Charting the Future of Configural Frequency Analysis II: Developments and Applications

This special double issue, edited by Erwin Lautsch and Alexander von Eye, gathers recent work on Configural Frequency Analysis (CFA) and splits the contributions into two groups. The first group deals with CFA as a statistical method and how it is being refined. The second collects studies that put the method to work on real data drawn from psychology and other behavioral sciences.

Von Eye and Lautsch open with an overview of where CFA stands and where it may be heading. Joachim Krauth follows with two pieces: one examines paradoxes that turn up in multidimensional contingency tables and asks what those paradoxes imply for CFA, while the other applies two-sample CFA together with Markov-chain analysis to questions in psychotherapy research.

Several contributions compare CFA against neighboring techniques. Von Eye and G. A. Bogat set logistic regression and prediction CFA side by side, and von Eye, Patrick Mair, and Bogat lay out prediction models built specifically for Configural Frequency Analysis. The question running through these papers is practical: when the aim is prediction, what does a configural approach add, and where does ordinary regression do the job just as well?

Sparse, large cross-classifications get their own attention. Von Eye and E. Gutiérrez Peña take up CFA for exactly that kind of data, and a companion paper by von Eye weighs the various measures available for analyzing large, sparse tables. On the computational side, M. Beier describes an algorithm and a tool for working out exact conditional probabilities of configuration frequencies, and S. von Weber, von Eye, and Lautsch present a combinatoric search procedure for finding types and antitypes.

The applications half begins with Lautsch and M. M. Plichta asking whether CFA and Latent Class Analysis produce findings that complement each other or simply duplicate them. M. M. Wagner-Menghin uses CFA to pin down achievement-motive types, revisiting a 2003 study of her own to see whether the earlier results hold up on replication.

Change detection in small samples is the focus for M. Ising and W. Janke, who apply paired-samples CFA to track multivariate shifts where sample sizes are too small for many standard tools. M. Stemmler and G. Bäumler hunt for types among decathletes, grouping athletes by their performance profiles across the ten events rather than by a single summary score.

A larger developmental study by Stemmler, Friedrich Lösel, A. Beelmann, S. Jaursch, and B. Zenkert looks at problem behavior in children across kindergarten and primary school, comparing what prediction CFA finds against what multiple regression reports on the same data. Closing the collection, Lautsch and U. Thöle identify and analyze personality types using a short measure of the Big-Five domains.