The poolr Package for Combining Independent and Dependent p Values
The poolr Package for Combining Independent and Dependent p Values
复制标题
DOI：
10.18637/jss.v101.i01
复制
发表时间：
2022-01-01
影响因子：
5.8
通讯作者：
Viechtbauer, Wolfgang
中科院分区：
文献类型：
--
作者：
Cinar, Ozan;Viechtbauer, Wolfgang
poolr 包提供了多种合并（即组合）p 值方法的实现，包括 Fisher 方法、Stouffer 方法、反卡方方法、二项式检验、Bonferroni 方法和 Tipp ett 方法。更重要的是，可以调整这些方法，以考虑测试之间的依赖性，并假设测试统计数据之间存在多元正态性，从而导出 p 值。所有方法都可以根据有效测试数量的估计或通过使用基于模拟适当排列测试的伪重复的经验导出的零分布进行调整。对于 Fisher、Stouffer 和反卡方方法，也可以直接将检验统计量推广以解释相关性，从而产生 Brown 方法、Strube 方法和广义反卡方方法。在本文中，我们描述了各种方法，讨论了它们在包中的实现，基于几个示例说明了它们的使用，并将 poolr 包与其他几个可用于组合 p 值的包进行了比较。
The poolr package provides an implementation of a variety of methods for pooling (i.e., combining) p values, including Fisher's method, Stouffer's method, the inverse chisquare method, the binomial test, the Bonferroni method, and Tipp ett's method. More importantly, the methods can be adjusted to account for dependence among the tests from which the p values have been derived assuming multivariate normality among the test statistics. All methods can be adjusted based on an estimate of the effective number of tests or by using an empirically-derived null distribution based on pseudo replicates that mimics a proper permutation test. For the Fisher, Stouffer, and inverse chi-square methods, the test statistics can also be directly generalized to account for dependence, leading to Brown's method, Strube's method, and the generalized inverse chi-square method. In this paper, we describe the various methods, discuss their implementation in the package, illustrate their use based on several examples, and compare the poolr package with several other packages that can be used to combine p values.