You are on page 1of 8

LowerConfidenceDrivesDesiretoCollaborate

NaomiMusgrave
KarthikPalaniappan
StevenVilwock

Evolution&HumanBehavior
SectionA
17March2015

Abstract

Anindividualsperceptionoftheirownknowledgelevelhasadirectimpactontheirchoicesand
behaviors.Wetheorizethisappliesspecificallytotheirwillingnesstocollaborateinordertoovercomea
challenge.Specifically,individualswithlowerconfidenceintheirabilitiesaremorelikelytocollaborate
withothers.Totestthehypothesis,weaskedparticipantstheirconfidenceintheirAmericanhistory
knowledge.WegavethemtheoptionofcollaboratingonanAmericanhistoryquestionforarewardif
successful,oransweringitbythemselvesforalargerrewarduponsuccess.Wefoundthatindividuals
markingthemselvesaslesssureoftheirknowledgeweremorelikelytocollaborate,andthattheytended
tobecorrectmoreoftenthanothergroups.Thesefindingsindicatethatpeoplewhoareunsureoftheir
abilitiesaremoresuccessfuliftheyfollowtheirselfevaluationandworkwithothers.

Previousresearchshowsthatwillingnesstocollaborateisrelatedtoneedforadditional
knowledge.
Gran
Melinfoundthatresearchersworkingtogetherontechnicalpaperstendtocollaborate
tosupplementtheirownabilities(Melin2000).Asimilarpatternemergedinthephasesofcompany
interactionsduringapartnership,withtheinitialpartnerselectionphasegearedtowardsevaluatingthe
othercompanysabilities(Emden2006).Aseparatefieldofresearchinspectstheinfluenceofconfidence
onperformance.Forinstance,astudentwithlowselfefficacyislesslikelytotacklechallengingtasksand
mayhavelowerinitiative(Seifert2004).Weproposethatanindividualsconfidenceintheirability
impactstheirdesiretocollaborateonachallengingtask.Ifstudentswithlowerconfidenceintheirskills
haveastrongerdesiretocollaborate,theywillfavorworkinginateam.Studentswhoseethemselvesas
lackingcompetencywillseektosupplementtheirabilitieswiththeskillsofothers,inordertoincrease
theirchancesofgainingaccesstoresources.
Wepresentedaquestionnaireto59UniversityofWashingtoncomputersciencestudentsand19
studentsstudyingintheOdegaardLibrary.Thefrontofthequestionnaireaskstheparticipanttoevaluate
herconfidencelevelinherU.S.historyknowledge.ThebackposesanAPU.S.historyquestion(Figure
1).WeselectedaquestionthatstudentsansweredcorrectlyabouthalfofthetimeontheAPexam,togive
participantsamoreevenchanceofgettingthecorrectanswer.Participantscouldchoosetowork
independentlyandpossiblywintwocookies,orattempttowinonecookiebyusingapreviousresponseto
helpthemanswerthequestion.Theresponsewouldbeselectedfromapilethatmimickedthedistribution
ofanswersontheAPexam.
Theuseofcookiesasamotivatorsimulatesstrivingforresourcesaspartofa
grouporalone.
Weexpectthatamajorityofindividualsreportinglowconfidencewillchooseto
collaborate,withasmallproportionofhighconfidenceindividualscollaborating(Figure2.b).Ifour
hypothesisisincorrect,thelowconfidencegroupwouldbelesswillingtocollaborateandshareresources
(Figure3.b).Duetotheirlowconfidenceintheirknowledge,thisgroupwouldfeelmorevulnerableand
seektoacquiremoreresources.Alternately,theremaybenocorrelationbetweenconfidenceanddesireto
collaborate.

