BasumcBosoI Ba*oune, A., Ross,D. & Ross, s.A. (1961).
Transmrssion ofaggressionthrough imitation of aggressive models.I
ournalof Abnor malandSocial p sy cholog A, 6.3, 575
_g2.INrRonucrroxThis studylooks at howaggressivebehaviourdevelopsin
children. It hasattracted a lot of attentionfrom a numberof
academicdisciplines andis still quoted in many textsdespite its
age. There are two social issuesthat the studyaddresses.First, is
aggression an innatefeature of ourbehaviour?Andto look at qne
particular aspectof this issue, can wesaythat
maleaggressiontowardswomenis a feature of ,natural,malebehaviouror
is it learnt? Thesequestions have a bearingon how wedevelop social
policies to dealwith aggressivebehaviour.The secondissue,
whichfollowson from thefirst,is if aggressionis learntthenhowis it
learnt?Banduras approachis an extensionof behaviourisizr and
basicallyseespeople as being mouldedby their life experiences. It
looks at howwe are affectedby the rewardsand punishments that we
experienceevery day Bandurais a leadingfigurein
socialLearningTheorywhichattempts to extend the concepts used in
operant andclassical condi_tioning to explaincomplexhuman social
behaviour. Keyconceptsinthisapproach are reinforcemenf and
imitation.THE SrunyIn this study,Bandurasetout to demonstrate that
if childrenare passivewitnessesto an aggressivedisplayby an adult,
they will imitate thisaggressive behaviourwhengiven the
opportunity. MorespecificallSqthe study wasguided by the
followingpredictions:. . . subjectsexposed to aggressivemodelswill
reproduceaggressiveacts resemblingthoseof themodels. . .. , . the
observationof subdued non-aggressive modelswill haveageneralized
inhibitingeffecton the subjects subsequentbeha_viour. . .MORAL
DEVELOPMENT 247. . . subjectswill imitate the behaviour of a
same-sex model to agreater degreethan a modelof the oppositesex. .
.. . . boys will be more predisposed than
girlstowardsimitatineaggression...(p.s7s)SubjectsThirty-sixboysand
36 girls agedbetween37 to 69 monthsweretested.The mean age was 52
months.They usedone male adult and onefemale adult to act as role
models.DesignThe study had three major conditions;a controlgroup, a
group exposedto an aggressivemodel, and a groupexposedto a passive
model. Thechildren whowere exposedto the adult models were further
sub-divided by their gender, and by the gender of the modelthey
wereexposedto. In otherwordsthere werethree
independentvariables.Asummary of the groups is shown in Table
13.2.This is quite a complicateddesignthat appears to cover a lot
ofdifferentpossibilities. However,the numberof children in each
group isquite small,and the results could be distortedif one group
containedafew childrenwhoare normally
quiteaggressive.Theresearchers
trieclTable13.2BanduraseishtexperimentalgroupsControlgroup - 24
subjectsEightexperimentalgroups (each with 6
subjects)rAggressivemodelcondition - 24 subjectsrNon-aggressive
model condition - 24subiectsAggressivemodel condition., 6 boyswith6
boys with6 girts with6 girtswithi.Same-sexopposite sexsame-sex
oppositesex.. modelmodelmodel modelNon-aggressive mod el control6
boyswith6 boys with6 girlswith 6 girls withsame-sex opposite
sexsame_sex oppositesexmodel modelmodel rnodelto reducethis
problemby pre-testing thechildrenand assessing
theiraggressiveness.They observedthe children in the nursery and
iudgedtheir aggressivebehaviouron four 5-pointrating
scales.rh.-.atingscaleswere:(a) physicalaggression;(b) verbal
aggression;(c) aggressiontowards inanimate objects;(d) aggressive
inhibition.A compositescorefor eachchildwasobtainedby adding the
results ofthe four ratings. It was then possible to matchthe
children in eachgroup so that they hadsimilarlevelsof aggressionin
theireverydaybehaviour.Theobservers werethe experimenter(female), u
,r.,rr".uschoolteacher(female), andthe model for maleaggression.The
studyreports thatthe first twoobservers'were well acquaintedwith
thechii-1r/oren (p. )/b).A disadvantage of usingrating scalesin
this way is that differentobserverssee differentthingswhen they
view the same event.Thismight mean that the ratings will vary from
oneobserver to another.Tocheck the inter-raterreliabilityof the
observations, 51 of the childrenwere ratedby two
observersworkingindependently andtheir ratingswere compared.
