Discrete Variables

Social science data commonly includes discrete variables, and we often use special models, such as logit, probit, and other limited dependent variable models, to deal with them. As it turns out, much evidence in the literature (discussed in our paper) indicates that the multivariate normal model used in ${\mathfrak{A}melia}$ usually works well for the imputation stage even when discrete variables are included and when the analysis stage involves these limited dependent variable models. Nevertheless, ${\mathfrak{A}melia}$ includes some limited capacity to deal directly with ordinal and nominal variables, which we now describe. In general, nominal variables must be declared to ${\mathfrak{A}melia}$, whereas ordinal (including dichotomous) variables often need not be, as described below. (For harder cases, see Schafer, 1997, for specialized MCMC-based imputation models for discrete variables.)