Shifting Expectations for Encoding Rules Mitigates Misinterpretation of Connected Scatterplots

Connected scatterplots visualize time-series data by connecting the points on a scatterplot based on temporal sequence. Viewers are prone to misinterpret the direction of time in these visualizations, possibly because they encode time with an unexpected rule - along the connected line (TIME IS A LINE) instead of from left to right (RIGHT IS LATER) as conventional in line charts. In this paper, we use the connected scatterplot to illustrate a perspective on visualization comprehension centered around expectations of encoding rules. People have initial expectations of encoding rules for visualizations that can stem from conventional practices or metaphors, and these expectations have been recognized as a potential factor influencing visualization comprehension. We present three preregistered experiments (n = 1429 in total) demonstrating two kinds of design interventions to strengthen the correct expectation for time and testing their effectiveness in reducing errors for understanding realistic connected scatterplots. We found that visual treatments that suppress the incorrect chart-type expectation and directional cues that emphasize the correct expectation both led viewers to expect TIME IS A LINE more. An explicit directional cue (arrows), ideally redundantly encoded with another cue (trace-line effect or animation), was most effective for reducing misinterpretations. Our findings not only provide practical guidelines for designing connected scatterplots but also contribute theoretical insights to inform the design of novel visualizations that challenge interpretability by defying expectations.