Estimating nonlinear models with multiply imputed data

Repeated-imputation inference (RII) techniques for estimating nonlinear models with multiply imputed data are described. RII techniques are used to estimate a logit model using the 1995 Survey of Consumer Finances. RII techniques use all information available in multiply imputed data and incorporate estimates of imputation error. The advantage of RII techniques for analysis of multiply imputed data is that RII techniques produce more efficient estimates and provide a basis for more valid inference. Researchers who do not use RII techniques when estimating nonlinear models on multiply imputed data may incorrectly conclude that some independent variables have statistically significant effects.