The value of online peer feedback in education has been widely established, and the use of online peer feedback tools is rapidly growing in practice. However, effect sizes appear to vary widely across studies, suggesting implementation details matter substantially. Further, there remain open questions about exactly which aspects of the multi-faceted peer feedback experience are most closely associated with learning outcomes. Within 13 different courses involving 2421 students, drawn across seven universities and six content disciplines, temporally lagged multiple-regression analyses were used to test the unique contributions of quantity, depth, and quality of received and provided comments to students’ growth in task performance across assignments. Meta-regression is applied to precisely estimating overall effect sizes and variation in effect sizes. Results reveal stronger relationships with growth in task performance for 1) provided rather than received comments, 2) longer rather than more comments, and 3) comments perceived to be helpful for revision. Further, there was largely quantitative and qualitative variation in observed relationships across courses that not attributable to statistical noise.

在线同伴反馈在教育中的价值已被广泛确立，在线同伴反馈工具的使用在实践中迅速增长。然而，不同研究的效果大小似乎差异很大，这表明实施细节非常重要。此外，关于多方面同伴反馈体验的哪些方面与学习成果最密切相关，仍然存在悬而未决的问题。在涉及 7 所大学和 6 个内容学科的 2421 名学生的 13 门不同课程中，使用时间滞后的多元回归分析来测试收到的数量、深度和质量的独特贡献，并为学生在任务表现方面的成长提供评论。任务。元回归用于精确估计整体效应大小和效应大小的变化。结果显示，对于 1) 提供的而不是收到的评论，2) 更长而不是更多的评论，以及 3) 被认为有助于修订的评论与任务绩效的增长有更强的关系。此外，跨课程观察到的关系在很大程度上存在定量和定性的变化，这不是统计噪声。