How the majority effect works

In 1951, psychologist Solomon Asch seated a student in a room with 7 other participants. Everyone was shown a card with one line and a card with 3 lines of different lengths, then asked to say aloud which of the 3 matched the first. The task was so simple that people solved it almost without errors when alone.

The other 7 were confederates. They answered correctly in the first rounds, then began unanimously naming an obviously wrong line. The real participant answered second to last and heard several identical, confident voices before her that contradicted her own eyes.

In about 37% of cases, the person repeated someone else's mistake. When answers were written on paper without hearing other voices, errors were under 1%. Their eyes saw correctly - only their words differed. About three-quarters of participants gave in at least once.

There are 2 reasons, and they work differently. The first is informational: if 7 people see one thing and you see another, the most economical hypothesis is that you're wrong. To the brain, the majority isn't an opinion but a cheap source of data about how the world works. By the same logic, you choose the longer queue and the cafe where people are sitting.

The second is normative: you don't want to be the one arguing with everyone. People usually recognize these concessions - they see the right line but name someone else's so they won't stand out. It's hard to separate the 2 mechanisms in an experiment, and the debate over whether perception itself changes or only what is said aloud is not fully settled.

Asch varied the group size. One confederate across from you means almost nothing, 2 already have a noticeable effect, and 3 produce almost the full effect. After that, the line hits a ceiling: 15 people don't exert more pressure than 3.

And there is one crack that breaks it all apart. If even 1 person in the group named the correct line, the pressure dropped about fourfold - even when the others kept getting it wrong. An ally doesn't need to be an authority or even to be right. It's enough for the agreement to stop being unanimous.

The effect has been replicated many times in different countries. It holds up, but not as a constant: conformity is lower where independence is valued, and lower in later decades than in the 1950s. The magnitude varies, not the direction.