Mazei Jens, Rudolph Cort W, Zacher Hannes, Hüffmeier Joachim
Department of Psychology, TU Dortmund University.
Department of Psychology, Wayne State University.
J Appl Psychol. 2025 May 5. doi: 10.1037/apl0001291.
A multiverse analysis allows researchers to systematically evaluate the support for a hypothesis across a range of sensible ways in which data can be prepared for statistical analysis and/or be analyzed. Accordingly, multiverse analysis provides insights into the relevance of different approaches to, for instance, dealing with outliers or attrition, creating scales, or using different measures for the same construct. The goal of this article is to illustrate the usefulness of multiverse analysis for research in applied psychology and to guide researchers in conducting a multiverse analysis. To do so, we provide a detailed process model of the typical stages involved in conducting a multiverse analysis (along with a shortened version depicting multiverse analysis "at a glance"), as well as a designated, corresponding preregistration template for multiverse analysis. To showcase the merits of a multiverse analysis, we also evaluate two exemplary hypotheses regarding employees' experience of commuting to and from work. We observed that the results of these hypothesis tests varied strongly depending on how common decisions were made. As such, multiverse analysis represents an important tool for exploring the robustness of knowledge at the level of individual studies, even before a replication is conducted. Hence, multiverse analysis can strengthen the transparency and openness of empirical work. (PsycInfo Database Record (c) 2025 APA, all rights reserved).
多宇宙分析使研究人员能够系统地评估在一系列合理方式下对假设的支持程度，这些方式涉及为统计分析准备数据和/或进行数据分析。因此，多宇宙分析能够洞察不同方法在处理异常值或损耗、创建量表或对同一构念使用不同测量方法等方面的相关性。本文的目的是说明多宇宙分析在应用心理学研究中的有用性，并指导研究人员进行多宇宙分析。为此，我们提供了一个详细的多宇宙分析典型阶段过程模型（以及一个描绘多宇宙分析“一目了然”的简化版本），还有一个专门的、相应的多宇宙分析预注册模板。为了展示多宇宙分析的优点，我们还评估了两个关于员工上下班通勤体验的示例性假设。我们观察到，这些假设检验的结果因常见决策方式的不同而有很大差异。因此，即使在进行复制研究之前，多宇宙分析也是探索个体研究层面知识稳健性的重要工具。因此，多宇宙分析可以增强实证研究工作的透明度和开放性。（PsycInfo数据库记录（c）2025美国心理学会，保留所有权利）
