I have a couple of questions on logistic regression:

1. In the stepwise procedure, I have two predictor variables --> one variable, X, is entered into the model before a second variable, Y. Variable X is a categorical variable (2 level -- 0 and 1) while variable Y is a continuous variable.

In the Type 3 Analysis of Effects, variable Y has a larger Wald chi-square statistic than variable X (both have p-values<0.001).

Is my interpretation below correct (based on the Type 3 Analysis of Effects):

If both variables (X and Y) are in the model, Y is a stronger predictor than X (due to the larger Wald chi-square statistic), keeping X constant. I can't say that X is more important (predictive) than Y because it was entered into the stepwise procedure first.

2. To resolve the issue of some missing data for one of the 2 independent variables (variable X), I created a flag of 1 or 0 for the presence/absence of data.

My question is if I need to include this newly created flag variable in the class statement, or should I just include it in the regression equation (model flag (event=1)=constant + X + Y + Flag_Var.

Thank you.