Developing the ability to self-assess is a crucial skill for students, as it impacts their academic performance and learning strategies, amongst other areas. Most existing research in this field has concentrated on the exploration of the students’ capacity to accurately assign a score to their work that closely mirrors an expert’s evaluation, typically a teacher’s. Though this process is commonly referred to as self-assessment, a more precise term would be self-assessment scoring accuracy. Our aim is to review what is the average accuracy and what moderators might influence this accuracy. Following PRISMA recommendations, we reviewed 160 articles, including data from 29,352 participants. We analysed 9 factors as possible moderators: (1) assessment criteria; (2) use of rubric; (3) self-assessment experience; (4) feedback; (5) content knowledge; (6) incentive; (7) formative assessment; (8) field of knowledge; and (9) educational level. The results showed an overall effect of students’ overestimation (g = 0.206) with an average relationship of z = 0.472 between students’ estimation and the expert’s measure. The overestimation diminishes when students receive feedback, possess greater self-assessment experience and content knowledge, when the assessment does not have formative purposes, and in younger students (primary and secondary education). Importantly, the studies analysed exhibited significant heterogeneity and lacked crucial methodological information.

培养自我评估的能力对学生来说是一项至关重要的技能，因为它会影响他们的学业成绩和学习策略等。该领域现有的大多数研究都集中在探索学生准确为其作业打分的能力，该分数密切反映了专家（通常是教师）的评估。尽管此过程通常称为自我评估，但更准确的术语是自我评估评分准确性。我们的目标是审查平均准确度是多少以及哪些审核者可能会影响该准确度。根据 PRISMA 建议，我们审查了 160 篇文章，其中包括来自 29,352 名参与者的数据。我们分析了 9 个可能的调节因素：（1）评估标准；(2) 量规的使用；（三）自我评价经历；（4）反馈；（五）知识内容；（六）激励；(7) 形成性评估；（八）知识领域；（九）文化程度。结果显示学生高估的总体影响（g = 0.206）  ，学生估计与专家测量之间 的平均关系为z = 0.472。当学生收到反馈、拥有更多的自我评估经验和内容知识、评估不具有形成性目的以及年龄较小的学生（小学和中学教育）时，高估就会减少。重要的是，分析的研究表现出显着的异质性，并且缺乏关键的方法学信息。