On what basis might one accept/reject a hypothesis when running ridge regression?

For example, if I have five predictor variables as part of a ridge regression, what criterion would I use to accept or reject the hypothesis that each of these predictors are linked to the outcome variable?

I know that bootstrapping can be used to produce standard errors in ridge regression but there are some convincing arguments against using them to calculate p-values or really reporting them at all (e.g. Goeman, Meijer & Chaturvedi, 2012). As such, what else could I use?

I've been running ridge regression in SPSS so the output is fairly restricted to a ridge trace (k and coefficients) and a plot of k and r-squared. Would anything in those suffice?


1 Answer 1


One possibility is discussed in Cule et al. BMC Bioinformatics 2011 http://www.biomedcentral.com/content/pdf/1471-2105-12-372.pdf and the associated R package : http://cran.r-project.org/web/packages/ridge/ridge.pdf


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.