Abstract:

Building on previous work where we investigated the correlation between perceived test complexity and student performance using a subset of data from a single subject, this study expands the scope to provide a more comprehensive analysis. This time, we engage with a more robust dataset that spans an entire academic year, including 1801 records from 47 students across 9 subjects, covering 89 distinct activities.

Our aim remains constant – to explore whether there is an alignment between the professor's and students' perceptions of assignment complexity and how these perceptions influence the students' final performance. Professors were asked to rate the difficulty level of each assignment they administered, and students were requested to provide their perceived complexity upon submitting their solutions. Notably, these perceptions were collected independently, with neither party privy to the other's assessment. Student time on assignment was also measured by looking at the assignment start date, submission date and assignment due date. Two variables were extracted from this to help better understand student perception of assignment difficulty by looking at the time each student spent on a given assignment.

We employed the k-means clustering algorithm to discern patterns in the relationship between the differences in complexity perceptions and student performance. The findings revealed a clear trend: students who perceive an assignment to be more difficult than their professor's rating tend to achieve lower scores. Although exploratory in nature, these findings have significant implications. They hint at the potential of leveraging perceptions of complexity as a tool for early identification of students likely to excel or struggle in a subject. This research reaffirms the importance of considering perception data as a valuable adjunct to performance metrics in higher education.

These conclusions emphasize the importance of monitoring not only students' performance but also their perception of test complexity. This can provide a more holistic view of their learning process and help educators create a more supportive and effective learning environment.

Keywords:

Student performance, student performance prediction, exam complexity, learning management systems, k-means clustering, student procrastination.