UNDERSTANDING GRAPH READING AND COMPREHENSION THROUGH EYE-TRACKING: EVIDENCE FOR THE DEVELOPMENT OF EXPERT STRATEGIES

The use of graphical information is the central to many aspects of our lives: personal finance, health, global change, education, etc. Given the importance, graph comprehension skills are critical for all informed citizens. Further, geoscience students must also learn to use new kinds of graphical forms (e.g., inverted axes, ternary plots, log plots, normalized data plots). However, relatively little is understood about the details of how individuals perform graph-reading tasks, how these skills develop over time, or how we might teach them more effectively. Ultimately, the goal with this research is to inform the design of instructional activities to promote the development of graph reading and comprehension.

Here we share our results from a collaborative study between the Macalester Geology department and Psychology eye tracking lab on the mechanisms behind graph reading and its implications for scientific literacy. Within our expert and novice pools (selected by education and experience levels) experts answered 85.7% of the multiple-choice questions about data interpretation correctly, whereas novices answered 64.7% correctly.

In addition to illuminating differences in the accuracy of expert/novice graph interpretation skills, this research also provides insights into the graph reading processes of an expert that are presumably developed over years of exposure. We measure this difference through various metrics (e.g. fixation dwell times and order, eye movement distance and velocity, interest-area regressions, interest-area dwell times) and scales (e.g. individuals, expert/novice groups, graphs, interest areas). Our early analysis suggests that the expert/novice continuum is both cognitive and strategic. Experts and novices averaged 100.9 and 381.1 eye movements per question, respectively, which means that novices adjusted their fixation points four times more often than experts. The aggregated expert graph reading approach shows a more structured, strategic pattern employed that is largely absent from the novice group. Before any questions were asked, 50.2% of expert fixations were on the data portion of the graph meaning they were spending a larger fraction of their time interpreting data after checking the title and axes; however, only 34.2% of novice fixations were in the same area.