A0201
Title: An intrinsic measure for quantifying the heterogeneity in meta-analysis
Authors: Ke Yang - Beijing University of Technology (China) [presenting]
Enxuan Lin - Innovent Biologics (China)
Wangli Xu - Renmin University of China (China)
Liping Zhu - Renmin University of China (China)
Tiejun Tong - Hong Kong Baptist University (Hong Kong)
Abstract: Quantifying the heterogeneity is an important issue in meta-analysis, and among the existing measures, the I2 statistic is most commonly used. A motivating example is first presented to demonstrate that the I2 statistic heavily depends on the study sample sizes. It is further shown, by a connection between analysis of variance and meta-analysis, that the I2 statistic was defined to quantify the heterogeneity between the observed effect sizes. Inspired by this, a new measure is introduced that aims to directly quantify the heterogeneity between the study populations involved in the meta-analysis in a way that avoids the influence of sample sizes through the observed effect sizes. More importantly, a new estimator, namely the IQ statistic, is also proposed to estimate the newly defined intrinsic measure for quantifying the heterogeneity. For practical use, the exact formulas for the IQ statistic are also specified under three different scenarios, including the mean, the mean difference, and the standardized mean difference. Simulations and real data analysis demonstrate that the IQ statistic provides an asymptotically unbiased estimator of the true heterogeneity between the study populations, and it does not depend on the study sample sizes as expected.
