AN EXAMINATION OF LEAST-SQUARES REGRESSION MODELING OF CATASTROPHE-THEORY

Although catastrophe theory appears to have substantial heuristic value in many areas of psychology, its acceptance has, to some extent, been hampered by the lack of a well-developed analytic framework. Two recent articles by Guastello (1987, 1988) suggest that a change-score least-squares regression model provides a method for testing catastrophe theory models. This article reviews these regression analysis procedures. The problems in the approach are detailed, then demonstrated with computer-generated data sets. The results show that such regression approaches cannot adequately distinguish between data that has arisen from a true catastrophe model and data from a true linear model.