Greene's finding that children's involvement was higher in the more pupil-controlled classrooms is in apparent conflict with CCEP's expectation that involvement is a measure independent of setting. A resolution is suggested in the study of a set of 20 all-day behavior stream observations of individual children in 3 Follow Through classrooms and one open classroom. In this study, children were observed to be more on-task in settings in which the distribution of controls between the teacher and the pupil was congruent or internally consistent. (The settings in Greene's study with low involvement may have been less congruently structured.) The study coded every 30-second interval of the 20 behavior stream observation records on each of 62 variables. Settings that were congruent on the four variables analyzed had the smallest percentage of off-task behavior. Comparisons between this study and Greene's are made, and three hypotheses are suggested: (1) the best predictor of academic skill and concept achievement in Follow Through classrooms should be the time children spend pursuing goals relating to partial individuation (the internalization of concepts, skills and dispositions) in congruently structured settings emphasizing these goals; (2) the degree of congruence in classroom settings will have something to do with pupil role development; and (3) because pupil role development represents an estimate of a person's individuation, pupil role development may correlate zero or negatively with certain affective estimates of self-concept. (For related documents, see TM 003 067-070.) (KM)