Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
CHAPTER1
Polarization
Whatexplainstheriseoffascisminthe1930s?Theemergenceofstudentradicalisminthe1960s?ThegrowthofIslamicterrorisminthe1990s?TheRwandangenocidein1994?EthnicconflictintheformerYugoslaviaandinIraq?ActsoftortureandhumiliationbyAmericansoldiersatAbuGhraibprison?TheAmericanfinancialcrisisof2008?Thewidespreadbelief,insomepartsoftheworld,thatIsraelortheUnitedStateswasresponsiblefortheattacksofSeptember11,2001?Andwhat,ifanything,dothesequestionshavetodowithoneanother?
Hereisaclue.Someyearsago,anumberofcitizensofFrancewereassembledintosmallgroupstoexchangeviewsabouttheirpresidentandaboutthe
intentionsoftheUnitedStateswithrespecttoforeignaid.1Beforetheystartedtotalk,theparticipantstendedtoliketheirpresidentandtodistrusttheintentionsoftheUnitedStates.Aftertheytalked,somestrangethingshappened.Thosewhobeganbylikingtheirpresidentendeduplikingtheirpresidentsignificantlymore.AndthosewhoexpressedmilddistrusttowardtheUnitedStatesmovedinthedirectionoffargreaterdistrust.ThesmallgroupsofFrenchcitizensbecamemoreextreme.Asaresultoftheirdiscussions,theyweremoreenthusiasticabouttheirleader,andfarmoreskepticaloftheUnitedStates,thansimilarpeopleinFrancewhohadnotbeenbroughttogethertospeakwithoneanother.
Thistalerevealsageneralfactofsociallife:Muchofthetime,groupsofpeopleendupthinkinganddoingthingsthatgroupmemberswouldneverthinkordoontheirown.Thisistrueforgroupsofteenagers,whoarewillingtorunrisksthatindividualswouldavoid.Itiscertainlytrueforthosepronetoviolence,includingterroristsandthosewhocommitgenocide.Itistrueforinvestorsandcorporate
Sunstein, Cass R.. Going to Extremes : How Like Minds Unite and Divide, Oxford University Press, USA, 2009. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/brown/detail.action?docID=431309.Created from brown on 2017-08-15 18:08:45.
Cop
yrig
ht ©
200
9. O
xfor
d U
nive
rsity
Pre
ss, U
SA
. All
right
s re
serv
ed.
executives.Itistrueforgovernmentofficials,neighborhoodgroups,socialreformers,politicalprotestors,policeofficers,studentorganizations,laborunions,andjuries.Someofthebestandworstdevelopmentsinsociallifeareaproductofgroupdynamics,inwhichmembersoforganizations,bothsmallandlarge,moveoneanotherinnewdirections.
Ofcourse,thebestexplanationsoffascismarenotadequatetoexplainstudentrebellions,andevenifweunderstandbothofthese,wewillnotbeabletoexplainethnicconflictinIraq,theRwandangenocide,abuseandbrutalityatAbuGhraib,conspiracytheoriesinvolvingIsrael,orthesubprimecrisis.Forparticularevents,generalexplanationscanuncoveronlypartsofthepicture.ButIdoaimtoshowstrikingsimilaritiesamongawiderangeofsocialphenomena.Theunifyingthemeissimple:Whenpeoplefindthemselvesingroupsoflike-mindedtypes,theyareespeciallylikelytomovetoextremes.Andwhensuchgroupsincludeauthoritieswhotellgroupmemberswhattodo,orwhoputthemintocertainsocialroles,verybadthingscanhappen.
Inexploringwhythisisso,Ihopetoseewhatmightbedoneaboutunjustifiedextremism—athreattosecurity,topeace,toeconomicdevelopment,andtosensibledecisionsinallsortsofdomains.Myemphasisthroughoutisonthephenomenonofgrouppolarization.Thisphenomenonofferslargelessonsaboutthebehaviorofconsumers,interestgroups,therealestatemarket,religiousorganizations,politicalparties,liberationmovements,executiveagencies,legislatures,racists,judicialpanels,thosewhomakepeace,thosewhomakewar,andevennationsasawhole.
GROUPSANDEXTREMISM
Whenpeopletalktogether,whathappens?Dogroupmemberscompromise?Dotheymovetowardthemiddleofthetendenciesoftheirindividualmembers?Theanswerisnowclear,anditisnotwhatintuitionwouldsuggest:Groupsgotoextremes.Moreprecisely,membersofadeliberatinggroupusuallyendupata
Sunstein, Cass R.. Going to Extremes : How Like Minds Unite and Divide, Oxford University Press, USA, 2009. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/brown/detail.action?docID=431309.Created from brown on 2017-08-15 18:08:45.
Cop
yrig
ht ©
200
9. O
xfor
d U
nive
rsity
Pre
ss, U
SA
. All
right
s re
serv
ed.
moreextremepositioninthesamegeneraldirectionastheirinclinationsbefore
deliberationbegan.2
Thisisthephenomenonknownasgrouppolarization.Grouppolarizationisthetypicalpatternwithdeliberatinggroups.Itisnotlimitedtoparticularperiods,nations,orcultures.Onthecontrary,grouppolarizationhasbeenfoundinhundredsofstudiesinvolvingmorethanadozencountries,includingtheUnited
States,France,Afghanistan,NewZealand,Taiwan,andGermany.3Itprovidesacluetoextremismofmanydifferentkinds.
