Respondent Attrition in Longitudinal Analysis of Panel Data

• In Event: 18.047 - Statistical Research in the Contexts of Propensity Scores, Longitudinal Models, and Bayesian Analyses

Abstract

Respondent attrition is a common problem in national longitudinal panel surveys such as the Early Childhood Longitudinal Study or the Longitudinal Study of American Youth. To make full use of the data, panel weights are provided to account for respondent attrition. Panel weight adjustments for nonresponse typically do not make use of information obtained from early waves in the survey program, which can lead to bias of growth model parameter estimates. Another method to address potential bias from nonresponse is full information maximum likelihood (FIML), which uses responses obtained during all data collection waves regardless of missing data. This study shows that FIML performs better in recovering parameter estimates when addressing nonresponse in a typical longitudinal data analysis.

Authors

• Daniel Y Lee, University of Maryland - College Park

• Laura M. Stapleton, University of Maryland

• Jeffrey R. Harring, University of Maryland