Qualitative interviews in psychology: problems and possibilities

UNCORRECTED PROOF

Qualitative interviews inpsychology: problems andpossibilities

Jonathan Potter and Alexa Hepburn

Discourse and Rhetoric Group, Department of Social Sciences, Loughborough University,Loughborough, Leicestershire, LE11 3TU, UK

This paper distinguishes a series of contingent and necessary problems thatarise in the design, conduct, analysis and reporting of open-ended orconversational qualitative interviews in psychological research. Contingentproblems in the reporting of interviews include: (1) the deletion of theinterviewer; (2) the conventions of representation of interaction; (3) thespecificity of analytic observations; (4) the unavailability of the interviewset-up; (5) the failure to consider interviews as interaction. Necessaryproblems include: (1) the flooding of the interview with social scienceagendas and categories; (2) the complex and varying footing positions ofinterviewer and interviewee; (3) the orientations to stake and interest onthe part of the interviewer and interviewee; (4) the reproduction ofcognitivism. The paper ends with two kinds of recommendation. First,we argue that interviews should be studied as an interactional object, andthat study should feed back into the design, conduct and analysis ofinterviews so that they can be used more effectively in cases where they arethe most appropriate data gathering tools. Second, these problems withopen-ended interviews highlight a range of specific virtues of basinganalysis on naturalistic materials. Reasons for moving away from the use ofinterviews for many research questions are described. Qualitative Researchin Psychology 2005; 2: 1!/27

Key words: cognitivism; discursive psychology; footing; qualitativeinterviews; research agenda; stake and interest; transcription

Correspondence: Jonathan Potter, Discourse and Rhetoric Group, Department of Social Sciences, LoughboroughUniversity, Loughborough, Leicestershire, LE11 3TU, UK.E-mail: [email protected]

# 2005 Edward Arnold (Publishers) Ltd 10.1191/1478088705qp045oa

Y:/Arnold/QP/articles/QP045OA/QP045oa.3d[x] Wednesday, 17th August 2005 15:34:32

Qualitative Research in Psychology 2005; 2: 1!/27www.QualResearchPsych.com

UNCORRECTED PROOF

This paper is about the use of open-endedor conversational interviews in psychologi-cal research. It has a series of aims. First itwill briefly document the centrality of thequalitative interview in contemporary psy-chology and describe some of the researchthat has been carried out on the conduct ofsocial research methods. Second it willhighlight some shortcomings in the wayinterview research has often been reportedthat can be reasonably easily rectified.Third it will identify a range of features ofqualitative interview interaction that areendemic and inescapable, and note someof the difficulties they raise for the satisfac-tory analysis of interviews. Fourth, it willconsider the implications of this discussionfor improvements in the use of open-endedinterviews. Finally, it will consider some ofthe advantages of working with naturalisticmaterials that are specifically highlightedby considering the endemic features ofopen-ended interviews. In effect this dis-cussion of interviews will highlight somemore systematic features of naturalisticmaterials.

The paper will take a schematicapproach. The aim is to pull together abroad range of issues and clarify theirimplication for understanding interviews.Each of these issues is worthy of muchfuller treatment but the intention here is tocapture the broader implications of thecombination of these issues. Likewise, therewill not be many attempts to show thegenerality of these problems !/ however,we expect researchers who work with inter-views to recognize them without difficulty.Moreover, it would be invidious to pick outparticular pieces of research, or particularmethodological expositions, for comment.We have chosen instead to illustrate ourpoints with an example from our owninterview research. Our approach will be

argumentative but constructive. The aim isto challenge the taken-for-granted positionof the open-ended interviews as the methodof choice in modern qualitative psychology.We will draw heavily on the resourcesof discursive psychology (Edwards, 1997;Edwards and Potter, 1992; Potter andEdwards, 2001) and conversation analysis(Sacks, 1992; Hutchby and Wooffitt, 1998)to develop the argument. Nevertheless, thepoints are intended to have a much broaderrelevance than specifically to those re-searchers with a discourse or conversationalinterest. We will also draw on an increasingbody of research (mostly from outside ofpsychology) that has started to topicalizethe research interview and the activitiesthat make it up, to raise questions aboutthe interpretations of interviews and sug-gest ways for developing this research in-strument (prominent examples includeBaker, 2001; Lee and Roth, 2004; Rapley,2001; van den Berg, 2003; Widdicombe andWooffitt, 1995). The immediate aim is togenerate debate about the role of qualitativeinterviews in psychology. The ultimate aimis to improve the quality of interviews andtheir targeting at particular research pro-blems. The ideal would be much less inter-view research, but much better interviewresearch.

Before we start a note on terminology. Thefocus is on interviews where participantsare not required to merely choose from aselected set of response categories, or tickboxes, to rate vignettes, or to engage in someother ‘structured’ procedure. That is, theseare interviews where participants are an-swering questions verbally and there issome attempt to capture their words (per-haps using notes, but more likely recordingsand transcriptions). Such interviews havebeen called conversational, active, qualita-tive, open-ended or even sometimes (con-


2 J Potter and A Hepburn

UNCORRECTED PROOF

fusingly) semi-structured. An interview ofthis kind will typically be guided by aschedule of topics or questions, althoughtheir order in the interview may vary andinterviewers are likely to depart from theschedule and use a variety of follow-upquestions (or comments, responses, orsome other contributions). We are deliber-ately not attempting to develop a morespecific definition at this point or to makefurther distinctions between different kindsof qualitative interview. See Kvale (1996),Mischler (1996), Fontana and Frey (2000)and Warren (2001) for detailed discussionsfrom a range of perspectives from across thesocial sciences. We will refer to the inter-views of this kind simply as interviews fromnow on in the paper. Many of the issuesraised here have broader relevance to socialscience focus groups, social surveys andother techniques for data generation andapproaches to analysis. However, for claritywe will restrict our focus here to interviews.

The interview in contemporaryqualitative psychology

There are many ways of documenting thecentral role that interviews play in contem-porary qualitative psychology. For brevity,we have chosen to do this by highlightingthe content of two excellent recent booksthat offer collections of how-to-do-it chap-

ters on qualitative research in psychology:Camic et al . (2003) and Smith (2003). Ineffect, these are the US and the Europeanstate-of-the-art collections. The US volumeis published through the American Psycho-logical Association (their first foray intoqualitative methods) and Sage, a majorpublisher of qualitative research in psychol-ogy, published the European version.

There is a core set of perspectives repre-sented in each book that are shared acrossthe two volumes, and provide an indicationof what might be considered standard incontemporary qualitative work. The thingwe wish to note is the relation between thebroad psychological perspective, the objectof study (broadly speaking), and the techni-que for data generation (often just called‘the method’). For clarity we have laid thisout in Table 1.

The point of this sketchy and somewhatrhetorical summary, of course, is that de-spite the highly varied topics that thesedifferent perspectives focus on, when itcomes to generating materials to studythey all have the qualitative interview(with the exception of discourse analysisand discursive psychology, of which moreis discussed below) as the approach ofchoice. Moreover, if we consider the wayinterviews are described in these chaptersmostly they are not only the technique ofchoice but also the choice to do interviewsis taken-for-granted. There is very little

Table 1 Psychological perspective, object of study and technique of data generation

Perspective Object of study Technique of data generation

Ethnography Cultures, rituals, groupings InterviewsPhenomenology Experience, consciousness InterviewsPsychoanalysis The unconscious InterviewsNarrative psychology People’s life stories InterviewsGrounded theory Highly varied InterviewsDiscourse analysis and discursivepsychology

Talk and texts Interviews and naturalistic data


Qualitative interviews in psychology 3

UNCORRECTED PROOF

explicit justification for the use of inter-views and their appropriateness for therelevant object of study. This sense of inter-views as the natural way to do nonexperi-mental, nonquestionnaire and nonsurveywork in psychology is one of the thingsthis paper is designed to dissipate.

Reservations and research studies

Ethnomethodologists and conversation ana-lysts have been sceptical about the use ofinterviews since the inception of thoseperspectives in the 1960s. However,although some important studies weredone concerning the operation of researchmethods (often of a more structured kind)by Garfinkel (1967) and Cicourel (1964,1974) the main reasons for scepticism weresimply that they had found more interestingand fruitful alternatives in the study ofpeoples’ organized practices (conversa-tional or otherwise) in natural settings.Nevertheless, some classic ethnomethodo-logical work, notably Garfinkel’s (1967)‘Agnes’ transexuality study, was based oninterviews.

Qualitative researchers in sociology haveconsidered a range of difficulties with theresearch interview. Silverman (2001) pro-vides an excellent summary of problems inusing interviews to do social research,focusing in detail on ‘positivist approaches’,which take interviews as a source of facts,and ‘emotionalist approaches’ that takeinterviews as a pathway to participants’authentic experiences (see also ten Have,2004). For the most part, these critiques arenot based on systematic research into whatgoes on in interviews, nor have they beendirected at, or picked up by, researchersworking within psychology.

