I am preparing a meta-analysis for a certain therapeutic intervention, and will be interested to quantify/summarise the change in various scores, pre-to-post intervention.

Which measure of effect size should I use, namely the effect size that I extract or compute from primary papers, and also that I later compute as a summary measure? I am not sure on which grounds to choose among the main choices:

  • Cohen's d

  • OR (the odds ratio)

  • SMD (standardized mean difference)

Cohen's d is I guess most typically reported in papers so makes life a bit easier. On the other hand, some papers recommend ORs, although not over Cohen's d but over correlation coefficients.

Are some measures of effect sizes more suitable for some meta-analyses but not for others, e.g. those that look at the effect of an intervention (longitudinal) vs those that do a cross-sectional comparison?

  • 1
    $\begingroup$ Use the one which is most clinically pertinent (typically the OR for a hard endpoint), unless it is reported by few studies. You can also pool all endpoints simultaneously with mvmeta in R or Stata, though. $\endgroup$ – Joe_74 Feb 7 '19 at 11:04

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.