Group Dynamics: Theory, Research, and Practice Editor: David Marcus, PhD

Journal Information Current issue feed Set PsycALERT: New issue update by email

2014 Volume 18, Issue 2 (Jun)

Teamwork Competency Test (TWCT): A step forward on measuring teamwork competencies. Pages 101­121 Aguado, David; Rico, Ramón; Sánchez­Manzanares, Miriam; Salas, Eduardo

Abstract | Full Text PDF | Full Text HTML | Permissions |

The effect of different phases of synchrony on pain threshold. Pages 122­128 Sullivan, Philip J.; Rickers, Kate; Gammage, Kimberley L.

Abstract | Full Text PDF | Full Text HTML | Permissions |

Reactions to social inclusion and ostracism as a function of perceived in­group similarity. Pages 129­137 Sacco, Donald F.; Bernstein, Michael J.; Young, Steven G.; Hugenberg, Kurt

Abstract | Full Text PDF | Full Text HTML | Permissions |

Single­case designs in group work: Past applications, future directions. Pages 138­158 Macgowan, Mark J.; Wong, Stephen E.

Abstract | Full Text PDF | Full Text HTML | Permissions |

Team task conflict resolution: An examination of its linkages to team personality composition and team effectiveness outcomes. Pages 159­173 O’Neill, Thomas A.; Allen, Natalie J.

Abstract | Full Text PDF | Full Text HTML | Permissions |

The impact of group norms and behavioral congruence on the internalization of an illegal downloading behavior. Pages 174­188 Sansfaçon, Sophie; Amiot, Catherine E.

Abstract | Full Text PDF | Full Text HTML | Permissions | Group Dynamics: Theory, Research, and Practice © 2014 American Psychological Association 2014, Vol. 18, No. 2, 101–121 1089-2699/14/$12.00 DOI: 10.1037/a0036098 Teamwork Competency Test (TWCT): A Step Forward on Measuring Teamwork Competencies

David Aguado Ramón Rico Autonomous University of Madrid and Knowledge Autonomous University of Madrid Engineering Institute

Miriam Sánchez-Manzanares Eduardo Salas Carlos III University of Madrid University of Central Florida

The Teamwork Knowledge, Skill, Ability Test (TWKSAT), designed to assess team- work competencies, has been widely used in both applied and academic contexts. However, studies have brought to light a number of reliability problems in the test. In this article, we describe 3 studies that (a) examined the functioning of the TWKSAT (N ϭ 135); (b) investigated a new measure, the Teamwork Competency Test, including its metric properties (N ϭ 120); and (c) analyzed the convergent and predictive validity of the Teamwork Competency Test compared with the TWKSAT (N ϭ 91). Based on our results, we conclude that the TWKSAT does not adequately reflect the initial substantive model and has limitations with regard to reliability. The Teamwork Com- petency Test improves the TWKSAT by enhancing reliability, content validity, and substantiating the dimensional structure of the test.

Keywords: teamwork test, teamwork competencies, team performance

Work teams have responded effectively to the out of a mere “group of experts” (Lawler & global challenges of our times, providing orga- Worley, 2006). nizations with the benefits of adaptability, pro- Work teams bring to their mission much ductivity, and creativity above and beyond the more than just the behaviors directly related to contributions individuals can make on their own the task at hand. Their members need to interact (Kozlowski & Ilgen, 2006; Salas, Sims, & and cooperate if they are to synchronize effec- Burke, 2005). However, teamwork requires sev- tively (Salas et al., 2005; Salas, Sims, & Klein, eral competencies to allow its members to ef- 2004), which requires a specific set of behaviors fectively integrate their contributions, function that further the attainment of team goals. Em- as a unified whole, and make an “expert team” pirical evidence shows that measures of knowl- edge and skills related to both a specific task and teamwork predict individual performance in work contexts (e.g., McClough & Rogelberg, This article was published Online First May 12, 2014. 2003; Schmidt & Hunter, 1983). At the same David Aguado, Autonomous University of Madrid, Ma- This document is copyrighted by the American Psychological Association or one of its allied publishers. drid and Knowledge Engineering Institute; Ramón Rico, time, teams formed by people with strong team- This article is intended solely for the personal use of the individualAutonomous user and is not to be disseminated broadly. University of Madrid; Miriam Sánchez- work competencies display a specific range of Manzanares, Carlos III University of Madrid; Eduardo Sa- behaviors, including the use of integrative las, University of Central Florida. (win–win) as opposed to distributive (win–lose) This research was partially supported by the Office of Naval Research Collaboration and Knowledge Interoper- negotiating strategies. These findings are rele- ability Program and Office of Naval Research Multidis- vant as teams are often affected by interpersonal ciplinary University Research Initiative Grant conflicts and faulty cooperation, which hinder N000140610446. optimum performance (Hackman, 2002). The Correspondence concerning this article should be ad- dressed to David Aguado, Autonomous University of Ma- availability of an appropriate measure of team- drid, Ciudad Universitaria de Cantoblanco, Madrid 28049 work competencies may help minimize these Spain. E-mail: [email protected] problems (Stevens & Campion, 1994, 1999).

101 102 AGUADO, RICO, SÁNCHEZ-MANZANARES, AND SALAS

Over recent decades, scholars increasingly Clough & Rogelberg, 2003; Stevens & Cam- have underlined the importance of teamwork pion, 1999). competencies for the effectiveness of groups However, studies employing the TWKSAT and work teams (e.g., Cannon-Bowers & Bow- have reported low levels of reliability (Athana- ers, 2011; Cannon-Bowers, Tannenbaum, Salas, saw, 2003; Chen et al., 2004; McClough & & Volpe, 1995; Chen, Donahue, & Klimoski, Rogelberg, 2003). Despite the TWKSAT’s ca- 2004; Leach, Wall, Rogelberg, & Jackson, pacity to predict individual performance in team 2005; Peeters, van Tuijl, Rutte, & Reymen, tasks, there are no studies addressing the causes 2006). In particular, Stevens and Campion of the test’s low reliability or explaining its (1994, 1999) proposed the Teamwork Knowl- dimensional structure. To resolve the problems edge, Skill, Ability Test (TWKSAT), which affecting the assessment of teamwork compe- identifies and measures five transportable team- tencies, we (a) examined the metric properties work competencies that are common to differ- of the TWKSAT items, as well as their content ent types of teams (see Figure 1) and have been validity and dimensional structure, and (b) pro- related to different performance criteria (Mc- pose a new measure based on the Stevens and

I. Interpersonal KSAs

1. Conflict Resoluon

Recognize team conflict.

Recognize type and source of conflict and implement conflict resoluon strategies.

Employ integrave (win-win) negoaon strategies.

2. Collaborave Problem Solving

Idenfy situaons requiring parcipave group problem solving.

Recognize the obstacles to collaborave group problem solving and implement correcve acons.

3. Communicaon

Understand communicaon networks and ulize decentralized networks to enhance communicaon.

Communicate open and supporvely.

Listen in a non-evaluave way and use acve listening techniques.

Maximize consonance between non-verbal and verbal messages, and recognize and interpret the non-verbal messages of

others.

Engage in ritual greengs and small talk.

II. Self-management KSAs This document is copyrighted by the American Psychological Association or one of its allied publishers. 4. Goal Seng and Performance Management This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Establish specific, challenging and accepted team goals.

Monitor, evaluate, and provide feedback on both overall team performance and individual performance.

5. Planning and Task Coordinaon

Coordinate and synchronize acvies, informaon, and task interdependences.

Establish task and role expectaons of individual team members, and ensure proper balancing of workload in the team.

Figure 1. The Stevens and Campion model. TEAMWORK COMPETENCY TEST 103

Campion (1994, 1999) model, analyzing its re- list) and self-management competencies (repre- liability and validity as a measure of teamwork sented by the last two items). This distinction is competencies. Although recently some tests based on the idea that team effectiveness de- have been proposed to capture knowledge on pends on its members’ ability to both manage teamwork (e.g., boundary-spanning activities, their relations appropriately and direct their ac- Marrone, Tesluk, & Carson, 2007; team roles, tions to carry out the tasks assigned by the Mumford, Van Iddekinge, Morgeson, & Cam- organization. pion, 2008), it is still necessary to develop tests Perhaps the most important contribution capable of assessing teamwork competencies to made by Stevens and Campion (1999) is the predict employees’ performance in teams and construction of the TWKSAT, which provides thus design and develop high-performance both practitioners and academics with an instru- teams. We aimed to improve the assessment of ment to measure key teamwork competencies. teamwork-related competencies in work set- The available evidence suggests that the tings, beyond the TWKSAT, one of the most TWKSAT has considerable predictive validity. popular measures in the field. Thus, the TWKSAT measure of employees in real work teams correlated with their perfor- The TWKSAT as a Measure of Teamwork mance in the team as evaluated by both super- Competency visors (correlations of between .23 and .52) and colleagues (correlations of between .21 and .34; Competencies are defined as the underlying McClough & Rogelberg, 2003; Stevens & Cam- characteristics integrated with an individual’s pion, 1999). Furthermore, Chen et al. (2004) knowledge, skills, and abilities that are causally found that the TWKSAT was sensitive to related to a referential criterion of effective changes in the individual competency of uni- and/or superior action in a specific job or situ- versity students after participation in a training ation (Spencer & Spencer, 1993). In particular, program designed to develop their teamwork researchers have stressed the key role of team- competencies. work competencies for the effectiveness of However, other studies in which the work teams (e.g., Cannon-Bowers et al., 1995; TWKSAT was used have also consistently re- Chen et al., 2004). ported low reliability of the measure. Stevens Cannon-Bowers et al. (1995) identified eight and Campion (1999) originally found reliability central teamwork competencies, namely adapt- of .80 (internal consistency), but this appears to ability, shared understanding of the situation, be an overestimate (McClough & Rogelberg, performance monitoring and feedback, leader- 2003). In fact, the alpha coefficients in studies ship, interpersonal relations, coordination, com- employing the TWKSAT were consistently munication, and decision making. These schol- lower. For example, McClough and Rogelberg ars also made a key contribution by defining (2003) found an alpha coefficient of .59, and competencies on the basis of two dimensions: Chen et al. (2004) found a coefficient of .64 task competencies versus team competencies before training in teamwork skills and .82 after and specific competencies versus general com- training. Athanasaw (2003) obtained coeffi- petencies. Given their relationship with perfor- cients of .66 for the complete scale and between mance in different teamwork situations, general .25 and .48 for each of the five competencies, team competencies (i.e., transportable compe- and Leach et al. (2005) found a coefficient of This document is copyrighted by the American Psychological Association or one of its allied publishers. tencies) are of particular interest to organiza- .70. As McClough and Rogelberg point out, This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. tions. With this in mind, Stevens and Campion however, the TWKSAT was designed from a (1994, 1999) identified five transportable team- multidimensional standpoint and other situa- work competencies that are common to differ- tional judgment tests share the low alpha coef- ent types of teams: (1) conflict resolution, (2) ficients found to date. In this type of test, inter- collaborative problem solving, (3) communica- nal consistency-based reliability measures tion, (4) goal setting and performance manage- should be complemented with test–retest esti- ment, and (5) planning and task coordination. mations (Clevenger, Pereira, Wiechmann, These competencies are grouped in two more Schmitt, & Harvey, 2001; McDaniel, Morge- general dimensions termed interpersonal com- son, Finnegan, Campion, & Braverman, 2001). petencies (defined by the first three items in the However, Chen et al. reported a test–retest cor- 104 AGUADO, RICO, SÁNCHEZ-MANZANARES, AND SALAS

relation of .60 in the control group for their Your team wants to improve the quality and flow of the study. Overall, the available evidence suggests conversations among its members. Your team should: (A) use comments that build upon and connect to what that the reliability of the TWKSAT could be others have already said; (B) set up a specific order for improved, despite the test’s predictive capacity. everyone to speak and then follow it; (C) let team No studies have been carried out to date to members with more to say determine the direction and identify the reasons for the TWKSAT’s low topic of conversation; (D) do all of the above. reliability, either by examining the metric char- The original version of the TWKSAT is writ- acteristics of the items or the degree to which ten in English. We therefore applied the back- the different contents proposed by Stevens and translation procedure (Brislin, 1970) to use the Campion (1994) are actually present in the scale in Spanish and ensure the equivalence of TWKSAT. Nor has any attempt been made to items (Gibson, 1999). Three certified translators explain the dimensional structure of the and a group of specialists from bilingual teams TWKSAT, although this is crucial to identify- in the United States and Spain were involved in ing the scoring obtained from the test. this process. The materials forming the test Overall, our main objective in this study was were first translated into Spanish by the three to present a new measure of the teamwork- translators. The Spanish version of the test was related competencies based on Stevens and then translated back into English by the bilin- Campion’s (1994) original model. To this end, gual specialists. Finally, the back-translation we conducted three empirical studies. The first was reviewed by the research team to ensure was devoted to analyzing the metric properties that the meaning of the items was consistent in of the items and the content validity and dimen- the different translations. As a final result, we sionality of the TWKSAT. In the second study, obtained a scale in Spanish equivalent to the we proposed a new measure, testing the im- English-language original. provements made on reliability and dimension- Participants and procedure. The partici- ality in a sample of university students. Finally, pants in the study were 135 students from dif- in the third study, we tested the validity of the ferent disciplines (78% information technology new measure in a sample of professional em- engineering and 22% psychology) at a public ployees. university in Madrid. Men made up 71% of the total sample, and the average age was 23 years Study 1 (SD ϭ 0.93). All of the participants were from Spain and had Spanish as their mother tongue. The aim of this first study was to analyze the Participation in the study was mandatory for items composing the TWKSAT, examining the students, being part of a program of prac- their content and dimensionality. To begin, we tices in different subjects. Participants signed a present the results of the descriptive analysis consent form regarding their participation in the based on the impact of each item on reliability. study. They could voluntarily decline to be en- Following the strategy employed by Rovinelli rolled in the research by not providing their and Hambleton (1977), we then go on to ana- records without any adverse effect on their class lyze the test items in terms of content validity. grades. At the beginning of the academic se- Next, the dimensionality of the test is examined mester, the researchers administered the using factor analysis. TWKSAT to the students as a part of their class

This document is copyrighted by the American Psychological Association or one of its allied publishers. requirements. The objectives and reasoning be-

This article is intended solely for the personal use of the individualMethod user and is not to be disseminated broadly. hind the study were explained by the professors at the end of the course. Each participant was TWKSAT translation. The TWKSAT also given a brief report with the scores ob- comprises 35 multiple-response items, which describe different situations that may arise 1 within a work team. Respondents answer the 1 The TWKSAT is a commercial test and we therefore do items by indicating how they would act in each not provide the content of the items included in the test so as situation. The questionnaire evaluates the five to protect the intellectual property rights of its authors. The test can be acquired at http://www.vangent-hcm.com/Solutions/ teamwork competencies identified by Stevens SelectionAssessments/SkillsAbilitiesAssessments/. We use the and Campion (1994). An example item is the original numbering of the items in our presentation of results to following (Stevens & Campion, 1999): facilitate understanding. TEAMWORK COMPETENCY TEST 105

tained on the test and the meaning of each of the asymmetry, kurtosis, and discrimination index teamwork competencies. for each item on the scale, as well as the Cron- The validity of the TWKSAT content was bach’s alpha estimated for all 35 items together. analyzed by three experts in the field of team- As shown, the range of difficulty varies between work. All raters had more than 10 years’ expe- .08 (Item 26) and .93 (Item 6). With regard to rience in conducting team-based research and the discrimination index, four items (7, 12, 27, consulting. The experts analyzed the TWKSAT and 35) are negatively correlated with the scale items as described in the Results section. and another series of items (2, 4, 10, 11, 15, 17, Results 24, and 25) display a positive but very low adjusted correlation (Ͻ.15). The total alpha for Description of items. Table 1 shows the the scale is .60. However, if the negatively mean (difficulty index), standard deviation, correlated items and those with a correlation of

Table 1 Descriptive Statistics for Study 1 Teamwork Knowledge, Skills, Attitudes Test Original Items

Discrimination index (corrected item-total Item Mean SD correlation) Skewness Kurtosis 01 0.59 0.49 .22 Ϫ0.80 Ϫ1.34 02 0.39 0.49 .03 0.68 Ϫ1.53 03 0.79 0.41 .18 Ϫ1.62 0.63 04 0.68 0.47 .01 Ϫ1.44 0.08 05 0.53 0.50 .17 Ϫ0.23 Ϫ1.94 06 0.93 0.25 .35 Ϫ5.39 27.02 07 0.49 0.50 Ϫ.07 0.02 Ϫ1.99 08 0.91 0.29 .37 Ϫ4.08 14.60 09 0.34 0.47 .19 0.25 Ϫ1.93 10 0.72 0.45 .13 Ϫ1.26 Ϫ0.40 11 0.50 0.50 .14 Ϫ0.46 Ϫ1.78 12 0.29 0.45 Ϫ.01 0.70 Ϫ1.49 13 0.71 0.46 .15 Ϫ1.01 Ϫ0.96 14 0.29 0.45 .22 0.69 Ϫ1.51 15 0.70 0.46 .10 Ϫ0.88 Ϫ1.22 16 0.11 0.32 .30 0.60 Ϫ1.63 17 0.69 0.46 .02 Ϫ0.79 Ϫ1.37 18 0.16 0.37 .31 0.15 Ϫ1.97 19 0.70 0.46 .30 Ϫ0.86 Ϫ1.25 20 0.76 0.43 .35 Ϫ1.62 0.63 21 0.56 0.50 .21 Ϫ1.06 Ϫ0.85 22 0.61 0.49 .29 Ϫ0.60 Ϫ1.63 23 0.50 0.50 .27 Ϫ0.33 Ϫ1.88 Ϫ Ϫ

This document is copyrighted by the American Psychological Association or one of its allied publishers. 24 0.40 0.49 .01 0.02 1.99 25 0.32 0.47 .02 0.43 Ϫ1.80 This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. 26 0.08 0.28 .24 0.92 Ϫ1.14 27 0.38 0.49 Ϫ.06 0.01 Ϫ1.99 28 0.68 0.47 .36 Ϫ1.18 Ϫ0.58 29 0.22 0.41 .15 1.15 Ϫ0.67 30 0.15 0.36 .41 0.13 Ϫ1.98 31 0.64 0.48 .15 Ϫ1.00 Ϫ0.99 32 0.09 0.29 .34 0.31 Ϫ1.89 33 0.61 0.49 .12 Ϫ0.40 Ϫ1.83 34 0.23 0.42 .10 0.76 Ϫ1.41 35 0.50 0.50 Ϫ.05 0.02 Ϫ1.99 Note. N ϭ 135. ␣ϭ.60. 106 AGUADO, RICO, SÁNCHEZ-MANZANARES, AND SALAS

less than .15 are discarded, the alpha scale rises was analyzed using Mplus (Muthén & Muthén, to .71. 2006). Following the substantive model pro- Content analysis. To examine the validity posed by Stevens and Campion (1994, 1999), of the test content, we employed the strategy we tested five models: (a) five-factor orthogonal proposed by Rovinelli and Hambleton (1977) in model, (b) two-factor orthogonal model, (c) which a group of experts expresses the degree to five-factor correlated model, (d) two-factor cor- which each item measures each of the previ- related model, and (e) one-factor model. ously defined evaluation objectives. Accord- Table 3 presents the fit indices established for ingly, three experts separately analyzed the 35 the different estimated models. The orthogonal items composing the test, indicating the extent solutions of the five-factor and two-factor mod- to which each item was representative of each els do not fit the data. Values in each index are of the teamwork subcompetencies defined by substantially below the usual standards. How- Stevens and Campion (1994) and relevant to the ever, when correlation between factors is al- measurement of the target. A score of 1 was lowed, the model fits the data better. Even so, assigned if the item was considered appropriate, the only adjustment index for which a good fit is Ϫ1 if not, and 0 if the expert in question was obtained is the root mean square error of ap- unsure. The resulting index will be 1 when all of proximation (Ͻ.05). A similar effect is ob- the experts concur that an item is appropriate in served with a single-factor model. view of the target measured. The experts were In addition, to deeply explore the dimension- also asked to make any general comments on ality of the items proposed, we conducted an the measurement items included. exhaustive exploratory factor analysis of the Table 2 summarizes the results. A number of different factorial solutions from one to seven items display a high level of item–target con- factors, including a review of both orthogonal gruence (Ͼ.80), other items show a congruence and oblique rotations. The results do not allow index of less than .50 for all of the measurement us to clearly label the factors according to the targets (27, 29, and 33), and Item 31 falls short Stevens and Campion model (1994). of the threshold value of .30 for inclusion in the Criterion validity analysis. To provide table (Thorn & Deitz, 1989). Table 2 also pro- further evidence of the criterion-related validity vides valuable information related to the degree of the TWKSAT, we correlated the scores in the to which the test items meet the proposed mea- TWKSAT against the criterion for individual surement objectives. There are no items associ- performance in a team task. A subsample of the ated with the subcompetencies A2 (“Recognize study (n ϭ 30; 56% women; average age ϭ 23 the type and source of conflict”), C1 (“Under- years) formed by the participants in an under- stand and use communication networks”), and graduate teamwork course was observed during D2 (“Monitor, assess and provide feedback on the resolution of a group decision-making task. individual and group performance”). Moreover, The participants were randomly assigned to the number of items that are congruent with the five-member teams. The task required the teams different measurement targets varies between a to generate effective measures to resolve traffic single highly congruent item (23) for A3 (“Use problems caused by improper parking on a uni- win–win strategies”) and four items (1, 4, 11, versity campus.2 The teams were allowed 30 and 19) for D1 (“Help set specific, challenging min to complete the task and were video- and accepted objectives”). This document is copyrighted by the American Psychological Association or one of its allied publishers. recorded for subsequent analysis. Both team Finally, all of the experts remarked in their This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. tasks and video recording were standard fea- comments that the test items allow assessment tures of the course, and the participants were of only the respondents’ knowledge, although therefore familiar with these procedures. An ad the original Stevens and Campion (1994) model hoc code (see Appendix A) was designed for refers to “teamwork knowledge, skills and abil- behavioral observation. The code categories de- ities.” scribe specific behaviors associated with the Dimensionality analysis. A confirmatory ϭ five teamwork competencies identified by Ste- factor analysis (N 135) was carried out on the vens and Campion (1994, 1999; e.g., “Request correlation matrix to examine the dimensional- ity of the test. As these are dichotomous (right/ wrong) items, the tetrachoric correlation matrix 2 The task materials are available on request. TEAMWORK COMPETENCY TEST 107

