Measuring Disparity over Inherently Ranked Groups
*Elsie Pamuk, Centers for Disease Control and Prevention
This presentation discusses the conceptual issues and methodological approaches to measuring disparity for domains with ordered categories - primarily domains that reflect some aspect of socioeconomic status, such as education and income. Justification is provided for using measures that retain the inherent ordering of the categories and weight by category size. Two separate approaches – regression measures and concentration indices – are described and compared to more traditional measures of disparity, using examples that involve comparisons of different populations and the same population at two points in time. The correspondence between regression-based measures and concentration indices is discussed, as well as practical considerations for choosing between these two approaches.