Tomofumi OkaAssociate Professor,
Department of Social Work,
Chiyoda-ku, Tokyo, 102-8554 Japan
Ian ShawReader in Social Work,
School of Social Sciences,
Cardiff CF1 3XH, Wales, UK
The Japanese version is included in
The purpose of this chapter is to demonstrate the nature and basic methods of qualitative research to social work students and novice researchers who plan to use qualitative research in their own field of study. First, we explain the nature and philosophical backgrounds of qualitative research in social work. Second, we describe how to conduct research using a qualitative study as an illustration. Third, we discuss how to collect, analyse, and display qualitative data. Fourth, we outline the quality of qualitative research. Finally, we discuss some of ethical issues pertaining to qualitative research.
Qualitative research can be defined as "multimethod in focus, involving an interpretive, naturalistic approach to its subject matter" (Denzin & Lincoln, 1994, p. 2). If you feel that this is too vague, it is probably because what this term means is too various. To illustrate this concept, we will describe the characteristics of qualitative research. Because many authors have pointed out many different characteristics, and because it is not possible to summarise them all in such a small space as this, we will discuss only three that are regarded as important, especially to social work students.
Qualitative researchers attempt to understand meanings that people give to their deeds or to social phenomena. In other words, researchers see people from the inside. For example, when you conduct interviews with users of a residential care house for people with physical disabilities, you will have pictures of how they feel about their ordinary lives in the house. How do they think of staying or working at the house? What sort of limitations do they notice about their residence? How do they deal with conflicts with care workers and roommates? What tacit rules cover their human relations? Such questions would be very interesting for social workers that want to improve their care houses from the viewpoints of the users.
Quantitative researchers may be able to do surveys without direct contact with research objects: they can collect data by using hired and trained interviewers or by mailing out questionnaires. Qualitative researchers, on the other hand, often enter into the natural fields of people whom they study, and have face-to-face interviews with them. Because of this, qualitative research is sometimes called "fieldwork." Although this direct connectedness with people and their lives attracts social workers, it also gives rise to several ethical problems, which quantitative researchers may not face. These ethical problems will be dealt with later in this document.
Analytic induction is a major logic of qualitative research. The rule is: take one case, and develop a working hypothesis to explain it. After that, you take another case, and examine whether the hypothesis can explain the new case. If it fails, you should revise the hypothesis to explain both of the cases. Then, take the third, and repeat the same process of examining and revising the hypothesis. When you do not need to revise the hypothesis further, and you expect that the hypothesis will fit any new cases you might take, you will have refined the hypothesis enough. As you may have noticed, your choice of cases to be examined has an important bearing on the trustworthiness of analytic induction and is related to the sampling procedures of qualitative research mentioned below. This approach has been developed most fully by Strauss and Corbin (1998). Not all qualitative research follows this approach, but the inductive approach to design, fieldwork and analysis is one of the most influential characteristics of qualitative inquiry.
It is essential for qualitative researchers to be aware of the influence of philosophy on strategies of research, because without knowledge of related philosophy, they are apt to be confused when analysing qualitative data. (What we mean by philosophy is views about how to recognise things that are to be researched.) Much has been written regarding the ways in which philosophical positions feed through to influence approaches to qualitative research. Four philosophical paradigms have been identified. They include positivism, postpositivism, critical theory, and constructivism (Guba & Lincoln, 1994).
Positivism is rarely evident in a fully developed form in contemporary research, but is most usually identified with quantitative research (Rotheray, 1993; Thyer, 1989). According to positivism, the "objective" world exists independently of any perspectives of the researchers. Therefore, researchers must disclose the "objective" facts. The distinguishing feature of positivism is the absence of any distinction between reality (as things that exist) and knowledge of reality (as things that are recognised). This paradigm is present in a diluted form in some qualitative research.
Postpositivism is a modified form of positivism that appeared after the end of World War II. It admits that human beings cannot perfectly understand reality, whereas with rigorous data collection and analysis, researchers can approach the truth. Postpositivist covers a range of positions so wide that it scarcely earns the name of a paradigm. It is widely influential within qualitative research and covers positions as different as the grounded theory approach (Glaser & Strauss, 1967; Strauss & Corbin, 1998), Herbert Blumer's brand of symbolic interactionism (Blumer, 1969), recent developments under the heading of scientific realism, and the detailed ways of analysing qualitative data devised by Miles and Huberman (1994). One of us has discussed the significance of postpositivist positions more fully elsewhere (Shaw, 1999, pp. 45-47).