There is a growing literature that usesethnomethodology, conversation analysisand/or discursive psychology to study theoperation of methods in practice. The moststudied topic here has been the standar-dized survey interview. Work on standar-dized surveys from an interactionalperspective was stimulated by Suchmanand Jordan (1990), who highlighted thefailure of survey researchers to appreciatethe centrality of interaction in administra-tion surveys and the consequences for theachievement of standardization. They ar-gued that survey researchers would needto respond much more flexibly to the con-tingencies of natural conversation if any-thing approaching standardization was tobe achieved. This tradition was continuedby others and has evolved into substantialbody of work (Houtkoop-Steenstra, 1995,1996, 1997; Maynard and Schaeffer, 1997,2000; Schaeffer and Maynard, 1996). Morerecently research has focused on the admin-istration of questionnaires (Antaki, 1999;Antaki and Rapley, 1996; Antaki, Hout-koop-Steenstra, Rapley, 2000; Rapley andAntaki, 1996) the organization of interactionin social science and market research focusgroups (Myers, 1998; Myers and Mac-naghten, 1999; Puchta and Potter, 1999,2002, 2004; Puchta, Potter and Wolff, inpress) and the administration of psycholo-gical tests (Marlaire and Maynard, 1990;Schegloff, 1999).

For some time discourse analysts havehighlighted the significance of the interac-tional nature of interviews (Potter andMulkay, 1985; Potter and Wetherell, 1995).Widdicombe and Wooffitt (1995) in theirwell-known study of talk about youth sub-cultures develop a major critique of theability of interviews to attend to the com-plicated categorization practices that areinvolved in them. However, they are less



UNCORRECTED PROOF

focused on how these are a function speci-fically of the interview features of thematerial. Rapley (2001), Lee and Roth(2004), and some contributions to van denBerg et al ., (2003) have focused on therelationship between the interactional orga-nization and research tasks. We will drawon this work as well as some of the otherliterature as we go along. Some of our pointswill build on observations in Antaki et al .,(2003). However, our goal is less descriptiveand analytic and more focused on using theapparatus of CA, DP and ethnomethodologyto highlight some important and largelyoverlooked problems with interviews.

PART 1: CONTINGENT PROBLEMS

We will break our discussion into two parts:contingent problems and necessary pro-blems. The rationale for this distinction isto separate out problems with interviewsthat are contingent in the sense that they arenot a necessary feature of doing interviewresearch, but could be (relatively easily)fixed, or at least attended to. The secondpart will focus on some problems that arenecessary (endemic and inescapable) to theenterprise of researching with interviews.We will start with five contingent problems:(1) the deletion of the interviewer; (2) theconventions of representation of interac-tion; (3) the specificity of observations; (4)the unavailability of the interview set-up;(5) the failure to consider interviews asinteraction. Our points here are not original.All of them have been made in one form oranother before, although often informallyrather than formally. However, the pointhere is to develop them, illustrate them,and collect them together as a package andnote that researchers still, regularly, fail totake them seriously.

The paper will focus on a concrete exam-ple to illustrate these points and the onesthat are to come. As indicated above, itwould be invidious to select one study fromanother researcher to illustrate problemsthat are very widely shared across interviewresearch. Instead, we will offer a reconstruc-tion (with actual materials) of some genericfeatures of the presentation and interpreta-tion of interview research. This comes froma research project written up in Hepburn(1995, 1997a,b, 2000); Hepburn and Brown,2001). From the original interview corpuswe selected haphazardly one interview and(roughly) one question and answer se-quence. The points we will make are in-tended to be generic to interviews sofindable just about anywhere. We expectthat interview researchers will have nodifficulty in recognizing them in their ownwork and that of others.

The deletion of the interviewer

This point has been made in a range ofdiscussions of interview research, usuallyin terms of the interviewee’s talk beingtaken out of context (see Bowers, 1988 fora reflexively applied example). We wish tofurther clarify what this means in practice.The following extract shows the sequencerendered in a style common to a wide rangeof contemporary qualitative research (again,we are not wishing to pick on particularexamples, however, illustrations can befound in Camic et al . (2003) and Smith(2003)).

Extract 1

Teacher: I think all teachers are stressed. Er becausethey’re stressed they may react um inappro-priately in certain situations because theyare near the edge themselves. Erm if you’re



UNCORRECTED PROOF

tired and stressed you’re not always in thebest situation to make good judgements. Thechildren I think at least are slightly moreaware of this than they used to be in thepast, but yes I would say it can affect it.

This extract uses a conventional ortho-graphic representation of talk. This con-structs it to look like a form of playscript.The talk is rendered into sentences withconventional textual punctuation. The firstpoint to note here is the deletion of theinterviewer. This works in two ways. Mostobviously the interviewer is not representedin the extract. We only have talk listed asfrom the Teacher. Yet more subtly the talk isrendered as an abstract statement pronoun-cing on the nature of teachers and theeffects of stress rather than a specific answerto a specific question put by a specificinterviewer.

This is clarified if we start to fill in moreof what is missing using the same ortho-graphic style. We have rendered the addi-tions in a lighter font to show the kind oftalk that is commonly omitted.

Extract 2

Teacher: What evidence do you have of discussion(Int: yes) and achieving a whole atmospherewithin a school

Int: Yes yes. So do you feel then that theconstraints on teachers’ time and the re-sources that are available to you actually erconstrain your ability to do your job well todeal effectively with with kids

Teacher: yes I think all teachers are stressed erbecause they’re stressed they may react uminappropriately in certain situations be-cause they are near the edge themselves(Int: yes yes) erm if you’re tired and stressedyou’re not always in the best situation tomake good judgements (Int: oh yeah yeah)the children I think at least are slightly moreaware of this than they used to be in the past(Int: mm mm) but yes I would say it canaffect it

This includes more of the interaction. Itgives some representation of the inter-viewer’s question and various ‘interjec-tions’. Yet it is still rendered as playscript.The precise actions going on here are hardto pin down because of the representation ofinteraction that is offered here.

The conventions of representation ofinteraction

For some thirty years conversation analystshave been developing styles of transcriptionthat capture elements of talk that are inter-actionally relevant. Gail Jefferson has beenthe major figure in these developments. See,for example, Jefferson (1985) and summa-ries in ten Have (1999) and Hutchby andWooffitt (1998) and conventions summar-ized in the appendix. The extract belowshows the interview sequence representedusing Jeffersonian transcription.

Extract 3

1 Teacher: What evidence do you have of discussion. an-

2 [and ] achieving a whole (0.1)

3 Int: [YEah.]

4 Teacher: atmosphere within 8a school.85 Int: YEAH.

6 (0.3)



UNCORRECTED PROOF

(An audio record of this interactionis available through the LoughboroughDiscourse and Rhetoric Group web site at

http://www.lboro.ac.uk/departments/ss/centres/dargindex.htm)

Even in this brief extract we are able tohighlight a wide range of hearable, andtherefore potentially conversationally live,features that are missed in the standardorthographic version (see Rapley, 2001, for

7 Int: Ye [ah.]

8 Teacher: [8mm8]9 (0.4)

10 Int: So d’you !/feel then that the constrai:nts

11 on teachers’ ti:me and the resources!/that are

12 available to youB/ actually (0.9) er constrain

13 your ability to do your job !/well to deal

14 effectively with- (0.2) 8with kids and8 (0.4)15 [((inaudible))]

16 Teacher: [U : : M : ] (1.0) ((swallows)) yes, (0.7)

17 I think all teachers are stressed.

18 (0.3)

19 Int: Mm.

20 Teacher: Er because they’re stressed (0.2) they may

21 react (0.5) u::m inap!/propriately,22 Int: Mhm.

23 (0.3)

24 Teacher: In certain situatio[ns,]

25 Int: [ M ]hm.

26 (0.4)

27 Teacher: Because they are near (.) the edge

28 themse[lves,]

29 Int: [Yeah.]

30 (0.5)

31 Int: 8Yeah.832 Teacher: Er::m (0.9) if you’re ti:red, (0.1) an stressed,

33 (.) er you’re !/not always in theB/ best !/situation

34 to makeB/ good judge[ments.]

35 In: [!/Oh ye]ah. 8yeah.836 Teacher: Er:m (0.4) the CHILdren I think at least are

37 slightly more aware of this [than they used to]

38 Int: [ M m : : : . ]

39 be in [the pa:]st.

40 Int: [M m: : .]

41 Int: Mm:.

42 (0.2)

43 Teacher: Er : :m :"/

44 Int: "/Mm.

45 (0.9)

46 Int: Mm.

47 (1.6)

48 Teacher: BUT huh !/YEs I would say hh (0.2) er it

49 can affect it.



UNCORRECTED PROOF

a further illustration of this kind). Again,we are not suggesting that all of thesefeatures are absent from all reports ofinterview research, but we did not haveany difficulty finding a range of currentpublications where such features aremissed. Examples are easily identified inthe Smith and Camic volumes. We havesummarized them in Table 2.

