Contact Us


Frequently Asked Questions

ETD Help

Policies and Procedures

Copyright and Patents

Access Restrictions

Search ETDs:
Advanced Search
Browse by:
Browse ProQuest
Search ProQuest

Laney Graduate School

Rollins School of Public Health

Candler School of Theology

Emory College

Emory Libraries

New ETD website is now LIVE and located here:

Random Estimating Functions to Accommodate Heterogeneity in Meta-Analysis

Bo, Na (2017)
Master's Thesis (28 pages)
Committee Chair / Thesis Adviser: Hanfelt, John
Committee Members: Zhang, Rebecca
Research Fields: Biostatistics
Partnering Agencies: Does not apply (no collaborating organization)
Keywords: Random Estimating Equation; Meta-Analysis; Between-Table Heterogeneity
Program: Rollins School of Public Health, Biostatistics and Bioinformatics (Biostatistics)
Permanent url:


Introduction: Meta-analysis is defined here as the statistical analysis of a collection of analytic results for the purpose of integrating the findings (DerSimonian & Laird, 1986). A major concern in meta-analysis is heterogeneity among the studies contributing analytic findings. Failure to account for heterogeneity could lead to misleading conclusions in a meta-analysis. The aim is to use statistical approaches to derive a common estimated odds ratio that represents the common truth behind multiple similar studies.

Methods: To accommodate heterogeneity, we propose to add a random perturbation to each component estimating function. The advantages of this proposal over a random-effects model are that, under reparametrization, the random estimating function remains unbiased, remains subject to an additive perturbation, and has a variance that is well-governed and easy to evaluate.

Results: Our new method can capture between- table heterogeneity and produces a valid estimate of the log odds ratio. An advantage of our new method is that it can be applied to further meta-analysis studies under reparametrization, by simply applying the Delta Method.

Discussion: A major advantage of our random estimating equation method over existing random-effects methods is that our new method can be implemented into meta-analyses for any 1-1 transformation of the odds ratio. Unlike a random-effects model, however, our approach does not easily suggest a data generation mechanism, which makes it challenging to conduct a simulation study. The ways of generating random observations under our model of a randomly perturbed Mantel-Haenszel estimating function need to be explored further.

Table of Contents


application/pdf Master's Thesis 28 pages (447 KB) [Access PDF copy of Master's Thesis]
Permission granted by the author to include this thesis or dissertation in this repository. All rights reserved by the author. Please contact the author for information regarding the reproduction and use of this thesis or dissertation.