Evaluation of Statistical Methods for Data Collection and Analysis on Racial and Ethnic Minorities and Other Hard-to-Reach Populations

Marc Elliott, RAND
Brian K. Finch, San Diego State University

The ability to assess health disparities among relatively small race/ethnic groups is a major area of emphasis for the Department of Health and Human Services. Our project attempts to address this concern, having carried out a systematic literature review which suggests that while field methods may have limited payoff—in terms of increasing effective sample size—the largest and potentially most useful payoffs may result from sampling and analytic techniques. As such, the feasibility study phase of our project will consist of several sampling techniques including: geographic over-sampling (individual-based, micro-clusters, and/or macro-clusters) and overlapping frame designs (surname lists). We will also pilot analytic strategies such as “borrowing strength” from contiguous years to improve the efficiency and accuracy of estimates. While our ultimate goal is to increase effective sample size, careful attention will be paid to trade-offs with cost, end-user capability, and feasibility for current national health data sets.