Cohen's d

[Salbod]

Good afternoon friend,

                I’ve run a series of t-tests with unequal groups (ns = 13 and 25). Moreover, in a few analyses the variances were unequal too.  I want  to report Cohen’s d. Does anyone have the reference for formula for Cohen’s d which handles the present situation (unequal n and unequal variance).

                Any suggestions or references will be greatly appreciated.

 

Stephen Salbod, Pace University, NYC

 

[Meyer, Gregory J]

Stephen,

 

You could use equation 2 from McGrath and Meyer (2006, “When Effect Sizes Disagree: The Case of r and d”, Psych Methods ), which reviews the various formulas in the literature for computing the standardized mean difference. This formula follows Cohen’s original conceptualization of d and computes the pooled descriptive standard deviation for the denominator.

 

Below is some syntax that could be used, both for raw data and for converting t-test output. Greg

 

* The Cohen's d formula requires that the SDs be computed as descriptive statistics.

* That is, the formula assumes that the SDs are sigma (computed with N in the

* denominator), not S, the inferential estimate of sigma (computed with N-1 in the

* denominator). Assuming the values entered in Grp1SD and Grp2SD are S not sigma,

* d is computed with the following formula (where Sgm = sigma).

COMPUTE Pool_Sgm = sqrt((((Grp1n - 1) * Grp1SD**2) + ((Grp2n - 1) * Grp2SD**2)) / (Grp1n + Grp2n)) .

COMPUTE Cohens_d = (Grp1Mean - Grp2Mean) / pool_sgm .

EXECUTE .

 

* To compute Cohen's d from an independent samples t-test, use the following.

COMPUTE d_by_t = (t_Value * n) / (sqrt(df) * sqrt(Grp1n * Grp2n)) .

EXECUTE .

 

 

 

From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Salbod, Mr. Stephen
Sent: Monday, November 23, 2009 4:52 PM
To: [hidden email]
Subject: Cohen's d

 

Good afternoon friend,

                I’ve run a series of t-tests with unequal groups (ns = 13 and 25). Moreover, in a few analyses the variances were unequal too.  I want  to report Cohen’s d. Does anyone have the reference for formula for Cohen’s d which handles the present situation (unequal n and unequal variance).

                Any suggestions or references will be greatly appreciated.

 

Stephen Salbod, Pace University, NYC

 

[Marta Garcia-Granero]

Salbod, Mr. Stephen wrote:

>
> Good afternoon friend,
>
> I’ve run a series of t-tests with unequal groups (ns = 13 and 25).
> Moreover, in a few analyses the variances were unequal too. I want to
> report Cohen’s d. Does anyone have the reference for formula for
> Cohen’s d which handles the present situation (unequal n and unequal
> variance).
>
> Any suggestions or references will be greatly appreciated.
>
>

Hi Stephen:

I might be wrong, but I always thought that if unequal variances are
present, then Cohen's d should not be used (since it pools both
variances into one single estimate), but Glass' delta could be used
instead (it is computed using only the control group standard deviation).

HTH,
Marta GG

--
For miscellaneous SPSS related statistical stuff, visit:
http://gjyp.nl/marta/

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD