1
$\begingroup$

I'm organizing a meta-analysis on performances with studies showing sometime positive and sometime negative estimates.

Sometime the result of the study is in a positive direction, with higher scores corresponding to better performance (e.g., number of tasks completed). Sometime the result of the study is in a negative direction, with lower scores corresponding to better performance (e.g., time to completion of the task).

How can I put together both types of studies? What kind of mathematical correction can I do?

I was thinking of estimating the effect size (in Hedges' g), then multiplying per -1 the positive-sided studies, so all studies will be in a negative direction. Is this a valid approach?

$\endgroup$
  • $\begingroup$ what other data eg S.D ? $\endgroup$ – Subhash C. Davar May 3 '16 at 11:47
  • $\begingroup$ This seems clear enough to me. $\endgroup$ – gung Jun 1 '16 at 12:48
1
$\begingroup$

Since you are using multiple measures for the outcome then I'm assuming that you will be using standardized mean difference. Therefore, you would multiple the mean effect sizes by -1 (not the measure of variance) for one group (either the ones that are better when more or the vice verse) to make them all in the same direction.

$\endgroup$

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.