Virginia Commonwealth University, VA, USA.
University of Richmond, VA, USA.
Pers Soc Psychol Bull. 2020 Sep;46(9):1344-1362. doi: 10.1177/0146167220903896. Epub 2020 Feb 24.
Selective reporting practices (SRPs)-adding, dropping, or altering study elements when preparing reports for publication-are thought to increase false positives in scientific research. Yet analyses of SRPs have been limited to self-reports or analyses of pre-registered and published studies. To assess SRPs in social psychological research more broadly, we compared doctoral dissertations defended between 1999 and 2017 with the publications based on those dissertations. Selective reporting occurred in nearly 50% of studies. Fully supported dissertation hypotheses were 3 times more likely to be published than unsupported hypotheses, while unsupported hypotheses were nearly 4 times more likely to be dropped from publications. Few hypotheses were found to be altered or added post hoc. Dissertation studies with fewer supported hypotheses were more likely to remove participants or measures from publications. Selective hypothesis reporting and dropped measures significantly predicted greater hypothesis support in published studies, supporting concerns that SRPs may increase Type 1 error risk.
选择性报告做法（SRPs）——在准备发表报告时添加、删除或改变研究元素——被认为会增加科学研究中的假阳性。然而，对 SRPs 的分析仅限于自我报告或对预先注册和已发表研究的分析。为了更广泛地评估社会心理学研究中的 SRPs，我们将 1999 年至 2017 年期间辩护的博士论文与基于这些论文的出版物进行了比较。选择性报告发生在近 50%的研究中。得到充分支持的论文假设比未得到支持的假设更有可能被发表，而未得到支持的假设则更有可能从出版物中删除。很少有假设被发现是事后修改或添加的。支持假设较少的论文研究更有可能从出版物中删除参与者或测量。选择性假设报告和删除的措施显著预测了发表研究中更大的假设支持，支持了 SRPs 可能会增加 1 型错误风险的担忧。
