proceedings of the human factors and ergonomics society annual meeting-2003-pruitt-686-90.pdf

Upload: sarahmich13

Post on 14-Apr-2018

215 views

Category:

Documents


0 download

TRANSCRIPT

  • 7/27/2019 Proceedings of the Human Factors and Ergonomics Society Annual Meeting-2003-Pruitt-686-90.pdf

    1/6

    http://pro.sagepub.com/Ergonomics Society Annual Meeting

    Proceedings of the Human Factors and

    http://pro.sagepub.com/content/47/4/686The online version of this article can be found at:

    DOI: 10.1177/154193120304700405

    2003 47: 686Proceedings of the Human Factors and Ergonomics Society Annual MeetingNicole Pruitt, Jessica Bolton, Andy Kearney, Benjamin Swan and Kip Smith

    The Effects of Trust, Communication and Proximity on Team Problem Solving

    Published by:

    http://www.sagepublications.com

    On behalf of:

    Human Factors and Ergonomics Society

    can be fouProceedings of the Human Factors and Ergonomics Society Annual Meetingdditional services and information for

    http://pro.sagepub.com/cgi/alertsEmail Alerts:

    http://pro.sagepub.com/subscriptionsSubscriptions:

    http://www.sagepub.com/journalsReprints.navReprints:

    http://www.sagepub.com/journalsPermissions.navPermissions:

    http://pro.sagepub.com/content/47/4/686.refs.htmlCitations:

    What is This?

    - Oct 1, 2003Version of Record>>

    at Australia Catholic University on May 22, 2013pro.sagepub.comDownloaded from

    http://pro.sagepub.com/http://pro.sagepub.com/http://pro.sagepub.com/content/47/4/686http://pro.sagepub.com/content/47/4/686http://pro.sagepub.com/content/47/4/686http://www.sagepublications.com/http://www.sagepublications.com/http://www.hfes.org/http://pro.sagepub.com/cgi/alertshttp://pro.sagepub.com/cgi/alertshttp://pro.sagepub.com/subscriptionshttp://pro.sagepub.com/subscriptionshttp://www.sagepub.com/journalsReprints.navhttp://www.sagepub.com/journalsReprints.navhttp://www.sagepub.com/journalsPermissions.navhttp://www.sagepub.com/journalsPermissions.navhttp://www.sagepub.com/journalsPermissions.navhttp://pro.sagepub.com/content/47/4/686.refs.htmlhttp://online.sagepub.com/site/sphelp/vorhelp.xhtmlhttp://online.sagepub.com/site/sphelp/vorhelp.xhtmlhttp://pro.sagepub.com/content/47/4/686.full.pdfhttp://pro.sagepub.com/http://pro.sagepub.com/http://pro.sagepub.com/http://online.sagepub.com/site/sphelp/vorhelp.xhtmlhttp://pro.sagepub.com/content/47/4/686.full.pdfhttp://pro.sagepub.com/content/47/4/686.refs.htmlhttp://www.sagepub.com/journalsPermissions.navhttp://www.sagepub.com/journalsReprints.navhttp://pro.sagepub.com/subscriptionshttp://pro.sagepub.com/cgi/alertshttp://www.hfes.org/http://www.sagepublications.com/http://pro.sagepub.com/content/47/4/686http://pro.sagepub.com/
  • 7/27/2019 Proceedings of the Human Factors and Ergonomics Society Annual Meeting-2003-Pruitt-686-90.pdf

    2/6

    THE EFFECTS OF TRUST, COMMUNICATION AND PROXIMITY ON TEAMPROBLEM SOLVINGNicole Pruitt, Jessica Bolton, Andy Kea mey , Benjamin SwanKansas State University

    Kip SmithLinkoping U niversityfinstitute of TechnologyThis paper discusses an experiment on the interaction of trust, communication , an dproximity on team problem solving. The research is motivated by the US Armys plan tochange the organizational structure of the combat infantry from a close-knit brotherhoodto an alliance of high-tech autonom ous individuals. We are interested in ascertaining thepossible consequences this transition may have on teamwork in a context that generalizesto the Armys mission. In our experiment, comm unication and proximity were fullycrossed in a 2x2 design. Trust at three levels - rust, distrust, and betrayal - was asubject variable revealed by play of an economic gam e. We find a significant main effectfor mode of communication: teams that comm unicate freely work more effectively.Further, we find a significant main effect for trust. When team mem bers trust each other,they work more efficiently. The lack of a significant interaction between proximity andcommunication reveals that the quality of team problem solving is not sensitive to themode of comm unication. Team s that coordinated their activity using a radio performedno worse than teams that worked face-to-face. This finding bodes well for organizationsthat rely on remo te comman d and control.