Considerfourexamples:
1.Whitepeoplewhotendtoshowsignificantracialprejudicewillshowmoreracialprejudiceafterspeakingwithoneanother.Bycontrast,whitepeoplewhotendtoshowlittleracialprejudicewillshowlessprejudiceafter
speakingwithoneanother.4
2.Feminismbecomesmoreattractivetowomenaftertheytalktooneanother—atleastifthewomenwhoaretalkingbeginwithaninclinationinfavorof
feminism.5
3.Thosewhoapproveofanongoingwareffort,andthinkthatthewarisgoingwell,becomestillmoreenthusiasticaboutthateffort,andstillmoreoptimistic,aftertheytalktogether.
4.Ifinvestorsbeginwiththebeliefthatitisalwaysbesttoinvestinrealestate,theireagernesstoinvestinrealestatewillgrowasaresultofdiscussionswithoneanother.
Intheseandcountlessothercases,like-mindedpeopletendtomovetoamoreextremeversionofwhattheythoughtbeforetheystartedtotalk.Supposeinthislightthatenclavesofpeopleareinclinedtorebellionorevenviolenceandthattheyareseparatedfromothergroups.Theymightmovesharplyinthedirectionofviolenceasaconsequenceoftheirself-segregation.Politicalextremismisoftena
Sunstein, Cass R.. Going to Extremes : How Like Minds Unite and Divide, Oxford University Press, USA, 2009. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/brown/detail.action?docID=431309.Created from brown on 2017-08-15 18:08:45.
Cop
yrig
ht ©
200
9. O
xfor
d U
nive
rsity
Pre
ss, U
SA
. All
right
s re
serv
ed.
productofgrouppolarization,6andsocialsegregationisausefultoolforproducingpolarization.
Infact,agoodwaytocreateanextremistgroup,oracultofanykind,istoseparatemembersfromtherestofsociety.Theseparationcanoccurphysicallyorpsychologically,bycreatingasenseofsuspicionaboutnonmembers.Withsuchseparation,theinformationandviewsofthoseoutsidethegroupcanbediscredited,andhencenothingwilldisturbtheprocessofpolarizationasgroupmemberscontinuetotalk.Deliberatingenclavesoflike-mindedpeopleareoftenabreedinggroundforextrememovements.Terroristsaremade,notborn,andterroristnetworksoftenoperateinjustthisway.Asaresult,theycanmove
otherwiseordinarypeopletoviolentacts.7Butthepointgoeswellbeyondsuchdomains.Grouppolarizationoccursinourdailylives;itinvolvesoureconomicdecisions,ourevaluationsofourneighbors,evenourdecisionsaboutwhattoeat,whattodrink,andwheretolive.
Tounderstandthenatureofthebasicphenomenonanditspowerandgenerality,letmeoutlinethreestudiesinwhichIhavepersonallybeeninvolved.
REDSTATES,BLUESTATES
In2005,ReidHastie,DavidSchkade,andIconductedasmallexperimentin
democracyinColorado.8AboutsixtyAmericancitizenswerebroughttogetherandassembledintotengroups,usuallyconsistingofsixpeople.Membersofeachgroupwereaskedtodeliberateonthreeofthemostcontroversialissuesoftheday.
Shouldstatesallowsame-sexcouplestoenterintocivilunions?
Shouldemployersengagein“affirmativeaction”bygivingapreferencetomembersoftraditionallydisadvantagedgroups?
ShouldtheUnitedStatessignaninternationaltreatytocombatglobalwarming?
Astheexperimentwasdesigned,thegroupsconsistedof“liberal”andSunstein, Cass R.. Going to Extremes : How Like Minds Unite and Divide, Oxford University Press, USA, 2009. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/brown/detail.action?docID=431309.Created from brown on 2017-08-15 18:08:45.
Cop
yrig
ht ©
200
9. O
xfor
d U
nive
rsity
Pre
ss, U
SA
. All
right
s re
serv
ed.
“conservative”members—theformerfromBoulder,thelatterfromColoradoSprings.ItiswidelyknownthatBouldertendstobeliberalandthatColoradoSpringstendstobeconservative.Thegroupswerescreenedtoensurethattheirmembersgenerallyconformedtothesestereotypes.Forexample,groupmemberswereaskedtoreportontheirassessmentofVicePresidentDickCheney.InBoulder,thosewholikedhimwerecordiallyexcusedfromtheexperiment.InColoradoSprings,thosewhodislikedhimweresimilarlyexcused.
Inthisway,theexperimentinvolvedgroupsoflike-mindedpeople.IntheparlanceofelectionyearsintheUnitedStates,theexperimentcreatedfive“BlueState”groupsandfive“RedState”groups—fivegroupswhosemembersinitiallytendedtowardliberalpositionsingeneralandfivewhosememberstendedtowardconservativepositions.Onthethreeissuesthatinterestedus,however,participantswerenotscreenedatall.Therewasnowayofknowingtheirpreciseviewsoncivilunions,affirmativeaction,andclimatechange.Participantswereaskedtostatetheiropinionsanonymouslybothbeforeandafterfifteenminutesofgroupdiscussion,andalsototrytoreachapublicverdictbeforethefinalanonymousstatement.Theiropinionswereregisteredonascaleof0–10,where0meant“disagreeverystrongly,”5meant“disagreeslightly,”and10meant“agreeverystrongly”withtherelevantproposition(statesshouldallowcivilunionsforsame-sexcouples,employersshouldmaintainaffirmativeactionprograms,theUnitedStatesshouldsignaninternationalagreementtocontrolglobalwarming).Wewereespeciallyinterestedinasinglequestion:Howwouldpeople’sprivate,anonymousstatementsoftheirviewschangeasaresultofabriefperiodofdiscussion?
