Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Add Firth logistic regression to BAS #63

Open
merliseclyde opened this issue Nov 9, 2022 · 0 comments
Open

Add Firth logistic regression to BAS #63

merliseclyde opened this issue Nov 9, 2022 · 0 comments

Comments

@merliseclyde
Copy link
Owner

merliseclyde commented Nov 9, 2022

Separation is a current problem with bas.glm in logistic regressions as which uses nested integrated Laplace approximations which depend on MLEs. With quasi or perfect separation, MLEs are not identifiable, which Firth Logistic regression addresses through data augmentation or equivalently a penalty term using Jeffreys invariant prior.

For DA, an augmented data set with $3n$ observations is constructed with stacking the design matrix 3 times and a response vector that is (the original Y, a nx1 vector of 1's, and a nx1 vector of 0's). The augmented response are accompanied by weights: $w_{i} = h_i /2$ where
$$H = W^{1/2}X(X^TWX)^{-1}X^TW^{1/2}$$
and $W$ is a diagonal matrix with $[W]_{ii} = \pi_i ( 1 - \pi_i)$.

Questions in implementation:

  • need to fix issue related to incorrect standard errors in weighted regressions and GLMS - issue Standard Errors for fitted values and predictions are incorrect for linear models with weights and glms #52
  • should weights be based on full model or different for each model. (seems like model dependent ideal)
  • can data augmentation be based on adding $n$ observations through sufficiency? i.e add proportion 0.5 and weights $w_i = h_i$?
  • should require iterative updating to get MLE's but should be feasible with current iteratively weighted least squares implentation for logistic regression

of interest for JASP @vandenman ?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

No branches or pull requests

1 participant