    INTRODUCTIONThere is a long tradition of research o n the issue of

    comm unication and trust and how they interact to affectindividual and team performance (e.g., LatanC, 1981;Milgram, 1974). This paper contributes to that traditionby testing the im pact of comm unication, trust, andproximity on two-person team performance in a series oflaboratory tasks.The research discussed here was triggered by thedevelopment of the US Armys O bjective Force Warrior,more familiarly known as the Army of One. TheArmy is changing the organizational structure of thecombat strike unit from larger, more costly troops to aone-person format in which the soldier receivescomm and and control by means of radio transmission.We are interested in ascertaining the possibleconsequences this transition may have on team work in acontext that generalizes to the Armys m ission. Does thetransformation from a close-knit brotherhood to analliance of high-tech autonomous individuals haveconsequences on mission success? In particular, what isthe impact of the change in mode of communication

    from face-to-face to radio transmission on teamefficiency and effectiveness?In previous eras of warfare the bond formed bysoldiers in their unit was a motivator. The face-to-facecomm unication aided assurance of the completion oforders. With the transformation to technology-drivenbattlefields, a smaller unit of soldiers will be given theresponsibiiity to carry o ut a m ission previously tasked toa larger unit. These soldiers will have to rely and placetheir trust in technology rather than in face-to-facecontact with each other and their commanders.Our research draws upon social impact theory(LatanC, 1981) to explain th e interaction betweencomm unication and trust on team performance. Thistheory suggests that the presence (or absence) of anaudience, or co-participant, influences individual andteam behavior. A leader who is on-site andcomm unicating directly with subordinates has a greaterimpact on their behavior than a leader who is distant orcomm unicating electronically (LatanC& LHerrou,1996). Empirical work supporting social impact theoryhas found that the possibility of being evaluated by co-participants may also alter behavior (LatanC & Wolf,1981). Social impact theory leads us to expect

    PROCEEDINGS of the HUMAN FACTORS AND ERGONOMICS SOCIETY 47th ANNUAL MEETING2003 68

    at Australia Catholic University on May 22, 2013pro.sagepub.comDownloaded from

    http://pro.sagepub.com/http://pro.sagepub.com/http://pro.sagepub.com/http://pro.sagepub.com/
  • 7/27/2019 Proceedings of the Human Factors and Ergonomics Society Annual Meeting-2003-Pruitt-686-90.pdf

    3/6

    erformance will be better when dec ision makers arecolocated rather than dispersed.We expect our research to show that three factorsinfluence team performance: comm unication mode,degree of proximity, and category of trust (trust, distrust,or betrayal). First, we expect performance in adistributed problem-solving task to improve ifparticipants are permitted to commun icate freely. Freecomm unication can be either direct or mediated. Freetalk is direct when spoken face-to-face. Mediated freetalk occurs over electronic links. We predict that teamperformance will be better und er both free-talkconditions than in the conditions in which w henparticipants are not allowed to com municate.In addition, we expect trust to influence teamperformance. Trust necessarily involves an interactionbetween two or mo re individuals or groups (Costa, Roe,& Taillieu, 2001). For trust to exist the party that actsfirst (or whose action is revealed first) must exhibittrusting behavior and the party that responds m ust proveto be trustworthy. Following Rousseau et al. (1 998), wedefine trusting behavior as the willingness to bevulnerable to the actions of others (the second party,those who are trusted) based upon positive expectationsof the others behavior. As such, trust is more thancooperation; it is cooperation given a palpablelikelihood of (opportunity) loss - risk - and anexpectation of positive reciprocation (Lewicki,McAllister,& Bies, 1998; Yates & Stone, 1992).Trustworthiness supports behavior that honors the firstpartys implicit expectation of positive reciprocation. Itoften, but not necessarily, incurs some m easure ofopportunity loss.

    A lack of trust can be m anifested in two distinctways, by either distrust or betrayal. Distrust is mark edby the absence of trusting behavior by th e party to actfirst while betrayal is marked by th e absence o ftrustworthy behavior by the party that responds(Burnham, McCabe, & Smith, 2000). If our participantsare both trusting and trustworthy, they a re more likely toperform w ell as a team than if they distrust each other orexperience betrayal.In addition to studying the interaction ofcomm unication, proximity and trust o n teamperformance, we are interested in their impact onindividual cognitive performance. We have no apriorireason to expect that these factors influence individualperformance. In particular, we expect to observe nodetrimental effect of communication m ode or trustcondition on individual cognitive perform ance.Accordingly, we expect to find a dissociation betweenour measures of team and individual Performance.

