Collaborative research: OPUS: Permutational Biometry: Synthesizing the Analytics of Data Analysis in Ecology and Evolution

在这个项目中，研究人员将综合他们在统计理论方面的工作。长期以来，生物数据的分析一直是使用标准（参数）统计方法完成的。然而，这些工具缺乏许多复杂问题的解决方案。作为一种选择，使用计算机对数据进行重新采样提供了一种通用程序，在传统方法不足的情况下可以取代传统方法。这些新方法有可能为生物学家提供一种比传统测试限制更少的工具。将制作一本完整的书，并免费提供，产生广泛的影响。还将提供基于网络的教程，演示本书中的所有主题。这两款产品都将利用研究人员开发的现有软件，并将其免费分发给科学界。通过综合25年来他们所开发的基于排列的分析，研究人员将得出一个在生物科学中执行数据分析的全新理念。这种新的范式源于他们在四个相关领域的方法的结合：1)置换过程中残差的重采样（RRPP）， 2)基于置换的效应大小，3)基于置换的两两比较，以及4)基于独立或相关数据集的基于置换的评估。结果将是强调假设检验和生物信号强度确定的生物计量学观点，其中经验抽样分布的生成使两个组成部分同时实现。该方法是单变量和多变量的，可用于评估独立或相关观测的趋势。完成后，综合将提供一个全面的学习平台，它结合了排列分析的彻底解释，并将这种新兴的统计观点与利用现有软件实现的实践教程集成在一起。这一进展有可能改变生物学家进行研究的方式，因为RRPP合成范式提供了能够解决参数化程序无法解决的假设的分析工具。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

In this project, the investigators will synthesize their work in statistical theory. The analysis of biological data has long been accomplished using standard (parametric) statistical methods. However, these tools lack solutions for many complex questions. As an alternative, resampling of data using computers provides a general procedure that can replace traditional methods in cases where they are insufficient. These new methods have the potential to provide a tool for biologists that is less restrictive than traditional tests. A full-length book will be produced and made freely available providing broad impacts. Web-based tutorials demonstrating all topics in the book will also be provided. Both products will leverage existing software the investigators have developed and distributed freely to the scientific community.By synthesizing 25 years of permutation-based analytics they have developed, the investigators will arrive at an entirely new philosophy for performing data analysis in the biological sciences. This new paradigm results from the union of their methods in four interrelated areas: 1) Resampling of Residuals in a Permutation Procedure (RRPP), 2) permutation-based effect sizes, 3) permutation-based pairwise comparisons, and 4) permutation-based evaluation of independent or correlated datasets. The result will be a biometric perspective that emphasizes both hypothesis testing and biological signal strength determination, where the generation of empirical sampling distributions enables both components simultaneously. The approach is both univariate and multivariate, and can be used to evaluate trends in independent, or correlated observations. When complete the synthesis will provide a comprehensive learning platform, that combines thorough explanation of permutational analytics, and integrates this emerging statistical perspective with hands-on tutorials that utilize existing software for implementation. This advance has the potential to transform how biologists conduct their research, as the RRPP synthetic paradigm provides analytical tools that are capable of addressing hypotheses that parametric procedures cannot.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.