Journal of Personality and Social Psychology
1979, Vol. 37, No. 6, 822-832
Many Hands Make Light the Work:
The Causes and Consequences of Social Loafing
Bibb Latane, Kipling Williams, and Stephen Harkins
Ohio State University
Two experiments found that when asked to perform the physically exerting
tasks of clapping and shouting, people exhibit a sizable decrease in individual
effort when performing in groups as compared to when they perform alone. This
decrease, which we call social loafing, is in addition to losses due to faulty coordination
of group efforts. Social loafing is discussed in terms of its experimental
generality and theoretical importance. The widespread occurrence, the
negative consequences for society, and some conditions that can minimize social
loafing are also explored.
There is an old saying that "many hands
make light the work." This saying is interesting
for two reasons. First, it captures one of
the promises of social life—that with social
organization people can fulfill their individual
goals more easily through collective action.
When many hands are available, people often
do not have to work as hard as when only a
few are present. The saying is interesting in
a second, less hopeful way—it seems that
when many hands are available, people actually
work less hard than they ought to.
Over 50 years ago a German psychologist
named Ringelmann did a study that he never
managed to get published. In rare proof that
unpublished work does not necessarily perish,
the results of that study, reported only in
summary form in German by Moede (1927),
have been cited by Dashiell (1935), Davis
(1969), Kohler (1927), and Zajonc (1966)
This research was supported by National Science
Foundation Grant GS40194.
The authors would like to thank Lewis Hinkle for
technical assistance and Edward Diener, John Harvey,
Norbert Kerr, Robert Kidd, George Levinger,
Thomas Ostrom, Richard Petty, and Ladd Wheeler
for their valuable comments.
S. Harkins is now an assistant professor of psychology
at Northeastern University.
Requests for reprints should be sent to Bibb
Latane, Behavioral Sciences Laboratory, Ohio State
University, 404B West 17th Avenue, Columbus, Ohio
43210.
and extensively analyzed by Steiner (1966,
1972) and Ingham, Levinger, Graves, and
Peckham (1974). Apparently Ringelmann
simply asked German workers to pull as hard
as they could on a rope, alone or with one,
two, or seven other people, and then he used
a strain gauge to measure how hard they
pulled in kilogramsof pressure.
Rope pulling is, in Steiner's (1972) useful
classification of tasks, maximizing, unitary,
and additive. In a maximizing task, success
depends on how much or how rapidly something
is accomplished and presumably on how
much effort is expended, as opposed to an
optimizing task, in which precision, accuracy,
or correctness are paramount. A unitary task
cannot be divided into separate subtasks—all
members work together doing the same thing
and no division of labor is possible. In an
additive task, group success depends on the
sum of the individual efforts, rather than on
the performance of any subset of members.
From these characteristics, we should expect
three people pulling together on a rope with
perfect efficiency to be able to exert three
times as much force as one person can, and
eight people to exert eight times as much
force.
Ringelmann's results, however, were strikingly
different. When pulling one at a time,
individuals averaged a very respectable 63 kg
of pressure. Groups of three people were able
to exert a force of 160 kg, only two and a
Copyright 1979 by the American Psychological Association, Inc. 0022-3514/79/3706-0822$00.75
822
SOCIAL LOAFING 823
half times the average individual performance,
and groups of eight pulled at 248 kg,
less than four times the solo rate. Thus the
collective group performance, while increasing
somewhat with group size, was substantially
less than the sum of the individual efforts,
with dyads pulling at 93% of the sum of
their individual efforts, trios at &5%, and
groups of eight at only 49%. In a way somewhat
different from how the old saw would
have it, many hands apparently made light
the work.
The Ringelmann effect is interesting because
it seems to violate both common stereotype
and social psychological theory. Common
stereotype tells us that the sense of team
participation leads to increased effort, that
group morale and cohesiveness spur individual
enthusiasm, that by pulling together groups
can achieve any goal, that in unity there is
strength. Social psychological theory holds
that, at least for simple, well-learned tasks
involving dominant responses, the presence
of other people, whether as co-workers or
spectators, should facilitate performance. It
is thus important to find out whether Ringelmann's
effect is replicable and whether it can
be obtained with other tasks.
The Ringelmann effect is also interesting
because it provides a different arena for testing
a new theory of social impact (Latane,
1973). Social impact theory holds that when
a person stands as a target of social forces
coming from other persons, the amount of
social pressure on the target person should
increase as a multiplicative function of the
strength, immediacy, and number of these
other persons. However, if a person is a
member of a group that is the target of social
forces from outside the group, the impact of
these forces on any given member should
diminish in inverse proportion to the strength,
immediacy, and number of group members.
