Criterion data were collected from 1,354 boys in 87 high schools throughout the country in four waves starting in the fall of 1966, and from principals, counselors, and teachers of each of the schools. A parallel prediction model was applied to subgroup analyses. Each subgroup's shifts over time were plotted. The value of raw change scores was doubtful at best. Raw change scores may greatly facilitate analysis, as long as they are interpreted with awareness of the potential for bias. The relationship between raw change and subgroup levels provided consistent indicators of differential subgroup shifts. The early identification of subgroups, then, has a facilitating effect in longitudinal studies. Tables, figures, bibliography, 3 appendixes, index