Table 2 Teamwork Knowledge, Skills, Attitudes Test Congruence Indexes

Item A1 A2 A3 B1 B2 C1 C2 C3 C4 C5 D1 D2 E1 E2 06 0.88 0.35 07 0.65 0.47 23 0.92 10 0.55 0.37 12 0.64 26 0.65 22 0.97 27 0.31 28 0.92 29 0.31 15 0.96 32 0.79 33 0.45 34 0.64 09 0.58 16 0.94 0.76 30 0.60 14 0.58 0.94 24 0.40 0.94 35 0.59 0.95 02 0.79 17 0.99 01 0.95 04 0.92 11 0.97 19 1.00 18 0.62 08 0.88 0.71 13 0.62 20 0.78 21 0.95 0.41 25 0.55 0.37 05 0.59 03 0.71 0.88 31 Note.A1ϭ The knowledge, skills, and attitudes (KSA) to recognize and encourage desirable, but discourage undesirable, team conflict; A2 ϭ The KSA to recognize the type and source of conflict confronting the team and to implement an appropriate conflict resolution strategy; A3 ϭ The KSA to employ an integrative negotiation strategy rather than the traditional distributive strategy; B1 ϭ The KSA to identify situations requiring participative group problem solving and to utilize the proper degree and type of participation; B2 ϭ The KSA to recognize the obstacles to collaborative group problem solving and implement appropriate corrective actions; C1 ϭ The KSA to understand communication networks and to utilize decentralized networks to enhance the communication where possible; C2 ϭ The KSA to communicate openly and supportively; C3 ϭ The KSA to listen nonevalu- atively and appropriately use active listening techniques; C4 ϭ The KSA to maximize consonance between nonverbal and verbal messages, and to recognize and interpret the nonverbal messages of others; C5 ϭ The KSA to engage in ritual greetings and small This document is copyrighted by the American Psychological Association or one of its allied publishers. talk, and a recognition of their importance; D1 ϭ The KSA to help establish specific, challenging, and accepted team goals; D2 ϭ

This article is intended solely for the personal use of the individualThe user and is not to beKSA disseminated broadly. to monitor, evaluate, and provide feedback on both overall team performance and individual team member perfor- mance; E1 ϭ The KSA to coordinate and synchronize activities, information, and task interdependencies between members; E2 ϭ The KSA to help establish task and role expectations of individual team members and to ensure proper balancing of workload in the team. To facilitate reading the indexes, indexes Ͻ.30 have been eliminated; indexes Ͼ.80 appear in bold italic type; and the greater index obtained for each item appears in bold type.

additional information from team members”). competency displayed by each of the team Two groups of five judges independently ana- members. All of the judges were postgraduate lyzed the recordings, noting the frequency of students who were blind to the objectives of the teamwork behaviors associated with a particular study and independent of the research group. 108 AGUADO, RICO, SÁNCHEZ-MANZANARES, AND SALAS

Table 3 tion networks”), and D2 (“Monitor, assess and Model Fit Indexes provide feedback on individual and group per- Model RMSEA CFI TLI formance”). The content analysis also revealed that certain items are not associated with any of Five factor (orthogonal) .082 .067 .008 the proposed measurement targets (Items 27, Two factor (orthogonal) .07 .324 .282 Five factor (correlated) .04 .776 .762 29, 31, and 33). In addition, the format of the Two factor (correlated) .038 .807 .792 items is designed to measure “knowledge,” but One factor .041 .774 .76 not skills or aptitudes. Finally, Stevens and Campion (1994) estab- N ϭ 135; RMSEA ϭ root mean square error of approxi- mation; CFI ϭ confirmatory fit index; TLI ϭ Tucker–Lewis lished 14 teamwork subcompetencies in their fit index. model, grouped into five competencies, which are, in turn, integrated in two dimensions. How- ever, our results indicate a structure that tends They were trained in the use of the code and toward unidimensionality. underwent a trial period to align criteria and To sum up, the results of this study provide discuss inconsistencies. The mean interjudge an explanation of the low reliability indices reliability for the different competencies was reported in previous research and suggest how a acceptable (␬ϭ.87); therefore, we computed more reliable measurement could be obtained the mean of their assessments to obtain a single from the TWKSAT without affecting its predic- score for each competency. Finally, we com- tive capacity. In line with prior studies using the puted the mean of the scores for each individual original English version of the TWKSAT (Chen and the five competencies, as in the case of the et al., 2004; McClough & Rogelberg, 2003; questionnaire, to obtain a total teamwork com- Stevens & Campion, 1999), the Spanish adap- tation of the test used in our study reveals sim- petency score. As expected, the TWKSAT score ϭ Ͻ was positively correlated with the observed ilar correlation indices (r .43, p .05). measure of teamwork skills (r ϭ .43, p Ͻ .05). It is necessary to recognize that our results may be influenced by cultural differences be- Discussion tween U.S.-based samples used by Stevens and Campion (1999) to develop and validate the The first study revealed certain weaknesses in TWKSAT and our Spain-based sample. Cross- the TWKSAT (Stevens & Campion, 1999) that cultural studies (e.g., Earley & Erez, 1997; Hof- affect its use in academic and professional con- stede, 1980, 1983a, 1983b) show that cultural texts. Specifically, the following deficiencies differences may affect the processes and out- were observed in three different facets: (a) comes of individuals and groups in organiza- Some items show very low reliability indices tions (Cox, Lobel, & McLeod, 1991; Gibson, affecting the general reliability of the scale, (b) 1999). Thus, in contrast to North Americans, the contents of the original model proposed by Spaniards tend to be (a) higher in collectivism, Stevens and Campion (1994) are not fully rep- so they tend to put group interests first (Trian- resented in the test, and (c) the dimensional dis, 1995) and look more actively for social structure obtained from the factor analysis is not acceptance, strong group identity, and the de- well aligned with the substantive model. velopment of personal relationships (Grimm, With regard to the first weakness, the results Church, Katigbak, & Reyes, 1999); (b) higher This document is copyrighted by the American Psychological Association or one of its allied publishers. of the study show that certain items should be in power distance, leading them to behave more This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. eliminated from the scale, as their adjusted cor- submissively with managers and avoid dis- relations with the total for the scale are less than agreements and feeling more comfortable work- the generally accepted standards. ing in teams with strong directivity (Earley & Turning to the second weakness, the content Erez, 1997); (c) lower in action orientation analysis indicates that certain teamwork sub- (Maznevski, DiStefano, Gomez, Noorderhaven, competencies identified in the general model & Wu, 1997), which could make them work less are not picked up by the test items. Thus, no hard (Hampden-Turner & Trompenaars, 1993) items exist associated with the subcompetencies and be more resistant to work by objectives A2 (“Recognize the type and source of con- (Kluckhohn & Strodtbeck, 1961); and (d) more flict”), C1 (“Understand and use communica- likely to believe that external forces determine TEAMWORK COMPETENCY TEST 109

their successes and failures (Trompenaars, four software development programmers’ teams 1993). According to these cultural differences from a small information technology company. between Spanish and North American popula- All team members responded to the question- tions, the way employees approach their profes- naire and provided feedback on the items. Most sional performance in teams and, therefore, the items were understood and perceived as unam- way employees respond to the TWKSAT may biguous. Some minor changes were made based be different. Thus, generalization of our find- on their comments, but all the items were re- ings to the context of teamwork in North Amer- tained. As a result, the new measure contains 36 ican cultures should be done with caution. items drafted in Spanish (see the complete scale To conclude, our findings suggest the need to in Appendix B) and uses a 4-point response improve the measurement of teamwork compe- scale of frequency (1 ϭ never/almost never and tencies. The TWKSAT allows measurement of 4 ϭ always/almost always). only the General Teamwork competency, which Participants and procedure. The sample constricts examination of the conceptual rich- comprised 120 final-year psychology students ness expressed in Stevens and Campion’s at a large Spanish public university, 68.7% of (1994) original model. Consequently, it is de- whom were women. The average age was 23 sirable to develop new items capable of captur- years (SD ϭ 0.96). The procedure was identical ing the different subcompetencies, as well as to that in Study 1. The assistance of the same obtaining more reliable scores in the five com- experts was used to analyze the validity of the petencies. test content as in Study 1.

Study 2 Results Description of items and scales. Table 4 The second study analyzed a new measure of summarizes the main descriptive statistics for the teamwork knowledge, skills, and abilities the questionnaire items. The mean scores for all based on Stevens and Campion’s (1994) model, items were above 2.5 (the theoretical midpoint which is called the Teamwork Competency Test on the response scale), except those related to (TWCT). For this purpose, we developed and the Collaborative Problem Solving dimension, adapted new items in light of the findings from in which the averages were somewhat lower Study 1. The new test comprises 36 items, (between 1.94 and 2.28). The reliability indices which were administered to a sample of 120 for both the scales and the full questionnaire university students to analyze the functioning of (.89) were satisfactory. the items and to test the TWCT’s reliability, Content analysis. Table 4 presents the content validity, and dimensionality. congruence indices (Rovinelli & Hambleton, Method 1977) for each item based on the subcompeten- cies proposed by Stevens and Campion (1994). Item development for TWCT. Based on The congruence indices for all of the items were results from Study 1, we developed new items satisfactory (minimum ϭ 0.56, maximum ϭ to represent the 14 subcompetencies established 1.00). Finally, the whole content domain estab- by Stevens and Campion (1994). The items lished in the model was represented by the items were worded in the observable behavior format. developed. This document is copyrighted by the American Psychological Association or one of its allied publishers. An initial set of 83 items was constructed, Dimensionality analysis. We carried out This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. which was progressively refined to the final 36 various factor analyses to explore the dimen- items. Based on interviews with three experts in sionality of the questionnaire above and beyond the fields of teamwork and organizational be- the adequacy and relevance of the items. Given havior, several items were rephrased (seven the high means for the items and the deviation items) or left out (31 items). The remaining 52 of distributions away from normal, we used the items were applied to different samples and, minimum unweighted least squares method for taking into account the item statistics (corrected factor extraction, employing the FACTOR pro- item-total correlation and factor loadings), 16 gram (Lorenzo-Seva & Ferrando, 2006). Bar- items were deleted. Using the selected 36 items, tlett’s sphericity test (␹2 ϭ 5557.6, df ϭ 703, we conducted a pilot study on 26 members of p Ͻ .001) and the Kaiser–Meyer–Olkin index 110 AGUADO, RICO, SÁNCHEZ-MANZANARES, AND SALAS

Table 4 Teamwork Competency Test Item Statistics, Congruence Indexes, and Scale Reliabilities

Content Item MSDA1 A2 A3 B1 B2 C1 C2 C3 C4 C5 D1 D2 E1 E2 1 3.34 0.67 1.00 16 3.00 0.75 1.00 19 2.91 0.74 0.56 0.83 21 3.12 0.75 0.77 0.63 0.5 7 3.41 0.60 0.69 0.83 10 2.47 0.89 0.64 3 2.88 0.79 0.94 15 3.27 0.68 0.81 0.81 11 3.00 0.86 0.81 25 3.04 0.84 1.00 14 2.71 0.74 0.70 24 2.51 0.81 0.83 26 2.61 0.76 1.00 5 2.83 0.70 0.95 17 3.27 0.70 0.95 28 3.28 0.69 1.00 29 3.31 0.68 0.69 2 3.43 0.61 0.78 27 3.37 0.63 1.00 8 3.57 0.61 0.89 9 3.13 0.75 0.83 12 3.25 0.79 1.00 30 3.23 0.78 1.00 35 2.70 0.75 1.00 6 2.99 0.76 0.87 20 3.03 0.81 0.95 22 2.91 0.83 0.95 31 3.00 0.77 1.00 32 3.06 0.78 1.00 36 2.86 0.80 0.95 4 3.06 0.72 0.87 23 3.33 0.75 1.00 33 3.11 0.68 0.87 34 3.07 0.68 0.87 13 3.65 0.60 0.89 18 3.05 0.73 1.00 Total 112.9 12.00 Note.A1ϭ The knowledge, skills, and attitudes (KSA) to recognize and encourage desirable, but discourage undesirable, team conflict; A2 ϭ The KSA to recognize the type and source of conflict confronting the team and to implement an appropriate conflict resolution strategy; A3 ϭ The KSA to employ an integrative negotiation strategy rather than the traditional distributive strategy; B1 ϭ The KSA to identify situations requiring participative group problem solving and to utilize the proper degree and type of participation; B2 ϭ The KSA to recognize the obstacles to collaborative group problem

This document is copyrighted by the American Psychological Association or one of its allied publishers. solving and implement appropriate corrective actions; C1 ϭ The KSA to understand communication networks and to utilize

This article is intended solely for the personal use of the individualdecentralized user and is not to be disseminated broadly. networks to enhance the communication where possible; C2 ϭ The KSA to communicate openly and supportively; C3 ϭ The KSA to listen nonevaluatively and appropriately use active listening techniques; C4 ϭ The KSA to maximize consonance between nonverbal and verbal messages, and to recognize and interpret the nonverbal messages of others; C5 ϭ The KSA to engage in ritual greetings and small talk, and a recognition of their importance; D1 ϭ The KSA to help establish specific, challenging, and accepted team goals; D2 ϭ The KSA to monitor, evaluate, and provide feedback on both overall team performance and individual team member performance; E1 ϭ The KSA to coordinate and synchronize activities, information, and task interdependencies between members; E2 ϭ The KSA to help establish task and role expectations of individual team members and to ensure proper balancing of workload in the team. TEAMWORK COMPETENCY TEST 111

(.66) confirmed that the item correlation matrix “recognizing obstacles to participative problem could be factorized. The factorial solution was solving” (Items 14, 24, and 26), and Factor 5 is obliquely rotated. Eleven factors were found to associated with items referring to “identifying have an eigenvalue of more than 1.00 after situations that require participation in decision extraction. The parallel analysis retained only making” (Items 11 and 25). Finally, Factors 7 the first eight factors extracted. In line with this and 8 reflect the Communication competency. analysis, the eight-factor model was the one that The items related to “active listening, nonverbal best reflected the initial substantive model, ex- communication and recognition of communica- plaining 56% of the total variance, with only tion networks” (Items 2, 3, 5, 7, 9, and 27) are 18% residual errors of more than .05. The eight- associated with Factor 7, and those related to factor solution was compared with a five-factor “informal communication acts with other team solution, as proposed in the substantive model. members” (Items 12 and 30) are associated with The percentage variance explained was found to Factor 8. be lower (43%), as was the percentage of resid- Finally, we performed a confirmatory factor ual errors greater than .05 (33%). In addition, analysis (N ϭ 120) on the scores obtained for the residual mean squares error average showed the eight factors to verify whether the eight a marginally acceptable value (.085) in the factors extracted are associated with two gen- eight-factor model, but a clearly unacceptable eral dimensions, as proposed in the substantive value in the five-factor model (.11). Therefore, model, or with a single dimension as Study 1 it appears that the eight-factor model better re- suggests. The following factors were assigned produces the analyzed data matrix compared to the Interpersonal competencies dimension in with the five-factor model. the two-dimensional model (see Figure 2): 1 Table 5 presents the results obtained in the (Conflict Resolution), 4 (Collaborative Problem configuration matrix after oblique rotation. Al- Solving: Group), 5 (Collaborative Problem though this matrix does not exactly reproduce Solving: Individual), 7 (Communication: Active the expected structure of 14 subcompetencies, a Listening), and 8 (Communication: Informal). detailed analysis indicates that it reflects the The following factors were assigned to the Self- substantive model—with some variations—on Management dimension: 2 (Planning), 3 (Per- which the development of the items was based. formance Objective Management: Monitoring), The first factor extracted represents the Conflict and 7 (Performance Objective Management: Resolution competency. The items developed in Feedback). relation to this competency (especially those In view of the weightings assigned to the referring to the recognition of conflict: Items 1, different variables, all of the parameters esti- 10, 15, 16, 19, and 21) are associated with this mated in the two-factor model are significant factor, as are the Communication competency (p Ͻ .001), except for the relationship between items related to open communication and sup- Factor 4 and Factor 5 and the Interpersonal port. They are joined by Item 18 (designed for Relations dimension (p ϭ .001 and p ϭ .040, the Planning and Coordination competency). respectively). All of the parameters estimated The second factor extracted reflects the Plan- are again significant (p Ͻ .001) for a one-factor ning and Coordination competency (Items 4, model, except for the relationship between Fac- 13, 23, 33, and 34). This factor also includes tor 4 and Factor 5 and the General Teamwork Item 8 (designed for the Communication com- factor (p ϭ .002 and p ϭ .042, respectively). This document is copyrighted by the American Psychological Association or one of its allied publishers. petency). The items for the Goal Setting and None of the standardized residuals attains a This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. Performance Management competency appear value of |res_z| Ͼ 2 for the one-factor model. in Factors 3 and 6. Items related to “monitoring, However, a high value, |res_z| Ͼ 2, appears in assessing and providing feedback on individual the two-factor model for the relationship be- and group performance” (Items 6, 22, 31, and tween Factor 1 and Factor 7, although this is not 35) are associated with Factor 3, and those an especially large deviation. referring to “offering teammates feedback on The values obtained for the goodness-of-fit their results” (Items 20, 32, and 36) are associ- estimators for both models indicate a good fit in ated with Factor 6. Factors 4 and 5 reflect the both models. For the one-factor model and two- Collaborative Problem Solving competency. factor model, the values were, respectively, Factor 4 is associated with items referring to ␹2 ϭ 0.181 and 0.242, standardized root mean 112 AGUADO, RICO, SÁNCHEZ-MANZANARES, AND SALAS

Table 5 Exploratory Factor Analysis for Teamwork Competency Test Items Configuration Matrix

Factor Item 1 2345678 28 0.93 Ϫ0.09 Ϫ0.05 Ϫ0.02 Ϫ0.10 Ϫ0.02 Ϫ0.03 0.05 17 0.92 Ϫ0.12 Ϫ0.02 Ϫ0.05 Ϫ0.07 Ϫ0.02 Ϫ0.04 0.08 29 0.90 Ϫ0.11 Ϫ0.03 Ϫ0.05 Ϫ0.10 Ϫ0.06 Ϫ0.06 0.12 15 0.47 0.08 0.05 0.04 0.16 0.07 0.02 Ϫ0.04 19 0.35 0.03 0.05 0.06 0.09 0.16 0.19 Ϫ0.03 18 0.29 0.00 0.25 0.09 0.03 0.06 0.17 0.03 01 0.24 0.07 0.12 0.10 0.24 0.14 0.16 Ϫ0.02 10 0.23 Ϫ0.02 0.15 0.08 0.09 0.06 0.14 Ϫ0.05 21 0.21 Ϫ0.05 0.06 0.13 0.04 0.29 Ϫ0.03 Ϫ0.08 16 0.18 0.13 0.20 Ϫ0.01 0.13 0.15 0.09 0.05 ؊0.94 Ϫ0.02 0.01 Ϫ0.03 0.10 Ϫ0.06 Ϫ0.03 0.05 34 ؊0.91 Ϫ0.04 Ϫ0.03 Ϫ0.04 0.06 Ϫ0.03 Ϫ0.01 0.02 33 ؊0.84 0.03 0.02 0.01 0.09 Ϫ0.07 Ϫ0.05 0.05 04 ؊0.55 0.13 Ϫ0.05 0.05 0.06 0.12 Ϫ0.02 0.01 23 ؊0.22 0.01 0.01 0.05 Ϫ0.07 0.11 0.08 0.04 08 ؊0.19 0.13 Ϫ0.02 0.05 Ϫ0.02 0.18 0.06 0.00 13 31 Ϫ0.06 Ϫ0.05 0.93 Ϫ0.03 Ϫ0.02 0.01 Ϫ0.06 0.03 06 Ϫ0.06 0.01 0.91 0.00 Ϫ0.02 0.02 Ϫ0.05 0.01 22 0.00 Ϫ0.05 0.87 0.02 Ϫ0.01 Ϫ0.01 Ϫ0.03 Ϫ0.07 35 0.12 Ϫ0.01 0.11 0.10 0.03 0.05 Ϫ0.01 0.08 26 Ϫ0.02 Ϫ0.01 Ϫ0.01 0.95 Ϫ0.06 0.03 Ϫ0.07 0.00 14 Ϫ0.02 0.01 Ϫ0.09 0.89 Ϫ0.07 Ϫ0.02 Ϫ0.06 0.01 24 Ϫ0.12 0.03 0.08 0.80 0.05 Ϫ0.02 0.03 0.01 11 Ϫ0.08 Ϫ0.02 Ϫ0.03 Ϫ0.02 0.98 Ϫ0.03 Ϫ0.08 0.02 25 Ϫ0.07 Ϫ0.02 Ϫ0.03 Ϫ0.05 0.93 Ϫ0.03 Ϫ0.10 0.03 20 Ϫ0.06 Ϫ0.10 Ϫ0.02 Ϫ0.01 Ϫ0.04 0.89 0.03 0.02 32 Ϫ0.09 Ϫ0.14 Ϫ0.02 Ϫ0.05 Ϫ0.01 0.85 0.03 0.08 36 0.11 0.06 0.13 0.04 0.00 0.53 0.08 0.09 27 Ϫ0.08 Ϫ0.06 Ϫ0.01 Ϫ0.04 Ϫ0.07 0.09 0.90 Ϫ0.08 02 Ϫ0.10 Ϫ0.05 Ϫ0.02 Ϫ0.06 Ϫ0.10 0.05 0.79 Ϫ0.03 03 0.16 0.04 Ϫ0.08 0.00 0.03 0.13 0.39 0.03 07 0.03 0.03 0.03 0.09 0.10 0.02 0.36 0.15 05 0.09 Ϫ0.03 0.14 0.01 Ϫ0.07 Ϫ0.08 0.25 0.09 09 0.08 Ϫ0.16 Ϫ0.03 0.17 0.11 Ϫ0.11 0.25 0.00 30 Ϫ0.02 0.01 Ϫ0.05 0.01 0.03 0.03 Ϫ0.03 1.00 12 0.02 0.04 Ϫ0.01 0.01 0.00 0.07 Ϫ0.06 0.90 Note. N ϭ 120; Extraction ϭ unweighted least squares; Rotation ϭ Oblimin–Kaiser; Convergence ϭ 12 iterations. Items rationally linked with the factor are in bold type. Items not rationally linked with the factor and up to .10 value are in bold italic type.