Thehigh correlationthat was achieved(r:.g9)showed
theseobservationsto be highlyreliable,suggesting tirat
theobserverswerein closeagreementaboutthe behavlouiof the
children.ProcedureThe childrenweretestedindividually.In stageone
they weretaken tothe experimentalroom which was set out for play.
one cornerwasarrangedas the child's play area, where therewas a
tableand chair.potatoprints and picture stickers, which were all
selectedas havinghigh interestfor these children. The adult
modelwas escortedto theopposite corner wheretherewasa smalltable,
chair,tinker toy malletand Bobo (a flve-footinflatabledoll).The
experimenterthen left theroom.In the non-aggressive condition,the
model assembled the tinkertoys in a quiet, subdued manner,ignoring
Bobo.In the aggressive condi-tionthe modelstarted to assemblethe
tinker toys, but after one minuteturned to Boboandwasaggressive to
the doll in a stylizedand distinc-tive way The aggression 'was both
physical (for example ,raised theMORAL DEVELOPMENT 249Bobo doll,
pickedup the mallet and struck the doll on the head',p.576),and
verbal (for example,'Pow!',and'Sock him in the noseip.576). After10
minutes the experimenter returnedand took thechild to
anothergamesroom.In stagetwo, the child was subjected
to'mildaggressionarousall Thechild wastakento a room with
attractivetoys, but afterstartingto playwith them the childwas told
that these weretheexperimenter's verybest toysand shehaddecided to
reserve them for the other children.Thenthe child was taken to the
nextroomfor stagethreeof thestudy The experimenterstayedin the
roombtherwise a number of chil-dren wouldeither refuse to
remainalone,or would leave beforetermi-nationof the sessionlIn this
room therewas a variety of toys, bothnon-aggressive(three bears,
crayons andso forth)andaggressivetoys(for example, a
malletpegboard,dartguns,nnd a three-footBobo).Thechild was kept in
this room for 2o minutes, and their behaviourwasobservedby judges
througha one-way mirror.observationsweremadeat
five-secondintervalsgiving 240 responseunits for eachchild.The
observersrecordedthree measuresof imitationin which theylooked for
responsesfrom the child that werevery similarto the displavby the
adult model:(1) imitativefor physicalaggression;(2)
imitativeverbalaggression;(3) imitativenon-aggressiveverbal
responses.Theyalsolooked at twotypesof behaviourthat
wereincompleteimita-tions of the adultmodel:(1) mallet
aggression;(2) sits on Bobo.In addition, they recordedthree typesof
aggressivebehaviourthat werenot imitationsof the adultmodel:(i)
punchesBobo;(2) non-imitativephysicaland verbal aggression;(3)
aggressive gun play.By looking at the resultswe can considerwhich
childrenimitatedthemodels,whichmodelsthey imitated,and
r,vhetherthey showeda gen-eralincrease in aggressivebehaviourrather
than a speci{icimitation ofthe adult behaviours.REsulrsThe results
aresulrlmarisedinTable13.3.They showthat:othe children who saw
theaggressrvemodelmade more aggressiveacts thanthe children whosaw
the non-aggressivemodel;oboys made more aggressiveacts
thangirls;.the boys in the aggressiveconditionsshowedmore
aggressionif themodelwas malethanif the model',vasfemale;ethe girls
in the aggressiveconditionsalso showed more physicalaggressionif
the model was malebut moreverbalaggression if themodelwas
female;othe exceptionto this generalpatternwas the observationof
howoften they punchedBobo, and in this casethe effects of
