Withhold live cumulative summaries until independent judgments are complete

Withhold live cumulative summaries until viewers have made their own judgment. For example, do not let person n+1 see the running histogram built from people 1 through n while estimating a proportion or a linear association.

Once early answers are visible, they become the seed that later viewers can follow. In this study, even a small visible seed could shape later judgments.

Mechanism: Sequentially visible summaries create an information cascade. The first visible answers establish a direction that later readers reuse, so error can propagate through the sequence.

Evidence: In the iterated experiment, the shown histogram mean strongly influenced the next person’s answer. Initial conditions affected later judgments, while the number of prior responses n was not significant, so a summary based on a few people influenced later viewers about as much as one based on many (Hullman et al., 2011).

Notes: Low sample counts are not a safe threshold by themselves.

• User Goal: Gather many individual judgments without earlier answers biasing later ones.
• Task: Objective estimation answered sequentially.
• Data: Quantitative values with responses submitted over time.
• Chart Setting: A live social interface updates a visible histogram or similar aggregate as new answers arrive.
• Audience: Later viewers can see earlier viewers’ estimates before submitting their own.
• Success Criterion: Stable accuracy regardless of the first few responses.

Break it when: Responses are collected independently and the aggregate is revealed only after the individual answer is complete. Why: The cascade condition in the study requires person n+1 to see the judgments from the first n people before answering.

Sacrifice: You lose live social feedback during the estimation step. Risk: Delaying the summary can reduce the immediate sense of community activity. Mitigation: Reveal the aggregate after independent submission rather than during the judgment itself.

Mistake: Leaving the running summary visible because only a few earlier users have contributed. Why it fails: The study found that the first visible judgments already set the stage for later answers.

Failure Sign: Later answers depend on how the first few visible responses were seeded. Quick Check: Seed the same chart with two different early cumulative summaries and compare later judgments. Stronger Test: Compare a live sequential condition against a batch-collected independent condition and inspect whether the live condition inherits the initial seed.

• Stop showing the public cumulative summary until the user has submitted an answer.
• Collect a batch of independent responses before publishing any aggregate.
• Remove the live cumulative summary from the estimation view when independent collection is not possible.