Professional Documents
Culture Documents
SERGE MOSCOVlCl
ELISABETH LAGE
Ecofe des Hautes Etudes en Sciences
Sociales. Paris
Abstract
1. Preliminary observations
The research we have carried out over the past few years (Faucheux and Mosco-
vici, 1967; Moscovici, Lage and Naffrechoux, 1969) indicates that thc behavioural
style of a minority constitutes an element for its becoming a pole of influence in a
group. Specifically, we assume that a minority must provoke a cognitive and
relational conflict in the group before it can exert influence and induce the
majority to change their opinions or judgments (Moscovici and Faucheux, 1972).
In so far as it arouses such conflict, the minority’s behavioural style constitutes a
source of change.’
Our intention in the present article is to explore experimentally this assump-
tion in greater detail. First, we wish to show that a consistent minority is more
effective than an inconsistent minority. Secondly, we wish to consider why a
minority of one individual does not modify the judgments of a majority with
respect to an explicit and objective norm, while a minority consisting of two
individuals does. Briefly, in situations in which the minority consists of a sub-
group of two people, one of them sees himself as the ‘innovator’ and the other
plays the role of ‘follower’. If an individual is alone and isolated, the majority
can reduce the conflict and reject his deviant response by attributing it to a
peculiarity specific to him. On the other hand, when a second person follows
him and adheres to his judgment, the conflict cannot be reduced so easily, nor
the deviant response rejected. In this case, a change is effected in the majority
which is brought to redefine itself by reference to the minority. Finally, the cost
of maintaining a deviant point of view is greater for a minority than for a major-
ity, and a minority must therefore have stronger motives and be more convinced
in order to do so. It follows that a deviant point of view held by a minority will
carry more weight and will create a deeper cognitive and relational conflict than
the same point of view held by a majority.
Within certain limits, a minority which conforms to a majority feels itself
socially justified in doing so, and no internal psychological processes are neces-
sary to bring about a change, while, on the other hand, a majority does not feel
thus justified in conforming to a minority, and quite complex internal processes
are required before it can give up its own position. Whether for either one of
these reasons or for both, it seems to us that a minority stimulates a more acute
conflict in a group than a majority, and, in consequence, its influence should
bring about greater changes. Thus we expect the influence of the minority to
1. We shall not elaborate on these notions readers to the articles mentioned in the
which are treated elsewhere. We refer text.
Studies in social influence III 151
operate at both the manifest and latent levels of judgment, while the influence
of the majority on the minority will be limited to superficial agreement or manifest
conformity.
We are well aware that these phenomena are more complex in real life, and
that it is difficult to provide incontestable proof of our assumption and the pro-
positions following from it. But we feel that progress may be made in this direc-
tion, given the importance of the problems involved, and, within limits, the novelty
of unexpected findings. The experiments reported here were designed to test the
following hypotheses:
1. A minority behaving consistently has influence at both manifest and latent
levels. On the other hand, an inconsistent minority has no influence at either level.
2. Where the changing of an explicit and well-structured norm is concerned
an effective minority must be composed of a group of individuals (in our case,
two), and not of a single individual.
3. A consistent minority exerts as much influence as a consistent but non-
unanimous majority.
4. A consistent minority exerts influence at both latent and manifest levels,
whereas a majority, whether unanimous or not, exerts influence only at the
manifest level?
5. The agents of minority influence are more readily accepted by the group
than the agents of majority influence.
These hypotheses suggest certain differences between the dynamics of con-
formity induced by a majority and the dynamics of innovation introduced by a
minority.
Before giving the results of this research, we shall describe briefly the experi-
mental design and the measures of the principal variables employed.
2. Experimental design
Since the research was intended to explore influence behaviour at three different
levels, the experimental design included the following sequential stages: (a) inter-
action, (b) investigation of latent response (test of colour discrimination), (c) post-
experimental questionnaire.
