Abstract
We used a policy-capturing approach to study 3,104 pay allocation decisions of 97 managers (54 men, 43 women) working for a large nonprofit organization with a pay-for-performance policy. Subordinates' performance, the consistency in subordinates' past job performance, the importance of the subordinates' jobs in meeting managerial goals, and the degree of disruption that would occur if subordinates quit significantly affected managerial pay allocations. A major proportion of the variance across the managers was accounted for by sampling error and criterion unreliability. The subjects' explicit rankings of the four subordinate-related factors did not correspond highly with the rankings of their regression weights.