0
feglm(y ~ x + z, 
      df, family = 'probit')

This does not seem to work in R. This is odd. The fixest package is really great. But for two step regressions such as the Heckman in a panel context, would involve a first stage probit. It doesn't seem to work with or without fixed effects. I would hate to have to move back to the plm package just to perform a probit. Any insight on whether people have to switch back and forth between these two packages would also be helpful.

This website claims that this works (https://stata2r.github.io/fixest/#models), although I cannot see any documentation explaining this phenomenon.

r-learning-machine
  • 105
  • 7
  • 1
    Try `family = binomial(link = "probit")` – Rui Barradas Jul 20 '22 at 21:14
  • This solved the problem. Thank you. Where did you find the documentation? This is the same format as in the plm package. – r-learning-machine Jul 20 '22 at 22:12
  • 1
    See the [function's documentation](https://search.r-project.org/CRAN/refmans/fixest/html/feglm.html) and follow the link [`family`](https://search.r-project.org/R/refmans/stats/html/family.html). – Rui Barradas Jul 21 '22 at 05:12
  • 1
    If something isn't working: check the version. Using "probit"/"logit" in the family works only since 0.10.0: https://lrberge.github.io/fixest/news/index.html#new-features-0-10-0 – Laurent Bergé Jul 22 '22 at 19:23

0 Answers0