Thank you very much for the previous answers. I wanted to convert log odds ratios variance to log odds ratios standard error for doing meta-analysis. can you suggest the formula. is it se = Sq.root(1/variance)
that is correct. the variance of the lnOR is (1/a+1/b+1/c+1/d), to get the Standard error you take the Sq.root of that. Afterwards you can calculate Confidence intervals and transform all back using the exp-function.
Whilst this formula is ok if you have an unadjusted effect estimate, for a meta-analysis you should use adjusted estimates and you won't have the info to calculate the SE this way. The SE can be calculated from the confidence interval (or p value) using calculator software in RevMan (free software from cochrane.org) - let me know if you want the formula to calculate by hand. BW, Dan.