Correlation Judgment and Visualization Features: A Comparative Study

Recent visualization research efforts have incorporated experimental techniques and perceptual models from the vision science community. Perceptual laws such as Weber's law, for example, have been used to model the perception of correlation in scatterplots. While this thread of research has progressively refined the modeling of the perception of correlation in scatterplots, it remains unclear as to why such perception can be modeled using relatively simple functions, e.g., linear and log-linear. In this paper, we investigate a longstanding hypothesis that people use visual features in a chart as a proxy for statistical measures like correlation. For a given scatterplot, we extract 49 candidate visual features and evaluate which best align with existing models and participant judgments. The results support the hypothesis that people attend to a small number of visual features when discriminating correlation in scatterplots. We discuss how this result may account for prior conflicting findings, and how visual features provide a baseline for future model-based approaches in visualization evaluation and design.

1 Introduction

In a recent study, Rensink and Baldridge demonstrated that the perception of correlation in scatterplots can be mathematically modeled using Weber's law [1]. In followup experiments, Rensink showed that this law is robust to changes in data characteristics and scatterplot design choices [2]. Based on these findings, Harrison et al. replicated the original study by Rensink and Baldridge, moving beyond scatterplots to measure and compare the effectiveness of a range of visualizations [3]. Their results indicate that the perception of correlation in all of these bivariate visualizations can be modeled using Weber's law. Together, these studies sparked a renewed interest in the information visualization community towards better understanding the underlying mechanics of visualization and modeling approaches, such as Kay and Heer's followup analysis of Harrison et al.'s released experimental data [4].