2. We would like to point out that influ- and private responses. While the individual
ence exerted on the manifest and latent can control his private responses, influence.
levels should not be confused with the on the latent level eludes such self-control.
distinction frequently made between public
152 Serge Moscovici and Elisabeth Lage
2.1 Interaction
This stage corresponds to the classic situation used in the experimental study of
social influence. It involves opposing the opinion of the agent of influence to the
general opinion expressed by the various members of the group. We selected a
situation involving responses to structured stimuli with regard to which there are
established norms. The agent of influence contradicted the expressed judgment
of the other individuals.
Within the field of perception, we specifically selected colour vision. Colour-
naming is part of an objective and stable body of knowledge in a given culture.
It is based on cultural obviousness and is learned from childhood on; individuals
regard judgments of colour as self-evident. We chose stimuli which were called
‘blue’ by almost the whole of the control group and for which the agent of in-
fluence gave the deviant response ‘green’. The latter response, although plausible
for the material concerned, was statistically very rare.
3. Responses in terms of light intensity will make the task more complex and more
not be analysed as they are not of direct realistic.
interest. We simply introduced them to
Studies in social influence 111 153
2.1.2.1 Minority
Three influence situations were created in which the source of influence was in a
minority, a consistent subgroup, a consistent individual and an inconsistent sub-
group:
1. The consistent minority of two individuals responded ‘green’ throughout
the interaction. The confederates were located at the first and fourth positions
in a group of six persons.
2. A consistent individual maintained the response ‘green’ throughout the
interaction. He was located at the first position in a group of four individuals.
3. Two inconsistent individuals sometimes gave the response ‘green’ and some-
times the response ‘Slue‘. The former response was given to 24 slides, the latter
to 12 slides in a random order independent of light intensity. The confederates
occupied positions one and four in a group of six individuals.
2.1.2.2 Majority
Two different situations were used here, unanimous and nonunanimous majority:
1. The unanimous majority, consisting of three confederates, gave the response
‘green’, with one naive subject. The subject was therefore confronted with a con-
sistent agent of influence, both synchronically and diachronically. The confeder-
ates occupied positions 1 , 2 and 3 in a group of four persons.
2. In the case of the nonunanimous majority. two naive subjects were con-
fronted with four confederates. These confederates always gave the response
‘green’, but for each naive subject the interindividual consistency of the majority
was broken by the other naive subject! The positions of the confederates were
varied in this situation, 1, 2. 3 and 4; 3, 4. 5 and 6 ; or 1. 2, 4 and 5 in a group
of six persons.
154 Serge Moscovici and Elisabeth Lage
4. The other way of disrupting this con- ‘green’ (50 per cent/50 per cent). Lower
sistency is to use a single naive subject in threshold: when the stimulus yields 75 per
the group and one confederate in the cent ‘green’ responses and 25 per cent
majority whose business it is to give respon- ‘blue’.
ses that are different from those of the Upper threshold: when the stimulus yields
other confederates (cf. Asch, 1956). 25 per cent ‘green’ responses and 75 per
5. Differential threshold: when the stimulus cent ‘blue’.
is judged ‘blue’ as frequently as it is judged
Studies in social influence Ill 155
change on the basis of these responses alone. At the present time, such a distinc-
tion seems to be quite a complex matter in colour vision (Goldiamond and Mal-
pass, 1958). That is why we called the object of study in this stage of our ex-
periment the ‘perceptual-cognitive code’, without specifying any further the level
involved. Since there has been very little research done on the latent effects of
influence, we regard the present experiment as a first attempt to approach this
problem. Since the problem is such an important one, we hope to be able to
develop more elaborate techniques of analysis, should the results of this experi-
ment prove to the positive?
We shall analyse the results from each experimental stage successively, examining
the influence of the minority and that of the majority, and suggest that two types
of phenomena are involved.
