0

Is it correct to control for variables that might have the inherited structure of the dependent variable?

Two cases below:

1) Variables that are not considered as predictors, but rather a consequence of the dynamics in the DV. Example: success rate of an online project (DV) and the comments on that project (Control)?

2) Binary transformed dependent variable included as a control?

Nick Cox
  • 48,377
  • 8
  • 110
  • 156
Delyan Peyankov
  • 103
  • 9
  • Relation to http://stats.stackexchange.com/questions/67057/unexpected-correlation-method-to-account-for-the-shared-variance ? – Nick Cox Aug 11 '13 at 15:11
  • 1
    A related term to search for is intermediate outcome for bullet 1. [See here for related discussion](http://stats.stackexchange.com/a/19615/1036). Bullet 2 is a bad idea as well, for related (although more obvious) reasons. – Andy W Aug 11 '13 at 15:42

0 Answers0