The Classic Group Influence Studies: Sherif, Asch and Milgram

1. Muzafer Sherif and the Formation of Social Norms (1936)


In 1936, Sherif reports a concern for the dramatic changes in social life associated with the '30s in America, the rise of totalitarian governments in Europe, widespread hunger and starvation, oppression of the powerless and the mobilization of mobs through political sloganeering. He suggested "the study of such unstable situations of oppression, hunger, and insecurity and their psychological consequences demand careful attention from social psychologists...especially in our time of transition" (1936:193). Again - "when social life becomes difficult...the equilibrium of life ceases to be stable and the air is pregnant with possibilities...Such a delicate, unstable situation is the fertile soil for the rise of doubts concerning the existing norms, and a challenge to their authority" (1936:85). Sherif was preoccupied with the important tensions in society that arose during the Depression.

When the normal historical patterns of interaction break down, people spontaneously evolve their own norms - i.e. standards of behaviour that regulate the conduct of individuals. Norms are patterns of action which people feel compelled to subscribe to because they appear to be appropriate, moral or ethical.

The Problem: to re-produce in the lab how norms could be seen to be evolving in society

The Question: what will the individual do when the individual is placed in situation where the environment cannot be defined objectively?

The Design: Use of the Autokinetic effect - an optical illusion produced when people are exposed to a pinpoint of light in a darkened room - "auto-kinesis" means that it appears to be self-moving or unstable because the darkness provides no frame to capture and stabilize the light source.

: Ss are placed in a dark room; Sherif instructs them that a light will appear and that it will move around (expectation effects?)

: a tiny point of light is projected onto a screen and the Ss are instructed to push an electrical key to signal when they perceive the movement

What was measured? the perceived "displacement" or "movement" of the light estimated in inches

Procedures: 3 conditions

(1) Subjects run alone - 100 trials during 4 separate blocks (400 estimates in all)

(2) Subject run alone during first block then moved into a group with other Ss for the remaining three blocks

(3) Subjects in Groups during first three blocks then run alone


1.    When individuals perceive movement which lack any other standard for comparison, they establish a subjective range of response limits - a normative range specific to each individual which remains stable for that individual

2.    Ss were all run alone in the first block, then put into groups with 2 other Ss for the next three blocks. During the second block (i.e. the first group session), the estimates began to converge. They became increasingly similar during the next two group sessions - in other words, individuals adopted estimates that tended to cluster around the same mean the longer they interacted.

3.    When Ss were put into groups with 1 or 2 others, the judgments of light displacement converged virtually from the very first few estimates - in these designs the Ss were required to give their estimates out loud in consistent turn taking over each of the 100 estimates in 3 consecutive sessions. In other words, they adopted a norm from the start which they all employed commonly. When tested alone, they retain the group estimates.


(1)         In groups making judgments about ambiguous stimuli (the autokinetic effect), there is a convergence towards a norm which then determines the subsequent individual choices

(2)         Ss did not report any conscious effort to follow the others. They reported that the displacement appeared to naturally behave the way the group was beginning to report it.


(1) the cause-effect relationship here is quite ambiguous. The more the people interact with one another, the more they converge in their definition of reality. But what is the cause here? Is it time? experience? What is the underlying process? Time is not a cause of anything - it merely marks that fact that changes are occurring not what is causing them.

(2) the Ss also claimed that the stimulus was ambiguous: "Darkness left no guide for distance. It was difficult to estimate the distance...There was no fixed point from which to judge the distance" (1936:97). Significantly, Sherif acknowledges that "the effect takes place even when the person looking at the light knows perfectly well that the light is not moving" (1936:92). How could this be a "norm" when people "knew" differently from what they "saw"?

At any rate, group pressure mediates how the world is perceived and defined, even of this study doesn't tell us exactly how this is occurring.


2. S.E. Asch: Effects of Group Pressure upon the Modification and Distortion of Judgments (1951)


Solomon Asch contributed a classic study of group influence on the individual. His work appeared after the Second World War. Where Sherif stressed how subjects were influenced by the group outlook, Asch was interested in the grounds of resistance to group pressure. In the introduction to his work, Asch reviewed the enormously important issue of propaganda which had been so influential in mobilizing the German and Italian populations during the second world war to support the war effort, and which, in the German case, promoted racial hatred resulting in genocide. For Asch it was important to identify the grounds for resisting such evil influence.

The immediate objective was to create laboratory conditions in which the Ss were asked to agree to views which were obviously contrary to fact - and to determine the conditions in which Ss resisted the majority errors.