Impact is divided up among the group members,
in much the same way that responsibility
for helping seems to be divided among witnesses
to an emergency (Latane & Darley,
1970). Latane further suggests that just as
psychophysical reactions to external stimuli
can be described in terms of a power law
(Stevens, 1957), so also should reactions to
social stimuli, but with an exponent having
an absolute value less than 1, so that the nth
person should have less effect than the
(» — l)th. Ringelmann's asking his workers
to pull on a rope can be considered social
pressure. The more people who are the target
of this pressure, the less pressure should be
felt by any one person. Since people are likely
to work hard in proportion to the pressure
they feel to do so, we should expect increased
group size to result in reduced efforts on the
part of individual group members. These reduced
efforts can be called "social loafing"—a
decrease in individual effort due to the social
presence of other persons. With respect to
the Ringelmann phenomenon, social impact
theory suggests that at least some of the effect
should be due to reduced efforts on the
part of group participants, and that this reduced
effort should follow the form of an
inverse power function having an exponent
with an absolute value less than one.
The Ringelmann effect is interesting for a
third reason: If it represents a general phenomenon
and is not restricted to pulling on a
rope, it poses the important practical question
of when and why collective efforts are less
efficient than individual ones. Since many
components of our standard of life are produced
through one form or another of collective
action, research identifying the causes
and conditions of inefficient group output
and suggesting strategies to overcome these
inefficiencies is clearly desirable.
For these three and other reasons, we decided
to initiate a program of research into
the collective performance of individuals in
groups.
Experiment 1
Clap Your Hands and Shout Out Loud
One of the disadvantages of Ringelmann's
rope pulling task is that the equipment and
procedures are relatively cumbersome and inefficient.
Therefore, we decided to keep our
ears open for other tasks that would allow us
to replicate the Ringelmann finding conceptually
and would provide the basis for extended
empirical and theoretical analysis. We
chose cheering and clapping, two activities
824 B. LATANfi, K. WILLIAMS, AND S. HARKINS
that people commonly do together in social
settings and that are maximizing, unitary,
and additive. As with rope pulling, output
can be measured in simple physical units that
make up a ratio scale.
Method
On eight separate occasions, groups of six undergraduate
males were recruited from introductory
psychology classes at Ohio State University; they
were seated in a semicircle, 1 m apart, in a large
soundproofed laboratory and told, "We are interested
in judgments of how much noise people make in
social settings, namely cheering and applause, and
how loud they seem to those who hear them. Thus,
we want each of you to do two things: (1) Make
noises, and (2) judge noises." They were told that
on each trial "the experimenter will tell you the
trial number, who is to perform and whether you
are to cheer (Rah!) or clap. When you are to begin,
the experimenter will count backwards from three
and raise his hand. Continue until he lowers it. We
would like you to clap or cheer for 5 seconds as
loud as you can." On each trial, both the performers
and the observers were also asked to make magnitude
estimates of how much noise had been produced
(Stevens, 1966). Since these data are not relevant
to our concerns, we will not mention them further.
After some practice at both producing and judging
noise, there were 36 trials of yelling and 36 trials
of clapping. Within each modality, each person performed
twice alone, four times in pairs, four times
in groups of four, and six times in groups of six.
These frequencies were chosen as a compromise
between equating the number of occasions on which
we measured people making noise alone or in groups
(which would have required more noisemaking in
fours and sixes) and equating the number of individual
performances contributing to our measurements
in the various group sizes (which would have
required more noisemaking by individuals and pairs).
We also arranged the sequence of performances to
space and counterbalance the order of conditions
over each block of 36 trials, while making sure that
no one had to perform more than twice in a row.
Performances were measured with a General Radio
sound-level meter, Model 1S6SA, using the C scale
and the slow time constant, which was placed exactly
4 m away from each performer. The C scale
was used so that sounds varying only in frequency
or pitch would be recorded as equally loud. Soundlevel
meters are read in decibel (dB) units, which
are intended to approximate the human reaction to
sound. For our purposes, however, the appropriate
measure is the effort used in generating noise, not
how loud it sounds. Therefore, our results are presented
in terms of dynes/cm2
, the physical unit of
work involved in producing sound pressure.
Because people shouted and clapped in full view
and earshot of each other, each person's performance
could affect and be affected by the others. For this
reason, the group, rather than the individual, was
the unit of analysis, and each score was based on
the average output per person. Results were analyzed
in a 4 X 2 X 2 analysis of variance, with Group
Size (1, 2, 4, 6), Response Mode (clapping vs. shouting),
and Replications (1, 2) as factors.
Results
Participants seemed to adapt to the task
with good humor if not great enthusiasm.
Nobody refused to clap or shout, even though
a number seemed somewhat embarrassed or
shy about making these noises in public.
