The Effect of Publication Bias on the Q Test and Assessment of Heterogeneity
The Effect of Publication Bias on the Q Test and Assessment of Heterogeneity
复制标题
DOI：
10.1037/met0000197
复制
发表时间：
2019-02-01
影响因子：
7
通讯作者：
van Assen, Marcel A. L. M.
中科院分区：
文献类型：
--
作者：
Augusteijn, Hilde E. M.;van Aert, Robbie C. M.;van Assen, Marcel A. L. M.
荟萃分析的主要目的之一是检验和估计效应量的异质性。我们研究了发表偏倚对Q检验的影响，以及作为真实异质性、发表偏倚、真实效应量、研究数量和样本量变化的函数的异质性评估。本研究有两个主要贡献，是相关的所有研究人员进行荟萃分析。首先，我们展示了发表偏倚何时以及如何影响异质性评估。通过分析推导出异质性测量H-2和I-2的预期值，并在Monte Carlo模拟研究中检查了Q检验的功效和I型错误率。我们的研究结果表明，发表偏倚对Q检验和异质性评估的影响是巨大的、复杂的和非线性的。发表偏倚可以显著降低和增加真实效应量的异质性，特别是当研究数量很大而群体效应量很小时。因此，我们得出结论，当存在发表偏倚时，同质性和异质性指标H-2和I-2的Q检验通常无效。我们的第二个贡献是，我们引入了一个网络应用程序，Q-sense，它可以用来确定发表偏倚对某一荟萃分析异质性评估的影响，并评估荟萃分析估计发表偏倚的稳健性。此外，我们将Q-sense应用于2项已发表的荟萃分析，显示发表偏倚如何导致效应量和异质性的无效估计。
One of the main goals of meta-analysis is to test for and estimate the heterogeneity of effect sizes. We examined the effect of publication bias on the Q test and assessments of heterogeneity as a function of true heterogeneity, publication bias, true effect size, number of studies, and variation of sample sizes. The present study has two main contributions and is relevant to all researchers conducting meta-analysis. First, we show when and how publication bias affects the assessment of heterogeneity. The expected values of heterogeneity measures H-2 and I-2 were analytically derived, and the power and Type I error rate of the Q test were examined in a Monte Carlo simulation study. Our results show that the effect of publication bias on the Q test and assessment of heterogeneity is large, complex, and nonlinear. Publication bias can both dramatically decrease and increase heterogeneity in true effect size, particularly if the number of studies is large and population effect size is small. We therefore conclude that the Q test of homogeneity and heterogeneity measures H-2 and I-2 are generally not valid when publication bias is present. Our second contribution is that we introduce a web application, Q-sense, which can be used to determine the impact of publication bias on the assessment of heterogeneity within a certain meta-analysis and to assess the robustness of the meta-analytic estimate to publication bias. Furthermore, we apply Q-sense to 2 published meta-analyses, showing how publication bias can result in invalid estimates of effect size and heterogeneity.