Just Five More Minutes: The Relationship Between Timed and Untimed Performance on an Introductory Programming Exam

This evidence-based practice paper explores the relationship between the performance on an untimed exam and performance on other course metrics including later timed exams in an introductory computer programming course. Background Introduction to computer programming courses are often viewed as being exceptionally hard for most engineering students not explicitly pursuing careers as professional programmers. The combination of the breadth of material, the complexity of that material, and students’ relative unfamiliarity with the material makes it exceptionally difficult to give a proctored exam during a traditional class period. Students frequently complain that they understood the material but needed additional time to complete exams. Purpose (Hypothesis) The purpose of this research is to explore the relationship between time needed to complete the exam and overall course performance. The hypothesis was that additional exam time is of little relevance – students who know the material do better on the exam than students who don’t know the material as well, regardless of the amount of time available. Design/Method During the Fall 2015 semester, the instructor of an introductory programming course scheduled the first exam on a Saturday with the provision that students had unlimited time. Completion time for the exam was noted for each student. Due to an administrative request, the second and third exam were held during a 50-minute class period. A comparison of time taken on exam 1, all three exam scores and other course performance indicators (i.e., homework and final project scores, final course grades, course attendance) was conducted. Results Analysis revealed two major findings. First, the hypothesis was confirmed; there is no relationship between exam time and course or exam performance. Second, homework, more so than timed or untimed exams, was a better predictor of overall course performance. Conclusion The conclusion this research made was that the value of exams in a programming class should be rethought altogether. Isolated exams are a poor predictor of overall performance and the granting of unlimited time does not result in stronger alignment to course performance.