More broadly our fictionalized typicaltranscript in Extract 1 misses potentiallyconsequential interviewer actions suchas the acknowledgement tokens (Claymanand Heritage, 2002) on lines 19, 22 and25; it also misses the news receipt andagreeing second assessment (Heritage,1984; Pomerantz, 1984) on line 35. Givethe significance of such elements to thedevelopment of this talk we suggest thatthey ought to be represented in thetranscript and therefore made available toreaders.

There are real tensions here with goodarguments in each direction. In the past ithas not been uncommon to advocate akind of Jefferson Lite for interview re-search, a form of transcription capturing

the words and some of the grosser ele-ments of stress and intonation, but leavingpauses untimed and not attempting tocapture more subtle elements such asclosing and continuing intonation, latch-ing, and so on. Potter and Wetherell (1987)advocates just this, and has often beencited since then as a warrant for thispractice (see Willig, 2001 for a recentpsychological example). Poland (2001)makes much the same kind of argument.Researchers against using fuller transcriptmay build the case that for the analysis ofthe broader ideological content of talk,where the key thing is the words, cate-gories and repertoires used, the represen-tation of features of speech delivery willonly get in the way. The ‘minutiae’ ofconversation will distract from the‘broader’ ideological organization of thetalk. There is some power to this observa-tion. However, we find the alternativeargument more compelling. This is thatthe analysis of broader patterns and ideo-logical talk should be able to deal with thespecifics of what is going on in the talkrather than simply a reconstructed, sim-

Table 2 Conversational features missed from the orthographic transcript

Feature Example

Emphasis (underlining) stressedClosing (full stop) intonation stressed.Continuing (comma) intonation Situations,Overlaps lined up with square brackets [ and ]

[YEah. ]Pause lengths timed in seconds (0.4)Elevated pitch !/wellElevated volume (capitals) YEAHLowered volume (enclosed in 8 8) 8kids8Elongated sound (colons) Er::mSpeeded up (enclosed in !/ B/) !/not always in theB/

One turn ‘latches’ onto another with Er::m:"/

no discernible pause "/Mm.Outbreath hhLaughter particle huh



UNCORRECTED PROOF

plified and distorted version of it. Notonly should it be able to deal with this,but it also will be most effective andpersuasive if it does deal with it. Mostpertinently for those with a focus onideological issues, the full Jeffersonianrepresentation of talk makes most apparentthe jointly constructed, socially engagednature of what is going on, including theclose dependence of what the intervieweesays on the interviewer’s question (andvice versa) in all its specifics.

Note that this is not just an issue for theresearchers conducting the analysis andtheir own theoretical and analytic procliv-ities and allegiances. Insofar as the evalua-tion of the work is a communal endeavourfor journal referees and readers there is astrong argument that the researchersshould provided a form of transcriptionof talk that will allow readers to make afull evaluation rather than one that mayalready embed their own theoretical as-sumptions within it. To some extent de-velopments in technology and the webdistribution of materials will make theseproblems less acute. As we have noted,the interaction we are working with isavailable on the web. But there is stillwork to do.

None of this is to underestimate theeffort involved here. While a professionalservice might be able to produce play-script interview transcript at a time ratioof 4!/6 hours of work per hour of inter-view, even a skilled Jeffersonian transcri-ber is unlikely to better a ratio of 20 to 1.And this will go down with recordingquality, quiet speakers, language and ac-cent complexities and so on. If the re-searcher’s overall time for a studystays constant they will need to makesacrifices somewhere else, probably insample size.

Global Observations

This point refers to the way in whichanalytic observations, of whatever kind,are linked to the interview transcript. Simi-lar points are made in Antaki et al ., (2003)with respect to the ways researchers canunderanalyse materials. Our observationsare designed to compliment theirs. Herethe issue is how an observation is madeexplicit or how a claim about the interviewis substantiated. In the conventional ortho-graphic representation common in inter-view research is it often not clear whatspecific elements of the talk are beingreferred to. This is partly because thisform of transcript collapses together (poten-tially) large numbers of different sorts ofconversational elements.

For example, if we consider Extract 1 wecan see that a range of different elements ofthe interviewee’s answer, which is con-structed interactionally with the inter-viewer, are collapsed together (with theinterviewer contributions absent). The se-parate lines for the interviewer and inter-viewee in the Extract 3 version allow theturn organization of the interview to beclarified and clearer reference to be madeto each. The line numbers allow a furtherspecificity of reference to be achieved. It isnot uncommon in contemporary qualitativeinterview papers in psychology to find alarge block of text reproducing interviewee’stalk, with some observations made about itthat are very hard to clearly link to specificelements of the talk.

Looked at another way, the challenge inanalysis is to show how your claims canaccount for the specifics of the talk, not justits broad themes. The block-of-text form ofrepresentation does not allow those speci-fics and their relation to the analysis to beclearly seen.



UNCORRECTED PROOF

The unavailability of the interviewset-up

The set-up of the interview is potentiallycritical in at least two, and most likelymore ways. First, what category have theparticipants been recruited under? Arethey taking part in the research on thebasis that they are a ‘lesbian mother’, an‘adolescent male’, a ‘recreational druguser’, or something less explicit? Interviewresearch typically recruits participants un-der categories of this kind. After all, this isa feature of proper attention to sampling.How are these categories constructed inthe various parts of the recruitment (in-cluding the introduction to the research,ethics procedures, administrative arrange-ments, and so on)? Second, what is thetask understanding offered to the partici-pant? This involves questions such as:what are they told that the interview willbe about , what it will be for, and what thetask of the interviewee will be?

These are complicated issues. Not onlyare full records of these things not kept,but there are (good) ethical reasons for notcollecting such records before full consentprocedures have been fulfilled. Moreover,it is hard to do adequately informativeconsent procedures before a range of im-portant details about the task and thereasons for recruitment have been de-scribed. Nevertheless, given the impor-tance of these (and other) features someattempt at recording and representingwhat went on seems to be required. Forexample, textual materials relating torecruitment could be included in thewrite up and some attempt could bemade to at least gloss the interactionbetween the participant and researcherprior to the interview. As far as possible,when the actual interview takes place the

recording could be started as early aspossible, so that the researchers gloss onthe interview and its goals is captured ontape and can be reproduced in materials.

This is not an issue that has receivedmuch discussion in the past. The pointhere is to signal both its neglect andimportance and to start, tentatively, toindicate some ways in which it could beattended to.

The failure to consider interviews asinteraction

These first four problems amount, takentogether, to a failure to treat interviews asan interaction. This is ironic given theagreement, within those approaches topsychology using qualitative interviewsthat appreciating their interactional natureis essential for their analysis (Gubriumand Holstein, 2001). The problem is infollowing the implications of this insightthrough into the design, practice, andrepresentation of interviews. The pointhere is that the recognition that interviewsare interactional has consequences forinterview research. For the points wehave made above the consequences areall addressable in a relatively straightfor-ward manner (although, as we have noted,there are principled arguments to be madewith some of our suggestions). That iswhy we have called these contingentpoints !/ they are problems that are moreor less fixable by changing the way inter-view talk is reported and represented interms of the form of the transcript andwhat sections of interviews are extractedand by including further informationabout the interaction that went on withthe participants as they were inductedinto the study. Their contingency is ap-



UNCORRECTED PROOF

parent by noting that the best interviewstudies do indeed concur with these fea-tures.

Our proposal is that research that isreporting interviews should include at leastthe following elements. (1) At minimum itshould include the relevant interview ques-tion(s). This is probably the topic initialquestion as well as any follow-ups or‘prompts’. This in itself might not be suffi-cient, but will allow at least an initialconsideration of the relation between thequestion construction and the answer. (2)The interview extracts should be tran-scribed to a level that allows interactionalfeatures to be appreciated even if interac-tional features are not the topic of the study.That is, they should be transcribed in morethan Jefferson Lite (or equivalent). (3) Theinterview extracts should be presented insuch a way, probably using line numbersand short lines that allow discrete connec-tions to be made between elements of talkand analytic interpretations. (4) The reportshould include information about howparticipants were approached, under whatcategories, with what interview tasks. Somecurrent interview studies include some ofthese elements; few include them all.Further research and argument in thisarea might well suggest more features forinclusion.

PART 2: NECESSARY PROBLEMS

So far we have discussed problems withinterviews for which there is a relativelystraightforward practical solution. For thesecond half of the paper we will identifysome problems that are less easily dealtwith. They represent a set of problems thatcan be highlighted by considering interviewinteraction from a conversation analytic and

discursive psychological perspective. Theseare problems that generate interpretationaldifficulties in the analysis of interview talk.And a further problem is that they areadditive or even multiplicative. That is,each can generate uncertainty about how itrelates to the others. That is not to say thatsensitive and skilled analysis cannot de-velop important claims (from a range ofperspectives) based on interviews. But ithighlights some of the elements that suchanalysis will need to grapple with. Pre-viously this has been done more of lessimplicitly !/ in future it may need to bemore explicit.

