Meta-regression: adjusting covariates during meta-analysis

Ravishankar, N and Nair, Sreekumaran N (2015) Meta-regression: adjusting covariates during meta-analysis. International Journal of Advanced Research, 3 (2). pp. 899-904. ISSN 2320-5407

[img] PDF
meta regression-848_IJAR-5080 (4).pdf - Published Version
Restricted to Registered users only

Download (616kB) | Request a copy

Abstract

Meta-regression is a method to access impact of covariates on the effect estimates of studies to be meta-analyzed. It involves a weighted linear regression with the dependent variable being the effect estimate of studies and study level covariates being the independent variables. The weights are assigned inversely proportional to variance of the effect estimates. Pooled estimate adjusted for the covariate can be obtained by centering the covariate at its mean. In such cases, the estimate of intercept provides the pooled estimate adjusted for the covariate. The present paper elucidates the procedure of meta-regression involving a single covariate and guides the readers to perform meta-regression without the aid of any software packages.

Item Type: Article
Uncontrolled Keywords: Meta-regression; Meta-analysis; Covariates; Fixed effect model; Random effects model; Weighted linear regression; Weighted least squares.
Subjects: Departments at MU > Statistics
Depositing User: KMC Manipal
Date Deposited: 07 Mar 2015 04:40
Last Modified: 07 Mar 2015 04:40
URI: http://eprints.manipal.edu/id/eprint/142121

Actions (login required)

View Item View Item