R-square in Multiple Regression

Home > Statistics Every Writer Should Know > The Stats Board > Discusssion

R-square in Multiple Regression
Message posted by mario (via 204.74.5.40) on February 5, 2002 at 7:28 PM (ET)

I want to calculate the percentage of variance accounted for by each predictor in a multiple regression. I have 8 predictors. Some predictors are significant and some are not, but I still want to do this.

I tried to do this 2 different ways and the results do not "match."

One way is to predict the dependent variable from each predictor (one at a time) and see what the R-square is.

The other way is to enter all predictors at the same time and take one of the predictors each time. See how the r-square changes when each predictor is taken out and all the others are left in.

Can anyone shed light on which is the correct way of doing this and why they do not match?

READERS RESPOND:
(In chronological order. Most recent at the bottom.)

Re: R-square in Multiple Regression
Message posted by JG (via 64.12.106.38) on February 7, 2002 at 8:21 AM (ET)

You have to transform your predictors into an independent or orthogonal set. Presently, the order in which you add or subtract predictors makes a difference, as you discovered. Also, there is adjusted R-squared, etc.

Your $5 contribution helps cover part the $500 annual cost of keeping this site online.