Search code examples
rpanelgoodness-of-fitlogits

Fixed effect logit:adjusted r square-bife package in R

I”m working on my fixed effect logit model using bife package in R.

But the problem is I need the adjusted r square, currently I only can calculate normal r square using the package result.

Does this package have any function to do that?

Solution

  • Logistic regression models do not have the concept of R-squared in the same sense as OLS. Nonetheless, there are a host of pseudo-R-squared metrics, that approximate R2 in the context of nonlinear models. Some of the enclosed metrics also adjust for model complexity in the same spirit as adjusted R2.

    You can access the log-likelihood of the full model by using the logLik(.) function. You will also need to calculate the model with intercept only to get the value for L(M_intercept).