Misleading Visualizations in Social and Personality Psychology

Abstract

Are researchers in social and personality psychology visualizing their data in an accurate and honest manner? The purpose of the current research was to examine the prevalence and extent to which visualizations in social and personality psychology were misleading. Every study published in the Journal of Personality and Social Psychology, Personality and Social Psychology Bulletin, and the Journal of Experimental Social Psychology in 2017 and 2018 were evaluated for the prevalence of misleading visualizations, that is, whether or not the full range of possible scale values were displayed on the y-axis, and the magnitude that axes were truncated. Results indicated that 62.08% (n = 131) of graphs that displayed a Likert-type variable were misleading. That is, almost two-thirds of visualizations plotted a variable with a truncated axis. Axes were truncated, on average by 48.89% (SD = 20.74%). There were also differences by graph type. Line graphs truncated their axes 18.75% more than bar graphs and were 27.47% more likely to have a truncated axis. The argument is made that guidelines and standards should be developed that explicate the importance of displaying the full range of scale values in visualizations of empirical data.

Daniel J. Chiacchia

PhD Student, Organizational Behaviour and Human Resource Management

My research interests include the social psychology of working in the digital economy (e.g., the automation of work and dehumanization, ‘gig’ work); leadership during threat and uncertainty (e.g., how leaders effectively adapt to change); as well as best practices in quantitative methodology, including open science, replicability, statistical power, data visualization, and effect sizes.