Can Student Characteristics be Used to Effectively Identify Students At-Risk in the Online STEM Environment?

EHR核心研究计划资助的提案将有助于综合，建立和/或扩大STEM（科学，技术，工程和数学）教育以下领域的研究基础：STEM学习，STEM学习环境，STEM劳动力发展，以及扩大STEM的参与。STEM教育管道在大学中显着缩小。社区大学服务于一些最多样化的受众，并且越来越多地使用在线学习作为提供STEM教学的更便宜的方式;此外，大规模开放式在线课程（通常称为MOOC）正在被提议作为学分制教学的替代方案。先前的研究表明，在线学习环境对不同类型的学生产生不同的影响。这个基于社区大学的研究项目提出了以下问题：社区大学层面的STEM在线学习是否可能比其他人更能影响代表性不足的群体，以及它会产生积极或消极的影响？我们能否确定哪些学生最适合在线教学还是面对面教学，或者对在线环境中的“高危”学生进行干预？该项目旨在通过使用两个重要的数据集来回答这些问题：一个是由纽约系统的六所学校组成的数据集，该系统为全国最多样化的学生团体之一提供服务，其中超过50，000名学生在线学习STEM课程。第二个是来自国家教育统计中心的大规模全国数据集，其中包含人口统计学，学术，个人和财务变量。只有一小部分关于在线学习的研究严格控制了学生对在线课程的自主选择。这项研究将探讨具有特定特征的学生不仅在在线STEM课程中表现更好或更差的程度，而且在大学之后，与参加类似的面对面STEM课程的学生进行匹配比较。该项目使用混合方法。定量分析将包括主成分因子分析、逻辑回归、线性回归、方差和协方差分析、广义线性混合模型、倾向分数匹配和敏感性分析，以检查课程和大学成果，包括课程保留率（到课程第十周结束时的出勤率）和成功完成课程（在课程中获得C-或更好的成绩），学生是否在课程结束后立即重新入学，以及在一年，两年，三年和六年的坚持。也将使用总平均成绩、累积的学分数以及这些间隔的转学率和毕业率。要建模的自变量和协变量包括在线与混合与离线STEM课程格式，以及各种人口统计变量，包括努力资本，社会资本，文化资本，金融资本，人力资本和习性。定性访谈和深入调查将被用来探索大规模数据集中发现的趋势，并将专门与纽约市立大学系统的在线教师进行调查。数据将被探索，以模拟什么变量有助于差异的“风险”在线。该项目的智力价值在于推进我们对在线选项如何差异化地帮助或阻碍不同类型的中学后STEM学生的理解。对于更广泛的影响，该模型的结果可以作为实施有针对性的干预措施的基础，可以为有风险的学生提供额外的辅导，辅导，技术支持，学习或在在线课程中取得成功所需的技能和行为培训;或者建议他们参加类似的面对面课程。这些政策影响将在该项目主办的为期一天的电子学习会议上讨论。

The EHR Core Research Program funds proposals that will help synthesize, build and/or expand research foundations in the following areas of STEM (Science, Technology, Engineering, and Mathematics) Education: STEM learning, STEM learning environments, STEM workforce development, and broadening participation in STEM. The STEM education pipeline narrows significantly in college. Community colleges serve some of the most diverse audiences, and are increasingly using online learning as a cheaper way to provide STEM instruction; additionally Massive Open Online Courses (commonly known as MOOCs) are being proposed as alternatives to credit-bearing instruction. Prior research shows that online learning environments impact different kinds of students differently. This research project based at a community college asks questions such as the following: Is this move towards STEM learning online at the community college level likely to impact underrepresented groups more than others, and will it have positive or negative impact? Can we identify which students are best served by online vs. face-to-face instruction or conduct interventions for students 'at-risk' in the online environment? This project aims to answer these questions by using two important datasets: one is a dataset to be assembled from six schools in the CUNY (City University of New York) system, which serves one of the most diverse student bodies in the country, and in which over 50,000 students have taken STEM courses online. The second is a large-scale national dataset from the National Center for Education Statistics which contains demographic, academic, personal, and financial variables.Only a small proportion of the research conducted on online learning has controlled for student self-selection into online courses in a rigorous way. This study will explore the extent to which students with particular characteristics fare better or more poorly not only in online STEM courses, but in college afterwards, with a matched comparison to students who take comparable face-to-face STEM courses. The project uses mixed methods. Quantitative analysis will include principal component factor analysis, logistic regression, linear regression, analysis of variance and covariance, generalized linear mixed models, propensity score matching, and sensitivity analysis to examine course and college outcomes including course retention (attendance through the end of the tenth week of classes) and successful course completion (earning a C- or better in the course), whether students re-enrolled in the semester immediately following the course, and persistence at one, two, three, and six years. Overall grade point average, the number of credits accumulated, and transfer and graduation rates at these intervals will also be used. Independent variables and covariates to be modeled include online vs. hybrid vs. offline STEM course format, and a variety of demographic variables including effort capital, social capital, cultural capital, financial capital, human capital, and habitus. Qualitative interviews and in-depth surveys will be used to explore the trends found in the large scale datasets, and a survey will be conducted specifically with online instructors in the CUNY system. Data will be explored to model what variables contribute to differential 'risk' online. The intellectual merit of the project rests on advancing our understanding of how online options differentially help or hinder different kinds of postsecondary STEM students. For broader impacts, the results of the model could be used as the basis for implementation of targeted interventions, either by providing at-risk students with additional mentoring, tutoring, technical support, advisement, or training in skills and behaviors necessary to succeed in an online course; or by advising them to enroll in a comparable face-to-face course instead. These policy implications will be discussed at a culminating one-day conference on elearning hosted by the project.