Does using a diversity of graph types help your audience remember your data?

Abstract

Visual memory is typically better when the memorized objects are different in their categories and features, presumably because their representations have less overlap and interference (Konkle et al., 2010). If true, this effect might have implications for improving memory in real world situations, such as trying to remember multiple patterns of data shown across the sequence of a research presentation. We tested whether participant memory was better for data contained across three sets of data visualizations of different types (a bar graph, pie graph, and a bubble graph), as opposed to when they were of the same type (three bar graphs, three pie graphs, or three bubble graphs). Participants (N=48) saw sets of three graphs of different types, and three of the same type. After each triplet they reproduced the four plotted values of each of the three graphs by drawing on a blank 'template' containing each graph's title, axes, and data labels. The reproduction error showed a non-significant trend toward a small memory advantage (10% error became 8% error) for the different set graphs, suggesting that, in this initial study, showing data across different visualization formats results in a tiny memory advantage at best.