This document is copyrighted by the American Psychological Association or one of its allied publishers. residual ϭ .048 and .047, root mean square stantive model (although it does not do so ex-

This article is intended solely for the personal use of the individualerror user and is not to be disseminated broadly. of approximation ϭ .037 and .031, com- actly). With regard to the first point, the inter- parative fit index ϭ .976 and .983, and Tucker– judge analysis indicates that the TWCT items Lewis index ϭ .966 and .975 (Figure 2). are representative of the 14 subcompetencies defined by the authors. In terms of reliability, Discussion the five scales present adequate Cronbach’s al- Overall, the results of the second study show phas of more than .80, except the Conflict Res- that the TWCT reasonably covers the whole olution scale, which has an alpha of .71. The content domain proposed by Stevens and Cam- alpha for the total scale is also adequate (.89). pion (1994), is reliable, and its dimensional Finally, the dimensionality analysis reveals a structure adequately reflects the original sub- latent structure for the TWCT that is substan- TEAMWORK COMPETENCY TEST 113

1,00 2,43 Conflict e Resoluon 1,00 2,05 Problem e Solving 1,00 3,48 1,60 Decision e ,50 Making

,25 1,00 2,10 1,00 e 1,44 Communicaon Teamwork ,52 1,00 1,41 1,28 Informal e Communicaon

1,44 1,00 2,63 Planning and e 1,40 Coordinaon

1,00 2,02 Monitoring e and Assessing

1,00 Offering1,98 Feedback e

1,00

-2,17 Conflict e Resoluon 3,65 1,00 1,00 2,05 Problem e ,52 Solving Interpersonal ,25 1,00 1,60 Decision e 1,40 Making

,53 1,00 2,13 e Communicaon

,88 1,00 1,40 Informal e Communicaon

1,00

This document is copyrighted by the American Psychological Association or one of its allied publishers. 2,61 Planning and e This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. 1,00 1,31 Coordinaon

1,00 Self-management 1,96 1,52 Monitoring and e Assessing 1,49 1,00 Offering1,91 e Feedback

Figure 2. Confirmatory factor analysis solutions for the one-factor and two-factor models. 114 AGUADO, RICO, SÁNCHEZ-MANZANARES, AND SALAS

tially associated with Stevens and Campion’s Method (1994) model. However, the structure observed Sample. A total of 91 employees working does not exactly reproduce either the structure in a major Spanish power utility took part in the of the 14 subcompetencies or the structure of study. Men made up 81% of the total sample, the five competencies. and the average age was 29.6 years (SD ϭ Our results suggest the need to reconfigure 3.38). The measurements were made as part of the originally designed scales. This is particu- the activities undertaken by the employees larly important regarding the Conflict Resolu- within the framework of a mandatory skills tion competency. The analyses of both dimen- development program set up by the company. sionality and congruence reveal a strong Participants signed a consent form regarding association between the Conflict Resolution their participation in the study. They could de- items and the Communication items (related to cline to be enrolled in the research by not pro- open communication and support). It seems rea- viding their records without any adverse effect sonable to integrate the items in the Communi- on their participation in the skills development cation subcompetency into the Conflict Resolu- program. We also obtained company consent to tion competency, as the factor analysis use the data in the research. indicates. This is also true with regard to Item Measures. 18, which both the judges and the factor anal- TWKSAT. We administered the original ysis associate with that competency. The Con- version of the Stevens and Campion (1999) test flict Resolution scale with an alpha of .84 would used in Study 1. thus be established with the same items as orig- TWCT. The version of the test designed in inally designed. The Collaborative Problem Study 2 was administered. Six different mea- Solving scale is also maintained intact, integrat- sures were established for each participant, con- ing the Factor 4 and Factor 5 items with an sisting of a total score and a score for each of alpha of .83. The Communication scale contains the five competencies. the Factor 7 and Factor 8 items associated with Team performance: Supervisor assessment. communication items, except for open commu- The assessments of each employee’s immedi- nication and support, obtaining an alpha of .72. ate superior were obtained specifically for the purposes of the study and did not form part of The Objective Management and Performance the company’s usual performance manage- competency would be formed by all of the items ment process. The information gathered on from Factors 3 and 6 with an alpha of .82. the employees was strictly training-related. Finally, the Planning and Coordination compe- The supervisor questionnaire contained eight tency integrates all of the items associated with ␣ϭ items describing different teamwork behav- Factor 2 ( .88), except Item 18, which is iors, to which they were asked to respond on moved to Conflict Resolution. a 4-point scale (1 ϭ completely disagree,4ϭ completely agree). An example item is “He or she (the employee) participates actively in Study 3 work meetings (giving his or her opinion, asking questions, etc.).” The items were The third study had a dual objective: (1) to drawn from a competency model used in the This document is copyrighted by the American Psychological Association or one of its allied publishers. obtain initial evidence for the convergent valid- company. This article is intended solely for the personal use of the individuality user and is not to be of disseminated broadly. the new measure of teamwork compe- Team performance: Self-assessment. The tency developed in Study 2, and (2) to compare same eight-item questionnaire expressed in the new measure with the original test proposed the first person was also answered by each by Stevens and Campion (1999). A sample of employee to assess his/her usual behavior in employees was used to analyze the relationship work team situations. between the TWCT and (a) the original Procedure. The supervisors’ assessments TWKSAT, (b) supervisor assessments of em- (Measure 3) and the self-assessments of the ployees’ teamwork competencies, and (c) self- employees (Measure 4) were obtained con- assessments of competency by employees currently at the beginning of the course. The themselves. supervisors gave their responses online via a TEAMWORK COMPETENCY TEST 115

user/key connection to a web page containing different (T ϭϪ1.78, p Ͻ .05). Analysis of the the questionnaire and instructions for comple- TWCT competencies shows that Competencies tion. The TWKSAT and TWCT measures (re- A (Conflict Resolution), D (Goal Setting and spectively, Measures 1 and 2) were collected Performance Management), and E (Planning at a first training session 1 week later. Both and Coordination) are positively correlated both questionnaires were administered in a paper- with the supervisor’s assessment (r ϭ .24, p Ͻ and-pencil format, counterbalanced to avoid .05; r ϭ .36, p Ͻ .01; r ϭ .27, p Ͻ .01, effects associated with the order of presenta- respectively) and with employees’ self-assess- tion. ments (r ϭ .30, p Ͻ .01; r ϭ .38, p Ͻ .01; r ϭ .37, p Ͻ .01, respectively). Results Finally, we performed a hierarchical regres- sion analysis examining the proportion of the Table 6 summarizes the descriptive statistics variance in the scores obtained from the super- and intercorrelations for all of the variables in visors’ assessments due to the TWKSAT and the study. The reliability of the two criteria TWCT, respectively. In the first step, the measures used was .70 for team performance TWKSAT score was introduced, and the TWCT supervisor evaluation and .54 for team perfor- score was introduced in a second step. Table 7 mance self-evaluation. The two versions of the ϭ indicates that the percentage of explained vari- TWKSAT were positively correlated (r .47, ance increases significantly at 6.3%, ⌬R2 ϭ p Ͻ .01), as were the assessments made by the Ͻ ϭ ␤ϭ Ͻ ϭ Ͻ .063, p .05; F(2.88) 6.60; .285, p supervisors and the employees (r .25, p .05, when the TWCT score is introduced. .05). As in Study 2, the competencies evaluated in the TWCT were positively correlated (except Discussion Competency B: Collaborative Problem Solv- ing). Also, the competencies as captured by the The results of the third study provide favor- TWCT were positively correlated with the orig- able empirical evidence for the convergent va- inal TWKSAT, except Competency B. lidity of the TWCT. The total score in this new The supervisor assessment correlation with version of the test is positively correlated with the TWCT was .34 (p Ͻ .01) compared with a the other measures in the study, namely (a) the correlation of .26 (p Ͻ .05) with the TWKSAT. score in the original Stevens and Campion test These correlations are not statistically different (1999), (b) the supervisors’ assessments of the (T ϭϪ0.76, p Ͻ .05). A similar effect was teamwork behavior of employees in their rou- observed in the employee self-assessments, tine work, and (c) the self-assessments made by which showed a correlation of .39 (p Ͻ .01) the employees themselves. Furthermore, the re- with the TWCT and .21 (p Ͻ .05) with the sults indicate that the Conflict Resolution, Goal TWKSAT. These correlations are statistically Setting and Performance Management, and

Table 6 Descriptive Statistics and Intercorrelations for Study 3 Measures

Measure MSD␣ 12345678

This document is copyrighted by the American Psychological Association or one of its allied publishers. 1. TWKSAT 21.9 2.3 .37 ءء

This article is intended solely for the personal use of the individual2. user and isTWCT not to be disseminated broadly. Total 109.0 10.9 .84 .47 ءء81. ءءTWCT KSA_A 30.5 4.6 .84 .36 .3 07. ءءTWCT KSA_B 13.6 2.6 .26 .10 .42 .4 00. ءء39. ءء59. ءءTWCT KSA_C 25.0 2.8 .35 .37 .5 ءء33. ء24. ءء59. ءء81. ءءTWCT KSA_D 19.9 3.3 .66 .41 .6 ءء51. ءء29. ءء36. ءء43. ءء72. ءءTWCT KSA_E 19.9 2.4 .55 .33 .7 ءء27. ءء36. 16. 13. ء24. ءء34. ءTeam performance supervisor evaluation 16.2 3.3 .70 .26 .8 ء25. ءء37. ءء38. 19. 08. ءء30. ءء39. ءTeam performance self-evaluation 23.4 2.8 .54 .21 .9 Note. TWKSAT ϭ Teamwork Knowledge, Skills, Attitudes Test; TWCT ϭ Teamwork Competency Test; N ϭ 91; KSA_A ϭ Conflict Resolution; KSA_B ϭ Collaborative Problem Solving; KSA_C ϭ Communication; KSA_D ϭ Goal Setting and Performance Management; KSA_E ϭ Planning and Task Coordination. .p Ͻ .01 ءء .p Ͻ .05 ء 116 AGUADO, RICO, SÁNCHEZ-MANZANARES, AND SALAS

Table 7 poor functioning of certain items, (b) the do- Hierarchical Regression Analysis Over the Team main contents proposed by the authors are not Performance Supervisor Evaluation satisfactorily represented by the test items, and Step df F ⌬R2 ␤ (c) the dimensionality of the test does not reflect .the original substantive model ء 1 1, 89 6.42 .067 Study 2 presents the TWCT, which is a new ءTWKSAT .259 -development focused on the Stevens and Cam ء063. 6.60 88 ,2 2 TWKSAT .125 pion (1999) model. The TWCT includes 36 items in a 4-point frequency scale format and ءTWCT .285 drafted in “observable behaviors” statements. Note. TWKSAT ϭ Teamwork Knowledge, Skills, Atti- tudes Test; TWCT ϭ Teamwork Competency Test. The dimensionality of the TWCT reproduces p Ͻ .05. the original substantive model better than its ء predecessor, reflecting the assessment contents domain included in the model and offering re- Planning and Coordination competencies are liable scores in both the test total and each of the the only competencies of the five contained in model’s five dimensions. the TWCT that correlate independently with the Finally, Study 3 provides initial evidence for criterion variables. the criterion validity of the TWCT compared Contrary to our expectations, however, the with the original test in a sample of profession- correlations between the different subscales in als. The total TWCT score is positively corre- the test are not all either significant or positive. lated with the teamwork assessments made by Specifically, Collaborative Problem Solving is both supervisors and the employees themselves. weakly related to the other dimensions except Moreover, the TWCT predicts the supervisor Goal Setting and Performance Management and assessment better than the TWKSAT. Planning and Coordination. Overall, our research provides empirical ev- Another interesting feature is the significant, idence concerning the method for valid, reliable positive correlation found between the original teamwork competency assessments. The TWCT TWKSAT and the supervisors’ assessments. offers a more accurate measure of teamwork This is in line with the original study carried out competency in terms of reliability than the orig- by Stevens and Campion (1999). Finally, com- inal measure proposed by Stevens and Campion parison of the predictive capacity of the two (1994), as well as greater conceptual richness in versions of the TWKSAT suggests that the terms of independent scores for the five compe- TWCT is a better predictor of the self- tencies identified by the authors. This impor- assessment of teamwork—but not of the super- tantly allows differential analysis of the effects visors’ assessments—than the original test. of specific competencies on the performance of different types of teams. In addition, it makes General Discussion possible the examination of whether improving the skills associated with a specific teamwork Overall, the three studies highlight certain competency benefits the others. Our findings significant limitations in the original TWKSAT also help both academics and practitioners to designed by Stevens and Campion (1999), im- better understand the low reliability indicators proving the metric characteristics of the test and for the TWKSAT and provide a new test This document is copyrighted by the American Psychological Association or one of its allied publishers. developing a new version. Available research (TWCT) that improves reliability and offers an This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. shows that the TWKSAT offers good predictive independent measure for each of the dimensions validity (McClough & Rogelberg, 2003), but it proposed by Stevens and Campion (1999). may be improved in terms of reliability (e.g., Considering the competencies measured by Athanasaw, 2003; Chen et al., 2004). Despite the TWCT from the standpoint of the five major the TWKSAT’s relevance as an appropriate teamwork dimensions proposed by Salas et al. measure of essential teamwork competencies, (2005), the competencies measured at the indi- the absence of studies examining the causes of vidual level are found to be associated with two these reliability problems is surprising. of the core dimensions these authors describe, Based on the results of Study 1, (a) the reli- namely follow-up and monitoring of team per- ability problems in the TWKSAT are due to the formance and team orientation. However, other TEAMWORK COMPETENCY TEST 117

core dimensions of the model, such as team skills to develop shared mental models (Can- leadership, replacement behaviors, and adapt- non-Bowers et al., 1995; Zaccaro, Rittman, & ability, do not appear in the TWCT. In addition, Marks, 2001), mutual trust (Webber, 2002), and recent studies have shown the capacity of team leadership (Cannon-Bowers et al., 1995; knowledge tests on team roles related to the Marks, Mathieu, & Zaccaro, 2001) may also be task, team, and boundary-spanning activities to critical in explaining team performance. As accurately predict performance in teams (Mar- mentioned above, the development of measures rone et al., 2007; Mumford et al., 2008). Future that integrate these aspects would improve our research should consider this issue as the ability to diagnose and predict teams’ perfor- TWCT would gain in both quality and utility if mance based on the personal abilities of the it included the mentioned dimensions. individuals who form them. Moreover, the use Despite the above-mentioned contributions, of advanced psychometric models like those our research is not without limitations. First, the based on item response theory (Hambleton & samples analyzed in the first two studies are of Swaminathan, 1985) would allow the design of a reasonable size but were provided by univer- computerized adaptive tests able to optimize the sity students. If our results are to be generalized, administration of a wide-ranging measure the studies would need to be replicated in larger (Lord, 1970; Owen, 1975). samples of employees. In particular, the factor Finally, let us consider the practical implica- analysis conducted on the TWKSAT should be tions of this research. The information provided replicated both in larger samples, given that our by the new TWCT is of great interest for people sample was smaller than the criterion of 10 management in organizations, as it shows man- participants per item, and in American samples agers, team leaders, and human resources pro- to solve the potential effects of cultural differ- fessionals which employees’ competencies are ences. As proposed by Ryan, Chan, Ployhart, critical to select, train, and develop to improve and Slade (1999), beyond the translation of the team performance. TWKSAT, it is necessary to adapt the test con- By throwing light on candidates’ skills, the sidering the equivalence of the measurements test would facilitate selection and team compo- obtained in culturally diverse populations. Sec- sition decisions. The access to information on ond, the factor analyses carried out on the five different competencies means that the con- TWCT reveal a dimensional structure that is figuration of teams can be supported by a richer similar to the substantive model employed but information base than would be possible using does not reflect it perfectly. Further studies only a single general estimate. In addition, the should look into the reasons for this, particu- differential diagnosis provided by the TWCT on larly with regard to the functioning of the Col- the stronger and weaker competencies of each laborative Problem Solving scale. In line with employee will allow teams to customize train- the underlying theoretical model, it was ex- ing and development interventions. This is par- pected that this competency would correlate ticularly valuable in view of the increasing de- with Conflict Management and Communica- mand for training programs to improve tion, but only a weak association was found. teamwork competencies (Chen et al., 2004). Moreover, Collaborative Problem Solving cor- Finally, the behaviors made explicit by the relates more closely with other competencies TWCT could be used as a guide to determine related to team self-management. Analysis of the behavioral anchors and sample behaviors This document is copyrighted by the American Psychological Association or one of its allied publishers. the content of items (e.g., Item 11, “To address used in the design of performance assessment This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. the trivial task-related issues, I do not need to and management tools. talk first with all team members in order to reach a decision”) indicates that behaviors of this kind are related to effective coordination. References The number and type of competencies se- Athanasaw, Y. A. (2003). Team characteristics and lected are critical issues. The competencies team member knowledge, skills, and ability rela- originally proposed by Stevens and Campion tionships to the effectiveness of cross-functional (1994) do not exhaust the set of skills that could teams in the public sector. International Journal be considered to explain what people actually of Public Administration, 26, 1165–1203. doi: do in effective teamwork. For example, the 10.1081/PAD-120019926 118 AGUADO, RICO, SÁNCHEZ-MANZANARES, AND SALAS