Despite this, they did manage to produce a
good deal of noise. Individuals averaged 84
dB (C) clapping and 87 dB cheering, while
groups of six clapped at 91 dB and shouted at
95 dB (an increment of 6 dB represents a
doubling of sound pressure).
As might be expected, the more people
clapping or cheering together, the more intense
the noise and the more the sound pressure
produced. However, it did not grow in
proportion to the number of people: The
average sound pressure generated per person
decreased with increasing group size, F(3, 21)
= 41.5, p < .001. People averaged about 3.7
dynes/cm2
alone, 2.6 in pairs, 1.8 in foursomes,
and about l.S in groups of six (Figure
1). Put another way, two-person groups performed
at only 71% of the sum of their individual
capacity, four-person groups at
51%, and six-person groups at 40%. As in
pulling ropes, it appears that when it comes
to clapping and shouting out loud, many
hands do, in fact, make light the work.
People also produced about 60% more
sound power when they shouted than when
they clapped, F(\, 7) = 8.79, p < .01, presumably
reflecting physical capacity rather
than any psychological process. There was no
effect due to blocks of trials, indicating that
the subjects needed little or no practice and
that their performance was not deleteriously
affected by fatigue. In addition, there were
no interactions among the variables.
Discussion
The results provide a strong replication of
Ringelmann's original findings, using a com-
SOCIAL LOAFING 825
pletely different task and in a different historical
epoch and culture. At least when people
are making noise as part of a task imposed
by someone else, voices raised together
do not seem to be raised as much as voices
raised alone, and the sound of 12 hands clapping
is not even three times as intense as the
sound of 2.
Zajonc's (1965) elegant theory of social
facilitation suggests that people are aroused
by the mere presence of others and are thus
likely to work harder (though not necessarily
to achieve more) when together. Although
social facilitation theory might seem to predict
enhanced group performance on a simple
task like clapping or shouting, in the present
case it would not predict any effect due to
group size, since the number of people present
was always eight, six participants and two
experimenters. Evaluation apprehension theory
(Cottrell, 1972) would also not predict
any effect as long as it is assumed that coactors
and audience members are equally effective
in arousing performance anxiety.
Therefore, these theories are not inconsistent
with our position that an unrelated social
process is involved. The results of Experiment
1 also can be taken as support for Latane's
(1973) theory of social impact: The impact
that the experimenters have on an individual
seems to decrease as the number of coperformers
increases, leading to an apparent
drop in individual performance, a phenomenon
we call social loafing.
However, there is an alternative explanation
to these results. It may be, not that
people exert less effort in groups, but that
the group product suffers as a result of group
inefficiency. In his invaluable theoretical analysis
of group productivity, Steiner (1972)
suggests that the discrepancy between a
group's potential productivity (in this case
» times the average individual output) and
its actual productivity may be attributed to
faulty social process. In the case of Ringelmann's
rope pull, Steiner identifies one source
of process loss as inadequate social coordination.
As group size increases, the number of
"coordination links," and thus the possibility
of faulty coordination (pulling in different
directions at different times), also increases.
l -
Clapping
1 2 4 6
GROUP SIZE
Figure 1. Intensity of noise as a function of group
size and response mode, Experiment 1.
Steiner shows that for Ringelmann's original
data the decrement in obtained productivity
is exactly proportional to the number of coordination
links.
Ingham et al. (1974) designed an ingenious
experiment to determine whether the process
losses found in rope pulling were mainly due
to problems of coordinating individual efforts
and the physics of the task, or whether
they resulted from reductions in personal exertion
(what we have called social loafing).
First, they conducted a careful replication
of Ringelmann's original rope-pulling study
and found similar results—dyads pulled at
91% of the sum of their individual capacities,
trios at 82%, and groups of six at only 78%.
In a second experiment, Ingham et al.
cleverly arranged things so that only the individual's
perception of group size was varied.
Individuals were blindfolded and led to believe
that others were pulling with them, but
in fact, they always pulled alone. Under these
conditions, of course, there is no possibility
of loss due to faulty synchronization. Still
there was a substantial drop in output with
increases in perceived group size: Individ-
826 B. LATANfi, K. WILLIAMS, AND S. HARKINS
uals pulled at 90% of their alone rate when
they believed one other person was also pulling,
and at only 85% with two to six others
believed pulling. It appears that virtually all
of the performance decrement in rope pulling
observed by Ingharn et al. can be accounted
for in terms of reduced effort or social
loafing.
