I have 2 categorically dependent variables(both binomial) in logistic regression which individually both give positive estimates against the response(binomial). However if modeled together one give positive and other give negative. I checked for the multicollinearity using VIF, which seems fine. What could be the possible reason for such behaviour.
Asked
Active
Viewed 27 times
0
-
1You should provide more details or give minimal reproducible example. The most probable reason is the pairwise correlation, another reason may be the improper hadling of factor variables – O_Devinyak Jun 07 '14 at 08:24
-
What is your sample size for this model? How many cases (1's)? How many non-cases (0's)? My guess is that you are seeing this as a result of inflated standard errors (resulting from a small sample size in at least one of the outcome categories). – Matt Reichenbach Jun 07 '14 at 16:15