0
$\begingroup$

In GEE, several methods have been proposed for correcting for bias when the cluster size is small to moderate (<40). Some have proposed alternative variance estimators, e.g. Morel, Bokossa, and Neerchal (2003), while others have described methods for obtaining bias-corrected estimates of regression coefficients, e.g. Lunardon and Scharfstein (2017).

Would it be advisable to employ both these methods simultaneously?

Lunardon & Scharfstein (2017) https://onlinelibrary.wiley.com/doi/full/10.1002/sim.7366

Morel et al. (2003) https://onlinelibrary.wiley.com/doi/abs/10.1002/bimj.200390021

$\endgroup$
0
$\begingroup$

Answering my own question here. Since the bias-correction methods target different sources of bias, I don't see why not use both. First get the bias-corrected estimates of regression coefficients, then apply the alternative variance estimator.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.