2
$\begingroup$

I am conducting a meta-analysis in the ecology discipline, such as on the effect of elevated CO2 on soil organic carbon. In my case, treatment and control groups have different sample sizes within the same study. For example, in 5 studies, combined control n is 43 and treatment n is 90. So, what will be the sample size for the overall meta-analysis? In ecology, the sample size of the mean effect size has only been reported if you study the multiple factors (e.g., elevated CO2, N fertilization, etc.) in the same paper. In other words, a forest plot is separately produced without including individual studies.

$\endgroup$
3
  • 4
    $\begingroup$ A sample size for a meta-analysis is typically described as the number of studies. You can't really sum the sample sizes from each study to get a total sample size. $\endgroup$ Commented Sep 27, 2023 at 4:47
  • $\begingroup$ Thank you for your reply. I would like to ask a follow up question. Then, what is difference between number of observation and number of sample size? Both are often reported in meta-analysis in ecology. $\endgroup$ Commented Sep 27, 2023 at 8:34
  • 1
    $\begingroup$ I'm not sure exactly how people use these terms in ecology (or anywhere really). $\endgroup$ Commented Sep 27, 2023 at 9:50

1 Answer 1

2
$\begingroup$

The point of meta-analysis is not to aggregate all the data together in one lump but to summarise in some way study-specific effects. In that context reporting the number of studies is best. If you have all the original data and do an appropriate individual participant data meta-analysis then report the number of participants but otherwise it is misleading for a conventional meta-analysis.

$\endgroup$
1
  • $\begingroup$ Thank you very much. I will follow your suggestion and do the same. $\endgroup$ Commented Sep 28, 2023 at 5:27

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.