Doing Qualitative Research Differently

A Psychosocial Approach

Wendy Hollway, Tony Jefferson

• Verlag: SAGE Publications Ltd
• Jahr: 2012
• ISBN:
  9781446290163
• Auflage: 2
• Thema: Psicologia
• Thema: Ricerche e metodologie nella psicologia

Hollway and Jefferson have updated their ground-breaking book for students and researchers looking to do qualitative research differently. The new edition critically reviews many of the assumptions, claims and methods of qualitative research and also acts as a `how to? guide to the method the authors call the Free Association Narrative Interview.

In the new edition, the authors situate their arguments firmly within a tradition of psychosocial research and show how their method has developed over the last decade. The book follows this approach through the phases of empirical research practice. At each stage they use examples from their own research and end with an extended case study which demonstrates the value of their method in producing a psychosocial research subject; that is, one with socially-imbued depth, complexity and biographical uniqueness.

ONE

Introduction: the need to do research differently

What do you, the researcher, assume about a person’s capacity to know, remember and tell about themselves? In this book, we argue and illustrate the proposition that social researchers need to revise their assumptions about the nature of that person – the research subject¹ – and that this revision should change their research practices.

Assume that you have to do a piece of research. As an example, let us suppose that you are interested in the fear of crime. Reading the literature, much of which is survey-based, you have learned that women consistently come out as more fearful than men, but there seem to be no satisfactory explanations for this, especially since young men seem more crime-prone. Moreover, you have women friends who claim to be unafraid of crime and you suspect that some of your male friends are more afraid than they are letting on. From this commonsense starting-point you frame your central research questions, namely: (a) what is the meaning of the finding that women are more fearful of crime than men; and (b) (if this finding is valid) why are they?

You then identify a number of female and male acquaintances (of a range of ages because age differences have also been found to be significant) who are willing to be informants to see if you can shed some light on these questions. This poses the problem of how you can find out relevant information from your informants. Do you just ask them directly the question(s) to which you wish to find an answer? If not, why not? How else would you approach them? Would it be feasible to observe them in relevant situations? If you decide on a face-to-face interview is it best to structure it through a series of questions? What should they cover and how many do you need? In other words, just how are you going to produce data that, when analysed, will help answer your starting questions?

Let us assume that you have constructed a series of questions, designed to explore face to face the fearfulness of your male and female respondents. This is the most common qualitative method used in the social sciences. You interview them. What then? Will you believe everything you are told? If not, how will you distinguish between truth and untruth? Even if you believe everything you are told, will you be satisfied that you have been told everything that is relevant? How would you define this, and how would you know? What do you assume about the effect of people’s motivations and memory on what they tell you? What will you assume about your effect as interviewer on the answers given? Does your sex, race, age and so on make a difference? Will men talk readily about their fear to a woman or a man, or to neither? How do you know?

Beyond questions of the veracity of the data produced are the questions of their significance. How will you analyse your interviewees’ answers to make some overall sense of them, especially when their accounts are littered with contradictions and inconsistencies? If you conclude that Tom is more fearful than he lets on, or that Anna is fearful out of all proportion to the risks she runs, what is informing these judgements? Why might your conclusions be more, or less, reliable than theirs? Can their answers help answer your starting questions, and if so, how? What theories of gender difference and of fear do you apply to your analysis and how?

All these questions faced us when we decided that existing approaches to research into the fear of crime seemed unable to explain satisfactorily the basic finding of women’s greater fear of crime. The finding seemed clear enough as it was endlessly reproduced (see below), but its meaning was puzzling given that women apparently are less likely to be victims of crime outside the home. Perhaps the problem lay with the question generally asked (‘How safe do you feel walking alone in your neighbourhood after dark?’); perhaps with the truthfulness or otherwise of the answers given; perhaps with the range of unexamined meanings such answers contained; perhaps a mixture of all three. In any event, existing, survey-based research into the fear of crime seemed ill suited to answer the very questions it made visible: ‘why’ questions and questions about what particular findings mean.

Survey research will do well enough to find out how many locks people have on their doors, or whether they have installed security lights, and other easily measurable factors. But, with something as complex (and hence unquantifiable) as fear, survey research has not been able to answer the ‘what’, never mind the ‘why’, of a given person’s, or community’s, fear of crime. To use another example that interests us, namely, sexuality, survey research might be able to find out how many sexual partners a person has had in a given period (although actually it might signally fail to do so), but it cannot find out what this means, nor why sexual behaviour of whatever kind is engaged in.

