Sampling minority groups in health surveys

Abstract

Sampling racial and ethnic minority groups in population-based health studies presents the statistical designer with a variety of notable challenges, many of which can be traced to the manner in which the sampling frame maps into the set of group members. A brief review of sampling approaches aimed at race/ethnic groups is followed by findings on two particularly important statistical issues one may face when sampling minorities: the implications of the geographic diffusion of minorities within sampled area clusters, and the relative utility of different forms of space and time sampling to deal with sometimes dramatic change over time in the linkage between entries on a sampling frame and members of relatively mobile minority groups. Simulated findings from practical illustrations involving the sampling of Hispanics are presented to address each issue.