Graphs with logarithmic axes distort lay judgments

Graphs that depict numbers of COVID-19 cases often use a linear or logarithmic scale on the y-axis. To examine the effect of scale on how the general public interprets the curves and uses that understanding to infer the urgency of the need for protective actions, we conducted a series of experiments that presented laypeople with the same data plotted on one scale or the other. We found that graphs with a logarithmic, as opposed to a linear, scale resulted in laypeople making less accurate predictions of how fast cases would increase, viewing COVID-19 as less dangerous, and expressing both less support for policy interventions and less intention to take personal actions to combat the disease. Education about the differences between linear and logarithmic graphs reduces but does not eliminate these effects. These results suggest that communications to the general public should mostly use linear graphs. When logarithmic graphs must be used, they should be presented alongside linear graphs of the same data and with guidance on how to interpret the plots. © 2020, Brookings Institution Press. All rights reserved.