I am running a logistic regression in Stata to try to determine which independent variables potentially cause a response in a binary dependent variable. When I run certain independent variables against the dependent variable I get very high significances around $p=.005$. However, when I add all of my variables together into one logistic regression equation all of the significances spike to $.1$ or higher, and many go from $.05$ to $.5$ or higher. It seems that the more variables added, the lower the significance goes, even if the variables are all significant by themselves against the dependent binary variable.

  • $\begingroup$ The answer here is better - more comprehensive, and more to the point - than answers in either of the indicated possible duplicates (though I think one Q. isn't even close to a duplicate, and the other not quite a duplicate). $\endgroup$ – Glen_b Mar 18 '15 at 12:21
  • $\begingroup$ @Glen_b: Think it's an FAQ that appears in various guises, e.g. here, here, here, & here. But this concise, germane, answer won't be lost if the q. gets closed as a duplicate, will it? $\endgroup$ – Scortchi - Reinstate Monica Mar 18 '15 at 12:38
  • $\begingroup$ @Scortchi No, answers are not lost upon closure. They can also be moved over to the duplicate thread via a "merge," which I have done here. $\endgroup$ – whuber Mar 18 '15 at 14:16