PRINCIPAL COMPONENT ANALYSIS OF BEHAVIORAL-DATA - A CASE-STUDY OF PREVIOUSLY PRESENTED DATA

Principal component analysis (PCA), like factor analysis, is a mathema tical method suitable for reducing many data variables into a few more comprehensible factors. In psychiatric research this method is increa singly taken into account, as it is now available in several PC versio ns. In the present article PCA is performed on a previously presented behavioral study, to demonstrate the method step by step. Another aim is to evaluate the profit obtained, by the psychiatric professional, f rom a more thorough statistical analysis. The data set used originates from a child-adult interactive study with 18 children classified into 4 diagnostic categories. The relative duration of a selected number o f defined behavioral elements, recorded from videotapes, was subjected to the analysis. The PCA was found to be very useful and informative, even though the number of cases was limited. The previous conclusions were still valid, although some specifications were possible. When th e method is fully understood, it is a powerful tool to discern differe nces in ethiologic data investigations such as these. It may be used i n any exploration and presentation of investigations with multiple var iables in which common trends are expected.