Comparing zero-imputation, reweighting, and drop the lowest grading strategies: A statistical simulation

Resumen de los autores

Because students often fail to complete assessments, this necessitates a course policy on how the missed assessment will affect a student’s grade. Three commonly-used methods are zero-imputation (i.e., assigning a zero for all missed assessments), drop the lowest (i.e., assigning a zero for missed assignments, but students can exclude a certain number of their lowest grades), and reweighting (i.e., moving the weight of a missed assignment to other assignments). Using statistical simulations, I tested which of these three methods of handling missed assessments produces a median grade closest to the true score. I simulated positively correlated (r = .50) data from 10 quizzes that were normally-distributed (M = 7.5, SD = 1.5), and then introduced 10% missing data (i.e., “sick days”). I also simulated satisficing, wherein students strategically skipped quizzes once an A+ grade was locked in. Next, I applied the three missing assessment strategies (zero-imputation, drop the two lowest, and reweighting two missed assessments) and compared their performance to the true scores without missing data. Median grades were extremely deflated when using the zero-imputation method. In contrast the other two methods were close to the true scores. However, the drop the two lowest method was biased towards students who did not become ill, essentially giving 2 or 3 bonus points to students who do not miss any assessments. In contrast, the reweighting method was comparatively unbiased. Overall, I recommend reweighting assessments as a good, relatively unbiased method for instructors to deal with missed assessments in the classroom.

Áreas y palabras clave

Cómo citar (APA 7)

Mackinnon, S. (2026). Comparing zero-imputation, reweighting, and drop the lowest grading strategies: A statistical simulation. https://doi.org/10.35542/osf.io/8ju4k_v1