Alternative approaches to CFA

This section presents two alternative approaches to CFA. These two approaches do not just introduce minor cosmetic changes. Rather, they go radically different ways while keeping the main idea of person-level research intact. Both approaches allow researchers to individually test cell frequencies against expected values. The first of the two approaches, proposed by Kieser and Victor (1991, 1999, 2000), uses different methods when estimating expected cell frequencies. Specifically, this approach uses the more general quasi-independence models (cf. Section 10.1, above). The second approach, due to Wood, Sher, and von Eye (1994), and Gutiérrez-Peña and von Eye (2000), views CFA from a Bayesian perspective. This approach allows one to take into consideration prior and subjective information in the search for types and antitypes.