Data from the 1969-70 and 1970-71 Head Start Planned Variation (HSPV)Study were used to examine program-child interactions. An effort was made to determine whether different preschool programs have different cognitive effects on different types of children. Seven hypotheses for the analysis of the data were generated from the results of the HSPV 1969-70 study; these hypotheses, which were tested on the 1970-71 data, involved the variables of ethnicity, socioeconomic status, sex, age, initial achievement, prior preschool experience, the response style variables of passivity and competence, and the dimension of program "directiveness". General findings are presented for each of the hypotheses. The concluding discussion focuses on three areas: (1) the relative importance of interactions of model with different variables; (2) the usefulness of various model groupings in predicting and interpreting interactions; and (3) the overall importance of the interactions of child characteristics and model in explaining cognitive outcomes. Due to limitations in the HSPV experimental design, the conclusions are presented as guidelines to profitable areas of inquiry rather than study findings. (SDH)