Brislin, R. W. (1970). Back-translation for cross- Hofstede, G. (1983a). The cultural relativity of orga- cultural research. Journal of Cross-Cultural nizational practices and theories. Journal of Inter- Psychology, 1, 185–216. doi:10.1177/ national Business Studies, 14, 75–89. doi:10.1057/ 135910457000100301 palgrave.jibs.8490867 Cannon-Bowers, J. A., & Bowers, C. A. (2011). Hofstede, G. (1983b). Dimensions of national cul- Team development and functioning. In S. Zedeck tures in fifty countries and three regions. In J. B. (Ed.), APA handbook of industrial and organiza- Deregowksi, S. Dziurawiec, & R. C. Annis (Eds.), tional psychology (Vol. 1, pp. 597–650). Wash- Expiscations in cross-cultural psychology (pp. ington, DC: American Psychological Association. 335–355). Lisse, Netherlands: Swets & Zeitlinger. doi:10.1037/12169-019 Kluckhohn, F., & Strodtbeck, F. L. (1961). Varia- Cannon-Bowers, J. A., Tannenbaum, S. I., Salas, E., tions in value orientations. Westport, CT: Green- & Volpe, C. E. (1995). Defining competencies and wood Press. establishing team training requirements. In R. A. Kozlowski, S. W. J., & Ilgen, D. R. (2006). Enhanc- Guzzo & E. Salas (Eds.), Team effectiveness and ing the effectiveness of work groups and teams. decision making in organizations (pp. 333–380). Psychological Science in the Public Interest, 7, San Francisco, CA: Jossey-Bass. 77–124. Chen, G., Donahue, L. M., & Klimoski, R. J. (2004). Lawler, E. E. III, & Worley, C. G. (2006). Built to Training undergraduates to work in organizational change. How to achieve sustained organizational teams. Academy of Management Learning and Ed- effectiveness. San Francisco, CA: Jossey-Bass. ucation, 3, 27–40. doi:10.5465/AMLE.2004 Leach, D. J., Wall, T. D., Rogelberg, S. G., & Jack- .12436817 son, P. R. (2005). Team autonomy, performance, Clevenger, J., Pereira, G. M., Wiechmann, D., and member job strain: Uncovering the Teamwork Schmitt, N., & Harvey, V. S. (2001). Incremental KSA link. Applied Psychology: An International validity of situational judgment tests. Journal of Review, 54, 1–24. doi:10.1111/j.1464-0597.2005 Applied Psychology, 86, 410–417. doi:10.1037/ .00193.x 0021-9010.86.3.410 Lord, F. M. (1970). Some test theory for tailored Cox, T., Lobel, S., & McLeod, P. (1991). Effects of testing. In W. H. Holtzman (Ed.), Computer as- ethnic group cultural differences on cooperative sisted instruction, testing and guidance (pp. 139– and competitive behavior on a group task. Acad- 183). New York: Harper & Row. emy of Management Journal, 34, 827–847. doi: Lorenzo-Seva, U., & Ferrando, P. J. (2006). 10.2307/256391 FACTOR: A computer program to fit the explor- Earley, P. C., & Erez, M. (1997). The transplanted atory factor analysis model. Behavioral Research executive. New York, NY: Oxford University Methods, 38, 88–91. doi:10.3758/BF03192753 Press. Marks, M. A., Mathieu, J. E., & Zaccaro, S. J. (2001). Gibson, C. B. (1999). Do they do what they believe A temporally based framework and taxonomy of they can? Group efficacy and group effectiveness team processes. Academy of Management Review, across tasks and cultures. Academy of Manage- 26, 356–376. ment Journal, 42, 138–152. doi:10.2307/257089 Marrone, J. A., Tesluk, P. E., & Carson, J. B. (2007). Grimm, S. D., Church, A. T., Katigbak, M. S., & Reyes, A multi-level investigation of antecedents and con- J. A. S. (1999). Self-described traits, values, and sequences of team member boundary spanning be- moods associated with individualism and collectiv- havior. Academy of Management Journal, 50, ism. Journal of Cross-Cultural Psychology, 30, 466– 1423–1439. doi:10.5465/AMJ.2007.28225967 500. doi:10.1177/0022022199030004005 Maznevski, M. L., DiStefano, J. J., Gomez, C. B., Hackman, J. R. (2002). Leading teams: Setting the Noorderhaven, N. G., & Wu, P. (1997, October stage for great performances. Boston, MA: Har- 8–12). The cultural orientations framework and This document is copyrighted by the American Psychological Association or one of its allied publishers. vard Business School Press. international management research. Paper pre- This article is intended solely for the personal use of the individualHambleton, user and is not to be disseminated broadly. R. K., & Swaminathan, J. (1985). Item sented at the Academy of International Business response theory: Principles and applications. Bos- annual meeting, Guadalajara, Mexico. ton, MA: Kluwer. McClough, A. C., & Rogelberg, S. G. (2003). Selec- Hampden-Turner, C., & Trompenaars, F. (1993). The tion in teams: An exploration of the Teamwork seven cultures of capitalism: Value systems for Knowledge, Skills and Ability Test. International creating wealth in the United States, Japan, Ger- Journal of Selection and Assessment, 11, 56–66. many, France, Sweden, and the Netherlands. Gar- doi:10.1111/1468-2389.00226 den City, NY: Doubleday. McDaniel, M. A., Morgeson, F. P., Finnegan, E. B., Hofstede, G. (1980). Culture’s consequences: Inter- Campion, M. A., & Braverman, E. P. (2001). Use national differences in work-related values. Bev- of situational judgement tests to predict job per- erly Hills, CA: Sage. formance: A clarification of the literature. Journal TEAMWORK COMPETENCY TEST 119

of Applied Psychology, 86, 730–740. doi:10.1037/ Boston, MA: Elsevier Academic Press. doi: 0021-9010.86.4.730 10.1016/B0-12-657410-3/00533-X Mumford, T. V., Van Iddekinge, C. H., Morgeson, Schmidt, F. L., & Hunter, J. E. (1983). Individual F. P., & Campion, M. A. (2008). The Team Role differences in productivity: An empirical test of Test: Development and validation of a team role estimates derived from studies of selection proce- knowledge situational judgment test. Journal of dure utility. Journal of Applied Psychology, 68, Applied Psychology, 93, 250–267. doi:10.1037/ 407–414. doi:10.1037/0021-9010.68.3.407 0021-9010.93.2.250 Spencer, L. M., & Spencer, S. M. (1993). Compe- Muthén, B., & Muthén, L. K. (2006). Mplus [Com- tence at work: Models for superior performance. puter software]. Los Angeles, CA: Author. New York, NY: Wiley. Owen, R. J. (1975). A Bayesian sequential procedure Stevens, M. A., & Campion, M. J. (1994). The for quantal response in the context of adaptive knowledge, skill, and ability requirements for mental testing. Journal of the American Statistical teamwork: Implications for human resource man- Association, 70, 351–356. doi:10.1080/01621459 agement. Journal of Management, 20, 503–530. .1975.10479871 Stevens, M. A., & Campion, M. J. (1999). Staffing Peeters, M. A. G., van Tuijl, H. F. J. M., Rutte, C. G., work teams: Development and validation of a se- & Reymen, I. M. M. J. (2006). Personality and lection test for teamwork settings. Journal of Man- team performance: A meta-analysis. European agement, 25, 207–228. doi:10.1016/S0149- Journal of Personality, 20, 377–396. doi:10.1002/ 2063(99)80010-5 per.588 Thorn, D. W., & Deitz, J. C. (1989). Examining Rovinelli, R. J., & Hambleton, R. K. (1977). On the content validity through the use of content experts. use of content specialists in the assessment of Occupational Therapy Journal of Research, 9, criterion-referenced test item validity. Dutch Jour- 334–346. nal of Educational Research, 2, 49–60. Triandis, H. C. (1995). Individualism and collectiv- Ryan, A. M., Chan, D., Ployhart, R. E., & Slade, ism. Boulder, CO: Westview Press. L. A. (1999). Employee attitude surveys in a mul- Trompenaars, F. (1993). Riding the waves of culture: tinational organization: Considering language and Understanding cultural diversity in business. Burr culture in assessing measurement equivalence. Ridge, IL: Irwin. Personnel Psychology, 52, 37–58. doi:10.1111/j Webber, S. S. (2002). Leadership and trust facilitat- .1744-6570.1999.tb01812.x ing cross-functional team success. Journal of Man- Salas, E., Sims, D. E., & Burke, C. S. (2005). Is there agement Development, 21, 201–214. doi:10.1108/ a “big five” in teamwork? Small Group Research, 02621710210420273 36, 555–599. doi:10.1177/1046496405277134 Zaccaro, S. J., Rittman, A. L., & Marks, M. A. Salas, E., Sims, D. E., & Klein, C. (2004). Cooper- (2001). Team leadership. The Leadership Quar- ation at work. In C. D. Spielberger (Ed.), Encyclo- terly, 12, 451–483. doi:10.1016/S1048-9843(01) pedia of applied psychology (Vol. 1, pp. 497–505). 00093-5

(Appendices follow) This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. 120 AGUADO, RICO, SÁNCHEZ-MANZANARES, AND SALAS

Appendix A

Observation Code Categories Conflict Resolution • She/he makes summaries along a commu- • She/he negotiates considering reasons in- nication interaction to demonstrate understand- stead of positions. ing. • She/he collects information about the dif- • She/he communicates openly and support- ferent sources of conflict. ively so that he/she recognizes the others. • She/he uses an integrative (win–win) nego- tiation strategy when possible. Goal Setting and Performance Management • She/he tries to define clearly the wished Collaborative Problem Solving results (SMART objectives). • She/he performs actions that promote oth- • She/he describes problems in terms of be- ers’ success instead of hindering their progress. haviors, consequences, and feelings. • She/he shows awareness about others’ per- • She/he avoids setting conclusions and as- formance (e.g., asking for participation of the cribing motives to others quickly. most active members). • She/he focused on communalities to gener- ate solutions (e.g., use brainstorming to gener- Planning and Task Coordination ate more options). • She/he requests additional information. • She/he makes questions for planning (e.g., How are we going to start the project? What resources are necessary? How much time we Communication have to invest?). • She/he produces complete and specific • She/he clarifies the procedures of execution messages. (what we make the first, the second, etc.). • She/he asks for feedback about message • She/he helps to associate tasks with understanding. teammates.

Appendix B

TWCT Items 1. When my work team is in conflict, I try to 4. I plan my tasks effectively. make it explicit to find solution pathways. 5. I try to use the most appropriate team network 2. When I interact with my team mates, I ask to communicate the different types of infor- questions to better understand what they say. mation, avoiding the same formal procedure 3. When I disagree with others, I make an effort all the time. to focus on what we have in common instead 6. I often get involved in monitoring the task of centering on what separates us. performance of other team members. This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

(Appendices continue) TEAMWORK COMPETENCY TEST 121

7. When we face an internal conflict because of 22. I try to establish milestones in my work team a communication problem or misunderstand- so that we can monitor our assigned tasks. ing, I try to solve it by asking questions and 23. When I am involved in a team project, I care listening to the people involved. about having clear plans concerning the tasks 8. I look at people when they talk to me and I and the timing to accomplish them. modify my body language to show real inter- 24. During group meetings, regulation is neces- est in what they tell me. sary to ensure that all members provide their 9. I can easily recognize people’s emotional opinions and to avoid that only a few partic- states by observing their nonverbal messages. ipate actively. 10. If someone in my team acts inappropriately, I 25. When performing tasks in which one is an talk privately with her/him, encouraging the expert, the contributions made by other mem- rest of the team to do the same. bers are not that important. 11. To address the trivial task-related issues, I do 26. In group decision meetings, it is more usual to not need to talk first with all team members so promote cohesion and reach a majority agree- we reach a decision. ment than to pay attention to divergent opin- 12. I make an effort to talk about less important ions. things with my peers for the sake of team 27. I try listening to my peers’ opinions without spirit and better internal communication. evaluating their positions as good or bad. 13. Having knowledge about people’s skills and 28. When working in a group, I say what I think situation requirements is critical to assign in an open and sincere way. tasks properly. 29. I expect my peers trust enough to tell me the 14. Discussions without directions or guides can aspects of my work that they most dislike. lead group members to make decisions that 30. I sometimes talk with my peers without an they would not make on their own. objective, just for sharing a while together. 15. When my personal interests are in conflict 31. It is important for me to monitor the tasks with others’ interests, I tend to be honest in assigned to each team member. the negotiation so that others understand my 32. I provide my peers with relevant information needs. on how well I think the team tasks are pro- 16. I care and act to make team conflicts explicit gressing. in a way that they can be solved. 33. When doing my job, I prioritize the tasks 17. I play an active role in team meetings by most necessary for my teammates to com- offering my opinions, asking questions, and plete their work. expressing my thoughts and ideas in a sincere 34. I try to ensure that my outputs match the and open way. inputs needed by my peers to perform their 18. I often help others in my team to make clear tasks. the roles and tasks they have to perform. 35. For the sake of team work, I set objectives 19. When I am upset about something, I express with moderate difficulty so that effort is my discomfort to the group in a constructive needed to accomplish them. way, asking for solution alternatives. 36. I often provide my peers with feedback on 20. I like to provide my peers with feedback their task performance. about what they do and to assess and value their work.

This document is copyrighted by the American Psychological Association or one of its allied publishers. Received October 15, 2012 21. If something upsets me in my team, I do not Revision received November 7, 2013 This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. like to act as if nothing has happened. Accepted December 14, 2013 Ⅲ Group Dynamics: Theory, Research, and Practice © 2014 American Psychological Association 2014, Vol. 18, No. 2, 122–128 1089-2699/14/$12.00 http://dx.doi.org/10.1037/gdn0000001 The Effect of Different Phases of Synchrony on Pain Threshold

Philip J. Sullivan, Kate Rickers, and Kimberley L. Gammage Brock University

Synchronization of behavior between individuals has been found to result in a variety of prosocial outcomes. The role of endorphins in vigorous synchronous activities (Cohen, Ejsmond-Frey, Knight, & Dunbar, 2010) may underlie these effects as endor- phins have been implicated in social bonding (Dunbar & Shultz, 2010). Although research on synchronous behavior has noted that there are 2 dominant phases of synchrony—in-phase and antiphase synchrony (Marsh, Richardson, Baron, & Schmidt, 2006), research on the effect of synchrony on endorphins has only incorporated in-phase synchrony. The current study examined whether both phases of synchrony would generate the synchrony effect. Twenty-two participants rowed under 3 counter- balanced conditions—alone, in-phase synchrony and antiphase synchrony. Endorphin release, as measured by pain threshold, was assessed before and after each session. Change in pain threshold during the in-phase synchrony session was significantly higher than either of the other 2 conditions. These results suggest that the synchrony effect may be specific to just in-phase synchrony, and that social presence is not a viable explanation for the effect of synchrony on pain threshold.

Keywords: behavioral synchrony, endorphins, pain threshold, in-phase synchrony

People tend to be inclined to move syn- ple have been shown to infer greater chronously, whether to an external stimuli or cohesiveness on individuals who move in syn- to partners. For example, children as young as chrony compared with those who move inde- two have been found to impulsively synchro- pendently (Lakens & Stel, 2011). Synchroniza- nize themselves with a drum beat (Kirschner tion of individuals’ movements with others has & Tomasello, 2009), and even minimal been shown to result in higher ratings of inter- groups, such as two individuals in rocking personal attraction and affiliation (Bernieri, chairs, will spontaneously synchronize their 1988; LaFrance & Broadbent, 1976; Miles, motions (Richardson, Marsh, Isenhower, Nind, & Macrae, 2009). Goodman, & Schmidt, 2007). The relevance Hove and Risen (2009) conducted a series of of synchronized behavior may be seen in a experiments on interpersonal attraction between variety of interpersonal outcomes that have individuals engaged in synchronized behavior. resulted from joint action or mimicry. In one study, participants engaged in synchro- Whether people spontaneously act in coordi- nized finger tapping with an experimenter for a nation with each other, or one person mimics 2.5 min trial. Degree of synchrony, as measured the actions of someone else, there appears to by percentage of finger taps occurring within be a robust effect of such actions on interper- 100 ms, was significantly correlated with higher sonal attraction and prosocial behavior. This document is copyrighted by the American Psychological Association or one of its allied publishers. ratings of liking for the experimenter. Their Acting in synchrony with others has repeat- This article is intended solely for the personal use of the individualedly user and is not to be disseminated broadly. been found to increase perceptions of in- second experiment manipulated interpersonal terpersonal attraction between participants. Peo- synchrony and included a baseline measure of liking. Results showed that individuals ran- domly assigned to a condition in which they tapped in synchrony with the experimenter re- Philip J. Sullivan, Kate Rickers, and Kimberley L. Gam- ported significantly greater liking for the exper- mage, Department of Kinesiology, Brock University. imenter than those in either the asynchrony or Correspondence concerning this article should be ad- alone conditions. There were no significant dressed to Philip J. Sullivan, Department of Kinesiology, Brock University, St. Catharines, ON, L2S 3A1. E-mail: baseline differences in liking for the experi- [email protected] menter.

122 SYNCHRONY EFFECT 123

Synchrony has also been repeatedly linked to relationship with some kind of deeply emo- a variety of prosocial behaviors. Individuals tional basis” (p. 782). who act in synchrony tend to display signifi- Because brain endorphins do not cross the cantly greater cooperation (Valdesolo, Ouyang blood–brain barrier, and can only be measured & DeSteno, 2010; Wiltermuth & Heath, 2009), through an invasive lumbar puncture (Boeker et norm compliance (Wiltermuth, 2012), compas- al., 2008; Dearman & Francis, 1983), pain sion (Valdesolo & DeSteno, 2011), and gener- threshold has been commonly used as an indi- osity (Stel, van Baaren, & Vonk, 2007) than cator of endorphin activity. (Dunbar et al., individuals who do not act in synchrony. 2012; Jamner & Leigh, 1999; Zillman, Rock- Wiltermuth and Heath (2009) investigated well, Schweitzer, & Sundar, 1993). Pain thresh- the effect of synchronized behavior and coop- old is understood to be a valid indicator of eration in small groups. Participants walked endorphin activity because endogenous opioids around campus in groups of three in one of two are released during pain sensation (DePue & conditions. In the synchronized condition, they Morrone-Strupinsky, 2005; Mueller, 1981). walked in step; in the nonsynchronized condi- Specifically, the use of a blood pressure cuff to tion, they walked normally. In a postwalk co- induce specific ischemic pain has been found to operative game, individuals in the synchronized be a valid and reliable, as well as noninvasive, condition behaved significantly more coopera- protocol to assess endorphin levels (Estebe, Le tively than those in the nonsynchronized condi- Naoures, Chemaly, & Ecoffey, 2000; Ryan & tion. A second experiment in the same study Kovacic, 1966). replicated this effect, even after controlling for Cohen, Ejsmond-Frey, Knight, and Dunbar common identity and common fate within the (2010) investigated synchronized behaviors in a groups. team of male collegiate rowers, using the pain In a study conducted by Valdesolo, Ouyange, threshold test as a noninvasive measure of cen- and DeSteno (2010), individuals were randomly tral endorphin activity. They found a difference assigned to one of two conditions. They rocked when teammates rowed in synchronicity in a in rocking chairs in pairs in either a synchro- virtual boat compared with when they per- nized or unsynchronized fashion for two 90 s formed the same workout alone. Specifically, trials. After this manipulation, the pairs had to the synchronized condition produced a signifi- participate in a cooperative joint action task. cantly greater increase in pain threshold from Subjects also completed ratings of similarity, pre- to postexercise than the solitary session. connectedness, and liking for their partners. This was despite the fact that workload was Pairs in the synchronized condition completed standardized between the two conditions. the cooperative task in significantly less time Cohen et al. (2010) concluded that vigorous than those in the unsynchronized condition. synchronized activity results in a “synchrony Synchrony also resulted in higher perceptions effect” whereby endorphin activity is increased. of connectedness and liking. This synchrony effect has been replicated by Recent research may have uncovered a po- Sullivan and Rickers (2013) who used the same tential mechanism for these social effects of protocol as Cohen et al. (2010), with partici- synchrony. It appears that synchronous behav- pants who were strangers as well as rowing iors, particularly vigorous synchronous behav- teammates. Because of the role of endorphins in iors, elevate participants’ endorphin levels, social bonding, the study investigated whether This document is copyrighted by the American Psychological Association or one of its allied publishers. which have been linked to social bonding in the use of teammates in Cohen’s study may This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. humans and primates (Dunbar, 2010; Dunbar & have influenced the results. Sullivan and Rick- Shultz, 2010). It has been suggested that endog- ers found that when individuals rowed for 45 enous endorphins may mediate between affiliative min with either teammates or strangers, they stimuli (e.g., shared synchronous movements) and still displayed significantly higher changes in affiliation (DePue & Morrone-Strupinsky, 2005. pain threshold than when they rowed alone. Dunbar and Shultz (2010) stated that social One aspect of synchrony that may be relevant grooming in primates builds intense social bonds to the synchrony effect is the phase of syn- through the release of endorphins, “which pro- chrony. There are two dominant phases of syn- vides a psycho-pharmacological mechanism chrony—in-phase and antiphase (Marsh, Rich- that enables two individuals to build a bonded ardson, Baron, & Schmidt, 2006). In-phase 124 SULLIVAN, RICKERS, AND GAMMAGE

synchrony is characterized by movements being Procedures in perfect unison whereas antiphase symmetry is characterized by alternating patterns of move- Before data collection, this study was ap- ment between two individuals. For example, if proved by the IRB of the primary author. Before two children were side by side on swings, in- participation, all participants completed in- phase movement would be if they both swing formed-consent forms and the Physical Activity forward and then backward together. Antiphase Readiness Questionnaire (PAR-Q; Canadian synchrony would be seen when one child Society for Exercise Physiology, 2002). The swings forward as the other swings back (Miles, PAR-Q is a 7-item measure to assess whether Nind, Henderson, & Macrae, 2010). Humans participants can safely partake in physical activity. appear to prefer in-phase synchrony (Marsh et Every participant was required to answer to ques- al., 2006; Richardson et al., 2007). Furthermore, tions using either yes or no responses. If they in-phase synchrony appears to have a more replied with “yes” to any of the questions, they were to obtain permission from their doctors to be powerful effect on some of the outcomes noted physically active. If they answered “no” to all above (Macrae, Duffy, Miles, & Lawrence, questions, it was safe for them to participate in 2008). physical activity. All participants were able to Although these two different phases may partake in the procedure. both affect outcomes of synchronized behav- The experiment used a randomized counter- ior, to date the synchrony effect has only been balanced repeated measures design. Participants investigated with in-phase synchrony. The rowed in all of three conditions—alone, and in current study investigated the role of different two synchronized condition with a partner. In phases of behavioral synchrony on the syn- the in-phase synchrony condition, both rowers chrony effect. Specifically, rowing under sol- kept in the same movement pattern (i.e., both itary, in-phase and antiphase synchrony con- rowers would be at the fully extended and full ditions were examined to see if all would contracted positions of the stroke at the same produce the same effect on pain threshold that time). In the antiphase synchrony condition, was seen by Cohen et al. (2010) and Sullivan rowers kept at opposite points of the movement and Rickers (2013). Because previous re- pattern (i.e., when one rower was fully ex- search has noted significantly greater in- tended, the other rower would be fully con- creases in endorphin levels after synchronous tracted, and vice versa). These conditions were compared with solitary activity, it was pre- counterbalanced so that one third of the partic- dicted that there would be a significant dif- ipants rowed alone first, one third rowed in the ference between solitary and in-phase syn- in-phase synchrony condition first, and the re- chrony. Because antiphase synchrony has not maining third rowed in the antiphase synchrony yet been studied with respect to the synchrony condition first. effect, no specific hypotheses were put for- Intensity of the workout was operationalized ward regarding this condition. by average 500 m split time (i.e., the time taken to row 500 m on the ergometer machine) over the 30 min session. This value was recorded Method during the first session and participants matched the intensity on subsequent trials. By slight ma- This document is copyrighted by the American Psychological Association or one of its allied publishers. Participants nipulations of the drag factor, the experimenters This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. were able to establish synchrony between par- Twenty-two (9 male, 13 female) individuals ticipants while allowing individuals to maintain participated in the study. Participants ranged in the same intensity (i.e., stroke rate) in all three age from 19 to 55 years (M ϭ 31.14, SD ϭ conditions. 13.13). Men’s average height was 73.11 in. Participants rowed with the same partner in (SD ϭ 1.54) and weight was 200.56 pounds both synchrony conditions and pairs always (SD ϭ 23.91). Women’s average height was consisted of rowers of the same gender. All 67.62 in. (SD ϭ 1.94) and weight was 145.00 participants were prohibited from listening to (SD ϭ 12.75). All participants had rowed for at music or consuming sport drinks during their least 1 year. trials. There was a week gap between sessions SYNCHRONY EFFECT 125