    METHODParticipants

    To test these hypotheses, we asked 72 pairs ofparticipants (18 female pairs, 23 m ale pairs, and 3 1mixed pairs; mean age 19, mode 18, maximu m 48) toengage in three games across a comp uter network.Participants were drawn from a population of generalpsychology students wh o received class credit forparticipation. All participants signed informed consentforms and were treated according to APA guidelines.Design

    The experiment manipulated proximity andcomm unication in a fully crossed 2 X 2 design with fourtrue experimental levels: colocated with free talk,colocated with no talk, remote with free talk, and remotewith no talk. In the colocated conditions, participants satside-by-side in the same room. In the remote conditions,they sat in separate rooms. In the no-talk treatmentsthey were not allowed to comm unicate at all. In thecolocated free-talk co ndition, they are allowed tocomm unicate freely. In the remote free-talk condition,they comm unicated using two-way radios.four conditions made choices th at revealed levels oftrust. Their choices self-selected the pair into one ofthree categories: trust, distrust, or betrayal. This quasi-variable serves as the third factor in our design.

    As part o f the experiment, pairs of participants in all

    Procedureof four conditions defined by two levels ofcomm unication (free and none) and two levels ofproximity (colocated and remo te) and asked to play threecomputer games. The first was a test of individualcognitive processing, the m ental rotation task of S hepardand Metzler (1 97 1) . The second task was the first of thetwo-person team tasks, the extensive form econom ictrust game of McC abe, Rassenti, and Smith (1 996),shown in Figure 1. The teams then played a variant ofthe classic Hobbits and Orcs river-crossing problem(Greeno, 1 974). Finally, each participant individuallyperformed a second trial of the m ental rotation task.The mental rotation task, an analo g coding approachto studying perception of m ental imagery, presented twothree-dimensional objects on a computer screen.Participants were asked to identify whether the objectswere the same or different (mirror images). To respond,the participant pressed the Y key to answer Yes tothe question Are they the same? and the N ey toanswer No. The depend ent variables were reaction

    Pairs of participants were random ly assigned to one

    PROCEEDINGS of the HUMAN FACTORS AND ERGONOMICS SOCIETY 47th ANNUAL MEETING2003687

    at Australia Catholic University on May 22, 2013pro.sagepub.comDownloaded from

    http://pro.sagepub.com/http://pro.sagepub.com/http://pro.sagepub.com/http://pro.sagepub.com/
  • 7/27/2019 Proceedings of the Human Factors and Ergonomics Society Annual Meeting-2003-Pruitt-686-90.pdf

    4/6

    This task was administered bothin a within-subjectsrepeated-play design. Learning shouldacross repetitions.suggest that the intervening teamindividual cognitiveThe software platform for the mentalotation task was SuperLab Pro (0 edrus Corporation).PIaye rs Outcomes for& turns Player 1, Player 2

    $1 , $9Figure 1 - The extensive-form trust game used as thestimulus in the quasi-man ipulation of trust. Thenumbers in the squares represent Player 1 and Player 2.Player 1 has the first move. A move to the right is adistrusting move that nets both players $2. A movedown is a trusting move that passes play to Player 2. Amove right by Player 2 is a trustworthy m ove that netsPlayer 1 $4 and Player 2 $5. A move down by Player 2betrays Player 1s rust by awarding $9 to Player 2 butonly $1 to Player 1.

    The first team gam e was the trust game show n inFigure I . Participants played for US dollars that werereceived immed iately upon completing the game. In thisgame, participants randomly assigned the role of player1 make a decision to move right or to move down.Moving to the right ends the gam e, gives both players$2 , and places the team in the distrust condition.Moving down passes play to player 2 and is an explicitstatement of trust. By moving do wn, player 1 is trustingplayer 2 to move to the right, netting both a largerpayoff. If Player 2 is given the chance to play, he or sh emay choose to move right or down. Mo ving right netsplayer 1 $4 and player 2 $ 5 , ends the game, and placesthe team in the trust (and trustworthy) condition.Mov ing down is an explicit betrayal of player 1strust.This move nets player 2 $9 while player 1 receives only$1. It also ends the gam e and places the team in thebetrayal condition. Each team played the game only

