oim vs bootstrap vs jacknife?

[ajay pasi]

Among the vce option oim vs bootstrap vs jacknife, which one should I go for to make my standard errors robust.

with xtlogit depVar indepVar i.Var1 i.Var2, fe vce () ---- for this regression "robust" option is not available in stata.

[Maarten Buis]

The first question you need to ask yourself is "robust to what?". There is only one method that is robust to everything. It is very simple. You just say "I don't know", and you will always be right. That answer will also be useless, but that is the price you have to pay for always being right. If you want a to be a bit more useful, you have to run a risk of being wrong. The art of doing research is to understand what can go wrong, and what dangers are most pressing for your research. Than you can make an informed choice of what dangers you want to counter and what dangers you are willing to accept, taking into account that you have to accept some dangers. So there is no general answer: all three standard errors have their uses. It depends on your situation which one is most appropriate.

[ajay pasi]

Thank you, sir, for such valuable help. Appreciate that

[ajay pasi]

Dear professor, I am not able to find out what OIM does, and what conditions where it is appropriate.

[Maarten Buis]

It uses the observed information matrix (hence the name) to estimate the variance covariance matrix. In practice, if you get different results, your results are unstable and questionable whatever choice you make.

[ajay pasi]

Noted ! , thanks Prof.