for all participants. All sessions were held in the scores (Dimitrov & Rumrill, 2003). Pre- and afternoon or early evening. postpain threshold scores by condition are Measures of pain threshold were taken before shown in Table 1. All three conditions dis- and after all trials using the blood pressure cuff played mean increases in pain threshold from protocol used by Cohen et al. (2010). All pain pre- to postactivity, which is consistent with the measurements were taken by the experimenter effect of vigorous exercise on endorphin activ- and recorded ϳ1 min before the start and con- ity (Boecker et al., 2008). Change in pain clusion of each trial. In all conditions, the same threshold was normally distributed in all three experimenter assessed pain threshold for all par- conditions. Average prescores did not differ be- ticipants. This was achieved by staggering the tween conditions (F (2, 63) ϭ 0.89, p ϭ .416), end of workouts of participants by ϳ30 s each. and there were significant pre- to postchanges in The pain thresholds of all participants were the solitary (F (1, 21) ϭ 24.10, p Ͻ 001), assessed away from all others being tested. The in-phase (F (1, 21) ϭ 69.69, p Ͻ 001) and blood pressure cuff was placed above the elbow antiphase conditions (F (1, 21) ϭ 22.04, p Ͻ on the nondominant arm of each participant. 001). There were no statistically significant dif- Ischemic pain was induced through manual ferences between genders in preactivity pain pumping of the cuff to increase pressure. Par- threshold scores or changes scores in any of the ticipants expressed the point at which pressure three conditions. became uncomfortable by saying “now,” at Before testing the hypotheses, the data was which point the pressure was recorded and the examined to see if it the assumption of spheric- cuff was removed. This protocol has been com- ity was upheld. Mauchly’s test was nonsignifi- monly used in studies on behavioral synchrony cant (W ϭ .929, p Ͼ .05), indicating that there (Cohen et al., 2010; Dunbar et al., 2012; Sulli- was homogeneous variance between the condi- van & Rickers, 2013), and previous research has tions, and that the data was appropriate for a supported its validity as a measure of pain repeated measures analysis. A repeated mea- threshold. It has been shown to be highly cor- sures analysis of variance (ANOVA) revealed a related with other measures of pain threshold, significant effect for condition (F (2, 42) ϭ including visual analog scales (Estebe et al., 4.12, p ϭ .023, ␩2 ϭ .16). Observed power 2000) and gross pressure ischemia (Ryan & within the design was .70. Bonferroni adjusted Kovacic, 1966). Furthermore, it has been found contrast analyses revealed that the in-phase syn- to have a high test–retest reliability within par- chrony condition was significantly different ticipants (r ϭ .87; Ryan & Kovacic, 1966). from both the solitary (F (1, 21) ϭ 5.27, p ϭ .03, ␩2 ϭ .20) and antiphase (F (1, 21) ϭ 8.99, Materials p ϭ .007, ␩2 ϭ .30) conditions. There was not a significant difference between the solitary and Two Concept2 Model D indoor rowing ma- antiphase conditions (F (1, 21) ϭ 0.41, p Ͼ chines (i.e., “ergometers”) were set up in the .05). rowing training center of the university. Pain threshold was measured using an AMG Medical Discussion Inc. Professional Series sphygmomanometer (blood pressure cuff). Units of pressure were The current study was designed to investigate recorded in millimeters of mercury (mmHg). if two different phases of synchrony would both This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. Results Table 1 Means and SDs of Raw and Change Pain Because pain threshold differs by individual Threshold Scores by Condition (Dunbar et al., 2012), change in pain threshold from pre- to postscore was used as the depen- Prescore Postscore Change score dent variable in analyses. Although there are Condition (M, SD) (M, SD) (M, SD) some concerns with using raw change scores in Solitary 178.64 (38.95) 210.00 (51.73) 31.36 (29.96) such analyses, research has shown that when In-phase 196.82 (55.41) 247.73 (50.89) 50.91 (28.60) using reliable scores, particularly physiological Antiphase 194.55 (51.89) 224.09 (52.34) 29.55 (29.52) measures, it is advisable to use raw change Note. Values are in millimeters of mercury (mmHg). 126 SULLIVAN, RICKERS, AND GAMMAGE

show the “synchrony effect” of elevated pain Therefore, it is possible that through the syn- threshold, which has previously only been doc- chrony effect, vigorous synchronized activity umented with in-phase synchrony (Cohen et al., releases endorphins, which in turn increase so- 2010; Sullivan & Rickers, 2013). Within the cial bonding. This process would be consistent current study, it was found that when individu- with speculation that collective activities in- als rowed in an in-phase synchronous condition, crease cohesion among group members in ac- they reported a significantly greater increase tivities such as marching in step and rhythmic (from pre- to postactivity) in pain threshold than dancing and singing during religious activities either the solitary or antiphase synchrony con- (McNeill, 1995). Through the inducement of ditions. There was no significant difference be- endorphins, the synchrony effect may serve as a tween the antiphase and solitary condition. mechanism for this increased cohesion. Therefore, the synchrony effect appears to be A secondary implication of the finding that only a result of in-phase synchrony. in-phase synchrony produces an effect on en- At this point, it appears that the synchrony dorphins whereas antiphase synchrony does not effect is a fairly robust phenomenon. It has is that this synchrony effect now appears to be repeatedly been found that individuals have a independent of any social facilitation effect. significantly higher change in pain threshold The designs of Cohen et al., (2010) and Sullivan after vigorous activity in synchrony with others and Rickers (2013) both included a confound- than after the same vigorous activity performed ing factor—the synchrony condition was also alone (Cohen et al., 2010; Sullivan & Rickers, the only group condition. In tasks requiring 2013). This has been found with different sam- physical conditioning demands, such as rowing, ples, including teammates and strangers, and the presence of others should result in perfor- both males and females. Whereas Cohen et al.’s mance increases (Strauss, 2002). Although per- (2010) original study only included male rowers formance was constant in both conditions in who were teammates, Sullivan and Rickers Cohen et al.’s and Sullivan and Ricker’s re- (2013) included strangers as well as teammates search, it may be that other responses were and males as well as females, and the current affected. The presence of others has been study also found the effect using both male and known to have an effect on unlearned re- female participants. sponses, particularly cardiovascular activity. As pain threshold has been interpreted as an Blascovich, Mendes, Hunter, and Salomon indication of endorphin activity (Cohen et al., (1999) found that the presence of others during 2010; Dunbar et al., 2012), it appears that vig- well-learned tasks could produce a “challenge” orous synchronized activity may produce a sig- response (i.e., an increase in cardiovascular ac- nificant “rush” in endorphins compared with the tivity). Because exercise stimulates endorphin same activity performed alone. Endorphins release (Boeker et al., 2008), such a change have been implicated in social bonding, al- could be responsible for the change in endor- though the exact mechanism is not clearly un- phins seen by Cohen and her colleagues. It may derstood. As noted above, it is possible that have been possible that the endorphin effect that endorphins mediate affiliative stimuli, such as Cohen attributed to synchrony was actually a synchronous movements, and affiliation (DePue mere presence effect. However, given the cur- & Morrone-Strupinsky, 2005). If this is the rent results, where a second group condition case, it may be that synchronized movements, (i.e., the antiphase synchrony) did not produce This document is copyrighted by the American Psychological Association or one of its allied publishers. particularly in-phase synchrony, which appears this synchrony effect, it appears that the effect This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. to be preferred by people (Marsh et al., 2006; on endorphins is because of synchrony, in par- Richardson et al., 2007), causes endorphin re- ticular in-phase synchrony, rather than any so- lease. In the case of vigorous synchronized ac- cial facilitation effect. tivities, such as rowing, this endorphin release is The present results extend our understanding above and beyond what is induced by the exer- of the synchrony effect. Consistent with previ- cise itself (Cohen et al., 2010). Because one of ous research, they have supported the finding the effects of endorphins, in addition to pain that vigorous synchronized activity causes an threshold, is feelings of well-being (Dunbar & increase in pain threshold, which appears to be Shultz, 2010), this shared synchrony would re- an indication of elevated endorphin activity. sult in stronger social bonds among participants. This effect now appears to be the result of SYNCHRONY EFFECT 127

synchrony, as opposed to social coaction, and Boeker, H., Sprenger, T., Spilker, M. E., Henriksen, specifically in-phase synchrony. Future research G., Koppenhoefer, M., Wagner, K. L.,...Tolle, is still required to fully understand this syn- T. R. (2008). The runner’s high: Opioidergic chrony effect. Specifically, the studies to date mechanisms in the human brain. Cerebral Cortex, have only used the synchronized activity of 18, 2523–2532. doi:10.1093/cercor/bhn013 rowing. Would the same synchrony effect be Canadian Society for Exercise Physiology. (2002). Physical Activity Readiness Questionnaire (PAR-Q). seen with other activities such as running in Retrieved from http://www.csep.ca/communities/c574/ step, or fitness classes? It would also be inter- files/hidden/pdfs/par-q.pdf esting to see if the effect could be replicated Cohen, E. E., Ejsmond-Frey, R., Knight, N., & Dun- with synchronized nonvigorous activities such bar, R. I. (2010). Rowers’ high: Behavioural syn- as tai chi or walking in step. Furthermore, we chrony is correlated with elevated pain thresholds. would note that the links between synchrony Biology Letters, 6, 106–108. doi:10.1098/rsbl and endorphins and cohesion are currently spec- .2009.0670 ulative and research is required to fully opera- Dearman, J., & Francis, K. T. (1983). Plasma levels tionalize this potential mechanism. To fully un- of catecholamines, cortisol, and beta- endorphins derstand this phenomenon, we need to know if in male athletes after running 26.2, 6, and 2 miles. synchrony directly effects endorphins and if so, The Journal of Sports Medicine and Physical Fit- does this synchrony-induced endorphin activity ness, 23, 30–38. affect cohesion. As interesting as the effect ap- DePue, R. A., & Morrone-Strupinsky, J. V. (2005). A pears to be, the time is ripe for more research on neurobehavioral model of affiliative bonding: Im- plications for conceptualization a human trait of the basic mechanics and ramifications of this affiliation. Behavioral Brain Science, 28, 313–385. synchrony effect. doi:10.1017/S0140525X05290064 However, there are several limitations to the Dimitrov, D. M., & Rumrill, Jr, P. D. (2003). Pretest- current design, which are probably not limited posttest designs and measurement of change. to the present study. The current sample, like Work: A Journal of Prevention, Assessment and Cohen at al.’s (2010) and Sullivan and Rickers Rehabilitation, 20, 159–165. (2013) comprised experienced rowers. It is pos- Dunbar, R. I. M. (2010). The social role of touch in sible that experienced rowers may associate humans and primates: Behavioural function and synchrony with good teamwork and the resul- neurobiological mechanisms. Neuroscience and tant pain threshold effect may be related to this Biological Reviews, 34, 260–268. doi:10.1016/j social effectiveness as opposed to any syn- .neubiorev.2008.07.001 chrony effect. Furthermore, there are some po- Dunbar, R. I. M., Baron, R., Frangoou, A., Pearce, E., tential demand characteristics is the study. In van Leeuwen, J. C., Stow, J.,...vanVugt, M. (2012). Social laughter is correlated with an ele- this protocol, the experimenter is not blind to vated pain threshold. Proceedings of The Royal the condition of the participants, and it is pos- Society Biological Sciences, 279, 1161–1167. doi: sible that this may have influenced the results. 10.1098/rspb.2011.1373 Future research on the effect of behavioral syn- Dunbar, R. I., & Shultz, S. (2010). Bondedness and chrony on pain threshold or endorphin activity sociality. Behavior, 147, 775–803. doi.10.1163/ should use more strict methodologies, including 000579510X501151 blinded conditions and exit questionnaires to Estebe, J. P., Le Naoures, A., Chemaly, L., & Ecof- probe participants’ suspicions about research fey, C. (2000). Tourniquet pain in a volunteer questions. study: Effect of changes in cuff width and pres- This document is copyrighted by the American Psychological Association or one of its allied publishers. sure. Anasthesia, 55, 21–26. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. References Hove, J. M., & Risen, L. J. (2009). It’s all in the timing: Interpersonal synchrony increases affilia- Bernieri, F. (1988). Coordinated movement and rap- tion. Social Cognition, 27, 949–960. port in teacher-student interactions. Journal of Jamner, L. D., & Leigh, H. (1999). Repressive/ Nonverbal Behavior, 12, 120–138. doi:10.1007/ defensive coping, endogenous opioids and health: BF00986930 How a life so perfect can make you sick. Psychi- Blascovich, J., Mendes, W. B., Hunter, S. B., & atry Research, 8, 17–31. doi:10.1016/S0165- Salomon, K. (1999). Social “facilitation” as chal- 1781(98)00134-6 lenge and threat. Journal of Personality and Social Kirschner, S., & Tomasello, M. (2009). Joint drum- Psychology, 77, 68–77. doi:10.1037/0022-3514.77 ming: Social context facilitates synchronization in .1.68 preschool children. Journal of Experimental Child 128 SULLIVAN, RICKERS, AND GAMMAGE

Psychology, 102, 299–314. doi:10.1016/j.jecp ment Science, 26, 867–891. doi:10.1016/j.humov .2008.07.005 .2007.07.002 LaFrance, M., & Broadbent, M. (1976). Group rap- Ryan, E. D., & Kovacic, C. R. (1966). Pain tolerance port: Posture sharing as a nonverbal indicator. and athletic participation. Perceptual and Motor Group and Organizational Management, 1, 328– Skills, 22, 383–390. doi:10.2466/pms.1966.22.2 333. doi:10.1177/105960117600100307 .383 Lakens, D., & Stel, M. (2011). If they move in sync, Stel, M., van Baaren, R. B., & Vonk, R. (2007). they must feel in sync: Movement synchrony leads Effects of mimicking: Acting prosocially and be- to attributions of rapport and entitavitiy. Social ing emotionally moved. European Journal of So- Cognition, 29, 1–14. doi:10.1521/soco.2011.29.1.1 cial Psychology, 38, 965–976. doi:10.1002/ejsp Macrae, C. N., Duffy, O. K., Miles, L. K., & Law- .472 Strauss, B. (2002). Social facilitation in motor tasks: rence, J. (2008). A case of hand waving: Action A review of theory and research. Psychology of synchrony and person perception. Cognition, 109, Sport and Exercise, 3, 237–256. doi:10.1016/ 152–156. doi:10.1016/j.cognition.2008.07.007 S1469-0292(01)00019-x Marsh, K. L., Richardson, M. J., Baron, R. M., & Sullivan, P., & Rickers, K. (2013). The effects of Schmidt, R. C. (2006). Contrasting approaches to behavioral synchrony in groups of teammates and perceiving and acting with others. Ecological Psy- strangers. International Journal of Sport and Ex- chology, 18, 1–38. doi:10.1207/s15326969eco ercise Psychology, 1–6. doi:10.1080/1612197X 1801_1 .2013.750139 McNeill, W. H. (1995). Keeping together in time: Valdesolo, P., & DeSteno, D. (2011). Synchrony and Dance and drill in human history. Cambridge, the social tuning of compassion. Emotion, 11, MA: Harvard University Press. 262–266. doi:10.1037/a0021302 Miles, L. K., Nind, L. K., Henderson, Z., & Macrae, Valdesolo, P., Ouyang, J., & DeSteno, D. (2010). The C. N. (2010). Moving memories: Behavioral syn- rhythm of joint action: Synchrony promotes coop- chrony and memory for self and others. Journal of erative ability. Journal of Experimental Social Experimental Social Psychology, 46, 457–460. Psychology, 46, 693–695. doi:10.1016/j.jesp.2010 doi:10.1016/j.jesp.2009.12.006 .03.004 Miles, L. K., Nind, L. K., & Macrae, C. N. (2009). Wiltermuth, S. (2012). Synchrony and destructive The rhythm of rapport: Interpersonal synchrony obedience. Social Influence, 7, 78–89. doi: and social perception. Journal of Experimental 10.1080/15534510.2012.658653 Social Psychology, 45, 585–589. doi:10.1016/j Wiltermuth, S. S., & Heath, C. (2009). Synchrony .jesp.2009.02.002 and cooperation. Psychological Science, 20, 1–5. Mueller, G. P. (1981). Beta-endorphin immunoreac- doi:10.1111/j.1467-9280.2008.02253.x Zillman, D., Rockwell, S., Schweitzer, K., & Sundar, tivity in rat plasma: Variations in response to dif- S. (1993). Does humor facilitate coping with phys- ferent physical stimuli. Life Sciences, 29, 1669– ical discomfort? Motivation Emotion, 17, 1–21. 1674. doi.10.1016/0024-3205(81)90069-2 doi:10.1007/BF00995204 Richardson, M. J., Marsh, K. L., Isenhower, R. W., Goodman, J. R., & Schmidt, R. C. (2007). Rocking Received February 17, 2013 together: Dynamics of intentional and uninten- Revision received July 18, 2013 tional interpersonal coordination. Human Move- Accepted July 22, 2013 Ⅲ This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. Group Dynamics: Theory, Research, and Practice © 2014 American Psychological Association 2014, Vol. 18, No. 2, 129–137 1089-2699/14/$12.00 http://dx.doi.org/10.1037/gdn0000002 Reactions to Social Inclusion and Ostracism as a Function of Perceived In-Group Similarity

Donald F. Sacco Michael J. Bernstein The University of Southern Mississippi Pennsylvania State University–Abington

Steven G. Young Kurt Hugenberg Baruch College, City University of New York Miami University

Although ostracism is a powerfully aversive experience, recent evidence identifies factors capable of moderating the impact of ostracism, such as in-group status and the group’s essential nature. In the current work, 67 Caucasian American participants (47 women) were included or ostracized by either same-race (i.e., Caucasian American) or other-race (i.e., African American) targets on a between-subjects basis while playing the game Cyberball. Participants then indicated the extent to which they felt similar to the other Cyberball players as well as how satisfied their basic needs (e.g., belonging- ness, self-esteem) were during the game. Consistent with past research, we found that in-group and out-group status moderated the magnitude of reactions to social inclusion and ostracism; that is, ostracism hurts more and social inclusion feels better when it is implemented by fellow in-group as opposed to out-group members. Importantly, we extend these previous findings by demonstrating that differential reactions to social inclusion and ostracism are mediated by changes in participants’ self-perceived simi- larity with in-group members. These results identify a potential mechanism responsible for the differential impact of in-group–out-group status on reactions to social inclusion and ostracism.

Keywords: ostracism, intergroup relations, in-group–out-group, threat

Humans are intensely social animals. As Leary, 2005; Williams, 2007). However, there such, we place great value on social connections has been some debate as to whether all experi- and group memberships, with our self-esteem ences of ostracism and social inclusion are and emotional states being closely linked to our equally impactful. Although initial research in- social interactions and relationships (Baumeis- dicated that aversive responses to ostracism are ter & Leary, 1995). As a result, being socially reflexive and unmoderated by situational factors rejected or ostracized is an intensely aversive (e.g., Gonsalkorale & Williams, 2007; Williams social experience, whereas being socially ac- & Zadro, 2005), more recent research suggests cepted is decidedly positive (MacDonald & that responses to social ostracism are sensitive to several factors, including characteristics of

This document is copyrighted by the American Psychological Association or one of its allied publishers. the individuals engaging in the act of ostracism.

