OCCUPATIONAL SEGREGATION IN THE MULTIDIMENSIONAL CASE - DECOMPOSITION AND TESTS OF SIGNIFICANCE

A new multidimensional version of the G-segregation index is developed and applied to the study of occupational segregation. U.S. Current Population Survey data are used to measure the difference in occupational segregation between races as well as the change between time periods. Decomposition of the difference (change) into 'occupation mix' and 'gender composition' components indicates the contribution of each factor. Because these inequality measures are computed from sample data, distributional information required to test hypotheses is lacking. Two computer-intensive methods for estimating the distributional properties are demonstrated. The approximate randomization and bootstrap methodologies are used to test for statistically significant differences in segregation between races and for changes over time. In addition, the components of the decomposition are examined for statistical significance.