Although the new evaluation system did not yield a shift in job scores more favorable to women, it is possible that alternative weighting schemes might have done so. As noted earlier, Treiman suggested that under certain conditions, changing the weights assigned to various factors may substantially alter the pay relationship between jobs, and that this would affect men and women differentially. This possibility is explored in this chapter.
The conditions identified by Treiman ( 1984) are two. First, different weighting schemes will make a difference only if the job content--i.e., the functions or work performed--of male- and female-dominated jobs varies in such a way that the former score higher on some factors, while the latter score higher on other factors. If this condition holds, assigning more weight to factors on which female-dominated jobs score higher and less weight to factors on which male-dominated jobs score higher may yield job scores that benefit women. Notable shifts in this direction will not occur, however, unless the second condition identified by Treiman holds. That second condition is that factor scores are not highly intercorrelated.
In Treiman's simulation, these conditions were met. In his study, he used variables as proxies for job factors that operationalize job content. He then generated mean factor scores and standard deviations on each factor for each race and sex grouping. The factors he used to describe job content were data, people, things, strength,