I performed Mann-Whitney U test and Fisher´s exact test as a univariate analysis (association) between independent and dependent variables. Now, I would like to create a multivariate model, including all those variables that were significant ($p$-value) on univariate analysis.

Does the increase of $p$-value as inclusion criterion for multivariate model (e.g. set at 0.2 instead of "standard" 0.05) decrease the probability of type II error?



Your Answer

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

Browse other questions tagged or ask your own question.