PhD defence G. (Gabrielle) Martins Van Jaarsveld

Below is a brief summary of the dissertation:

Every January, millions of people make New Year’s resolutions, promising to exercise more, save money, or read more books, and by February most of those plans have been abandoned. University students are no different. Each course or academic year they set study goals but often get lost somewhere between setting those goals and actually achieving them. This dissertation asks how universities can help students not only set better goals, but actually follow through and achieve them, as well as how that help can reach every student who needs it.

Across four studies, this research tested the use of a chatbot that guided students through setting weekly study goals, checked in on their progress, and helped them to reflect on how things went. The results showed that guidance and feedback during goal setting helped students write better goals, but did not make them more likely to achieve them. However, a mid-week prompt for students to monitor their progress made it more likely students achieved their goals.

A final study used a large language model to “read” the students’ conversations and score the quality of their planning, monitoring, and reflection, showing how AI can help track the development of students’ self-regulation in near-real time, at a scale impossible for human raters.

Together, these findings show that effective goal setting support is not just a single intervention, but a series of smaller check-ins, and that chatbots can help scale that to as many students as necessary, affordably, and reliably.