Researchers often ask subjects to commit considerable time and effort to completing tasks that are not especially enjoyable. In a multistage investigation of sixth-grade boys and their families, we hypothesized that boys who were prone to high levels of distress (i.e., anxiety, depression, low self-esteem, and low well-being) but only low or moderate levels of self-restraint (i.e., consideration of others, impulse control, suppression of aggression, and responsibility) would be particularly unlikely to agree to participate. Consistent with this hypothesis, boys from 33 classrooms who were nominated by their peers as high in distress and moderate or low in self-restraint were significantly less likely than other boys to take part in an in-class survey. In addition, the families of boys who scored high in distress and moderate or low in self-restraint on the Weinberger Adjustment Inventory (WAI) were less likely than other families to agree to an observation session in their homes and to a series of telephone interviews about daily events in the boys' lives. Across the 3 phases of the research, the cumulative attrition rate of these at-risk boys was approximately 80%, compared to only 50% for those low in distress and/or high in self-restraint. These results suggest that the children of greatest interest in studies of social competence and family interaction may often be among those least adequately represented.