genderwerereversed.Table13.3 Mean'aggiessionscores for
experimentaland controlsubjectsExperimental
groupsAggresslveNon-aggresslveFemale MalemodelFemale
MalemodelmodelmodelControlgroupResponse categorYlmitative
physicalaggressionFemalesubjects5.5Male
subjects1-2.42.5o.27.225.84at.z2.Oo.0_L-3lmitativeverbalaggressionFemalesubjectsI3.7Male
subjects4.30.0o.0o.3r.7L5.-L13.5o.56.70.5LB.777.2L8.715.528.85.8,13.O2.OL2.7o.71".7MalletaggressionFemalesubjectsMale
subjectsPunches BoboFemalesublectsMale subjects6.316.518.911.9
r+.41L7.7L5.7Non-imitativeaggressionFemalesubjectsMalesubjects7.2lo.r8.436.7z)-.JLO.t1A6.1lL+
.oAggressive gun playFemalesubjectsMale
subjects2.68.94.515.91.87.32.5LO. i3.7!4.3Source: Bandura,Ross&
Ross (1961)MORALDEVELOPMENT251DrscusstoxOne of the
issuescommentedon by Bandura, Rossand Rossis the affectthat the
gender of the model had on the children.Theynoted
thattheaggressionof the female model had a confusingeffecton them.
Forexampleoneof the childrensaid,'Who is thatlady? Thats not the
wayfor a lady to behave.Ladies are supposedto act like ladies. . 1
(p.581),and another childsaid,'Youshouldhaveseen what thatgirl did
inthere.She was just actinglike a man. I neversaw a girl act like
thatbefore.She waspunchingand fightingbut no swearing'(p.581).On
theother hand,the aggressivebehaviourof the male model
fittedmorecomfortably into a culturalstereotypeof appropriate
behaviour. Forexample,oneboy said,Al'sa good socker,he beat up
Bobo. I wantto socklike Al' (p.581), and one of the girls
said,'Thatman is a strongfighter,he punchedand punchedand he could
hit Bobo rightdownto the floorand if Bobo got up he said, "Punch
your nose".Hes a goodfighterlikeDaddy' (p.581).If we lqok back at
thequestionswe raised in thebackgroundsectionof thissummary,then
whatcan we learn fromthe study?First,isaggressioninnate?Like all
examples of the nature-nurturedebate,itis very hard to get
clearevidenceoneway or the other,This studyshowsthat
aggressivebehaviour can be learnt,but it doesnot offerany
evi-denceon the questionof whethersomefeatures of aggressionare
alsoinnate.On the issue of maleviolence,it is rvorthnotingthat
thechil-dren in this study already had an expectation that men will
behavemoreaggressivelythan women.Thiswasshownby the
childrenscomments.The secondquestion rvas how is aggression
learnt?Bandurabelievesthat we can learnby being witnessesto the
behaviour ofothers, and his studyoffers some supportfor this idea.
if this is so, thenit would suggest that the regular viewing of
violentbehaviouron televi-sionprogrammeswouldencouragethelearning
of violentbehaviourin the viewer. A later variationof the
experiment(Bandura, Ross&Ross,1963)showed the children the
violent behaviouron a videoratherthan in reallife,and he
foundtheywerestill likely to imitatethe aggres-sion
behaviourtorvardsthe Bobodoll.Thereare,however,a numberof
reasonswh1'weshould be cautiousabout makingtoo many
connectionsbetwcen this studyandthe every-day experienceof
children.F'orexample,we harreno evidenceaboutany long-termeffects
of the study,anci also, it is very uncommonforchildrento be in a
situation wheretheyof their experience will be with peopleopinions
on whateveris going on.ouEs?Iol{s1Howis aggressionmeasuredin this
study?2Howelse couldit be meaSured?3Whatare the threeindependent
variables?4What ethical guidelines doesBandura appearto
break?SuggestedAnswers start at P'455are alonewith strangers.
Muchthey know who will give their