This article is intended solely for the personal use of the individual user and isDonald not to be disseminated broadly. F. Sacco, Department of Psychology, The Uni- For example, Bernstein and colleagues (2010) versity of Southern Mississippi; Michael J. Bernstein, De- demonstrated that being included or ostracized partment of Psychological and Social Sciences, Pennsylva- by in-group members exacerbates the intensity nia State University–Abington; Steven G. Young, of the experience (i.e., being ostracized by an Department of Psychology, Baruch College, City University of New York; Kurt Hugenberg, Department of Psychology, in-group hurts more than being ostracized by an Miami University. out-group). Correspondence concerning this article should be ad- Whereas these past findings demonstrate that dressed to Donald F. Sacco, Department of Psychology, The in-group status is an important moderator of University of Southern Mississippi, Owings-McQuagge Hall, 118 College Drive 5025, Hattiesburg, MS 39406. reactions to social inclusion and ostracism, re- E-mail: [email protected] search has yet to identify the mechanism under-

129 130 SACCO, BERNSTEIN, YOUNG, AND HUGENBERG

lying these differential reactions. To address and material support, offering protection and this research lacuna, we tested the hypothesis access to mates (e.g., Correll & Park, 2005). that inclusion within and ostracism by an in- Indeed, in-group bias, or the tendency to favor group can have more powerful influences on the in-groups over out-groups (Tajfel, 1982), is re- self (e.g., influencing belongingness and self- flected in numerous behaviors, ranging from esteem) precisely because such experiences can more positive evaluations of in-groups, greater change perceptions of self-similarity to the in- levels of prosocial behavior toward fellow in- group, which serves as a psychological media- group members, and allocation of more re- tor of the increased intensity of in-group inclu- sources to in-groups (e.g., Ferguson & Kelley, sion and ostracism. To this end, we first review 1964; Hein, Silani, Preuschoff, Batson, & the literature on the need to belong, in-group– Singer, 2010; Tajfel, Billig, Bundy, & Flament, out-group status, and perceptions of similarity, 1971). Consistent with the importance of social and then provide direct experimental evidence in-groups, Bernstein and colleagues (2010) re- for our proposed mediational hypothesis. cently reported two studies demonstrating that ostracism is more painful and social inclusion The Need to Belong more positive when carried out by fellow in- group members (cf. Goodwin, Williams, & Whereas stable social bonds have numerous Carter-Sowell, 2010). In their first study, partic- physical and psychological benefits, unfulfilled ipants were either ostracized or socially in- belongingness needs have a decidedly negative cluded by racial in-group or out-group mem- impact on well-being (Baumeister & Leary, bers; participants reported more threatened 1995). Specifically, acute experiences of ostra- affiliation needs following racial in-group ostra- cism induce negative mood and thwart basic cism (relatively to ostracism by racial out-group social needs satisfaction (e.g., self-esteem, be- members), but more satisfied basic affiliation longing; Williams & Zadro, 2005). Importantly, needs following social inclusion by racial in- these strong reactions to the experience of os- group as opposed to out-group members. Their tracism are largely unaffected by various indi- second study replicated this basic finding with vidual difference measures that seem intuitively political group membership; that is, participants relevant (e.g., trait self-esteem and social anxi- reported more social pain when ostracized by ety; Leary, Haupt, Strausser, & Chokel, 1998; individuals who shared their own party affilia- Zadro, Boland, & Richardson, 2006). tion (as opposed to ostracism by confederates of Indeed, prior work has even found that ostra- a different party affiliation), but greater satisfac- cism lowers mood and basic need satisfaction tion when socially included by individuals with equally, regardless of the characteristics of the the same political party affiliation as their own. ostracizing party. For example, Williams and Importantly, these authors found that this colleagues (2000) found that responses to ostra- strengthened experience of in-group inclusion cism did not differ based on in-group–out- and ostracism occurred most strongly for per- group distinctions (e.g., PC vs. Mac users; see ceivers who were made to see the in-group as also Smith & Williams, 2004). Furthermore, highly essentialized (i.e., participants who were individuals’ basic needs satisfaction is threat- led to believe that these groups possessed char- ened regardless of whether participants believe acteristics defined as relatively inborn, immuta- that they were ostracized by fellow participants ble, or genetically based; see Prentice & Miller, This document is copyrighted by the American Psychological Association or one of its allied publishers. (i.e., peers), by confederates who were trained 2007). This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. to ostracize them, or even by a computer pro- grammed to arbitrarily ostracize them (Zadro, Perceived Self-Similarity and Reactions to Williams, & Richardson, 2004). Finally, Gon- In-Group–Out-Group Inclusion and salkorale and Williams (2007) even found that Ostracism ostracism perpetrated by a disliked out-group (i.e., KKK members) is experienced as pain- In the current work, we tested the hypothesis fully as that perpetrated by a valued in-group. that the experience of becoming psychologi- Nonetheless, not all sources of affiliation cally similar to the including in-group or dis- equally satisfy belonging needs. Specifically, similar from the ostracizing in-group would me- valued social groups are a vital source of social diate the effect of in-group–out-group status on SOCIAL INCLUSION, OSTRACISM, AND SIMILARITY 131

reactions to inclusion and ostracism. Across nu- out-group members, and thus may be unrelated merous kinds of relationships, individuals pre- to the experiences of ostracism and social in- fer to interact with others who are similar across clusion. a variety of domains, including intelligence and To test our hypothesis, we manipulated the physical attractiveness (e.g., Feingold, 1988; race of supposed confederates in the context of Jensen, 1978). Furthermore, similarity is posi- the Cyberball paradigm (Williams & Jarvis, tively related to relationship satisfaction and 2006), thereby allowing us to independently longevity in both platonic and romantic rela- manipulate whether participants experienced tionships (e.g., Acitelli, Kenny, & Weiner, social inclusion or ostracism and whether that 2001; Gonzaga, Campos, & Bradbury, 2007; experience was initiated by members of one’s Russell & Wells, 1991). own racial in-group or out-group. Following Interestingly, this relation between similarity each of these experiences, we assessed partici- and liking is bidirectional, such that positive pants’ perceived similarity with the Cyberball interactions with others lead to increased per- “confederates” they played with as well as their ceptions of similarity. According to Morry’s basic needs satisfaction following the game. (2005) attraction-similarity model, attraction This design allowed us to replicate previous leads to perceptions of similarity between the findings that inclusion and ostracism perpe- self and others and these perceptions of simi- trated by racial in-groups produce more intense larity provide numerous benefits, ranging from interpersonal reactions (both positive and neg- feeling more understood and validated to in- ative) as well as to test the hypothesis that the creased feelings of positive affect and decreased effects of in-group inclusion and ostracism on loneliness. Put simply, we are likely to affiliate basic affiliation needs (dis)satisfaction are me- with similar others (e.g., Acitelli et al., 2001) diated by changes in perceived similarity with but also perceive those we affiliate with as more in-group members. similar to the self, and these feelings of inter- personal similarity are important for prompting Method both initial liking and securing the stable and long-term relationships that best satisfy belong- Participants ing needs (Baumeister & Leary, 1995). Based on these previous findings, we suggest The sample was composed of 67 (47 women; ϭ ϭ that inclusion by in-group members creates per- Mage 18.97 years, SD 1.28) Caucasian ceptions of enhanced similarity with members American undergraduate students at a large of that group, an outcome we propose may be midwestern university in the United States who responsible for enhanced need satisfaction fol- participated in exchange for partial course cred- lowing inclusion by in-group members. Con- it; non-Caucasian Americans did not participate versely, ostracism by in-group members may be in this study.1 A 2 (inclusionary status: ostra- interpreted as evidence of dissimilarity between cism, inclusion) ϫ 2 (agent race: same-race, the self and group, which may sharpen the neg- other-race) between-subjects design was em- ative experience of social exclusion. Specifi- ployed. cally, to the extent that inclusion by in-group members makes the in-group more attractive to Materials an individual, that person should indicate This document is copyrighted by the American Psychological Association or one of its allied publishers. heightened perceived similarity with that in- Basic Needs Scale. To assess participants’ This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. group, resulting in higher levels of interpersonal reactions to social inclusion and ostracism, we satisfaction; conversely, ostracism by in-group administered the Basic Needs Scale, used ex- members should lead individuals to perceive tensively in prior research (Williams et al., that group as less attractive, resulting in reduced 2000; Zadro, Williams, & Richardson, 2004). perceptions of similarity with the in-group and This questionnaire assesses four social needs potential interpersonal dissatisfaction (i.e., known to fluctuate with experiences of social Morry, 2005). However, because out-groups are inclusion and ostracism: belonging, perceived categorized as essentially dissimilar from the self, simple acts of inclusion and exclusion may 1 No effects of participant sex were observed; therefore, lead to little change in perceived similarity to this variable is not discussed further. 132 SACCO, BERNSTEIN, YOUNG, AND HUGENBERG

control, self-esteem, and meaningful existence ticipants were directed to play Cyberball, an (for a review, see Williams, 2007). The ques- Internet ball-tossing game ostensibly designed tionnaire included 16 items assessing perceived to help people hone their mental visualization threats to these four needs (e.g., “I felt discon- skills. Importantly, our White participants saw nected,” “I felt I had control over the situation,” either two White faces (i.e., in-group members) “I felt liked,” “I felt meaningless”); we asked or two Black faces (i.e., out-group members) as participants to provide their responses based on fellow players (on a between-subjects basis, to how they felt during the game (reflexive) rather manipulate racial in-group and out-group mem- than based on how they currently felt (reflec- bership). In addition, participants were ran- tive). Participants responded to each of the 16 domly assigned to either receive the ball questions using a 7-point Likert scale (1 ϭ not roughly one third of the time (inclusion condi- at all to 7 ϭ extremely), scored such that higher tion) or to receive the ball only twice at the numbers indicated more fulfillment of the re- beginning and then never again for the remain- spective need (i.e., higher numbers indicated a ing throws (ostracism condition). greater experience of belonging, higher self- Participants then completed the similarity esteem, etc.). question and the Basic Needs Scale. Partici- Perceived similarity. Using the same pants were debriefed and thanked for their par- 7-point Likert scale used to assess basic needs ticipation. We used a funneled debriefing ques- satisfaction, we also asked participants the ex- tionnaire to determine whether participants tent to which they agreed with the statement were suspicious about any aspect of our exper- “The other participants are similar to me.” imental procedures, including the Cyberball manipulation. Using an open-ended format, par- Procedure ticipants were asked, “Did you find anything odd or suspicious during the study?” and “Did The procedure was adapted closely from that you think anything in the study was fake?” If of Bernstein and colleagues (2010). Upon their participants answered affirmatively to either arrival in a university laboratory, participants question, they were asked to elaborate on their provided informed consent and were escorted to initial answer. individual cubicles. Specifically, they were in- formed that they were participating in a study Results involving online communication and mental vi- sualization, and that because the study was oc- Manipulation Check Effectiveness curring in multiple departments on the univer- sity campus, they would be playing with other To determine whether our Cyberball manip- students located elsewhere on campus. In actu- ulation was believable to participants, we coded ality, the other participants were computer- participants’ responses to our funneled debrief- controlled agents programmed to either socially ing questions for suspicion. Specifically, we include or ostracize the participants on a be- created a “player suspicion” variable and noted tween-subjects basis. To facilitate our experi- any participants who indicated that they found mental design, we told participants that they the two Cyberball players generally or their race would be able to see a photograph of the two to be fake, odd, or suspicious. Anyone mention- participants with whom they would be playing ing such a comment in either question was This document is copyrighted by the American Psychological Association or one of its allied publishers. during the game itself, and that the two partic- assigned a 1, and all others who did not mention This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. ipants would be able to see a picture of them anything were assigned a 0. This coding re- (thus allowing for our manipulation of agent sulted in a total of six of 67 respondents having race). Participants were then instructed to stand made comments regarding some aspect of the against a white wall while the experimenter players being fake, odd, or suspicious. Impor- took a digital photograph of them, which the tantly, only one participant’s suspicion was ac- experimenter then ostensibly uploaded to the tually based on the race of the Cyberball play- computer network prior to the start of the game. ers. After having their picture taken and complet- We conducted analyses to determine whether ing a series of filler questions supposedly mea- these differences varied across experimental suring “mental visualization” capabilities, par- conditions. We first conducted a chi-square SOCIAL INCLUSION, OSTRACISM, AND SIMILARITY 133

analysis, using Cyberball target race (Caucasian of variance (ANOVA).2 A main effect of inclu- vs. African American) as the independent vari- sionary status emerged, F(1, 63) ϭ 105.73, p Ͻ 2 able. This analysis yielded a nonsignificant dif- .001, ␩p ϭ .63, such that included participants ference, ␹2(1) ϭ 0.67, p ϭ .41, indicating that (M ϭ 4.44, SE ϭ 0.14) reported substantially suspicion about the players did not vary based more fulfilled basic needs than did ostracized on whether participants were playing with participants (M ϭ 2.50, SE ϭ 0.13). Impor- White or Black agents. We next conducted a tantly, this main effect was qualified by a sig- similar chi-square analysis using Cyberball con- nificant Inclusionary Status ϫ Agent Race in- 2 dition (social inclusion vs. ostracism) as the teraction, F(1, 63) ϭ 5.41, p ϭ .023, ␩p ϭ .079 independent variable; again, this analysis re- (see Figure 1). Simple effects analyses revealed vealed no difference between conditions, that participants ostracized by same-race agents ␹2(1) ϭ 0.35, p ϭ .56. To determine whether (M ϭ 2.22, SD ϭ 0.58) had less fulfilled basic these factors (social experience and agent race) needs than did those ostracized by other-race interacted, we conducted a binary logistic re- agents (M ϭ 2.78, SD ϭ 0.88), t(35) ϭ 2.20, gression, dummy coding (1) inclusion ϩ Black p ϭ .032, d ϭ 0.75. Although participants in- agents, ostracism ϩ White agents, and ostra- cluded by same-race agents were directionally cism ϩ Black agents against the referent group more satisfied (M ϭ 4.61, SD ϭ 0.83) than were of inclusion ϩ White agents (0). The regression those included by other-race agents (M ϭ 4.28, was nonsignificant, ␹2(3) ϭ 2.82, p ϭ .42, for SD ϭ .73), this effect did not reach statistical each of the three dummy-coded variables en- significance, t(28) ϭ 1.15, p Ͼ .20, d ϭ 0.42. tered into the logistic regression. This significant interaction between inclusion- In a final analysis, we dummy coded both ary status and target race, however, is consistent social experience (0 ϭ exclusion, 1 ϭ inclu- with previous findings indicating that although sion) and agent race (0 ϭ Black, 1 ϭ White) ostracism always feels worse than social inclu- and entered them simultaneously as factors in a sion, such effects can be qualified by the in- binary logistic regression, using our “player group–out-group relationship of the rejecter suspicion” factor as the dependent variable. and rejected (Bernstein et al., 2010). This analysis allowed us to examine the effect of each of these factors while also accounting Mediation Analysis for the variability due to the other factor. This analysis also yielded a nonsignificant regres- We then investigated whether the stronger sion, ␹2(2) ϭ 0.98, p ϭ .61. responses to in-group inclusion and ostracism All evidence suggested that suspicion about were mediated by changes in the perceived sim- the Cyberball players, including their race, did ilarity of the self to in-group members.3 We not differ between conditions of our experi- predicted that participants included by same- ment. Furthermore, this identical experimental race participants should see the self as more procedure has been used in previous work and similar to the agents, relative to participants similarly low levels of suspicion were docu- experiencing ostracism. We further predicted mented (see Bernstein et al., 2010). As such, the that increased perceived similarity should be current procedure possessed an adequate level positively correlated with increased fulfillment of experimental realism. of basic affiliation needs. Finally, we predicted that the relationship between inclusionary status This document is copyrighted by the American Psychological Association or one of its allied publishers. and basic needs would be mediated by the This article is intended solely for the personal use of the individualBasic user and is not to be disseminated broadly. Needs

The individual needs were highly related 2 (␣ϭ.94); as such, we formed a composite basic Treating each of the four subscales as separate depen- dent variables in four separate ANOVAs had no effect on needs score for each participant (higher values the interaction pattern (ps Ͻ .07). represent more satisfied basic needs; see Mc- 3 Because perceived similarity and belongingness are po- Connell, Brown, Shoda, Stayton, & Martin, tentially psychologically similar constructs, we conducted a 2011, for similar procedures). We then sub- correlational analysis to determine how strongly they were related to one another, r(67) ϭ .49, p Ͻ .01. Because these jected these scores to a 2 (inclusionary status: two concepts were only moderately correlated, perceived inclusion vs. ostracism) ϫ 2 (agent race: same- similarity was an adequate variable to include as a psycho- race vs. other-race) between-subjects analysis logical mediator. 134 SACCO, BERNSTEIN, YOUNG, AND HUGENBERG

Figure 1. The effect of inclusion and ostracism on basic needs satisfaction as a function of the in-group–out-group status of the agents.

aforementioned increase in perceived similarity. statistically significant in the direction predicted However, we believe this relationship should by the mediation hypothesis” (Preacher & only be true for participants playing with same- Hayes, 2004, p. 719). Having already estab- race agents. We hypothesized that de facto dis- lished these requirements, we next performed similarity with racial out-groups would be un- the bootstrapping procedures using the macros affected by the relatively minor inclusionary provided by Preacher and Hayes (2008) to de- and exclusionary behaviors used in our task, termine whether the direct effect of inclusionary and as such, we did not expect similarity to status on basic needs was mediated by percep- mediate the relationship between inclusionary tions of similarity. This analysis yielded a 95% status and basic needs satisfaction. bias-corrected confidence interval (based on We therefore conducted a moderated media- 5,000 bootstrap samples) for the indirect effect tion analysis (i.e., a conditional indirect effect; that did not include zero (95% CI [0.066, 0.70]). see Preacher, Rucker, & Hayes, 2007) to exam- This analysis indicated that perceptions of per- ine our specific hypothesis, and we used the ceived similarity mediated the direct effect. Preacher et al. (2007) existing SPSS macro to Greater social inclusion led to stronger percep- conduct this analysis. Before testing the moder- tions of similarity (B ϭ 0.92, p ϭ .002), and as ated mediation, we began by dummy coding the similarity increased, so too did participants’ ba- independent variable of inclusionary status (1 ϭ sic needs satisfaction (B ϭ 0.28, p Ͻ .001). The inclusion, 0 ϭ ostracism) and entered it as a direct effect, however, was mediated when in- predictor of the composite basic needs score. cluding similarity (B ϭ 1.68, p Ͻ .001). We also included our similarity measure as our However, we wished to examine whether this mediator and target race as the moderating vari- was true both when playing with same-race and able (0 ϭ White, 1 ϭ Black). This analysis other-race agents (i.e., did target race moderate produced a significant overall model, F(3, the mediated effect?). We conducted a moder- 63) ϭ 5.82, p ϭ .014, R2 ϭ .22, which included ated meditation analysis using Model 7 with the This document is copyrighted by the American Psychological Association or one of its allied publishers. our previously described direct effect on the W moderator being target race (0 ϭ White, 1 ϭ This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. basic needs, such that inclusion increased basic Black). This yielded two bootstrapping models, needs satisfaction relative to ostracism (B ϭ each with 5,000 iterations. When participants 1.94, p Ͻ .001). We next examined whether the played with White agents, similarity did indeed direct effect was mediated by perceptions of mediate the effect of inclusion and ostracism on similarity. To test this hypothesis, we employed basic needs satisfaction (95% bias-corrected the bootstrapping method developed by confidence interval did not include zero; 95% Preacher and Hayes (2004, 2008). This method CI [0.079, 0.91]). However, similarity did not requires that a relationship between an indepen- mediate the effect when participants played dent variable and a dependent variable exist to with Black agents (95% bias-corrected confi- be mediated, and that the “indirect effect be dence interval did include zero; 95% CI [Ϫ0. SOCIAL INCLUSION, OSTRACISM, AND SIMILARITY 135

054, 0.39]). Thus, our moderated mediation hy- tancing oneself from the aggressor and perhaps pothesis was supported. also protecting generally positive perceptions of the in-group (e.g., Doosje, Ellemers, & Spears, Discussion 1995). Alternatively, ostracism from the in- group may threaten one’s perceived standing Consistent with past research, we found that within the in-group (Williams, 2007), causing the social pain stemming from ostracism is the observed dissimilarity and intensified nega- more severe when administered by fellow in- tivity. group members (e.g., Bernstein et al., 2010). Although past theory and research (e.g., Wil- Conversely, social inclusion is experienced as liams, 2007) proposed that early stage reactions relatively more positive when administered by to ostracism are essentially reflexive and insen- in-group relative to out-group members. Impor- sitive to situational factors, the current results tantly, we found that these enhanced effects of and other recent work (Bernstein et al., 2010) social inclusion and ostracism observed in the suggest that responses to ostracism are not sim- in-group condition are mediated by perceived ply “all or nothing,” and instead can be more similarity with those agents of inclusion or os- nuanced and contextualized. Furthermore, the tracism. Not only do in-group members seem current results also mesh nicely with social pain more similar to the self when they include par- theory (MacDonald & Leary, 2005), which pro- ticipants (and more dissimilar when they ostra- poses that social and psychical pain are phe- cize participants), but changes in the perception nomenologically and physiologically linked ex- of similarity with agents mediate the relation- periences (e.g., Eisenberger, Lieberman, & ship between inclusionary status and the fulfill- Williams, 2003). To the extent that reactions to ment of basic affiliation needs. Critically, this ostracism are similar to responses to physical relationship was only true when participants pain, it is sensible that the most serious social played the game with racial in-group members; injuries (e.g., being ostracized by in-groups) perceptions of similarity did not change, regard- should result in the most acute experiences of less of the inclusionary experience, in the racial social pain. out-group condition (i.e., moderated media- Nonetheless, a number of open questions still tion). remain. First, given that the current work used Several theoretical positions are congruent race as the operational definition of in-group with our findings. First, out-group members are and out-group, it may appear unclear whether naturally constrained with respect to how simi- the current effects are due to an in-group effect lar they may be perceived to in-group members. or due to status or power differences among the Among racial out-groups, for whom the signals in-group and out-group. Although a sensible of group differences (i.e., skin tone) are salient alternative explanation, this hypothesis is not (e.g., Ito & Urland, 2003), it may be difficult to easily congruent with previous findings. Indeed, ever completely overcome that difference so Bernstein and colleagues (2010) found that both long as “race” is the primary category that de- Black and White participants showed equiva- fines the targets. This explanation is further lent effects when socially included or ostracized supported by evidence showing that out-groups by racial in-group versus out-group members. are perceived as being more homogenous as Clearly, this pattern of data does not align easily compared with in-groups (Park, Ryan, & Judd, with a status or power interpretation of race. This document is copyrighted by the American Psychological Association or one of its allied publishers. 1992). It may be that when individuals think of Second, in the current work, we did not test This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. in-group members, they are better able to imag- whether similar results would be observed for ine a greater degree of variation in how similar other in-group–out-group distinctions. Here, they are to themselves; receiving positive feed- too, we can point to past research to address this back (such as acceptance) may enhance these issue. Given that the differential responses to perceptions of similarity, because positive inter- ostracism and social inclusion by in-group and personal interactions can increase perceived out-group members only appear to occur for similarity (Morry, 2005). At the same time, highly essentialized groups (e.g., race; see receiving negative feedback may result in per- Bernstein et al., 2010), but not for more arbi- ceiving the agent as less similar to themselves trary groups or groups with weak affiliations (perhaps the group as well) as a means of dis- (e.g., PC vs. Mac users; Smith & Williams, 136 SACCO, BERNSTEIN, YOUNG, AND HUGENBERG