8. Three main indices were used in this The difference in the criteria adopted for
analysis: (a) the per cent of ‘green’ respon- these two measures is justified by the fact
ses in relation to the total number of that there must be a minimum of ‘green’
judgments; (b) the per cent of groups in responses in the group for the other naive
which this judgment appears (‘influenced subjects to be aware of its existence,
groups’); and fc) the per cent of naive whereas a single ‘green’ response retains its
subjects who respond ‘green’ (‘influenced salience for the subject who has given it in
subjects’). the course of interaction.
We regard as an ‘influenced group’ any Since the distribution of the results it not
group in which at least three ‘green’ re- normal, nonparametric methods will be used
sponses are given, and as an ‘influenced in the analysis: the Mann-Whitney U-test
subject’ any individual who gives a ‘green’ for the first measure, and the chi-square
response at least once during the interaction. test for the other two.
Studies in social influence III 157
Indices of influence
~
show that judgments of ‘green’ represent 10.07 per cent of the responses. This
differs significantly from the responses of the control group ( z = 4.09, p < .001).
This effect is not due to a few exceptional individuals. In fact, this response
appears in 50 per cent of the groups and is given by 42.50 per cent of the indivi-
d u a l ~ (b)
. ~ Let us now consider the influence of a single consistent individual.
Responses of ‘green’ represent only 1.22 per cent, which does not differ signifi-
cantly from the control group. Thus it seems that a single individual, even if he
is consistent, does not modify the majority judgment in our experimental situa-
tion. He has significantly less influence than a consistent subgroup, which sup-
ports Hypothesis 2 (per cent ‘green’ responses, z = 3.39, p < .003; per cent
of ‘groups influenced’, X 2 = 4.83, .05 > p >
.01; per cent ‘individuals influ-
enced’, X z = 9.81, p < .Ol). (c) An inconsistent subgroup of two individuals has
no influence on the judgments of the majority. The results do not differ signifi-
cantly from those of the control group, but there is a significant difference with
respect to the results obtained from the consistent minority (per cent of ‘green’
responses, z = 3.38, p < .003; per cent of ‘individuals influenced’, X = 10.50,
p < .Ol). A minority subgroup thus has immediate influence only if it behaves
consistently. This finding confirms our first hypothesis.
On account of this quantitative analysis we feel more confident in recommend-
ing as a basic theoretical notion the idea that a minority’s style of behaviour may
be a decisive factor in the process of innovation. Moreover, at least in our ex-
9. These are the results from our previous tions referred to, in which we give complete
research (Moscivici et al., 1969). They cor- results for the three stages of the design.
respond to one of the experimental varia-
158 Serge Moscovici and Elisabeth Lage
Indices of influence
10. We also analysed the influence of the this factor does have some effect. Subjects
nonunanimous majority as a function of its are less influenced when they answer at the
position in the group. The results show that beginning than at the end.
Studies in social influence I l l 159
difference in this respect (cf. Tables 1 and 2). The predictions of Hypothesis 3
are thus verified."
We are still left with the problem of understanding the nature of the minority
and majority influences. We must decide whether they involve similar or dif-
ferent mechanisms, whether they induce similar or different effects. The second
stage of our experimental design enables us to approach this problem.
11. In order to generalise this finding to minority and majority subgroups. This issue
larger groups, it would be necessary to could not be reduced to a mere question
first study the criteria for the perception of of numbers.
160 Serge Moscovici and Elisabeth Lage
In this analysis, we shall use as indices the differential threshold and the upper
and lower thresholds established from individual responses. The displacement
of these indices with respect to the control group data will be taken to indicate a
change in the perceptual-cognitive code. A wide dispersion of scores will be
taken as an index of conflict to which the subjects react in different ways.12 It
will also be interesting to study the responses of subjects according to their
behaviour in the interaction phase. Let us first examine the effects of minority
influence and then those of majority influence.
3.2.1 Minority
Table 3 indicates the displacement of response of the experimental groups from
those of the control group. Table 4 shows the dispersion of these responses.