With respect to clapping and especially
shouting, however, there are several possible
sources of coordination loss that might have
operated in addition to social loafing: (a)
Sound cancellation will occur to the extent
that sound pressure waves interfere with each
other, (b) directional coordination losses will
occur to the extent that voices are projected
toward different locations, and (c) temporal
coordination losses will occur to the extent
that moment-to-moment individual variations
in intensity are not in synchrony. Our second
experiment was designed to assess the relative
effects of coordination loss and social
loafing in explaining the failure of group
cheering to be as intense as the sum of individual
noise outputs.
Experiment 2
Coordination Loss or Reduced Effort?
For Experiment 2 we arranged things so
that people could not hear each other shout;
participants were asked to wear headphones,
and during each trial a constant 90-dB recording
of six people shouting was played
over the earphones, ostensibly to reduce auditory
feedback and to signal each trial. As a
consequence, individuals could be led to believe
they were shouting in groups while actually
shouting alone. Ingham et al. (1974)
accomplished this through the use of "pseudosubjects,"
confederates who pretended to be
pulling with the participants but who in fact
did not pull any weight at all. That is an
expensive procedure—each of the 36 participants
tested by Ingham et al. required the
services of 5 pseudosubjects as well as the
experimenter. We were able to devise a procedure
whereby, on any given trial, one person
could be led to believe that he was performing
in a group, while the rest thought
he was performing alone. Thus, we were able
to test six real participants at one time.
Additionally, although we find the interpretation
offered by Ingham et al. plausible
and convincing, the results of their second
experiment are susceptible to an alternative
explanation. When participants were not pulling
the rope, they stood and watched the
pseudosubjects pull. This would lead people
accurately to believe that while they were
pulling the rope, idle participants would be
watching (Levinger, Note 1). Thus, as the
number of performers decreased, the size of
the audience increased. According to Cottrell's
evaluation apprehension hypothesis (1972),
the presence of an evaluative audience should
enhance performance for a simple, welllearned
task such as rope pulling, and, although
there is little supportive evidence, it
seems reasonable that the larger the audience,
the greater the enhancement (Martens &
Landers, 1969; Seta, Paulus, & Schkade,
1976). Thus, it is not clear whether there
was a reduced effort put forth by group members
because they believed other people were
pulling with them, or an increase in the effort
exerted by individuals because they believed
other people were watching them. In Experiment
2, therefore, we arranged to hold the
size of the audience constant, even while varying
the number of people working together.
Method
Six groups of six male undergraduate volunteers
heard the following instructions:
In our experiment today we are interested in
the effects of sensory feedback on the production
of sound in social groups. We will a'sk you to
produce sounds in groups of one, two, or six, and
we will record the sound output on the soundlevel
meter that you can see up here in front. Although
this is not a competition and you will not
learn your scores until the end of the experiment,
we would like you to make your sounds as loud
as possible. Since we are interested in sensory feedback,
we will ask you to wear blindfolds and earphones
and, as you will see, will arrange it so
that you will not be able to hear yourself as you
shout.
We realize it may seem strange to you to shout
as loud as you can, especially since other people
are around. Remember that the room is soundproofed
and that people outside the room will not
SOCIAL LOAFING 827
be able to hear you. In addition, because you will
be wearing blindfolds and headsets, the other participants
will not be able to hear you or to see
you. Please, therefore, feel free to let loose and
really shout. As I said, we are interested in how
loud you can shout, and there is no reason not to
do your best. Here's your chance to really give it a
try. Do you have any questions?
Once participants had donned their headsets and
blindfolds, they went through a series of 13 trials,
in which each person shouted four times in a group
of six, once in a group of two, and once by himself.
Before each trial they heard the identification letters
of those people who were to shout.
Interspersed with these trials were 12 trials, two
for each participant, in which the individual's headset
was switched to a separate track on the stereophonic
instruction tape. On these trials, everybody
else was told that only the focal person should
shout, but that individual was led to believe either
that one other person would shout with him or that
all six would shout.
Thus, each person shouted by himself, in actual
groups of two and six, and in pseudogroups of two
and six, with trials arranged so that each person
would have approximately equal rest periods between
the trials on which he performed. Each trial was
preceded by the specification of who was to perform.
The yells were coordinated by a tape-recorded voice
counting backwards from three, followed by a constant
90-dB 5-sec recording of the sound of six
people shouting. This background noise made it impossible
for performers to determine whether or
how loudly other people were shouting, or, for that
matter, to hear themselves shout. Each trial was
terminated by the sound of a bell. This sequence of
25 trials was repeated three times, for a total of 75
trials, in the course of which each subject shouted
2 4 times. ; . • : • •
As in Experiment 1, the data were transformed
into dynes/cma
and subjected to analyses of variance,
with the group as the unit of analysis and
each score based on the average output per person.
Two separate 3 X 3 analyses of variance with group
size (1,2,6) and trial block (1-3) were run, one on
the output of trials in which groups actually shouted
together, and one on the pseudogroup trials in which
only one person actually shouted.
