ISSN 1671-3710
CN 11-4766/R
主办:中国科学院心理研究所
出版:科学出版社

心理科学进展 ›› 2016, Vol. 24 ›› Issue (9): 1504-1518.doi: 10.3724/SP.J.1042.2016.01504

• 研究方法 • 上一篇    

心理学研究中的可重复性问题:从危机到契机

胡传鹏1;王 非1;过继成思1;宋梦迪1;隋 洁2;彭凯平1   

  1. (1清华大学心理学系, 北京 100084)
  • 收稿日期:2016-01-02 出版日期:2016-09-15 发布日期:2016-09-15
  • 通讯作者: 彭凯平, E-mail: pengkp@mail.tsinghua.edu.cn; 隋 洁, E-mail: jie.sui@gmail.com
  • 基金资助:

    国家自然科学基金项目(31371017/31471001)资助。

The replication crisis in psychological research

HU Chuanpeng1; WANG Fei1; GUO Jichengsi1; SONG Mengdi1; SUI Jie2; PENG Kaiping1   

  1. (1 Department of Psychology, Tsinghua University, Beijing 100084, China) (2 Department of Experimental Psychology, University of Oxford, Oxford, UK, OX1 3UD )
  • Received:2016-01-02 Online:2016-09-15 Published:2016-09-15
  • Contact: PENG Kaiping, E-mail: pengkp@mail.tsinghua.edu.cn; SUI Jie, E-mail: jie.sui@gmail.com

摘要:

可重复性问题是当前科学界面临的共同问题。最近, 心理学研究领域的可重复性问题也受到广泛关注, 引起了研究者的积极讨论与探索。通过对2008年发表的100项研究结果进行大规模重复实验, 研究者发现, 心理学研究的成功重复率约为39%, 但该研究仍然存在着巨大的争议, 不同的研究者对其结果的解读不尽相同。针对可重复性问题, 研究者通过数据模拟、元分析以及调查等多种方法来分析和探索其原因, 这些研究表明, 可重复性问题本质上可能是发表的研究假阳性过高, 可疑研究操作是假阳性过高的直接原因, 而出版偏见和过度依赖虚无假设检验则是更加深层的原因。面对可重复性问题, 研究者从统计方法和研究实践两个方面提出了相应的解决方案, 这些方法与实践正在成为心理学研究的新标准。然而, 要解决可重复性问题, 还需要心理学研究领域的多方参与, 尤其是在政策上鼓励公开、透明和开放的研究取向, 避免出版偏见。心理学研究者为解决可重复性问题做出的努力, 不仅会加强心理学研究的可靠性, 也为其他学科解决可重复问题提供了借鉴, 推动科学界可重复问题的解决。

关键词: 可重复性, 假阳性, 基于估计的统计, 开放科学, 革新

Abstract:

The recent replication crisis in psychology has raised increasing concerns. Given that it is important for psychologists to understand the replication crisis and the consequent changes in the field, the present review aimed at providing an overview of the relevant events and highlighting the methodological reforms in psychology. First, we briefly recapitulated the events that raised doubt about reliability of psychological research. Then, we focused on a recent large-scale replication study (Reproducible Project: Psychology) and the following debates. In addition, three underlying causes of the replication crisis were examined: 1) the overabundance of false positive results, which is mainly caused by the prevalence of questionable research practices, 2) publication bias, and 3) the over-reliance on Null-Hypothesis Significant Testing (NHST) and misunderstanding of p-values, which was confirmed by our survey among Chinese psychology students and academics. More importantly, the proposed solutions provided in the literature were also reviewed, e.g., shifting from NHST to alternative statistical methods and adopting new ways to enhance research integrity (i.e., pre-registration, full disclosure, and open data and materials of studies). Finally, we call for joint efforts by policy makers, researchers and educators to solve the replication crisis.

Key words: replicability, false positive, estimate-based statistic, open science, reform