(a) Subjects who were confronted with a consistent minority of two individuals
modified their perceptual-cognitive code after their interaction in the influence
situation. They displaced their judgments, as compared with those of the control
group, calling those stimuli ’green’ which the control group still called ‘blue’.
This displacement was significant for all three measures, as shown in Table 3.
This reaction is not exclusively due to the subjects who were ‘influenced’ during
the interaction phase, for it is found to the same degree in those individuals who
responded ‘green’ as in those who persisted in giving the response ‘blue‘. There
is no significant difference between these two groups.
Thresholds
* One-tailed test
12. The displacement is measured by the homogeneity test (Winer, 1962, p. 33), com-
Student t-test, comparing the results of the paring the variance in the experimental
experimental groups with the results of the group with that of the control group: F =
control group. VI
- the degrees of freedom being, respec.
Dispersion is measured by the variance v2’
tively, (n, - 1) and (nz - 1).
Studies in social influence III 161
Our examination of the latent response therefore reveals more important effects of
the influence than was initially indicated in the manifest responses. In other words,
subjects were influenced by the minority whether or not they showed the effects
of this influence in the interaction stage of the experiment. This aspect of our
findings would seem to be of particular importance in that it shows clearly that
the manifest response does not reflect all aspects of influence behaviour. A certain
range of responses, however, indicates that the displacement of threshold varied
from subject to subject (differential threshold and upper thre’shold). This latent
influence effect, which we had been hoping to demonstrate, depends upon the
number of persons that make up the minority and upon its behavioural style.
Thresholds
(b) A single consistent individual, isolated in the group, does not produce this
effect. No significant displacement of thresholds is to be found in this situation.
The only effect which occurs concerns the range of responses (differential thres-
hold and lower threshold). This finding appears to be due to the fact that some
subjects generalised the notion ‘blue’, and others the notion ‘green’, but it is
impossible to establish a systematic effect at this level in view of the disproportion
between the ‘influenced’ and the uninfluenced subjects. (c) The same is true, in
this experiment, of the influence effects of an inconsistent minority. There is no
difference between the experimental groups and the control group in either
displacement or range of responses.
The influence of a consistent minority therefore has deep-lying effects. It is not
limited to a modification of manifest responses but involves a change at the level
of the perceptual-cognitive code underlying the manifest responses. Moreover,
these effects are more pervasive than the foregoing quantitative analysis indicates,
162 Serge Moscovici and Elisabeth Lage
since they affect the majority of the individuals who were confronted with this
minority. But latent influence, like manifest influence, depends upon the behav-
ioural style of the agent of influence and its numerical strength in the group. A
minority which has no manifest effect has no latent effect either. It remains to be
seen whether this phenomenon is specific to minority influence, as we hypothesised,
or if it is also to be found with majority influence.
3.2.2 Majority
It will be recalled that a naive subject exposed to the influence of a unanimous
majority was, in general, persuaded to share the majority opinion. What lies be-
hind this influence? Did the subjects modify the basis of their colour-naming
after the interaction, or did this basis remain unchanged in spite of the influence
at the manifest level? (a) The results of the second stage of the experiment show
that there is apparently no modification of latent judgments. The differential
threshold was not displaced in comparison with that of the control group; and
there was no significant difference in the dispersion of responses. The only positive
result concerns the behaviour of the subjects who were ‘influenced’ during the
interaction phase. These subjects generalised the response ‘green’ for the upper-
limit threshold to a greater extent than the control group did ( t = 2.02, .05 >
p > .01). For the other measures, no such results were found. Thus no deep-
lying modification of judgment can be inferred from the effects of majority in-
fluence.
