[R-meta] [External] R-sig-meta-analysis Digest, Vol 33, Issue 16

Viechtbauer, Wolfgang (SP) wo||g@ng@v|echtb@uer @end|ng |rom m@@@tr|chtun|ver@|ty@n|
Fri Feb 21 13:36:50 CET 2020

Interesting -- will take a look.

There is also:

Aloe, A. M. (2015). Inaccuracy of regression results in replacing bivariate correlations. Research Synthesis Methods, 6(1), 21-27.


>-----Original Message-----
>From: R-sig-meta-analysis [mailto:r-sig-meta-analysis-bounces using r-project.org]
>On Behalf Of Cort Rudolph
>Sent: Friday, 21 February, 2020 13:28
>To: r-sig-meta-analysis using r-project.org
>Subject: Re: [R-meta] [External] R-sig-meta-analysis Digest, Vol 33, Issue
>I did a small simulation study on this once. Including even small numbers of
>regression coefficients in a MA model will really bias the estimate.
>Cort W. Rudolph, Ph.D.
>Associate Professor
>Industrial and Organizational Psychology
>Saint Louis University
>Morrissey Hall 2827
>St. Louis, MO, 63103
>rudolphc using slu.edu
>Office:  +1(314) 977-7299
>Mobile: +1(313) 720-7082
>www.cortrudolph.com <http://www.cortrudolph.com/>
>    Message: 2
>    Date: Fri, 21 Feb 2020 08:36:15 +0000
>    From: "Viechtbauer, Wolfgang (SP)"
>    	<wolfgang.viechtbauer using maastrichtuniversity.nl>
>    To: Yingkai <yykjiyisuipian using gmail.com>,
>    	"r-sig-meta-analysis using r-project.org"
>    	<r-sig-meta-analysis using r-project.org>
>    Subject: Re: [R-meta]  standardized betas transformed to correlation
>    	coefficients
>    Message-ID: <4a2254ab2c754542b161b14598ca06cb using UM-MAIL3213.unimaas.nl>
>    Content-Type: text/plain; charset="utf-8"
>    Dear Yingkai,
>    I would strongly caution against using this formula. It was empirically
>derived and has no proper statistical grounding. In general, one cannot
>transform a standardized regression coefficient into a correlation
>coefficient (without any further information about how the x and y variables
>of interest correlate with all of the other predictors in the model). The
>one exception is the case where there are are no other predictors in the
>model, in which case the standardized regression coefficient is exactly
>equal to the correlation coefficient (and which illustrates that at least in
>this scenario, the transformation equation would be incorrect as it would
>incorrectly inflate the 'estimated correlation' when the association is
>    Best,
>    Wolfgang
>    -----Original Message-----
>    From: R-sig-meta-analysis [mailto:r-sig-meta-analysis-bounces using r-
>project.org] On Behalf Of Yingkai
>    Sent: Friday, 21 February, 2020 7:42
>    To: r-sig-meta-analysis using r-project.org
>    Subject: [R-meta] standardized betas transformed to correlation
>    Dear all,
>    I am confused about how to transform standardized betas  to correlation
>    coefficients.
>    Can I used the simple imputation formula proposed by Peterson and Brown
>    (2005) where r = β + 0.05 λ ?  λ is an indicator variable that equals 1
>    when β is nonnegative and 0 when β is negative. Is this correct? or
>    acceptable ?
>    best,
>    --
>    Yingkai, Ph.D Candidate
>    Southwest University, ChongQing, CHINA.

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