Critical theory under a narrow definition originated in the Frankfurt School, which was founded in Germany, during the pre-war years. In this document, however, we would use such a wide definition of Critical Theory that it would include the basic paradigms of any qualitative research directed at generating empowering or emancipatory social change directly through research (Harvey, 1990, Popkewitz, 1990). In Japan, a well-known one based on this paradigm is probably feminist research. Critical ethnography and participatory action research are also related to this. Reality cannot be grasped without researcher bias that is caused by historical, political, societal, ethnic, or gender conditions. Research should be very much related to social values, while the realisation of social values is the purpose of research.
As with postpositivist, contructivism includes a wide span of positions, from those indistinguishable from post-positivist approaches to relativist positions. Lincoln and Guba are perhaps the best known advocates of constructivism, and they are well towards the relativist end of the continuum. Their critique of conventional inquiry positions has become well known. As Lincoln summarises it in her personal account of her journey to constructivism:
Egon and I rejected conventional inquiry on three basic grounds: its posture on reality; its stance on the knower-known relationship, and its stance on the possibility of generalisation. (Lincoln, 1990, p. 68)
On reality, they came to advocate multiple, socially constructed realities which, "when known more fully, tend to produce diverging inquiry" (Lincoln & Guba, 1986, p. 75). Realities cannot be studies "in pieces" (for example, as variables) but only holistically and in context. The traditional image of the relationship between knower and known, researcher and subject, was rejected. "Knower and known not only could not remain distanced and separated in the process of evaluation, but probably should not" (Lincoln, 1990, p. 68). Hence, "the relationship, when properly established [in the process of research], is one of respectful negotiation, joint control, and reciprocal learning" (Lincoln & Guba, 1986, p. 75). Finally, because there are no enduring, context-free truth statements, and all human behaviour is time and context bound, "we [those who support this paradigm] began to doubt seriously the possibility of generalisation from one site to the next" (Lincoln, 1990, p. 68).
Is it the case that to be consistent, researchers must choose one paradigm? Is there an inherent inconsistency in subscribing to the worldview of one approach but employing the methods of the other? Lincoln answers in the affirmative:
The adoption of a paradigm literally permeates every act even tangentially associated with inquiry, such that any consideration even remotely attached to inquiry processes demands rethinking to bring decisions into line with the worldview embodied in the paradigm itself. (Lincoln, 1990, p. 81)
Our position is that there is a real but imperfect link between paradigm and method. We are likely to make a trade-off between ideas of research rigour and the relevance of our work to the social work community. This means that if we make the research more relevant to social work fields, the rigor of the research is likely to be decreased, and that if the ways of conducting research are too rigor, the research is apt to be less relevant. Becker's reminder is also salutary. He would have us give up " preaching about how things [related to research] should be done and settling for seeing how they are in fact done" (1996, p. 54). Purist attitudes toward paradigms are not appropriate in qualitative researchers. The essence of qualitative research is flexibility. In fact, qualitative researchers are called "bricoleur" (Denzin and Lincoln, 1994, p. 2) because they promote research by using whatever is immediately available.
Qualitative research is flexible in nature, therefore simple outlines of research procedures might mislead readers. We will therefore introduce one of our research projects as an example, which is done by Oka, and related to patient groups.
One day I (Oka) received a telephone call at my office. The caller, a social worker who was involved in patient self-help groups, was unknown to me. She asked me to visit her office and discuss any ideas I may have for research I could carry out on her groups.
This might be an unusual example of how to enter into a field, because it is not usually so easy for a researcher to gain access. Persons who introduce a researcher into the field in qualitative research are called "gatekeepers". It is vital for the researcher to maintain a trusting relationship with the gatekeeper because he or she can make it easy for the researcher to build trust with the people in the field. As Lincoln and Guba (1985) point out:
The building of trust is a developmental task; trust is not something that suddenly appears after certain matters have been accomplished ("a specifiable set of procedural operations"), but something to be worked on day to day. Moreover, trust is not established once and for all; it is fragile, and even trust that has been a long time building can be destroyed overnight in the face of an ill-advised action (p. 257).
Gaining entry is a very important and complex process (Bailey, 1996, pp. 49-52), and it is apt to provoke the researchers to anxiety and even a sense of guilty (Gans, 1991). For students starting qualitative research, often the first setback they meet occurs during this process.
