What is Qualitative Research?

Front matter

RESEARCH METHODS SERIES

What is Qualitative Research?

Martyn Hammersley

Contents

Series foreword vii Preface ix 1 Defining qualitative research 1 2 Methodological philosophies 21 3 Divergent analytic styles 47 4 Two methodological disputes 67 Conclusion 95 References 101

Index 119

Series foreword

Preface

Qualitative research, in a wide variety of forms, has become very influential across many fields of social science over the past fifty years. It originally emerged out of resistance to the previously common assumption that quantitative measurement, experimental method, and/or statistical analysis are essential if the knowledge produced by social science is not to be ‘meagre’ and ‘unsatisfactory’ (Thomson 1889: 73-4; see Merton et al. 1984). However, while qualitative researchers are agreed in their opposition to this definition of scientific research, or at least its application to social inquiry, beyond this there is little general consensus amongst them today. There are sharp disagreements about how research should be pursued, the epistemological and ontological assumptions on which it ought to rely, and even about its purpose or function. Given this, we might ask: What exactly does the term ‘qualitative research’ now mean? And also: Is it useful to think of it, any longer, as referring to a distinctive, coherent, and worthwhile approach? These are the questions around which this book is organized.

Many books, very different in character, could be, and indeed have been, written on this topic, each reflecting, to some extent, the experience and preferences of the author, these necessarily being the product of a particular social, cultural, and intellectual path. However, my main aim here has not been to present my own views about qualitative research, but rather to outline its character and variety, and to explore some of the disputes that surround and shape it, and their implications.

| am very grateful to several people for their comments on an earlier draft of the book: Graham Crow, the editor of the series in which it appears, Stephanie Taylor, Anna Traianou, and two colleagues who served as readers for the publisher. | am sure that it has been improved as a result, even though | know that | have failed to address adequately all of the issues they raised.

The question ‘What is qualitative research?’ demands a definition of some sort: an account of what the phrase means. There is, of course, a very large methodological literature dealing with qualitative inquiry, and many definitions have been provided. Here are a couple of examples:

Qualitative research is a research strategy that usually emphasizes words rather than quantification in the collection and analysis of data. (Bryman 2008a: 366)

Qualitative research is an umbrella term for an array of attitudes towards and strategies for conducting inquiry that are aimed at discovering how human beings understand, experience, interpret, and produce the social world. (Sandelowski 2004: 893)

While both these definitions are certainly broadly accurate — and neither, of course, exhausts what these authors say about qualitative research — they pick out very different defining features. Other definitions offer additional features. For example, in a chapter entitled ‘What is Qualitative Research?, Pertti Alasuutari (1995: 7) identifies its central feature as a particular kind of analysis: whereas quantitative work seeks to explain outcomes by examining the frequency with which they are empirically associated with possible causes, qualitative analysis employs a type of reasoning that is analogous to riddle-solving. He explains this as follows:

Any single hint or clue could apply to several things, but the more hints there are to the riddle, the smaller the number of possible solutions. Yet each hint or piece of information is of its own kind and equally important; in unriddling — or qualitative analysis — [...] [e]very hint is supposed to fit in with the picture offered as the solution.

As these three definitions make clear, what are taken as the defining criteria of qualitative research can vary considerably. This perhaps tells us that it is not a simple phenomenon, not one that is easily characterized.

Ideally, in trying to understand what qualitative research is we are looking for a set of features that are shared by all examples of it, and that are not found together in other kinds of research. None of the definitions we have discussed is successful in these terms. While that provided by Alan Bryman captures something important, the main feature it identifies is, in effect, a negative one — the absence of quantification. After all, the presence of words in data collection and analysis is not distinctive to qualitative research: words are central to questionnaires, a common source of quantitative data; and there are generally more words than numbers in the analysis sections of quantitative research reports.'

Nor does Margarete Sandelowski’s definition fit these requirements. Interpreted at face value, ‘discovering how human beings understand, experience, interpret, and produce the social world’ is the goal of a great deal, if not all, of social inquiry, not just that normally listed under the heading of ‘qualitative. For example, much survey research has been concerned with documenting differences in attitude with a view to explaining people’s behaviour. Of course, it may be that what she intends by words like ‘understand; ‘experience’ and ‘interpret’ is distinctive, but at face value it is not clear what marks out qualitative research here.

