This article reviews educational efforts to promote a responsible conduct of research (RCR) that were reported in scientific publications between 1990 and early 2020. Unlike previous reviews that were exploratory in nature, this review aimed to test eleven hypotheses on effective training strategies. The achievement of different learning outcomes was analyzed independently using moderator analysis and meta-regression, whereby 75 effect sizes from 30 studies were considered. The analysis shows that the achievement of different learning outcomes ought to be investigated separately. The attainment of knowledge strongly benefited from individualized learning, as well as from the discussion and practical application of ethical standards. Contrarily, not covering ethical standards tended to be a feature of successful courses, when looking at other learning outcomes. Overall, experiential learning approaches where learners were emotionally involved in thinking about how to deal with problems were most effective. Primarily intellectual deliberation about ethical problems, often considered the “gold standard” of ethics education, was significantly less effective. Several findings from previous reviews, e.g., the preferability of mono-disciplinary groups, could not be replicated with multivariate analysis. Several avenues for future research efforts are suggested to advance knowledge on the effectiveness of research integrity training.

本文回顾了 1990 年至 2020 年初科学出版物中报道的促进负责任研究行为 (RCR) 的教育工作。使用调节器分析和元回归独立分析了不同学习成果的成就，其中考虑了来自 30 项研究的 75 个效应量。分析表明，应分别调查不同学习成果的成就。个性化学习以及道德标准的讨论和实际应用极大地受益于知识的获得。相反，不涵盖道德标准往往是成功课程的一个特征，在查看其他学习成果时。总的来说，学习者在情感上参与思考如何处理问题的体验式学习方法是最有效的。主要是关于道德问题的智力审议，通常被认为是道德教育的“黄金标准”，但效果明显较差。先前审查的一些发现，例如单一学科组的可取性，无法用多变量分析复制。建议了未来研究工作的几种途径，以提高有关研究诚信培训有效性的知识。通常被认为是道德教育的“黄金标准”，但效果明显较差。先前审查的一些发现，例如单一学科组的可取性，无法用多变量分析复制。建议了未来研究工作的几种途径，以提高有关研究诚信培训有效性的知识。通常被认为是道德教育的“黄金标准”，但效果明显较差。先前审查的一些发现，例如单一学科组的可取性，无法用多变量分析复制。建议了未来研究工作的几种途径，以提高有关研究诚信培训有效性的知识。