Thresholds
Apparently then, subjects yield to majority pressure during the interaction, but
this submission is superficial, and the subjects do not really assimilate the judg-
ment of the agent of influe2ce. This suggests that minority influence and majority
influence involve different mechanisms. Further confirmation of this is provided
by the results from the ponunanimous majority situation which indicate the same
tendency. (b) Here the perceptual-cognitive code of the subjects also remained
Studies in social influence 111 163
Thresholds
unchanged after the influence situation. However, there was a greater range of
responses than in the control group (differential threshold and lower threshold).
This finding seems to be due to the fact that some subjects persisted in judging
stimuli as ‘blue’ which were regarded as ‘green’ by the control group. Unfor-
tunately we cannot show that there is a systematic tendency here. If we could,
this would have indicated a polarisation of judgments. But the responses of the
‘influenced’ subjects do not differ from those of the ‘noninfluenced‘ subjects.
Thus, there does not appear to be any major finding concerning the modification
of latent response as a result of majority influence, unanimous or nonunanimous.
This stage of the experiment has enabled us to show that minority and majority
influences differ in nature. A minority, without obtaining substantial acceptance
of its point of view at the manifest level, can nevertheless influence subjects to
revise the very basis of their judgments, while a majority can make them almost
all accept its point of view, if it is unanimous, without affecting the underlying
perceptual-cognitive system. In other words, majority influence works on the
surface while minority influence has deep-lying effects. Our fourth hypothesis is
therefore confirmed.
This finding has important implications for the analysis of influence processes
because it suggests that different long-term effects may be attributed to innova-
tion and conformity. Since the present research is only the first of its kind, it
would seem that further research is called for. There is a need for new ex-
perimental techniques capable of analysing latent processes.” But it would appear
indisputable at the present time that the quantitative study of manifest judgments
will not suffice to increase our knowledge of the processes of influence, and, in
particular, will not suffice to enable us to comprehend the particular nature of
innovation compared with conformity.
The data from these two experimental phases support our hypotheses and
contribute to the analysis of innovation in relation to our previous findings. It
seems at present to be the case that a consistent minority can influence a majority
and modify structured and explicit norms. This modification occurs at a deep-
lying level, unlike the superficial effects induced by majority inf1~ence.l~
But the
behavioural style of the source of influence has a decisive role in this process,
and the numerical strength of the minority is also significant. To clarify the in-
fluence conflict, let us now consider the findings of the third stage of the ex-
periment.
3.3.1.1 Minority
Inconsistent minority. Here the responses of the subjects differed little from those
of the control group. They were not inclined to adopt the response ‘green’ as their
own (question 4). In judging the slides, they did not use this response any more
frequently than the control group (question 8). At the very most, they more fre-
quently regarded the response ‘green’ as a possible response to the stimulus
(question 3; z = 2.86, p .004). The number .of shades distinguished indicates
that they were not entirely immune to the response conflict induced by the in-
166 Serge Moscovici and Elisabeth Lage
3.3.1.2 Majority
Unanimous majority. The questionnaire shows that in this situation subjects tried
to resolve the response conflict in which they found themselves by regarding
the responses ‘blue’ and ‘green’ as equally plausible for the stimulus material.
Their responses did indeed differ from those of the control group in this way. The
response ‘green’ was accepted by the subjects as plausible (question 3; z = 3.61,
p < .003) and their personal judgment supported this (question 4; z = 2.13,
p .03). At the same time, however, the response ‘blue’ loses being convincing as
perceptual evidence (question 5, z = 3.52, p < .003; question 6, z = 2.01, p .04).
Both responses were regarded as equally likely for these slides. They also distin-
guished significantly more shades (question 8; z = 1.70, p .06). These findings
are further emphasised by the fact that subjects returned to the response ‘green’
during interaction.
Studies in social influence III 167
All these findings clearly confirm our findings for manifest responses. The
naive subject, exposed to a unanimous judgment at variance with his, is induced
to accept this point of view and to doubt the generality of his own response, al-
though it rests on sound evidence.
This agreement, due to influence and persisting in the postexperimental ques-
tionnaire, nevertheless represents a superficial agreement, as we showed by means
of the colour discrimination test. The responses to the questionnaire enable us to
propound an explanation of the acute conflict which the subjects experienced.
