Effects of Misclassification of Race/Ethnicity Categories in Sampling Stratification Affects Survey Estimates

Misclassification of race/ethnicity occurs when there is a discrepancy of classifications based on two different sources, e.g., administrative data and reported values. As a result of this error, a sample designed to meet analytic objectives could, when implemented, result in the loss of effective sample sizes in key domains involving the race/ethnicity group. In assessing misclassification error in the race/ethnicity category, the true values and the misclassified values are established. In practice, the true value is often unknown and can only be assumed. In our study, we assessed whether the misclassification of race/ethnicity occurred during the sampling frame construction, assuming that the data obtained from the respondents is more accurate than the frame and will serve as the true values. This assumption is aligned with survey practice where estimates for race/ethnicity are often derived based on reported values rather than the frame values. We estimate the misclassification matrix in which misclassification parameter/proportion can be calculated with the usual weighted survey estimate. We also investigated the impact of misclassification on survey estimates (weighted totals), where these estimates were produced using the weights that had been raked into three different marginal population totals.