Let us return to my story. To begin with, I planned to start informal conversational (unstructured) interviews with group leaders. Accompanied by the gatekeeper, I met the leaders. The gatekeeper worked as a facilitator, helping to make the leaders and me feel relaxed and interpreting words that were difficult for me to understand, because the leaders used many indigenous words and abbreviations for medical or social services terms.
Before the interview began, I had planned to base my research on how the groups were started, but I found during the interviews that many leaders entered their groups after they had been established, and that it would be impossible for them to answer my questions about how their groups were started. I decided, therefore, to revise my research design and to investigate the problems presently confronting the groups. This is an example of the flexibility of the qualitative research design. Rubin and Rubin (1995) mention that:
Adjusting the design as you go along is a normal, expected part of the qualitative research process. As you learn how the interviewees understand their world, you may want to modify what it is you are studying and rethink the pattern of questioning. Such flexibility is much better than persisting in a design that is not working well or that doesn't allow you to pursue unexpected insights. (p. 44)
After developing a set of questions about the groups' problems, I planned to conduct focus group interviews, which are a sort of qualitative group interviews (Morgan & Krueger, 1997). I asked many different groups to send him participants, and from these formed several focus groups. While analysing the interview data, I identified several management problems. In order to explore the most severe of these, I next decided to conduct individual guided (semi-structured) interviews. To select interviewees in qualitative research, researchers use "purposeful sampling." Although there are many strategies governing purposeful sampling, one of those I used is "intensity sampling." This is to choose "information-rich cases that manifest the phenomenon of interest intensely" (Patton, 1990, p. 171).
By using different types of interviews, I was able to collect data. This is important because if the data collected using different methods show the same pattern, that pattern is more credible. This logic supports a well-known qualitative research technique called "triangulation." If you collect data from different sources or from different researchers, you can also use "triangulation". Although this technique has received much criticism because it "assumes a single fixed reality that can be known objectively," it is believed to be a useful tool for qualitative research (Seale, 1999, pp. 53-61).
After analysing the interview data, I made copies of the research report and handed them to the group leaders, some of whom were the interviewees. I then asked the group leaders to evaluate the report. This process is called various names: member validation, member checking, or member check. Although novice researchers may assume that research objects would best know the meanings of their words, this technique is not free from criticism as a validation technique (Bloor, 1997). Bloor has concluded that, rather than validation of evidence, we should regard such exercises as valuable additional evidence.
Ways of collecting data vary in qualitative research. Because our purpose is to outline qualitative research, we will describe three classic methods of collecting qualitative data: interviews, participatory observation, and documentary analysis (Burgess, 1991a). However, we should remember that these older methods have been challenged by newer arrivals such as analysis of visual images and sound.
Rogers and Bouey (1996, p. 52) point out, "Without a doubt, the most utilized data collection method in qualitative research studies is the interview." Many authors classify qualitative interviews into three types: structured interviews, unstructured interviews, and semi-structured interviews.
First, structured interviews: sometimes called standardised interviews, these are often used in quantitative research. In structured interviews, researchers ask the same set of questions, in the same order, using the same words, to different interviewees. Structured interviews are convenient for comparing different interviewees' answers to the same questions, and when a team of researchers is involved in conducting the interviews.
Second, unstructured interviews: also called informal conversational interviews, these interviews do not have any predetermined set of questions; instead the researchers and interviewees talk freely (Burgess, 1991b). Unstructured interviews are often used in combination with participatory observation (mentioned below) and although they may look easy to conduct, novices usually find them difficult because the researchers have to generate and develop questions according to what the interviewees say.
Third, semi-structured interviews: sometimes called guided interviews, these are somewhere between structured and unstructured interviews in format in that the researchers prepare interview guides that consist of a set of questions. The guides allow researchers to generate their own questions to develop interesting areas of inquiry during the interviews. This type of interview is widely used as the qualitative interview (Flick, 1998, p. 76).
All three types of interview can be used in combination (Patton, 1990, p. 287). For example, after conducting structured interviews, researchers can conduct semi-structured interviews, and finally unstructured interviews or, conversely, they may start with unstructured interviews to relax the interviewees, and move to a semi-structured interview format.
Participant observation has been described as "the most intimate and morally hazardous" form of social research (Lofland, 1972). Patton (1990, pp. 205-216) has demonstrated that there are variations in observational methods. First, the extent of the researcher's participation can vary from full participation to onlooker observation. For example, researchers can enter the homes of people with disabilities as care workers or users (full participation), or as onlookers or spectators (least participation).
