683-90. 22 Evered and Louis, op. cit.; Emerson, op. cit., pp. 353, 374, 375; Magoon, op. cit., p. 653; Patton, op. cit., pp. 9-10; Rist, op. cit., p. 42. 23 J. M. Johnson, Doing Field Research,New York, Free Press, 1975, p. 3. 24 H.J. Gans, The UrbanVillagers, Glencoe, Free Press,p. 34. 25 Zelditch, op. cit. 26 D. P. Warwickand C. A. Lininger, The Sample Survey: Theory and Practice, New York, McGraw-Hill, 1975, pp. 9-10, as quoted with approval in the text-book by A. erOrensteinand W. R. F. Phillips,Und standing Social Research, Boston, Allyn and Bacon, 1978, pp. 411-12. 27 H. J. Gans, The Levittowners, London, Allen Lane, The Penguin Press,p. 450. 28 W. F. Whyte, Street Corner Society, second edition, Chicago, Universityof ChicagoPress,p. 308. 29 D. Light, 'Surface data and deep structure: observing the organization of professional training', AdministrativeScience 0uarterly, vol. 24, no. 4, December 1979, pp.551-9. 30 J. S. Coleman, K. Q. Campbell, C.J. Hobson, J. McPartland, A. M. Mood, F. D. Weinfeldand R. L. York, Equality of EducationalOpportunity, Washington,U.S. GovernmentPrinting Office,1968. 31 M. Rutter, B. Mangham, P. Mortimore, and J. Ouston, Fifteen Thousand Hours, Cambridge, M.A., UniversityPress, 1979. Harvard 32 Light, op. cit., p. 558.

92
33 Patton, op. cit. 34 Mishler,op. cit., p.11. 35 E. Shapiro, 'Educationalevaluation: rethinking the criteria of competence', School Review, November 1973, pp.523-49. 36 Ibid., pp.528-9. 37 Patton, op. cit., p.15. 38 Light,op. cit.,p.558. 39 Patton, op. cit., p .14. 40 Gans, The Urban Villagers, op. cit., p. 350. 41 For example W. F. Whyte, 'Research methods for the study of conflict and co-operation', The A merican Sociologist, vol. 11, no. 4, November 1976. 42 Lofland, op. cit., p. 6. 43 Evered and Louis, op. cit., p. 390. 44 Shaffiret al., op. cit., pp. 10-11. 45 Faraday and Plummer,op. cit., p. 778. 46 E. J. Webb, D. T. Campbell, R. D. Schwartz, and L. Sechrest,

Alan Bryman

M. G. Trend, 'On the reconciliation of qualitativeandquantitative analyses: a case study',HumanOrganization, vol. 37, no. 4, Winter 1978, pp. 345-54. 56 V. Myers, 'Toward a synthesis of ethnographic and field methods', Human Organization,vol. 36, no. 3, Fall 1977, pp. 244-51. 57 J. James, 'Ethnography and social problems', in Weppner,Street Ethnography, Beverley Hills, Sage, 1977,pp.184, 193. 58 van Maanen,op. cit., p. 520. 59 Rist, op. cit., p. 62. 60 An obvious example is J. H. Goldthorpe, D. Lockwood, F. Bechhofer, and J. Platt, The Affluent Worker in the Class Structure, London, Cambridge UniversityPress. 61 Gans,op. cit., p. 34. 62 This is very evident in W. G. Runciman, A Treatise on Social Theory Volume One: The Meth. odology of Social Theory, London, Cambridge UniversityPress,1983. Unobtrusive Measures: Non-reactive 63 W. E. Snizek, 'An empirical Research in the Social Sciences, assessment of "Sociology: A Multiple Paradigm Science"', The American Chicago,Rand McNally,1966. 47 N. K. Denzin, The Research Act Sociologist, vol. 11, no. 4, November in 50 cio logy, London, Butterworth, 1976, pp. 217-19. 1970, p. 310. 64 C. Marsh, 'Problems with sur48 Whyte,op. cit. veys: method or epistemology', Soci49 Ibid., p. 216. ology, vol. 13, no. 2, May 1979, 50 Cf. Bogdan and Taylor, op. cit., pp. 293-305. See also C. Marsh, The ch. 1. Survey Method, London, Allen & 51 Douglas,op. cit., p. 30. Unwin,1982. 65 Whyte, Street Corner Society, 52 Agar,op. cit., p.53. 53 Wilson,op. cit., p. 261. op. cit.; Gans, op. cit.; J. Skolnick, 54 Rist, op. cit., p.48. Justice Without Trial, New York: 55 S. D. Sieber, 'The integration Wiley, 1966. of fieldwork and survey methods', 66 For example, D. Willer and American Journal of Sociology, vol. J. Willer,Systematic Empiricism,New 78, no. 6, May 1973, pp. 1335-59; Jersey, Prentice-Hall, 1973.