[R-meta] effect size similarities

Viechtbauer, Wolfgang (SP) wo||g@ng@v|echtb@uer @end|ng |rom m@@@tr|chtun|ver@|ty@n|
Wed Jun 23 09:57:48 CEST 2021

This is for the case of two dependent samples, that is, we want to compute

d = (mean1 - mean2) / SD_change

where mean1 and mean2 are the means at the two measurement occasions and SD_change is the standard deviation of the change scores. Then indeed

d = t / sqrt(n)

where t is the t-statistic from the two-dependent samples t-test.

But this whole discussion started out with the case of a z-test of a binomial proportion. This is neither the case of two independent samples, nor the case of two dependent samples, and so none of this is relevant.


>-----Original Message-----
>From: Patrizio E Tressoldi [mailto:patrizio.tressoldi using unipd.it]
>Sent: Wednesday, 23 June, 2021 9:42
>To: Viechtbauer, Wolfgang (SP); r-sig-meta-analysis using r-project.org
>Subject: Re: [R-meta] effect size similarities
>Il 23/06/2021 09:28, Viechtbauer, Wolfgang (SP) ha scritto:
>> I understand. What I tried to say is that this equation does not apply for a
>one-sample t-test. Cohen's d is
>> d = (mean1 - mean2) / SD,
>> where mean1 and mean2 are the means of two groups and SD the (pooled) SD and
>hence it reflects the difference between two groups. If there is only one group,
>then Cohen's d cannot be computed.
>In Lakens (2013) it is stated: "The effect size estimate Cohen’s dz can
>also be calculated
>directly from the t-value and the number of participants using the
>formula provided by Rosenthal (1991): Cohen′s dz = t√n
>Lakens, D. (2013). Calculating and reporting effect sizes to facilitate
>cumulative science: a practical primer for t-tests and ANOVAs. Frontiers
>in psychology, 4, 863.

More information about the R-sig-meta-analysis mailing list