The previous two subsections explained how to estimate the confirmatory factor model when the observed variables represent ordered categories. However, in order to focus specifically on the issue of categorical indicators, all cases with missing observations on at least one indicator were dropped, reducing the original sample of 1200 to 1160. It is possible to maximize the information available in the raw data file using Mplus by adding an ANALYSIS statement specifying TYPE = missing h1, which has the effect of using pairwise rather than listwise deletion for missing observations. Thus rather than losing all information about cases with missing data on a single variable, correlations will be estimated using all cases with complete observations available on both variables (that is, even if there is missingness on a third variable). The Mplus syntax is the following:

TITLE:           Two Factor Model with Categorical and Missing Data;
DATA            FILE IS values_ord_miss.dat;
VARIABLE:   NAMES ARE privtown govtresp compete
                    homosex abortion euthanas;
                    CATEGORICAL ARE privtown govtresp
                    compete homosex abortion euthanas;
                    MISSING ARE all (-1)
MODEL:       economic BY privtown govtresp compete;
                    morals BY homosex abortion euthanas govtresp;
OUTPUT:      standardized;

The full output will not be displayed, but the results are summarized in Table 4 below.