Many educators will have observed a familiar trend every term: a classroom which was full in the first week becomes noticeably emptier by the last, and initial enthusiasm cools as the year progresses. We seek to quantify and analyse the process of student disengagement, identifying those demographics who are most at risk.

Though intuitively familiar, the concept of student engagement has been given a great many definitions, ranging from the emotional and organisational to the cognitive and behavioural. For the present purposes, we analyse the behavioural aspect, of participation in learning activities, motivated by the constructivist view that students’ construction of their own knowledge is at the heart of education.

To investigate this, we monitor the responses to classroom polls in a series of 6 workshops of a large first year undergraduate finance course with over 1000 students. We ask students at the start of the workshop whether, and to what extent, they have prepared for the workshop, and at the end of the workshop, about their own perceived increase in learning that day. We also use in-class quizzes as a formative assessment tool, asking all students one quantitative and one qualitative question each workshop. We link this to a pre-course demographic questionnaire.

We find:
1. It matters whether students correctly answer in-class quizzes, since some students who get the answer wrong cease to answer any more in-class quizzes for the rest of the semester. Using a tobit regression, we find the effect to be statistically significant for four quizzes during the workshops. The effects persist even when controlling for student age, gender, geographical origin, language status, and whether they have been the first in their family to go to university. We hypothesize that some students experience disappointment when questions are too difficult, and lose motivation to engage in quizzes, even though all responses are anonymous.
2. Male students, students from Eastern Europe and from Asia, and students who speak English as a second language, are significantly more likely to reduce their participation in quizzes in the second half of the course. We find that student age, and being first in the family to attend university, have no effect.
3. However, female students, students from the Middle East and Asia, and students with English as a second language, are significantly less likely to decrease the extent of their preparation, between the first and last three workshops, showing greater resilience.
4. There is no significant linkage at any point between students’ perception of the extent of their own learning during one workshop, and the extent of their preparation for the subsequent workshop.

Measurement of student engagement in universities has been previously measured either by general surveys of the campus and the institution as a whole, for instance, the US National Survey of Student Engagement and the Australian First Year Experience Questionnaire, or by surveys representing a snapshot during single lecture, and close observation of a limited subset of students, e.g., the ‘Behavioural Engagement Related to Instruction’ survey. However, we have not been able to find any prior studies investigating the decline in students’ engagement over a semester on a single course. We add to the literature by tracking different forms of disengagement over the life of the course, and the demographic groups who are more or less liable to each.

Keywords: Student Disengagement, Preparation, Perception of learning, Formative assessment.