Wednesday, June 23, 2010

What's the 'Average Normalized Gain' ? #2

Some blog followers might be interested in a post of the above title. The abstract reads:


ABSTRACT: In response to my post "What's the 'Average Normalized Gain' ?" EvalTalk's Robert Walker made three points to which I respond in turn:

a. "Why "normalized"? There seems to be no relationship to the normal curve." See my discussion of the physics/psychometric language gap in my *complete* post online at .

b. "Why not use Cohen's effect size?" Primarily because, for pre/-post data for a class, it depends on the actual bare (unrenormalized) average <%Gain> = <%post> - <%pre>, whose magnitude, tends to be negatively correlated with the average pretest score <%pre>. On the other hand, in "Lessons from the Physics Education Reform Effort" [Hake (2002) I calculated a relatively large effect size of 2.43 for the average of the average normalized gains for 48 Interactive Engagement courses vs the average of the average normalized gains for 14 Traditional courses.

c. "I have found a meta analysis by Anja Zwingenberger of 36 original studies very useful. They all used effect size." But I doubt that Zwingenberger considered pre/post studies (please correct me if I'm wrong) because, despite their value, they are very rare, in part due to the pre/post paranoia that plagues some PEP's (Psychologists-Education specialists-Psychometricians).


To access the complete 17 kB post please click on .

Richard Hake, Emeritus Professor of Physics, Indiana University

Honorary Member, Curmudgeon Lodge of Deventer, The Netherlands

President, PEdants for Definitive Academic References which Recognize the Invention of the Internet (PEDARRII)>

REFERENCES [[URL's shortened by .]

Hake, R.R. 2010. "What's the 'Average Normalized Gain' ? #2" online on the OPEN AERA-L archives at Post of 23 Jun 2010 14:14:45 -0700 to AERA-L & Net-Gold. The abstract and link to the complete post are being transmitted to various discussion lists.

No comments: