Patterns
The patterns chart displays missing value patterns for the analysis variables. Each pattern corresponds to a group of cases with the same pattern of incomplete and complete data. For example, Pattern 1 represents cases which have no missing values, while Pattern 33 represents cases that have missing values on reside (Number of people in household) and address (Years at current address), and Pattern 66 represents cases which have missing values on gender (Gender), marital (Marital status), address, and income (Household income in thousands). A dataset can potentially have 2number of variables patterns. For 10 analysis variables this is 210=1024; however, only 66 patterns are represented in the 1000 cases in the dataset.
The chart orders analysis variables and patterns to reveal monotonicity where it exists. Specifically, variables are ordered from left to right in increasing order of missing values. Patterns are then sorted first by the last variable (nonmissing values first, then missing values), then by the second to last variable, and so on, working from right to left. This reveals whether the monotone imputation method can be used for your data, or, if not, how closely your data approximate a monotone pattern. If the data are monotone, then all missing cells and nonmissing cells in the chart will be contiguous; that is, there will be no "islands" of nonmissing cells in the lower right portion of the chart and no "islands" of missing cells in the upper left portion of the chart.
This dataset is nonmonotone and there are many values that would need to be imputed in order to achieve monotonicity.
When patterns are requested a companion bar chart displays the percentage of cases for each pattern. This shows that over half of the cases in the dataset have Pattern 1, and the missing value patterns chart shows that this is the pattern for cases with no missing values. Pattern 43 represents cases with a missing value on income, Pattern 30 represents cases with a missing value on address, and Pattern 20 represents cases with a missing value on marital. The great majority of cases, roughly 4 in 5, are represented by these four patterns. Patterns 14, 60, and 56 are the only patterns among the ten most frequently occurring patterns to represent cases with missing values on more than one variable.
The analysis of missing patterns has not revealed any particular obstacles to multiple imputation, except that use of the monotone method will not really be feasible.