论文部分内容阅读
概述:回归是应用统计学中最受欢迎的工具之一。然而,回归分析结果的误用和误解在生物医学研究中是常见的。本文运用统计理论和模拟研究来说明有关这种普遍使用的统计方法的一些悖论。我们还特别指出在顶级医学期刊发表的很多文章中广泛使用的模型选择程序事实上是错误的。模型选择使用哪一种步骤化程序需基于可靠的统计理论。
Overview: Regression is one of the most popular tools in applied statistics. However, the misuse and misunderstanding of regression analysis results are common in biomedical research. This article uses statistical theory and simulation studies to illustrate some of the paradoxes about this commonly used statistical method. We also point out in particular that the model selection process widely used in many articles published in top medical journals is in fact wrong. Which step-by-step procedure the model chooses to use depends on sound statistical theory.