A Comparative Study on Methods for Measuring Heterogeneity in Binomial Proportion – Meta Analytic Approach

Abstract

Statistical inference on a common quantity of interest arising from multiple related studies is quite pervasive in medical, social, statistical genetics, clinical trials, and epidemiological research. However, a scope is identified to study the issues related to pooling proportion across many Bernoulli type of events. The Main objective of the study is to exploit the asymptotic approximation of proportion and logit transformation in random effects model. Seven procedures for estimating between-variance in frequentist inferential methods have been used. Number of Bernoulli trials, number of studies, and events occurred at boundaries of the parameter are prime data characteristics considered in the study. From the results it is possible to device a methodology to choose more appropriate methods under different problem situations. All the procedures are implemented in R and major functions are presented in Appendix.