The purpose of this study is to investigate the feasibility of using model-based imputation methods for record nonresponse in a longitudinal survey. Record nonresponse means that the responses to an entire set of questions (record type) are missing for a wave. In this study we have selected four variables to model and impute: (i) receipt = receipt of earnings; (ii) w pay = weeks worked with pay; (iii) earn = earnings amount; and (iv) maid = Medicaid coverage. Maid is on the person (P) record and the others on the wage and salary (WS) record. For any wave, a person may respond to neither record type, to P, or to both. So the first three variables are reported or missing simultaneously and maid may or may not be missing at the same time as the others.