We will focus on four issues: (1) theflooding of the interview with social scienceagendas and categories; (2) the complex andvarying footing positions of interviewer andinterviewee; (3) the possible stake andinterest of interviewer and interviewee; (4)a drag toward cognitive and individualexplanations. We will take them in turn.

Issue 1: Flooding the interview with asocial science agenda and categories

By social science (or psychological)agenda we are referring loosely to the setof concerns and orientations that arecentral for the researcher. Some of thesemay be quite explicit and reflect thespecific research questions that are a focusof the research. Others may be extremelyinexplicit and reflect the disciplinary em-beddedness of the research enterprise.This might include the factors and vari-ables approach characteristic of modernempirical psychology, the various theore-tical frames that interview researchersuse, the assumptions about what a personcan know about her or his own practice



UNCORRECTED PROOF

and so on. The list is a potentially longone.

The social science agenda is bound upwith (although goes beyond) the categoriesthat are used. These include the variousmore or less technical terms and descrip-tions that appear in interview questionsand interview responses. This is a rathercomplicated topic as it may be quitehard to judge what terms are social scien-tific and what are not. They are rarelylikely to be as explicit as the followingexample:

Please describe a time in your life when youexperienced internalized homophobia (Giorgiand Giorgi, 2003: 263).

However, as social representations research-ers, Foucaultians and others have argued,everyday talk can involve a range of ‘sedi-mentations’ of earlier ‘theoretical’ notionsas the languages of psychoanalysis, Marx-ism, symbolic interactionism and so onbecome parts of people’s everyday conver-sational currency.

Let us consider the interview questionfrom Extract 3 again.

Extract 4


11 on teachers’ ti:me and the resources !/that are




Let us start with a very basic observation.Although the question may appear rathercasually developed, with its hesitations

(e.g., the delay and ‘er’ on line 12) andtrailing off (the quiet on 14 and the inaud-ible elements on 15) it is nevertheless arecognisable type of social research ques-tion. Indeed, it is arguable that it is these‘casual’ elements that, in part, constitute itssocial science features. For example, Puchtaand Potter (2004) identify such features inmarket research focus groups as proceduresfor generating informality and managingpotential epistemic asymmetries betweeninterviewer and interviewee. And theynote that uncompleted listings in multiplechoice questions such as this can bothprovide candidate answers to model whatthe researcher expects and show that thelisting offered does not exhaust answeringpossibilities. The point we are making hereis that the question construction with itsinformality and candidate answer carefully(and appropriately) coaches the participantin the relevant social science agenda.

Although this question does not involveobviously recognisable technical terms(such as ‘internalized homophobia’) wecan note the way that the question isconstructed in terms of abstract processesand structures (‘teachers’ time’ line 11,‘constraints’ lines 10 and 12) and genericcategories (‘teachers’ line 11, rather than, forexample, specific teachers at specificschools). Looked at another way, this ques-tion is part of an abstract approach to socialprocesses rather different from, say, a se-quence of staff room troubles telling talk.

Note also that the social science agenda isnot only developed in questions but also invarious other types of interviewer turn. Takethe following extract.

Extract 5

27 Teacher: Because they are near (.) the edge

28 themse[lves,]



UNCORRECTED PROOFNote the difference between the acknowl-edgement tokens in 29 and 31 and the newsreceipt/agreement combination in 35. Theelements of the teacher’s talk in 27!/28 and32!/34 are being receipted rather differently.Such differences could be consequential,and could reflect the researcher’s agenda indifferent ways.

Issues about social science agenda andcategories are difficult and hard to disentan-gle. What is social science and what is not,what are interviewer actions and what arenot, are hard questions to answer. However,this is not to claim that they are unimpor-tant. At its most basic these issues face uswith the possibility that a piece of interviewresearch is chasing its own tail, offering upits own agendas and categories and gettingthose same agendas and categories back ina refined or filtered or inverted form.

Issue 2: Interviewer’s and Interviewee’sFooting

The second necessary issue involves thefooting or speaking position of the inter-viewer and interviewee. The notion of foot-ing was introduced by Goffman (1981) tocharacterize conversational practices suchas the current speaker reporting another’s

speech. He makes a contrast between theanimator, the current speaker who is doingthe talking, and the composer, the personwho originally made up the words. And henotes that a further distinction is neededbetween the composer and origin of theviewpoint; for example, a political speech-writer may write words to express ideas fora leader. At the same time Goffman distin-guished a range of different reception roles:e.g., addressed versus unaddressed recipi-ent, overhearer versus eavesdropper.

We can extend these footing categories toconsider the different basis on which parti-cipants are speaking. For example, are theyspeaking as individuals or category mem-bers? And if a category member, what is therelevant category? Looked at another way,are they speaking as individuals with aninstitutional identity or as persons withtheir own unique and idiosyncratic prefer-ences? This brings the apparatus of member-ship categorization analysis to bear oninterviews (Baker, 2001, 2004). And asPomerantz and Zempel (2003) show, otherkinds of contextualizing devices are alsoavailable in interviews.

Let us first consider the talk of theinterviewee. Note again the question she isasked, and focus this time on the categoriesthat are used.

29 Int: [Yeah.]

30 (0.5)

31 Int: 8Yeah.832 Teacher: Er::m (0.9) if you’re ti:red, (0.1) an stressed,



35 Int: [!/Oh ye]ah. 8yeah.8

Extract 6


11 on teachers’ ti:me and the resources !/that are



UNCORRECTED PROOFNote the way that the interviewee isaddressed with a mix of direct personalterms (‘you’, ‘your job’ lines 12 and 13),but also as a category member (‘teachers’time’ line 11). This interviewee has beenrecruited as a category member (a teacher)and is being addressed as such. Herspeaking position then for this interactionis as a teacher. Yet her ‘personal’ feelingsare being asked about. And as we havealready noted, in the transcript and re-search publication the interviewee is ren-dered using the category Teacher (seeBillig, 1999; Watson, 2004, on this issue).The interviewee could exploit the contin-gency of answering questions to develop adistinction between constraints on her andon teachers generally. She is not forced toaccept the terms of the question. However,this would involve a bit of work, as such adistinction is not projected by the ques-tion. She would have to roll back themultiple choice listing in lines 12!/15 todo that. This could be just the kind ofproblem that generates the trouble at thestart of the reply in line 16 (note theuncertainty marker, the long delay, and theswallow) although there are other plausi-ble candidates.

Of course, one way to look at this is thatthe required psychological research taskhas been achieved !/ the participant isanswering as a representative of the cate-gory that they have been recruited under.The problem, however, is that the precisecategory that the interviewee is speakingfrom can be quite hard to identify con-

fidently; indeed, it may be profoundlyambiguous.

Now consider the footing of the Inter-viewer. Are they the addressed reci-pient? Or are they conduit to some otherrecipient? For example, if this was atelevision news interview the interviewerand interviewee might both treat theoverhearing audience as the proper recipi-ent of the talk. This is shown, for exam-ple, by a lack of change of state markers(such as ‘oh’) in news interviewertalk. After all, they may be asking ques-tions they already know the answer to,or are not interested in the answer to.The issue is not the news interviewer’schange of knowledge state but how in-formed the audience has been (Claymanand Heritage, 2002; Heritage and Great-batch, 1991).

The situation in qualitative interviewsis complex !/ the interviewer addressesthe interviewee. But are they who thetalk is designed for? Sometimes the inter-view is set up with the interviewerpresenting as strongly involved. Atother times they may be presenting asneutral. There might be all kinds ofdelicacy in this set-up when interviewingminority groups, or extremist groupssuch as fascists or racists (see Billig,1977). Moreover, it is not just a matterof the overall set-up at the start of theinterview. Different kinds of interviewerturns can display different footing posi-tions. Take the following from our targetexample:




16 Teacher: [ U : : M : ] (1.0) ((swallows)) yes, (0.7)




UNCORRECTED PROOFNote here the interviewer’s news receiptand agreement in line 35 and the extendedagreeing mms on 38 and 40. These turnsshow a different kind of involvement tothat common in news interviews. Theypresent a footing as a full recipient. Theinterviewer presents as not merely a con-duit to the collection of knowledge but anactive participant with their own knowl-edge and views. We can gloss this asactive interviewing. But note that this isnot a consistent feature of even thisextract. Elsewhere the interviewer providesmore sparse acknowledgement tokens (e.g.,Extract 3, lines 19, 22). The footing isvariable.

The general point, then, is the footingfor both interviewer and interviewee ispotentially convoluted and variable. Thereare considerable complexities when ad-dressing footing in interviews (see forexample the debate: Leudar and Antaki,1996a,b; Potter, 1996). We have consideredfooting mainly in relation to the categoriesinterviewee and interviewer; others haveconsidered some of the more fine-grainedfooting shifts in interviewees talk (Ensink,2003; Lee and Roth, 2004; Wilkinson,2000). In general, there is a major chal-lenge here for anyone analysing interviewmaterial.