Astheexperimentunfolded,peopleinbothBoulderandColoradoSpringswerepolite,engaged,andsubstantive.Theytreatedeachotherwithcivilityandrespect.Ihaveseenthevideosofseveralofthesediscussions,anditisfairtosaythatformostoftheparticipants,therewasanefforttothinkhard,tolistentoothers,andtobereasonable.Whatwastheeffectofdiscussion?Therewerethreecritical
Sunstein, Cass R.. Going to Extremes : How Like Minds Unite and Divide, Oxford University Press, USA, 2009. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/brown/detail.action?docID=431309.Created from brown on 2017-08-15 18:08:45.
Cop
yrig
ht ©
200
9. O
xfor
d U
nive
rsity
Pre
ss, U
SA
. All
right
s re
serv
ed.
findings.
MoreExtremism
Inalmosteverygroup,membersendedupwithmoreextremepositionsaftertheyspokewithoneanother.MostoftheliberalsinBoulderfavoredaninternationaltreatytocontrolglobalwarmingbeforediscussion;theirenthusiasmincreasedafterdiscussion.MostoftheconservativesinColoradowereneutralonthattreatybeforediscussion;theystronglyopposeditafterdiscussion.Discussionmadesame-sexcivilunionsmorepopularamongtheliberalsinBoulder;discussionmadecivilunionslesspopularamongconservativesinColoradoSprings.Mildlyfavorabletowardaffirmativeactionbeforediscussion,liberalsbecamestronglyfavorabletowardaffirmativeactionafterdiscussion.Firmlynegativeaboutaffirmativeactionbeforediscussion,conservativesbecameevenmorenegativeaboutaffirmativeactionafterdiscussion.
MuchLessInternalDiversity
Theexperimenthadaseparateeffect,onethatisequallyimportant:Itmadebothliberalgroupsandconservativegroupssignificantlymorehomogeneous—andthussquelcheddiversity.Beforemembersstartedtotalk,manygroupsdisplayedafairbitofinternaldisagreement.Thegroupdisagreementswerereducedasaresultofamerefifteen-minutediscussion.Notethattheprimarytesthereinvolveswhathappenedtotheiranonymousstatements.Howdiversewerepeople’spredeliberationviews,ontheseissues,comparedwiththeirpostdeliberationviews?Intheirprivatestatements,groupmembersshowedfarmoreconsensusafterdiscussionthanbefore.
GreaterRifts
Itfollowsthatdiscussionhelpedtowidentheriftbetweenliberalsandconservativesonallthreeissues.Beforediscussion,someliberalgroupswere,on
Sunstein, Cass R.. Going to Extremes : How Like Minds Unite and Divide, Oxford University Press, USA, 2009. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/brown/detail.action?docID=431309.Created from brown on 2017-08-15 18:08:45.
Cop
yrig
ht ©
200
9. O
xfor
d U
nive
rsity
Pre
ss, U
SA
. All
right
s re
serv
ed.
someissues,fairlyclosetosomeconservativegroups.Theresultofdiscussionwastodividethemfarmoresharply.
Here,then,isaninitialindicationofwhygroupsgotoextremes.Whenpeopletalktolike-mindedothers,theytendtoamplifytheirpreexistingviews,andtodosoinawaythatreducestheirinternaldiversity.Weseethishappeninpolitics;ithappensinfamilies,businesses,churchesandsynagogues,andstudentorganizationsaswell.
FEDERALJUDGESANDPOLARIZEDDIFFERENCES
Formanydecades,theUnitedStateshasbeenconductingatrulyextraordinarynaturalexperimentinvolvinggroupbehavior,moderation,andextremism.Theexperimentinvolvesfederaljudges,whoarerandomlyassignedintogroupsthatlookabitlikeBoulderandColoradoSprings.Whatcanwelearnfromthisexperiment?Thesimplestlessonisthatnolessthanordinarycitizens,like-mindedjudgesgotoextremes.Thisisastrikingfinding,becausejudgesarespecialistsandlearnedinthelaw;theyarenotsupposedtobesovulnerabletothepoliticalinclinationsoftheircolleagues.
Onfederalcourtsofappeals,judicialpanelsconsistofthreejudges.Thepossiblepanelcompositionsarejustfour:(a)threeRepublicanappointees,(b)threeDemocraticappointees,(c)twoRepublicanappointeesandoneDemocraticappointee,and(d)twoDemocraticappointeesandoneRepublicanappointee.Panelassignmentsarerandom,andthesampleisverylarge.Forthisreason,itispossibletotestwhetherjudicialvotesareaffectedbypanelcomposition—thatis,whetherRepublicanandDemocraticappointeesvotedifferentlydependingonwhethertheyaresittingwithRepublicanorDemocraticappointees.Doweobserveanythinglikegrouppolarizationamongfederaljudges?