    once and, in the process, placed itself into one of threecategories of trust -trust (and trustworthy), distrust, andbetraya l. This self-selection is a participan t variable andthe resulting treatments are quasi-experimental variables.The design is accordingly 2 x 2 x 3 (comm unication byproximity by trust). We did not collect subjectiveratings of trust to avoid inducing response b ias.and Orcs river-crossing problem. Th e game starts withthree Hobbits and three O rcs on one side of a river.They want to cross the river using a small boat. Thegoal of game is to move them all safely across the river.Three rules constrain play: (1 ) the Orcs may notoutnumber the Hobb its anywhere at any time (or else theOrcs eat the Hobb its), (2) a character must be in the boatto get it across the river, and ( 3 ) no more than twocharacters may be in the boat at one time. In our variantof the game, one player is responsible for loading andlaunching the boat from the left bank, the other playerfor loading and launching the boat from the right bank.Players are given 100 moves or 45 m inutes to completethe game. Play in the Hobbits and Orcs game gives usthree measures of team problem solving: the amount oftime taken, the number of moves, and the number ofillegal moves (e.g., feeding a Hobbit to the Orcs). Theplatform for both the trust game and the Hobbits andOrcs river-crossing problem was Java script written forthis experiment.

    The second team task w as a variant of the H obbits

    Analysisrotation task in which the individual is the unit ofanalysis. We used a within subject t-test to analyze themental rotation data and 2 x 2 x 3 ANOV A to analyzethe dependent measures from the Hobbits and Orcsgame.

    The team is the unit o f analysis except for the mental

    RESULTSBy manipulating comm unication and proximity, andby allowing participants to reveal trusting behavior atthree levels, and by measuring performance in the

    Hobbits and Orcs game, we test whethercomm unication, proximity, and trust interact to influenceteam performance in a cooperative problem-solving task.A box plot analysis of the performance data from the 72teams was conducted. Seven outliers were revealed andeliminated from all analyses.across the four conditions of comm unication andproximity to show the ma in effect of trust onperformance. The ANO VA reveals that trust issignificant F(2, 53) = 4.52, p < ,016. Teams that were

    Figure 2 shows the time on task d ata aggregated

    PROCEEDINGS of the HUMAN FACTORS AND ERGONOMICS SOCIETY 47th ANNUAL MEETING2003 688

    at Australia Catholic University on May 22, 2013pro.sagepub.comDownloaded from

    http://pro.sagepub.com/http://pro.sagepub.com/http://pro.sagepub.com/http://pro.sagepub.com/
  • 7/27/2019 Proceedings of the Human Factors and Ergonomics Society Annual Meeting-2003-Pruitt-686-90.pdf

    5/6

    ng (and trustworthy) successfully co mpleted theess time. N o other main effects or interactionsime on task.

    1 2 0 0 -1000 - T T

    0Trust D s rust Betrayal

    Figure 2 -Graph sh owing the statistically significantmain effect of trust on the time for task com pletion.Errors bars are one standard deviation. Trusting teamstook less time than teams that distrusted or betrayed.Figure 3 shows the data for the total number of

    moves aggregated across proximity and trust to show themain effect of comm unication on performance. TheANOV A reveals that commun ication is significant F( 1,53) = 8.69, p < .005. Team s that conversed freely (eitherface-to-face or over the radio) completed the task m oreeffectively. The three-way interaction is also significantF(2, 53) = 3.77, p < .030. Team s in the two no-talkconditions who trusted and teams in the remote no-talkcondition who experienced distrust needed mo re moves.Teams that experienced betrayal were equally(in)efficient across all conditions.Figure 4 hows the data for the num ber of illegalmoves showing the main effect of communication onerrors made. The ANO VA reveals that communicationis significant F( l, 53) = 10.79, p < .002. Teams thatcomm unicated freely made fe wer errors. No other maineffects or interactions were significant for illegal moves(errors made).Of the 130 participants, 125 were both faster andmore accurate in the second trial of the m ental rotationtask. This improvement reflects learning. As expected,the manipulations of comm unication, trust, and

    Totalnumberof moves30

    2 5

    20

    1 5

    1 0

    5

    0

    T

    Free talk No talk

    Figure 3 -Graph show ing the statistically significantmain effect of communication on total number of movestaken to complete the game. Errors bars are onestandard deviation. Team s that were allowed to talkmade few er errors.Numberof illegalmoves

    Free talk No talkFigure 4- Graph sh owing the statistically significantmain effect of communication on errors. Errors bars areone standard deviation. Team s that were allowed to talkmade fewer errors.

