Mencommunicatemoreassertivelythanwomen.Ifwomenhavean“opendoor”,theyemphasize moreoftenthattheleaderisanequalmemberofthegroup.Womenhaveastabletraitofsending relevant messages, which represent the transformational leadership style, while men use relevant messages more as a tool if necessary for the improvement of group coordination. Despite the differencesinthecommunicationpathsoftheleaders,menandwomenareequallylikelytousethe transformational leadership style, in that they send relevant messages, which relates positively to leader effectiveness. Men and women are equally effective leaders, who are equally credible and equallylikelytorequestforty.Wefindaweakhintthatwomenwithan“opendoor”mighthaveit harder to make their followers adapt to them, as followers might perceive to have a stronger influenceonwomenratherthanmen.However,overtime,womengaincredibilityandtheirgroups becomesuccessful.Asoneofthefemaleleadersexpressedit“everyoneisonboardnow”,aftershe finally managed in the last periods that her group simultaneously chose the most efficient effort level. Initially, before followers learn that women can be as effective leaders as men, they might
41
We use OLS regressions for the leader’s evaluation as the dependent variable, and interaction of the communicationtreatmentandthegenderoftheleaderastheexplanatoryvariables.Werestricttofollower observationsinPart2andcorrectstandarderrorsforclusteringatthegrouplevel.
thinkthatothersholdstereotypebeliefsagainstfemaleleaders.However,theyarewillingtolearn theoppositeandtogivefemaleleadersachancetoletthemlead.Whatspeaksforthisspeculation is that we find no discrimination in followers’ effort, in followers’ messages, in aggregate leader evaluations,orinfollowingassertive,masculineͲactingmaleandfemaleleaders,althoughliterature reportsaboutevidenceforsuchphenomena(Ridgeway2001;Eaglyetal.2003;Day2014).
Oneshortcomingofourdataistherelativelylownumberofobservationsandthatourdatamightbe overͲfitted.Exceptforonemaleleaderwhosentonceasuggestionofzero,allmaleleadersalways requested forty if they sent any message. Thus, the limitation of our data is that we have no variationintheobservationsformaleleaders.Wecannotcomparethewholespectrumofmaleto female behavior, or the impact of message value within gender. What we can compare is male leaders requesting forty to female leaders requesting forty, and follower behavior in these cases. Anothershortcomingmightbethepresenceofanegativeexperimenterdemandeffect(Zizzo2010). Such an effect would go into the opposite direction than our prediction of male leaders being followedtoagreaterextentthanfemaleleaders,thusleavingroomforanalternativeinterpretation ofourresultsthatdoesnotruleoutthatmenmightbemoreeffectiveleadersthanwomen.
Therandomizationoftheleaderpositionmightinvolveashortcomingoftheexperimentaldesign, since leaders are forced into leadership. Followers might stereotypically think that the female leadersinthelaboratoryarelesseffectiveleaders,whoareonlyforcedintotheposition,whichthey would not hold in the real world.42In the real world, leaders rather voluntarily selfͲselect into leadership, for example, when starting a business, or they selfͲselect at least into candidacy for leadership,forexample,whencompetingforatopposition.Relatedtogenderandleadership,some studiesalreadyexploredfactorslikecompetition,orelectionaversion(Reubenetal.2012;Kanthak andWoon2014).Brandtsetal.(2015)studythedemocracyeffectonleadereffectiveness,finding that elected leaders are more effective than randomly selected leaders, while leaving it an open question for future research whether leaders of different gender would be equally effective. Althoughitisaweakcontrolforwhetherleaderswouldhavevolunteeredinourstudyanyway,and whethertheyembracedtheleaderposition,butusingdatafromourfinalquestionnaire,wefindno differencesinthepreferencefortheleaderrolebythegenderoftheleader.
AspointedoutinEaglyetal.(2003),thetransformationalleadershipstyleisindependentfromthe underlying organizational structure. In our study, the organizational structure is rather flat. The leader is closely tied to the group, and faces the same incentives as other group members. These circumstancesmighteasetheuseofthetransformationalleadershipstyleforbothmenandwomen, andthereforemightnotbringgenderdifferencesontothesurface.Amorecompetitivesettingora stronger hierarchy might magnify maleͲdominated leadership qualities. Nevertheless, due to the findingsinourpresentstudyweseenoreasonwhywomenshouldnotbeempoweredtothesame extentlikemen.Womenhaveadifferentstyletocommunicateandtolead,andsinceweshowed that followers adapt over time to their leaders, we have reason to expect that an increase in the number of female leaders could also affect the structure of organizations on the long run. Such a “feminine” development might lead to more balanced, more diverse, and more effective leaderͲ
followerinteractionswithinorganizations.
