i

Description

Montagnes et forêt se reflétant dans l'eau immobile d'un lac

Source

Yellowstone National Park (National Park Service) · original page

Licence

PD Public domain

Original photo, unretouched. The background is a version of it stylised by the site.

False consensus

What seems obvious to us, we assume seems obvious to everyone.

In brief

• What: we overestimate how many others share our choices, opinions, and preferences.
• Why: our own view is the most accessible data; we spend time with people like us; we interpret the situation as “the” reality.
• Detect: “everyone knows that…,” “no one would want…,” a typical user who looks like the team.
• Avoid: ask instead of assuming, run anonymous polls, test with real users, assume the opposite view exists.

On this page

1. Definition
2. The mechanism
3. Where we encounter it
   1. In project management
   2. In management
4. Famous cases
   1. Ross, Greene, and House: “Eat at Joe's” 1977
   2. “Tappers” and “listeners” (curse of knowledge) 1990
5. Detection signals
6. Avoidance reflexes
7. With AI: detect, fix, mitigate
   1. Detect
   2. Fix
   3. Mitigate
   4. Prompts to integrate into your workflows
   5. How to integrate them
   6. Precautions
8. Limits
9. Sources and verification
10. See also
    1. Combines with, or reinforces
    2. Related biases
    3. Counterweights and related reflexes

Definition#

Fictitious illustration

False consensus (false consensus effect) is the tendency to overestimate the extent to which others share our opinions, beliefs, preferences, and behaviors. It was described in 1977 by Lee Ross, David Greene, and Pamela House (Stanford).

What defines this effect is not simply thinking others are somewhat like us, which can be reasonable, but that the estimate depends on our own choice: two people who made opposite choices each estimate that their choice is the more common one.

The mechanism#

• Availability. Our own views and those of people around us are the easiest to recall; they form the sample we reason from.
• Biased sample. We mostly interact with people from the same profession, background, or department: their agreement feels representative.
• Situation interpretation. Everyone believes they see the situation “as it is”; if it is obvious, others should react like we do. Ross calls this naive realism.
• Motivation. Knowing we are not alone in thinking this way is reassuring and justifies our choices.

Where we encounter it#

In project management#

• Interface design: the team designs “for the user” using its own habits as the norm; what seems intuitive to them is not intuitive for an occasional user (a neighbor of The curse of knowledge).
• Requirements gathering: “everyone wants this feature” often means “the sponsor and I want it.”
• Quick decisions in meetings: “it's obvious, we agree, right?”; silence is read as agreement.
• Team standards and rules: a naming convention or a tool is judged “obviously better” because it suits those who chose it.
• Client relationship: we assume their priorities, constraints, and wording are the same as ours.

In management#

• Motivation: a manager motivated by autonomy assumes everyone likes it, and removes reference points that part of the team wanted.
• Communication: “everyone understood” after an announcement; no one asked questions because each person thinks they are the only one who did not understand (pluralistic ignorance).
• Team climate: a manager convinced that “the team is happy” because they are, or because close collaborators say so.
• Change: the sponsor of a reorganization underestimates the share of people who oppose it because their own circle supports it.

Famous cases#

Detection signals#

• “Everyone knows that…,” “no one would do that,” “people want…” with no data.
• The “typical” user is a coworker, or yourself.
• Silence is read as agreement; people who are silent were not asked.
• We are surprised by the opposite opinion: “I didn't think anyone could think that.”
• Different views are attributed to incompetence or bad faith rather than different preferences.
• Decisions are made by a small group, without consultation, because “everyone agrees.”