Second, there is a continuum between overt and covert observations. In overt observations, people know they are being observed whereas in covert observations, they do not. One reason why covert observations are conducted is that people might behave differently when they know they are being observed. There is much ethical discussion about covert observations (for example, Bulmer, 1982), in particular about the difficulties in deciding the extent to which the observations should be overt or covert. For example, in day centres for people with mental disabilities, should all the users be informed of the observations, even though some might fear for them unduly or others may attend the centres for a limited time?
Despite these difficulties, observations are powerful tools for researchers. For example, researcher visiting the setting to conduct interviews may unintentionally uncover much interesting data and many themes while waiting for interviewees to arrive (Hornsby-Smith, 1993). Even if you plan to use only interview data, visiting and observing the setting in which the interviewees are located must help you find important topics around the research. You might notice something that nobody has talked about in your interviews because it is so familiar to them that they think it unworthy to refer to or they may simply not want to talk about it.
There are usually many documents associated with our research settings. According to Burgess (1991a), there are three sorts of distinctions among documents. One is made between primary sources and secondary sources. Primary sources have a direct relationship with those who are studied. They include letters, diaries, and reports. Although these documents describe first-hand accounts of situations, we should not accept them uncritically and "it is essential to locate them in context." where they were produced (Burgess, 1991a, p. 124). On the other hand, secondary sources are transcribed or edited from primary sources, and we should remember that they may include errors that the transcribing and editing processes made.
Another distinction is made between public and private documents. For example, in a residential home for children, the public documents would include newsletters and books published by the organisation running the home about their history. Private documents would include the children's diaries, and letters written by children to their parents and vice versa. Among public documents, for social work researchers, official documents records (Berg, 1998, p. 182) may be important. They include case files and staff diaries that social workers share as information resources for treatment.
The third distinction is made between solicited and unsolicited documents. Solicited documents are produced with requests of the researchers. For example, researchers can ask users of a social service to keep diaries in which users jot down whatever come into their mind. Unsolicited documents are naturally produced and later taken by researchers. They include personal diaries and letters.
As mentioned above, there are several different philosophical backgrounds to qualitative research. The different bases of epistemology give rise to a wide variety of ways of analysing data, and therefore we will put forward very general principles or rules for analysing qualitative data.
Unlike in quantitative research, data analysis in qualitative research can occur before the data collection process has been completed. Indeed, Coffey and Atkinson (1996) suggest that in qualitative research:
We should never collect data without substantial analysis going on simultaneously. Letting data accumulate without preliminary analysis along the way is a recipe for unhappiness, if not total disaster. (p. 2)
In unstructured interviews, for instance, you should start to analyse what is being said while talking with people, otherwise, you will not be able to decide what questions to ask next in the conversation. In qualitative research, doing data analysis while collecting data is called the "principle of interaction between data collection and analysis" (Erlandson, Harris, Skipper & Allen, 1993, p. 114).
Moreover, it is important to remember that the researcher is not the only person engaging in such a simultaneous analysis. The person being interviewed is also engaging in what Holstein and Gubrium call "indigenous coding" (Holstein & Gubrium, 1995, p. 56; Shaw, 1999, pp.175-176). That is, interviewees may also analyse what they say while being interviewed. The researchers should take this into their consideration when they start to analyse the interview data.
A researcher conducting an interview with social workers about work-related stress might notice that what the interviewees say in the interview rooms of their agency is quite different from what they say in a location where different social rules operate such as a restaurant. It may be very hard for the interviewees to describe their work-related stress in the very place that produces their stress. Or the interviewees might overstate particular aspects of their stress in order to make the conversation in the restaurants more interesting and entertaining. Thus, it is vital for qualitative researchers to consider the context in which the data are obtained.
The importance of context has been brought to the forefront through the general availability of extensive electronic text manipulation in qualitative data analysis, as Seale (2000) pointed out. In other words, when "cut-and-paste techniques" (techniques for cutting off relevant segments and pasting them together) are applied to the analysis of "interview transcripts" (verbal and non-verbal records of interviews), consequently novice researchers often pay attention to the fragments of the transcript, and lose sight of the context in which the fragment is located, which can be misleading.
Many authors say, "In qualitative research, the investigator serves as a kind of 'instrument' in the collection and analysis of data" (McCracken, 1988, p. 18). This means that reflexivity is an important aspect of qualitative research. According to Hammersley and Atkinson (1995, p. 16), reflexivity implies that "the orientations of researchers will be shaped by their socio-historical locations, including the values and interests that these locations confer upon them."
