2

I am currently fitting a penalized logistic regression model using the package logistf (due to quasi-complete separation). I chose this package over brglm because I found much more recommendations for logistf. However, the brglm seems to integrate better with other functions such as predict() or margins::margins(). In the documentation of brglm it says:

"Implementations of the bias-reduction method for logistic regressions can also be found in thelogistf package. In addition to the obvious advantage ofbrglmin the range of link functions that can be used ("logit","probit","cloglog"and"cauchit"), brglm is also more efficient computationally."

Has anyone experience with those two packages and can tell me whether I am overlooking a weakness in brglm, or can I just use it instead of logistf?

I'd be grateful for any insights!

elehna
  • 106
  • 5
  • 2
    why don't you run both of them on some simple problems and see if they give similar answers? Unfortunately this isn't a very well-focused programming problem ... – Ben Bolker Jul 15 '20 at 22:31
  • 2
    I ran my model with both functions and received the same results regarding the p-values and estimates. Maybe I was just overly sceptical because of the widespread recommendations of logistf rather than brglm and because of Scortchi's comment (https://stats.stackexchange.com/a/68917) in 1a). But after re-reading it, it sounds more like a recommendation than a note of caution. – elehna Jul 15 '20 at 22:42
  • 1
    I don't see any cautions there. I'd actually recommend https://CRAN.R-project.org/package=brglm2 , which is newer (same authors I believe) and has some additional functionality such as diagnostic tools ... – Ben Bolker Jul 15 '20 at 23:27

0 Answers0