Generalizing From Single Case Studies

Abstract

Although studies of single cases occur throughout research fields, they are only recently beginning to appear in evaluation. These studies are difficult to generalize from, because statistical techniques do not apply. This article offers a variety of suggestions for logically analyzing the relationship between a single case and a population so that reasonable generalizations may be possible. Methodologies used by clinicians and by court justices are also examined for their relevance to evaluation methodology.

What Is the Use of a Single-Case Study in Management Research?

“…(This limitation is considered even more severe if the study involves a single case.) Often this is expressed as absence of external validity -the impossibility of extending the case study findings to a population of other cases (Donmoyer, 1990;Kennedy, 1979;Yin, 2009). Supporters of single case studies have developed a series of arguments -to be examined below -to solve the problem of "a sample of one" but none of the arguments has convinced those in favor of generalization by means of multiple cases or statistical samples.…”

“…Authors such as Donmoyer (1990), Kennedy (1979) and Yin (2009), believe increasing the number of data points in a single case would eliminate the problem. However, even with a larger number of data points, the single case remains unique (Ruddin, 2006).…”

“…He is the one who reads the case and decides whether it can be applied to his/her case (be "generalized") or not. According to the legal tradition, the judge's decision is guided primarily by four attributes: i) whether the material facts of the two cases are similar, ii) whether the decision taken in the previous case would still be fair considering possible changes in context, iii) reasons for a particular decision and iv) the level of generalization upon which the decision was made (Kennedy, 1979). Kennedy also points out that, while many researchers are not accustomed to the idea of leaving generalization to the reader of the case study, this is common practice, particularly in the legal and clinical fields.…”

Applying Enterprise Architecture to Business Networks

“…Kennedy [1979] proposed two approaches for cross-case analysis: a case-survey approach and a case comparison approach. Huberman and Miles [2002] indicated that cross-case analysis may carry the risk of achieving premature or false conclusions.…”

“…One way to the manage this is to analyse data from different perspectives. Kennedy [1979] proposed two conditions to conduct the comparison; first, isolated factors within particular case studies must be worthy of substantive attention; second, the number of case studies must be large enough to warrant cross-case tabulations. When these conditions are in place, cross-case comparisons can be made to establish cross-case patterns.…”