Case Selection, Case Studies, and Causal Inference: A Symposium
Newsletter of the American Political Science Association Organized Section for Qualitative and Multi-Method Research, Vol. 6, No. 2, pp. 2-16, 2008
15 Pages Posted: 8 Aug 2011
Date Written: September 2008
For scholars concerned with causal inference, how should cases be selected? The statistician David Freedman evaluates case selection strategies proposed by political scientists. He dissents from Fearon and Laitin, who suggest that when case studies are used to deepen findings derived from regression analysis, cases should be selected at random. Freedman also urges caution regarding Gerring’s approach, as he claims Gerring inappropriately proposes to identify cases as typical, diverse, extreme, etc. on the basis of criteria associated with large-N and experimental methods. Freedman likewise disagrees with Goertz’s recommendation about case-selection within the 2 x 2 matrix formed by dichotomous independent and dependent variables. Goertz suggests that analysts should ignore the 'null cell,' in which both the purported cause and the effect are absent, but Freedman argues against this practice, contending that all cells in the matrix can potentially be important. Through this discussion, these authors come to agree on the importance of grasping the strengths and limitations of alternative case-selection strategies.
Suggested Citation: Suggested Citation