The Evolution of Statistical Methods in Speech, Language, and Hearing Sciences.
The Evolution of Statistical Methods in Speech, Language, and Hearing Sciences.
复制标题
DOI：
10.1044/2018_jslhr-h-astm-18-0378
复制
发表时间：
2019-03
期刊：
影响因子：
--
通讯作者：
J. Oleson;Grant D. Brown;R. McCreery
中科院分区：
文献类型：
--
作者：
J. Oleson;Grant D. Brown;R. McCreery
目的
言语、语言和听力科学领域的科学家依靠统计分析来帮助揭示从其研究中收集的数据的复杂关系和模式。然而，来自沟通科学与障碍领域的研究数据很少符合许多传统统计方法的基本假设。幸运的是，统计学领域提供了许多成熟的统计技术，可用于应对当今涉及人类行为数据复杂研究的挑战。在这篇综述文章中，我们重点介绍了几种在言语和听力科学分析中有应用前景的技术和一般方法。
方法
这篇综述文章的目的是概述在言语、语言和听力科学中有应用前景但可能未得到充分利用的统计方法。
结果
我们提出了一些建议，用于确定在分析比例数据和重复测量数据时，替代统计方法何时可能更具优势。我们还介绍了贝叶斯范式和统计学习，并就科学家何时可能考虑这些方法提出了建议。
结论
现代统计技术提供了更大的灵活性，使科学家能够提出更直接和更有信息量的研究问题。
Purpose Scientists in the speech, language, and hearing sciences rely on statistical analyses to help reveal complex relationships and patterns in the data collected from their research studies. However, data from studies in the fields of communication sciences and disorders rarely conform to the underlying assumptions of many traditional statistical methods. Fortunately, the field of statistics provides many mature statistical techniques that can be used to meet today's challenges involving complex studies of behavioral data from humans. In this review article, we highlight several techniques and general approaches with promising application to analyses in the speech and hearing sciences. Method The goal of this review article is to provide an overview of potentially underutilized statistical methods with promising application in the speech, language, and hearing sciences. Results We offer suggestions to identify when alternative statistical approaches might be advantageous when analyzing proportion data and repeated measures data. We also introduce the Bayesian paradigm and statistical learning and offer suggestions for when a scientist might consider those methods. Conclusion Modern statistical techniques provide more flexibility and enable scientists to ask more direct and informative research questions.