The Journal of Sor'uil l\<yrlnj!,ujy. S. I'. S. S. I. Biillri'm, 1939, 10, 271-299,
PATTERNS OF ACCiRKSSIVE BEHAVIOR IN EXPERIMENTALLY
CREATED "SOCLAL CLIMATES"
ChilJ IFfljiirif Ri-sciirili Sttiiioii, State Uni-vcisily of lo-u-ii
KURT LEWIN, RONALDLIPPITT, AND RALPH K. WHITE
A. PROBLEMS AND METHODS
The present report: is apreliminary summary on one phase of a series of experimentalstudies of group life which has as its aim a scientific approach to such questions as the following: What underlies such differing patterns of group behavior as rebellion against authority, persecution of a scapegoat, apathetic submissiveness toauthoritarian domination, or attack upon an outgroup? How many differences in subgroup structure, group stratification, and potencyof ego-centered and group-centered goals he utilized as criteria for predicting the social resultants of different group atmospheres?
Is not democratic group life more pleasant, but authoritarianism more efficient? These are the sorts of questions to which “opinionated” answers are many and varied today, and to which scientific answers, are, on that account, all the more necessary. An experimentalapproach to the phenomena of group life obviously raises manydifficulties of creation and scientific control, but the fruitfulness ofthe method seems to compensate for the added experimental problems.
In the first experiment Lippitt organized two clubs of l0-year-oldchildren, who engaged in the activity of theatrical mask-makingfor a period of three months. The same adult leader, changing hisphilosophy of leadership, led one club in an authoritarian mannerand the other club in accordance with democratic techniques, whiledetailed observations were made by four observers, This study,reported in detail elsewhere (6), suggested more hypotheses thananswers and led to a second and more extensive series of experimentsby White and Lippitt. Four new clubs of 10-year-old boys wereorganized, on a voluntary basis as before, the variety of club activitieswas extended, while four different adult leaders participated. Tothe variables of authoritarian and democratic procedure was addeda third, "laissez-faire" or group life without adult: participation.
Also the behavior of each club was studied in different "social climates." Every six weeks each group had a new leader with adifferent technique of leadership, each club having three leadersduring the course of the five months of the experimental series. Thedata on aggressive behavior summarized in this paper are drawnfrom both series of experiments.
Some of the techniques usedd for the equating of groups havebeen described previously (4). But will be summarized here withthe improvements in method of the second experiment. Before theclubs were organized the schoolroom group as a whole was studied.Using the sociometric technique developed by Moreno (8) the interpersonal relations of the children, in terms of rejections, friendships, and leadership, were ascertained. Teacher ratings on relevant items of social behavior (e.g., teasing, showing off, obedience, physicalenergy) were secured, and observations were made on the playground and in the schoolroom by the investigators. The school records supplied information on intellectual status, physical status,and socio-economic background. From the larger number of eager volunteers in each roomit was then possible to select from eachschoolroom two five-member clubs, which were carefully equatedon patterns of interpersonal relationships, intellectual, physical, andsocio-economic status, in addition to personality characteristics. The attempt was not to equate the boys within a particular club, but to ensure the same pattern in each group as a whole.
In spite of the methods described abo\ e to control by selectionsome of the more elusive social variables, it was essential to use anumber of experimental controls which would help to make theresults more clear-cut. First of all, to check on the "individuality" of the club as a whole, each group was studied in different socialatmospheres so that it could be compared with itself. A second question raised by the first experiment was that concerning the personality of the leader as a factor m the creating of social atmospheres. The second experiment, with four leaders, makes possible a comparisonof the authoritarianism and democracy of four different leaders,and the "laissez-faire" method of two different leaders. In twocases it is also possible to compare the same atmosphere, created bytwo different leaders with the same club.
One other type of control seemed very important, the nature of
the club activity, and the physical setting, losing the same clubrooms
(two clubs met at the same time in adjacent hut distinctly separate
areasof the same large room) seemed to answer the latter
problem, but the question of activity was more complex. The
following technique was developed: a list of activities "which were
of interest to all the children was assembled (e.g., mask-making,
TABLE 1
Autthoritarian
All determination
of policy by the
leader.
2. Techniques and activity
steps dictated
Liy the authority,
one at :i
time, so that future
steps were always
uncertain to a largtdegree.
3. The leader usually
dictated the
particular work task
and work companions
of each member.
