Search

The digitalization of the educational domain has led to the availability of more data on students. Instead of only focussing on student performance, researchers and teachers nowadays are able to analyse student behaviour during educational tasks. Many studies explore this behaviour with the concept of item response time. Research on student navigation behaviour is lacking. This research extends the exploration of educational data and addresses specifically the analysis of navigational test behaviour. With the help of the ACET 2018 high-stakes test dataset and the literature study, this research proposed measures for navigational test activities, including steps, jumps, skips, changes and hops. Additionally, we made a distinction between the navigation activity before and after reaching the last item. We used k-means clustering on a total of 21.565 students and their tests and identified three consistent test-taking navigation strategies. It was found that the majority of students mostly follow the linear order of the tests, the second group used more activities, and the last group was more active after the last item. No differences in the mean of student ability were found using the clusters. Our results suggested there is little relation between the navigational activities and student ability, and difficult items were skipped and changed more often. Predetermined ‘easy’ items with a higher number of non-linear activities could serve as an indicator for hard-to-understand questions. As we discovered that most answer changes were advantageous, this research advocates for the use of free navigation in tests, giving the students an extra chance to show their full potential.

The digitalization of the educational domain has led to the availability of more data on students. Instead of only focussing on student performance, researchers and teachers nowadays are able to analyse student behaviour during educational tasks. Many studies explore this behaviour with the concept of item response time. Research on student navigation behaviour is lacking. This research extends the exploration of educational data and addresses specifically the analysis of navigational test behaviour. With the help of the ACET 2018 high-stakes test dataset and the literature study, this research proposed measures for navigational test activities, including steps, jumps, skips, changes and hops. Additionally, we made a distinction between the navigation activity before and after reaching the last item. We used k-means clustering on a total of 21.565 students and their tests and identified three consistent test-taking navigation strategies. It was found that the majority of students mostly follow the linear order of the tests, the second group used more activities, and the last group was more active after the last item. No differences in the mean of student ability were found using the clusters. Our results suggested there is little relation between the navigational activities and student ability, and difficult items were skipped and changed more often. Predetermined ‘easy’ items with a higher number of non-linear activities could serve as an indicator for hard-to-understand questions. As we discovered that most answer changes were advantageous, this research advocates for the use of free navigation in tests, giving the students an extra chance to show their full potential.