Forpresentpurposes,thekeyquestionsarethese:HowdoRepublicanappointeesvoteonpanelsconsistingsolelyofRepublicanappointees(RRRpanels)?HowdoDemocraticappointeesvoteonpanelsconsistingsolelyof
Sunstein, Cass R.. Going to Extremes : How Like Minds Unite and Divide, Oxford University Press, USA, 2009. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/brown/detail.action?docID=431309.Created from brown on 2017-08-15 18:08:45.
Cop
yrig
ht ©
200
9. O
xfor
d U
nive
rsity
Pre
ss, U
SA
. All
right
s re
serv
ed.
Democraticappointees(DDDpanels)?RRRpanelsareabitlikeColoradoSprings,andDDDpanelsareabitlikeBoulder.DofederaljudgesbehaveascitizensdointheColoradoexperiment?Morespecifically,wemightaskwhetherRepublicanappointees,onRRRpanels,behavedifferentlyfromRepublicanappointeesonRRDpanelsorRDDpanels,andwhetherDemocraticappointees,onDDDpanels,behavedifferentlyfromDemocraticappointeesonDDRorDRRpanels.Dolike-mindedjudgesshowespeciallydistinctivevotingpatterns?
Thephenomenonofgrouppolarizationtellsuswhattoexpect.BothDemocraticandRepublicanappointeesshouldshowextremebehavioronpanelsthatareunified,thatis,onDDDandRRRpanels.WhereverDemocraticappointeesandRepublicanappointeesshowageneraldifferenceinvotingpatterns,thatdifferencewillbeamplifiedifwecompareDemocraticappointeesonDDDpanelswithRepublicanappointeesonRRRpanels.Totestthisclaim,wemightwanttocomparetwofigures:(a)thetotaldifferencebetweentheliberalvotingratesofDemocraticappointeesandthatofRepublicanappointeesand(b)thedifferencebetweentheliberalvotingratesofDemocraticappointeesonall-DemocraticpanelsandtheliberalvotingratesofRepublicanappointeesonall-Republicanpanels.Thelatterdifference—betweenDemocraticappointeesonDDDpanelsandRepublicanappointeesonRRRpanels—mightbecalledthepolarizeddifference.
Incountlessareas,Democraticappointeesshowespeciallyliberalvotingpatternsonall-Democraticpanels.Republicanappointeesshowespeciallyconservativevotingpatternsonall-Republicanpanels.Ifweaggregateallcasesshowinganideologicaldifferencebetweenthetwogroups,wefinda15percentdifferencebetweenRepublicanandDemocraticappointeesinliberalvotingrates.Thatisaprettybigdifference.Butthepolarizeddifferenceisfarhigher—34percent!
Ourmethodwasquitesimple.Wecollectedtensofthousandsofjudicialvotes,mostlyinideologicallycontestedcases,includingracediscrimination,sexdiscrimination,disabilitydiscrimination,affirmativeaction,campaignfinance,
Sunstein, Cass R.. Going to Extremes : How Like Minds Unite and Divide, Oxford University Press, USA, 2009. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/brown/detail.action?docID=431309.Created from brown on 2017-08-15 18:08:45.
Cop
yrig
ht ©
200
9. O
xfor
d U
nive
rsity
Pre
ss, U
SA
. All
right
s re
serv
ed.
environmentalprotection,labor,andfreespeech.Weusedsimple,relativelyuncontroversialteststocodedecisionsas“liberal”or“conservative.”Forexample,ajudicialrulinginfavorofanAfricanAmericanplaintiff,allegingracediscrimination,wascodedasliberal.Similarly,wecharacterizedasliberalavotethatfitstheusualpoliticalstereotypes—toupholdanaffirmativeactionprogram,acampaignfinancerestriction,anenvironmentalregulationchallengedastooaggressive,oradecisionoftheNationalLaborRelationsBoardinfavorofemployees.True,thesetestsofwhetherajudicialdecisionisliberalareprettycrude.Butbecausethesampleissobig,weareabletodiscernclearandilluminatingpatterns;thecrudenessofthetestsdoesnotseemtohaveintroduceddistortions.
Considerjustafewkeyexamples.9
•Ingayrightscases,theoverallspreadbetweenRepublicanappointeesandDemocraticappointeesis41percent—Republicanappointeesvoteinfavorofgayrights16percentofthetimecomparedwitha57percentrateforDemocraticappointees.ButifwecomparehowDemocraticappointeesvoteonDDDpanelstohowRepublicanappointeesvoteonRRRpanels,thepolarizeddifferenceturnsouttobemorethandouble—86percent!Inourdataset,Republicanappointeesvotepro–gayrights14percentofthetimeonRRRpanels—comparedwith100percentforDemocraticappointeesonDDDpanels.
•Incasesinvolvingdisabilitydiscrimination,theoveralldifferenceis18percent;thepolarizeddifferenceisnearlydouble,at33percent.
•IncasesinvolvingdecisionsbytheEnvironmentalProtectionAgency,theoveralldifferenceinvotingis15percent;thepolarizeddifferenceisnolessthan36percent.
•Inaffirmativeactioncases,theoveralldifferenceisasignificant28percent;thepolarizeddifferenceisawhopping49percent.
