I am conducting a meta-analysis but the raw study coefficients are probit coefficients. I am planning to convert these to log odds to meta-analyse through the formula: 1.61 * probit_coefficient.

Q1) Is this conversion correct?

Q2) Also, is it appropriate to then convert the standard errors of the probit coefficients to log odds standard errors via 1.61*probit_SE?

Q3) And then calculate the variance for the log odds as logodds_SE^2 * sample_size?

  • $\begingroup$ Is there any reason why you do not want to use the probit coefficients and their se directly? $\endgroup$ – mdewey Feb 14 '20 at 13:22
  • $\begingroup$ I wasn't aware that probit coefficients could be used for meta-analysis, as they are not typical standardised effect sizes that can be easily interpreted. E.g. I thought effect sizes for meta-analysis need to be Cohen's d, r, Fisher's Z, or log odds (which I would likely convert to odds ratios for interpretation). $\endgroup$ – JRB Feb 15 '20 at 14:17
  • $\begingroup$ You can use anything for which you have a standard error. The ones you quote are the ones which are most often seen. $\endgroup$ – mdewey Feb 15 '20 at 16:29

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.