ľ¹Ï¸£ÀûÓ°ÊÓ

Introductory statistics courses are essential for many university students, but are often also challenging. Students struggle to understand the abstract concepts involved, such as significance level and p-value. Appropriate feedback could help, but is difficult for teachers to provide when student groups are large. In this thesis, a solution is sought in automated feedback in a computer-based learning environment. The central question is: How can automated intelligent feedback support first-year university students in developing statistical proficiency?

Two feedback types were designed and evaluated within two first-year introductory statistics courses for social sciences students. The first type, inner loop feedback, provided feedback on steps that students made in constructing hypothesis tests. It provided information about errors within single steps the student added and about inconsistencies in the student’s entire solution so far. The second feedback type, outer loop feedback, provided students with an overview of their understanding of the statistical concepts involved. The estimates in this overview were based on the students’ work on homework sets within the computer-based learning environment.

The evaluations showed that students valued the feedback types and used both of them extensively. Inner loop feedback helped students to solve more hypothesis-testing tasks, and to make fewer errors within these tasks. Students needed time to learn how to work with this feedback, though. Outer loop feedback seemed more straightforward to use. It encouraged students to work slightly longer on their homework sets and, for weaker students, this seems to have resulted in slightly higher exam grades.