Re: how to calculate an effect size

["Regina Johnson" <vencass@xxxxxxxxxxxxxx>]

you did not provide the website in your e-mail.

----- Original Message -----

From: Jill Levenson, Ph.D.

To: 'Child Maltreatment Researchers'

Sent: Thursday, June 04, 2009 7:34 PM

Subject: RE: how to calculate an effect size

Here is a very good website discussing in simple terms the basic concepts of power analysis. Cohen did provide guidelines for interpreting effect sizes for comparisons and correlations. I use GPower and find it to be very helpful. You can download it for free.

 

Jill Levenson, Ph.D., LCSW

Fort Lauderdale, FL

 

From: bounce-3988070-6833436@xxxxxxxxxxxxxxxx [mailto:bounce-3988070-6833436@xxxxxxxxxxxxxxxx] On Behalf Of Todd McDonald
Sent: Thursday, June 04, 2009 1:53 PM
To: Child Maltreatment Researchers
Subject: RE: how to calculate an effect size

 

Dr. Koh

 

Thank you for the web page.  This seems to be limited to when you compare two means.  Is that correct?  What about when you have a logistic regression model?  Should I just use the odds ratio as the effect size? 

 

TM

 

--- On Wed, 6/3/09, Eun Koh <eunkoh@xxxxxxxxxxxx> wrote:

From: Eun Koh <eunkoh@xxxxxxxxxxxx>
Subject: RE: how to calculate an effect size
To: "'Child Maltreatment Researchers'" <child-maltreatment-research-l@xxxxxxxxxxxxxxxx>
Date: Wednesday, June 3, 2009, 5:16 PM

I recently looked into the issue of effect size, and hopefully, this site will help:

 

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

 

It is noted in the above site, but again many people refer to Cohen's article when discussing the size (small, medium, large) of effect size: Cohen (1988). Statistical power analysis for the behavior sciences (second ed.).

 

Hope this helps.

 

Eun 

 

Eun Koh

Postdoctoral Research Associate

Children and Family Research Center

School of Social Work

University of Illinois at Urbana-Champaign

Tel) 312-641-2505 (ext.18)

E-mail) eunkoh@xxxxxxxxxxxx

 

 

---

From: bounce-3984573-9402306@xxxxxxxxxxxxxxxx [mailto:bounce-3984573-9402306@xxxxxxxxxxxxxxxx] On Behalf Of Todd McDonald
Sent: Wednesday, June 03, 2009 2:01 PM
To: Child Maltreatment Researchers
Cc: jmpolstra@xxxxxxx
Subject: how to calculate an effect size

I notice some of the journal articles use very large samples.  Is there a way to calculate an effect size from a crosstabs table and a logistic regression table?  Is there a rule on what is considered a big effect? 

 

TM