Testing cognitive models by a joint analysis of multiple dependent measures


One central aim of psychological research in the field of judgment and decision making is to identify properties of the underlying cognitive processes as well as to develop and critically test cognitive models. This chapter describes the general multiple measure maximum likelihood approach. This approach provides the mathematical foundations for the other more complex methods presented—that is (non-) hierarchical Bayesian methods and multinomial processing trees. Applying hierarchical Bayesian models to both outcome and process measures seem to be a promising avenue for future research. A simplified version of the Bayesian method that shares central parts with the multinomial processing trees (MPT) approach is a maximum likelihood approach that takes into account multiple dependent measures simultaneously. Compared to the extended MPT approach by D. W. Heck and E. Erdfelder, multiple measure maximum likelihood considers response times in a continuous manner and can be estimated per person with relatively few observations.

In M. Schulte-Mecklenbeck et al. (Eds.), Handbook of Process Tracing"
Marc Jekel
Marc Jekel
Post-Doctoral Researcher (Akademischer Rat)

My research interests include coherence-based reasoning, artificial neural networks, information distortion, and methods & statistics.