Crosse Michael J, Foxe John J, Molholm Sophie
Segotia, Galway, Ireland.
Trinity College Dublin, Ireland.
ArXiv. 2024 Jan 17:arXiv:2401.09401v1.
Statistical hypothesis testing and effect size measurement are routine parts of quantitative research. Advancements in computer processing power have greatly improved the capability of statistical inference through the availability of resampling methods. However, many of the statistical practices used today are based on traditional, parametric methods that rely on assumptions about the underlying population. These assumptions may not always be valid, leading to inaccurate results and misleading interpretations. Permutation testing, on the other hand, generates the sampling distribution empirically by permuting the observed data, providing distribution-free hypothesis testing. Furthermore, this approach lends itself to a powerful method for multiple comparison correction - known as max correction - which is less prone to type II errors than conventional correction methods. Parametric methods have also traditionally been utilized for estimating the confidence interval of various test statistics and effect size measures. However, these too can be estimated empirically using permutation or bootstrapping techniques. Whilst resampling methods are generally considered preferable, many popular programming languages and statistical software packages lack efficient implementations. Here, we introduce PERMUTOOLS, a MATLAB package for multivariate permutation testing and effect size measurement.
统计假设检验和效应量测量是定量研究的常规组成部分。计算机处理能力的进步通过重采样方法的可用性极大地提高了统计推断的能力。然而，当今使用的许多统计方法都基于传统的参数方法，这些方法依赖于对总体的假设。这些假设可能并不总是成立，从而导致结果不准确和解释具有误导性。另一方面，置换检验通过对观察到的数据进行置换来凭经验生成抽样分布，提供无分布假设检验。此外，这种方法适用于一种强大的多重比较校正方法——称为最大校正——它比传统校正方法更不易出现II类错误。传统上，参数方法也用于估计各种检验统计量和效应量测量的置信区间。然而，这些也可以使用置换或自举技术凭经验进行估计。虽然重采样方法通常被认为更可取，但许多流行的编程语言和统计软件包缺乏高效的实现。在这里，我们介绍PERMUTOOLS，一个用于多变量置换检验和效应量测量的MATLAB包。
