Personal and environmental variables that characterize students who are involved in different types of interactions in different types of classroom settings were studied. Organizational, behavioral, task, response, and initiation interactions were studied in four eighth-year science classrooms using both student-centered and teacher-directed curricula. Three teachers and 108 students in a large metropolitan secondary school in Brisbane (Australia) participated in the study. The study was conceptualized within the Lewinian B=f(P,E) framework (K. Lewin; 1936, 1951), and involves multivariate analysis of selected "P" variables: student personality, motivational, and cognitive characteristics, and "E" variables: curriculum materials and characteristics of the learning environment. Attempts to interpret the significant regression coefficients by referring to the original raw data indicate that the significance is an artifact of the regression model trying to accommodate extreme scores. Whenever data from complex environments are to be analyzed, the techniques applied must not attempt to fit the data into an articifical model. Alternative modeling procedure may be more appropriate. Thirteen tables and 8 figures present study data, and there is a list of 24 references. (SLD)