One way to become aware of one's own values and preconceptions is to use "a cultural review." For example, Shaw (1999, p. 148) gives the illustration of a researcher "preparing to interview a woman having difficulties coping with caring for her son of 25 who has serious learning disabilities and lives at home". The researcher could make a self-inquiry by asking the following questions: "Who have I cared for?" what would be "my own feelings about being cared for by a close relative?" and "my reaction to carrying out personal care tasks" and so on.
Another way of being reflexive is to keep "a reflexive journal" in which researchers put "information about their schedule and logistics, insights, and reasons for methodological decisions" (Erlandson et al., 1993, p. 143).
As Coffey and Atkinson (1996, p. 3) have repeatedly mentioned, "There are many ways of analyzing qualitative data. . . . We certainly want to discourage our readers from uncritically adopting particular approaches to analysis without making principled decisions from among the available alternatives." In particular, they worry about the misconception that the way to analyse data is to use some form of coding and "stress . . . that although coding may be part of the process of analysis, it should not be thought of as the analysis in itself" (p. 26).
Misconceptions about coding have been popularised by the widespread use of CAQDAS (Computer-assisted qualitative data analysis software: Main CAQDAS is described at <http://www.scolari.co.uk/>.). This is because present main CAQDAS programs have been given "a particularly strong influence" by grounded theory, where coding is central to the analysing processes (Lonkila, 1995) and the main functions of the programs are only related to codes or coding processes (Tesch, 1990, pp. 150-155). (However, newer programmes are overcoming these limitations.)
Although we have insufficient space to detail various strategies of analysis, qualitative researchers are allowed to use as many strategies as they like, because flexibility is one of the principles of qualitative research.
In their reports, researchers should describe what decisions were made and actions were taken through the analysing processes. Many novice researchers using qualitative research say, for example, "Through qualitative interviews I have derived such and such a concept," without clarifying the decisions they made during the analysis. As Miles and Huberman (1994) have mentioned, however, "Keeping a precise record of the actual criteria and decision rules used . . .is essential" (p. 100). An example of a simple decision rule given by Miles and Huberman is "two confirmations and no contradiction" (p. 131). When applying this rule to qualitative interviews, researchers would regard a concept as important if it is confirmed by at least two interviewees and negated by nobody.
One of the basis rules in writing a qualitative interview report is to use quotes of the interview transcripts that support your ideas. How to edit and use quotes is, however, a rather difficult task for novices.
Kvale (1996) introduces the following guidelines for reporting interview quotes: First, the quotes should be linked to the related text. Second, the contexts of the quotes should be clarified. Third, the quotes should be given interpretation, otherwise the readers cannot understand why they are quoted. Fourth, a proper balance between quotes and text should be kept. When the first three principles are neglected one result is that a surfeit of half-digested quotations dominates the text. Fifth, the quotes should usually be relatively short in length. Long quotes are often vague in their meaning. However, Atkinson has demonstrated the value of longer quotations as contextualising narrative (Atkinson, 1992; for a recent social work example see White, 1999). Sixth, only the most illustrative quotes should be presented. Too many quotes and too much repetition makes a report dull. Seventh, "interview quotes should be rendered in a written style." We are not sure how strictly this rule should be applied to Japanese editing because in the Japanese language there seems to be wider differences between written and spoken styles than in Kvale's language. However, it is important to edit all quotes sufficiently so that readers of the report can understand them easily. Lastly, Kvale emphasises that "In order that the reader will know about the extent of editing of the quotes, the principles for editing should be given, preferably with a simple list of symbols for pauses, omissions, and the like." (pp. 266-267).
Having accumulated a huge amount of transcripts, students often ask us what they should find through their analysis. To put it very simply, successful qualitative research reports often include new concepts, new typology, or new causal networks.
As Hammersley and Atkinson (1995, p. 209) point out, "The initial task in analysing qualitative data is to find some concepts that help us to make sense of what is going on in the scenes documented by the data." This process is often called "conceptualisation." New concepts may be taken from "folk terms" (Spradley, 1979), which are used in the culture of the interviewees, or the researchers may find useful concepts though literature reviews on the subjects. A familiar example of the power of conceptualisation is Goffman's work on "stigma" (Goffman, 1963) and "total institutions" (Goffman, 1961).