•Insexdiscriminationcases,theoveralldifferenceis17percent;thepolarizedSunstein, Cass R.. Going to Extremes : How Like Minds Unite and Divide, Oxford University Press, USA, 2009. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/brown/detail.action?docID=431309.Created from brown on 2017-08-15 18:08:45.
Cop
yrig
ht ©
200
9. O
xfor
d U
nive
rsity
Pre
ss, U
SA
. All
right
s re
serv
ed.
differenceisnearlytriple,at46percent.
Ifalloftheevidenceistakenasawhole,thelessonisunmistakable.ItisnotexactlyshockingtofindthatRepublicanandDemocraticappointeesshowsignificantlydifferentvotingpatterns.Buttheoveralldifferenceismuchsmallerthanthepolarizeddifference—thedifferencebetweenhowRepublicanappointeesvotewhensittingonlywithRepublicanappointeesandhowDemocraticappointeesvotewhensittingonlywithDemocraticappointees.Onthisscore,judgesdonotlookawholelotdifferentfromcitizensinColoradoSpringsandBoulder.Whentheysitwithlike-mindedothers,theybecomemoreextreme.
Onequalification:Whilethisisthecentralpatterninmanyareasofthelaw,therearethreeareasinwhichjudgesarenotaffectedbythepanel’scomposition.Inthoseareas,bothRepublicanandDemocraticappointeesvotethesamewhethertheyareintheminorityorpartofaunifiedpanel.Thethreeareasareabortion,capitalpunishment,andnationalsecurity.Apparentlyjudgeshavesuchstrongconvictionsinsuchcasesthattheyarenotaffectedbywhattheircolleaguessayordo.Iwillreturntothispointlater;itoffersanimportantcautionarynoteaboutmycentralclaims.Sometimespeoplefeelreallystrongly,andtheviewsofothersdonotmovethem.
PUNISHINGWRONGDOERS
Nowletusturntothebehaviorofjuriesand,inparticular,totheeffectsofdeliberationonpunitivedamageawards.Thisisaprettytechnicalarea,butanunderstandingofthoseeffectswill,Ihope,illuminateanumberofissuesincludingbutextendingwellbeyondpoliticsandlaw.
InAmericanlaw,punitivedamageawardsareofmajorimportanceintheirownright.Companiesaregreatlyconcernedaboutunpredictableandsometimesveryhighawards,inthehundredsofmillionsofdollars.Manypeoplehavetriedtodevelopwaystodisciplinejurydecisions,andtheSupremeCourthastakenan
Sunstein, Cass R.. Going to Extremes : How Like Minds Unite and Divide, Oxford University Press, USA, 2009. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/brown/detail.action?docID=431309.Created from brown on 2017-08-15 18:08:45.
Cop
yrig
ht ©
200
9. O
xfor
d U
nive
rsity
Pre
ss, U
SA
. All
right
s re
serv
ed.
activeinterestintheproblem.Moreimportantstill,punitivedamageawardsprovideanexcellentareainwhichtostudytheconsequencesofdiscussionongroupbehavior,especiallyforpeoplewhodisplayadegreeofoutrage—andoutrageisoneofmycentralconcernshere.
Ifgroupmembersbeginwithadegreeofoutrage,dodeliberatinggroupsbecomemoreoutragedorlessso?Theanswerbearsonsocialmovementsandpoliticalprotestsofmanydifferentkinds.Asweshallsee,italsobearsonfeuds,ethnicconflict,andevenfamilybehavior.Whenachildisupsetatunfairbehavioratschool,howareparentslikelytoreact?Whenahusbandisangryaboutunfairnessdirectedathimatwork,howwillawifereact,andhowwillhiswife’sreactionaffecthim?
Tounderstandthejuryexperiments,conductedwithDanielKahnemanandDavidSchkade,wemustbeginwithastudyofindividuals,notgroups,involvingabout1,000people,whowereaskedtoregistertheirjudgmentsaboutmisconduct
byacorporatedefendant.10Thegoalwastounderstandwhypunitivedamageawardsaresovariable:Whydosomejuriescomeupwithawardsof$100,000andotherswithawardsof$1million,incasesthatseemprettysimilar?Weaskedpeopletorecordtheirjudgmentsonthreedifferentscales.Thefirstwasaboundedscaleof0to6,involvingtheoutrageousnessofthecompany’sbehavior.Eachofthepointsalongthescalewasclearlymarked,sothat0meant“notatalloutrageous”and6meant“exceptionallyoutrageous.”Thesecondwasalsoaboundedscaleof0to6,butthisscalemeasuredthedesiredlevelofpunishment;0meant“none”and6meant“extremelysevere”punishment.Thethirdscalewastheunboundedoneofdollars.Shouldthecompanyhavetopay$10,000?$100,000?$1million?More?
Ourcentralfindings,involvingpersonalinjurycases,werestraightforward.Peopleagreeonhowoutrageouscorporatemisconductis.Theyalsoagreeontheappropriateseverityofpunishmentontheboundedscale.Butthedollarscalecreatesalotoftroubleandconfusion.
Sunstein, Cass R.. Going to Extremes : How Like Minds Unite and Divide, Oxford University Press, USA, 2009. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/brown/detail.action?docID=431309.Created from brown on 2017-08-15 18:08:45.
