We have had several PhD's from Sheffield that have been done with small N's. Likewise, my own PhD from Clark University, focused on a few (maybe 15, I can't remember) in- depth case studies (though I also did a postal survey as well with lots of respondents--I found the case studies to be much more interesting than the material from the questionnaire, and indeed the in depth material is where all the publications have come from, but that is another issue). I have also published work based on just two in depth examples. It depends on what you are aiming to do with the material and what kinds of questions you are asking. Susan Hanson used to say it isn't the number of respondents that matter is is how variable the population is. So an n of 1 in a homogenous population is going to capture *all* the variability. Having said this, in depth research with few respondents is typically not about trying to identify central tendency, nor is it about capturing *all* the possible variations but is instead about trying to understand how certain issues/concepts/problematics play out in peoples lives. These ideas are all reflected in the Critical Realism methods debates mentioned below (Massey and Meegan had a great book called politics and method in the late 80's that talked about the different advantages of qualitative and quantitative methods for doing political economy research--there is a chapter by Sayer in it). Also the paper by Flyvbjerg that Tim mentions is one I often refer students to with regard to sample size.
Sorry a bit long winded, but this is one of those things that gets my goat, so to speak....
Megan Blake
Senior Lecturer in Human Geography
University of Sheffield, Sheffield S10 2TN
Adjunct Associate Professor of Geography
Hong Kong University, Hong Kong
|