Measuring industrial variation in racial discrimination using log-linear models

This paper modifies previous research on discrimination by measuring different aspects of discrimination across 39 industrial segments of the labor market during an interesting period—the late 1960s. Operating within a standardization framework, we show how simple functions of multiplicative log-linear parameters can be meaningfully interpreted as measures of racial discrimination across and within industry categories. Our results indicate that there is substantial variation across industries in the degree and kind of discrimination.