Cop
yrig
ht ©
200
9. O
xfor
d U
nive
rsity
Pre
ss, U
SA
. All
right
s re
serv
ed.
Toestablishthesepoints,weusedasimpletechnique,inwhichindividualresponsesarepooledtoproduce“statisticaljuries,”whoseverdictisthejudgmentofthemedianmember.Havingdonethis,wefoundthatsmallgroupsofsixpeople,orstatisticaljuries,usuallyagreeaboutoutrageousnessandappropriatepunishment.Importantly,theagreementcutsacrossdemographicdifferences.Withthemagicofthecomputer,wecancreatestatisticaljuriesofanyimaginablekind—allmale,allfemale,allwhite,allHispanic,allAfricanAmerican,allrich,allpoor,allold,allyoung,allwelleducated,allpoorlyeducated.Demographydoesnotmatter.Allthesegroupsessentiallyagreewithoneanother!
Bycontrast,statisticaljuriesshowalotofvariabilitywithrespecttodollarawards.Thedollarjudgmentofonejuryisnotagoodpredictorofthedollarjudgmentsofotherjuries.Butdemographyisnotthesourceofthevariability;itisnotasifrichpeopledisagreewithpoorpeople,oroldpeopledisagreewithyoungpeople,ormendisagreewithwomen.Theproblemisthedollarscale.Thereasonforthevariabilityisthatwhatevertheirdemographicgroup,peopledonothaveaclearsenseofhowtotranslatetheirpunitiveintentions,onaboundedscale,ontothescaleofdollars.Doesa“6”meanapunishmentof$50,000,or$100,000,or$1million,or$10million,ormore?Peoplejustdon’tknow.Thedollarscale,boundedatthelowerend($0)andessentiallyunboundedattheupperend,lackssignpoststhatgivemeaningtothevarious“points”onthescale.Forthisreason,peoplewhoagreethatthecaseisa“4”onascaleof0–6maynotagreeontheappropriatetranslationofthatfigureintosomemonetaryequivalent.
ThestudyIhavejustdescribedinvolvedanefforttopoolindividualresponses;itdidnotinvolvegroupdiscussion.Ifwewanttounderstandhowjuriesactuallybehave,orhowoutragedevelopsintherealworld,thisisabigdefect.
Henceweconductedafollow-upexperiment,involvingabout3,000jury-eligiblecitizensand500deliberatingjuries,eachwithsixpeople.Ourgoalwastolearnhowpeoplewouldbeinfluencedbyseeinganddiscussingtheviewsofothers.Hereishowtheexperimentworked.Peoplereadaboutapersonalinjury
Sunstein, Cass R.. Going to Extremes : How Like Minds Unite and Divide, Oxford University Press, USA, 2009. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/brown/detail.action?docID=431309.Created from brown on 2017-08-15 18:08:45.
Cop
yrig
ht ©
200
9. O
xfor
d U
nive
rsity
Pre
ss, U
SA
. All
right
s re
serv
ed.
case,includingtheargumentsmadebybothsides.Theywerealsoaskedtorecord,inadvanceofdeliberation,anindividual“punishmentjudgment,”nowonascaleof0to8,where(again)0indicatedthatthedefendantshouldnotbepunishedatall,and8indicatedthatthedefendantshouldbepunishedextremelyseverely.Aftertheindividualjudgmentswererecorded,jurorsweresortedintosix-persongroupsandaskedtodeliberatetoreachaunanimous“punishmentverdict.”Youmightpredict(aswedid)thatpeoplewouldcompromiseandthattheverdictsofjurieswouldbethemedianofpunishmentjudgmentsofjurors.Butyourpredictionwouldbebadlywrong.
Instead,theeffectofdeliberationwastocreatebothaseverityshiftforhigh-punishmentjurorsandaleniencyshiftforlow-punishmentjurors.Whenthemedianjudgmentofindividualjurorswas4orhigheronthe8-pointscale,thejury’sverdictendeduphigherthanthatmedianjudgment.Consider,forexample,acaseinvolvingamanwhonearlydrownedonadefectivelyconstructedyacht.Jurorstendedtobeoutragedbytheideaofadefectivelybuiltyacht,andgroupsweresignificantlymoreoutragedthantheirmedianmembers.Highlevelsofoutrageandseverepunitivejudgmentsbecamehigherandmoresevereasaresultofgroupinteractions.
Butwhenthemedianjudgmentofindividualjurorswasbelow4,thejury’sverdictwastypicallybelowthatmedianjudgment.Consideracaseinvolvingashopperwhowasinjuredinafallwhenanescalatorsuddenlystopped.Individualjurorswerenotgreatlybotheredbytheincident,seeingitasagenuineaccidentratherthanacaseofseriouswrongdoing.Insuchcases,juriesweremorelenientthanindividualjurors.Here,then,isalessonaboutwhathappenswhenpeoplediscusswrongdoing.Ifgroupmembersareupset,theywillprobablygetmoreupsetaftertalkingtoeachother.Ifgroupmembersthinkthatwhathappenedisnotabigdeal,theywillusuallythinkthatwhathappenedisbasicallynothingafteraperiodofdiscussion.
Withdollarawards,bycontrast,juriesweresystematicallymoresevereintheir
Sunstein, Cass R.. Going to Extremes : How Like Minds Unite and Divide, Oxford University Press, USA, 2009. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/brown/detail.action?docID=431309.Created from brown on 2017-08-15 18:08:45.