If quantitative survey-based research is not up to addressing ‘what does this mean’ and ‘why’ questions, it does not follow that the other, qualitative, research tradition has ready answers to such questions. In our experience, it has not. Primarily, this is because of the widespread assumptions in the tradition, by ethnographers, participant observers and interviewers alike, that their participants are ‘telling it like it is’, that participants know who they are and what makes them tick – what we might call the ‘transparent self problem’ – and are willing and able to ‘tell’ this to a stranger interviewer – what we might call the ‘transparent account problem’. Neither selves nor accounts are transparent in our view. Treating people’s own accounts as unproblematic flies in the face of what is known about people’s less clear-cut, more confused and contradictory relationship to knowing and telling about themselves. In everyday informal dealings with each other, we do not take each other’s accounts at face value, unless we are totally naïve; we question, disagree, bring in counterexamples, interpret, notice hidden agendas. Research is only a more formalised and systematic way of knowing about people, but in the process it seems to have lost much of the subtlety and complexity that we use, often as a matter of course, in everyday knowing. We need to bring some of this everyday subtlety into the research process.

One of the good reasons for believing what people tell us, as researchers, is a democratic one: who are we to know any better than the participants when it is, after all, their lives? If we are prepared to disagree, modify, select and interpret what they tell us, is this not an example of the kind of power that we, as researchers, have that should be kept in check by being faithful to the voices of those we are researching? Feminists, in their efforts to diminish the power differentials between researcher and researched, have been strong advocates of the principle of giving voice to hitherto voiceless women. But, as Riessman claims, ‘we cannot give voice’ since we ‘do not have direct access to another’s experience. We deal with ambiguous representations of it – talk, text, interaction and interpretation’ (1993: 8). If we wish to do justice to the complexity of our subjects an interpretative approach is unavoidable. It can also be fair, democratic and not patronising, as long as this approach to knowing people through their accounts is applied to the researcher as well as the researched; as long as researchers are not seen as neutral vehicles for representing knowledge in an uncontaminated way (sometimes called ‘God’s eye view’ or ‘the view from nowhere’). In other words, it is legitimate as long as there is no special objective status that excludes us from being theorised as the same kind of subjects as our informants (albeit in a different position from them).

However, the now widespread recognition of the need to interpret accounts has led to the problem of just how to escape the ‘hermeneutical circle’ (Denzin, 1989: 141), the idea that there is no end to the interpretative process. If experiences can only ever be ambiguously represented, is interpreting these various representations, rather than the experiences themselves, the only possible activity for researchers? We think not. We think that, though it is far from transparent, there is a relationship between people’s ambiguous representations and their experiences. This position is sometimes called ‘critical realism’ (Bunge, 1993; Watkins, 1994–5). But, tracking this relationship relies on a particular view of the research subject: one whose inner world is not simply a reflection of the outer world, nor a cognitively driven rational accommodation to it. Rather, we intend to argue for the need to posit research subjects whose inner worlds cannot be understood without knowledge of their experiences in the world, and whose experiences of the world cannot be understood without knowledge of the way in which their inner worlds allow them to experience the outer world. This research subject cannot be known except through another subject; in this case, the researcher. The name we give to such subjects is psychosocial: our justifications for such a starting-point, and the consequent theoretical, methodological and ethical implications for research, constitute the subject matter of our book.

The book’s content

In Chapter 2 we continue to work with the example of the fear of crime, showing how the survey research on which this area of enquiry was initially based made a series of unwarranted methodological and theoretical assumptions. However, we argue that qualitative research into the fear of crime has failed, as it has in other areas, to look critically at the assumptions it shares with survey research about the nature of the research subject on whose accounts knowledge about the fear of crime depends. We use the case examples of Roger and Joyce to illustrate how our understanding of psychosocial subjects can explain differences in the fear of crime in two people who share many social characteristics.

Chapter 3 is about producing data through the interview method. It looks closely at our initial failures in a pilot interview and makes the case, as a result, for moving from question-and-answer-based interviewing to narrative interviewing. However, the theory of the subject implicit in narrative interviewing leaves no space for what we call the ‘defended’ subject (understanding the effects of defences against anxiety on people’s actions and stories about them). The method that we developed to accommodate the psychoanalytic principles of the defended subject is based on eliciting and paying attention to free association. We illustrate how it works in the case of Jane’s opening response in one such interview. In this way we introduce the free association narrative interview method. Qualitative researchers take seriously the need to understand the role of the interviewer in the production and analysis of data and in this chapter we begin a theme that continues throughout the book, namely, the unconscious intersubjective dynamics in the interview relationship, which we explain and illustrate using concepts such as countertransference, recognition and containment.

Following the focus in Chapter 3 on data production, Chapter 4 is all about data analysis. Here, we take some extended and detailed examples from one family we interviewed (Ivy and two of her adult children, Tommy and Kelly) to demonstrate our approach to analysing unstructured qualitative data. In contrast to the widespread tendency in qualitative research to fragment data by using code and retrieve methods, we illustrate a method based on the principle of working with the whole data and paying attention to links and contradictions within that whole. We demonstrate not only the need for theoretically informed interpretation, but how we do it, based on what principles.