Finally, while Alasuutari’s use of the riddle-solving metaphor is instructive, | doubt that the mode of reasoning he identifies is limited to qualitative work. Indeed, Susan Haack (2009) has argued that natural science, and all forms of inquiry, can best be understood as operating in a manner similar to solving a cross-word puzzle, a task that involves relating ideas about what the solution to a clue might be (empirical data) to hints from the letters supplied by answers to other clues (existing knowledge) - a process that parallels very closely what is involved in riddle-solving, as described by Alasuutari.

As | have already indicated, the problems with these definitions do not reflect a failure on the part of the authors concerned. Rather, they reveal that the task of providing an account of the distinctive features of qualitative research is far from straightforward. Indeed, trying to produce a list of

1 Campbell (1975) famously made the point that quantitative research is always founded on acts of ‘qualitative knowing’.

features that all of what might be identified as qualitative inquiry shares, and that are exclusive to it, is almost certainly a hopeless venture. Later we will see why this is.

Nevertheless, we can, perhaps, produce a list of features that, in combination, frequently characterize what would be referred to as qualitative research. A number of strategies could be used. One is to appeal to the root terms from which the phrase has been historically derived. This etymological method is illustrated by Fred Erickson’s (2011: 43) definition: ‘From Latin, qualitas refers to a primary focus on qualities, the features, of entities — to distinctions in kind — while the contrasting term quantitas refers to a primary focus on differences in amount. Appeals to etymology can sometimes be illuminating — in this case what it produces is similar to definitions produced in others ways, for example that of Bryman.

Another strategy would be to look at examples and to try to identify features that are frequently present. Box 1.1 provides summaries of a small selection of studies that would probably be regarded as clear instances of qualitative inquiry.

Box 1.1 Exemplars of qualitative research

Morton on ‘becoming tongan’

Having visited Tonga as a teenager, lived in Tongan households in Australia, and later married a Tongan, the anthropologist Helen Morton (1996) carried out eight months’ ethnographic fieldwork living in a village there, along with her young son. She observed what went on and talked to people in the village, recording her data in the form of fieldnotes; though she also used a questionnaire distributed to secondary school students. Her aim was to study the experience

of childhood in that society - what it means to become Tongan - seeking as part of this to understand the rather harsh treatment to which Tongan children are often subjected.

Rai on ‘positive loitering’

This was a ‘critical’ ethnographic study of a ‘community policing’ practice introduced by the Chicago Alternative Policing Strategy (CAPS) as a way for community members to combat ‘gang presence, public drinking, criminal behavior, and drug activity’ (Rai 2011: 66). The author examines how ‘positive loitering’ was used in a ‘gentrifying

What is qualitative research?

Chicago neighborhood’ to try to eradicate an informal street labour market. It involved ‘passive-aggressive flash mob actions where people simply show up and hangout in the spaces where the laborers gather’ (Rai 2011: 67). She observed monthly CAPS meetings, and interventions on the street, and interviewed ‘positive loiterers’, some of the people seeking work, and the organizer of the market. She argues that positive loitering highlights how ‘democratic and neoliberal rhetoric can dovetail in local practices that obscure systemic inequality [...]; and expose the ambivalence of civic participation’ (Rai 2011: 68).

Wacquant on boxing

Loic Wacquant, a French sociologist working in the United States, became an apprentice boxer ‘by default and by accident’ as part

of a study of ‘the everyday reality of the black American ghetto’ (Wacquant 2004: viii). He emphasizes the value of ‘theoretically armed’ participant observation, drawing on the work of the French sociologist Pierre Bourdieu. His aim was ‘to plumb the inner depths’ of what it means to be a boxer, so as to capture and represent ‘the taste and the ache’ of this profession, as pursued in this community (Wacquant 2004: vii). In sociological terms, the task was to document the dispositions and skills involved, and how they are developed and deployed. His data were fieldnotes written up each evening after being in the gym (Wacquant 2004: ix-x).

