Home > Statistics Every Writer Should Know > The Stats Board > Discusssion
Multiple regressions aargh
Message posted by kate (via 195.107.47.211) on June 21, 2001 at 11:46 AM (ET)
I'm embarassed about this question, but here goes...why do correlations which were previously significant "drop out" when put into a multiple regression. When I am writing up my results does it mean that the previous correlations are not worth recording. Can you tell I don't understand multiple regressions...?! Kate
READERS RESPOND:
(In chronological order. Most recent at the bottom.)
Re: Multiple regressions aargh
Message posted by Darius (via 200.23.217.10) on June 21, 2001 at 1:48 PM (ET)
Kate.Remember that when you use multiple regression the adjusted r^2 depend on the number of variables that you put into the analysis (if some of them don愒 contribute to the correlation reduce the signifficance of the correlation).
Re: Multiple regressions aargh
Message posted by JG (via 128.8.23.72) on June 21, 2001 at 3:53 PM (ET)
Are you talking about the Beta coefficients or about adjusted R-squared. While adjusted R-squared can go down if one has many variables in the regression that do not contribute to a better fit - any reasonable regression process only adds variables if they lower the adjusted R-squared. As far as Beta coefficients are concerned, if you do not have orthogonality the significance of any given independent variable depends on the other variables in the regression equation. That is, X1 may be very significant by itself but become insignificant once X2 and X3 are included in the regression equation.
Your $5 contribution helps cover part the $500 annual cost of keeping this site online.