Avoidance reflexes#

1. Ask instead of assume. Turn “everyone thinks that…” into a question asked to a varied sample before deciding.
2. Poll anonymously. A quick poll (two or three questions) gives the real proportion; answers do not depend on rank or discomfort.
3. Test with real users. Five people who did not design the product reveal, according to Jakob Nielsen (2000), about 85% of interface usability problems; they do not replace a representative sample, but they correct the team's estimate.
4. Seek the opposite view. Ask: “Who on the team or client side would disagree, and why?” Assign someone whose task is to answer that.
5. Broaden your sample. Talk to people from other jobs, other levels, other sites, and real clients.
6. Distinguish silence from agreement. Round-robin or simultaneous voting, not “does anyone have an objection?”
7. Make your assumptions explicit. Write “I assume users know X”; a written sentence is easier to verify than implicit obviousness.
8. Treat a contrary view as data, not as an anomaly. If it exists, the effect is in play: add it to your estimate.

With AI: detect, fix, mitigate#

Prompts to integrate into your workflows#

1. List “obvious” assumptions. Before a decision or a design.

Here is a decision or design we are considering: [description] for [target audience].
List all implicit assumptions about what users or stakeholders know, want, and will do (“everyone knows that…”). For each one: the simplest way to verify it (question, test, or data) and what would change if it were false.
Rank them from most risky to least risky.

2. Neutral questionnaire in three questions. To measure rather than guess.

We want to know whether [population] prefers [option A] or [option B] for [item]. Write a questionnaire of no more than five questions, without wording that steers the response, with balanced choices, and a final open-ended question.
Then explain how to build the sample so it is not made up only of people who resemble us, and what we will not be able to conclude from a small sample.

3. Review a message from the point of view of someone who does not know. To test a message or documentation.

Here is a message intended for [recipient, knowledge level]: [text]. Read it as someone who has none of the author's context.
Indicate: terms or references they would not understand, skipped steps, what they might misunderstand, and the questions they would have without daring to ask. Propose a clearer version.

How to integrate them#

• Before the decision: prompt 1 makes the “obvious” assumptions visible; the team chooses which ones to verify.
• To measure: prompt 2 helps draft the survey, which you send to real people.
• Before publishing: prompt 3 serves as a first review, followed by a review by a real recipient.

Precautions#

• A “simulated user” is not a user: never present its responses as survey results.
• Do not submit the question to AI while giving it your opinion first: it will return the consensus you are looking for.
• Do not send personal or confidential data to an unauthorized tool.

Limits#

• The effect is robust (review by Marks and Miller, 1987), but its interpretation is debated: using your own opinion as a clue to others' opinions can be rational when you have no other information (Dawes, 1989; Hoch, 1987). The bias starts when the estimate does not get corrected by data.
• Krueger and Clement (1994) nevertheless showed that the effect persists under conditions where simple reasonable inference should reduce it: it cannot be reduced to that.
• It varies by topic: stronger for opinions and preferences than for abilities, and stronger when the opinion matters to oneself.
• Not to be confused with pluralistic ignorance, its opposite: you believe you are alone in thinking something that many others also think, and you stay silent.

Sources and verification#

• Ross, Greene & House (1977), “The “false consensus effect”: An egocentric bias in social perception and attribution processes”, Journal of Experimental Social Psychology, 13(3), 279-301
• Marks & Miller (1987), “Ten years of research on the false-consensus effect: An empirical and theoretical review”, Psychological Bulletin, 102(1), 72-90
• Dawes (1989), “Statistical criteria for establishing a truly false consensus effect”, Journal of Experimental Social Psychology, 25(1), 1-17.
• Krueger & Clement (1994), “The truly false consensus effect: An ineradicable and egocentric bias in social perception”, Journal of Personality and Social Psychology, 67(4), 596-610
• Hoch (1987), “Perceived consensus and predictive accuracy: The pros and cons of projection”, Journal of Personality and Social Psychology, 53(2).
• Nielsen, J. (2000), “Why You Only Need to Test with 5 Users”, Nielsen Norman Group.
• Newton (1990), doctoral dissertation, Stanford University (title cited in several forms depending on sources, including Overconfidence in the communication of intent: Heard and unheard melodies) — the “tappers” experiment.
• Heath & Heath (2007), Made to Stick: Why Some Ideas Survive and Others Die, Random House — account of the “tappers” experiment.

See also#