Although they came to doubt the evidence for their judgments, they did not
modify their criteria of ‘blue’ and ‘green’. Powerful constraint seems to exert
great influence and to provoke acute conflict without apparently inducing any
profound change in the subjects’ thinking.
effects produced by these two influence situations are thus not identical, although
the manifest responses are quantitatively similar. The colour discrimination test
indicates two distinct phenomena: Innovation leads to a deep-lying transformation
of judgments while conformity is represented mainly by superficial agreement.
Competence assigned to the minority. Our findings indicate that the minority is
judged incompetent in evaluating colour, regardless of behavioural style (con-
sistent or inconsistent) and regardless of its numerical significance (one or two
confederates). In fact, in all three experimental conditions, the naive subjects
considered themselves to be more competent in this respect than the agent of
16. In this analysis, Student’s t-test was used was applied when more than two measures
when appropriate, namely, for independent were being compared. In question 12, the
or correlated samples. The Friedman test measure used was the chi-square.
Studies in social influence 111 169
influence, whether they responded ‘green’ or not during the interaction stage.
Question 10:
Consistent minority of two confederates t = 7.02, p < . 0 0 1
Consistent minority of one confederate t = 10.58, p < .001
Inconsistent minority of two confederates t = 5.00, p < .001
From these resuIts it appears that behavioural style is irrelevant to this assess-
ment. The inconsistent minority, which answered ‘blue’ intermittently, was not
judged to be more competent than the minority which exclusively gave the an-
swer ‘green’.
Where the minority is represented by two individuals, the subjects seem to
judge the first of the two confederates more severely than the second (consistent
minority, t = 2.27, .05 > p > .02; inconsistent minority, t = 3.05, .01 >
p < .001). It is as if the subjects held the first individual responsible for the
response ‘green’, while in fact both confederates make the same judgment.
It is therefore not surprising to find that a single consistent individual is judged
as even less competent than two consistent individuals (question 10; t = 1.97,
p .05). Naive subjects do not regard him as fit to judge either the colour of the
slides ( t = 10.85, p < .OOl) or their light intensity ( t = 2.70, p .01).
These findings all clearly indicate that minority influence can not be explained
by the fact that subjects attribute more competence to the minority in colour-
naming. How then are we to explain this influence?
Choices, rejections and leader designation. The individuals who form a con-
sistent minority do not differ in any way sociometrically from naive subjects.
They are not more frequently designated as leaders, nor more often chosen, nor
more often rejected. On the other hand, a consistent isolated individual is signifi-
cantly more rejected than the other members of a group (question 12; X 2 = 21.71,
p < .001). The same is true of the inconsistent minority of two confederates
(question 12; x2 = 12.15, p < ,001). In both these cases, the agent of influence
is regarded above all as a source of disturbance.
From these findings it would seem that a consistent minority appears attractive
in so far as subjects regard it as being more assured than they are. This does not,
however, imply that the minority is seen as being more powerful.
Assurance attributed to the majority. All the naive subjects regarded them-
selves as less assured when confronted with a unanimous majority, whatever their
response during the interaction stage (question 1 1 ; t = 3.98, p < .001). This
finding is once more a function of their isolation; if the unanimity of the majority
is broken, this is enough for the two naive subjects to regard themselves as just
as assured as the confederates.
The conflict induced by the majority is thus essentially the consequence of its
behavioural style. The pressure exercised on the isolated subject makes him doubt
both his competence and his assurance. On the other hand, when two naive sub-
Studies in social influence III 171
jects are confronted with a majority, they question its competence and retain
their own assurance.
Choices, rejections and leader designation. For a unanimous majority, we can
only compare the sociometric responses concerning the different confederates.
since a single naive subject cannot chose or reject himself. The only significant
result here indicates that the first confederate is less rejected than the other two
(question 12: X = 3.49, .I0 > p > .05). For a nonunanimous majority, the
naive subjects reject the confederates significantly more frequently than they
reject their partner (question 12: x2 = 5.82. p .02).
