Date:         Thu, 29 Sep 2005 09:31:52 +0200
Reply-To:     Marta García-Granero
              <biostatistics@terra.es>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:         Marta García-Granero
              <biostatistics@terra.es>
Organization: Asesoría Bioestadística
Subject:      Re: Hedges Unbiased Effect Size Estimator
In-Reply-To:  <54CF82BF475EF74583773795A396AD0C06BF4214@nt207mxsep.corporate.hdwa.health.wa.gov.au>
Content-Type: text/plain; charset=ISO-8859-15

Hi Jonathon,

To my knowledge, this formula is for 2 independent samples. Right now
I don't have Egger's book on meta-analysis at hand, but I'll check if
that particular topic is covered.

I seem to remember that I read (somewhere...) that the formula could
be used for Cohen's d obtained from paired data considering n1=n2=n,
but I'm not really really sure, because somewhere else I read that the
Hedges estimator is only for independent samples.

See:
http://web.uccs.edu/lbecker/Psy590/es.htm

I haven't had time to read the following paper slowly, but I think you
can find something interesting in it:
http://www.ucm.es/info/Psi/docs/journal/v2_n1_1999/art32.pdf

(Although it's a Spanish journal, only the summary is in Spanish, the
rest of the paper is in English).

LJ> My question relates to the calculation of the confidence intervals for
LJ> Hedges Unbiased Effect Size Estimator.  Can someone clarify if this
LJ> formula pasted below is appropriate for a single group pre-test
LJ> post-test design (that is, repeated measures).  I ask because Hedges and
LJ> Olkin (1985, somewhere between 80-90) use the example of an experimental
LJ> and control group (that is, independent groups, not repeated measures).
LJ> In addition is anyone is able to offer any references which deal with
LJ> the issue, then that would also be appreciated.
LJ> I have pasted a portion of syntax provided by Marta Garcia-Granero
LJ> below:
LJ> COMPUTE poolvar=((n1-1)&*(sd1&**2)+(n2-1)&*(sd2&**2))/(n-2) .
LJ> COMPUTE cohen=(mean1-mean2)/SQRT(poolvar).
LJ> COMPUTE hedges=cohen&*(1-(3/(4&*(n)-9))).
LJ> COMPUTE se_hedge=SQRT((n/(n1&*n2))+((hedges&**2)/(2&*(n-3.94)))).

Regards,
 Marta García-Granero          mailto:biostatistics@terra.es