Avoid random scatter when comparing multiple icon arrays

Use a sequential block arrangement when readers must compare two or more icon arrays at a glance. For example, line up the highlighted figures in each array instead of scattering them randomly when showing before-and-after risks or two risks that differ by about 10 percentage points.

Comparison works better when each proportion appears as one contiguous mass. Random scatter obscures moderate differences, so viewers can reverse the ordering of two values that are actually distinct.

Mechanism: Sequential blocks let viewers compare adjacent areas directly, while random layouts add visual noise and force mental aggregation before the comparison can happen.

Evidence: More than one quarter of respondents confused random icon arrays showing 29% versus 40%, while far fewer made that error with sequential arrays; random arrangements also produced higher first-glance estimates than sequential ones for most tested proportions (Ancker et al., 2011).

Notes: The paper states that small to moderate differences might become detectable after a longer examination period.

• User Goal: Help viewers see which of two or more proportions is larger.
• Task: Rank values or judge a moderate difference quickly.
• Data: Multiple part-whole proportions shown in separate icon arrays.
• Chart Setting: Arrays are compared in one session, likely at first glance, without relying on counting.
• Audience: Consumers or patients reading risk graphics.
• Success Criterion: Readers correctly order the arrays and notice moderate differences immediately.

Break it when: Viewers can inspect the graphic for longer and immediate discriminability is not required. Why: The paper notes that small to moderate differences may become detectable after a longer examination period.

Sacrifice: The arrays lose the scattered appearance of random placement. Risk: Sequential layout improves comparison but does not make the values exact; large blocked proportions were still overestimated. Mitigation: Do not rely on first-glance comparison alone when the exact percentage difference is important.

Mistake: Use random scatter in every array and assume the difference will still be obvious because the counts differ. Why it fails: Moderate differences can be hard to see and may even be ranked in the wrong order.

Failure Sign: Viewers reverse the order of two arrays with a known moderate difference. Quick Check: Show the pair for about 10 seconds and ask which one is larger. Stronger Test: Compare a random version and a sequential-block version of the same pair, and keep the version with fewer rank reversals.

• Replace the random scatter in each array with a contiguous block of the same size.
• Keep the grid size and highlighted count constant across the compared arrays while changing only the arrangement.
• Re-test whether viewers can rank the arrays correctly at first glance.