Charting the future of Configural Frequency Analysis II: developments and applications

In 2003, Lautsch and von Eye edited a special issue of this journal, in which the past development of Configural Frequency Analysis (CFA; Lienert, 1969; Lienert & Krauth, 1975; von Eye, 2002; Lautsch & von Weber, 1995) was described, and current trends of development were charted. Only two years have past, and the development of CFA is proceeding at a rapid pace. In the preface to the 2003 special issue, the editors listed six fields of current research on the method of CFA. These fields were (1) simulation studies on the behavior of CFA statistical tests; (2) studies concerning the dependency structure of CFA tests; (3) studies concerning the size of tables that can be analyzed using CFA; (4) statistical bases of CFA; (5) reformulation of existing models of CFA; and (6) improvement of computer programs. It certainly comes as no surprise that, after only two years, these fields are still being actively researched. It is impressive to see the large number of results. Many of these results are presented in the current special issue. This special issue contains 14 articles. It is divided into two sections. The first is devoted to articles in which new methodological questions are asked, and new methods and tools are presented. It contains eight articles. The second is devoted to applications of CFA. It contains six articles. The first, the methodological section begins with an article by Krauth on the analysis of statistical paradoxes using CFA. This article is important in three respects. First, it shows that CFA can fruitfully be applied to data structures that are paradoxical from certain perspectives. Second, it shows that, under particular conditions, CFA allows one to identify the paradoxical structures. Finally, it also shows that, when the paradoxial aspects in a data set are unknown, CFA may not be sensible to all of these aspects, thus indicating that CFA, just as any other method of analysis, is not a cure-all, but should be applied keeping its limitations in mind. The second article, also by Krauth, tackles the important problem of statistical analysis of success in psychotherapy. The author proposes combining methods of two-sample CFA with methods of first-order Markov modeling. The article shows that and how methods of CFA can be used to analyze the dependence structure of longitudinal data.