By listing concepts and relating them to one other, we can produces typologies. An example is shown below. Social work practitioners appear to be strongly preoccupied with accountability. Research has elicited recurring instances of ways in which practice regarded by social workers as having not gone well is likely to trigger deep uncertainties in the mind of the reflective practitioner (Shaw & Shaw, 1997). This research suggests there was a widespread sense that outcomes, both positive and negative, were beyond prediction, and will never cease to jump out and surprise the practitioner. An example of a typology has been developed by Bull and Shaw to illustrate the accounts that may be adopted by social workers in relation to actions and events for which they may be called to account.
Source: Bull & Shaw (1992, p. 641)
A matrix is a useful form for displaying typologies. The following example was produced, in research that Oka participated in, when leaders of parent groups were asked what they regarded as the most important of their activities. This question was asked of over twenty leaders, and only those concepts were selected when they were included in the quotes of two or more interviewees.
Source: Oya-no-Kai (1998)
"Having children with the same diseases makes our psychological distance much smaller."
"We are like relatives. Parents who have children with the same diseases won't cut their relations with each other even if they had quarrels."
"Knowing the existence of this group, I was relieved. I felt, oh, I am not alone."
"Thinking that I am alone and knowing that somebody is feeling like me are completely different!"
"Speaking about each others' sufferings, and talking without reservations. That's important."
"Nobody understands me in the community. But here, everybody understands me."
Matrices are also useful for describing or explaining situations, time-ordered changes, and relations between two lists of concepts and so on. Miles and Huberman (1994) introduced sophisticated techniques of matrices for display of qualitative data.
It is commonly argued that qualitative data are used for generating hypotheses, or describing process, whereas quantitative data are used to analyse outcomes or verify hypotheses. We believe that this division of labour is rigid and limiting. Having said this, the division is reinforced by the frequent absence of discussions of cause in qualitative methodology and analysis texts (Coffey & Atkinson, 1996). Qualitative evaluation cannot resolve the problems of causal conclusions any more than quantitative evaluation, but it can assess causality "as it actually plays out in a particular setting" (Miles & Huberman, 1994, p. 10). As well as matrices, networks are a good way of displaying data and causalit because they can describe complex stories in a simple way.
The following is adapted from a research report one of whose authors is Oka: this is a causal network on the shortage of group leaders in self-help groups for parents whose children have rare and serious diseases.
This figure describes an answer comprising the combined responses of the interviewees: because their children have diseases, the parents have to stay home, and this does not allow much interaction to take place among group members. On the other hand, because of the rarity of their children's diseases, group members are geographically scattered so that they have difficulty in interacting with one another. The rarity of the diseases creates difficulties in obtaining information on the diseases, and on the other hand, the parents' groups act as important sources of the precious information. Thus, many members join the groups to obtain information and many of them become dependent on the group leaders. Such dependent members increase the leaders' workload, preventing them from facilitating members' chances of interaction. Members, who have little interaction with each other, fail to develop mutual trust and to understand their groups' purposes and activities. Consequently, many members become more dependent. This complex story is simply described in the network figure.
Although the arrows used in this figure are simple, various forms can be used to convey complex meanings. For example, the influences confirmed by over three interviewees are regarded as strongly confirmed, which is indicated by using bold arrows.
Probably many qualitative researchers have been criticised by critics who have claimed that their research it is too subjective, or the number of cases is too small, or that mere talking is never a scientific method, and so on. Qualitative inquiry, moreover, has a short history in Japanese social work research and therefore Japanese social work researchers may need to strongly defend the validity of their qualitative research. However, they should do so in the confidence that qualitative research is no less rigorous than more traditional forms of inquiry.
Although "validity" and "reliability" of qualitative research is discussed by many researchers (for example, Cook & Campbell, 1979 is a classic discussion; Kirk & Miller, 1986; Silverman, 1983), the most often quoted concept of the problem of establishing validity is probably the notion of trustworthiness that was developed mainly by Lincoln and Guba (1985, p. 290). The notion of trustworthiness has four elements: credibility, transferability, dependability, and confirmability. These are analogous to "internal validity", "external validity", "reliability", and "objectivity" in conventional criteria. Because their philosophical base is constructivism, these criteria are not applied to all the qualitative research methods. We believe, however, that these criteria are useful for novice researchers to understand the validity and reliability issues of qualitative research and so we will describe them and some of the techniques related to them.
