I have an appended dataset where one sample was assessed on one version of a math test and another sample for a different group of students who were assessed on a different version of a math test. I would like to know how correlated the two different math tests are for these two different samples. Is there a way that I can do this in Stata? It is difficult for me to conceptualize how this can be done when students are considered missing on the other math test variable that they didn't take.