Taking the cross-wave file you generated above, and taking a single pair of waves (ideally waves 1 and 2), define an outcome variable indicating attrition (w1=1 and w2=0 as a starting point). You may want to recreate the file with more variables: copy the old syntax and edit it to add them.
Fit a logistic regression with the variables you think may have an effect on attrition - take care to declare your categorical variables.
Interpret what you find.