Bogdan on the autobiography of Jane Fry

In the early 1970s, Robert Bogdan (1974) carried out a series of

life history interviews, several times a week over a period of three months, with ‘Jane Fry’ (pseudonym), a transsexual born as a biological male who believed that she was a woman. He transcribed and edited around a hundred hours of audio-recorded material with a view to producing a sociological life history that conveyed her experience and perspective on life. Her ‘autobiography’ - an account in her own words, albeit composed from what she said in interviews - forms the central part of the book. In addition, Bogdan provides a commentary, drawing on medical records, that juxtaposes her perspective with those of the professionals who had been in contact with her at various times.

MacLure and Walker on parents’ evenings in UK secondary schools

This is a study of the discursive practices displayed in 184 meetings (typically lasting around five minutes each) between teachers and parents in five UK secondary schools (MacLure and Walker 2000).

The audio-recordings were made by two teachers and two parents, without the presence of the researchers; though the latter had observed a parents’ evening in each school. Follow-up interviews were carried out with a small number of parents, teachers, and students. The authors examine the typical structure of these meetings in sociolinguistic terms. In particular, they document the ‘interactional absence’ of students even when they were physically present, the strong similarity in the pattern of the interaction despite considerable variation in the characteristics of the parties involved, and the dominant (though occasionally contested) role played by teachers.

Wetherell and Edley on masculinity

This research was based on thirty audio-recorded interviews with groups of British men (totaling 61) from diverse occupational backgrounds, varying in age from 20 to 64 (Wetherell and Edley 1999). A typical discussion group consisted of the interviewer (Nigel Edley) and two volunteers, although a few sessions involved three and others just one. In some interviews photographs of possible role models were used to stimulate conversation. On the basis of detailed discourse analysis of this material, the authors challenge the theoretical concept of ‘hegemonic masculinity’, examining specific practices by which men construct themselves as masculine. The authors state that ‘We chose discourse as a site for investigating men’s identities because we are persuaded of the central role discursive practices play in the constitution of subjectivity’ (Wetherell and Edley 1999: 337).

Wright and Decker on armed robbers

Active armed robbers in St Louis, Missouri were contacted via a specially recruited and paid fieldworker who was an ex-criminal known to local criminals (Wright and Decker 1997). The researchers met their informants on the streets, and carried out semi-structured interviews asking about robberies in which the informants had been involved. They sometimes took the informant to the scene of a robbery to test his account; and also compared accounts provided by co-offenders. On the basis of these data, they drew conclusions about the role of motivational factors, situational features, and environmental cues in generating armed robberies.

Mitchell on survivalists

Survivalists are people who ‘anticipate various kinds of imminent cataclysm - economic collapse, race war, nuclear attack, and so

forth - and take steps to ensure their own postdisaster welfare’ (Mitchell 1991: 97). They ‘often eschew telephones, launder their mail through letter exchanges, use nicknames and aliases, and carefully conceal their addresses from strangers’ (Mitchell 1991: 100). In

order to document survivalist beliefs and practices Mitchell carried out participant observation, covert and overt, and interviews, and began to edit a group newsletter, with the result that he became ‘the recipient of a steady stream of members’ written opinions and perceptions’ (Mitchell 1991: 100). This role then allowed him to use tape recorders and cameras at group meetings. In his book Dancing at Armageddon he provides a picture of the ‘world’ of survivalists and how it reflects and responds to salient features of modern industrial civilization (Mitchell 2001).

Frake on ‘how to ask for a drink in Subanum’

In a classic study of an ethnic group who were farmers occupying part of the mountainous interior of an island in the Philippines, the author provides a detailed description of their drinking practices (Frake 1964a). This covers the types of drink available within the culture and focuses on how one of these, what he refers to as a type of beer, is drunk at religious festivities that serve important socio-political functions. His aim was to describe in detail what someone must know in order to be a competent member of this culture (Frake 1964b: 112). The information was obtained through observation but also by

highly structured interviews designed to elicit the terminological distinctions - about types of drink and the activities surrounding them - employed by the Subanum.

