Cautions on the Use of Multiple Imputation When Selecting Between Latent Categorical versus Continuous Models for Psychological Constructs