An Experimental Study of the Effects of Representa

AnExperimentalStudyoftheEffectsofRepresentationalGuidanceonCollaborativeLearningProcessesDanielD.SuthersandChristopherD.HundhausenLaboratoryforInteractiveLearningTechnologiesUniversityofHawai’iatManoaTheimportanceofbothsocialprocessesandofrepresentationalaidsforlearningiswell-established,yetfewexperimentalstudieshaveaddressedthecombinationofthesefactors.Theresearchreportedinthisarticleevaluatestheinfluenceoftoolsforconstructingrepresentationsofevidentialmodelsoncollaborativelearningprocessesandoutcomes.Pairsofparticipantsworkedwith1of3representations(Graph,Ma-trix,Text)whileinvestigatingcomplexscienceandpublichealthproblems.Depend-entmeasuresincluded(a)thecontentofparticipants’utterancesandrepresentationalactionsandthetimingoftheseutterancesandactionswithrespecttotheavailabilityofinformation;(b)amultiplechoicetestoftheabilitytorecallthedata,hypotheses,andevidentialrelationsexplored;and(c)thecontentsofawrittenessay.Theresultsshowthatrepresentationalnotationscanhavesignificanteffectsonlearners’interac-tions,andmaydifferintheirinfluenceonsubsequentcollaborativeuseoftheknowl-edgebeingmanipulated.Forexample,GraphandMatrixuserselaboratedonprevi-ouslyrepresentedinformationmorethanTextusers.RepresentationanddiscussionofevidentialrelationswasquantitativelygreatestforMatrixusersaspredicted,yetthiscameatthecostofexcessiveconsiderationandrevisionofunimportantrelations.Graphusersmayhavebeenmorefocusedintheirconsiderationofevidence,andtheworkdoneintheGraphrepresentationhadthegreatestimpactonthecontentsoftheessays.Althoughlimitedtoinitialuseofrepresentationsinalaboratorysetting,theworkdemonstratesthatrepresentationalguidanceofcollaborativelearningisworthyofstudyandsuggestsseverallinesoffurtherinvestigation.THEJOURNALOFTHELEARNINGSCIENCES,12(2),183–218Copyright©2003,LawrenceErlbaumAssociates,Inc.CorrespondenceandrequestsforreprintsshouldbesenttoDanielD.Suthers,LaboratoryforInter-activeLearningTechnologies,UniversityofHawaiiatManoa,1680EastWestRoad,POST317,Hono-lulu,HI96822.E-mail:[e-mail]eofsocialprocessestolearning,includingthepotentialutilityofcollaborativelearning,iswellestablished(Brown&Campione,1994;Lave&Wenger,1991;Scardamalia&Bereiter,1991;Slavin,1990;Webb&Palincsar,1996).Likewise,priorworkhasshowntheimportanceofrepresentationalaidstoindividualunderstandingandproblemsolving(Koedinger,1991;Kotovsky&Si-mon,1990;Larkin&Simon,1987;Novak,1990;Novick&Hmelo,1994;Zhang,1997).Yetfewstudieshaveaddressedthecombinationofthesefactors;namely,theroleofrepresentationalaidsinsupportinggrouplearningprocesses.Excep-tionsincludeBakerandLund(1997),Guzdial(1997),andSchwarz,Neuman,Gil,andIlya(2003).Theresearchreportedinthisarticleaddressesthequestionofhowwemightdesignrepresentationalsystemsthatguideandsupportcollaborativelearningprocesses(suchasdiscourse)inapositiveway.Ourstudyhasseveralconvergingmotivations.PriorworkbySuthersandcol-leaguesonBelvedere,anetworkedenvironmentforcollaborativeconstructionof“evidencemaps”(Suthers,Toth,&Weiner,1997;Suthers&Weiner,1995),sug-gestedthattherepresentationalbiasoftoolssuchasBelvederemightinfluencestu-dents’discussion.Concurrently,otherresearcherswereusingdifferentrepresenta-tionsforsimilarobjectives(supportingepistemicreasoninginscience).Forexample,SenseMaker(Bell,1997)usedacontainerrepresentation(inwhichdataaresortedintotheorycontainers);WebCamile(Guzdialetal.,1997)andSpeakEasy(Hoadley,Hsi,&Berman,1995)usedthreadeddiscussions;andPuntambekar,Nagel,Hübscher,Guzdial,andKolodner(1997)studiedmatrixrepresentations.However,thechoiceofrepresentationsforthesesystemswasgenerallynotbasedonsystematiccomparisonsoftheeffectsofrepresentationsoncollaborativelearning.TheoreticalinspirationsforundertakingsuchacomparisoncamefromRoschelle’s(1994)observationthatsharedrepresentations(animationsandsimulationsinhiscase)servetomediatecollaborativeinquiry;andfromCollinsandFergusons’(1993)discussionofrepresentationsas“epistemicforms”withassociated“epistemicgames.”OthertheoreticalmotivationsarediscussedinSuthers(1999b,2001)aswellasinthisarticle.Basedonthesemotivations,Suthersandcolleaguesundertookaclassroomstudyofrepresentationaleffects(Toth,Suthers,&Lesgold,inpress)thatwillbediscussedintheconclusionsofthisarticle.Pragmaticcon-straintslimitedthatstudytoanalysisofstudents’artifacts.Wewereunabletodi-rectlyobservetherolesrepresentationsplayedinstudents’collaborativelearningprocesses,furthermotivatingthelaboratorystudyreportedhere.ROLESOFEXTERNALREPRESENTATIONSINCOLLABORATIVELEARNINGExternalrepresentationshavelongbeenasubjectofstudyinthecontextoflearn-ingandproblem-solvingtasks,withresearchshowingthatthechoiceofrepresen-184SUTHERSANDHUNDHAUSENDoNotCopytationcaninfluenceanindividual’sconceptionofaproblemandhencetheeaseoffindingasolution(seepreviouscitations).Onemightaskwhetheritissufficienttoextrapolatefromthiswork,predictingrepresentationaleffectsongroupsbyaggre-gatingeffectsonindividuals.Althoughwebelievethatmuchcanbegainedfromsuchreasoning,wealsobelievethattheshareduseofrepresentationsbydistrib-uted