combining predictions and confidence values

Hello.  This is a general process questions and it may simply show that I have a fundamental misunderstanding of what I am doing.  I have two example sets with the same outcome labels and mostly the same variables, the smaller set of the two have some new values that were not recorded in the more distant past.

I have a test set of new students that I would like to predict the outcome for.  Both models now do this nicely (thank you messages at the end of my other post in this forum).  I am wondering, then, if there is an appropriate way of taking the two sets of predictions and combining them.  Obviously the students who are high risk in both sets of results are of interest, but I am wondering if there is an accepted method for reducing the number of false positives by cross checking the predictions between the two sets of results.  Cheers,

--crhis