Afteranalyzingtheresults,afewpatternsemerged.Aroundhalfmarkedthemselvesashaving
averageconfidence,withaminorityselectingaboveaverage.(Figure4.a).Thisdistributionwas
consistentacrossbothgroups(Figure4.b).Withinthelowconfidencegroup,44%decidedtocollaborate,
andhalfasmanyintheaverageconfidencegroup.Nonewithaboveaverageconfidencechoseto
collaborate(Figure5.a,5.b).Acrosstheboard,eachconfidencegroupwascorrectaround70%ofthe
time,witha10%increasefromlowconfidencenoncollaboratorstohighconfidence(Figure6.a,Figure
7.a).Thediscrepancyamongnoncomputersciencestudentsismostlikelyduetotheextremelysmall
samplesize,andisnotstatisticallysignificant(Figure6.b,7.b).Thehighestproportionofcorrectanswers
was89%,fromthelowconfidencecollaboratorsincomputerscience(Figure7.b).
Theseresultsalignwithouroriginalprediction.AsshowninFigure5.a,thehighestproportionof
collaboratorswereinthelowconfidencegroup.Thisisagoodchoiceforthisgrouptomake,sincethey
enduphavingthehighestproportionofcorrectanswers.However,highconfidencenoncollaborators
alsoperformedwell,showingthattheirdecisiontoworkalonepaidoffwithaccesstomoreresources
thantheywouldhaveotherwise.Theseresultshaveafewflaws,however.Thesampleofprimarily
computersciencestudentsisnotrepresentativeoflargerpopulations.Thedifferencebetweenthereward
ofoneandtwocookiesisveryslight,andmanypeoplewerewillingtorisknotgainingtheextracookie
fortheconvenienceofworkingalone.Oursimulationofcollaborationbyallowingpeopletopeekat
previousresponsesdoesnotcarryanyofthebenefitsoffacetofacecollaboration.Finally,thestudywas
conductedverypublicly,withrespondentsandsurveyorsinteractingconstantly.Participantshadno
privacywhenfillingoutthequestionnaire,whichmayhaveimpactedtheirselfreportedconfidence
levels.Inordertoironouttheseflaws,afollowupstudyshouldallowcollaborationbetweenparticipant
pairsandoperateonalargerpopulationmoreevenlydistributedacrossdifferentmajors.Additionally,
participantsshouldberandomlyselected,andcompletethefrontofthequestionnaireinprivacy.Ifour
initialresultsarestillsupported,thenwewouldseethesamepatternsariseaboutconfidenceand
collaborationdistributions.

FiguresandSupplements
HowwouldyourateyourUShistoryknowledgecomparedtotheaverageUWstudent?
[]BelowtheaverageUWstudent
[]AboutthesameastheaverageUWstudent
[]AbovetheaverageUWstudent
StudentsstagedasitininGreensboro,NorthCarolina,in1960toprotest:
a.

povertyintheSouth

b.

cutbacksinstudentaid

c.

segregationofpublicfacilities

d.

thewarinVietnam

e.

theoppressionofwomen

Figure1:Questionnairepresentedtoparticipants.Thefirstquestionwasonthefront,andthesecondon
theback.Aftertheparticipantcompletedthefirstquestion,theyweregivenanoptiontocollaborateby
lookingatapreviousrespondentsanswer,orworkbythemselves.Iftheychoosetoanswerthehistory
questionbythemselvesandwerecorrect,theywouldgettwocookies.Iftheycollaboratedandwere
correct,theywouldgetonecookie.Ifincorrect,theywouldnotgetanycookies.

Figure2.a:Predictedconfidencedistribution.H
alfwouldmarkthemselvesashavinglowconfidence,
sinceoursamplewasprimarilyComputerSciencestudents.Weexpectedthispopulationwouldhaveless
recentexposuretoU.S.historytopics,sowouldbelesssureoftheirknowledgelevel.
Figure2.b:Predictedcollaborationdistribution.WithlowerlevelsofselfconfidenceintheirU.S.history
knowledge,higherproportionsofthatgroupwouldwanttocollaborateonaresponse.


Figure3.a:Confidencedistribution,ifthehypothesisisunsupported.Agreaterproportionwouldmark
themselvesashavingaverageorhigherconfidence.
Figure3.b:Collaborationdistribution,ifthehypothesisisunsupported.Thelowandhighconfidence
groupswouldbelesslikelytocollaborate.

Figure4:Participantsconfidenceleveldistribution,takingtheentiresample(Figure4.a)andbreaking
thesamplingintocomputersciencestudentsandstudentsworkinginOdegaard(Figure4.b).

Figure5:Proportionofeachconfidencelevelwillingtocollaborate,withtheentiresample(Figure5.a)
andparticipantbreakdown(Figure5.b).

Figure6:Correlationsbetweenconfidencelevelandansweringcorrect.Correctnessisrelatively
consistentacrossallconfidencelevels,asshowninFigure6.a.ParticipantbreakdowninFigure6.b.

Figure7.a:Correctnessratesofcollaboratorswithineachconfidencegroup.Thetwonumbersbeloweach
confidencecategoryindicatethesizeofthetotalnumberofpeopleinthatcategory(computerscienceand
Odegaardparticipants,respectively).
Figure7.b:Correctnessratesofnoncollaboratorswithineachconfidencegroup.

WorksCited
Emden,Zeynep,Calantone,RogerJ.Droge,Cornelia.2006CollaboratingforNewProduct
Development:SelectingthePartnerwithMaximumPotentialtoCreateValue.Journalof
ProductInnovationManagement23(4):330341.
Melin,Gran.2000.Pragmatismandselforganization:Researchcollaborationontheindividuallevel.
ResearchPolicy29(1):3140.
Seifert,Timothy.2004.UnderstandingStudentMotivation.EducationalResearch46(2):137149.

You might also like