Abstract
Children in a Headstart program N=113) were assessed on two occasions using a naturalistic observation system and peer sociometric measures. Thirtytwo children were selected by O'Connor's (1969, 1972) convergent criteria of teacher rankings and being below 15% peer interaction frequency. Half of the children saw an experimental modeling film designed to teach children how to initiate entry into peer groups and half of the children saw a control film. The present investigation included methodological elements left uncontrolled in previous investigations. Results indicated no significant multivariate Fratios for treatment main effects, sociometric main effects, or treatment by sociometric interactions. This failure to replicate previous results with the O'Connor film raises serious methodological criticisms of previous work with socially withdrawn children.