Uemocrntic
1. All policies a matter
of group discussion
and decis
i o n , enoouragec!
and assistiitl by the
lender,
2. Ai-tivity porspectiv-
u gaincil during
first tUscussion
period. G e n e r a l
steps to group goal
s k e t c h e d , and
wliere technical advice
-was needed
the leader suggested
Iwo or three alternative
1) r o c e-
(lures from which
choice could be
mnde.
3. 'I'he members were
free to work with
w h o m e v e r they
chose, and the division
of tasks was
li-ft up to the
Compk-te freedotn
for group or individual
decision,
without any leader
participation.
2. Various materials
supplied by the
leader, who inad'j
it clear thai he
would supply inforinauon
w h e n
askt'd. He took tio
other part in work
discussions.
3. Ccimplufii nonpart
i c i p a l i i i i i liy leade;-.
4. The diHiiHiatoi- WM'^
"personal"' in his
praise and criticism
of the woi-k
of each member,
but remained aloof
from active group
participation except
when demonstrating.
He was friendly
nr impersonal
rather than openiy
hostile.
4. The leader was
"o b 3 e c t i V e" or
"f:ut-minded" i n
his praise anti criticism,
and tried to
be a regular group
member in spirit
without dointi too
much of ihe woj'k.
V e r y infrequent
cotiiments on member
activities nnless
i|uestioned, and
no attempt to partii'ipati-
or interfere
\vi[h the course
of events.
274 JO^_Mi^^•^^, OF SOCLAI. PS^CMOI.OG-,'
mural painting, soap carviny;, model airjilane construction, etc.).
!^•'Ieeti^f;; iir?;t, ir-i chrono!oy;ical time, the democratic g:roups used
these possibilities as liic basis foi" liiscussioti and \i)ted lijion their
club activity. Ihe authoritarian leaders were then ready, as their
clubs met, to launch the same activity without choice by tlie members.
The "lai.<xrz-fai>-i'" groups -were acquainted \\'\t\'i the \'ariety of materials
-vvhicli were available, hut the\- \\-ere not otherwise influenced
in their choice of activity; in their case, consequently, the activity
factor could not be completely controlled.
The contrasting methods of the leaders m creatuii^ the three t}'pes
of group atmosphere may he briefly sLinnnari7,i:(.i as m 'I able I.
It should be; clear that due to the voluntary nature of the group
participation, and the cooperation of the parents and school systems,
no radically autocratic methods (e.g., use of threats, instilling fear,
etc.) were used. Fairly conj^enia! extra-cluh relatiiinships "were
maintained with each member by the leadei.
The kinds of data collected during the course of the experiments
may be classed roughly as: (;/) pre-cluli data, described above in
reh-ition to the problem of equating the groups; (/>) ohservalions of
behavior\n the experimental situation; and (c) extra-club information.
Observations of club behavior consisted of:
{iij. A quantitativf running accoLint of the social interaci.
iiins of the live children anti (he leader, in terms of symbols
lor dii-ecti\'e, compHiini, and dhjc-rtive (fact-minded) appro
iirhi,-';. and rL'^pdnsc's, including a category of purposeful rcfiibiai
to respond to a social apprDach.
(/;), .A iniuute hy minute group structure analysis giving a
record of: acMvity subgroupings, the activity goal of each subgroup
was initiated by the leader or spontaneou^ily formed by
the children, and ratings on degree of unity of each sub-
{c). An iuiLM-iiretiye running account of signiticant member
nriion?, and chnngL"; in dynamics of Ihe grotip as a whole,
(il). Continuous stenographic records of all conversation,
{':). An interpretive running account of inter-club relationi^
liips.
(/). An "iinprcs^ioni-^tir" writf-up b\' the leader as to u-hat
he saw and felt from -vvilhin the group atmosphtre during
eacli tnectln^.
((/). Cotntnents by guest observers.
(11). Movie records of se\'eral liegiueiirs of club life.
KURT J.HWIN, RON M.D Lil'FITT, AND RALPH K. WHti L: 2 I :>
All of these observations {except /, g, and /() were synchronized
at minute intervals so that side by side they furnish a rather complete
cross sectional picture of the ongoing life of the group. The major
purpose of this experiment in methodology of obscr\^ution was to
record a> fully and with as miicli nisight as possible the total
behavior of the group, a distinct break away from the usual procedure
of recortlnig onl\ cert;:in pre-determined symptoms of behavior.
The second aim \\as to ascertain whether data collected
by this method could be fruitfully anal\'zcd from bolli a sociological
and psychological point of view (5).
