I did learn about collinearity issue where it is possible that 2 significant independent variables when used together in the model can make either or both insignificant. However does the vice-versa case exist? and under which circumstances?
Asked
Active
Viewed 68 times
2
-
+1 first, i think the answer is yes, but i am struggling to give you a numerical example. – Haitao Du Aug 22 '16 at 04:56
-
See this thread, especially about "suppressor variables". https://stats.stackexchange.com/questions/73869/suppression-effect-in-regression-definition-and-visual-explanation-depiction – Peter Flom Jan 01 '24 at 12:46
-
Although it's unclear what the "vice-versa" case might be, we have many posts explaining how significance can change as variables are introduced in regression -- and they cover all the possibilities. – whuber Jan 01 '24 at 14:29
1 Answers
-2
In general and dismissing collinearity , linear models become more representative as variables are introduced even though R^2 is relatively low in business application. Literature exists explaining this rational. During data exploratory practice one may find two less significant variables resulting in higher significance when clustered (unsupervised data mining ).
moka
- 7
-
1This answer is here is lacking and unclear. It may behoove the OP and other readers here to provide a more thorough explanation. Otherwise this answer may be deleted for being low quality. – Shawn Hemelstrand Jan 01 '24 at 14:28