STEM-R: Modeling STEM Retention and Departure across Physics, Mathematics, and Engineering

西弗吉尼亚大学的研究人员将开发一个新的STEM离职理论框架，详细说明学生离开STEM专业的原因。该研究扩展了Tinto的大学离校模式，包括学生离开STEM但留在大学的职业探索过程。该框架将通过广泛测量人口统计、社会、学术、情感（自我效能感、自尊、归属感）、职业探索/抱负和心理变量，在许多STEM专业要求的物理和数学入门课程序列的四个纵向点上进行开发和测试。纵向测量监测学生STEM职业承诺和职业认同的演变，并确定影响决定开始STEM离开过程或导致经过充分调查的决心追求STEM职业的因素。该研究将探讨影响大学STEM离职的关键问题，包括通过修改建议协议可以在多大程度上预防离职，在这些门卫班的表现及其对自我效能的影响在多大程度上影响转专业的决定，哪些心理/社会因素标志着学生开始探索非STEM职业，以及归属在多大程度上影响留任，特别是对代表性不足的女性和农村学生。这项研究将有助于更深入地了解这一重要现象。它还将通过产生一个在大量经济多样化的学生群体中得到验证的STEM离校理论框架，为全国关于STEM保留的讨论提供信息。研究设计将实施多阶段分析和测量，目标是对STEM离职过程进行建模，并量化离职过程的原因和标志。研究人员将深入调查：(1)大学离职的理论框架需要哪些修改来解释大学内部从STEM学科的离职，(2)预测STEM离职所需的最低信息以及STEM离职的来源，(3)可以通过干预措施预防STEM离职的比例，(4)对于服务不足的人群，STEM离职标志有何不同，以及(5)如何确定干预措施最有效的学生亚群体。研究人员将利用机构的上课和结果数据，为个别班级和上课序列建立一套“生存”概率。然后使用线性回归分析来分析这些概率如何受到以下因素的影响：(1)学生的内在特征（通过ACT/SAT、种族和民族、社会经济地位、第一代地位和个性特征衡量的能力），(2)学生的学业准备（高中GPA和高中课程学习模式），(3)学生在当前班级的学习成绩（通过出勤模式和作业成绩衡量），(4)学生当前与家庭、大学结构和社区的社会联系；(5)学生当前的情感状态、自我效能感、归属感、自尊和STEM认同。潜在增长模型将跟踪这些因素随时间的变化。同时，对学生的职业认同状态、职业探索和职业决策状态进行监测。使用四点纵向测量，学生职业认同状态的转变（特别是表明新的职业输出过程的变化，因此STEM离开的威胁）将与先前生存概率的变化以及学生情感、学术或社会状态的变化相关。将制定签名通知有风险学生的顾问，并对咨询协议进行一般性修改。将确定促进留校的校园社会/学术结构。在所有层面上，对代表性不足的学生的差异结果进行调查，目的是设计针对这些人群的干预措施。本项目由美国国家科学基金会EHR核心研究（ECR）项目支持。ECR项目强调基础STEM教育研究，在三个领域产生基础知识：STEM学习和学习环境，扩大参与和劳动力发展。

Researchers at West Virginia University will develop a new theoretical framework for STEM departure that will detail the reasons why students leave STEM majors. The research extends Tinto's university departure model to include the career exploration process where a student leaves STEM but remains in college. The framework will be developed and tested by extensive measurement of demographic, social, academic, affective (self-efficacy, self-esteem, belonging), career exploration/aspirations and psychological variables at four longitudinal points in physics and mathematics introductory class sequences required for many STEM majors. The longitudinal measurement monitors evolution of a student's STEM career commitment and vocational identity, and determines factors that influence a decision to begin the process of STEM departure or lead to a well-investigated resolution to pursue a STEM career. The research will explore crucial questions influencing university STEM departure, including to what extent departure is preventable by modifying advising protocols, to what extent performance in these gate-keeper classes and its effect on self-efficacy influence the decision to change major, what psychological/social factors mark students beginning to explore non-STEM careers, and to what extent belonging influences retention, particularly of underrepresented women and rural students. The research will contribute to a deeper understanding of this important phenomenon. It also will inform the national discussion of STEM retention by producing a theoretical framework of STEM departure validated across a large, economically diverse pool of students. The research design will implement a multi-stage analysis and measurement with the goal of modeling the process of STEM departure and quantifying the reasons for and markers of the departure process. Researchers will investigate in depth: (1) what modifications are needed in theoretical frameworks for college departure to explain intra-university departure from STEM disciplines, (2) the minimum information required to predict STEM departure and where STEM departure originates, (3) what fraction of STEM departure could be prevented by interventions, (4) how STEM departure markers differ for underserved populations, and (5) how to identify subpopulations of students where interventions will be most effective. Researchers will use institutional class-taking and outcome data to build a set of "survival" probabilities for individual classes and class-taking sequences. Linear regression analysis then be used to analyze how these probabilities are affected by (1) the student's intrinsic characteristics (ability measured by ACT/SAT, race and ethnicity, socioeconomic status, first generation status, and personality profile), (2) the student's academic preparation (high school GPA and high school course-taking patterns), (3) the student's academic performance in his or her current class (measured by attendance patterns and assignment grades), (4) the students current social connectedness to both family and university structures and communities, and (5) the student's current affective state, his or her self-efficacy, sense of belonging, self-esteem, and STEM identity. Latent growth modeling will track changes in these factors over time. In parallel, the student's vocational identity status, his or her state of career exploration and career decision making, will be monitored. Using the four-point longitudinal measurement, transitions in the student's vocational identity state (particularly changes that indicate a renewed career exportation process and therefore the threat of STEM departure) will be correlated with previous changes in survival probabilities and with changes in the student's affective, academic, or social state. Signatures to inform advisers of at risk students and general changes to advising protocols will be developed. Campus social/academic structures that promote retention will be identified. At all levels, differential results for underrepresented students will be investigated with the goal of designing interventions that target these populations.This project is supported by NSF's EHR Core Research (ECR) program. The ECR program emphasizes fundamental STEM education research that generates foundational knowledge in three areas: STEM learning and learning environments, broadening participation, and workforce development.