Colleges and universities have long sought to improve their students’ adjustment, retention, and likelihood of graduation (Berger, Ramírez, & Lyons, 2012). To help with this goal, institutions have increasingly attempted to identify students who are struggling early in college so that they can provide timely support and assistance. These early efforts can include a variety of institutional assessments or commercial tools. Some research has shown that social adjustment and engagement measures gathered approximately 1 month into the first semester can predict short-term college retention (Bowman, Miller, Woosley, Maxwell, & Kolze, 2019; Sun, Hagedorn, & Zhang, 2016). These results support the potential importance of understanding these initial experiences and perceptions; however, a problem with this strategy is that it depends on students voluntarily completing a survey, and students who are struggling to adjust may be the least likely to respond. A second problem is that surveys are introspective by nature, so even data obtained from students who do respond are prone to error (e.g., Herzog & Bowman, 2011). Therefore, the present study explored an alternative to traditional early-alert systems at residential institutions: using campus dining data to create an indicator of students’ social networks. Key benefits of this approach are that students do not need to respond to institutional requests to provide information and that data collection is automatic and starts almost immediately when students arrive on campus. As a result, it is possible to use student behavior from as early as the first week or two of classes to inform individualized outreach efforts and support.

中文翻译：

---

学生社交网络的早期识别：通过校园用餐预测大学留校率和毕业率

长期以来，学院和大学一直致力于提高学生的适应能力、保留率和毕业可能性（Berger、Ramírez 和 Lyons，2012 年）。为了帮助实现这一目标，机构越来越多地尝试识别在大学早期遇到困难的学生，以便他们能够及时提供支持和帮助。这些早期的努力可以包括各种机构评估或商业工具。一些研究表明，第一学期大约 1 个月收集的社会适应和参与措施可以预测短期大学留校率（Bowman、Miller、Woosley、Maxwell 和 Kolze，2019 年；Sun、Hagedorn 和 Zhang，2016 年）。这些结果支持理解这些初始体验和感知的潜在重要性；然而，这种策略的一个问题是它取决于学生自愿完成调查，而努力适应的学生可能最不可能做出回应。第二个问题是调查本质上是内省的，因此即使从做出回应的学生那里获得的数据也容易出错（例如，Herzog & Bowman，2011）。因此，本研究探索了住宿机构传统早期预警系统的替代方案：使用校园餐饮数据创建学生社交网络的指标。这种方法的主要好处是学生不需要响应机构要求提供信息，并且数据收集是自动的，并且在学生到达校园时几乎立即开始。因此，