We set out in this article to give a more detailed analysis of the phenomenon of
innovation, starting from our previous research on the subject, and to compare
it with the phenomenon of conformity. The analysis was carried out at three
levels - that of manifest influence behaviour. that of latent influence behaviour,
and that of interindividual relations.
172 Serge Moscovici and Elisabeth Lage
One result of this research suggests that a minority may influence a structured
and explicit norm. This influence is first noticeable at the manifest level and is as
great here as that of a majority in a comparable situation. But the impact of the
minority extends beyond this. Individuals assimilate the opinion of the agent of
influence at a latent level and their underlying response code is thereby modified.
This process effectively holds for all individuals, whatever their attitude during
interaction; and this indicates that the extent of minority influence is much greater
than analysis of explicit judgments would suggest.
All our results agree in attributing to consistent behavioural style the function
of a genuine source of influence in the process of innovation. If a minority can
influence majority opinion, it is due neither to the competence of the individuals
who constitute the minority. nor to their position of leadership, but solely to their
conviction and coherence. Such a behavioural style engenders an influence con-
flict, both cognitive and relational, which can explain the observed phenomena.
It would appear that the minority must be represented by a subgroup of indivi-
duals to become influential. What is involved is not, to repeat. a matter of num-
bers but of the dynamics of conflict in which each member of the minority plays,
in the eyes of the others. a differentiated role which is an essential part of the
process by which the majority opinion is brought into question. The initiative for
the deviant response falls upon the first individual to introduce it into the group,
while the second individual brings a kind of social warning to this point of view
and ensures that it is not regarded as a purely individual aberration, as may happen
in the case of a single individual giving the deviant response.
Thus we see that consistency acts not so much as an individual variable but
rather as a variable which affects the structure of interaction. A consistent minor-
ity achieves social recognition of its qualities in the eyes of others; individuals
seem to admire the courage and audacity of the agent of influence. The azent of
influence thus becomes a pole of persuasion, without however being invested
with authority and without the members of the group joining it because of a need
for affiliation. The attraction exerted by individuals who are coherent and con-
sistent. although in a minority, cannot be explained in terms of the factors to
which social influence is traditionally ascribed. The study of innovation thus
forces us to reconsider the theoretical analysis of influence processes.
We also made some attempt to compare innovation and conformity. In order
to do so. we reproduced the Asch phenomenon with our experimental material.
We confirmed the findings of Asch himself, namely, that the influence of the
majority is overwhelming at the manifest level, provided the majority is unani-
mous. The disruption of the interindividual consistency of the majority consider-
ably diminishes its impact. Behavioural style thus appears to play the role of a
Studies in social influence ZII 173
REFERENCES
Rbumk Zusammenfassung
Cette Ctude expCrimentale avait pour but Diese experimentelle Studie zielte darauf ab,
d’analyser les rnkcanismes de l’influence die Beeinflussungsmechanismen zu untersu-
impIiquCs dans Ies deux phtnornknes fonc- chen, die rnit den zwei funktional entgegen-
tionnellement contraires de l’innovation et gesetzten Phanomenen der Innovation und
174 Serge Moscovici and Elisabeth Lage
d u conformisme. Nous avons pendant plu- Konformitat verkniipft sind. Aus Griinden
sieurs anntes effect& des recherches sur der besonderen Bedeutung und der be-
I’innovation par suite de son importance in- schrankten Forschung in diesem Bereich
trinssque et du nombre limit6 d’6tudes qui beschaftigten wir uns bereits seit einigen
lui avaient et6 consacr6es. Jahren mit dem ersten dieser beiden Phano-
Dans le prtsent article, nous nous som- mene.
