Student Achievement in Undergraduate Statistics: The Value of Encouraging Failure

Abstract

This article details what resulted when I re-designed my undergraduate statistics course to allow failure as a learning strategy and focused on achievement rather than performance. A variety of within and between class t-tests are utilized to determine the impact of unlimited test and quiz opportunities on student learning on both quizzes and subsequent assignments in two undergraduate statistics courses (one a 15-week hybrid and one a 6-week online course). The results show that the previous DFW rate was reduced, and no significant difference in outcomes was found between the two different course modalities. Furthermore, students achieved significantly higher on their last quiz and test attempts on every instrument in both semesters, with large effect sizes. Corresponding assignments showed students achieved significantly higher than the mean first attempt, but significantly lower than final mean quiz attempt scores, showing that some knowledge was not carried over to application. The article concludes by evaluating the results of unlimited testing with minimum standards and the limitations of the study and the pedagogical model.