Search for: All records

Multiple stakeholders are interested in measuring undergraduate student success in college across academic fields. Different metrics might appeal to different stakeholders. Some metrics such as the fraction of first-time, full-time students who start in the fall who graduate within six years, the graduation rate, are federally mandated by the U.S. Department of Education, Integrated Postsecondary Education Data System (IPEDS). We argue that this calculation of graduation rate is inherently problematic because it excludes up to 60% of students who transfer into an institution, enroll part-time, or enroll in terms other than the fall. By expanding the starters definition, we propose a graduation rate definition that includes conventionally excluded students and provides information on progression in a specific program. Stickiness is an even more-inclusive alternative, measuring a program’s success in graduating all undergraduates ever enrolled in the program. In this work, programs are grouped into six academic fields: Arts and Humanities, Business, Engineering, Other, Social Sciences, and STM (Science, Technology, and Mathematics. Stickiness is the percentage of students who ever enroll in an academic field that graduate in the same field. We use the Multiple Institution Dataset for Investigating Engineering Longitudinal Development (MIDFIELD) 2023 which contains unit-record data for over 2 million individual students at 19 institutions. For the academic fields studied, Engineering has the highest graduation rate and third highest stickiness. Social Sciences and Business also have higher graduation rates and stickiness than the other fields. We also track the relative fraction of students migrating to and from each academic field. This paper continues our work to derive better metrics for understanding student success. 

A substantial investment by the National Science Foundation (NSF), including awards from Engineering Education and Centers in the Engineering Directorate and the Division of Undergraduate Education in the Education and Human Resources Directorate, has led to the creation and study of the Multiple Institution Database for Investigating Engineering Longitudinal Development (MIDFIELD). This large database of student records has yielded groundbreaking research on student pathways by a small interdisciplinary team of researchers. The team has shown that while individual engineering programs may have poor graduation rates, a multi-institutional view reveals that engineering programs as a whole graduate a larger fraction of students than other groups of disciplines. The team has also shown that women and men have similar graduation rates in engineering, likely a result of efforts to make engineering education a welcoming environment for women and the high academic credentials of the women who do study engineering. As with the overall graduation rate, individual institutions and programs can and do have outcomes that depart from this aggregate perspective. A comprehensive study of student pathways in various engineering disciplines provided practitioners with rich information specific to their disciplinary context. The team has also designed a variety of metrics that have provided researchers and practitioners with an improved understanding of student pathways. The quality of the data source and the research team is attested by these substantial findings, multiple best paper awards, and other recognitions. This paper provides updates on transitioning MIDFIELD to the American Society for Engineering Education (ASEE), documentation of institutional policies, and supporting a growing community of researchers in using the database including the second offering of the MIDFIELD Institute. This work is supported by the NSF Division of Engineering Education and Centers. 

Longitudinal, student-level data are a rich resource for characterizing how students navigate the terrain of higher education. Learning to work effectively with such data, however, can be a challenge. In this paper, we share some of our experiences over years of conducting research with the Multiple Institution Database for Investigating Engineering Longitudinal Development (MIDFIELD). MIDFIELD contains individual student-level records for all undergraduate students at 19 US institutions with over 1.7 million unique students. This paper focuses on our lessons learned about processing longitudinal data to prepare it for analysis. We describe and define the steps that we take to process the data including filtering for data sufficiency, degree-seeking, and program (major), then classifying by completion status and demographics. We use the examples of calculation of graduation rate and stickiness to show the details of how the processed data is used in analysis. We hope this paper will help introduce the landscape of longitudinal research to a wider audience and provide tips for working with this valuable resource.