Oliver and O'Reilly on lifestyle migration from Britain to Spain

This study involved a re-analysis of data, derived from observation and interviews, collected by the two authors in earlier, separate projects that had each led to a published account. The aim was to use secondary analysis to explore the continued relevance of social class in light of arguments about its declining influence in Western societies. Oliver and O'Reilly (2010: 49) argue that ‘despite attempts to rewrite their own history and to mould a different life trajectory through geographical mobility’ migrants reproduced class distinctions in and through their attitudes to other migrants and to tourists. Bourdieu’s theoretical approach was employed here to explore the limited possibilities for reconstructing institutionalized dispositions in this field.

Olson et al. on suicide notes

This is a study in the sociology of health concerned with motivations for suicide (Olson et al. 2011). It offers a cross-cultural analysis of suicide notes written by Native Americans, Hispanics, and Anglos

in New Mexico. Official records were used to identify those suicides where a suicide note was available. For ethical reasons the researchers did not contact family members to obtain additional information.

The suicide notes ranged in length from several words to several pages, and took various forms, including a message on the back

of a grocery bag. Five categories emerged describing motivation: feelings of alienation, of failure or inadequacy, being psychologically overwhelmed, the wish to leave problems behind, and a desire for reunification with another in an afterlife. It was concluded that the ethnic differences in patterns of response across these categories were surprisingly small.

Levitas on political rhetoric

This study examined official documents relating to social inclusion produced in the early years of the New Labour government in the United Kingdom. The author identifies three different discourses in these documents, and studies the relationships amongst them: a redistributive discourse (acronym, RED) with a primary focus on poverty and inequality; a moral underclass discourse (MUD) which

What is qualitative research?

concentrated on what is portrayed as the delinquent behaviour

of those excluded; and a social integrationist discourse (SID) that emphasized the capacity of paid employment to facilitate people’s ‘reintegration’ into society. The three discourses differ in what the excluded are portrayed as lacking: from the point of view of RED

they have too little money, for SID the problem is that they have no work, while from the perspective of MUD they have no morals (Levitas 1998: 7). Levitas offers a critical analysis of how these discourses operated in New Labour policies.

Lewis on an online support group for irritable bowel syndrome

As part of work for a Masters thesis, Lewis (2006) became a participating member of an online support group for people with Irritable Bowel Syndrome (IBS), analyzing postings on bulletin boards and discussions in chatrooms. He supplemented these naturally occurring on-line data with a ‘qualitative questionnaire’, in other words a questionnaire with open-ended questions, sent and returned by email to participants who had agreed to complete it. His research examined how people with this condition experience their illness and use online support groups to cope with it.

Kaplan on engaging in action research with disadvantaged students through participatory photography

At the behest of a local education authority, Kaplan (2008) worked over a nine-month period with disadvantaged students and their teachers in several schools in the northwest of England. The

aim was to facilitate students using photography to explore, and share, their perspectives on, and to improve their experiences

of, education. Students were given a workshop on ““reading”/ interpreting photographs’ and issues about ethics and consent, and were instructed in basic photographic technique. Subsequently, the photographs were discussed by the group and some of these selected and captioned for a powerpoint presentation. In one school, some of the photographs generated considerable conflict, and Kaplan reports that he was left with a sense that the school’s senior management ‘was unwilling or unable’ to acknowledge the value of students’ perspectives which challenged existing structures of teaching and learning (Kaplan 2008: 188).

The differences among the studies described in Box 1.1 are probably more striking than any commonalities, not just in terms of substantive focus but also in the methods employed. Some of them involve forms of participant observation — researchers going to particular social settings, observing and perhaps also taking on a participant role. Others rely entirely on interviews or on documentary evidence, of varying kinds. Also, some of the studies were concerned with portraying an aspect of the lives of the people concerned while others focus on analyzing forms of language-use or on explicating the categories characteristic of a particular culture.

So, the label ‘qualitative research’ covers a heterogeneous field. Moreover, we cannot conclude that any feature shared by two or more of the studies listed is relevant to defining ‘qualitative research’. This is true, for example, of the fact that a couple of them used the work of Pierre Bourdieu: while his ideas are currently very influential in some quarters, qualitative inquiry draws on a very wide range of theoretical perspectives that differ quite sharply from one another.