Having given a detailed account of methods of data production and analysis, we turn to the implications that this different approach has for some of the key issues in qualitative research, namely ethics and the generalisability of knowledge derived from case-based analyses. In Chapter 5, we take available ethical guidelines for the conduct of social-science research and assess their appropriateness in three cases chosen from our own research. We demonstrate their inadequacy in this new domain and map out the contours of an alternative approach, based on principles of honesty, sympathy and respect.

In Chapter 6 we examine the tension in our research into the fear of crime between the use of case studies and the need for research to be able to generalise its findings. We show the weaknesses involved in coding and clustering our 37 cases for the purposes of generalisation. We use a series of mini-examples to demonstrate how cases with identical codings were not similar once personal meanings were taken into account and how women’s shared risks of sexual assault did not account for their fear of crime. We conclude that generalisations about the fear of crime need to be based on biography as well as demographic factors.

Throughout the book, we conduct our argument using detailed case-study illustrations. In Chapter 7, we demonstrate, through a single, extensive and multi-faceted case study of a 24-year-old burglar, Ron, what can be achieved through the free association narrative interview method, our approach to data analysis and our psychosocial theory of the subject. This is a case study that goes well beyond description and uses and refines theory to provide insights into the importance of Ron’s biography and his inner world in understanding his criminal activities. In addition, we address some central issues in qualitative research concerning the role of memory and the possibility of gaining access to truth through interview-based research.

We end with a short Afterword which outlines the difference that adopting our view of the research subject might make to qualitative research.

Note

---

1  The history of the term ‘subject’ in this – research methods – context may have come full circle. We use ‘subject’ not in the tradition of experimental psychology (where the term is criticised, paradoxically, for objectifying the people taking part in research) but in the philosophical sense. Here ‘subject’ refers to the person and how s/he is theorised.

---

TWO

Researching the fear of crime

In researching any topic, there are two overarching questions that have to be addressed: what is the object of enquiry and how can it be enquired into? In respect of the fear of crime, these translate into ‘What is the fear of crime?’ and ‘How can it be measured?’ For the sake of simplicity, in what follows we shall refer to ‘what’-related questions as theoretical, and ‘how’-related ones as methodological. In research into the fear of crime there have been far more attempts to measure the fear of crime than to ask what it is that is being measured. Although this problem has been recognised by some authors,¹ even these have failed to notice the importance of a further question: namely, who is the assumed subject of research? In this respect, research into the fear of crime is typical. As we intend to show during the course of this chapter, the problems this common oversight poses for research are as grave as the failure to define the fear of crime.

‘How safe do you feel ...?’

It may seem remarkable now that, without defining what the fear of crime was, early researchers in the field, such as those conducting the first British Crime Survey (Hough and Mayhew, 1983), felt able to measure it. They found that women, especially elderly women, are more fearful of crime than men. Because this finding was ‘discovered with monotonous regularity’ (Gilchrist et al., 1998), the fearful old lady, afraid to venture out after dark, has become a common stereotype, as the authors of the 1996 British Crime Survey came to bemoan (Mirrlees-Black et al., 1996: 55). Yet, when we remind ourselves of the original source of this knowledge, we find it stems from the answers of large national samples to the following question: ‘How safe do you feel walking alone in this area after dark?’ Moreover, the answer was required to fit into one of four categories: ‘very safe’, ‘fairly safe’, ‘a bit unsafe’ or ‘very unsafe’.

Survey research interviews of this kind, where answers can be quantified on a Likert scale, are so prevalent that their capacity to produce evidence is taken for granted. Mishler’s extensive consideration of research interviewing concluded that the ‘standard approach to interviewing [the survey interview] is demonstrably inappropriate for and inadequate to the study of the central questions in the social and behavioural sciences’ (1986: ix). The main reason for this is because the approach fails to address the way in which respondents’ meanings are related to circumstances. Reliance on coding isolated responses strips any remaining context from these responses:

The problem raised by so radical a decontextualization of the interview at so many different levels ... is that respondents’ answers are disconnected from essential sociocultural grounds of meaning. Each answer is a fragment removed from both its setting in the organized discourse of the interview and from the life setting of the respondent. (Mishler, 1986: 23)

Of course, these responses, duly coded, have to be reassembled so as to make sense of them. However, ‘when these [fragmented] responses are assembled into different subgroups by age, gender and the like, the results are artificial aggregates that have no direct representation in the real world’ (Mishler, 1986: 26). These are the processes which have generated the findings about gender and age differences in the fear of crime. As Josselson puts it: ‘when we aggregate people, treating diversity as ...