Knowledge that Transforms

Four laboratory studies documented systematic performance illusions for individuals and group members. After completing a task, most individuals typically rank their performance below the median (a negative performance illusion) and most group members rank their group performance above the median (a positive performance illusion). This evaluation gap surfaced in all studies; group members consistently assign their group a higher rank than individuals assign themselves. This gap contradicts a number of theories assuming that group biases merely extend individual biases.

The stability, predictive validity, and construct validity of constraints were investigated in 3 studies. College students' situational constraint scores pertaining to a more specific performance domain ( specific classes) better predicted a similarly specific criterion measure (class grade) relative to constraint scores pertaining to an aggregate performance domain (all classes) and correlated with a global criterion measure (semester grade point average or GPA). In 2 other studies, it was found that self-reported constraints were independent of attributional style. Finally, data from a 3rd study indicated that goal difficulty mediated the constraint-performance relationship. These data suggest that the criterion-related validity of self-report measures of constraints might be attenuated when accounting for global success criteria relative to more specific component success criteria; constraints may be less prone to cognitive-style correlates than heretofore assumed; and the constraint-performance relationship may be mediated by more proximal motivational constructs such as aspiration level.

In this study, researchers examined the effect of task engagement and experience on frequency rating accuracy with an objective measure of task frequency. By contrasting findings from the memory literature on frequency estimation with current task analysis practices, these investigators proposed that task performers and respondents with low levels of experience would generate more accurate frequency estimates compared with task observers and respondents with high levels of experience. Participants were randomly assigned to a task engagement condition (performer vs. observer) as well as to an experience condition (low vs. high). In support of the hypotheses, participants in the performer and low-experience conditions demonstrated greater accuracy in their frequency ratings. These findings were illustrated with several different measures of accuracy. Implications for task analysis research and practices are discussed.