The contrast with quantitative research

In trying to define ‘qualitative research, it is important to remember that all questions, or all interpretations of them, involve an implicit or explicit ‘frame’. In the case of ‘what is?’ questions, this relates to a typology of some sort: such questions are concerned with how one type of thing differs from something else. And the most obvious contrast in this case, indicated in some of the definitions | have already quoted, is: what is qualitative, as opposed to quantitative, research?

This is almost certainly the primary contrast that determines the meaning of the phrase ‘qualitative research’ in most contexts; though some have suggested that it is uninformative (Grahame 1999: 4; Silverman 2006: 33), and others that it is ‘unadventurous’ albeit necessary (Holliday 2002: 1). Yet, if we know the features of quantitative research, this can help us to some degree in identifying those of qualitative work, even if it does not tell us all that we might wish to know. Furthermore, this contrast is not just semantically primary but also reflects the historical development of qualitative inquiry: it emerged as a distinct kind of social science in competition with an already established tradition of quantitative method.

The label ‘qualitative research’ began to be widely used in the 1960s to identify deviation from the quantitative forms of research that were then dominant: experimental work in social psychology and some applied fields; survey research in sociology, political science, and other areas; the use of ‘official statistics’, notably in demography, economics, and health research; content analysis of media material; and structured observation, notably in education and criminology. It is not that qualitative data had not been used prior to this — indeed the development of specific quantitative methods was frequently motivated by a commitment to replace earlier types of non-quantitative data that were judged to be scientifically unsatisfactory. For example, questionnaires and tests were introduced as a more efficient and reliable method of gaining information about attitudes and values than the life history interviews that had previously been used (Stouffer 1930); predictive psychological tests were employed as an alternative to clinical interviews by psychologists (Meehl 1954); and structured forms of observation and interviewing were used as a substitute for more open-ended forms that failed to produce standardized data (see Platt 1996; Gobo 2011).

So, it was through opposition to previously used ‘unscientific’ methods that quantitative work came to be formulated as a distinct, labeled approach. And it increasingly came to be presented as generally being committed to:

1 Hypothesis-testing. An explicit research design is developed at the start of inquiry, which is aimed at testing some set of hypotheses. This is done through operationalising key variables: in other words, specifying the data that would be relevant to them, along with the instruments needed to produce these data, as well as the form of analysis to be employed.

2 The use of numerical data. Data take the form of specific counts of instances or of rankings/measurements of objects according to the degree to which they possess some feature. Measurement is often treated as superior to counting and ranking, and the ideal is to employ scales modeled on those used in natural science, such as in the measurement of temperature.

3 Procedural objectivity. Counting, ranking and measurement procedures must operate in ways that are ‘objective’, in the sense that they are standardized. This is designed to rule out bias caused by the social and personal characteristics of the researcher, and to enable checks on the operation of such bias through replication. Generalization. Samples may be studied with a view to generalizing to some larger population, and statistical techniques can be used to maximize the validity of the generalization and to assess the chances of serious error in the process.

Identifying systematic patterns of association. Statistical techniques are also used to describe the patterns to be found in the data, and perhaps also to test the likelihood that these could have been the product of random processes rather than of a systematic causal relationship.

Controlling variables. There is usually an attempt to control variables, physically via experimental method or ‘statistically’ through cross-case analysis of a large sample or population. This is designed to separate out what effects a ‘treatment’ or ‘causal’ variable has, on a particular type of outcome, from the systematic influence of other variables that might impact on it.

It was in the process of challenging the dominance of quantitative method, formulated in this way, that qualitative research itself came to be framed as a general style, approach, or ‘paradigm’ (see, for example, Filstead 1970). This challenge was based upon a number of arguments, including:

the importance of studying what normally happens in the ‘real’ world, rather than what happens under experimental conditions;

the need to observe what happens rather than to rely solely upon respondents’ accounts in formal interviews or questionnaires;

the need to allow people to speak in their own terms in interviews if we are to be able to understand their distinctive perspectives;

the danger that quantification results in the meaning of central concepts being lost;

the concern that the kind of variable analysis employed by quantitative researchers ignores the complex, contingent and context-sensitive character of social life, and the extent to which actions and outcomes are produced by people interpreting situations in diverse ways, and acting on the basis of these interpretations, rather than passively responding to external causes.

