10
$\begingroup$

I know that it is possible when you realize a multiple linear regression to calculate the confidence interval on the R-squared and on the adjusted R-squared. Does somebody know how to do it with R?

$\endgroup$

1 Answer 1

6
$\begingroup$

You probably can be served by the CI.Rsq function in the psychometric package. The equation is provided in the help. An alternative would be bootstrapping the CI.

$\endgroup$
2
  • $\begingroup$ I believe this requires n>60 or so - wonder if there is a more general or exact solution? Troels $\endgroup$
    – user37217
    Jan 12, 2014 at 18:38
  • $\begingroup$ I'm not sure recommendations for n > 60 for the CI are really any different from recommendations of n for the regression in the first place. Poor parameter estimates lead to poor CI estimates. That will be true regardless of how they are calculated. A more complex approach would be to bootstrap the model directly but there's not enough information in the question to help with that. $\endgroup$
    – John
    Jan 13, 2014 at 6:18

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.