Issue 3: Interviewer’s and Interviewee’sStake and Interest

One of the basic claims of discursivepsychology is that in their interaction peo-ple orient to issues of stake and interest(Edwards and Potter, 1992; Potter andEdwards, 1990). That is, they may respondto what others say as based on particularinterests, and they may manage issues ofinterest in their own talk. Note that this isnot an attempt by discursive psychologiststo understand what people say in terms ofits interestedness; it is an attempt to take theissue of interestedness for participants as atopic.

In research interviews issues of stakeand interest are both profound and com-plex. Let us take the interviewee to startwith. As we have already noted, intervie-wees are typically recruited as members ofa social category of some kind. There maywell be an expectation that they have astake in that category. Yet this is oftencombined with questions that treat theparticipant as a broadly neutral informanton their own practices. Nevertheless, thesekinds of issues have often been a topic ofanalyses of qualitative interviews. Ques-tions of how interview participants ad-dress and manage issues of stake andinterest has been a thread running through

Extract 7

32 Teacher: Er::m (0.9) if you’re ti:red, (0.1) an stressed,



35 Int: [!/Oh ye]ah. 8yeah.836 Teacher: Er:m (0.4) the CHILdren I think at least are


38 Int: [ M m : : : . ]


40 Int: [M m: :.]



UNCORRECTED PROOF

a wide range of discursive psychologicalstudies involving interviews. Issues of theinterviewer’s stake and interest have beenless addressed.

It is routine, for example, for qualitativeinterviews to be conducted as parts of PhDprojects where the interviewer and theresearcher are combined in one person (afooting issue). It is also common (andappropriate) for PhD researchers to caredeeply about the topics they are studying.The issue for us is not the interestedness ornot of qualitative interviewers, but howissues of stake and interest are managed.One thing that is clearly absent from theinterview analysed above is the sort ofintroductory element that is common inmarket research focus groups for example.This is a quote from near the start of a focusgroup where the moderator is describingwhat is going to happen in the group (seePuchta and Potter, 2004 for further exam-ples and analysis)

Note how much emphasis is placedon the independence of the moderatorfrom the companies involved in produ-cing either the advertising ideas or thecars themselves (lines 1!/3). There is a

very explicit construction of the adverti-sing and car manufacturing interests,of the researchers’ independence fromthese (repeated on line 9), and the rela-tionship of what the participants sayto the researcher’s happiness (lines11!/12).

The kind of separation between theresearcher and the direct concern withthe topic is much less common in aca-demic qualitative interview research. Sosuch a strong emphasis on disinterest isprobably not possible, whether desirableor not. Researchers can and do introduceissues of stake and interest explicitly ininterviews, although our sense is that thishappens more in the interview set-up(where it is often therefore not availablefor considering its analytic implications,as we noted above). How such introduc-tions relate to the trajectory of what goeson is an important, difficult and currentlyunder researched topic.

As we have already noted with theexample of footing, issues such as thisare likely to vary across the interview.Issues of interest and involvement maycome into play in different places in

Extract 8

1 Moderator: As I say I B/!/don’t make!/ the adver¡/tising,2 (0.5) I don’t sell cars."/I don’t work for

3 ei:ther company that doe:s:. .hhh s:o: er::

4 !/whilst (0.3) !/this research has clearly

5 been commissionedB/ by: (0.1) er a

6 !/company that doesB/ both. An !/you’ll see (as we go through) who ‘tis.B/

7 (0.2)

8 !/I don’t have a vested interest.

9 (0.4)

10 8Right,8 !/so I don’t really mind what

11 youB/ say:.



UNCORRECTED PROOF

different ways. Let us just take the exam-ple of the previously quoted fragment.

Note again here the interviewer’s displaysof investment in the topic in lines 35,38 and 40. Agreements (and disagree-ments, of course) can display broaderalignments and interests in topics (cf.Koole, 2003). For the moment we willjust note again that if we take this ser-iously it makes the process of interviewanalysis considerably more complicatedthan it is often presented. Again, ourgeneral point is that there is a profoundcomplexity in interview material that israrely explicitly addressed. To take it intoaccount during analysis is a major chal-lenge (for an example that highlights thesubtlety of the challenge, see Edwards,2003).

Issue 4: Reproduction of Cognitivism

For many interview researchers some kindof cognitive perspective will be entirelyappropriate. Our point is not to directlyquestion this desirability but to note howit can be an interactional product of theway interviews are conducted and a

source of confusion. There are a numberof potential facets here but we will con-

centrate on two: the privileging of concep-tual rumination over action and thetreatment of cognitive language as descrip-tive.

Privileging conceptual ruminationA basic feature of qualitative interviews isto treat the interviewee as a reporter onevents, actions, social processes and struc-tures, and cognitions. The point here isthat this kind of explicit conceptual rumi-nation is treated as providing a way intoparticipants lives or social organizations orwhatever the topic of the research project.This shows up in the way participantsanswer questions. Note the (at least appar-ently) abstract, syllogistic logic of thefollowing:

Extract 10

16 Teacher: [ U : : M : ] (1.0) ((swallows)) yes, (0.7)


And note the way causal relationships areidentified as the participant theorizes as aproto-social scientist.

Extract 9

32 Teacher: Er::m (0.9) if you’re ti:red, (0.1) an stressed,



35 Int: [!/Oh ye]ah. 8yeah.836 Teacher: Er:m (0.4) the CHILdren I think at least are


38 Int: [ M m : : : . ]


40 Int: [M m: :.]



UNCORRECTED PROOF

There is a lot more we could say about howthese descriptions are put together. How-ever, the basic point to note is that here theteacher is being asked as a teacher not to bea teacher but to formulate features of thelives of teachers and what causes them toact in particular ways. Again, this is notdoing much more than restate the basicrationale often assumed when doing inter-views. You ask people about what they doand think and they helpfully tell you aboutit. The point, however, is to highlight pre-cisely the assumption that interviewees canhelpfully tell you about social processes,causal relations and so on.

Of course, the most straightforward ver-sion of this assumption was criticized byearly discourse analytic work (Gilbert andMulkay, 1984; Potter and Mulkay, 1985;Potter and Wetherell, 1987; Wetherell andPotter, 1988). Partly on the basis of varia-bility in interviewee talk, the argument wasmade for changing the focus to the activitiesdone in interview talk and the resourcesused in that talk. A large amount of researchhas been done since this time (of varyingquality). However, interview analysis of thiskind is still challenging. In the first place itoften requires the analyst to make difficultextrapolations from the kinds of actionsdone in the interview talk to the kinds ofactions done elsewhere. In the second placethe activities done in interviews are parti-cularly complicated because of the sorts ofagenda, footing, stake and interest issuesthat we have noted above. Moreover, it islikely that much of this discourse analyticwork using interviews has underestimatedthe pervasiveness of interview identitiesand practices in its analysis, and it has

been supported in this underestimation bya range of the contingent problems noted inthe first section of the paper.

Cognitive Language as DescriptiveLanguageA second feature characteristic of qualita-tive interviews is the treatment of cognitivelanguage. The focus on the practical role ofcognitive terms has been a major topic ofdiscursive psychology (see Edwards, 1997,for overview). More recently some work hasconsidered the role of cognitive terminologyin social research settings. For example,Puchta and Potter (2002, 2004); Myers,2004; Potter and Puchta, forthcoming) haveconsidered the practical role of ‘POBA’terminology; the use of terms such asPerceptions, Opinions, Beliefs and Atti-tudes (the acronym is from Henderson,1991 although Puchta and Potter, 2004,highlight a broader family of such terms).These studies have highlighted both the waythat such notions are related to issues ofaccountability (important for generating an-swers) and how social research interactionis often organized to accomplish POBAs asobjects within individuals.

In our current extract, for example, notethe use of the POBA term ‘feel’ at in initialelement of the topic-initial question.

Extract 12

Int: So d’you !/feel then that the constrai:nts

on teachers’ ti:me and the resources !/that are

In market research focus groups POBAconstructed questions were used to headoff both ‘don’t know’ responses and todiscourage participants from asking the

Extract 11

20 Teacher: Er because they’re stressed (0.2) they may21 react (0.5) u::m inap!/propriately,



UNCORRECTED PROOF

moderator. The interactional logic of POBAsis that people are the best experts on theirown POBAs and they should not be directlyundermined by factual considerations. Itseems likely that something very similar isgoing on in Extract 12. It would be too muchof a digression to elaborate on these issueshere (see Potter and Puchta, forthcoming).Our point is that to fully understand thequalitative interview as an interaction wewill need to pay attention to the practicaland interactional role of cognitive terms andbe very cautious about treating such termsas if they referred to psychological objects ofsome kind within individuals.

Similar sorts of issues arise with theinterviewee’s use of psychological or cogni-tive terminology. To take one example, theinterviewee uses the term ‘stressed’ on threeoccasions in this sequence (lines 17, 20, 32).Whatever referential role this term has,careful analysis will need to consider whatit is being used to do in this sequence. Forexample, Hepburn and Brown (2001) in ananalysis of these interviews highlight someof the practical uses of stress talk in mana-ging accountability and linking individualactions in with broader institutional rolesand relationships. Our point again is that tofully understand what is going on in quali-tative interviews, researchers will need to beattentive to the practical role of psychologi-cal language.

