The Impact of Assessment Policy on Learning: Replacement Exams or Grade Dropping

Abstract

Instructors often debate the merits of alternate grading policies such as dropping the lowest exam or offering an additional exam to replace the lowest score. To date, there has been little research conducted on the impact of these policies on performance. In this study, the author investigates student performance in intermediate macroeconomics over three semesters at a small Eastern college. In each semester, a different assessment policy was applied: Count all three in-term exams; drop the lowest of three exams; offer an additional exam to replace the lowest in-term exam score. Contrary to previous research and conventional wisdom, the author finds that allowing students to drop their lowest grade improved performance on a cumulative final exam, while offering a replacement test had no significant effects.

Keywords:

• grade dropping
• grading
• intermediate macroeconomics
• replacement exams

JEL code:

• A2

Acknowledgments

The author especially thanks Francis Bush and Lester Hadsell for their valuable advice and insights.

Notes

1. Approaches vary as to whether the replacement score replaces the previous low only if it exceeds the previous low, or it may replace the previous low in any situation. In the current study, the replacement score only counts if it improves upon the previous low.

2. There are other studies, such as MacDermott (2009) and Hadsell and MacDermott (2010), which investigate the effects of grade dropping. However, these focus on the effects on student behavior rather than on the impact on learning.

3. It is important to acknowledge that there are differences between the current sample and that used in Sewell (2004). This is discussed at length later in the article.

4. The multiple choice format was chosen to balance competing incentives: interest in providing students an opportunity to take another exam while trying to limit the grading burden and provide a fast turnaround for students. It is certainly valid to ask whether a multiple choice exam can aid a student in preparation for a final exam of essays and problems. Arguing the merits of multiple choice exam vs. short-answer exam is beyond the scope of this article. On the other hand, if the replacement test was meant to provide feedback going in to the exam, the format is of little difference.

5. The two students who withdrew from the class before the third exam are not included in this discussion.

6. The largest gain was 54 points (25 to a 79), while the smallest gain was a single point. On the flip side, the greatest shortcoming was 49 points (previous low was an 86 while the score on the replacement was a 37).

7. Further, there was a lack of uniformity in this data. Student scores, where available, were reported for the two-part and three-part SAT as well as the ACT.

8. These 14 nonmajors have been removed from the dataset due to insufficient data.

9. The complete survey is available upon request.

10. Recall, not every student availed themselves of the replacement test. Substituting taking the replacement test for being offered the replacement test does not change the results appreciably. That is, it has no effect on sign and significance and only a minor impact on magnitude of the coefficients.

11. Aggregating dummy variables such as Sophomore and Freshman into Underclass, or Asian, Black, Hispanic, and Other into Minority do not change the results appreciably. Further, the use of Major GPA rather than Cumulative GPA fails to alter the results in any noteworthy way.

12. Recall, 55 of 63 students took the replacement exam.

13. As stated earlier, seven students took zeros on the third exam. It may be interesting to include a dummy for students who strategically underperformed on an exam. However, the cutoff would be subjective (below 50 percent?). On the other hand, taking a zero is clear evidence of strategic behavior.