Making and Analyzing Thematic Connections

A more conventional way of presenting and analyzing interview data than crafting profiles is to organize excerpts from the transcripts into cat­egories. The researcher then searches for connecting threads and patterns among the excerpts within those categories and for connections between the various categories that might be called themes. In addition to present­ing profiles of individuals, the researcher, as part of his or her analysis of the material, can then present and comment upon excerpts from the interviews thematically organized.

During the process of reading and marking the transcripts, the re­searcher can begin to label the passages that he or she has marked as interesting. After having read and indicated interesting passages in two or three participants’ interviews, the researcher can pause to consider wheth­er they can be labeled. What is the subject of the marked passages? Are there words or a phrase that seems to describe them, at least tentatively? Is there a word within the passage itself that suggests a category into which the passage might fit? In Sheehan’s transcript, some of the labels for the passages included in the Appendix might be “background of provider,” “support groups,” “impact on family,” “abuse,” and “parents.”

The process of noting what is interesting, labeling it, and putting it into appropriate files is called “classifying” or, in some sources, “coding” data. (See Dey, 1993, p. 58, for a critique of the term coding as applied to qualitative research.) Computer programs are available that will help classify, sort, file, and reconnect interview data. By telling the computer what to look for, the program can scan large amounts of data quickly and sort material into categories according to the directions.

For those who choose to work with either a dedicated analytical program or even a word-processing program, I suggest caution in doing significant coding or editing on screen. I recommend working first on a paper copy and then transferring the work to the computer. My experi­ence is that there is a significant difference between what one sees in a text presented on paper and the same text shown on screen, and that one’s response is different, too. I have learned, for example, that it is fool­ish of me to edit on screen, because I invariably miss issues that are easily evident to me when I work with a paper copy. I would not recommend relying on reading an interview text on screen for the process of catego­rizing material. Something in the mediums of screen and paper affects the message the viewer retrieves (see McLuhan, 1965, for an early and influential commentary on this process).

At this point in the reading, marking, and labeling process it is impor­tant to keep labels tentative. Locking in categories too early can lead to dead ends. Some of the categories will work out. That is, as the research­er continues to read and mark interview transcripts, other passages will come up that seem connected to the same category. On the other hand, some categories that seemed promising early in the process will die out. New ones may appear. Categories that seemed separate and distinct will fold into each other. Others may remain in flux almost until the end of the study.¹ (See Charmaz, 1983, for an excellent description of the pro­cess of coding; also Davis, 1984.)

In addition to labeling each marked passage with a term that places it in a category, researchers should also label each passage with a notation system that will designate its original place in the transcript. (Dey, 1993, points out that many dedicated analytical computer programs will do this automatically.) I use, for example, the initials of the participant, a Roman numeral for the number of the interview in the three-interview sequence, and Arabic numbers for the page number of the transcript on which the passage occurs. Later, when working with the material and considering an excerpt taken from its original context, the researcher may want to check the accuracy of the text and replace it in its full context, even go­ing back to the audiotape itself. The labeling of each excerpt allows such retracing.

The next step is to file those excerpts either in computer files under the name of the assigned category or in folders. Some excerpts might fit reasonably into more than one file. Make copies of those and file in the multiple files that seem appropriate.

After filing all the marked excerpts, reread all of them file by file. Start sifting out the ones that now seem very compelling, setting aside the ones that seem at this stage to be of less interest. At this point, the researcher is in what Rowan (1981) calls a “dialectical” process with the material (p. 134). The participants have spoken, and now the interviewer is responding to their words, concentrating his or her intuition and intel­lect on the process. What emerges is a synthesis of what the participant has said and how the researcher has responded.

Some commentators regard this sorting and culling as an entirely intuitive process (Tagg, 1985). It is important, however, that researchers also try to form and articulate their criteria for the winnowing and sorting process. By doing so, they give their readers a basis for understanding the process the researchers used in reducing the mass of words to more manageable proportions.

I do not begin to read the transcripts with a set of categories for which I want to find excerpts. The categories arise out of the passages that I have marked as interesting. On the other hand, when I reflect on the types of material that arouse my interest, it is clear that some patterns are present, that I have certain predispositions I bring to my reading of the transcripts.

When working with excerpts from interview material, I find myself selecting passages that connect to other passages in the file. In a way, quantity starts to interact with quality. The repetition of an aspect of ex­perience that was already mentioned in other passages takes on weight and calls attention to itself.

I notice excerpts from a participant’s experience that connect to each other as well as to passages from other participants. Sometimes excerpts connect to the literature on the subject. They stand out because I have read about the issue from a perspective independent of my interviewing.

Some passages are told in a striking manner or highlight a dramatic incident. Those are perhaps the most troublesome for me. They are at­tractive because of their style or the sheer drama of the incident, but I know that I have to be careful about such passages. The dramatic can be confused with the pervasive. The researcher has to judge whether the par­ticular dramatic incident is idiosyncratic or characteristic (Mostyn, 1985).

Some passages stand out because they are contradictory and seem de­cisively inconsistent with others. It is tempting to put those aside. These in particular, however, have to be kept in the foreground, lest researchers exercise their own biased subjectivity, noticing and using only materials that support their own opinions (Kvale, 1996, p. 212; Locke, Silverman, & Spirduso, 2004, pp. 222-223). The researcher has to try to understand their importance in the face of the other data he or she has gathered (Miles & Huberman, 1984).

The process of working with excerpts from participants’ interviews, seeking connections among them, explaining those connections, and building interpretative categories is demanding and involves risks. The danger is that the researcher will try to force the excerpts into catego­ries, and the categories into themes that he or she already has in mind, rather than let them develop from the experience of the participants as represented in the interviews. The reason an interviewer spends so much time talking to participants is to find out what their experience is and the meaning they make of it, and then to make connections among the experi­ences of people who share the same structure. Rowan (1981) stresses the inappropriateness of force-fitting the words of participants into theories derived from other sources.

There is no substitute for total immersion in the data. It is important to try to articulate criteria for marking certain passages as notable and selecting some over others in order for the process to have public cred­ibility. It is also important to affirm your judgment as a researcher. You have done the interviewing, studied the transcripts, and read the related literature; you have mentally lived with and wrestled with the data, and now you need to analyze them. AsJudi Marshall (1981) says, your feeling of rightness and coherence about the process of working with the data is important. It is your contribution as the researcher.

Source: Seidman Irving (2006), Interviewing As Qualitative Research: A Guide for Researchers in Education And the Social Sciences, Teachers College Press; 3rd edition.