Cop
yrig
ht ©
200
9. O
xfor
d U
nive
rsity
Pre
ss, U
SA
. All
right
s re
serv
ed.
awardsthanthemedianjuror.Eventhesmallawardsweretypicallyhigherthantheawardselectedbythemedianjurorbeforepeoplestartedtotalk.Hereisthemoststrikingfinding:In27percentofthecases,thejury’sawardwasatleastashighasthatofthehighestpredeliberationjudgmentofthemembersofthatparticularjury!Hencetheshifttowardmoreseverity,andmoreextremism,wasespeciallypronouncedwithdollars.Itfollows,bytheway,thatthemonetaryawardsbydeliberatingjurieswereevenmoreunpredictablethanthemonetaryawardsbystatisticaljuries.
Letmeunderlineourtwokeyfindings.Thefirstisthatwhenpeoplebeginwithahighlevelofoutrageandfavorsomekindofaggressiveresponses,groupsaremoreaggressivethanindividuals.Thesecondisthatformonetaryawards,juriesaresignificantlymoreextremethanjurors.
TAKINGRISKS
Whathappenswhenpeoplewhoareinclinedtotakeriskstalkwithotherpeoplewhoareinclinedtotakerisks?Theansweristhattheybecomestillmoreinclined
totakerisks.11
Consider,forexample,thequestionswhethertotakeanewjob,toinvestinaforeigncountry,toescapefromaprisoner-of-warcamp,ortorunforpolitical
office.12Withrespecttomanydecisions,membersofdeliberatinggroupsbecamesignificantlymoredisposedtotakerisksafterabriefperiodofcollectivediscussion.Onthebasisofsuchevidence,itbecamestandardtobelievethatdeliberationproducedasystematic“riskyshift.”Forasignificantperiod,themajorconsequenceofgroupdiscussion,itwasthought,wastoproducethatriskyshift—athoughtthatwouldbearonmanypartsofsociallife,becausegroupsareoftenaskedtodecidewhethertotakeagambleor,instead,totakeprecautions.
Butlaterstudiesdrewthisconclusionintoseriousquestion.Theyevenraisedthequestionwhetherculture,ratherthangroupdynamics,isresponsiblefortheriskyshift.OnmanyofthesamequestionsonwhichAmericansdisplayedariskyshift,
Sunstein, Cass R.. Going to Extremes : How Like Minds Unite and Divide, Oxford University Press, USA, 2009. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/brown/detail.action?docID=431309.Created from brown on 2017-08-15 18:08:45.
Cop
yrig
ht ©
200
9. O
xfor
d U
nive
rsity
Pre
ss, U
SA
. All
right
s re
serv
ed.
Taiwanesesubjectsshoweda“cautiousshift.”13Onmostofthetopicsjustlisted,deliberationledcitizensofTaiwantobecomesignificantlylessrisk-inclinedthantheywerebeforetheystartedtotalk.NorwasthecautiousshiftlimitedtotheTaiwanese.AmongAmericans,deliberationsometimesproducedacautiousshiftaswell,asrisk-aversepeoplebecamemorereluctanttotakecertainrisksafterthey
talkedwithoneanother.14Therearetwomajorexamplesofcautiousshifts:thedecisionwhethertomarry(!)andthedecisionwhethertoboardaplanedespitesevereabdominalpain,possiblyrequiringmedicalattention.Inthesecases,themembersofdeliberatinggroupsmovedtowardgreatercaution.
Atfirstglance,itseemedhardtoreconcilethesecompetingfindings,butthereconciliationturnedouttobesimple:Thepredeliberationmedianisthebest
predictorofthedirectionoftheshift.15Whengroupmembersaredisposedtowardrisk-taking,ariskyshiftisobserved.Whenmembersaredisposedtowardcaution,acautiousshiftisobserved.ItfollowsthatthestrikingdifferencebetweenAmericanandTaiwanesesubjectsisnotaproductofanyculturaldifferenceinhowpeoplebehaveingroups.Itresultsfromadifferenceinthepredeliberation
mediansoftheAmericansandtheTaiwaneseonthekeyquestions.16WhenAmericansshowapre-deliberationmedianinfavorofcaution,discussionmovesthemtowardgreatercaution;thesameistrueofTaiwanese.WhenAmericangroupsshowariskyshift,andTaiwaneseacautiousshift,itissimplybecauseofadifferenceintheirinitialinclinations.Thustheriskyshiftandthecautiousshiftarebothsubsumedunderthegeneralrubricofgrouppolarization.
Itistemptingtowonderwhethergrouppolarizationisaproductofparticularculturesandparticular“types.”ButasIhavenoted,thereisnonationonearthinwhichgrouppolarizationhasbeenfoundnottooccur.Iwillreturn,however,tosomewaysofcounteractingit.
Inthebehaviorallaboratory,grouppolarizationhasbeenshownina
remarkablywiderangeofcontexts.17Howgood-lookingarecertainpeople?Groupdeliberationproducesmoreextremejudgmentsaboutthatquestion:If
Sunstein, Cass R.. Going to Extremes : How Like Minds Unite and Divide, Oxford University Press, USA, 2009. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/brown/detail.action?docID=431309.Created from brown on 2017-08-15 18:08:45.
