A total of 112 female teachers and student teachers acted as participants or observers in an experimental situation in which the participant taught a simulated elementary school child a mathematics lesson for three 5-minute trials while the observer watched. The child's performance supposedly either improved over trials (Low-High), deteriorated over trials (High-Low), or ramained stable, but low (Low-Low). Contrary to prediction, participants appeared to attribute change in the child's performance (improvement or deterioration) to themselves. Participants were somewhat more likely to attribute poor performance to situational factors than observers were. However, both groups attributed low or descending performance to situational factors more often than they did high performance. The Low-High child was evaluated higher on measures of sentiment and skill than the High-Low child, who in turn was evaluated higher on sentiment, skill and effort than the Low-Low child. (Author/SET)