Aims: To establish the extent that group pressure can influence an individual to conform to that group's way of thinking.
Procedure: A set of participants is seated around a table. Each participant in turn is shown a card which has a "reference" line drawn on it, and another card which has three labelled lines. The task of the participant is to state publically which of the three labelled lines matches the reference line. In each case, the task is designed to be easy. In fact, all but one of the participants are confederates of the examiner, and, on a secret signal from the examiner, will deliberately provide an incorrect answer. The test is designed so that the "innocent" participant is the penultimate person to have to answer, and so by the time his turn comes, he will have heard most of the other "participants" answer, and so be subject to pressure to conform with the rest of the group.
Findings: Depending on the size of group, it was found that participants agreed with the incorrect majority answer in 32% of trials. This rate of conformance was lower when the group size (and number of confederates) was smaller, but remained fairly constant when there were three or more confederates in the experiment.
Conclusion: From this experiment alone, we could conclude that a group exerts a strong influence on an individual to conform, especially when the individual is in a minority of one.
Strength: Asch had not expected to see such a high degree of conformity. The fact that the results of the experiment were not what he expected suggests that this was a well-designed and useful experiment: rather than confirming the experimenter's prejudice, it provided information which challenged it.
Weakness 1: It is likely that the high degree of conformance observed by Asch was in part a product of the prevailing social climate (in the time of McCarthyism). Subsequent attempts to repeat the experiment have not shown such high rates of conformance, which casts doubt on the generalisability of the results.
Weakness 2: The experiment must have been very time-consuming, since only one "real" participant could be tested on each iteration.