INTERVIEWS !/ PROBLEMS AND PROSPECTS

We will organize this concluding section intwo parts. First we will consider the im-plications of this paper for the use ofqualitative interviews in social research.Second, we will discuss the relative virtuesof work with records of naturalistic interac-tion.

Qualitative Interviews

We started by overviewing a set of contin-gent problems with the design and presen-tation of qualitative interviews: (1) thedeletion of the interviewer; (2) the conven-tions of representation of interaction; (3) thespecificity of observations; (4) the unavail-ability of the interview set-up; (5) the failureto consider interviews as interaction. With-out attempting an extensive survey we haveno doubt that readers can select just aboutany journal that regularly publishes psycho-logical work based on qualitative interviewsand find examples which display some orall of these problems. This is not surprisingas even some of the ‘how-to-do-it’ manualsfor research involving qualitative interviewsin psychology show the same problems.Furthermore, as we noted some researchershave argued that these are not really short-comings. However, our conclusion is thatthese are problems, and that qualitativeresearch in psychology would in generalbe improved by correcting them. Moreover,research opting for alternative practicesshould justify precisely how and why itwould be improved by, say, deleting theinterviewer or using a more playscriptreconstruction of talk.

The second set of problems we consid-ered were ones that are a necessary part ofdoing interviews: (1) the flooding of theinterview with social science agendas andcategories; (2) the complex and varyingfooting positions of interviewer and inter-viewee; (3) the possible stake and interest ofinterviewer and interviewee; (4) the repro-duction of cognitivism. These necessaryproblems are intimately connected to thecontingent ones because they are oftenobscured by the common representationalpractices used in interview research. Ourargument here is that none of these ele-



UNCORRECTED PROOF

ments can be eliminated (issues of agendas,footing, interest, and so on will always berelevant to some extent) rather they presentthe interview researcher with particularlyacute set of analytic problems. Whatever theanalytic perspective, inferring things appro-priately from interviews involves under-standing what is going on in theminteractionally, and that in turn involvesthe complex and demanding task of analys-ing the development of an implicit researchagenda, identifying footing shifts, explicat-ing orientations to stake and so on. Asresearchers with some expertise in interac-tion analysis we would like to emphasizethat this is a challenging analytic require-ment. Such analysis is rarely done with anydegree of seriousness in current interviewresearch, and where it is the analysis oftenhighlights just how much the interviewee’stalk is a product of specific features of theinterview. Widdicombe and Wooffitt (1995)is one of the best exemplifications of this.

Furthermore, these necessary problemshighlight issues for the design and conductof interviews. Interviewer introductions,questions, responses and so on have re-ceived surprisingly little systematic study(Rapley (2001) and some contributions tovan den Berg et al . (2003 are the excep-tions). Such study could start to tease outthe way different kinds of social scienceagendas are established, the way footingshifts are marked and so on. Alternativeways of designing questions, for example,might offer up clearer routes to analysis.Moreover, what precisely is involved in thestrand of work that has emphasized thelimits of traditional ‘neutral interviewing’and has pressed the case for ‘active inter-viewing’ (Gubrium and Holstein, 2004)could be explicated. Closer analysis is likelyto show up these categories are much toosimple and that ‘active interviewing’ is

made up of a range of different elements,often combined with elements that lookmore like ‘neutral interviewing’. One con-clusion, then, is that much more research isneeded into the social and interactionalnature of the research interview itself. De-spite its ubiquity it has remained surpris-ingly under studied.

Our second conclusion is that althoughqualitative interviews are treated as rela-tively easy to perform (students, for exam-ple, often perform open-ended interviewswith almost no training) they are very hardto do well. On top of this they are hard toanalyse and even harder to analyse well (forexample, students, who seemed to have notrouble conducting the interviews them-selves, often report major difficulties inmaking informative or appropriate analyticinferences from them). For these reasons wepropose that interviews are overused inqualitative psychology. At the very least,the set of contingent problems we identifiedshould be attended to. And rather thaninterviews being the default technique thereshould be a more careful weighing up oftheir virtues and limitations, and the precisereasons for their use should be given.Researchers should ask more often andmore seriously: are interviews essential?

One reason often given for the use ofinterview research is that due to the sensi-tivity of the topic it would be impossible todo the research in any other way. Thepracticalities and ethics of access to delicatematerial are complex, of course, however wehave found that such arguments are oftengiven without the researchers having tried toget access. The right kind of approach (withthe appropriate understanding of the risksand work for all the different parties in-volved) can be effective in very delicateareas (see Hepburn and Potter, 2003 andHepburn, forthcoming, for discussions and



UNCORRECTED PROOF

some styles of approach). We will end withsome comments on the use of naturalisticrecords for research as an alternative.

Naturalistic Records

Naturalistic records can include audio andvideo recordings of conversations in every-day or work settings, records ofprofessional!/client interaction, televisionprogrammes, documents such as medicalrecords or personal diaries and so on.Although the definition of naturalistic hasbeen a source of some controversy (see, forexample, Lynch, 2002; Potter, 2002; Speer,2002a,b; ten Have, 2002) the criterion usedhere is that the activity being recordedwould have happened as it would haveanyway; it is not got up by the researcher,for example by way of an open-ended inter-view. The records are dubbed naturalisticrather than natural in recognition of a rangeof potential sources of what would tradi-tionally be called reactivity involved in therecruitment, the recording and so on (fora highly relevant debate on this, seeHammersley, 2003; Speer and Hutchby,2003). Nevertheless, they are generatedwith the aim of avoiding active researcherinvolvement, even if the full realization ofthis ideal is often impossible. Note thatwe are advocating naturalist records as aresearch topic, much like interviews areresearch topics, but not suggesting that allresearchers should use them in discursivepsychology or conversation analysis pro-jects. Such records are potentially a liveand important source of material for all ofthe perspectives listed in Table 1. After all,they are records of people living their liveswhich is one of the main topics of theenterprise of psychology.

One way of considering the value ofnaturalistic records it to assess them inrelation to the four necessary problemsdiscussed above. We can see that naturalis-tic records: (1) avoid flooding the interactionwith psychology and social science agen-das; (2) avoid some of the complex inter-viewer/interviewee footing complexities; (3)have stake and interest tied to particularrelevant practices in the domain understudy; (4) avoid cognitivism by collectingmaterial where participants are not requiredto offer abstract conceptual rumination onsome aspect of their lives. These advantagesare in addition to more familiar advantagesof working with naturalist materials thatthey (1) throw up novel questions andissues; (2) go beyond familiar limits ofmemory, attention and perception that un-derpin peoples’ accounts of their practicesor the organizations in which they work; (3)get representations and ‘cognitions’ in ac-tion; (4) provide resources for appreciatingissues of application (see Potter, 2003, 2004for overview and discussion of thesepoints).

Our identification of problems with qua-litative interviews, then, provides the basisof a more systematic account of the virtuesof working with naturalistic records. Suchmaterial is not the most appropriate in allcases and qualitative interviews are stilllikely to be the technique of choice for arange of research issues. Nevertheless, wehave tried to lay out some considerationsthat would at least start to question theperhaps surprising and perhaps unsuppor-table dominance of the qualitative interviewin qualitative research in psychology.

Acknowledgements

We would like to thank audience membersat departmental seminars in the London



UNCORRECTED PROOF

School of Economics, February 2004, andthe University of Rome, La Sapienza, July2004, for helpful comments on an earlierversion of this paper. The original idea arosein a conversation with Sandra Jovchelovitchand Caroline Howarth. The paper hasgreatly benefited from comments by SusanSpeer and Elizabeth Stokoe.

References

Antaki, C. 1999: Assessing quality of life ofpersons with a learning disability: howsetting lower standards may inflate well-being scores. Qualitative Health Research9, 437!/54.

Antaki, C. and Rapley, M. 1996: ‘Quality of life’talk: the liberal paradox of psychologicaltesting. Discourse and Society 7, 293!/316.

Antaki, C., Houtkoop-Steenstra, H. and Rapley,M. 2000: ‘Brilliant. Next question...’: high-grade assessment sequences in the comple-tion of interactional units. Research onLanguage and Social Interaction 33, 235!/

62.

Antaki, C., Billig, M., Edwards, D. and Potter, J.2003: Discourse analysis means doing ana-lysis: a critique of six analytic shortcomings.Discourse Analysis Online 1, [http://www.shu.ac.uk/daol/previous/v1/n1/index.htm]. This is reprinted as Antaki, C., Billig,M., Edwards, D. and Potter, J. 2003: ElAnalisis del discurso implaca analizar: Cri-tica de seis atajos analiticos. Athenea Digital3, [http://antalya.uab.es/athenea/num3/an-taki.pdf]

Baker, C. 2001: Ethnomethodological analyses ofinterviews. In Gubrium, J.F. and Holstein,J.A., editors, Handbook of interview re-search: context and method . London: Sage.