Extra-club information is of the following t\'p('s:
{a). Interviews ivith each child by a friendly "non-club"
person during cacli transition period (from one kind ui group
atmosphere and iiMdcr lo aiiiiilier) and ni the tnd of the
experimcnr, coiuei^nin.tr such itfins as roiuparison ol present
chib leader wirh |)ie\'irnis ones, with rhe leacher, and wilh
parents: opinions on ciub accivilius; hnw ihv cliil) L:ouki be run
betttr; who were the best ajid ]i(iore^t club nieiiiliers; what an
idea! club leader would be lilce, etc.
(/'). Interviews with the parents by the investij^ntni-s, concentrating
on kinds of discipline used in the home, status of the
child in the fiimily group (relations with siblings, etc.), personality
ratings on the same scale used by the teache.'rs, discussion
of chilli's attitude toward ihc elub, school, and other
gronp activities.
{c). Talks witli the teachers concerning the transfer to the
schoolroom, of behavior patterns acquired in the club.
{d). Administration of a Rorschach test to each club member.
{e). C^o rivers at ions with the children during fsvo sLimrner
hikes arranged after the experiment was over.
These data were gathered -with a view to correlating the individtinl
pattern of beba\-ior in tbe club situation witb the types of
grai.i|) membei-shiii wJiich existed outside the experiment, and with
tbe more or less stable individual personality' structure. The individual
differences in "social plasticity" seem to he rather striking.
Two other points of experimental technique seem of interest.
The lirst concerns the introduction of observers into the club situation.
In Lippitt's first experiment it was found that four observers
grouped around a table in a physically separated part of the club
room attracted virtually no atiention if it -was explained at the first
276 JOUR.X.M. or soci,-\;, PS~I'CIJOI.OGY
meetmg that "tHo.se are some people interested in learnm<j bow a
mask-making- club goes; tbey bave plenty to do so they won't bother
us and \-\e won't bother them." In the second experiment tbe
arrangement was c^'en more advantageous and seemed to make for
equally unselfconscious behavior on the part of the cUihs. In this
set-up the lighting arrangement was such that tbe observers were
grouped behind a low burlap wall in a darkh' shaded area, and
seemed "not to exist at all" as far as tbe children and leaders were
concerned.
Tbe second point of interest is the development of a miniber of
"group test" Situations, whicb aided greatly in i^etting at tbe actual
social dynamics of a given group atmosphere. One test used systematically
was loi' the leader to leave the room on business during
the course tjf the cluh meeting, s(-j that the "social pressure" factor
could be analyzed more realisticalh'. Another practice was for
the leader to arrive a iew n-iinutes late so that the obser\-ers could
record tbe individual and "atmospheric" difterences in spontaneous
work initiation and work perspective. A third fruitful technique
i\"as that of having a strar-igcr (a .Lrraduate student who played the
role of a janitor or electrician) enter the club situation and criticize
tbe group's work efforts. A rather ilramatic i)icture of tbe results
of this type of situatior-i may be seen in figures 5 and 6. Further
vanations (jf -vich evjienmental luampulations are heino; utilized
in a research now in progress.
B, RESULTS
I'he ai-ialysis of tbe results froiu the second experiment is now
proceeding in various directions, following two main trends: (a) interpretation
of socio!oq:ical or "group-centered" data; (h) interpretation
of psychological (;r "individual-centered'' data. The socioloszical
approach includes such analyses as dilYcrences in volun-ie of social
interaction related to social atmosphere, nature of club acti\'ity, outgroup
relationship, dift'eiences in pattern of iiiteraetion related lo
outgroup and ingroup orientation, atmosphere differences in leadt:rgroup
relationship, effect upon group structure pattern of social
atmosphere and types of activity, poup differences in language behavior,
etc. The p,s\chological approacii includes such anahse.-; ;]s
relation of Iiome background to pattern of club beha\-ioi-, range
of \-ariatioii of member behavior in different t\pes of social atmosKLRT
I.I'WIN, RON'ALD LIPI'TTT, .AXD RAi.lMi K. 277
phcif, |i:itt(nns of iiuli\'idn;t[ reaction t:o atmospliere transitions in
relation tu ca>e Iii>toi-\ data, coirclation between position in j^roup
stratilication and pattern of social action, etc. In this paper will
he presented onl\- certain data from the partialK' completed .L'eneral
analysis which arc reh-vant to the ilynamics of individLial and .s^roup
aggression.
