I am trying to write a Economics meta-analysis. The papers that I have collected so far contain a 50/50 split between OLS models and Logistic regressions (coefficients reported as odds ratios). If we assume for simplicity that all the outcome variables are measured in the same way, I have used the method described here to convert the odds ratios to SMDs.

My plan is to estimate the model using both the OLS and Logistic coefficients. I will include dummy variables for the different model types, as a form of subgroup analysis. Is this okay to do? Any help is much appreciated!

  • $\begingroup$ Seems reasonable. Do you have some specific doubt? $\endgroup$ – mdewey Sep 7 '20 at 14:57
  • $\begingroup$ I'm getting some strangely low transformed coefficients and high standard errors. Also i've read a few papers online that highlight problems with using models that contain different numbers of covariates. Do you know how this is usually dealt with? $\endgroup$ – os301 Sep 8 '20 at 10:58
  • $\begingroup$ If the other covariates differ then the coefficients will not be strictly comparable between studies even using just OLS or just logistic. $\endgroup$ – mdewey Sep 8 '20 at 13:35

Your Answer

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

Browse other questions tagged or ask your own question.