Credibility is analogous to "internal validity" in conventional criteria. It relates to how the reconstruction of the researchers fits the realities and views the participants express in the process of the inquiry. To establish credibility, researchers use a variety of techniques, but because we have limited space, we will discuss only the following.
One is "prolonged engagement." That is "the investment of sufficient time to achieve certain purposes; learning the 'culture [of the participants],' testing for misinformation introduced by distortions either of the self or of the respondents, and building trust [with the participants]" (Lincoln & Guba, 1985, p. 301).
Another is "peer debriefing." This is the process of "allowing a peer who is a professional outside the context and who has some general understanding of the study to analyse materials, test working hypotheses and emerging designs, and listen to the researcher's ideas and concerns" (Erlandson et al., 1993, p. 140).
Transferability refers to the possibility that what was found in one context by a piece of qualitative research is applicable to another context. As Lincoln and Guba (1985) point out:
If there is to be transferability, the burden of proof lies less with the original investigator than with the person seeking to make an application elsewhere. The original inquirer cannot know the sites to which transferability might be sought, but the appliers can and do. . . . The responsibility of the original investigator ends in providing sufficient descriptive data to make such similarity judgements possible. (p. 298)
Providing sufficient descriptive data is often called "thick description" (Geertz, 1973). For example, if you conduct qualitative research on a day centre for people with mental disabilities, you should describe the context so thickly that users of your research report can judge whether the findings are applicable to their own settings.
Dependability is the qualitative researcher's equivalent of the conventional term "reliability", which is equal to replicability. In quantitative research, reliability means that the same tests should produce the same results. For qualitative researchers, this kind of replicability is impossible to realise because the research design is so flexible and the research findings are produced by constantly changing interactions between researchers and participants. Therefore, as Guba and Lincoln (1989) states,
Far form being threats to dependability, such changes and shifts are hallmarks of a maturing - and successful - inquiry. But such changes and shifts need to be both tracked and trackable (publicly inspectable). (p. 242)This idea leads to a technique of "auditing," mentioned below.
On the other hand, confirmability is parallel to "objectivity" in conventional criteria, and is "concerned with establishing the fact that the data and interpretations of an inquiry were not merely figments of the inquirer's imagination." Researchers need to link "assertions, findings, and interpretations, and so on to the data themselves in readily discernible ways" (Schwandt, 1997, p. 164).
Although dependability and confirmability correspond to different notions in conventional criteria, both are realised by similar techniques in qualitative research. Of these techniques, "auditing" is emphasised strongly by many authors. Auditing in qualitative research is analogous to a fiscal audit. Schwandt (1997, p. 6) states that auditing is "a procedure whereby a third-party examiner systematically reviews the audit trail maintained by the inquirer". In the case of qualitative interview research, the audit trail includes recorded materials such as cassette tapes, interview transcripts, interview guides, lists of interviewees, lists of categories and hypotheses the researcher used while analysing the data, notes about research procedures, and so on.
Lincoln and Guba (1985, p. 326) claim that "even for a complex project, a week to ten days will be sufficient" to complete auditing. However, that may sounds too expensive for our readers, most of whom may be novices, and they might be relieved to see what other researchers have written:
The researcher's work in preparing an "audit trail" and the auditor's analysis, with its very detailed procedures, are at least as expensive. . . . We should probably expect that detailed documentation and auditing will continue to be restricted to high-stakes studies, or to those in which the researcher has a special interest in documentation or auditing as such (Huberman & Miles, 1994, p. 440).
Huberman and Miles (1994) also warn that this sort of re-analysis through auditing "raises questions about invasion of privacy and about potential harm to informants" (p. 440). In spite of these risks and limitations, simpler types of auditing could be useful tools for improving the quality of qualitative research.
Ethical issues related to social work research are very important and have been much discussed. All qualitative researchers in social work should give serious thoughts to these issues particularly as the nature of qualitative research adds its own complications. In this document, we will deal with some ethical issues pertaining to qualitative research. They include confidentiality, informed consent, emotional safety, and reciprocity.
Very few people would willingly express their most private details, opinions and emotions in public documents knowing that their names would be published. Thus, confidentiality is a vital requirement for credible research.
More importantly, mere anonymity is insufficient for confidentiality to be safeguarded (Berg, 1998, pp. 48-50). For example, suppose a paper says, "We visited a university-attached day care centre for people with mental disabilities in City Y, Prefecture K." If the author's office is in the Kanto Area, readers can readily assume Prefecture K to be Kanagawa Prefecture. They might also associate City Y with Yokohama City, the second largest city in Japan. Even a large city does has not very many university-attached day care centres, consequently although this paper may have concealed the name of the site, it fails to maintain confidentiality. If this paper includes quotes from service users who criticise professionals very strongly, the professionals may suspect that certain of their own clients are the critics. Suspicion such as this can only harm people, consequently, in particular those who are socially disadvantaged.
