Some split file groups with valid data not analyzed by logistic regression

Problem

I have been running the LOGISTIC REGRESSION command (Analyze->Regression->Binary Logistic) in SPSS. I have a split file structure in place with five split file groups. When running a model with one particular dependent variable, the output contained only the Case Processing summary for split file group 1 and the warning: "For split file location = 1.00 the dependent variable has less than two non-missing values. For logistic regression, the dependent value must assume exactly two values on the cases being processed This command is not executed." Examination of the data revealed that the dependent variable was indeed a constant for cases with full data on the predictors in the model. However, the dependent variable was not a constant for valid cases in groups 2 to 5, yet they were not analyzed or even referenced in this output. In the analysis of a different dependent variable, I received a similar warning for split file group number 3: For split file location = 3.00 the dependent variable has less than two non-missing values. For logistic regression, the dependent value must assume exactly two values on the cases being processed This command is not executed. This dependent variable was indeed a constant for valid cases in group 3, whereas it was not a constant for groups 1, 2, 4, or 5. The logisitc regression analysis proceeded for groups 1 and 2 with all of the requested output. Group 3 was not analyzed, as predicted by the warning. However, groups 4 and 5 were not even referenced in the output for this procedure, although the warning did not apply to these groups. Why does the presence of inapplicable data in one split file group, in the form of a constant dependent variable, prevent the analysis from proceeding on all of the subsequent split file groups?