42
This would explain the leader evaluation results. Unsuccessful women are blamed less, but successful womenareevaluatedmoreharshlythanmen.
Chapter4
AReplicationoftheDemocracyEffectonLeaderEffectivenessinthe
TurnaroundGame
withErnestoReuben4.1Introduction
Economicactorsareinterdependentinmanysituations.Ifinputsoftheactorsarecomplements,the lowestͲperforming actor is likely to determine group performance (Brandts et al. 2015). Globally, exertingefforttoreduceemissionswouldbeefficientforeveryoneifothersactinthesameway.In a small organization, like a project team or a startͲup team, economic actors team up to be more successful and achieve higher outcomes through coordinated action. Leaders can be useful “coordinationdevices”insuchsituations(KrissandEil2012).Whatmakessomeleadersmoreeffectivethanothers?Leadershipisoneofthemostwidelystudied universal phenomena of human behavior (Burns 1978; Bass and Bass 2008).43Our interest lies in exploring the determinants of leader effectiveness, and in this study we focus on the selection process of leaders. How does a more democratic selection procedure alter leader effectiveness? First, democracy can have a selection effect on the leader’s abilities, as democracy attracts and stronglypositivelyselectsmorecompetentleaders(DalBóetal.2015b).DalBóetal.(2015b)show, using Swedish data, “politicians are on average significantly smarter and better leaders than the populationtheyrepresent”.Earlier,usinganextensivedataset,Besleyetal.(2011)showthatmore highlyeducatedleadershaveapositiveeffectongrowth.Second,literatureonthedemocracyeffect showsthat,besideshavingeffectsthroughthechoiceofpolicies,democraticinstitutionscanaffect behavior directly (Dal Bó et al. 2010).44Moreover, employee participation in decisions has been showntohaveanenhancingeffectonproductivity(LevineandTyson1990;Boninetal.1983;Black andLynch2001). Brandtsetal.(2015)usedtheturnaroundgametostudytheroleoftheleaderselectionprocesson leaders’effectiveness.Theirtreatmentdesignhas3x2factors,varyingtheselectionoftheleader(no leader,randomlyselectedleader,electedleader)andtheincentivescheme(noincentiveincrease, incentiveincrease).Theyfindthatcommunicationfromaleaderhasagreatereffectthanincentives, andthatelectedleadersaremoreeffectivethanrandomlyselectedleaders.Further,theyfindthat electedleadersdoabetterjobbecausetheysendmoreperformanceͲrelevantmessages,andthus theelectioneffectisnotdrivenbyadifferenceinfollowerreactions.Inananalogouspaper,Brandts etal.(forthcoming)showthatinternallyselectedleadersaremorelegitimatethanexternalleaders. Internal leaders have a lower social distance from the group, which grants them increased social credibility, so that followers follow them to a greater extent. Additionally, the authors compare randomly selected internal leaders with internally elected leaders, and, after controlling for
43
Recent reviews of studies from psychology, political science, economics and management provide an overviewaboutthedevelopmentinleadershipliterature(Avolioetal.2009;AhlquistandLevi2011).
44
Dal Bó et al. (2015a) even validate these earlier results by using a novel analytical approach ruling out selectioneffectsundercertainconditions.
observable leader characteristics and content of leader messages, they find that the increased effectivenessofelectedleadersrootsinfollowerreactions.
We decided to replicate the paper by Brandts et al. (2015), since the context and methodology in their paper comes close to our interests. We share the common view that reproducibility is an important principle in science: if a result is valid, then it is possible to replicate it independently. Previewing our results, our main contribution is that we are able to replicate the main finding in Brandts et al. (2015). We also contribute to the strand of literature on democratic institutions solving social dilemmas and coordination problems, especially in the context of organizational hierarchycombinedwithemployeeparticipation,inthatweshowthatvotinghasadirecteffecton groupperformance,inthesensethatfollowersaremorelikelytofollowelectedleaders,andvoting hasalsoanindirecteffectmediatedbyelectedleadersbehavingmoreresponsibly.