We might first recall one or two of the most striking residts of
the iirst experitiient (6). As tiie club meetings progressed the
authoritiirian clvdi members developed a pattern of ai^^ressive dotnination
to\\ard one another, and their relation to the leader was
one of suhmission or of persi^^tent demands for attention. 'I'hc
interactions in the democratic club \\ ere tnore spontaneous, tnore
fact-minded, and triendlv. Rehitions to the leader were free and
on an "equality hasis." Comparing the t"\vo groups on the one
Item of o\'ert hostility the authoritarian group was surprisingh inure
aggressive, the ratio being 40 to I. Comparini; a constellation of
"ego-in\'olved" types of language behavior (e.g., hostile, resistant,
demands for attetUion, hostile criticism, expression of competition}
•\vith a group of objective or "iionemotive"' beha\ lors, it was found
that in tbe authoritarian group 7.^ pei" cent of the aiialyzed language
1 [a.
— i
-/
/
Vv /
«o
iK
SO
JB
G
O
o
•
_J
—
-
1
FK;UKK 1
CI: ay SCAPECO.ATS IN* .^K" ALTocn.\'rH:
(Lippirr, 1937)
The curves (which indicate the amount of aggression directet! against
each iiulividual) shoiv a much loner K^-'nei'al level of dominating behiivior
in the dcinocratic (D) than In the autocratic (A) groiiii. Twice during the
nieetinfis of the atithoritarian cluh the aggression of four members was
focused tipon the fifth {a niul / ' ) . In Imth teases the scapegoat dioii[)ed oiit
of the group imaiediatcly or soon afterwards.
278 JOURNAL OF SOCLIL I'SI'CJ iOLOGY
behavior was of the "ego-in\ olved" type as compared to 31 per cent
in the democratic cltih. Into the- ohjective category ^\ent 69 per
cent of the helia\'ior of the democratic group as compared to J7
per cent of the language activities oi the authoritarian group.
A second type ot data related to the dynamics of aggression
as It existe(i m tbe first experiment may be seen in 1"lgure 1. Twice
during the course of the meetings of the authoritarian club the
situation shiited from one of mutual aggres-;ion hc't::'een all member?
to one of concentrated aggression to'i\"ard one lnemher by the other
four. In both cases the lowered status of n. scapegoat position w;is
so acutely unpleasant that the member left the group, rationali/.ing
his break from the cluh by sucli remarks as, "Tbe doctor says my
e\cs are so had I'll ba\e to pla\' outiio(u"s in the sunshine instead
of coining to cluh meetings." Interestingly enough the two memhers
who were singled out for persecution had been rnted by the teachers
a,-' the t\\ o le:idcrs in tbe group, one of rhem scoring second m
popularity hy the sociometric technique, as ^yQ\\ as being physically the
strongest. After the emergence of hoth scapegoats, there was a
ratber brief rise m friendly cooperatu'e behavior between thp other
members of the group.
In the second experiment (see previous discussion, p. 4) there
were five democratic, five autocratic, and ti\o "hiissez-faire" atmospheres.
The fact tfiat the leaders were successful in modifying tlieir
behavior to correspond to these three philo-ophies of leadership is
clear on the hrtsis of several quantitati\'e indices. For instance,
the ratio of "directive'' to "compliant" beha\-ior on the part of the
autocratic leaders was 63 to 1 ; on the part of the democratic leaders
it was 1.1 to L The total amount of leader participation "was
less than half as great in "laisscz-i<i'rc" ;is in either autocracy or
democracy.
The data on aggression a\erages \n these three atmospheres are
summarized in Figures 2, 3, and 4. All of them indicate average
amounts of aggression per 50-minnte, five-member cluh meeting.
They re|)rcscnt b(_*ha\"ior records, as reeorded by the interaction
ohservLM", and include ail social actions, both verbal and physical,
which he designated as "hostile"' or "joking hostile.'' Figure 2 sho-ivs
especially the bi modal character of the asisression averages in
autocracy; four of the h\v_ autocracies bad an extremely \Q\\ ]e\"el
of aggression, and the lifth had an extremely bigh one. For comKURT
I.i;\\iN, RONALD LH'ITii, AND RALPH K, \\ 279
FI&2-AGGRESSI0N IN SIX AUTOCRATIC CLUB-PERIODS
ffl -"^
IN CL
10
O
lO
^ 3D