Multiple imputation prior to model estimation was used to estimate the very minimal missing data. Five data sets were imputed using the EM algorithm in NORM (Schafer, 1999), and the IMPUTATION option was used in Mplus to repeat the analyses for each of the data sets and produce parameter estimates that were averaged over the five sets. Standard errors were computed by combining the average of the squared standard errors across the set of analyses and the parameter variation between analyses.
