Meta-Analysis: Assessing Homogeneity between Study Variances in Categorical Models of Effect Sizes Friday, November 13, 2015 | 2:30 p.m. - 4:00 p.m.

Peik Hall Room 28

Hedges discussed the rationale for fitting categorical models to effect sizes in meta-analysis. Under mixed-effect meta-analytic models, when conducting meta-regression, the assumption is that the between-studies variance is constant.  However, one can opt for a likelihood function that computes a between-studies variance within each factor level. Typically, the decision on which specification to adopt has been made on a theoretical basis or by ad-hoc comparisons of within group variation. The presenter will consider the likelihood ratio test of the null hypothesis that residual variances are equal.

CanAM Online Symposium Series in Educational Research Methods 

Meta-Analysis: Assessing Homogeneity between Study Variances in Categorical Models of Effect Sizes

Presented by Ariel M. Aloe, University of Iowa

Abstract
Hedges discussed the rationale for fitting categorical models to effect sizes in meta-analysis. Under mixed-effect meta-analytic models, when conducting meta-regression, the assumption is that the between-studies variance is constant.  However, one can opt for a likelihood function that computes a between-studies variance within each factor level. Typically, the decision on which specification to adopt has been made on a theoretical basis or by ad-hoc comparisons of within group variation.  The presenter will consider the likelihood ratio test of the null hypothesis that residual variances are equal.

If you have questions about this seminar, contact Professor Mark Davison, mld@umn.edu.

To be notified about future seminars, contact sawye100@umn.edu.

The CanAm Online Symposium, formerly known as the Big Ten Online Symposium, is a series of presentations on advanced measurement and research methods in education.  In 2015-16, the Symposium will include four online seminars.