Theturnaroundgameisavariationoftheminimumeffortcoordinationgame(BrandtsandCooper 2007). The basic minimum effort coordination game models an organization’s production with strong complementarities (Van Huyck et al. 1990). Individuals simultaneously choose among differenteffortlevelsandthelowestindividualeffortdeterminesgroupperformance.Thegamehas multipleequilibria,oneateveryeffortlevelthatissimultaneouslychosenbyallgroupmembers.To choose the highest effort is the most efficient, payoffͲdominant outcome for all group members. ThoughthereisatradeͲoff,thehighertheindividualeffortchoiceis,theriskieritbecomes,sincethe possibilityincreasesthatothergroupmemberschoosealowereffort.AveryriskͲaverseindividual couldsecurethelowestequilibriumpayoffbychoosingthelowesteffort,andthusavoidlossesfrom coordination failure. Coordination on the lowest effort is the usual outcome if preͲplay communication is not possible (Blume and Ortmann 2007). The turnaround game introduces organizational hierarchy to the minimum effort coordination game, but only after a failure to coordinateontheefficientequilibrium(BrandtsandCooper2007).Therefore,theturnaroundgame hastwoparts.Inthefirstpart,groupsofplayershavetocoordinaterepeatedlyinabasicminimum effortcoordinationgame.Usuallygroupsestablishahistoryoffailureandconvergetotheinefficient equilibrium,whichmakestheintroductionofaleadermoremeaningful(BrandtsandCooper2006a). In the second part, leadership is introduced. The task of the leader is to try to “turnaround” the group,whichrequiresacoordinatedchangefromtheinefficienttothemostefficientequilibrium.A turnaroundishardertoachieveifthegrouphasexperiencedfailure(DevetagandOrtmann2007).
ComparedtoBrandtsetal.(2015),ourfocusissolelyontheselectionprocess,andthuswedonot vary the tools leaders can use. We allow selected leaders only to use voluntary, nonͲbinding, free form preͲplay communication. Effective leaders in the game manage to convince others to take a certaincourseofaction,butthisismorethanpureauthorityoverothers.Ordersalonewouldnotbe sufficient to move followers. To be effective, leaders have to create the belief in followers that others will follow the leader too. Leaders have to make their team believe in success in order to motivateteammemberstochoosetheirmaximumeffort.
Whatmakesleaderscredible?Credibilityreferstothebeliefsofthefollowerswhoareled.Followers ofcredibleleadersbelievethattheleadershipis“appropriate,properandjust”(Tyler2006),sothat the leader has legitimate authority to make decisions, to send orders, and to expect followers to
follow her orders.45Moreover, also followers expect others to follow the leader. Therefore,
credibility roots in the legitimacy of the leader. A leader can prove her legitimacy by sending the relevant message content, and through a successful group history. However, besides legitimacy rooting in competence, the selection process itself can be important. Election might increase the legitimacy of the leader, and Brandts et al. (2015) already showed that election increases leader effectiveness. In their study, more legitimate leaders send more relevant messages that facilitate efficient coordination. Thus, in their paper, it is not the property of being elected that improves leader effectiveness. Rather, it is elected leaders who become more active in influencing their followers by sending more relevant messages, as if elected leaders would be more motivated to provethattheydeservedthefollowers’trust,andthattheresultoftheelectionwas“appropriate, properandjust”(Tyler2006).
One change we made compared to the design in Brandts et al. (2015) is that group members volunteer before the leader selection process takes place. Volunteering is an important feature presentinrealͲworldenvironments.Noteveryonehasapreferencetobecomealeader,andusually leaders are not forced into a leader position. Rather, they selfͲselect into candidacy and subsequentlyorsimultaneouslygetselectedfromsuperiorsorelectedfromemployees.Therefore, wethinkthataddingvolunteeringtothedesignisagoodchange,reproducingrealͲworldfeaturesin moredetailinthelaboratorygame.Weareawarethatvoluntaryleadershipitselfalreadyenhances performance compared to exogenous leadership (Rivas and Sutter 2011). However, this fact just underlines the importance of our findings: the democracy effect remains valid even after adding volunteeringtothedesign.Ourfindingsunderlinetheimportanceofemployeeparticipationinthe leaderselectionprocesswhichhasperformanceenhancingeffectsforthewholeorganization.