The Experiment involved a "Line Discrimination Task". Eight Ss met together in groups. Each individual was shown a series of cardboard displays with lines of different sizes. They were asked to match these lines to one of three lines on a standard chart. Each S made an estimate - all 8 people out loud in a turn taking procedure - then the procedure was repeated with each new stimulus line. The key manipulation was that sometimes one of the Ss patently disagreed with the majority.

Main features of the Experiment

(1) The Ss have two sources of information - what they can see with their own eyes and what they hear the others reporting

(2) Both sources of information were present simultaneously

(3) The public declaration made the inconsistencies highly obvious

(4) According to Asch, the Ss appeared quite spontaneous and natural - i.e. the Ss were surprised and alarmed at their inconsistencies

Basic Procedure: 18 trials of the match making task. 12 of the trials were designated as "critical trials". While there were 8 people in the group, in fact 7 were 'confederates' of the experimenter, graduate student assistants. In the 12 critical trials, they deliberately responded incorrectly - making the real remaining S the lone person at variance with the majority.


(1) Taking the critical group as a whole, there was a clear movement toward the majority view - the incorrect view. Specifically, about one-third of the critical estimates (not estimators) by real Ss were either identical to the Majority Error, or in the direction of the Error. The optimistic view is that two-thirds of the estimates on the critical trials were correct despite external pressure.

In the control group, the Confederates responded correctly in the line matching task. Here only two guesses by real Ss were in error.

(2) There appears to have been remarkable individual variation among the conduct of the real Ss. One-quarter never were swayed by the majority view, never made any errors. But about one-third were swayed in 50% or more of their trials. This led Asch to explore the personal characteristics of those who tended to be either "Independents" or "Yielders". These ideas were pulled together following the de-briefing of the Ss.

Among the Independents:

(1) Independence based on a strong self confidence about their perceptions and experiences. Vigorously resisted the pressure from others often by making their choices known in a loud and assertive voice - as though the others had got the job wrong.

(2) A second posture was that Ss became clearly withdrawn from the group, as though they were withholding their emotional investment in the group - to some extent neutralizing the costs of disagreement

(3) The last posture was one of grave uncertainty. These people resisted the pressure but exhibited obvious nervousness, tension, self doubt and anxiety.

Among the Yielders:

(1) Distortion of Perception. Ss reported that they came to believe that what the others said was accurate. I'm not sure about this - it reminds me of Leon Levy's study of verbal conditioning where the Ss who were told the point of the experiment denied being so told. Here the de-briefing seems to overlook the possibility that the Ss are merely telling a lie to repair a lie - and so ensure the integrity of the experiment. Another point, unless the lines were self-evidently different, no inference could be drawn about group pressure since mistakes or confusions could happen naturally - but the discrimination task was designed to make the differences appear self evident.

(2) Many Ss doubted their own self perception and decided that they would "go-along" with the majority since the majority, because it was larger, would be more reliable. Others said that perhaps they had missed some of the instructions and didn't understand but that surely the other 7 had paid more attention.

(3) And finally some Ss admitted that they were afraid of being different, afraid of standing out from the crowd, and being viewed as inferior.

Asch abandoned the quasi-clinical post hoc line of explanation since it seemed to point in the direction of personality differences - which it does - and which may simply lead to tautological explanations. I.E. people "cave-in" because they have yielding personalities. What was the evidence of the personality?- the outcome! Any independent measure of that personality? No. If not, the explanation is circular or tautological.

Effects of Group Structure:

Asch explored (i) the effect of partners (ii) group size and (iii) size of the group error


A: The Ally. In further variations of the experiment, Asch introduced conditions in which the lone S would have an ally. In the first instance, 2 real Ss were used - one in the 4th and one in the 8th position. In the critical trials, the level of aggregate errors dropped to 10.4%. In another variation, a confederate in the 4th position always chose correctly - and the real S in the 8th position was obviously assisted by this. Here the errors dropped to 5.5%. (vs 32% in the basic design) Therefore, he concluded that minimum human support erodes the pressure of the majority.

B: The Withdrawal of the Ally. In another variation, the conduct of the supporting ally was manipulated. Again, real Ss appeared in the 8th position, the confederate in the 4th. Half way through the trials, the confederate goes with the erroneous majority position - and as a result the real S loses courage. Errors jump from 5.5% to 28% of all estimates given - almost as high as the basic design.

C: The Late Ally. In a third variation, Asch introduces a confederate ally who begins to support the real S half way through the critical tests. Here the errors drop from 32% to 8%. However, those who had been most conforming ignored the late ally and continue with the majority. A change would have made their prior behaviour appear to be erroneous, so they ignore the ally and act in a cognitively consistent way.

