The achievement of students who participated in a computer-assisted instruction unit on social studies directionality skills was examined in relation to group size. Also analyzed were students’ interaction in small groups and the structure of those groups. Sixty-six second-grade students in solo, pair, or triad treatment conditions were given a pretest and equivalent-form posttest. The computer task concerned cardinal and intermediate directions, which students applied in a computer tictactoe game. Interaction among the students was recorded by audiotape and observer notes. The coding protocol contained the four major categories: giving help, receiving no explanations, receiving explanations, and procedural assistance. Verbatim records of twenty-five randomly chosen students were coded, tallied, and analyzed. For students in pairs or triads (n=55), group structure was analyzed for cooperation, competition, dominance, “odd-man-out” phenomenon, and how the group developed over time. Students made significant gains in their social studies knowledge. Analysis of covariance, with the pretest as the covariate, showed no significant relationship between achievement and group size; achievement and sex; or achievement and ability. For triads, but not for pairs, students in mixed-ability groups had significantly greater gains in achievement than students in uniform-ability groups; this is consistent with prior research. Procedural help was the most frequent type of student interaction. There was no significant relationship between achievement and any of the four major categories of interaction. This is in contrast to the work of Webb (1982a, 1982b, 1982c, 1983, 1985). Because these students were much younger, they probably could not generate effective explanations. Of the twenty-two groups, sixteen were characterized by cooperation and three by competition. Neither ability nor sex was related to cooperation. Ability had no bearing on competition, but boys were more likely to compete. Cooperative interaction was least at the beginning of the session, increased during the middle, and slightly decreased at the end of the computer session. Administrators can utilize the conclusion that students learned in small groups equally as well as when they were alone. Small groups were useful for C.A.I. in that students mastered the content, appeared motivated, and participated actively. / Ed. D. / incomplete_metadata