Baker, C. 2004: Membership categorization andinterview accounts. In Silverman, D., editor,Qualitative research: theory, method andpractice , second edition. London: Sage.

Billig, M. 1977: The new social psychology andfascism. European Journal of Social Psychol-ogy 7, 393!/432.

Billig, M. 1999: Whose terms? Whose ordinari-ness? Rhetoric and ideology in conversationanalysis. Discourse and Society 10, 543!/58.

Bowers, J.M. 1988: Review essay: discourseanalysis and social psychology. British Jour-nal of Social Psychology 27, 185!/192.

Camic, P.M., Rhodes, J.E. and Yardley, L., editors,2003, Qualitative research in psychology:expanding perspectives in methodology anddesign . Washington: American Psychologi-cal Association.

Cicourel, A.V. 1964:Method and measurement insociology. New York: Free Press.

Cicourel, A.V. 1974: Theory and method in astudy of argentine fertility. New York: JohnWiley & Sons Ltd.

Clayman, S.E. and Heritage, J. 2002: The NewsInterview: journalists and public figures onthe air. Cambridge: Cambridge UniversityPress.

Edwards, D. 1997: Discourse and cognition .London: Sage.

Edwards, D. 2003: Analysing racial discourse:the discursive psychology of mind-worldrelationships. In van den Berg, H., Wetherell,M. and Houtkoop-Steenstra, H., editors,Analysing race talk: multidisciplinary ap-proaches to the interview. Cambridge: Cam-bridge University Press.

Edwards, D. and Potter, J. 1992: Discursivepsychology. London: Sage.

Ensink, T. 2003: The frame analysis of re-search interviews: social categorization andfooting in interview discourse. In van denBerg, H., Wetherell, M. and Houtkoop-Steen-stra, H., editors, Analyzing race talk: multi-disciplinary approaches to the interview.Cambridge: Cambridge University Press,156!/177.

Fontana, A. and Frey, J.H. 2000: The interview:from structured questions to negotiated text.In Denzin, N.K. and Lincoln, Y.S., editors,



UNCORRECTED PROOF

Handbook of qualitative research , secondedition. London: Sage.

Garfinkel, H. 1967: Studies in ethnomethodol-ogy. Englewood Cliffs, NJ: Prentice-Hall.

Gilbert, G. N. and Mulkay, M. 1984: OpeningPandora’s box: a sociological analysis ofscientists’ discourse . Cambridge: CambridgeUniversity Press.

Giorgi, A.P. and Giorgi, B.M. 2003: The descrip-tive phenomenological method. In Camic,P.M., Rhodes, J.E. and Yardley, L., editors,Qualitative research in psychology: expand-ing perspectives in methodology and design .Washington: American Psychological Asso-ciation.

Goffman, E. 1981: Forms of talk . Oxford: BasilBlackwell.

Gubrium, J and Holstein, A. 1997: Active inter-viewing. In Silverman, D., editor, Qualitativeresearch: theory, method and practice , sec-ond edition. London: Sage.

Gubrium, J and Holstein, A. 2001: From theindividual interview to the interview society.In Gubrium, J.F. and Holstein, J.A., editors,Handbook of interview research: context andmethod . London: Sage.

Hammersley, M. 2003: ‘Analytics’ are no sub-stitute for methodology: a response to Speerand Hutchby. Sociology 37, 339!/51.

Henderson, N. 1991: The art of moderating: ablend of basic skills and qualities. Quirk’sMarketing Research Review 18, 19!/39.

Hepburn, A. 1995: Deconstructing secondaryschool bullying: a postmodern discourseanalysis. Unpublished PhD, Glasgow Cale-donian University.

Hepburn, A. 1997a: Teachers and secondaryschool bullying: a postmodern discourseanalysis. Discourse and Society 8, 27!/48.

Hepburn, A. 1997b: Discursive strategies inbullying talk. Education and Society 15,13!/31.

Hepburn, A. 2000: Power lines: Derrida, discur-sive psychology and the management of

accusations of school bullying. British Jour-nal of Social Psychology 39, 605!/28.

Hepburn, A. forthcoming: Getting closer at adistance: theory and the contingencies ofpractice, In Gergen, K., editor, Theory inaction, special issue of Theory & Psychology.

Hepburn, A. and Brown, S.J. 2001: Teacher stressand the management of accountability. Hu-man Relations 546, 531!/55.

Hepburn, A. and Potter, J. 2003: Discourseanalytic practice. In Seale, C., Silverman,D., Gubrium J. and Gobo, G., editors,Qualitative research practice . London: Sage.

Heritage, J.C. 1984: A change-of-state token andaspects of its sequential placement. In At-kinson, J.M. and Heritage, J.C., editors,Structures of social action: studies in con-versation analysis . Cambridge: CambridgeUniversity Press.

Heritage, J.C. and Greatbatch, D.L. 1991: On theinstitutional character of institutional talk:the case of news interviews. In Boden, D.and Zimmerman, D.H., editors, Talk andsocial structure: studies in ethnomethodol-ogy and conversation analysis . Oxford: Po-lity Press.

Houtkoop-Steenstra, H. 1995: Meeting both ends:standardization and recipient design in tele-phone survey interviews. In: ten Have, P. andPsathas, G., editors, Situated order: studiesin the social organization of talk and embo-died activities . Washington, D.C.: UniversityPress of America.

Houtkoop-Steenstra, H. 1996: Probing behavior ofinterviewers in the standardized semi-openresearch interview. Quality & Quantity 30,205!/30.

Houtkoop-Steenstra, H. 1997: Being friendly insurvey interviews. Journal of Pragmatics 28,591!/623.

Hutchby, I., and Wooffitt, R. 1998: Conversationanalysis: principles, practices and applica-tions . Cambridge: Polity Press.

Jefferson, G. 1985: An exercise in the transcrip-tion and analysis of laughter. In Van Dijk, T.,



UNCORRECTED PROOF

editor, Handbook of discourse analysis, Vo-lume 3 . London: Academic Press.

Koole, T. 2003: Affiliation and detachment ininterviewer answer receipts. In van denBerg, H., Wetherell, M. and Houtkoop-Steenstra, H., editors, Analyzing race talk:multidisciplinary approaches to the inter-view. Cambridge: Cambridge UniversityPress.

Kvale, S. 1996: InterViews: an introduction toqualitative research interviewing . London:Sage.

Lee, Y-J. and Roth, W-M. 2004: January: Making ascientist: discursive ‘doing’ of identity andself-presentation during research interviews.Forum: qualitative social research [On-lineJournal] 51, Art. 12. Available at: http://www.qualitative-research.net/fqs-texte/1-04/1-04leeroth-e.htm

Leudar, I. and Antaki, C. 1996a: Discourseparticipation, reported speech and researchpractices in social psychology. Theory &Psychology 6, 5!/29.

Leudar, I. and Antaki, C. 1996b: Backing footing.Theory & Psychology 6, 41!/46.

Lynch, M. 2002: From naturally occurring data tonaturally organized ordinary activities: com-ment on Speer. Discourse studies 4, 531!/37.

Marlaire, C. and Maynard, D. 1990: Standardizedtesting as an interactional phenomenon.Sociology of Education 63, 83!/101.

Maynard, D.W. and Schaeffer, N.C. 1997: Keep-ing the gate: declinations of the request toparticipate in a telephone survey interview.Sociological Methods & Research 26, 34!/79.

Maynard, D.W. and Schaeffer, N.C. 2000: Towarda sociology of social scientific knowledge:survey research and ethnomethodology’sasymmetric alternates. Social Studies ofScience 30, 323!/70.

Mischler, E. 1986: Research interviewing: contextand narrative . Harvard.

Myers, G. 1998: Displaying opinions: t opics anddisagreement in focus groups. Language inSociety 27, 85!/111.

Myers, G. 2004: Matters of opinion: talking aboutpublic ideas . Cambridge: Cambridge Univer-sity Press.

Myers, G. and Macnaghten, P. 1999: Can focusgroups be analysed as talk? In Barbour, R.S.and Kitzinger, J., editors, Developing focusgroup research . London: Sage.

Poland, B.D. 2001: Transcription quality. InGubrium, J.F. and Holstein, J.A., editors,Handbook of interview research: contextand method . London: Sage.

Pomerantz, A.M. 1984: Agreeing and disagreeingwith assessments: some features of pre-ferred/dispreferred turn shapes. In Atkin-son, J.M. and Heritage, J.C., editors,Structures of social action: studies in con-versation analysis . Cambridge: CambridgeUniversity Press, 57!/101.

Pomerantz, A.M. and Zemple, A. 2003: Perspec-tives and frameworks in interviewers’queries. In van den Berg, H., Wetherell M.and Houtkoop-Steenstra, H., editors, Analys-ing race talk: multidisciplinary approachesto the interview. Cambridge: Cambridge Uni-versity Press.

Potter, J. 1996: Right and wrong footing. Theoryand Psychology 6, 31!/9.

Potter, J. 2002: Two kinds of natural. DiscourseStudies 4, 539!/42.

