Fig. 7

The Plots depict the interaction between Gradient Steps (continuous versus dichotomous) and the true polarization of U.S. states in the 2020 presidential election, in predicting participants’ estimates of how polarized the states were (on the left plot) and participants’ perceived voting influence for each state/trial (on the right plot). Each State is represented twice, once as a dark dot for the average dichotomous voting margin and once as a light dot for the average continuous voting margin. Overall, the left plot demonstrates that dichotomous maps exacerbated the polarization of participants’ estimates except for a few extremely polarized states for which dichotomous colors led participants to underestimate polarization. The plot on the right demonstrates that perceived voting influence (measured on a 5-point Likert Scale) increased when continuous gradient steps were used as opposed to dichotomous gradient steps, but the effect is particularly strong in the states  that had less (actual) voting polarization during the 2020 presidential election