D: The Compromise Ally. Again in this design, the Ss are in the 8th position, the confederate is in the 4th. In this design, the group as a whole chooses the line least consistent with correct line. The ally chooses the next closest - but not the right one. The real Ss appear to take some moral support from the split in the group, but the level of decline in errors did not reach statistical significance. The compromise is only a weak consolation to the S. The S does not follow the compromise because it is still seen to be wrong.

Role of the Majority Size

The size of the confederate group was varied from between 1 and 15 people. The maximum number of errors during the 18 trials would be 12 - arising from the 12 critical trials. So what were the average number of errors for a majority of various sizes. Obviously, we are not all that much influenced by a single other person who seems to make mistakes systematically. But what is the critical number which extracts the maximum amount of compliance from the real Ss?


Number of Confederates







Mean Errors







So a maximum effect appears to be extracted when there are just 3 other erroneous opinions (4 versus 4.2 is not a significant difference) - groups of a larger size don't seem to extract accordingly larger conformity.

Size of the Group Error

Asch's point here is simple. The larger the deviation of the whole group from the correct choice, the smaller the effect of control on the real Ss - although again individual differences remain important - a minority of people follow even when the majority appear to be wildly off-base.

Rather than summarizing, let me leave you with a question: what did Asch's work really tell us about WW2 propaganda and the conditions under which it can be resisted? What did Sherif tell us about the Depression. What did either of them tell us that wasn't common knowledge? Or did they merely "borrow" from the real world to sketch a little 'dramatization" - one which required us already to be literate about the things being explained? I'll leave off today with that little insurgent question.

The next important contribution to group influence was the work of Stanley Milgram - Obedience to Authority.


Stanley Milgram, Obedience to Authority and Obedience (The Film).


Sherif's work suggested that our knowledge of the world was influenced by others in the absence of objective cues. Asch suggested that even with the physical cues, our perception of the world is coloured by others (or at least our verbal behaviour regarding the perception of the world is coloured by the group). Asch also suggested that the size of the group, the role of an ally and the magnitude of group error all influence our perceptions. Milgram's position is that aside from group influence, there is another kind of influence, the influence of authority, especially the authority of just one other person, a person of superior status. The task of the experiment was to make the subject obey commands that involved hurting another person. The event which motivated the study was the prosecution of Adolph Eichmann in Jerusalem for his role in deporting the Jews of Europe to the ghettos and eventually to the death camps in Poland, principally Auschwitz. Several million people were murdered at Auschwitz. A good recent contribution to this period can be found in Daniel Goldhagen's Hitler's Willing Executioners (1996).

Summary points regarding the design and results:


wide subject base (not college students) - skilled and unskilled workers (40%) - white collar, sale, business.(40%) - and professionals (20%)


The Learning Task - use of a shock to "teach" the Learner pair associations of words. The 2 recruits "flip" for the roles of Learner and Teacher - but the flip is rigged - the Learner is a confederate and the Teacher is a real S. The Learner leaves room to occupy learning desk and the Teacher reads instructions on an intercom.


The Impressive Shock Generator with graduated electrical switches from 15 volts up to 450 volts (marked "XXX") -misleading device which does not deliver shocks but which measures the highest level of shocks administered. Each shock is given once and the Teacher graduates to the next switch.


"Sample" shock given to Teacher to lend validity to the experiment.


The Experimenter who directs the Teacher engages in prods and directives to encourage Teacher to continue in spite of resistance.


Various Levels of "Feedback" from the Learner (Design Conditions).


Milgram asks various groups to estimate the maximum shock level which Ss would use and the percent who would defy the experimenter


Outcomes (see #6 above for main findings)-> change of personnel: women as teachers - voice feedback compliance was 65%



Mixon's Objections: Mixed Signals versus the Certainty of Auschwitz


Baumrind's Ethical Objections - was it ethical to make Ss suffer? Did Ss agree to have their dark natures revealed to them? Did we learn anything about conduct we didn't already know?


Steve Patten: Milgram knew that the experiment was traumatizing some Ss but continued after learning this because he believed science would benefit. Yet the Ss are denied the same rationale - they believe they are helping science and are told that this sacrifice is required of the Learner - yet we view Milgram as justified but don't give the same consideration to.the real Ss.


Brannigan's Objection: every manipulation which reduced ambiguity about harm, that made it clear that someone was suffering reduced compliance. This is grounds to be optimistic about human nature - not skeptical...Back to Goldhagen's alternative model: the Holocaust was .not about obedience to authority but deep-seated "exterminationist anti-Semitism."