Production of engineers by universities in the United States has been a concern amidst public discourse about future American economic strength. Low completion rates in American engineering programs pale in comparison to increases in the engineering production of other countries. Responding to such concerns, leaders in engineering have worked to revise undergraduate engineering curricula and pedagogies. Project-based learning is one new pedagogy that has gained popular attention. In this approach, students work on an extended-duration project to address an open-ended problem, collaborate intensively with each other, and experience close student-faculty interaction. Projects are designed to provide opportunities for students to apply theoretical knowledge in real-world contexts.

Despite enthusiasm surrounding project-based learning, its effectiveness has been evaluated rarely, and most evaluations have not employed rigorous research design and analytic methods. Numerous small-scale studies have found positive effects on student outcomes. However, often these studies were observational in design and did not control for relevant unobserved differences among participants sufficiently well to support causal conclusions. In this case study, I extend past research by working with a large-scale dataset, applying innovative quasi-experimental methods to evaluate the effectiveness of project-based pedagogies implemented at the Massachusetts Institute of Technology (MIT). I incorporated an interrupted time-series analysis within a framework provided by multiple-spell discrete-time survival analysis to support causal inferences about the effects of increasing project-based course offerings on student persistence in engineering. I examined whether students exited the field of engineering and, if they exited, at what point in their academic career they did so. For those who exited, I also investigated whether they returned to engineering and, if they returned, when this occurred.

I found that MIT's decision to increase project-based course offerings increased the probability that students initially interested in engineering would remain in the field, and decreased the probability that interested students who left engineering would return to the field later in their studies. Importantly, I found that the impact of the increase in project-based courses differs for students according to their academic preparation such that as academic preparation increases, the effects noted above decline and, in some cases, reverse direction.