The aim of this study was to perform multiple dataset imputation in order to depict the observational dataset.
However, the patients who were non-respondents and those who stopped during follow-up differed from the whole-course respondents in a number of aspects, so we used multiple imputation in order to diminish selection bias.
The resultant data set was subjected to column mean imputation in order to fill in the missing trait values, given that the data were missing at random, i.e. not missing over all measures within a battery.
However, regarding the weights and imputations, in order to verify the robustness of our results, we also performed exploratory factor analysis and MG-CFA with the weights (we recall that using weights in the MG-CFA with alignment is not feasible in Mplus).
We used multiple imputation procedures in order to analyse changes in religious participation across all five waves included in our analyses.
Reporting alcohol consumption is strongly related to the sociodemographic characteristics [ 35] that can be included in the imputation model in order to reduce non-response bias.
Missing parental data were a potential source of selection bias, and we performed a multiple imputation procedure in order to obtain complete parental data to help minimise this bias.
Although we recognise the importance of correct specification of the imputation model in order to provide valid results of MI, we focus on whether a detailed description of variables included in the imputation model and information on the imputation method are provided, rather than judging the adequacy of the imputation model.
Previous work (Schafer, 1997; Moons et al., 2006; Sterne et al., 2009; White et al., 2011) emphasizes the importance of including the outcome in imputation models in order to maintain observed relationships between covariates and the outcome as required for regression modeling.
We also assessed the sensitivity of the results after imputation by changing the order of imputed variables, adding two-factor interactions in the imputation models, and by imputing maternal BMI from previous or following pregnancies with known maternal BMI (appendix B and table C of data supplement).
