A Larger-N, Fewer Variables Problem? The Counterintuitive Sensitivity of QCA

Qualitative & Multi-Method Research, Spring 2014, Vol. 12, No. 1

9 Pages Posted: 23 Jan 2015

See all articles by Chris Krogslund

Chris Krogslund

University of California, Berkeley

Katherine Michel

University of California, Berkeley

Date Written: January 20, 2015

Abstract

Studies employing Qualitative Comparative Analysis (QCA) often analyze a relatively small number of cases (n) to assess the impact of a substantial number of variables (k) on a given outcome. In the tradition of writing on the comparative method and multi-method research, however, this ratio of cases-to-variables is viewed as an analytic problem. In this exploratory research note, we raise questions about the implications of this ratio for the stability of findings in QCA. One common method of assessing result stability is the “drop-one” sensitivity test, which repeatedly reruns a particular analysis, each time dropping a single case. We find that, for the number of cases to which analysts most routinely apply QCA, this type of sensitivity analysis produces paradoxical results. Directly contrary to the standard expectation that more robust findings emerge with a higher n/k ratio, we discover that QCA findings based on a lower n/k ratio prove to be more stable. This result calls into question the validity of the drop-one test as a sensitivity metric for QCA.

Keywords: Qualitative Comparative Analysis, qualitative methodology, sensitivity tests, simulation methods

JEL Classification: C10, C15, C63

Suggested Citation

Krogslund, Chris and Michel, Katherine, A Larger-N, Fewer Variables Problem? The Counterintuitive Sensitivity of QCA (January 20, 2015). Qualitative & Multi-Method Research, Spring 2014, Vol. 12, No. 1. Available at SSRN: https://ssrn.com/abstract=2552940

Chris Krogslund

University of California, Berkeley ( email )

Katherine Michel (Contact Author)

University of California, Berkeley ( email )

No Address Available

Register to save articles to
your library

Register

Paper statistics

Downloads
83
Abstract Views
653
rank
303,521
PlumX Metrics