We examine millions of grades and pre-college test scores earned by undergraduates between 2006 and 2019 at a large public US university that became increasingly selective, as measured by admitted student test scores, over that time. A persistent correlation between test score and grades over the period motivates us to employ a simple importance sampling model to address the question, “How much is increased selectivity driving up campus grades?”. Of the overall 0.213 rise in mean undergraduate grade points over the thirteen-year period, our model indicates that nearly half, 0.098±0.004, can be ascribed to increased selectivity. This fraction is nearly 70%, within the more codified domains of engineering, business and natural sciences. Removing selectivity’s influence surfaces curricular-related grade inflation which, over the study period, range by a factor of four, from a low of ∼0.05 in business and engineering to a high of 0.18 in the humanities. The concept of federating this and other types of analysis across multiple campuses is compelling.
This book project, tentatively titled From College to Career, utilizes rich qualitative data to investigate whether graduates of Business and Engineering departments enjoy employment advantages over English and Communications graduates because they learn more in-demand skills for the 21st century. The book shows that in fact most Business and Engineering graduates move seamlessly into jobs that provide good pay, benefits, and room for promotion, but usually engage in mundane office work that rarely requires them to draw on degree-specific skills. On the other hand, many English and Communications majors struggle to enter the labor market and even when they find jobs that require their classroom learning these skills are not treated as valuable or remunerated as if they were. The book, therefore, argues that the shift in higher education from promoting the general liberal arts to the more specific practical disciplines is a misguided practice leaving students not only with a narrower education but also, ironically, less prepared for the workforce.
College students' course-taking decisions are consequential for their academic trajectories, but little is known about how students make these decisions and their implications for demographic inequality in fields of study. This project leverages students' clickstream data from the University of Michigan's Atlas tool, combined with comprehensive student transcript data, to understand what factors are associated with decisions about what courses to consider, and to enroll in conditional on considering. The project utilizes decision models that allow for distinct rules at different stages to examine the decision rules students apply at each stage of the choice process. Preliminary descriptive results show that there are demographic differences by gender and race/ethnicity in the number of courses and distinct fields that students consider before enrolling.
Students' first formal interaction with STEM typically occurs in any one of a number of introductory courses. Outcomes in these courses can affect retention in STEM. Those outcomes are often disparate across student populations, manifesting as grade inequities among students with different social identities. In this study, we consider student grades in their first introductory STEM course across five large, public, research-intensive US universities over ten years. In every one of these institutions, course outcomes were positively associated with the number of systemic advantages (by ethnicity, gender, low income, and first generation status). This begins to explicate the truly systemic character of inequality in STEM as being commonplace within universities, and present across the higher education landscape.
The courses that students take - and their grades in these courses - critically shape academic pathways. Yet, most instructors are unaware of many characteristics of their students, their academic pathways, and how equitable their grades are. At Michigan, a few groups are working towards making these data available to more instructors and learning how to make them actionable. At this time, these data are provided when kicking off intensive reform efforts (through the Foundational Course Initiative), or ‘pushed’ to instructors with large courses with ‘inequitable’ outcomes (a collaboration between the Assessment Toolkit and Associate Deans of Undergraduate Education). Our immediate next step is to learn more about what instructors think about and do in response to this information, to help us improve the information that we share, how it is shared, and who we share these data with.
This project considers why some students are able to declare majors in a timely manner when attending public, less-selective colleges and universities while others do not. We specifically evaluate the roles of individual clarity and decisiveness, academic resources, social resources, and organizational experiences in shaping students' declaration outcomes.
Using digital trace data to observe course consideration, together with qualitative student interviews, we provide a novel empirical study of course consideration as an important component of course selection. We find that students experience consideration as complex process, and that consideration patterns during students’ first years are predictive of major selection in later terms.
Using natural-language techniques, we forecast student majors with course enrollment data. We find that a student's first course is strongly predictive of their initial major. With the first year of academic data, a student's major can be predicted with more than 50% accuracy.
