Purchase this article with an account.
Elsie Lee, Jason Rubinstein, Eileen Kowler; Eye movement strategies during search of graphs with relevant and irrelevant information. Journal of Vision 2017;17(10):537. doi: https://doi.org/10.1167/17.10.537.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Eye movements provide insights about strategies used to interpret graphical information (Carpenter & Shah, 1998). Subjects searched bar or line graphs to answer simple two-alternative questions. Each graph contained four data values; two were relevant to the question. Graphs were inspected for a mean of 15.5 s (SD 5.6). The first half was spent reading the question. Initial fixations on the graph focused on the legend and the X-axis in the attempt to locate relevant data. Legend and X-axis were re-fixated about 5 times each, even though one glance should have been sufficient. Fixations on data values became prominent during the third quarter of the trial, with preferences to fixate relevant data emerging during the final quarter. Fixations on the Y-axis occurred more often when the question required a decision about the value of the data, rather than ordinal comparisons. During the final seconds of the trial, fixations focused on the two alternatives of the question and either the legend or X-axis, depending on which corresponded to the alternatives. These results show that eye movement strategies evolve over time, reflecting a rational strategy to locate and then evaluate information. Early fixations focused on identifying the referents, and later fixations moved to evaluating the data, with the discrimination between relevant and irrelevant data emerging late. Frequent refixations may reflect attempts to conserve memory, as in visual tasks (Hayhoe & Ballard, 2005; Epelboim & Suppes, 2001). However, given the complexity of the cognitive decisions required to interpret graphs, the frequent re-fixations of critical locations, as well as the orderly progression of fixation preferences from referents to data, may support (and reveal) the gradual build-up of evidence that is used to generate the final interpretation of the graph.
Meeting abstract presented at VSS 2017
This PDF is available to Subscribers Only