Is there a gender pay gap among graduates in some science, technology, engineering, and math (STEM) fields? Women and men have near-identical human capital at college exit, but cultural beliefs about men as more fit for STEM professions than women may lead to self-beliefs that affect pay. We hypothesized that women and men would be paid differently upon college exit, and that gender gaps in self-beliefs about one’s abilities, or self-efficacy, would correspond to this difference. Using data from a three-wave longitudinal study of graduates of engineering programs from 2015–2017, we find a confidence gap that aligns with a gender pay gap. Overall, these findings point to the role that cultural beliefs play in creating gender disparities among STEM degree-holders.

As low-status academic offerings purveyed by high-status institutions, massively open online courses (MOOCs) are ambiguous credentials. In interviews with 60 people who devoted substantial time to at least one MOOC between 2014-2017, we find that people use MOOCs to build skills for application at work and home, build relationships, navigate life transitions, and enhance formal presentations of self.

College and university students submit millions of course reviews each year, yet these instruments are only rarely leveraged for scientific inquiry. This paper examines 11,255 reviews submitted to computer science courses to illustrate how such inquiry might proceed.

The US federal government has serially called upon colleges and universities to assist the nation in moments of national crisis. This brief outlines an ambitious plan to enlist the postsecondary sector in helping millions of Americans get back to work in the wake of the pandemic and equip them for ongoing prosperity in a highly dynamic economy.

Using digital trace data to observe this process at scale at a private research university, together with qualitative student interviews, we provide a novel empirical study of course consideration as part of the course selection process.

Our study suggests new design goals for learning analytics systems, the importance of real-time analytics to many instructors, and the value of flexibility in data selection and aggregation for an instructor when working with an analytics system.

We show that a Reinforcement Learning (RL) model produces better learning gains using fewer educational activities than a linear assignment condition, and produces similar learning gains to a self-directed condition using fewer educational activities and with lower dropout rates.

We use a variety of text classification algorithms on a large corpus of college admissions essays to model the extent and ways that AI might inform human evaluation.