Describing occupational segregation in sparse and incomplete arrays

The study of sex segregation is increasingly based on log-multiplicative and related models that allow analysts to characterize the amount and structure of segregation independently of (a) the mix of occupations in the economy and (b) the relative size of the male and female labor forces. Although these models are elegant and powerful, methodological complications can arise when the segregation arrays are sparse and small occupations frequently show up as perfectly segregated (i.e., all male or all female). The authors develop a general approach that makes it possible to analyze such sparse arrays with log-multiplicative and related models.