I have the following for the data for the meta-analysis:

OR , standard errors, logged OR , variance of logged OR

I have generated a forest plots first using "OR" with "variance of logged OR" using jamovi software, the results were as expected and significant. then I realized that it may be wrong to do that and then I plotted again but using "OR" with "standard error" , what I have noticed that results were about the same and that the confidence intervals were slightly wider when I used the "variance of logged OR" .

so that I wonder if its correct to generate forest plot using "variance of logged OR" with "OR" ?



You generally do all your calculations with the log-odds ratios and the standard errors of the log-odds ratios, because these behave a lot better than the untransformed odds ratios (for very large studies the difference may occasionally be small). You back transform back to the odds ratio scale after doing your calculations.

What a particular software package does when given some values may depend on whether it understands what they are, if so it may actually have done appropriate transformations of odds ratios, but you would have to consult the documentation to find out.


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.