gmeta: Meta-Analysis via a Unified Framework of Confidence Distribution
An implementation of an all-in-one function for a wide range of meta-analysis problems. It contains three functions. The gmeta() function unifies all standard meta-analysis methods and also several newly developed ones under a framework of combining confidence distributions (CDs). Specifically, the package can perform classical p-value combination methods (such as methods of Fisher, Stouffer, Tippett, etc.), fit meta-analysis fixed-effect and random-effects models, and synthesizes 2x2 tables. Furthermore, it can perform robust meta-analysis, which provides protection against model-misspecifications, and limits the impact of any unknown outlying studies. In addition, the package implements two exact meta-analysis methods from synthesizing 2x2 tables with rare events (e.g., zero total event). The np.gmeta() function summarizes information obtained from multiple studies and makes inference for study-level parameters with no distributional assumption. Specifically, it can construct confidence intervals for unknown, fixed study-level parameters via confidence distribution. Furthermore, it can perform estimation via asymptotic confidence distribution whether tie or near tie condition exist or not. The plot.gmeta() function to visualize individual and combined CDs through extended forest plots is also available. Compared to version 2.2-6, version 2.3-0 contains a new function np.gmeta().
||stats, BiasedUrn, binom
||Guang Yang, Jerry Q. Cheng,
Minge Xie and Wei Qian
||Jerry Q. Cheng <jcheng18 at nyit.edu>
||GPL-2 | GPL-3 [expanded from: GPL (≥ 2)]
Please use the canonical form
to link to this page.