Peers influence students’ academic decisions and outcomes. For example, several studies with strong claims to causality demonstrate that peers affect the choice of and persistence in majors. One remaining issue, however, has stymied efforts to translate this evidence into actionable interventions: the literature has not grappled adequately with the fact that in natural settings, students typically select most of their peers. The bulk of causal evidence for peer influence comes from exogenously assigned peers (e.g., roommates) because peer effects are easier to identify in such cases. However, students do not form their most important ties for the convenience of scientific inference. In order to link theory and practice, we need to understand which peers are influential. We employ longitudinal, multiplex network data on students’ choices of and persistence in their majors from 1260 students across 14 universities to identify likely causal pathways of peer influence via self-selected peers. We introduce time-reversed analysis as a novel tool for addressing some selection concerns in network influence studies. We find that peers with whom a student reports merely spending time, rather than—e.g., close friends, study partners, esteemed peers—consistently and potently influence their college major choice.

中文翻译：

---

同伴影响对专业选择的途径

同伴影响学生的学术决定和结果。例如，几项强烈主张因果关系的研究表明，同龄人会影响专业的选择和坚持。然而，剩下的一个问题阻碍了将这一证据转化为可行的干预措施的努力：文献没有充分考虑这样一个事实：在自然环境中，学生通常会选择大多数同龄人。同伴影响的大部分因果证据来自外生分配的同伴（例如室友），因为在这种情况下同伴效应更容易识别。然而，学生并不是为了科学推理的方便而形成他们最重要的联系。为了将理论与实践联系起来，我们需要了解哪些同行具有影响力。我们利用来自 14 所大学 1260 名学生的纵向、多重网络数据来了解学生对专业的选择和坚持，以确定通过自我选择的同伴产生同伴影响的可能因果路径。我们引入时间反转分析作为解决网络影响力研究中一些选择问题的新工具。我们发现，学生自称只是共度时光的同龄人——例如亲密的朋友、学习伙伴、受人尊敬的同龄人——持续而有力地影响着他们的大学专业选择。