@swasheck determine which two are correlated if u can, use PCA or something so that each is compared to each (check out WEKA for fast functions to do this so you can see at a glance which they are)
then run it with the one but not the other, then vice versa
see if the results are similar
if not, go for something like ridge regression as recommended, if so - the variable didn't affect the results all that much anyways
the other answer recmmends x2 and x3
try doing simple linear regression with those to see if they are the culprit
the other thing is, the data may not have a strong association
that's about the best I can help most likely
I passed stats, and it took me two tries for the first course
