Mabire-Yon Renaud
UMR1296 'Radiations: Defense, Health, Environment', INSERM, University of Lyon 2, Bron, France.
Int J Psychol. 2025 Apr;60(2):e70006. doi: 10.1002/ijop.70006.
In applied psychology, traditional statistical methods often provide only a broad overview, potentially overlooking nuanced variable relationships. This article presents a comprehensive tutorial on quantile regression (QR), a statistical modelling technique ideally suited for psychological data analysis. Unlike conventional regression, QR examines relationships across different quantiles of the data distribution, revealing complex dynamics and offering robustness to non-normality and heteroscedasticity. We demonstrate its utility through a practical example, analysing the relationship between age and life satisfaction, supported by annotated R code. The tutorial emphasises grounding QR in a sound theoretical framework and introduces the quantile loss approach as an alternative to p value interpretation. By providing both theoretical understanding and practical tools, this tutorial aims to empower researchers to improve the depth and reproducibility of their findings in psychological research.
在应用心理学中，传统统计方法往往只能提供一个宽泛的概述，可能会忽略细微的变量关系。本文提供了一篇关于分位数回归（QR）的全面教程，分位数回归是一种非常适合心理数据分析的统计建模技术。与传统回归不同，QR研究数据分布不同分位数之间的关系，揭示复杂的动态关系，并对非正态性和异方差具有稳健性。我们通过一个实际例子展示了它的实用性，分析年龄与生活满意度之间的关系，并辅以带注释的R代码。本教程强调将QR建立在坚实的理论框架基础上，并引入分位数损失方法作为p值解释的替代方法。通过提供理论理解和实用工具，本教程旨在使研究人员能够提高其在心理学研究中发现的深度和可重复性。