    PROCEEDINGS of the HUMAN FACTORS AND ERGONOMICS SOCIETY 47th ANNUAL MEETING2003689

    at Australia Catholic University on May 22, 2013pro.sagepub.comDownloaded from

    http://pro.sagepub.com/http://pro.sagepub.com/http://pro.sagepub.com/http://pro.sagepub.com/
  • 7/27/2019 Proceedings of the Human Factors and Ergonomics Society Annual Meeting-2003-Pruitt-686-90.pdf

    6/6

    n this individual cognitiverocess. Accordingly, we have documented aissociation of the effect of trust, com munication on, andproximity on team and individual performance.DISCUSSION

    The performance data indicate that trusting teamsare faster at completing the task. Further, teams thatcomm unicate complete the task m ore effectively andmake fewer errors. It comes as no surprise that opencomm unication improves performance. However, theeffect of trust on team efficiency is relativelyunexpected. Both colocated and distributed teamscompleted the problem-solving task m ore efficiently ifthey previously experienced trust. Distrust and betrayalare detrimental to team efficiency.In contrast, our data show n o main effect forproximity. Team s that were located in different roomswere both as efficient and as effective as team s thatworked face-to-face. This bodes well for the design ofdistributed work groups. For distributed groups to workwell, however, they must trust each other andcomm unicate freely.The long-term goal of this study is to identify anydeleterious effect on cognitive performance that m ightbe induced or exacerbated by the erosion of trust thatmay be induced by the US Armys planned transitionfrom cohesive fighting units to ad-hoc assemblies ofspecialists. This study suggests that team problemsolving suffers in the absence of trust. It also sugge ststhat the extensive form trust game is an effectiveexperimental tool for eliciting trust, distrust, andbetrayal. We plan to continue to use the trust game toidentify situations in w hich cogn itive performancesuffers in the absence of trust. We also plan to conductexperiments that introduce manipulations that areexpected to m itigate those effects.

    ACKNOWLEDGEMENTSThe views expressed in this work are those of the

    authors and do not necessarily reflect official Armypolicy. This work was supported by the DoDMultidisciplinary University Research Initiative (MUM)program administered by the A rmy R esearch Officeunder grant DAAD 19-0 1 1-062 1. Dr. ElmarSchmeisser is the Technical Monitor. The Java scriptsfor the trust game and the H obbits and Orcs river-crossing problem were written by Blaine W illiams.

    REFERENCESBurnham, T., McCabe, K., and Smith, V. L. (2000).Friend-or-foe intentionality priming in an extensive formtrust game. Journal of Econom ic Behavior andOrganization 43 , 57-73.Costa, A. C., Roe, R. A., & Taillieu, T. (2001).Trust within teams: The relation with performanceeffectiveness. European Journal of Work andOrganizational Psychology 10 (3), 225-244.Greeno, J. G. (1974). Hobbits and Orcs:Acquisition of a sequential concept. Co mi tivePSyChOlO~ (2), 270-292.LatanC, B. (1981). The psychology of social impact.Am erican Psychologist 36, 343-356.LatanC, B., & LHerrou, T. (1 996). Spatialclustering in the conformity game: Dynam ic socialimpact in electronic groups. Journal of Personality and

    Social Psychology 70 (6 ) , 1218-1230.LatanC, B., & Wolf, S. (1981). The social impact ofmajorities and minorities. Psychological Review 88 (S),438-453.Lewicki, R. J., M cAllister, D. J. , & Bies, R. J.(1 998). Trust and distrust: New relationships andrealities. Academy of Management Review 23 (3), 438-458.McCabe, K. A., Rassenti, S. J., & Smith, V. L.(1996). Gam e theory and reciprocity in some extensiveform experimental games. Proceedings of the NationalAcademy o f Sciences, USA 93 , 13421 13428.Milgram, S. (1 974). _Obedience o Authority. NewYork: Harper & Row.Rousseau, D. M., Sitkin, S. B., Burt, R. S.,&Camerer, C . (1 998). Academ y of Management Review23 (3), 393-404.Shepard, R. N . & Metzler, J. (1971). Science 171Yates, J. F., & Stone, E. R. (1992). The riskconstruct. In J. F. Yates (Ed.) Risk Taking Behavior.New York: Wiley.

    (3972), 701-703.

    PROCEEDINGS of the HUMAN FACTORS AND ERGONOMICS SOCIETY 47th ANNUAL MEETING2003 690

    at Australia Catholic University on May 22, 2013pro.sagepub.comDownloaded from

    http://pro.sagepub.com/http://pro.sagepub.com/http://pro.sagepub.com/http://pro.sagepub.com/