Page 1 of 1
Contribution of each beta coefficient to the model's R2?
Posted: Wed Sep 21, 2011 11:27 am
by PCA
Does anyone know of a way to calculate this in EViews?
I am trying to find the individual contribution of each beta coefficient to the overall R2 of the model. In other words, if a model has 4 statistically significant explanatory variables, how much does each one of those contribute to the model as a whole?
Re: Contribution of each beta coefficient to the model's R2?
Posted: Wed Sep 21, 2011 12:10 pm
by startz
This is a good question, but unfortunately it doesn't have a unique answer unless the right hand side variables are uncorrelated.
Re: Contribution of each beta coefficient to the model's R2?
Posted: Wed Sep 21, 2011 12:20 pm
by PCA
Assuming they aren't correlated, is the only way to figure out their individual contribution to the R2 figure by stepping each variable into the regression one at a time?
Take a peek at the attachment picture - does that seem logical?
Re: Contribution of each beta coefficient to the model's R2?
Posted: Wed Sep 21, 2011 1:05 pm
by startz
If they're uncorrelated--which is unusual--you can just run separate regressions on each explanatory variable and use the R^2 from each.
Re: Contribution of each beta coefficient to the model's R2?
Posted: Thu Sep 22, 2011 12:32 pm
by PCA
Does EViews calculate the partial sum of squares or sequential sum of squares for variables?