Generalizing from an unique case? : Outline of an idiographic theory of samples

Although qualitative methods are now recognized in the methodological literature, they are still regarded with skepticism by some methodologists, mainly those with statistical training. One reason for this skepticism concerns whether qualitative research results can be generalized, which is doubted not only because they are derived from only a few cases, but also because even where a larger number is studied these are generally selected without observing the rigorous criteria of statistical sampling theory. In this regard, the methodology textbooks still distinguish samples into two types: probability samples (simple random, systematic, proportional stratified, non-proportional stratified, multistage, cluster, area, and their various combinations), and non-probability ones (haphazard or convenience, quota, purposive, of the emblematic case, snowball, telephone). About the latter is stated: the obvious disadvantage of nonprobability sampling is that, since the probability that a person will be chosen is not known, the investigator generally cannot claim that his or her sample is representative of the larger population. This greatly limits the investigator's ability to generalize his or her findings beyond the specific sample studied (…) A nonprobability sample may prove perfectly adequate if the researcher has no desire to generalize his or her findings beyond the sample (Bailey, 1978: 92) This position again tends to relegate qualitative research to the marginal role of furnishing ancillary support for surveys, which is precisely as it was conceived by Barton and Lazarsfeld (1955) and the methodologists of their time. The aim of this study is to show that this methodological denigration of qualitative research is overly severe and unjustified, for three reasons. Firstly, because also the use of probability samples and statistical inference in social research often proves problematic. Secondly, because there are numerous disciplines, in both the social and human sciences, whose theories are based exclusively on research conducted on only a few cases. Thirdly because, pace the methodological orthodoxy, a significant part of sociological knowledge is idiographic. My intention is therefore not to criticize sampling theory or its applications; rather, it is to remedy a situation where statistical inference is deemed the only acceptable method, and idiographic generalization as scientifically ill-founded. Finally qualitative researchers do not need to throw away the baby generalization with the bathwater of probability sampling, because we can have generalizations without probability.