It has been hypothesized that current methods of determining pay rates value the characteristics of jobs held primarily by men differently than the characteristics of jobs held primarily by women, resulting in lower earnings for women. A policy capturing approach using numerically rated job characteristics (PAQ data) was applied separately to the earnings of both men and women to determine which work behaviors were predictive of the earnings of each group. Resulting "male" and "female" job evaluation models were found to place occupations in essentially the same pay hierarchy, casting doubt on the hypothesis. It appeared that current systems in general use for valuing occupations contained little, if any, gender bias and were relatively neutral in their effects on the earnings differential. However, women were found to have lower representation in the higher paying positions within occupations and to be employed in somewhat greater proportions in occupations which paid less. Younger women were also found to earn about the same as older women while older men earned substantially more than younger men. Consequently, the earnings differential increased with worker age. (Author)