Nonresponse in follow-back surveys of ethnic minority groups: an analysis of the Puerto Rican Maternal and Infant Health Study.

Abstract

This study documents the levels and sources of nonresponse in the first large-scale maternal-infant health survey administered to representative samples of Puerto Rican mothers on both the U.S. mainland and the island of Puerto Rico. The data source is the Puerto Rican Maternal and Infant Health Study, which was administered to a vital records-based sample of 2763 mothers of infants. An additional 805 women were nonrespondents. Nonresponse is examined as a function of several characteristics measured from vital records using logistic regression. The response rate for this survey compares favorably to response rates for similar surveys. Although nonresponse is not associated with most characteristics measured from vital records, it is higher among mainland residents and mothers of infants who died. The absence of significant associations is due to opposite relationships between several covariates and the failure to locate and refusal. For example, nonresponse in the birth sample is not associated with migration, despite the difficulty of locating migrants. The lower likelihood of locating migrants is offset by their willingness to participate. Selectivity due to nonresponse is minimal. Nevertheless, researchers who design "binational" surveys should be aware of setting-specific circumstances that affect the ability to locate sampled individuals and secure their cooperation.