Results
Overall, participants shouted with considerably
more intensity in Experiment 2 than
in Experiment 1, averaging 9.22 dynes/cm2
when shouting alone, as compared to 4.73
dynes/cm2
, t ( l 2 ) - 4.05, p < .01. There are
several plausible reasons for this difference.
The new rationale involving the effects of reduced
sensory feedback may have interested
GROUP SIZE
Figure 2. Intensity of sound produced per person
when cheering in actual or perceived groups of 1, 2,
and 6, as a result of reduced effort and faulty coordination
of group efforts, Experiment 2.
or challenged individuals to perform well.
The constant 90-dB background noise may
have led people to shout with more intensity,
just as someone listening to music through
headphones will often speak inappropriately
loudly, (the Lombard reflex). The performers
may have felt less embarrassed because the
room was soundproof and the others were
unable to see or hear them. Finally, through
eliminating the possibility of hearing each
other, individuals could no longer be influenced
by the output of the others, thereby
lifting the pressure of social conformity.
As in Experiment 1, as the number of actual
performers increase, the total sound output
also increased, but at a slower rate than
would be expected from the sum of the individual
outputs. Actual groups of two shouted
at only 66% of capacity, and groups of six
at 36%, F(2, 10) = 226, p < .001. The comparable
figures for Experiment 1 are 71% and
40%. These similarities between experiments
suggest that our procedural changes, even
though they made people unable to hear or
see each other, did not eliminate their feel-
828 B. LATANfi, K. WILLIAMS, AND S. HARKINS
ing of being in a group or reduce the amount
of incoordination or social loafing.
The line connecting the solid circles in
Figure 2 shows the decreased output per person
when actually performing in groups. The
dashed line along the top represents potential
productivity—the output to be expected if
there were no losses due to faulty coordination
or to social loafing. The striped area at
the bottom represents the obtained output
per person in actual groups. Output is obviously
lower than potential productivity, and
this decrease can be considered as representing
the sum of the losses due to incoordination
and to reduced individual effort.
In addition to shouting in actual groups,
individuals also performed in pseudogroups in
which they believed that others shouted with
them but in which they actually shouted
alone, thus preventing coordination loss from
affecting output. As shown in Figure 2, people
shouted with less intensity in pseudogroups
than when alone, F(2, 10) =37.0, p
< .0001. Thus, group size made a significant
difference even in pseudogroups in which coordination
loss is not a factor and only social
loafing can operate.
When performers believed one other person
was yelling, they shouted 82% as intensely as
when alone, and when they believed five
others to be yelling, they shouted 74% as
intensely. The stippled area defined at the
top of Figure 2 by the data from the pseudogroups
represents the amount of loss due to
social loafing. By subtraction, we can infer
that the white area of Figure 2 represents the
amount of loss due to faulty coordination.
Since the latter comprises about the same area
as the former, we can conclude that, for shouting,
half the performance loss decrement is
due to incoordination and half is due to social
loafing.
Discussion
Despite the methodological differences between
Experiments 1 and 2, both experiments
showed that there is a reduction in sound
pressure produced per person when people
make noise in groups compared to when
alone. People in Experiment 1 applauded and
cheered in full view of each other, with all
the excitement, embarrassment, and conformity
that goes along with such a situation. In
Experiment 2, no one could see or hear any
other person. Only the experimenters could
see the people perform. And finally, the rationale
changed drastically, from the experimenters'
interest in "judgments of how much
noise people make in social settings" to their
interest in "the effects of sensory feedback
on the production of sound in social groups."
Yet, despite differences in the task characteristics
and supposed purpose, the two studies
produced similar results. This points to the
robust nature of both the phenomenon and
the paradigm.
General Discussion
Noise Production as Group Performance
Although we do not usually think about it
that way, making noise can be hard work, in
both the physical and the psychological sense.
In the present case, the participants were
asked to produce sound pressure waves, either
by rapidly vibrating their laryngeal membranes
or by vigorously striking their hands
together. Although superficially similar in
consequence, this task should not be confused
with more normal outbreaks of shouting
and clapping that occur as spontaneous outbursts
of exuberant expressiveness. Our participants
shouted and clapped because we
asked them to, not because they wanted to.
This effortful and fatiguing task resulted in
sound pressure waves, which, although invisible,
can be easily and accurately measured
in physical units that are proportional to the
amount of work performed. The making of
noise is a useful task for the study of group
processes from the standpoint both of production
and of measurement—people are practiced
and skilled at making noise and can do
so without the help of expensive or cumbersome
apparatus, and acoustics and audio engineering
are sufficiently advanced to permit
sophisticated data collection. We seem to
have found a paradigm wherein people get
involved enough to try hard and become somewhat
enthusiastic, yet the task is still effortful
SOCIAL LOAFING 829
theenough so that they loaf when given
opportunity.