Performance in introductory STEM courses is consequential for STEM retention. In this project, we have examined the extent and the underlying causes for demographic performance gaps in introductory STEM courses across different STEM fields and institutions. We have further explored solutions for addressing these performance gaps and promoting diversity in STEM retention.
We are conducting a longitudinal study intended to elucidate factors around course and major selection. We began interviewing a cohort of 85 students in the summer of 2019 before students matriculated to an elite university and have been interviewing them every academic quarter. In addition, we incorporate transcript and CARTA-use data into our analyses.
Content repetition is a strategy in which students elect to enroll in college courses which repeat material they covered in schoolwork prior to college entry. Repetitions seem to offer benefits of strengthened skill and confidence for those who pursue them, but may exacerbate problems of classroom climate and instructors’ assumptions about instructional quality and student ability. Furthermore, students who do not employ this strategy often underestimate their ability and lose confidence as learners, with implications for STEM pathway completion.
The AskOski project (http://askoski.berkeley.edu) seeks to improve equity and achievement in higher education by making colleges and universities first-class beneficiaries of human-centered AI research.
We will cover highlights from the 11 papers produced from this ongoing effort, including uncovering relational semantics from enrollments (https://tinyurl.com/UniversityMap) and generating personalized academic plans (AAAI'21).
UC Irvine is serving as a pilot demonstration site to develop and implement a state-of-the-art measurement project to improve our understanding of the value of undergraduate educational experiences and promote evidence-based models of undergraduate student success.
In this project we track a random sample of more than 1,200 UCI undergraduates for two years, collecting academic transcript data, survey data, and engagement data from learning management systems for each of the students.
Using administrative, survey, and clickstream data from UC Irvine, we investigate differences between male and female students in course choices, between and within majors, and in course navigation. We first document how much gender segregation is attributable to major choice, as compared to course choices within majors, and how the curricular structure of programs is related to within-program segregation. Second, using survey data from current undergraduates, we examine how students make decisions about which classes to take. Third, using data from our learning management system, we examine differences between male and female students in course engagement and navigation.
Much recent work in STEM education has focused on ensuring equitable access for underrepresented groups in STEM, but studies of the digital divide suggest that inequalities of access may be magnified in the rapid move to online education. We deploy a mixed methods study of three higher education STEM departments to understand and characterize emergent gaps in learning trajectories stemming from the rapid shift to online course delivery, with focus on implications for underserved groups in STEM and the development of strategies or resources to remedy new cracks in the pipeline.
Millions of U.S. high school students graduate academically underprepared for college and enroll in developmental courses that are generally deemed ineffective and stigmatizing. We study the effects on students' academic pathways of a large-scale intervention to improve college-level mathematics instruction and eliminate developmental mathematics at a large public university serving a diverse population of on-campus and online students.
We seek to understand how inequities persist through an academic pathway to find the highest impact (highly inequitable or exclusionary courses with high enrollment) in order to distribute limited resources aimed to improve courses and thus academic pathways for students. To highlight inequities in higher education courses, we created an interactive dashboard for higher education administrators, faculty, and staff.
I will share two work-in-progress projects – First on how computational analytics can describe the evolving uniqueness of academic pathways; Then on a user-facing platform that provides a novel lens for researchers to investigate the interaction and evolvement of students’ dispositions and choices under academic context.
Using longitudinal data from the Department of Education, this project examines how students' occupational plans and beliefs about how to realize those plans (1) vary by gender, race and ethnicity, social class, and attribute of where students grew up; and (2) affect outcomes such as major selection, attrition, and completion.
I pose three questions about the enacted, espoused, and latent curriculum that I believe can help guide future analyses of students’ pathways through undergraduate general education. These questions relate to classic debates in curricular theory, academic planning, and educational design, and provide a framework for discussion of learning analytics initiatives.