Entry for 17 November 2007
Sue Wheeler, at the University of Leicester, invited me down to do a day-long workshop on qualitative data analysis. This sounded interesting, and seemed like a good way to prepare the two-day qualitative research workshop John McLeod and I are doing next month, so I accepted. At dinner on Friday night with Sue, Julie Folkes-Skinner and others, I rashly promised Sue that I would present a list of “Qualitative Analysis secrets" at the day training on Saturday. I have had a feeling for some time that there is set of basic understandings about good practice in qualitative research that students in general and most folks in the UK don’t seem to share with North American qualitative researchers like David Rennie and I. The following is what I managed to come up with between dinner on Friday and breakfast on Saturday and then used as a kind of theme for the workshop, which turned out to be a great success. I’m sure that the list is incomplete, and that it could use sharpening, but at least it is a start:
I. Nine Truths about Qualitative Research in General:
A. On how hard it is:
1. Qualitative Research is harder (more work and more difficult) than quantitative research.
2. A qualitative research study is only as good as you are (as an interviewer and analyst).
3. If you don’t have good listening/empathic exploration skills, or if you aren’t logical and good with language, you should be doing quantitative research instead.
B. On Research Questions:
4. A clear definition of your topic and a clear statement of your research question is the most important requirement of good qualitative study. Everything follows from that!
5. Qualitative research is really good at answering open (exploratory) research questions. Conversely, if your main research question is closed (confirmatory), you should be doing a quantitative study instead.
C. Diversity and sameness:
6. Everyone does qualitative analysis differently. This is good.
7. The Brand Name Problem. Brand names for qualitative research are social fictions. Empirical phenomenology, Interpretative Phenomenological Analysis (IPA), Grounded Theory Analysis (GTA), and Consensual Qualitative Research are more similar than they are different. In fact, they refer to different but overlapping research traditions rather than different methods.
D. On Grounded Theory:
8. Grounded Theory Analysis. No one knows what Grounded Theory Analysis really is. Most is what is called GTA is based on Rennie, Elliott, Pigeon and others' construction of what they thought Glaser & Strauss (and Corbin) were talking about.
9. Consensual Qualitative Research (CQR) was not invented by Clara Hill; it is a form of GTA constructed by Elliott in the late 1980’s to deal with positivists (see #8).
II. Twelve Secrets about Qualitative Analysis:
A. On short cuts:
1. Selecting the bits you find interesting from a transcript is not qualitative analysis; it’s journalism. Qualitative analysis holds itself responsible for all meaning units. Journalism is a noble profession, but it's not qualitative research.
2. Computer software, such as NVIVO or MaxQDA) can be useful for housekeeping (i.e., counting categories, keeping things neat), but does not do qualitative data analysis; only people can do that.
3. The easiest, cheapest qualitative analysis software is Word, configured with two windows (one for your data, the other for your analysis). It won’t do your housekeeping, though; you’ll have to do that yourself.
B. On Domains vs. Categories:
4. Finding “categories” that correspond roughly to your interview schedule questions is not an analysis. If you stop there, you have wasted everyone’s time, including your informants’.
5. A category tells you something specific in answer to one of your research questions (=substantive categories).
6. But it can be very useful to divide your data up into broad, organizing domains (e.g., context, client contributions). These often but not always correspond to your research or interview topics. GTA call these formal categories, but they are not real categories, because they don’t tell us anything new about the topic.
C. On Categories:
7. Stomach coding: When you read your data, pay attention to how it feels in your gut (or wherever you feel things). When you make a new category or code a piece of data into a new category, make sure your stomach agrees with it. (Credit for this one to Gendlin and Rennie.)
8. The 37 category problem: Avoid the unnecessary multiplication of categories (Occam’s razor). Don’t let your categories multiply like rabbits until they overrun your analysis. (Also known as the flat, boring analysis problem.)
9. Analyses aren’t democracies (or large groups): Some categories are bigger/ more abstract/ broader/ more important than others. Stack them up like coat hanger trees or organizational charts, 3, 4, 5 or even 6 deep.
10. Constant comparison (GTA): Every time you come to a new meaning unit or add a new category, compare it to all your other meaning units/categories. (This can be tedious at first but becomes easy as your category system develops. Use your stomach to help with this!)
11. The rule of four: Whenever you get 4 or more categories at a particular level in your analysis, look to see how they relate to each other:
•They might go under an higher order category;
•They might form a sequence/narrative;
•They might go on some kind of dimension or continuum;
• … or maybe not.
But at least check!
12. Make a picture, flow chart or table that tells a story with your categories.
No comments:
Post a Comment