The Causes, oj Social Loafing
The present research shows that groups
can inhibit the productivity of individuals so
that people reduce their exertions when it
comes to shouting and clapping with others.
Why does this occur? We suggest three lines
of explanation, the first having to do with attribution
and equity, the second with submaximal
goal setting, and the third with the
lessening of the contingency between individual
inputs and outcomes.
1. Attribution and equity. It may be that
participants engaged in a faulty attribution
process, leading to an attempt to maintain an
equitable division of labor. There are at least
three aspects of the physics and psychophysics
of producing sound that could have led people
to believe that the other persons in their
group were not working as hard or effectively
as themselves. First, individuals judged their
own outputs to be louder than those of the
others, simply because they were closer to
the sound source. Second, even if everyone
worked to capacity, sound cancellation would
cause group outputs to seem much less than
the sum of their individual performances.
Finally, the perception of the amount of
sound produced in a group should be much
less than the actual amount—growing only
as the .67 power of the actual amount of
sound, according to Stevens's psychophysical
power law (1975).
These factors may have led individuals to
believe that the other participants were less
motivated or less skillful than themselves—
in short, were shirkers or incompetents. Thus,
differences in the perception of sound production
that were essentially the result of physical
and psychophysical processes may have
been mistakenly attributed to a lack of either
skill or motivation on the part of the others,
leading individuals to produce less sound in
groups because there is no reason to work
hard in aid of shirkers or those who are less
competent.
This process cannot explain the results of
Experiment 2, since the capacity to judge the
loudness of one's own output, much less that
of others, was severely impaired by the 90-dB
background masking noise used to signal the
trials. However, rather than "discovering" social
loafing while participating in the experiment,
the participants may have arrived with
the preexisting notion that people often do
not pull their own weight in groups. Thus,
despite being unable to hear or see one another,
lack of trust and the propensity to attribute
laziness or ineptitude to others could
have led people to work less hard themselves.
2. Submaximal goal setting. It may be
that despite our instructions, participants redefined
the task and adopted a goal, not of
maki: g as much nr as possible, but merely
of making enough je or of matching some
more or less well-denned standard. Individuals
would clearly expect it to be easier to
achieve this goal when others are helping, and
might work less hard as a consequence. This,
of course, would change the nature of noise
production from what Steiner (1972) would
term a maximizing task to an optimizing
task. A maximizing task makes success a function
of how much or how rapidly something
is accomplished. For an optimizing task, however,
success is a function of how closely the
individual or group approximates a predetermined
"best" or correct outcome. If participants
in our experiments perceived sound
production as an optimizing rather than a
maximizing task, they might feel the optimal
level of sound output could be reached more
easily in groups than alone, thereby allowing
them to exert less effort.
The participants in Experiment 2 could
hear neither themselves nor others and would
not be able to determine whether their output
was obnoxious or to develop a group standard
for an optimal level. Furthermore, in both
experiments, the experimenters reiterated
their request to yell "as loud as you can,
every time," over and over again. Before the
first trial they would ask the group how loud
they were supposed to yell. In unison, the
group would reply, "As loud as we can!" We
think it unlikely that participants perceived
the task to be anything other than maxi-
mizing.
3. Lessened contingency between input and
830 B. LATANfi, K. WILLIAMS, AND S. HARKINS
outcome. It may be that participants felt
that the contingency between their input and
the outcome was lessened when performing
in groups. Individuals could "hide in the
crowd" (Davis, 1969) and avoid the negative
consequences of slacking off, or they may
have felt "lost in the crowd" and unable to
obtain their fair share of the positive consequences
for working hard. Since individual
scores are unidentifiable when groups perform
together, people can receive neither precise
credit nor appropriate blame for their performance.
Only when performing alone can
individual outputs be exactly evaluated and
rewarded.
Let us assume that group members expect
approval or other reward proportional to the
total output of a group of n performers, but
that since individual efforts are indistinguishable,
the reward is psychologically divided
equally among the participants, each getting
l/n units of reward. Under these assumptions,
the average group, if it performed up
to capacity and suffered no process loss, could
expect to divide up n times the reward of
the average individual, resulting in each member's
getting n X I/ n, or «/», units of reward,
the same amount as an individual.
Although the total amount of reward may
be the same, the contingency on individual
output is not. Any given individual under
these assumptions will get back only one wth
of his own contribution to the group; the
rest will be shared by the others. Even though
he may also receive unearned one wth of
each other person's contribution, he will be
tempted, to the extent that his own performance
is costly or effortful, to become a "free
rider" (Olson, 1965). Thus, under these assumptions,
if his own performance cannot be
individually monitored, an individual's incentive
to perform should be proportional to
l/n.
