Meta-analysis: SD for Cohen's d and Hedge's g

My colleague and I are conducting a meta-analysis. We are looking for reference to compute Cohen’s d and Hedge’s g. In Paul Ellis’s excellent Effect Size equations (https://www.polyu.edu.hk/mm/effectsizefaqs/effect_size_equations2.html) we noted that the SD pooled for Hedge’s g and Cohen’s d differed in that term (n-1) in the formula.

We also noted that from the book of Borenstein (2009 version, Wiley, downloadable here:
https://www.meta-analysis.com › downloads
, the SD within seems to be carrying the (n-1) terms (Equation 4.19 on P.26 to 27 in Chapter 4 “Effect Size and Precision”), irrespective whether it is for Cohen’s d or Hedge’s g.

Since the stat package we tried (CMA and R) did not specify how they compute SD for these methods and we would like to include the formula in the report, we would like to clarify with people in the field on whether the SD for d and g should be different. Thank you very much in advance.

Check out:
The Handbook of Research Synthesis and Meta-Analysis: Second Edition. (2009) Cooper, Harris; Hedges, Larry; Valentine, Jeffrey (Editors); pp. 225-228.

IIRC, Hedge’s g has slightly better statistical properties, in that it is an unbiased estimate for small samples.

1 Like