2004), this indicates that the current mediational Journal of Abnormal Psychology, 69, 223–228. pattern for in-groups would likely hold only for doi:10.1037/h0046572 subjectively relevant and highly essentialized Gonsalkorale, K., & Williams, D. (2007). The KKK in-groups. won’t let me play: Ostracism even by a despised In conclusion, consistent with past research, outgroup hurts. European Journal of Social Psy- we found that reactions to social inclusion and chology, 37, 1176–1186. doi:10.1002/ejsp.392 Gonzaga, G. C., Campos, B., & Bradbury, T. (2007). ostracism are exacerbated when perpetrated by Similarity, convergence, and relationship satisfac- fellow racial in-group members (compared with tion in dating and married couples. Journal of racial out-group members). Importantly, these Personality and Social Psychology, 93, 34–48. exacerbated reactions stemming from in-group doi:10.1037/0022-3514.93.1.34 inclusion and ostracism are explained by fluc- Goodwin, S. A., Williams, K. D., & Carter-Sowell, tuations in perceived self-similarity with in- A. R. (2010). The psychological sting of stigma: group members. Future research would benefit The costs of attributing ostracism to racism. Jour- by understanding additional mechanisms poten- nal of Experimental Social Psychology, 46, 612– tially responsible for differentiated reactions to 618. doi:10.1016/j.jesp.2010.02.002 inclusion and ostracism by in-group members. Hein, G., Silani, G., Preuschoff, K., Batson, C. D., & Singer, T. (2010). Neural responses to ingroup and outgroup members’ suffering predict individual References differences in costly helping. Neuron, 68, 149– 160. doi:10.1016/j.neuron.2010.09.003 Acitelli, L. K., Kenny, D. A., & Weiner, D. (2001). Ito, T. A., & Urland, R. (2003). Race and gender on The importance of similarity and understanding of the brain: Electrocortical measures of attention to partners’ marital ideals to relationship satisfaction. the race and gender of multiply categorizable in- Personal Relationships, 8, 167–185. doi:10.1111/ dividuals. Journal of Personality and Social Psy- j.1475-6811.2001.tb00034.x chology, 85, 616–626. doi:10.1037/0022-3514.85 Baumeister, R. F., & Leary, M. R. (1995). The need .4.616 to belong: Desire for interpersonal attachments as Jensen, A. R. (1978). Genetic and behavioral effects a fundamental human motivation. Psychological of nonrandom mating. In R. T. Osborne, C. E. Bulletin, 117, 497–529. doi:10.1037/0033-2909 Noble, & N. J. Weyl (Eds.), Human variation: .117.3.497 Biopsychology of age, race, and sex (pp. 51–105). Bernstein, M. J., Sacco, D. F., Young, S. G., Hugen- New York, NY: Academic Press. berg, K., & Cook, E. (2010). Being “in” with the Leary, M. R., Haupt, A. L., Strausser, K. S., & in-crowd: The effects of social exclusion and in- Chokel, J. T. (1998). Calibrating the sociometer: clusion are enhanced by the perceived essentialism The relationship between interpersonal appraisals of the ingroup and outgroup. Personality and So- and state self-esteem. Journal of Personality and cial Psychology Bulletin, 36, 999–1009. doi: Social Psychology, 74, 1290–1299. doi:10.1037/ 10.1177/0146167210376059 Correll, J., & Park, B. (2005). A model of the ingroup 0022-3514.74.5.1290 as a social resource. Personality and Social Psy- MacDonald, G., & Leary, M. R. (2005). Why does chology Review, 9, 341–359. doi:10.1207/ social exclusion hurt? The relationship between s15327957pspr0904_4 social and physical pain. Psychological Bulletin, Doosje, B., Ellemers, N., & Spears, R. (1995). Per- 131, 202–223. doi:10.1037/0033-2909.131.2.202 ceived intragroup variability as a function of group McConnell, A. R., Brown, C. M., Shoda, T. M., status and identification. Journal of Experimental Stayton, L. E., & Martin, C. E. (2011). Friends Social Psychology, 31, 410–436. doi:10.1006/jesp with benefits: On the positive consequences of pet This document is copyrighted by the American Psychological Association or one of its allied publishers. .1995.1018 ownership. Journal of Personality and Social Psy- This article is intended solely for the personal use of the individualEisenberger, user and is not to be disseminated broadly. N. I., Lieberman, M. D., & Williams, chology, 101, 1239–1252. doi:10.1037/a0024506 K. D. (2003, October 10). Does rejection hurt? An Morry, M. M. (2005). Relationship satisfaction as a fMRI study of social exclusion. Science, 302, predictor of similarity ratings: A test of the attrac- 290–292. doi:10.1126/science.1089134 tion-similarity hypothesis. Journal of Social and Feingold, A. (1988). Matching for attractiveness in Personal Relationships, 22, 561–584. doi:10.1177/ romantic partners and same-sex friends: A meta- 0265407505054524 analysis and theoretical critique. Psychological Park, B., Ryan, C. S., & Judd, C. M. (1992). Role of Bulletin, 104, 226–235. doi:10.1037/0033-2909 meaningful subgroups in explaining differences in .104.2.226 perceived variability for in-groups and out-groups. Ferguson, C. K., & Kelley, H. H. (1964). Significant Journal of Personality and Social Psychology, 63, factors in overevaluation of own-group’s product. 553–567. doi:10.1037/0022-3514.63.4.553 SOCIAL INCLUSION, OSTRACISM, AND SIMILARITY 137

Preacher, K. J., & Hayes, A. F. (2004). SPSS and Williams, K. D. (2007). Ostracism. Annual Review of SAS procedures for estimating indirect effects in Psychology, 58, 425–452. doi:10.1146/annurev simple mediation models. Behavior Research .psych.58.110405.085641 Methods, Instruments, & Computers, 36, 717–731. Williams, K. D., Cheung, C. K. T., & Choi, W. doi:10.3758/BF03206553 (2000). Cyberostracism: Effects of being ignored Preacher, K. J., & Hayes, A. F. (2008). Asymptotic over the Internet. Journal of Personality and So- and resampling strategies for assessing and com- cial Psychology, 79, 748–762. doi:10.1037/0022- paring indirect effects in multiple mediator mod- 3514.79.5.748 els. Behavior Research Methods, 40, 879–891. Williams, K. D., & Jarvis, B. (2006). Cyberball: A doi:10.3758/BRM.40.3.879 program for use in research on ostracism and in- Preacher, K. J., Rucker, D. D., & Hayes, A. F. terpersonal acceptance. Behavior Research Meth- (2007). Addressing moderated mediation hypoth- ods, 38, 174–180. doi:10.3758/BF03192765 eses: Theory, methods, and prescriptions. Multi- Williams, K. D., & Zadro, L. (2005). Ostracism: An variate Behavioral Research, 42, 185–227. doi: indiscriminate early detection system. In K. D. 10.1080/00273170701341316 Williams, J. P. Forgas, & W. von Hippel (Eds.), Prentice, D. A., & Miller, D. T. (2007). Psychologi- The social outcast: Ostracism, social exclusion, cal essentialism of human categories. Current Di- rejection, and bullying (pp. 19–34). New York, rections in Psychological Science, 16, 202–206. NY: Psychology Press. doi:10.1111/j.1467-8721.2007.00504.x Zadro, L., Boland, C., & Richardson, R. (2006). How Russell, R. J., & Wells, P. A. (1991). Personality long does it last? The persistence of the effects of similarity and quality of marriage. Personality and ostracism in the socially anxious. Journal of Ex- Individual Differences, 12, 407–412. doi:10.1016/ perimental Social Psychology, 42, 692–697. doi: 0191-8869(91)90057-I 10.1016/j.jesp.2005.10.007 Smith, A., & Williams, K. D. (2004). R U there? Zadro, L., Williams, K. D., & Richardson, R. (2004). Effects of ostracism by cell phone messages. How low can you go? Ostracism by a computer is Group Dynamics: Theory, Research, and Practice, sufficient to lower self-reported levels of belong- 8, 291–301. doi:10.1037/1089-2699.8.4.291 ing, control, self-esteem, and meaningful exis- Tajfel, H. (1982). Social psychology of intergroup tence. Journal of Experimental Social Psychology, relations. Annual Review of Psychology, 33, 1–39. 40, 560–567. doi:10.1016/j.jesp.2003.11.006 doi:10.1146/annurev.ps.33.020182.000245 Tajfel, H., Billig, M. G., Bundy, R. P., & Flament, C. (1971). Social categorization and intergroup be- Received June 22, 2013 haviour. European Journal of Social Psychology, Revision received August 27, 2013 1, 149–178. doi:10.1002/ejsp.2420010202 Accepted August 27, 2013 Ⅲ

E-Mail Notification of Your Latest Issue Online!

Would you like to know when the next issue of your favorite APA journal will be This document is copyrighted by the American Psychological Association or one of its allied publishers. available online? This service is now available to you. Sign up at http://notify.apa.org/ and This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. you will be notified by e-mail when issues of interest to you become available! Group Dynamics: Theory, Research, and Practice © 2014 American Psychological Association 2014, Vol. 18, No. 2, 138Ð158 1089-2699/14/$12.00 DOI: 10.1037/gdn0000003 Single-Case Designs in Group Work: Past Applications, Future Directions

Mark J. Macgowan and Stephen E. Wong Robert Stempel College of Public Health & Social Work, Florida International University

This paper examined how group work researchers and practitioners have used single- case designs (SCDs) to evaluate interventions for improving group processes and outcomes. Fifty-one group work studies published from 1972 to 2012 using SCDs and addressing a wide range of problems were identified. The 8 types of SCDs applied in these studies are discussed in relation to their methodology, relative strengths and limitations, and applications in group work research and practice evaluation. Selected studies, which varied in quality and rigor, are described to illustrate design features and utility for group work. The review reveals the challenges and opportunities in using SCDs, and it provides recommendations for designs best suited to answer particular questions.

Keywords: group work, single-case designs, single-subject designs, single-system designs, practice evaluation

Single-case designs (SCDs), also known as numerous tests for determining statistically sig- single-subject, single-participant,N-of-1, and nificant change. single-system evaluation designs, can evaluate SCDs have been structured to examine spe- effects of an intervention with only one subject cial features of the group modality. The design or group. In most SCDs, repeated measures are may focus on functioning of individuals within taken on a single subject or group under no- the group, a subgroup of individuals, or the treatment (baseline) and treatment conditions, whole group. SCDs can also include variables and the subject or group serves as its own con- that reflect the group experience, such as mem- trol. Controlled SCDs demonstrate experimental ber satisfaction, nature of interactions, and lead- control by replicating treatment effects within ership. There are many potential focal points each single-case study through procedures such as such as intervention theory, individual group repeatedly introducing and removing the interven- members, group structures, group processes and tion; successively introducing the intervention leadership, all of which relate to group work across different targets, settings, or subjects; outcomes (Burlingame, Strauss, & Joyce, randomly alternating between various condi- 2013). tions; or incrementally changing the level of the intervention. Results of SCDs are typically graphed for visual analysis, but there are also Benefits of Single-Case Designs for Group Work

This document is copyrighted by the American Psychological Association or one of its allied publishers. SCDs are considered a useful research design This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. for clinical practice across professions includ- Mark J. Macgowan and Stephen E. Wong, Robert Stem- ing psychology, social work, special education, pel College of Public Health & Social Work, Florida Inter- and counselor education (Bloom, Fischer, & national University. Both authors contributed equally to the writing of this Orme, 2009; Heppner, Kivlighan, & Wampold, paper and are listed in alphabetical order. 2008; Horner et al., 2005; Kazdin, 2003; Lun- Correspondence concerning this article should be ad- dervold & Belwood, 2000; Vonk, Tripodi, & dressed to Mark J. Macgowan or Stephen E. Wong, School Epstein, 2007). They can be a valid source of of Social Work, Robert Stempel College of Public Health and Social Work, Florida International University, 11200 data in determining empirically supported inter- SW 8th Street, Miami, FL 33199. E-mail: macgowan@fiu ventions (Chambless et al., 1998). In group .edu; wongse@fiu.edu work it has been recommended as a valuable

138 SINGLE-CASE DESIGNS GROUP WORK 139

evaluation method (Garrett, 2005; Toseland & Regardless of the reasons, the focus in group Rivas, 2012; Zastrow, 2006). research has been on between-groups designs, There are a number of important advantages with little attention on SCD technology. With of SCDs for research and practice. SCDs offer a respect to clinical practice, relatively few prac- means of obtaining valuable pilot data before titioners use SCDs, but instead rely on informal larger between-groups designs are utilized methods to determine progress (Foster, Watson, (Petry, Martin, & Finocche, 2001; Petry, Wein- & Young, 2002; Galassi & Gersh, 1993; Ven- stock, & Alessi, 2011). Between-groups designs timiglia, Marschke, Carmichael, & Loew, require a control group and a sufficient number 2000). In group work, the findings are similar. of subjects to satisfy the power requirements of A survey of the use of SCDs by 54 school the statistical test to be used for data analysis. practitioners found that none used SCDs in their However, SCDs can be conducted with a single groups but rather relied on subjective methods person or one group containing as many mem- (Garrett, 2005, p. 251). bers who attend. SCDs are also the most prac- There have been a number of reviews of tical technology available for evaluating one’s SCDs in group work (Fike, 1980; Johnson, own practice, which is required by the Ameri- Beckerman, & Auerbach, 2001; Patterson & can Group Psychotherapy Association (AGPA, Basham, 2002; Robison, Morran, & Hulse- 2007), Association for Specialists in Group Killacky, 1989; Rose, 1977), a few of which can Work (ASGW, 2000, 2007), and the Associa- be summarized. In his book on behavioral group tion for the Advancement of Social Work with work, Rose (1977, pp. 66Ð72) described seven Groups (AASWG, 2006). Another benefit of applications of SCDs, illustrating five simple SCDs is their flexibility, as they can accommo- baseline-treatment designs (AB), one reversal date a trial-and-error process allowing clini- design (ABAB), and one multiple-baseline de- cians to try out different group interventions in sign (MBD). These examples illustrated how to the search for an effective one (Newman & use SCDs to assess change in variables such as Wong, 2004). This also makes SCDs particularly member-to-member interactions, member well suited for practitioners using evidence-based praise statements, leader-to-member interac- group treatments with diverse populations. Group tions, leader suggestion giving, positive affec- workers using SCDs can determine whether an tive responses, suggestion-opinion giving re- evidence-supported intervention or technique, sponses, information-giving responses, and validated with other populations, is effective subjective appraisal of intervention procedures with the particular cultural groups they are serv- by group members. ing (Macgowan, 2008; Macgowan & Hanbidge, Fike (1980) noted how SCDs could be used 2014). flexibly to capture the nature of change in groups. He described the possible relationships between multiple formative/process variables Previous Reviews and outcome variables in groups. He argued that designs simply recording individual outcomes, Compared with between-groups designs, “do not give attention to all of the relevant SCDs seldom appear in published group work group information” (Fike, 1980, p. 42). Accord- research. Reviews of the history of group work ing to Fike, SCDs should also include other research since the 1950s made no mention of group variables such as member satisfaction This document is copyrighted by the American Psychological Association or one of its allied publishers. SCDs (Barlow, Burlingame, & Fuhriman, 2000; with the group, quality of interactions, and This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. Barlow, Fuhriman, & Burlingame, 2004; Feld- group leadership. man, 1987; Frank, 1975; Horne & Rosenthal, Robison and colleagues (Robison et al., 1997; McGrath, 1997). In a review of 54 studies 1989) described two SCDs (ABAB and MBD) in a decade of group work research up to 1994, and statistical procedures (time-series analyses only two SCDs were identified (Tolman & Mo- and nonparametric tests) that group workers could lidor, 1994). As the authors noted, the lack of use to evaluate the effectiveness of their interven- studies might have been a reflection of “publi- tions. They cautioned that ABAB designs might cation biases against such studies, rather than not be able to determine causality with some researcher rejection of such designs” (Tolman dependent variables, such as cohesion, as the & Molidor, 1994, p. 152). developmental processes may occur irrespec- 140 MACGOWAN AND WONG

tive of an intervention and might not be revers- In Table 1 and the following section we re- ible through withdrawal of treatment. In such view SCDs used in extant group work research, cases, a MBD across clients or groups could be beginning with the simplest and progressing to used to demonstrate causality. To improve the the more complex designs. We briefly describe internal validity of findings from a multiple- the logic behind each design, what the design baseline across groups design, the authors rec- can accomplish, and how it was properly or ommended random assignment of clients into less-than optimally applied in previous studies. the different groups. We also note when these designs were used to investigate particular questions about group Purpose and Scope of Review processes or outcomes. Discussion and recom- mendations for research and practice follow. This paper advances previous reviews by in- cluding more recent studies; examining a Review of Single-Case Designs in Group broader range of SCD designs, their individual Work Studies methodologies, and relative strengths and weak- nesses; and making recommendations for prac- The first SCD in group work appeared in tice and research with groups. This review is 1972 with an additional five published in that restricted to SCDs involving facilitated small decade (see Table 1). There were 15 published groups whose purpose is to improve the psy- in the 1980s, 11 in the 1990s, and 19 since the chosocial or socioemotional functioning of its year 2000, representing a considerable increase members. The term small group “implies the from the previous decade. The discipline with ability of members to identify themselves as the most publications is Psychology followed members, to engage in interaction, and to ex- by Social Work. Most of the studies have orig- change thoughts and feelings among themselves inated in the United States but studies were also though verbal, nonverbal, and written commu- conducted in Argentina, New Zealand, and the nication processes. Members can meet face-to- United Kingdom. face, by telephone or video, or through com- puter networks” (Toseland & Rivas, 2012, B Design (Treatment Only) p. 11). We used a number of methods to locate and Four of the publications in Table 1 used a B identify SCDs in the literature. We restricted the design, which first appeared in the 1980s. It is search to scholarly journals of all years, up to the simplest SCD and it is uncontrolled, con- 2012. We searched a number of databases in- sisting of merely taking repeated measurements cluding social service abstracts, PsycINFO, So- during treatment. When outcome measures are ciofile, Medline, and Social Work Abstracts. collected in this manner, various results are The reference lists of other reviews (Garrett, possible: (a) measures might change in the de- 2005; Johnson et al., 2001; Maughan, Christian- sired direction, indicating a clinical improve- sen, Jenson, Olympia, & Clark, 2005; Smith, ment; (b) measures might not change signifi- 2012) were also examined, in addition to those cantly, indicating a lack of improvement despite in recent SCD studies. The following terms treatment; (c) measures might change in the related to group work and SCDs were used to undesired direction, indicating clinical deterio- locate the literature, searching in the title, key- ration; or (d) some intermediate result that falls This document is copyrighted by the American Psychological Association or one of its allied publishers. words, and abstract: single subject, single sys- between a and b or between b and c. These This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. tem, single case, multiple baseline, withdrawal, results are important for deciding whether to reversal, alternating treatments design, AB, continue treatment and for making future treat- ABA, BAB, ABAB, appearing with any of the ment recommendations, making the B design following terms: group work, group counseling, valuable for clinicians. However, although the group treatment, group intervention, group ther- B design is easy to do and it is superior to apy, group prevention, group approach, group unsystematic or no collection of outcome data, psychotherapy, group training, and group in- it has serious limitations. If the client does im- struction. After an iterative process to select prove during treatment, numerous alternative studies meeting the above criteria, 51 studies explanations such as history, maturation, testing were found. (Campbell & Stanley, 1966; Cook & Campbell, This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Table 1 Single-Case Designs Previously Applied in Group Work Research

Considerations for Studies Using this Design Brief Description Advantages Disadvantages Group Work Design Case Study & Uncontrolled Designs B Monitor outcome measures Simplest & easiest SCD to Unable to draw causal Previous researchers (Cetingok & Hirayama, & Fontao ;ءduring treatment. conduct. Does not inferences. have used this design 1983 ;ءGraphic display can require a no-treatment Susceptible to to monitor if Mergenthaler, 2008 show improvement, no baseline phase. confounds from formative group O’Brien et al., & Patterson ;ءchange, or worsening history, maturation, processes are 1999 associated with testing, natural occurring as Basham, 2002)

intervention. healing, & expected. WORK GROUP DESIGNS SINGLE-CASE extraneous events. AB & related After collecting stable Relatively easy & practical Sometimes not possible Must consider the (Hall, 2006; Johnson, ;ءdesigns baseline data or data to conduct. to delay treatment to possibility that et al., 2001 moving in a counter- collect baseline data. emergent changes Ledgerwood, et al., therapeutic direction, Substantial changes coinciding with 2008 apply the intervention evident in a single treatment are due to [counterbalanced & Nelson ;ء[for a period at least AB design can only developmental or AB equal to the length of suggest a treatment internal group Kelley, 2001; Rubin, ,Stevens ;ءbaseline. effect. Should not processes. 1991 make causal 2006; Toseland, inferences from this Krebs, & Vahsen, design because it is 1978; Wade, Ortiz, susceptible to & Gorman, 2007 confounds from [interrupted time- extraneous events. series design]) ABC, ABCD When initial treatment (B Accommodates a flexible Like an AB design, is Developmental & (Fry, et al., 2009 & Phaneuf ;ء[AB1B2, or B1) proves trial & error process. A susceptible to internal group [ABC AB1B2B3,& unsatisfactory, apply a series of interventions confounds from processes may also McIntyre, 2011 (ء[related designs second treatment (C) or can be applied until a extraneous events confound the internal [ABCD the same treatment at a satisfactory outcome is that might coincide validity of these higher intensity (B2). If attained. with the seemingly designs. performance under C or effective treatment. B2 proves unsatisfactory, Additional threats to apply a third treatment this design include (D) or the same carryover effects treatment at an even from prior higher intensity (B3). treatments.