Seligman (197S) has shown that animals
and people become lethargic and depressed
when confronted with tasks in which they
have little or no control over the outcomes.
Likewise, in our experiments, people may
have felt a loss of control over their fair
share of the rewards when they performed in
groups, leading them also to become, if not
lethargic and depressed, at least less enthusiastic
about making lots of noise.
Since people were asked to shout both alone
and in groups, they may have felt it smart to
save their strength in groups and to shout as
lustily as possible when scores were individually
identifiable, marshalling their energy
for the occasions when they could earn rewards.
This line of reasoning suggests that if
inputs were made identifiable and rewards
contingent on them, even when in groups, it
would be impossible for performers to get a
free ride and they would have an incentive to
work equally hard in groups of different sizes.
Social Loafing and Social Impact Theory
Each of these three lines of explanation
may be described in terms of Latane's (1973)
theory of social impact. If a person is the
target of social forces, increasing the number
of other persons also in the target group
should diminish the pressures on each individual
because the impact is divided among
the group members. In a group performance
situation in which pressures to work come
from outside the group and individual outputs
are not identifiable, this division of impact
should lead each individual to work less
hard. Thus, whether the subject is dividing
up the amount of work he thinks should be
performed or whether he is dividing up the
amount of reward he expects to earn with his
work, he should work less hard in groups.
The theory of social impact further stipulates
the form that the decrease in output
should follow. Just as perceptual judgments
of physical stimuli follow power functions
(Stevens, 1957), so also should judgments of
social stimuli, and the exponent of the psychosocial
power function should have an exponent
of less than one, resulting in a marginally decreasing
impact of additional people. Thus,
social impact theory suggests that the amount
of effort expended on group tasks should decrease
as an inverse power function of the
number of people in the group. This implication
cannot be tested in Experiment 1 or with
the actual groups of Experiment 2, inasmuch
as coordination loss is confounded with social
loafing. However, a power function with an
SOCIAL LOAFING 831
exponent of —.14 accounted for 93% of the
variance for the pseudogroups of Experiment
2. It appears that social impact theory provides
a good account of both the existence
and the magnitudeof social loafing.
The Transsituational and Transcidtural
Generality of Social Loafing
The present research demonstrates that
performance losses in groups occur with tasks
other than rope pulling and with people other
than prewar German workers. There are, in
addition, other instances of experimental research
that demonstrate similar cases of social
loafing. For example, Marriott (1949) and
Campbell (1952) have shown that factory
workers produce less per person in larger
groups than in smaller ones. Latane and Darley
(1970) have found that the likelihood
that a bystander will intervene in a situation
in which someone requires assistance is substantially
reduced by the addition of other
bystanders who share in the responsibility for
help. Wicker (1969) has found that the proportion
of members taking part in church
activities is lower in large than in small
churches, presumably because the responsibility
for taking part is more diffuse. Similarly,
Petty, Harkins, Williams, and Latane (1977)
found that people perceived themselves as
exerting less cognitive effort on evaluating
poems and editorials when they were among
groups of other unidentifiable evaluators than
when they alone were responsible for the task.
These experimental findings have demonstrated
that a clear potential exists in human
nature for social loafing. We suspect that the
effects of social loafing have far-reaching and
profound consequences both in our culture
and in other cultures. For example, on collective
farms (kolkhoz) in Russia, the peasants
"move all over huge areas, working one
field and one task one day, another field the
next, having no sense of responsibility and
no direct dependence on the results of their
labor" (Smith, 1976, p. 281). Each peasant
family is also allowed a private plot of up to
an acre in size that may be worked after the
responsibility to the collective is discharged.
The produce of these plots, for which the
peasants are individually responsible, may be
used as they see fit. Although these plots occupy
less than 1 % of the nation's agricultural
lands (about 26 million acres), they produce
21% of the total value of Soviet farm
output (about $32.5 billion worth) (Yemelyanov,
1975, cited in Smith, 1976, p. 266).
It is not, however, that the private sector is
so highly efficient; rather, it is that the efficiency
of the public sector is so low (Wadekin,
1973, p. 67).
However, before we become overly pessimistic
about the potential of collective effort,
we should consider the Israeli kibbutz, an
example that suggests that the effects of social
loafing can be circumvented. Despite the
fact that kibbutzim are often located in remote
and undeveloped areas on the periphery
of Israel to protect the borders and develop
these regions, these communes have been very
successful. For example, in dairying, 1963
yields per cow on the kibbutz were 21%
higher than for the rest of Israel's herds, and
in 1960 yields were 15% higher than in England.
