Abstract
The accuracy with which 4 small group leaders could rank order the relative interaction frequency and speech duration of their group members was evaluated by comparing their rankings to criterion data provided by a highly reliable group-interaction recording system. It was demonstrated that the leader judgments were reasonably accurate, with leaders evaluating duration more accurately than frequency. The data were discussed in terms of the acceptability of leader judgments as a data source in screening studies. However, a non-participant observer data source was more accurate.