Dowdy Art, Hantula Donald A, Travers Jason C, Tincani Matt
Department of Teaching and Learning, Temple University, Philadelphia, PA USA.
College of Education and Human Development, Temple University, 1301 Cecil B. Moore Avenue, Philadelphia, PA 19122 USA.
Perspect Behav Sci. 2021 Jul 21;45(1):37-52. doi: 10.1007/s40614-021-00303-0. eCollection 2022 Mar.
Publication bias is an issue of great concern across a range of scientific fields. Although less documented in the behavior science fields, there is a need to explore viable methods for evaluating publication bias, in particular for studies based on single-case experimental design logic. Although publication bias is often detected by examining differences between meta-analytic effect sizes for published and grey studies, difficulties identifying the extent of grey studies within a particular research corpus present several challenges. We describe in this article several meta-analytic techniques for examining publication bias when published and grey literature are available as well as alternative meta-analytic techniques when grey literature is inaccessible. Although the majority of these methods have primarily been applied to meta-analyses of group design studies, our aim is to provide preliminary guidance for behavior scientists who might use or adapt these techniques for evaluating publication bias. We provide sample data sets and R scripts to follow along with the statistical analysis in hope that an increased understanding of publication bias and respective techniques will help researchers understand the extent to which it is a problem in behavior science research.
发表偏倚是一系列科学领域中备受关注的问题。尽管行为科学领域对此记录较少，但仍有必要探索评估发表偏倚的可行方法，特别是对于基于单病例实验设计逻辑的研究。虽然发表偏倚通常通过检查已发表研究和灰色研究的元分析效应大小之间的差异来检测，但在特定研究语料库中确定灰色研究的范围存在困难，这带来了几个挑战。我们在本文中描述了几种元分析技术，用于在既有已发表文献又有灰色文献时检查发表偏倚，以及在无法获取灰色文献时的替代元分析技术。尽管这些方法大多主要应用于组设计研究的元分析，但我们的目的是为可能使用或改编这些技术来评估发表偏倚的行为科学家提供初步指导。我们提供样本数据集和R脚本以配合统计分析，希望对发表偏倚和相关技术的更多了解将有助于研究人员了解它在行为科学研究中的问题程度。
