Evaluating the Patterns Table
The tabulated patterns table shows whether the data tend to be missing for multiple variables in individual cases. That is, it can help you determine if your data are jointly missing.
There are three patterns of jointly missing data that occur in more than 1% of the cases. The variables employ (Years with current employer) and retire (Retired) are missing together more often than the other pairs. This is not surprising because retire and employ record similar information. If you don't know if a respondent is retired, you probably also don't know the respondent's years with current employer.
The mean income (Household income in thousands) seems to vary considerably depending on the missing value pattern. In particular, the mean Income is much higher for 6% (60 out of 1000) of the cases, when marital (Marital status) is missing. (It is also higher when tenure (Months with service) is missing, but this pattern accounts for only 1.7% of the cases.) Remember that those with a higher level of education were less likely to respond to the question about marital status. You can see this trend in the frequencies shown for ed (Level of education). We might account for the increase in income by assuming that those with a higher level of education make more money and are less likely to report marital status.
Considering the descriptive statistics and patterns of missing data, we may be able to conclude that the data are not missing completely at random. We can confirm this conclusion through Little's MCAR test, which is printed with the EM estimates.