Potter, J. 2003: Discourse analysis and discursivepsychology. In Camic, P.M., Rhodes, J.E. andYardley, L., editors, Qualitative research inpsychology: expanding perspectives in meth-odology and design . Washington: AmericanPsychological Association, 73!/94.

Potter, J. 2004: Discourse analysis. In Hardy, M.and Bryman, A., editors, Handbook of dataanalysis . London: Sage.

Potter, J. and Edwards, D. 1990: Nigel Lawson’stent: discourse analysis, attribution theory,and the social psychology of fact. EuropeanJournal of Social Psychology 20, 24!/40.

Potter, J. and Edwards, D. 2001: Discursive socialpsychology. In Robinson, W. P. and Giles, H.,editors, The new handbook of language and



UNCORRECTED PROOF

social psychology. London: John Wiley &Sons Ltd.

Potter, J. and Mulkay, M. 1985: Scientists’ inter-view talk: interviews as a technique forrevealing participants’ interpretative prac-tices. In Brenner, M., Brown, J. and Canter,D., editors, The research interview: uses andapproaches . London: Academic Press.

Potter, J. and Puchta, C. forthcoming: The powerof the POBA!/Psychological language andsocial research practice. In Hepburn, A. andWiggins, S., editors, Discursive research inpractice: new approaches to psychology andinteraction . Cambridge: Cambridge Univer-sity Press.

Potter, J. and Wetherell, M. 1987: Discourse andSocial Psychology: beyond Attitudes andBehaviour. London: Sage.

Potter, J. and Wetherell, M. 1995: Discourseanalysis. In Smith, J., Harre, R. and vanLangenhove, L., editors, Rethinking methodsin psychology. London: Sage.

Puchta, C. and Potter, J. 1999: Asking elaboratequestions: focus groups and the managementof spontaneity. Journal of Sociolinguistics 3,314!/35.

Puchta, C. and Potter, J. 2002: Manufacturingindividual opinions: market research focusgroups and the discursive psychology ofattitudes. British Journal of Social Psychol-ogy 41, 345!/63.

Puchta, C. and Potter, J. 2004: Focus grouppractice . London: Sage.

Puchta, C., Potter, J. and Wolff, F. in press: Repeatreceipts: a device for generating visible datain market research focus groups. QualitativeResearch .

Rapley, T. 2001: The artfulness of open-endedinterviewing: some considerations on ana-lysing interviews. Qualitative Research 1,303!/24.

Rapley, M. and Antaki, C. 1996: A conversationanalysis of the ‘acquiescence’ of people withlearning difficulties. Journal of Communityand Applied Social Psychology 6, 207!/27.

Sacks, H. 1992: Lectures on conversation . Vols. Iand II, In Jefferson, G., editor, Oxford: BasilBlackwell.

Schegloff, E.A. 1999: Discourse, pragmatics, con-versation, analysis. Discourse Studies 1,405!/36.

Silverman, D. 2001: Interpreting qualitative data:methods for analysing talk, text and inter-action . London: Sage.

Smith, J.A., editor, 2003: Qualitative psychology:a practical guide to research methods. Lon-don: Sage.

Speer, S.A. 2002a: ‘Natural’ and ‘contrived’ data:a sustainable distinction? Discourse Studies4, 511!/25.

Speer, S.A. 2002b: Transcending the ‘natural/contrived’ distinction: a rejoinder to tenHave, Lynch and Potter. Discourse Studies4, 543!/48.

Speer, S.A. and Hutchby, I. 2003: From ethics toanalytics: aspects of participants orienta-tions to the presence and relevance ofrecording devices’. Sociology 37, 315!/37.

Suchman, L. and Jordan, B. 1990: Interactionaltroubles in face-to-face survey interviews.Journal of the American Statistical Associa-tion 85, 232!/41.

ten Have, P. 1998: Doing conversation analysis: apractical guide . London: Sage.

ten Have, P. 2002: Ontology or methodology?Comments on Speer’s ‘natural’ and ‘con-trived’ data: a sustainable distinction? Dis-course Studies 4, 527!/30.

ten Have, P. 2004: Understanding qualitativeresearch and ethnomethodology. London:Sage.

van den Berg, H, Wetherell, M. and Houtkoop-Steenstra, H., editors, 2003: Analyzing racetalk: multidisciplinary approaches to theinterview. Cambridge: Cambridge UniversityPress.

Warren, C.A.B. 2001: Qualitative interviewing. InGubrium, J.F. and Holstein, J.A., editors,Handbook of interview research: contextand method . London: Sage.



UNCORRECTED PROOF

Watson, D.R. 2004: Ethnomethodology and tex-tual analysis. In Silverman, D., editor,Qualitative Research: theory, method andpractice , second edition. London: Sage.

Wetherell, M. and Potter, J. 1988: Discourseanalysis and the identification of interpreta-tive repertoires. In Antaki, C., editor, Analys-ing lay explanation: a case book . London:Sage.

Widdicombe, S. and Wooffitt, R. 1995: Thelanguage of youth subcultures: social iden-

tity in action . Hemel Hempstead, UK: Har-vester/Wheatsheaf.

Wilkinson, S. 2000: Women with breast cancertalking causes: Comparing content, biogra-phical and discursive analyses. Feminism &Psychology 10, 431!/60.

Willig, C. 2001: Introducing qualitative researchin psychology: adventures in theory andmethod . Buckingham: Open UniversityPress.

About the authorsJONATHAN POTTER has written on constructionism, discourse analysis, cognitivism,psychology and institutions, child protection, quantification rhetoric, riots, literature,eating, reflexivity, relativism, racism, science, method and reality. He is one of thefounders of Discursive Psychology and is Professor of Discourse Analysis atLoughborough University.ALEXA HEPBURN is a Senior Lecturer in Social Psychology at Loughborough University.She has published on constructions of authority, control and bullying in schoolsituations. Her research has also developed a critical perspective on more traditionalforms of psychology, reflected in her book An Introduction to Critical Social Psychology(Sage, 2003). Her most recent work has focused on emotion in interaction on the NSPCCChild Protection Helpline, and in particular crying and its reception.



UNCORRECTED PROOF

Appendix: Transcription Symbols

[ ] Square brackets mark the start and endof overlapping speech. Position themin alignment where the overlap occurs,as shown below.

!/¡/ Vertical arrows precede marked pitchmovement, over and above normalrhythms of speech. They are formarked, hearably significant shifts */

and even then, the other symbols (fullstops, commas, question marks) mopup most of that. Like with all thesesymbols, the aim is to capture inter-actionally significant features, hearableas such to an ordinary listener */

especially deviations from a commonsense notion of ‘neutral’ which admit-tedly has not been well defined.

0/ Side arrows are not transcription fea-tures, but draw analytic attention toparticular lines of text. Usually posi-tioned to the left of the line.

Underlining signals vocal emphasis; the extent ofunderlining within individual wordslocates emphasis, but also indicateshow heavy it is.

CAPITALS mark speech that is obviously louderthan surrounding speech (often occurswhen speakers are hearably competingfor the floor, raised volume rather thandoing contrastive emphasis).

8!/I know it,8 ‘degree’ signs enclose obviously quieterspeech (i.e., hearably produced */ asquieter, not just someone distant).

that’s r*ight. Asterisks precede a ‘squeaky’ vocaldelivery.

(0.4) Numbers in round brackets measurepauses in seconds (in this case, 4 tenthsof a second). Place on new line if notassigned to a speaker (i.e after a TRP).

(.) A micropause, hearable but too short tomeasure.

((text)) Additional comments from the transcri-ber, e.g., context or intonation.

she wa::nted Colons show degrees of elongation ofthe prior sound; the more colons, themore elongation. I use one per syllable-length.

hhh Aspiration (out-breaths); proportion-ally as for colons.

.hhh Inspiration (in-breaths); proportionallyas for colons.

Yeh, ‘Continuation’ marker, speaker has notfinished; marked by fall!/rise or weakrising intonation, as when enunciatinglists.

y’know? Question marks signal stronger, ‘ques-tioning’ intonation, irrespective ofgrammar.

Yeh. Periods (full stops) mark falling, stop-ping intonation (‘final contour’), irre-spective of grammar, and notnecessarily followed by a pause.

bu-u- hyphens mark a cut-off of the precedingsound.

!/he saidB/ ‘greater than’ and ‘lesser than’ signsenclose speeded-up talk. Sometimesused the other way round for slowertalk.

solid."/"/

We had‘Equals’ signs mark the immediate‘latching’ of successive talk, whetherof one or more speakers, with nointerval. Also used as below (lines 3!/5), where an unbroken turn has beensplit between two lines to accommo-date another speaker on the transcriptpage.

heh heh Voiced laughter. Can have other sym-bols added, such as underlinings, pitchmovement, extra aspiration, etc.

sto(h)p i(h)t Laughter within speech is signalled byh’s in round brackets.

uh um How to spell ‘er’ and ‘erm’ the Jeffersonway. (Can be added to, etc.)



Documents

Qualitative interviews in psychology: problems and possibilities