Use gender-matched restroom person icons instead of head-outline icons for calibrated risk judgments

Use gender-matched restroom person icons instead of head-outline person icons in icon arrays when the goal is to make felt likelihood track the displayed probability. For example, keep the same icon-array counts and colors but swap a head-and-shoulders outline set for male or female restroom icons matched to the reader.

Among more numerate and more graphically literate readers, icon shape changed how tightly perceived risk followed the displayed percentage. Restroom person icons produced stronger perceived-to-actual risk alignment than the weakest-performing person-like alternative.

Mechanism: For readers who can work effectively with quantitative and graphical information, restroom person icons appear to support a reading strategy that keeps subjective judgment closer to the displayed value.

Evidence: In the higher graphical literacy and higher numeracy subgroups, icon type significantly changed the correlation between perceived risk and displayed risk. Restroom person icons showed notably higher correlations than head-outline icons, which had the weakest correlations in those higher-skill groups (Zikmund-Fisher et al., 2014).

Notes: The paper also reports strong correlations for photographs and, in some higher-skill analyses, ovals, but restroom person icons are the clearest directly actionable option emphasized by the authors.

• User Goal: Make readers’ felt likelihood move in step with the displayed probability.
• Task: Present one probability and support well-calibrated subjective judgment about that value.
• Data: A single risk estimate shown as event versus non-event counts in an icon array.
• Chart Setting: A static icon array where only the icon type changes and the displayed numbers stay constant.
• Audience: Readers with higher numeracy or higher graphical literacy.
• Success Criterion: A stronger correlation between perceived risk ratings and the displayed percentage.

Break it when: Your audience is lower in numeracy or graphical literacy. Why: In those lower-skill groups, the paper found no overall icon-type effect on the correlation between perceived and displayed risk, and restroom person icons offered no clear advantage.

Sacrifice: The benefit is audience-dependent rather than universal. Risk: If you apply this icon swap to a mixed or lower-skill audience without checking subgroup performance, you may see no improvement in calibration. Mitigation: Evaluate perceived-to-actual risk alignment separately for higher- and lower-skill readers.

Mistake: Using head-outline person icons and assuming any anthropomorphic icon will calibrate subjective risk equally well. Why it fails: Head-outline icons produced the weakest perceived-to-actual risk correlations in the higher-skill subgroups.

Failure Sign: Perceived-risk ratings change only weakly as the displayed percentages increase. Quick Check: A/B test the same icon array with restroom person icons versus head-outline icons in a higher-skill segment and compare the correlation between perceived and displayed risk. Stronger Test: Repeat the analysis separately for higher- and lower-numeracy or graphical-literacy groups to confirm that the advantage is concentrated in the higher-skill group.

• Replace head-outline person icons with gender-matched restroom person icons while keeping the same numerator, denominator, and color coding.
• Re-measure how closely perceived risk tracks the displayed value in the higher-skill audience segment.
• If most readers are lower-skill, do not assume this icon change will improve calibration without subgroup testing.