0

I know there has been a similar question posted before Why do my p-values differ between logistic regression output, chi-squared test, and the confidence interval for the OR?

but im still not sure what to do-

I have a similar issue with the above question in that for my univariate tests I have used either welches two sample t test, wilcoxon rank sum test, chi-squared test or fishers' exact test depending on the type of data and whether it was normal or not.

In the univariate logistic regression, I get different P-values from the univariate tests above- but I think the test for p-value for the glm function assumes that the data is normal and I have a number of variables that are not normal (although as far as Im aware normality is not an assumption for logistic regression).

Ofcourse, I could just use the p-value for the univariate analysis from the tests i mentioned above but I'm not sure what to do for the p-values of multivariate logistic regression- would these be valid if the data was not normal and can they be fairly compared to the univariate?

Danielle
  • 1
  • 1
    There is no assumption of normality of the predictors for p-values to be accurate, so it is unclear what your concern is. – Noah Apr 18 '21 at 20:40
  • 1
    The normality condition in the Wald test concerns the estimated parameter, not the variables themselves. You should not conclude that because the variables in your regression are non-normal, the p-values from the glm function are wrong. – Jonny Lomond Apr 18 '21 at 20:49

0 Answers0