Hi all,
On the old listserver there was dicussion around use of penalized likelihood estimates with complex survey data: http://www.stata.com/statalist/archi.../msg01126.html
It doesn't look like there was any definitive answer at the time, but these authors suggest Firth (eg, -firthlogit- on SSC) can be useful for rare event logistic analyses with complex survey data, but it's an untested aside(?):
"Because of its computational simplicity, the standard bias-corrected logistic regression approach of Firth, where qhij is estimated under the naive assumption of independence, appears to greatly reduce the bias. The latter approach may be somewhat easier to implement using standard statistical software for logistic regression with complex survey data[3]."
https://www.ncbi.nlm.nih.gov/pubmed/26265769
and a preprint copy: http://www.people.fas.harvard.edu/~k...irth/firth.pdf
On the old listserver there was dicussion around use of penalized likelihood estimates with complex survey data: http://www.stata.com/statalist/archi.../msg01126.html
It doesn't look like there was any definitive answer at the time, but these authors suggest Firth (eg, -firthlogit- on SSC) can be useful for rare event logistic analyses with complex survey data, but it's an untested aside(?):
"Because of its computational simplicity, the standard bias-corrected logistic regression approach of Firth, where qhij is estimated under the naive assumption of independence, appears to greatly reduce the bias. The latter approach may be somewhat easier to implement using standard statistical software for logistic regression with complex survey data[3]."
https://www.ncbi.nlm.nih.gov/pubmed/26265769
and a preprint copy: http://www.people.fas.harvard.edu/~k...irth/firth.pdf