Learning From Strangers: The Art and Method of Qualitative Interview Studies
CHAPTER 1: INTRODUCTION
Interviews can be as prepackaged as the polling or survey interview in which questions are fixed and answers limited: Studies whose ultimate aim is to report how many people are in particular categories or what the relationship is between being in one category and another are justly called quantitative. because their results can be presented as a table of numbers (for example, in a table Quantitative studies pay a price for their standardized precision. Because they ask the same questions in the same order of every respondent, they do not obtain full reports. Instead, the information they obtain from any one person is fragmentary, made up of bits and pieces of attitudes and observations and appraisals. Interviews that sacrifice uniformity of questioning to achieve fuller development of information are properly called qualitative interviews, and a study based on such interviews, a qualitative interview study. because the fuller responses obtained by the qualitative study cannot be easily categorized, their analysis will rely less on counting and correlating and more on interpretation, summary, and integration. In general, if statistical analysis is our goal, we would do better to use a survey approach. The report we ultimately write can provide readers with a fuller understanding of the experiences of our respondents. We need not restrict ourselves to just the one approach. what is common to them all is that they ask the respondent to provide an observer's report on the topic under study. The style of the qualitative interview may appear conversational, but what happens in the interview is very different from what happens in an ordinary conversation. In an ordinary conversation each participant voices observations, thoughts, feelings. Either participant can set a new topic, either can ask questions. In the qualitative interview the respondent provides information while the interviewer, as a representative of the study, is responsible for directing the respondent to the topics that matter to the study. judging when the respondent's report was adequate and when it needed elaboration, mostly the interviewer expressed a desire to understand whatever it was the respondent was saying. A qualitative interviewing study can be enormously time consuming, but it need not be. It is entirely possible for investigators who do quantitative work to end a study knowing more about the statistical packages they have used for computer analysis than about the topic of their study. By contrast, those who do qualitative interview studies invariably wind up knowing a lot about the topic of their study. While it can be valuable for the results of qualitative interview studies to be verified by other methods, it can also be valuable for the results of studies done by other methods to be illuminated by qualitative interview studies. A COMPROMISE? FIXED QUESTION, OPEN RESPONSE Here respondents are asked carefully crafted questions but are free to answer them in their own words rather than required simply to choose one or another predetermined alternative. Unfortunately, the fixed- question- open- response approach to data collection turns out to sacrifice as much in quality of information as it gains in systematization. The fixed- question— open- response approach would have succeeded in getting a headline but would have missed the story. Even though fixed- question— open- response interviewing may at first appear to be a systematic approach to qualitative interviewing, it is not. It is a different approach entirely. Qualitative interview studies generally begin with decisions regarding the sample to interview, move on to data collection, and conclude with analysis. the phases of work in qualitative research overlap and are intermeshed. The chapters that follow trace the likely sequence of the investigator's concerns in a qualitative interview study: sampling, preparing for interviewing, conducting the interviews, analyzing the data, and, finally, writing the report. CHAPTER 2: RESPONDENTS: CHOOSING THEM AND RECRUITING THEM
The set of topics the study explores, taken together, might be said to constitute the substantive frame of the study. who should be talked with, and about what, can be worked out. One good reason for doing pilot interviews is to clarify the aims and frame of the study before interviewing its primary respondents. develop the study's substantive frame in consultation with members of its primary audience, If representatives of that audience are available, the frame might usefully be discussed with them. The study's substantive frame decides who should be interviewed and what they should be asked. There are two distinct categories of potential respondents: people who are uniquely able to be informative because they are expert in an area or were privileged witnesses to an event; and people who, taken together, display what happens within a population affected by a situation or event. produce a dense description of what happened, try to talk with everyone in a position to know what happened in the hope that each would provide part of the story and that all of their accounts together would provide the story in full. Often the study of an issue can be cast in a way that requires a panel of informants but with what seems to be only slight redefinition can be recast to require a sample of representatives. We might, of course, decide to do both studies. We might want a panel of informants to tell us about the institution of child visitation and a sample of parents to tell us how it works in practice.We would then be doing two distinct studies. They would enrich each other, but our work load would be greater. people of different backgrounds, with different perspectives, who became involved in different ways. A study of an organization requires that the investigator succeed in obtaining informants without being perceived as an intrusive foreign presence. success is dependent on a certain amount of social grace, including sensitivity, considerateness, and tact; self- confidence; awareness of the politics of the institution; and persistence. Being unobtrusive can help. A good person to start with in any study requiring a panel is a knowledgeable insider willing to serve as an informant on informants. need to feel confident of you before they can comfortably be candid. vouched for by a mutual acquaintance can be useful. implied sponsorship of government or foundation funding for the project may also help. One principle is to start with people who are available to you and easy to interview, especially if having interviewed them will make you more informed and legitimized when you proceed to interview others. A second principle is to have your early interviews with people who are of marginal importance to the study so that if you make mistakes it won't matter so much. One approach is to develop a sample that can be argued on grounds of mathematical probability to be not too different from the population in which we are interested. A sample can be a probability sample only if respondents are selected randomly. Random selection is not the same as haphazard selection. a procedure that could equally well have selected absolutelyanybody in the population. Rather than choose respondents randomly, and thus risk unwaned duplication in our sample, we may prefer to select respondents purposively so that we obtain istances of all the important dissimilar forms present in the larger population. This