Cop
yrig
ht ©
200
9. O
xfor
d U
nive
rsity
Pre
ss, U
SA
. All
right
s re
serv
ed.
individualsthinkthatsomeoneisgood-looking,thegroupislikelytothinkthat
personisdevastatinglyattractive.18(Moviestarsundoubtedlybenefitfromthisprocess.)Grouppolarizationalsooccursforobscurefactualquestions,suchashow
farSodom(ontheDeadSea)isbelowsealevel.19Evenburglarsshowashiftinthe
cautiousdirectionwhentheydiscussprospectivecriminalendeavors.20Inarevealingfindingattheintersectionofcognitiveandsocialpsychology,groupshavebeenfoundtomakemore,ratherthanfewer,“conjunctionerrors”(believingthatAandBaremorelikelytobetruethanAalone)thanindividualswhenindividualerrorratesarehigh—thoughfewerwhenindividualerrorratesare
low.21
Togetasenseofthepowerofgrouppolarizationinthedomainsoflawandpolitics,considerjustafewmorestudies.Afterdeliberation,groupsofpeopleturnouttobefarmoreinclinedtoprotestapparentlyunfairbehaviorthanwastheir
medianmemberbeforediscussionbegan.22Consider,forexample,theappropriateresponsetothreedifferentevents:policebrutalityagainstAfricanAmericans,anapparentlyunjustifiedwar,andsexdiscriminationbyalocalcitycouncil.Ineveryoneofthesecontexts,deliberationmadegroupmembersfarmorelikelytosupportaggressiveprotestaction.Groupmembersmoved,forexample,fromsupportforapeacefulmarchtosupportforanonviolentdemonstration,suchasasit-inatapolicestationorcityhall.Interestingly,thesizeoftheshifttowardamoreextremeresponsewascorrelatedwiththeinitialmean.Whenpeopleinitiallysupportedastrongresponse,groupdiscussionproducedagreatershiftinthedirectionofsupportforastillstrongerresponse.Asweshallsee,thisfindingisstandardwithintheliterature:Theshifttowardextremismisoftenlargerwhenthe
averagepersonstartswithaprettyextremeposition.23
Peopleoftenmakeindividualjudgmentsaboutfairnessandunfairness;theyalsomakethosejudgmentsingroups.Whathappenstoourjudgmentsaboutunfairnesswhenwespeakwithoneanother?Theanswershouldnowbeclear:Whenweareindividuallyinclinedtobelievethatunfairnesshasoccurred,our
Sunstein, Cass R.. Going to Extremes : How Like Minds Unite and Divide, Oxford University Press, USA, 2009. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/brown/detail.action?docID=431309.Created from brown on 2017-08-15 18:08:45.
Cop
yrig
ht ©
200
9. O
xfor
d U
nive
rsity
Pre
ss, U
SA
. All
right
s re
serv
ed.
discussionwillintensifyourbeliefsandmakeusveryangry.24Therelevantstudieswerequiterealistic.Peoplewereaskedtoengageintasksdesignedtosimulateactivitiesthatmightactuallybeundertakeninabusinesssetting—suchasclassifyingbudgetitems,schedulingmeetings,androutingaphonemessagethroughtheproperchannelswithassignmentoftheproperlevelofpriority.Goodperformancecouldproducefinancialrewards.Aftercompletingthetasks,peoplewereabletoaskfortheirsupervisors’judgmentsandreceivefeedbackfromthem.Someoftheanswersseemedrudeandunfair,suchas“I’vedecidednottoreadyourmessage.Theinstructionssayit’suptome...sodon’tbothersendingmeanyothermessagesorexplanationsaboutyourperformanceonthistask”and“Ifyouwouldhaveworkedharder,thenyou’dhavescoredhigher.Iwillnotacceptyourmessageonthisround!”
Peoplewereaskedtoratetheirsupervisorsalongvariousdimensions,includingfairness,politeness,bias,andgoodleadership.Theratingsoccurredinthreeperiods.Thefirstincludedindividualratings,thesecondincludedagroupconsensusjudgment,andthethirdincludedindividualratingsaftergroupjudgment.Itturnedoutthatgroupjudgmentswerefarmorenegativethanthe
averageofindividualjudgments.25Inmanycases,groupmembersdecidedthatthebehaviorwasreallyveryunfair,eventhoughindividualsbelievedthatthebehaviorwasonlymildlyunfair.Interestingly,thegroups’conclusionsweretypicallymoreextremethanwerepeople’sindividualjudgmentsafterdeliberation.Butsuchjudgmentswerenonethelessmorenegative,andthusmoreextreme,thanpredeliberationindividualjudgments.
Thesefindingsareremarkablysimilartothoseinvolvingjuroroutrage,where,aswehaveseen,groupsaremoreoutragedthantheirmedianmember.Wenowhaveastrongclueaboutthesourcesofprotestmovements,atopicthatIexploreinduecourse.Forthemoment,letustrytoexplaingrouppolarization.
Sunstein, Cass R.. Going to Extremes : How Like Minds Unite and Divide, Oxford University Press, USA, 2009. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/brown/detail.action?docID=431309.Created from brown on 2017-08-15 18:08:45.
Cop
yrig
ht ©
200
9. O
xfor
d U
nive
rsity
Pre
ss, U
SA
. All
right
s re
serv
ed.