In 1959, kibbutz chickens were producing
22% of the eggs with only 16% of
the chickens (Leon, 1969). The kibbutz and
the kolkhoz represent the range of possibilities
for collective effort, and comparisons
of these two types of collective enterprise may
suggest conditions under which per person
output would be greater in groups than in-
dividually.
Social Loafing as a Social Disease
Although some people still think science
should be value free, we must confess that we
think social loafing can be regarded as a
kind of social disease. It is a "disease" in
that it has negative consequences for individuals,
social institutions, and societies. Social
loafing results in a reduction in human
efficiency, which leads to lowered profits and
lowered benefits for all. It is "social" in that
it results from the presence or actions of other
people.
The "cure," however, is not to do away
with groups, because despite their inefficiency,
groups make possible the achievement of
many goals that individuals alone could not
832 B. LATANfi, K. WILLIAMS, AND S. HARKINS
possibly accomplish. Collective action is a
vital aspect of our lives: From time immemorial
it has made possible the construction of
monuments, but today it is necessary to the
provision of even our food and shelter. We
think the cure will come from finding ways
of channeling social forces so that the group
can serve as a means of intensifying individual
responsibility rather than diffusing it.
Reference Note
1. Levinger, G. Personal communication, June 1976.
References
Campbell, M. Group incentive payment schemes:
The effects of lack of understanding and group
size. Occupational Psychology, 1952,26, 15-21.
Cottrell, N. Social facilitation. In C. McClintock
(Ed.), Experimental social psychology. New York:
Holt, Rinehart & Winston,1972.
Dashiell, J. F. Experimental studies of the influence
of social situations on the behavior of individual
human adults. In C. Murchison (Ed.), A handbook
of social psychology. Worcester, Mass.: Clark
University Press, 1935.
Davis, J. H. Group performance. Reading, Mass.:
Addison-Wesley, 1969.
Ingham, A. G., Levinger, G., Graves, J., & Peckham,
V. The Ringelmann effect: Studies of group size
and group performance. Journal of Experimental
Social Psychology, 1974,10, 371-384.
Kbhler, O. Ueber den Gruppenwirkungsgrad der
menschlichen Korperarbeit und die Bedingung
optimaler Kollektivkroftreaktion. Industrielle Psychotechnik,
1927,4, 209-226.
Latane, B. A theory of social impact,. St. Louis, Mo.:
Psychonomic Society, 1973.
Latane, B., & Darley, J. M. The unresponsive bystander:
Why doesn't he help? New York: Appleton-Century-Crofts,
1970.
Leon, D. The kibbutz: A new way of life. London:
Pergamon Press, 1969.
Marriott, R. Size of working group and output. Occupational
Psychology, 1949, 23, 47-57.
Martens, R., & Landers, D. M. Coaction effects on
a muscular endurance task. Research Quarterly,
1969, 40, 733-737.
Moede, W. Die Richtlinien der Leistungs-Psychologie.
Industrielle Psychotechnik, 1927,4, 193-207.
Olson, M. The logic of collective action: Public goods
and the theory of groups. Cambridge, Mass.: Harvard
University Press, 1965.
Petty, R., Harkins, S., Williams, K., & LataniS, B.
The effects of group size on cognitive effort and
evaluation. Personality and Social Psychology Bulletin,
1977, 3, 579-582.
Seligman, M. Helplessness. San Francisco: Freeman,
1975.
Seta, J. J., Paulus, P. B., & Schkade, J. K. Effects of
group size and proximity under cooperative and
competitive conditions. Journal of Personality
and Social Psychology, 1976, 34, 47-53.
Smith, H. The Russians. New York: Ballantine
Books, 1976.
Steiner, I. D. Models for inferring relationships between
group size and potential group productivity.
Behavioral Science, 1966, 11, 273-283.
Steiner, I. D. Group process and productivity. New
York: Academic Press, 1972.
Stevens, S. S. On the psychological law. Psychological
Review, 1957,64, 153-181.
Stevens, S. S. A metric for the social consensus.
Science, 1966, 151, 530-541.
Stevens, S. S. Psychophysics: Introduction to its
perceptual, neural and social prospects. New York:
Wiley, 1975.
Wadekin, K. The private sector in Soviet agriculture,
Los Angeles: University of California Press, 1973.
Wicker, A. N. Size of church membership and members
support of church behavior settings. Journal
of Personality and Social Psychology, 1969, 13,
278-288.
Zajonc, R. B. Social facilitation. Science, 1965, 149,
269-274.
Zajonc, R. B. Social psychology: An experimental
approach. Belmont, Calif.: Brooks/Cole, 1966.
Received June 23, 1978 •