mes efforcC, non seulement d‘analyser plus Im vorliegenden Artikel haben wir ver-
attentivement la question mais aussi d’effec- sucht, die Position nicht nur griindlicher zu
tuer une comparaison entre les effets de analysieren, sondern auch die Effekte von
I’innovation et ceux du conformisme. Nous Innovation mit denen von Konformitat zu
avons en particulier essay6 de dimontrer vergleichen. Insbesondere versuchten wir zu
que le style de coniportement constituait zeigen, daI3 der Verhaltensstil als allgemei-
une source gCnCrale d’influence pour les ne Quelle von EinfluSnahme bei beiden er-
deux phinomtnes itudits dans lesquels des wahnten Phanomenen wirksam ist, sofern
jugements manifestes Ctaient exprim& Ce- manifeste Urteile erforderlich sind. Auf der
pendant, Ies effets latents de I’influence peu- anderen Seite konnen die latenten Effekte
vent ne pas Stre les m h e s pour I’innovation von EinfluBnahme bei Innovation und Kon-
et le conformisme. formitat verschieden sein.
Pour pouvoir analyser ces probltmes, Um diese Fragen niiher zu untersuchen,
now avons mis au point une expkrience en haben wir ein experimentelles Design aus
trois parties. La premiere avait pour but drei Teilen entwickelt. Der erste Teil sollte
ti’Ctudier le mtcanisme de I’influence mani- die manifeste EinfluBnahme auf ein quasi-
feste sur un jugement quasi physique fond6 physikalisches Urteil erforschen, bei wel-
sur un truisme culturel. La deuuikme partie chem hohe kulturelle Ubereinstimmung vor-
portait sur l’analyse des modifications la- herrscht. Der 7weite Teil diente der Erfor-
tentes que cette influence apportnit au code schung von latenten Veranderungen im ‘per-
perceptif et cognitif. La troisitme partie. ceptual-cognitive code’, die auf die EinfluB-
prtsentCe sous forme d’un questionnaire nahme zuriickgefiihrt werden konnen. Der
post-exp&rimental,visait ?I fournir des infor- dritte Tei! sollte in Form eines postexperi-
mations sur divers aspects de la question, y mentellen Fragebogens Informationen iiber
compris sur la manikre dont les sujets per- verschiedene Aspekte liefern, einschlieDlich
cevaient l’agent d’influence. Lc principal dariiber, wie die Vpn die EinfluR ausiibende
but de cette experience etait d‘etudier les Tnstanz (Einzelperson oder Untergruppe)
variations, au sein d’un groupe, du rapport w a hrnahmen.
majorit6 ou minorit6 par rapport B I’agent Der Hauptzweck der experimentellen Ma-
d’influence ainsi que son style de compor- nipulationen diente der Variation des Mi-
tement. cohtrent ou non coherent. noritats- oder des Majoritatsverhaltnisses
Les rCsultats que nous avons obtenus in- der EinfluRnahme ausubenden Instanz in-
diquent que la coherence de comportement nerhalb einer Gruppe, sowie der Variation
constituait le facteur determinant de l’in- ihres Verhaltensstils, der entweder konse-
fluence exercte aussi bien par la majoritt quent oder unbestandig sein konnte. Die
que la minoriti. Mais tandis que dam le wichtipsten Ergebnisse zeigen. daf3 die Ver-
cas du conformisme I’influence se limite B haltenskonsequenz den Hauptfaktor fur die
modifier les jugements manifestes. dans ce- EinfluBnahme darstellt, egal, ob sie von der
lui de l’innovation. elle modifie le code Majoritat oder von der Minoritat ausgeiibt
perceptif-cognitif qui est Four-jacent ?I ces wird. Aber wahrend in der Konformitatsbe-
jugements dingung die EinffuRnahme auf die Veran-
derung manifester Urteile beschrankt bleibt,
wird in der Innovationsbedingung auch der
‘perceptual-cognitive code’ verandert. der
den Urteilen zugrunde liegt.