kind of sample might be referred to as a sample chosen o maximize range. We are particularly likely to want a sample chosen to maximize range rather thana probability sample if our sample will be small. If we plan to work with samples much smaller than (samples of 30, say) we may not trust random selection to provide us with instances of significant developments that occur infrequently. Random sampling will provide us with a picture of the population as well as of particular instances, and sampling for range will ensure that our sample includes instances displaying significant variation. To know whether potential respondents have characteristics you want, you can include "filter" questions in the telephone calls you make to arrange for interviews. The third approach to obtaining a sample of respondents, in addition to choosing them on a probability basis or choosing them to provide a useful range of instances, is to accept pretty much whomever we can get. This is a sample of convenience. In attempting to learn about a group difficult to penetrate— gypsies, migrant workers, the very rich— it can be a breakthrough to find any member of the group, any member at all, willing to serve as an informant and respondent. You might find a congregating place for people of the kind you want to study. a support group. check an encyclopedia of associations to see if a group has been established With a probability sample, generalization is straightforward, based on mathematical argument. With a sample in which it has been possible to maximize range, it can be argued that instances of every important variation have been studied. With other sorts of samples other arguments must be relied on. Here are five arguments that might be advanced to justify the attempt to generalize from the findings of convenience samples— and one that should not be, although it sometimes is. it is likely to be a good idea to include at least a few comparison cases. Sometimes cases that occur infrequently should be sought out because they are significant conceptually. Compared with survey research studies, qualitative interview studies collect more material from fewer respondents. Case research is different primarily because it anchors its potential for generalization in the welter of detail of the single instance. Generalization can then become uncertain (and rest heavily on the theory we bring to the case), but in compensation we have the coherence, depth, development, and drama of a single fully understood life. People marooned at home tend to welcome interviewers. the hospitalized or the retired. people in crisis, On the other hand, interviewers may need the right sponsorship or topic or approach to avoid being turned down we have generally sent the potential respondents a letter explaining the study, arguing for the importance of their participation, and saying someone would telephone. much to be said for letting respondents know that their participation will be valued. undesirable to interrupt a respondent's account. CHAPTER 3: PREPARATION FOR INTERVIEWING
A good report would inform its audience about matters of importance to them. It would tell them about experiences that affect them, provide them with explanations for things that have puzzled them, and give them maps to situations they may enter. It would contribute to their competence, their awareness, or their well- being. interview guides should be seen as provisional and likely to change as more is learned. Most survey studies try to keep interviews to an hour or less. interview will go for an hour and a half or 2 hours. It is almost always desirable, if time and costs permit, to interview respondents more than once. a first meeting is partly about establishing the research partnership. A research project that compared telephone and face- to- face interviewing found that telephone respondents broke off contact more quickly, were both more acquiescent and more evasive, and were more cautious about self- revelation. CHAPTER 4: INTERVIEWING
When I can, I begin the interview where the respondent seems already to be. my role will be as interviewer and propose to the respondent that his role will be try to get in tune with the respondent by extending his comment the interviewer will not question the respondent's appraisals, choices, motives, right to observations, or personal worth. admire their knowledge and authority and was, indeed, already awed to be in the presence of someone so important. Some interviewers are willing to act as the respondents? antagonists. If they suspect the respondent is holding back information, they are ready to confront the respondent: Journalists sometimes read up on respondents, the better to confound the respondents? efforts to dissemble. Being a good interviewer requires knowing what kind of information the study needs and being able to help the respondent provide it. scenes and events external to the respondent and the respondent's own thoughts and feelings. We obtain descriptions of specific incidents by asking respondents to particularize. "Is there a specific incident you can think of that would make clear what you have in mind?"; describing a specific incident. 1 Respondents often prefer to provide generalized accounts rather than concrete instances, Helping Respondents Develop Information Extending. Filling in detail. Identifying actors. Others the respondent consulted. Inner events. Making indications explicit. marker as a passing reference made by a respondent to an important event or feeling state. Because markers occur in the course of talking about something else, you may have to remember them and then return to them when you can, saying, "A few minutes ago you mentioned…" But it is a good idea to pick up a marker as soon as you conveniently can if the material it hints at could in any way be relevant for your study. The first rule of interviewing is that if the respondent has something to say, the respondent must be able to say it. If you find yourself talking over the respondent, interrupting, or holding the floor while the respondent tries to interrupt, something is going wrong in the interview. Never, never fight for control of the interview. The interview is a collaboration. collaboration; it's your responsibility to set topics. You can usually manage the redirection without discouraging the respondent from talking freely. It is usually enough for the interviewer to give business card information— location and profession— along with the study's aims and sponsorship. It can be hard to know what is relevant, especially in early interviews, before the frame of the study is firmly established. My policy is: If in doubt, see what's there. How do you know whether you are being told enough, whether you are being given enough development and enough detail? One test is visualizability. Be alert to indications by the respondent of discomfort, antagonism, or boredom. CHAPTER 5: ISSUES IN INTERVIEWING
you have to acknowledge the respondent's distress and, for a time, simply sit and listen and permit the person to feel whatever he or she feels. "That's too bad." Beyond this the interviewer does best to convey a middle distance in response to the respondent's feelings, in touch with them and responsive to them, but not overwhelmed by them. There is no reason for an interviewer to feel guilty about intruding on a respondent's grief or sorrow. the interviewer should bear in mind that sensitivity, tact, and respect for the respondent, always important, are essential with a respondent who displays pain.