0

I have 1 dependent variable (a type of behaviour) and 5 independent variables. The independent variables all have statistically significant positive correlations with the DV (p<.01). However, I conducted a regression analysis and only one of the independent variables is now statistically significant.

After reading some other posts, I am assuming that this is due to multicollinearity between my IVs. My question is: how do I report this in my study? Is it ok to say there is a relationship due solely on the siginificant correlation? and include an argument on multicollinearity as a limitation to my regression analysis?

dissertationhelp
  • 511
  • 4
  • 6
  • 16
  • This sounds very much like [your question from yesterday](http://stats.stackexchange.com/questions/86437/testing-the-relationship-between-variables-very-basic-question), so it would be a good idea to provide the link for the context. – amoeba Feb 14 '14 at 21:56
  • Regarding your question: are your independent variables correlated between each other? – amoeba Feb 14 '14 at 21:58
  • some are and some are not, and if they are, it's only weak correlations – dissertationhelp Feb 14 '14 at 22:02
  • You can have substantial multicollinearity even if all pairwise correlations are small. This topic has come up regularly. Please search our site & read some of the related threads. If you still have a question afterwards, edit your Q w/ what you've learned & what you still need to know. – gung - Reinstate Monica Feb 14 '14 at 22:08
  • @gung: actually there is even a closer thread: http://stats.stackexchange.com/questions/68162/two-independent-variables-both-correlate-with-the-dependent-variable-but-none-a. Definitely a duplicate. – amoeba Feb 14 '14 at 22:09

0 Answers0