Detecting and Correcting Attrition Bias in Longitudinal Family Research

Bias due to the attrition of respondents poses a threat to the internal and external validity offind- ings in longitudinal family research. Differences in characteristics between two samples at initial and subsequent waves is a threat to external va- lidity, and differences in the relationships be- tween variables in the samples is a threat to inter- nal validity. Methods of detecting attrition bias are discussed, and Heckman's procedure to cor- rect attrition bias is presented. Data from the University of Southern California Longitudinal Study of Generations (N = 2,044) are used to il- lustra