Also, as mentioned earlier, qualitative researchers require prolonged engagement in the field in order to carry out effective research. If you conduct research at Hospital A for some years and publish a qualitative case report on an anonymous hospital, many people around you will surely conclude that the hospital mentioned in the report is Hospital A. In this case, you should pay as much attention as possible to maintaining the confidentiality of the individual participants, changing the facts where necessary as long as these changes do not distort the essential elements of the report.
Gaining informed consent is essential for all sorts of research and the flexible nature of the qualitative research design causes particular problems. Because of such an emergent design, Bartunek and Louis (1996) emphasise the importance of repeatedly confirming informed consent. In a qualitative research project,
prospective participants often do not have full knowledge . . . of the types of events that will unfold during a study. . . Informed consent . . . must then reflect an awareness that such events cannot entirely be predicted. As a result, a revised view of informed consent seems warranted, in which consent is negotiated at different points in the research cycle. Informed consent is not something that can be handled once and for all at the beginning of a study. (p. 58)
Qualitative interview research gives inquirers many opportunities to involve the participants emotionally about sensitive topics (Renzetti and Lee, 1993). As Padgett (1998) states:
Many qualitative interviews elicit intense discussions of painful life events such as divorce, death of a family member, and domestic abuse. Sensitivity to research ethics dictates that we do not introduce these topics gratuitously; they should either be volunteered by the respondents or inquired about when they are the focus of the study. (p. 37)
On the other hand, if the interviewer has good listening skills, they may provide participants with a chance to release their emotions. Weiss (1994) points out:
There are obvious resemblances between the research interview and therapeutic interviewing. The research interviewer resembles a therapist by encouraging the respondent to develop thoughts and memories, by eliciting the respondent's underlying emotions, and by listening closely to the respondent's utterances. (p. 134)
This "therapeutic" nature of qualitative interviews might cause a more complicated ethical dilemma (Patton, 1990, p. 354). If, while interviewing, the participants begin to regard you as "therapists" and open their mind more than they would usually for "researchers", should you stop them or should you allow them to continue? If you are also a social worker and are naturally expected to play a therapeutic role, you might be confused as to which type of interview you are conducting, a research interview or a therapeutic one.
The relationship between the methodology of qualitative research and professional practice is a major question that falls outside the scope of this chapter. However, the question of reciprocity is important to note. Qualitative researchers tend to have more personal relations with the research participants, and the reciprocity of research will be more keenly noticed among both researchers and participants. As Glesne states (1999):
As research participants willingly open up their lives to researchers - giving time, sharing intimate stories, and frequently including them in both public and private events and activities - researchers become ambivalent, alternatively overjoyed with the data they are gathering, but worried by their perceived inability to adequately reciprocate. (p. 126)
In particular, our participants are often socially disadvantaged, and they need social advocates for them. Social workers who co-operate with you on research surely hope that your research will help improve social welfare.
Novices may be worried so much about the ethical issues mentioned above that they hesitate about publishing their research. However, the principle of reciprocity will not allow you to keep your research results as private "beautiful memories" only. By taking up the opportunity to do research, you assume its obligation and responsibility.
Although there are special issues in ethical practice in qualitative research, we do not agree with Lincoln and Guba when they argue that qualitative research is more ethical than conventional research (Lincoln & Guba, 1989). No research methodology is ethically privileged, and formulations of ethical principles are no different for quantitative and qualitative methodologies.
We can state that qualitative research has been conducted in the early days of Japanese social work. Also, numerous research projects with a name of "case study" have been continuously carried out since a long time ago. However, Japanese social work researchers have seemingly paid their attention only to the research results and application to practice, and much less attention to the qualitative research methodology. In these days, qualitative research are more interested among researchers of education, nursing, and sociology, and several textbooks have been published. Social work researchers will be more and more interested in qualitative research, and we should remember the relations between the research methodology and helping practice and the unique ethical issues in our arguments.
This article was first written in English by Tomofumi Oka, and later revised and added to by Ian Shaw. This was then translated by Oka into Japanese. Readers can access the original English version at http://socwork.com/2000/qr.html.