Normative influence exercises as practitioners see it

You cannot practice courage in a meeting. You can practice the small mechanical moves that make courage unnecessary, and most take under a minute.

What to take away

The mechanism the drills work against

Normative influence is the pull toward matching what a group does. It differs from informational influence, where you treat the group as evidence about the world and update because you think it knows something you do not.

The two produce the same behavior and need different responses. Informational influence is corrected by better evidence. Normative influence is corrected by moving the decision somewhere the group cannot see it.

Solomon Asch's line-judgment studies, published in 1951, are the standard demonstration. Confederates gave an obviously wrong answer about which of three lines matched a reference line, and a share of real participants went along.

The number who conformed varied by trial and by subject, so treat any single percentage as a property of that setup, not a constant.

The useful finding is the shape: the wrong answer was social, not perceptual, and many participants who conformed said afterward that they had seen the lines correctly.

The drills

1. The one-line prior

Before any meeting where a decision is expected, write one sentence: what you currently think, and why. Twenty seconds. Do not share it.

The value is entirely retrospective. Two weeks later you can see whether your view moved because of something said in the room or because of the room, and memory will not give you that comparison.

2. The outside phone call

Take a live decision to one person who is not in the group. Describe the situation and the options. Do not tell them what the group thinks.

You are testing whether your reasons survive translation. That is the practical end of the questions in social epistemology: what you are entitled to believe on the strength of the people around you.

Reasons that depend on the group tend to come out as "it is just what we do." You will hear yourself say it, which is more convincing than reading about it here.

3. The written round

If you run a meeting, collect positions in writing before discussion opens. Every attendee gives a sentence, at once, before hearing anyone else.

This is the only drill that changes the group's output rather than your record of it. A room where the first spoken view sets the norm produces one kind of decision, and a room where nine views exist before anyone speaks produces another. Run it three times before judging it; the first round will feel stilted.

4. The refusal budget

Decide in advance how many small norm-following moves you will spend this month without argument, and which one you will not. The point is not to refuse more. It is to stop treating every norm as equally worth complying with, so the one that matters is not spent in advance on the round of drinks and the late finish.

5. Ask for the number

Whenever you are told what most people do, ask how many and how they know. Once a week is enough. You are not conducting an interrogation, and the answer is not usually the interesting part: the shift in how confidently the claim is repeated afterward is.

6. The delayed agreement

When you notice yourself about to agree quickly with a group, say instead that you want to think about it until tomorrow. Then actually think about it, alone, and record whether the answer changed.

Most of the time it will not, and that is a useful result: it tells you the group was not deciding for you on that occasion. When it does change, you have found a live case, and one live case teaches more than a month of reading.

7. The exception rehearsal

Say out loud, alone, the sentence you would use to decline. "I am not going to do that." "I want to think about it." "I do not agree, and I am fine being outvoted."

This feels absurd. It also works, for the plain reason that the cost of dissent is mostly the cost of composing the sentence under pressure, and a sentence you have said before does not have to be composed.

8. The norm audit

Once a month, list the unwritten rules your team actually follows: when people arrive, when they leave, who speaks first, what nobody questions. Mark each one as load-bearing or inherited. You are not proposing to break the inherited ones. You are separating the rules that hold the work up from the ones nobody chose.

A four-week run

Four weeks is long enough to collect cases and short enough to finish. At the end you should have a small pile of written priors, which is the only real evidence you will ever have about how much the groups you are in move you.

What the drills cannot do

They cannot make you comfortable being the only one. That discomfort is a real feature of being a social animal, and the sensible response is to reduce how often it is required rather than to try to feel nothing.

They cannot tell you the group is wrong. A drill that shows the group moved you does not show it moved you in a bad direction, and the further question needs evidence rather than introspection.

They also cannot be run on other people. Somebody who has decided to press you will not be slowed by your having written a sentence beforehand, and that situation calls for the field guide under manipulation resistance rather than for a drill.

What they do give you is a record. Whether discomfort at being outnumbered should count as evidence at all remains contested, and the Stanford Encyclopedia entry on epistemic self-doubt is a fair guide to why reasonable people disagree.

The mechanism these drills work against is set out under normative influence. The numbers-based cousin is under social proof, and the framework behind all of it is under social influence basics.

Common questions

Which drill should I start with if I only do one?

The one-line prior. It costs twenty seconds, it needs nobody's cooperation, and every other drill on this page is more informative once you have a few weeks of priors to compare against.

Does the written round work in a group that already has a strong leader?

It works better there, and only if the leader writes last or not at all. A written round in which the senior person's sentence is read out first has reproduced the problem in a new format.

How do I run these without seeming difficult?

Seven of the eight are invisible to everyone else. The one that is not, asking for the number, is a plain question and lands badly only in groups where plain questions land badly, which is itself the finding.

How long before any of this changes anything?

The record starts immediately. Whether your behavior in rooms changes is a slower and less certain matter, and anyone promising you a timeline for it is guessing.