The matching line is the long one. Set the card in front of a person who is on their own and they will say so. Seat them last in a row of students who have all just named a shorter line, and the same card becomes a public problem. The fact did not change. The order of speaking did.
Solomon E. Asch, then professor of psychology at Swarthmore College, described the series in Opinions and Social Pressure, Scientific American, November 1955. A group of seven to nine young men, all college students, was told it was an experiment in visual judgement. Two cards. One held a standard line. The other held three lines, one of them the same length as the standard, the other two off by somewhere between three quarters of an inch and an inch and three quarters. People answered out loud, in the order they were sitting. All but one had been told to agree on the wrong line on certain trials. The one real subject sat near the end and answered in public, after a unanimous majority had already spoken.
What a clear card did not protect
There were 18 trials, and on 12 of them the majority was wrong. Asch ran this with 123 students from three colleges, not including his own. People matching the lines alone make mistakes less than 1 per cent of the time. Under the unanimous wrong majority, the real subjects accepted the majority's error on 36.8 per cent of those selections. About a quarter never went along. Seventy-five per cent agreed with the majority in varying degrees. The independent people tended to stay independent. The people who yielded tended to keep yielding. It was not a one-off stumble that the next card repaired.
The interviews afterwards are the part a buyer should not skip. Among those who yielded, some concluded that they were wrong and the room was right. Others went along, as Asch records it, so as not to spoil the results. Plenty suspected the majority were simply following the first speaker, or that the card was some kind of trick, and those suspicions did not free them at the moment they had to speak. All of the yielding subjects underestimated how often they had conformed. A survey after the meeting will undercount the times the room won. The live call is the record.
Three people were already enough
Asch then varied the size of the opposition, from a single person up to much larger majorities. A single opponent swayed the subject very little. With two opponents, wrong answers rose to 13.6 per cent. With three, they jumped to 31.8 per cent. Past that, a larger majority added little. You do not need an auditorium. Three people who have already agreed, in a meeting of four, are almost the whole effect.
He also made the error blatant, hoping to find a gap so obvious that nobody would follow it. He did not find that gap. Even when the difference between the lines was seven inches, some people still gave the majority's answer. "The fact is obvious" is not a design. Obvious is the condition he already ran.
What one other voice changed
Unanimity was the load, and it was fragile. When the real subject had one partner who answered correctly throughout, either another unwitting person or someone instructed to tell the truth, wrong answers fell to a quarter of the rate under a unanimous majority. People felt warmly toward that partner and credited him with confidence, and they denied that he was the reason they had stayed independent. The denial is not the result. The drop is the result.
The partner did not have to be right in the same way. A dissenter who disagreed with the majority and also with the subject, picking a moderately wrong line, cut the majority's effect by about a third, and the extreme yielding disappeared. A dissenter who always picked the most wrong line dropped the subjects' errors to 9 per cent, and the errors that remained were the moderate ones. Dissent itself, not a perfectly correct ally, did a large part of the work. Asch is plain that the direction of the dissent still mattered. The practical point is earlier. One public disagreement, before the subject spoke, changed the errors.
Taking the partner away did not work in only one way. In one version the partner answered correctly for the first six trials, and 18 of 27 subjects were completely independent while he did. Then he joined the majority. Errors rose at once, back to about the level of a person who had been alone the whole time. Having stood with an ally did not keep them independent once the ally switched sides. When the partner simply left the room, the errors rose less. Desertion was worse than absence.
What a safety day has to put in the room
A psychological safety session that talks about speaking up, and then lets the sponsor state the answer before anyone else has spoken, has built the seat at the end of Asch's row. The card can be a launch, a hiring call or a customer's complaint. If three people have already named the short line, the person with the long line is in the 31.8 per cent condition before the facilitator has asked for courage. Courage was not what moved the number. A second voice, spoken first, was.
The standard I would take into a proposal is specific. On at least one live disagreement, someone other than the sponsor states what they see before the rest of the room speaks. The group then hears what changes when that person is present, and what changes if that person later sides with the majority. Asch's desertion result is the warning for a leader who backs the dissent in the morning and folds after lunch. And because the people who yielded undercounted their own yielding, do not accept a retrospective "we spoke up" as the receipt. You had to hear the other line when it was still costly.
Collective Campus has done capability building for teams since 2015. The psychological safety workshop is for an intact team that will practise one real disagreement with a second voice in the room before the sponsor settles it. Why capable people go quiet is in why your best people stopped speaking up. What happens once the boss has already given the answer is in nobody disagreed once the boss had spoken. The rule people are already holding, before any exercise starts, is in the rule a safety day has to surface.
What to bring
Bring one decision the team is actually in, and permission for someone who is not the sponsor to name what they see before anyone else agrees. If the answer has to be unanimous before it can be said, the long line loses.
