Measurement of Race

Epidemiologists often rely on self-reports to measure the value of a predictor or outcome variable. In most cases, the self-reports are proxies, and their accuracy rests on how well they match an objective measure of the variable. However, in the case of a predictor variable like race there is no objective measure to which the self-reports can be compared. A self-report of race can be compared to another measurement of race, but there is no reason why one should be the proper reference standard rather than another. Without a reference standard, there is no way to measure the error in a count of the number of Blacks in the population or the accuracy of a report of a Black-White difference in the rate of a disease. I propose a method of establishing a reference standard for self-reports of race similar to one used in economics for variables like poverty and unemployment. Race, according to my proposal, is not an intrinsic characteristic of members of a population, and a member's race can vary from one disease to another. As a result, a member can be White in relation to an inherited hemoglobin disease but Black in relation to high blood pressure. I show how a relative rather than absolute measure of race can lead to more effective strategies for reducing racial disparities in health.