2
$\begingroup$

I am conducting a meta-analysis where the effect size is the individual group raw mean (i.e., not raw mean difference). A couple of questions:

  1. How (in R: metafor or otherwise) is the estimated sampling variance calculated for these raw means, given sd and n as inputs?

  2. If for some of my data points the sd is not reported, can I use some other statistic (I'm not sure what, given there are no group comparisons) to estimate sd so that the estimated sampling variance can be computed?

Improve this question
$\endgroup$

1 Answer 1

Reset to default
5
$\begingroup$

Well, if you are using metafor and R you can examine the code to see how metafor does it. The sampling variance of a raw mean is $\sigma^2/n$ where $\sigma$ is the standard deviation and $n$ the sample size.

If you are missing standard deviations you could impute them from the other studies if you have the same measure being used in other studies. If you have several to choose from then either average them, take the largest (conservative), or repeat the analysis with all options to see if it affects your substantive conclusion.

Improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.