A clear answer can hide an unanswered question
Imagine a team choosing a launch date. One colleague says Friday should work. Another pauses, mentions an unfinished test and asks for another day. The first answer is easier to act on. It offers a date without asking the room to sit with uncertainty.
But the pause and the steady voice tell us little about whether the test will pass. To decide well, the team needs information about the work. Delivery supplies information about the speaker, and the two can become confused.
Confidence can be a useful clue. Someone who saw a street sign clearly may be more certain than someone who glimpsed it through a moving window. The difficulty is deciding whether two people use certainty in comparable ways, and whether their certainty concerns something they actually know.
Three things that need separate names
Keep three judgments distinct:
- Competence: the ability to perform this task reliably.
- Decision confidence: how likely a person thinks this particular answer is to be correct.
- Communicated confidence: the rating, wording or manner that reaches the group.
The launch meeting can contain all three without measuring them in the same way.
Those signals can line up, but the alignment requires evidence. A person can know a subject well and struggle to explain it under pressure. Another can deliver a polished answer while missing a basic constraint. Neither fluency nor hesitation establishes which situation the group is facing.
There is also a difference between confidence about one answer and a general tendency to feel certain. The research on confidence matching addresses why these personal differences matter when judgments are combined. Knowing that someone is more confident on their stronger answers does not establish that they are better than another person who speaks more cautiously.
How overconfidence can become status
In six studies published in 2012, Cameron Anderson and colleagues investigated whether overconfidence could help people gain standing in groups. Their findings supported a route through perceived competence: people who overrated their abilities could look more capable to others.
One analysis used recordings of small-group work. Observers associated competence with speaking more, an assured vocal tone and a calm demeanor, among other cues. Overconfident participants displayed several of these behaviors. The studies concern particular tasks and groups; they do not establish that confidence always defeats ability.
Nor does overconfidence necessarily mean deliberate deception. A mistaken self-assessment can be sincere. That makes a simple instruction to watch for bluffing inadequate: the speaker may believe the reassuring account as firmly as everyone else does.
The practical question is therefore about the claim. What has been checked? Which part rests on an estimate? A calm delivery remains useful communication, but the group needs a way to examine what it communicates.
People can start using the same confidence scale
Dan Bang and colleagues’ 2017 study found that partners could adapt their expressed confidence toward one another, a pattern called confidence matching. This can help people with similar expertise communicate despite different starting habits. It can also obscure differences in who is more likely to be right.
A shared vocabulary is useful only if the group understands what it measures. Suppose two colleagues learn to use “fairly sure” consistently. That solves a communication problem. It does not give both colleagues the same knowledge of a database migration.
The distinction is easy to miss because convergence feels cooperative. Agreeing on how certain to sound can make a conversation smoother without resolving whose evidence is stronger. Matching is neither automatically harmful nor a substitute for evaluating expertise.
Discussion carries more than certainty
A 2026 study of 210 participants, forming 105 pairs, compared general-knowledge performance alone, with access to a partner’s answer and numerical confidence, and through verbal discussion. Pairs containing at least one person with high trait confidence improved more through discussion than through the numerical exchange. Pairs with two low-confidence members benefited similarly from both formats.
Confidence matching occurred during discussion, but did not predict its accuracy gains. The richer conversation evidently cannot be explained just by partners aligning their certainty. The study also used a fixed order, with passive communication before active discussion, leaving practice or fatigue as possible influences.
This is evidence from a controlled knowledge task, not a recipe for choosing every project team. Its useful correction is narrower: talking can convey reasons and help people revise judgments. Confident participation is not inherently a problem; the information available during that participation matters.
Expertise belongs to a task
In a 2022 perceptual-decision experiment, participants placed bets on their own answers and those of other players. Their judgments reflected both another player’s ability and the difficulty of the decision. They could account for a strong player facing a hard problem, or a weaker player facing an easy one.
That experiment did not test office persuasion. It does show a more informative way to think about trust than a general impression of someone’s confidence.
For the hypothetical launch, relevant questions might concern previous estimates on comparable work, knowledge of the unfinished test and access to its latest result. A good record in a different role may deserve respect without settling this particular deadline. Evidence has to travel with the task.
Give the missing information somewhere to go
The Royal Society’s guidance for group decisions recommends independent input, recognition of expertise and deliberate opportunities for minority information to emerge. Where hierarchy matters, anonymous collection can help preserve that independence.
For the launch meeting, one practical application is to collect each person’s proposed date and main reason before discussion. Then ask whoever owns the testing to explain the unresolved dependency. The team can agree on a condition: Friday remains the plan if the test passes tomorrow; otherwise the date is reconsidered.
This example is an application of those principles, not a tested guarantee of a successful launch. Its benefit is concrete: people can inspect the condition instead of interpreting how comfortable someone looked.
A useful recommendation can contain uncertainty. “I recommend Friday, provided the payment test passes tomorrow” gives the room a proposal, a dependency and a check. It is easier to challenge or update than an unexplained assurance.
Return to the launch date with the test result, its owner and the condition for reconsidering Friday recorded together. That gives a confident proposal something the group can check. For the speaker’s side of the same problem, quiet executive presence focuses on making reasoning clear without performing certainty.
