此文旨在研究影子教育对青少年学生成绩的影响，参考中国人民大学中国教育追踪调查数据（CEPS），选取七年级、九年级学生及其家长数据，分类研究。在研究方法上主要使用倾向得分匹配法（PSM），结合Logit逻辑回归模型、Tobit模型，研究影子教育决策、时间、支出与学生文化课成绩之间的关系。通过研究发现：1.青少年学生在学业成绩上的表现，受影子教育决策影响，参加影子教育对成绩起到一定程度上促进作用，但不同的科目在显著性上不尽一致。2.学生分配在影子教育的时间影响学生的文化课成绩，二者存在统计上正向影响的关系。3.家庭在影子教育经费上支出也影响成绩，在一定程度上，虽呈正向关系，但不同的科目显著性存在差异。4.在城乡户口的群组分析中，影子教育决策对语文、数学成绩的效应不受城乡户口区别的影响，对英语成绩的效应受城乡因素的影响，且城镇学生参加影子教育决策对成绩的正向效应大于农村学生参加影子教育决策对成绩的正向效应。5.在性别的群组分析中，影子教育决策对语文、英语成绩的效应受性别影响，而对数学成绩的效应不受性别因素影响。且男生参加影子教育在语文成绩上的影响效应小于女，而在数学成绩上的影响效应大于女。与此同时，分析协变量对影子教育决策的逻辑倾向，父母文化程度、家庭结构、科技水平、籍贯、户口类别、教育期望、经济条件对影子教育决策存在联系。此文政策建议：常规教育体制应与影子教育并驾齐驱。1.影子教育方面：影子教育应配合常规教育发展，结合互联网，应用“共享+”思维，同时提高影子教育机构的准入门槛，实现平台与学习资源均衡。2.教育体制方面：改革教育模式、选拔机制，限制影子教育机构的质量，选拔综合型教育人才，提高教师队伍教学水平。

The purpose of this paper is to study the influence of shadow education on the achievement of young students. Referring to the China Education tracking survey data (CEPs) of Renmin University of China, the data of seventh and ninth grade students and their parents are selected for classified research. In terms of research methods, we mainly use the tendency score matching method (PSM), combined with logit logistic regression model and Tobit model, to study the relationship between shadow education decision-making, time, expenditure and students' cultural course performance. The findings are as follows: 1. The performance of adolescent students in academic performance is affected by shadow education decision-making, and participation in shadow education plays a role in promoting performance to a certain extent, but different subjects are not consistent in significance. 2. The time allocated by students in shadow education affects students' achievement in culture course, and there is a statistically positive relationship between them. 3. Family expenditure on shadow education also affects performance. To some extent, although there is a positive relationship, there are significant differences in different subjects. 4. In the group analysis of urban and rural household registration, the effect of shadow education decision-making on Chinese and math performance is not affected by the difference between urban and rural household registration, and the effect on English performance is affected by urban and rural factors. The positive effect of urban students' participation in shadow education decision-making on performance is greater than that of rural students' participation in shadow education decision-making on performance. 5. In the group analysis of gender, the effect of shadow education decision-making on Chinese and English performance is influenced by gender, while the effect on mathematics performance is not affected by gender. Moreover, the effect of boys' participation in shadow education on Chinese achievement is less than that of girls, and the effect on mathematics achievement is greater than that of girls. At the same time, it analyzes the logical tendency of covariates to shadow education decision-making, and the relationship between parents' education level, family structure, technology level, native place, household registration type, education expectation and economic conditions to shadow education decision-making. This paper suggests that the regular education system should keep pace with shadow education. 1. Shadow education: shadow education should cooperate with the development of conventional education, combine with the Internet, apply "sharing +" thinking, improve the access threshold of shadow education institutions, and achieve the balance between platform and learning resources. 2. Education system: reform the education mode and selection mechanism, limit the quality of shadow education institutions, select comprehensive education talents, and improve the teaching level of teachers.