Is there any reference or guideline regarding the minimum number of studies for conducting a meta-analysis?

  • 2
    $\begingroup$ No, but two studies is enough. $\endgroup$
    – zkurtz
    Sep 13 '13 at 21:13
  • 1
    $\begingroup$ Sure, mathematically two studies are enough to calculate a summary effect size. But is there any source i could reference to justify e.g. that I report individual results when there are less than seven studies but calculate a meta-analysis for seven studies or more? $\endgroup$
    – jokel
    Sep 15 '13 at 20:19
  • $\begingroup$ @jokel The question is quite pertinent and needs further investigation. I came across a doctoral thesis by Ronco L. Sharon. This thesis brings out that k = 20 is sufficient for an effect-size. The detailed reference is available in the article by me- meta-analysis of sample correlations. It can be downloaded from ssrn.com or EBSCO-HOST $\endgroup$ May 11 '14 at 12:05

There's actually no minimal (or maximal) number needed for a pooled analysis (e.g. meta-analysis). I've conducted systematic reviews including one trial to over 250 trials. As long as you can analyze data from two separate sources then it is a meta-analysis. If you have a limited number of studies contributing to the meta-analysis then some analyses are not advisable (e.g. publication bias tests, meta-regressions, subgroup/ sensitivity analyses, etc.) due to low power.

Hope this helps.



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.