Abstract:
Scheduling assessment activities across multiple time slots is an increasingly common institutional
response to student diversity, particularly the growing proportion of working students in higher
education. However, when distinct question sets are used across slots to prevent information leakage,
differential difficulty may emerge, inadvertently penalizing students assigned to a harder version. This
paper presents a practical case study from the course Marketing Management II (second year of the
Bachelor in Business Administration and Management, Technical University of Cartagena, Spain),
where an online formative quiz was administered in two non-overlapping time slots with different
question sets. Building on Spanos’s (2019) misspecification testing philosophy, a two-sample t-test
was applied after systematically validating its underlying assumptions through runs tests, t-plots,
Levene’s test for equality of variances, and the Shapiro–Wilk normality test. Results confirmed a
statistically significant mean difference of 2.29 points favouring the morning group (T = 5.24, df =
124, p < .001). As a corrective measure, 1.2 points were added to the scores of the afternoon group,
substantially reducing the performance gap. The case illustrates both the feasibility and the importance
of applying rigorous inferential tools to detect and correct unfair grading differentials in educational
assessment. |
