Being Well-Prepared for Regular Pair-Programming Helps At-Risk Students

This paper studies 80 final-year undergraduate computing students in an advanced programming course, and aims to relate their reported self-study time for the regular pair-programming labs to their examination performance. We expect that more regular self-studying time would lead to a better examination performance. It is interesting that this is true only for the 27 at-risk students, which are defined as those students failing in the course's mid-term test. We use the k-means algorithm to divide these students into 2 groups according to their self-study time for lab and the difference between their exam and test scores. For the other 53 non-at-risk students, the same analysis using the k-means algorithm shows that more self-study time for lab does not lead to better examination performance, while it is the self-study time right before the exam which affects the performance. It casts doubt on the commonly given unqualified advice for students to study routinely over cramming right before exam. The result leads us to look for other factors that determine student performance.