What is qualitative research?

In light of this contrast with quantitative social science, we can define

‘qualitative research’ along the following lines: a form of social inquiry that tends to adopt a flexible and data-driven research design, to use relatively unstructured data, to emphasize the essential role of subjectivity in the research process, to study a small number of naturally occurring cases in detail, and to use verbal rather than statistical forms of analysis. Each of these features requires further elaboration:

A flexible, ‘inductive; ‘abductive; or data-driven orientation. Qualitative researchers place more emphasis on generating and developing descriptions and explanations than upon testing pre-defined hypotheses. This means that a flexible research design is adopted, rather than one in which a detailed plan is laid out at the start of the research and then ‘implemented’. This is also reflected at the stage of analyzing data, where the task is to generate categories rather than to place data into pre-determined ones. And the categories initially developed tend to be open-ended and flexible in character, so that each data item can be assigned to more than one of them. In other words, the categories do not form a mutually exclusive and exhaustive set, at least not at the start of the analytic process.

Relatively unstructured kinds of data are used. There is little pressure to engage in formal counting, ranking, or measurement. For example, in the case of observation, qualitative researchers watch carefully what is happening, and often try to write concrete descriptions in natural language that capture relevant aspects of what is observed and of how events unfold. Alternatively, or as a complement, audio-or video-recording is often used, with transcripts being produced on the basis of these. Similarly, in the case of interviews, qualitative research typically involves a relatively unstructured approach where the aim is to invite informants to talk at length about matters that are broadly relevant to the research, with the interviewer following up to encourage more elaboration, detail, or exemplification where necessary. Qualitative researchers may also use documentary data, such as official reports, newspapers and magazines, photographs, maps, diaries, and so on, without seeking to quantify their content in the manner of much content analysis. In recent years, there has been a growth in the use of visual data, seeking to counter the more common reliance upon text. There has also been increasing use of material available electronically on the Internet. Qualitative researchers may also sometimes elicit documentary data, for example asking people to write diaries, produce drawings, take photographs, make videos, etc.

Subjectivity. There is acceptance, perhaps even celebration, of the fact that data, and inferences from them, are always shaped by the social and personal characteristics of the researcher. It is recognized that it is impossible to eliminate the effect of these, and indeed that they may facilitate insight as well as leading to error. It is sometimes argued that reflexivity — the provision of detailed information about the researcher and the research process — can enable readers to allow for any effects of the researcher’s characteristics, or of how the research was carried out, that might obscure or threaten the validity of the analysis. In short, there is opposition to the idea that research should be a standardized and impersonal process — to any requirement that the personal be suppressed in the name of science. However, this does not necessarily imply opposition to the kind of concern with objectivity that requires researchers to assess and try to counter potential threats to the validity of their conclusions.

The study of ‘natural’ settings. Experimental research creates settings, for example in a laboratory, that are specifically designed to allow control over the treatment variable and to rule out confounding variables. And much non-experimental quantitative research relies upon questionnaires or formal interviews that are structured with the aim of standardizing the stimuli to which respondents are subjected, in order to render responses comparable. By contrast, most qualitative work investigates what goes on in the ordinary settings in which people live and work, and/or uses interviews that are designed to approximate to ordinary conversations in key respects.

Small number of cases studied. Survey research generally studies large samples in order to generalize across many cases, and/or to provide enough cases to employ comparative analysis so as to control variables. By contrast, qualitative inquiry often involves investigation of a small number of naturally occurring cases, perhaps just one. This stems from insistence on the need for in-depth examination of each case, in order to document complexity. Also involved is the argument that each feature of a case can only be understood within the context of that case, because features will shape one another, rather than having pre-determined and fixed characters. There is opposition, then, to the tendency in quantitative research to rely upon data that have been extracted from their local contexts, for example questionnaire responses whose relationship to people's lives is uncertain. Furthermore, in-depth investigation of cases allows the checking of interpretations through comparison of data of different kinds, for example that from observation with that from interviews, or accounts from different informants. This is rarely possible in surveys.