Dissertation Diary #7: Qualitative Sampling
A how-to guide for selecting cases, with a *sample* (pun intended) of my dissertation
One of the first steps of any research study design is thinking through sampling. Not that kind of sampling (unfortunately), but the pool of cases from which you will extract data and make conclusions about the phenomenon of interest you’re studying.
Quantitative studies (typically) aim to have a statistically representative sample of some larger population; this goal requires a large amount data, ideally randomly selected from the population of interest.
Qualitative studies, on the other hand, depend on purposeful sampling1, which has the goal of selecting cases rich with information pertaining to the research interests of the study. The goal here is not empirical generalizability, but rather conceptual generalizability2 — that the patterns and conceptual framework uncovered through the in-depth investigation of the selected cases can be useful for understanding the phenomenon of interest in other contexts.
There is a wide variety of strategies for purposeful sampling — which one is best depends on your research questions. Generally, though, you’ll need to decide whether you want to learn from typical or atypical cases. Typical cases, or observations that match the normative, ordinary experiences of a population of interest, present an ideal context for evaluating prior conceptual findings. Atypical cases, on the other hand, investigate little-known contexts, which offers the opportunity for scholars to problematize existing systems of categorization, illuminate relationships that are ordinarily unseen in most settings, and highlight outliers that reveal a valuable phenomenon or perspective. Once you decide on typical vs atypical cases, the next key step is to strategically structure in variation — as Collins and colleagues write, “A central tenet in the scholarly enterprise is that generating explanations requires variation” (2024:31). Within your case set, how can you ensure there are meaningful differences that can support your conceptual uncoverings?
For my dissertation, I chose atypical cases. The “typical” situation for a traditional neighborhood public school is that the school would serve students who live in the assigned attendance zone surrounding the school building. The “atypical” scenario, then, is a traditional neighborhood public school that serves a student body that primarily comes from neighborhoods that are not the assigned attendance zone. By selecting elementary schools that have fewer than 50% of the neighborhood children attending their school *and* less than 50% of the enrollment coming from the assigned neighborhood, I investigate atypical schools with the goal of improving the explanatory power for answering my driving question about the impact of geographic diversity on the work of educators in traditional neighborhood schools.
I designed my case selection to maximize the diversity of experiences in order to ensure variation. Specifically, I selected schools that varied in location (i.e. what section of the city they were in — West Philly, South Philly, Northeast Philly, North Philly, Center City, etc.) and in strength of ties. As I’ve talked about a bit in a previous post, I am also studying the patterns of neighborhood-school ties that emerge in a school choice district. The strength of a neighborhood tie (i.e. how many kids as a percentage of the student body come from a particular neighborhood) may influence the work and culture of a school. Already I am seeing differences between sites, even just with half of the schools formally on board at the writing of this post. A good sign!
But what about the number of interviews total and per school? This is probably the most frequently asked question in qualitative research (by quantitative folks) because it is so different than quantitative sample size calculation — and, admittedly, seemingly hand-wavey at first blush. The goal in a case model is to reach saturation3. The number of interviews needed can’t be known until the study is completed, as you’ll need to continue interviewing people until you are no longer learning anything new or surprising. Once you keep hearing the same things over and over again, that’s a sign that you’ve likely reached saturation for your set of research questions.
Since most grant/fellowships/dissertation proposals will still require some estimate number of participants, you need to give a best guess (and then a line about reaching saturation, often citing Small 2009 as I just did above). For my dissertation, I said I would do about 60 interviews — roughly 20 interviews per school. At one school I have done 13 interviews already, and I’m starting to feel the “this is pretty consistent with what I’ve learned thus far” mark for that school. But that doesn’t mean I’m done, because as I increase my count at other schools, there may be things I need/want to follow up with prior participants, or dig into more deeply with new participants from this early school. Qualitative research is beautifully, and painfully, iterative. You really need the flexibility, patience, and grace to go with the flow (so long as you have a strong design in place at the start).
Patton, Michael Q. 2002. Qualitative Research and Evaluation Methods. 3rd ed. Thousand Oaks, CA: SAGE Publications, Inc.
Collins, Caitlyn, Megan Tobias Neely, and Shamus Khan. 2024. “‘Which Cases Do I Need?’ Constructing Cases and Observations in Qualitative Research.” Annual Review of Sociology 50(1):21–40. doi:10.1146/annurev-soc-031021-035000.
Small, Mario Luis. 2009. “`How Many Cases Do I Need?’: On Science and the Logic of Case Selection in Field-Based Research.” Ethnography 10(1):5–38. doi:10.1177/1466138108099586.