Using eye tracking to differentiate student difficulties reasoning with data

Findings from a past project studying students in algebra-based physics courses indicate significant issues working with graphed and pictured data. Students overuse reasoning schemes that work for one cause - one effect data but do not work for multiple causes. Students make incorrect claims like, "If variable A is unchanged and variable B does change than variable A must not affect variable B." In this study, we use eye tracking data to investigate students' attention to the variables in the graphs and pictures. We observe different student gaze patterns on questions answered incorrectly vs. correctly for single-trend questions. This demonstrates that control-of-variables (and not just logical reasoning) affects student skills interpreting this data. In addition, we present other results. For example, the pictorial data requires significantly less attention time for students to analyze than does the single-trend graphed data even though these present the same data sets.