(table continues) 141 This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. 4 AGWNADWONG AND MACGOWAN 142

Table 1 (continued) Considerations for Studies Using this Design Brief Description Advantages Disadvantages Group Work Design Experimental & Controlled Designs ABA, BAB, ABAB, Remove an apparently Replication of treatment Ethical concerns about Certain group processes (Hauserman, et al., ABCBC, effective treatment or effects provides a strong intentionally (e.g., cohesion, role 1972 [ABAB]; AB1B2B1B2,& return to the condition demonstration of reversing gains differentiation) are Helmus, et al., 2003 other reversal that preceded the experimental control associated with a unlikely to be [ABA]; Kirby, et al., designs apparently effective over the dependent treatment. Not reversible making the 2008; Linsk, et al., Marr ;ء[treatment to reverse variable (good internal applicable with some design unworkable. 1975 [ABAB observed improvements, validity). Design can be interventions (e.g., & Fairchild, 1993 & then reintroduce the conducted with only one psychoeducational) [ABAA]; Martinez treatment to reproduce participant or one because their effects & Wong, 2009 & Mouton ;ء[and replicate its effects. group. are not reversible. [ABAB Stanley, 1996 [ABA]; Petry, et al., 2001 [AB1B2CA]; Vaccaro, 1990 [ABAB]; Weber, (ء[ABAB] 1980 Concurrent Simultaneously collect data Design does not require a Design requires two or MBD across (Bates, 1980; Clark, multiple-baseline on two or more reversal & therefore is more independent participants within Cunningham, & design equivalent baselines. not complicated by baselines (i.e., the same group is Cunningham, 1989; After obtaining stable possible ethical similar behaviors, vulnerable to carry- Foxx, McMorrow, pre-intervention data, concerns & settings, participants, over across members Bittle, & Ness, 1986; apply the intervention to irreversibility. As the or groups). due to modeling Foxx, et al., 1983; the first baseline. After treatment is replicated Intervention is effects & changes in Gronna, Serna, observing clear gains on across different delayed for second group interaction Kennedy, & Prater, the first baseline, apply behaviors, members, or & subsequent patterns. 1999; Hall, et al., ,Hall ;ءthe intervention to the groups, evidence for the baselines. Generalization across 2000 second baseline. Repeat generalizability (external participants or groups Schlesinger, & this process with third & validity) of the cannot be Dineen, 1997; subsequent baselines. intervention is obtained. distinguished from Hansen, et al., .Hansen, St ;ءloss of experimental 1985 control (i.e., Lawrence, & extraneous effects). Christoff, 1989; This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Table 1 (continued) Considerations for Studies Using this Design Brief Description Advantages Disadvantages Group Work Design Kelly, Laughlin, Claiborne, & Patterson, 1979; Kelly, Wildman, Urey, & Thurman, 1980; Lumpkin, Silverman, Weems, WORK GROUP DESIGNS SINGLE-CASE Markham, & Kurtines, 2002 [partial nonconcurrent MBD]; Phaneuf & McIntyre, 2007; Roessler & Lewis, 1984; Serna, Schumaker, Sherman, & Sheldon, 1991; Tracey, et al., 1974; Tymchuk, Andron, & Rahbar, 1988) ;ءNoncurrent The NCMBD resembles a Design does not require Requires multiple Because they are (Connors, et al., 1984 multiple-baseline succession of AB running multiple participants or started at different Edleson, et al., Kettlewell, et ;ءdesign designs. In its strictest baselines groups. points in time, there 1985 & Lees ;ءform the length of the simultaneously. The is less likelihood of al., 1992 different baselines vary NCMBD replicates carryover effects Ronan, 2008) & are randomly treatment effects across across participants & assigned to a series of participants or groups & groups. clients or groups. therefore demonstrates generality with each successive baseline. (table continues) 143 This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. 4 AGWNADWONG AND MACGOWAN 144

Table 1 (continued) Considerations for Studies Using this Design Brief Description Advantages Disadvantages Group Work Design ;ءAlternating Baseline sessions & Efficient design that Difficult to use with Can be used to test (Blake, et al., 1990 treatments design treatment sessions are allows the comparison complex therapy effects of Gajar, et al., 1984; rapidly alternated of multiple treatments protocols that interventions on Kastner & May, ,.Wong, et al ;ءaccording to a random within a short period of involve a cumulative outcomes & for 2009 (ءschedule. time. Can show building process. comparing the effects 1988 treatment effects even Given rapidly of interventions on when baseline data is alternating within-group trending in the desired treatments, there is conditions (e.g., direction. Brief the risk of engagement). presentation of interaction effects treatments may with typical group minimize irreversibility development events. & multiple treatment interference. Mixed or combined Combining SCDs or Can permit more Mixed & combined Complex designs could (Kohler & Fowler, ,.Wong, et al ;ءdesigns embedding one SCD sophisticated inquiry & designs require be structured to 1985 (ءwithin another can refinement of higher proficiency in produce sequential 1996 increase design interventions (e.g., after the use of SCD changes in group manipulations to provide demonstrating an methodology. processes & group more demonstrations of effective therapy outcomes to test the experimental control & assessing strategies to relationships between answer complex promote generalization). those variables. questions. .Article is further reviewed in the text of this paper ء SINGLE-CASE DESIGNS GROUP WORK 145

1979), and natural healing (Wong, 2010) pre- significant gains in any of their dependent mea- vent one from concluding that treatment caused sures. Cetingok and Hirayama explained their those improvements. unexpected findings by noting that their elderly Group process measures can also be plotted subjects probably entered the program well in- during a B design, revealing fluctuations in vari- formed about their health care needs and appar- ables such as verbal interaction patterns and ently became suspicious that the mental health group cohesion, which might mediate outcomes assessment data might be used to expel them for group members. A few group researchers from their rent-subsidized, high-rise apart- have monitored group process during treatment ments. in this manner (Fontao & Mergenthaler, 2008; O’Brien, Korchynsky, Fabrizio, McGrath, & AB Design (Baseline; Treatment) Swank, 1999; Patterson & Basham, 2002). For example, O’Brien and colleagues (1999) mea- The AB design is a substantial improvement sured positive process, satisfaction, and nega- over the B design and unlike the latter could be tive process during stress management groups, considered an actual “design” in that it provides and reported the findings to the group. In addi- a minimal degree of experimental control. Con- tion, the intervention included increasing posi- trol comes from the baseline data (during which tive interactions (e.g., active helping) during the no treatment is administered) to which treat- latter sessions of the group program. The re- ment data is compared. If the baseline data are searchers found that ratings of positive process stable or moving in a countertherapeutic direc- and satisfaction increased significantly across tion and this pattern changes shortly after the sessions. These investigators further reported introduction of the intervention, these results that satisfaction ratings were inversely corre- can give evidence that the intervention might be lated to cardiovascular reactivity to stress, while effective. Unlike the B, the AB design elimi- negative process ratings were positively corre- nates maturation (i.e., processes associated with lated to this physiological measure. The authors the mere passage of time) and testing (i.e., mea- speculated that the planned emphasis on posi- surement reactivity) as alternative explanations tive interaction may have improved ratings over of observed change (Campbell & Stanley, time. More recently, Fontao and Mergenthaler 1966). Unfortunately, the AB design is still (2008) examined the verbal content of female susceptible to “history” or extraneous events patients at four measurement points in a psy- that happen to coincide with treatment (Camp- chodynamic group for eating disorders. As pre- bell & Stanley, 1966; Hersen & Barlow, 1976), dicted by the researchers, group members’ lan- such as the arrival or departure of group guage patterns (“emotion-abstraction patterns”) members. Eight studies used this design since significantly related to particular therapeutic first appearing in the literature in the 1990s, factors, which also suggested the presence of a with most published since the year 2000 (see positive therapeutic cycle. The researchers Table 1). noted that such language patterns may be used Rubin (1991) aggregated a set of AB designs to identify the occurrence of therapeutically im- to evaluate outreach counseling and support portant points in the process of group work. groups for battered women. A formidable chal- Although process measures can serve as lenge to his evaluation effort was that interven- proximal indicators of change and are important tions used at the agency were “somewhat unsys- This document is copyrighted by the American Psychological Association or one of its allied publishers. for assessing whether desired therapeutic pro- tematic,” theoretically eclectic, “idiosyncratic,” This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. cesses unfold as planned, the ultimate test of and “. . . varied from session to session and across group therapy is showing improvement in client practitioners (group leaders)” (p. 340). With two outcomes. Cetingok and Hirayama (1983) used sets of measures collected on each woman, one a B design consisting of four assessments con- scoring desired thoughts, feelings, and behav- ducted at the first, third, fifth, and seventh iors and the other scoring abusive behavior by weeks of their program to assess whether par- the spouse, neither measure showed positive ticipation in groups was associated with gains in change that could be attributed to agency inter- health care knowledge and mental health status ventions. for elderly persons with diabetes, heart disease, Johnson et al. (2001) obtained more favor- and hypertension. The investigators found no able results when examining whether leader- 146 MACGOWAN AND WONG

initiated changes influenced group processes in magnitude (B3)inanAB1B2B3 design. This three AB design studies implemented by under- highlights one of the advantages of SCDs; graduate students and practitioners. The three namely, their flexibility and ability to incorpo- studies included (a) student leaders of an ado- rate an evidence-based, trial-and-error process lescent girls’ group encouraging the develop- within clinical practice. This flexibility is cru- ment of trusting relationships between members cial for accommodating the variable responses and the effect on subsequent group attendance, of individuals and groups to particular interven- (b) interventions to increase participation by the tions. “silent majority” in a group of veterans and Although the ABC, ABCD, AB1B2, the number of members later speaking up in the AB1B2B3, and related designs permit the mod- group, and (c) encouragement of self-expres- ification of interventions until the client or sion through puppetry in children who had ex- group has improved to a satisfactory level, these perienced loss and emotional trauma and the designs are unable to determine which of the number of children then using this mode of interventions, if any, actually produced the im- expression. Changes in dependent measures provement. This is because these designs are from the A to B phases were statistically sig- merely complicated versions of AB designs and nificant in the first two studies. In the third are also susceptible to extraneous events that study, statistical tests were not used, but there might have coincided with any of the seemingly was 200% improvement from baseline to inter- effective treatment phases. In addition, se- vention mean scores. quence and carryover effects are possible in Ledgerwood, Alessi, Hanson, Godley, and ABC and ABCD designs in that positive Petry (2008) used counterbalanced AB designs changes in C or D phases actually might be due (A1A2B; A1BA2) across four clinics to gauge to delayed effects of previous treatment phases. the effects of contingency management on Interpretation of group work data is even trick- attendance at substance abuse therapy groups. ier, because the conduct and affect of group Three randomly selected clinics were exposed members and the nature of the group experience to an A1A2B design composed of the following is expected to change as a group moves through conditions: An 8-week “baseline” of standard developmental stages. treatment (A1), 16 additional weeks of standard Two studies have recently utilized these de- treatment (A2), and 16 weeks of standard treat- signs. Fry, Botterill, and Pring (2009) evaluated ment plus contingency management (material two formats of group-based speech-language incentives for attendance) (B). A fourth ran- therapy on the stuttering of an 18-year-old male domly selected clinic was exposed to an A1BA2 client (in a group of up to 10 members) using an design, which resembled the previous design ABC design with a maintenance phase. After a except that it presented the last two phases in 5-week baseline (A), 2 weeks of intensive group reverse order. Clients who were enrolled in the instruction was provided during one treatment clinics when contingency management was im- phase (B), followed by 5 weeks of consolida- plemented attended a significantly greater per- tion/self-management training in a second treat- centage of therapy sessions than clients who ment phase (C). Finally, follow-up measures were enrolled in treatment without contingency were taken during a 10-month maintenance management. phase. Compared to the A phase, there was a significant reduction in the percent of stuttered This document is copyrighted by the American Psychological Association or one of its allied publishers. ABC, ABCD, AB1B2,AB1B2B3, and Related syllables per session and in the duration of This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. Designs stuttering in the B and C phases, and this im- provement continued or increased in the main- If the outcome in the B phase of an AB tenance phase. Although introduction of the design is unsatisfactory, a second intervention interventions was associated with decreased (C) can be tested in an ABC design, or the same stuttering, as intended, it was not known if the intervention at a higher magnitude (B2) can be interventions were responsible for this improve- tested in an AB1B2 design. If the outcome of C ment. or B2 proves unsatisfactory, a third treatment A group intervention has also been added to (D) can be tested in an ABCD design, or the other therapeutic modalities in a stepwise ABC same treatment can be tested at an even higher treatment design. Phaneuf and McIntyre (2011) SINGLE-CASE DESIGNS GROUP WORK 147

used such a design in a three-tier parent training In a landmark study, both for its use of pre- program involving parents of preschool chil- cise observational measures of group interac- dren with developmental disabilities (n ϭ 8 tion and SCD evaluation methodology, Linsk, parentÐchild dyads). Each tier included an in- Howe, and Pinkston (1975) included several tervention component followed by an evalua- ABAB designs to gauge the effect of question- tion to determine who either responded favor- asking on appropriate verbalizations by elderly ably to the intervention, or who would need residents in activity groups in a nursing home. further service in the next intervention tier. The Following a baseline phase in which group ac- first phase included 3 weeks of individual par- tivities were conducted in the agency’s usual ticipants listening to or reading materials drawn manner, a second phase gauged the effects of from Webster-Stratton’s Incredible Years pro- the group leader asking residents as many ques- gram, followed by an evaluation to determine tions as possible during the group activity. In a either who responded and would be scheduled third phase, the group leader returned to the for a 3-month follow up assessment, or who customary manner of directing the group. In the would move on to the second intervention tier. fourth and final treatment phase, the group The second tier involved 11 weeks of Webster- leader resumed the procedure of increased ques- Stratton’s group-based Incredible Years Parent tion-asking. Linsk and colleagues showed that Training Program, followed by another evalua- more frequent questions doubled the number of tion. Those who responded were scheduled appropriate verbalizations relative to baseline for the 3-month follow up, and those who sessions using the agency’s usual procedure. needed the final tier received a single session of Weber (1980) investigated the impact of vid- individualized video feedback training and were eotape feedback on adolescents in inpatient scheduled for the last 3-month follow up. All group therapy, hypothesizing that it would in- but one child displayed reductions in observed crease verbal expressions of warmth and de- crease expressions of hostility and flight. Dur- inappropriate behavior or parental ratings of ing experimental sessions, the first 45 minutes child behavior problems. of each group session was videotaped and then 1 2 1 2 immediately played back to the members. The ABA, BAB, ABAB, ABCBCB, AB B B B , effects of videotaping and playback were eval- and Other Reversal Designs uated within an ABAB design with each phase lasting 2 weeks. There were inconsistent results Inherent weaknesses of AB, ABC, and across the three measures of verbal affect. There ABCD designs can be rectified by removing the was some support for an increase in warmth seemingly effective treatment and returning to expressed during videotaping. However, trends the previous experimental condition to reverse within phases were unstable with warmth in- observed improvements, and then reintroducing creasing in the initial baseline phase without the treatment to replicate its positive effects. videotape feedback and decreasing rapidly at Replication of treatment effects by reapplying the end of the final videotape feedback phase. an intervention and reproducing its results is a For flight there was some reduction in the num- powerful demonstration of causality (Sidman, ber of responses during videotape feedback; 1960). The drawbacks of reversal designs in- but, here too, flight responses were declining clude ethical concerns about removing an ap- steadily at the end of both baseline phases with- This document is copyrighted by the American Psychological Association or one of its allied publishers. parently effective treatment merely for the pur- out videotape feedback. Finally, hostility This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. pose of demonstrating its efficacy, irreversible showed no consistent reduction during the pres- effects of certain interventions (such as psy- ence of videotape feedback. Different statistical choeducational therapies conveying knowledge tests for change in the SCD data yielded con- that cannot be removed), and changes in certain flicting results, showing how these analyses can group-based dependent variables (e.g., cohe- be contradictory. Contributing to the inconsis- sion) attributable to group maturation that also tent findings was high variability within the might be irreversible. Ten reversal designs have phases and a predetermined length for all phases been used in group work since the early 1970s, (2 weeks). Sometimes experimental conditions representing the second most utilized SCD (see were altered before data within each phase had Table 1). stabilized. Had the investigator waited until data 148 MACGOWAN AND WONG

within phases had stabilized before changing lems by simultaneously collecting data on two conditions, he might have been able to more or more independent baselines—consisting ei- clearly identify the levels and trends within ther of different behaviors/targets, the same be- phases and then been able to detect changes in haviors/targets in different settings, or the same subsequent phases. behaviors/targets in different group members Martinez and Wong (2009) utilized an (or groups)—and then sequentially introducing ABAB design to evaluate the effects of written the intervention across these baselines. If im- and telephone prompts on attendance at group provements occur when, and only when, an support meetings for women survivors of do- intervention is successively applied to two or mestic abuse. Participants of this study were more of these baselines, then the intervention’s residents of a transitional housing program who effectiveness is demonstrated. Because concur- upon admission to the program were notified of rent MBDs replicate treatment effects across the group support meetings along with other behaviors or across group members, they can services offered by the agency. After 2 weeks of demonstrate some treatment impact across baseline in which no additional steps were taken those two dimensions. Drawbacks to using con- to increase attendance at the support meetings, a current MBDs can be the lack of two or more colorful written invitation to attend the meeting independent but equivalent baselines (which was delivered and a single phone call was made must respond similarly to the same interven- to each resident the morning of the meeting for tion) and the requirement to delay the interven- 2 weeks. Next, the baseline condition of no tion for the second and subsequent baselines. prompts was reinstituted for 2 weeks, which The concurrent MBD was used in 17 group was followed by a final 3-week phase in which work studies. written and telephone prompts were reintro- Hansen, St. Lawrence, and Christoff (1985) duced. During baseline phases attendance was delivered a problem-solving skills intervention limited to the same three residents. During the for challenging situations in community living first prompting phase the number of residents in to seven patients in a partial hospitalization attendance doubled. However, when prompts program. Hansen and colleagues utilized in- were discontinued attendance fell back to the structions, modeling, behavior rehearsal, and level observed in baseline. In the final prompt- verbal reinforcement during group sessions to ing phase the number of residents again in- successively teach skill components of problem creased, this time to more than triple what was identification, goal definition, solution evalua- recorded in baseline. tion, evaluation of alternatives, and selection of the best alternative. Group data plotted within a Multiple-Baseline Designs (MBDs) concurrent MBD across skill components showed that the five skills were rarely exhibited MBDs have been the most frequently used during baseline phases, but these same skills SCD in group work (k ϭ 21), probably because showed immediate and significant increases fol- they lack troublesome reversal phases and be- lowing training. cause group work involves multiple clients Hall, Dineen, Schlesinger, and Stanton whose progress sometimes can be monitored (2000) used a MBD to evaluate the effective- within separate baselines. Also contributing to ness of group training in verbal interactive skills the common use of this design is the prevalence for developmentally disabled adults in a voca- This document is copyrighted by the American Psychological Association or one of its allied publishers. of social skills training in group work where tional training program. Six complex verbal This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. MBDs have been utilized to assess the effects of skills (e.g., “social conversation”) were succes- sequential training of specific social skills. This sively taught with instructions, modeling, re- section will review several illustrative group hearsal, and feedback over a 7-week period, and work studies using concurrent and nonconcur- the effects of training were evaluated with a rent MBDs. concurrent MBD across responses. There were Concurrent MBD. As noted above, ethical mixed or null effects of the training procedure concerns and irreversibility of certain types of on clients’ rated skills. Hall and colleagues change are limitations of ABAB and other re- noted that this poor outcome might have been versal designs. Concurrent MBDs do not con- due to their training schedule that required lead- tain reversal phases and circumvent these prob- ers to begin training on additional skills before SINGLE-CASE DESIGNS GROUP WORK 149

previously trained skills had been mastered, and participating in the program, which they pointed the low proficiency of group trainers who were to as evidence of the program’s effectiveness. A mostly first-year graduate students. couple of features of this study, however, limit Nonconcurrent MBD (NCMBD). its ability to demonstrate treatment efficacy: (a) NCMBDs are a variant of the MBD in which the baseline data for all three groups were re- the baselines do not all start at the same time or constructed retrospectively, which greatly in- run simultaneously. If implemented strictly ac- creased the possibility of error; and (b) group cording to the first published description of this treatment was preceded by 2 weeks of intake design (Watson & Workman, 1981), the pre- interviews, during which all groups showed im- treatment baseline phases are of different lengths mediate “decreases” in spouse abuse from base- and the assignment of these different baseline line levels. The decline during the interview lengths to individual clients is randomized. In the period, before any treatment was administered, published literature, however, NCMBDs contain- suggests that decreases in abuse may have been ing different baseline lengths and random as- due to the intake interviews or participants’ signment of these baselines has been inconsis- admission to the program rather than the treat- tent. Because the separate baselines of a NCMBD ment itself. are initiated at different times, and can even be Kettlewell, Mizes, and Wasylyshyn (1992) started after the completion of other baselines, a met all of the strict requirements for a NCMBD NCMBD is essentially a series of AB designs in a study evaluating group cognitiveÐbehav- and is relatively easy to carry out. Furthermore, ioral treatment for bulimia. Kettlewell and his because replication of intervention effects in the associates used a comprehensive treatment that separate AB designs of the NCMBD are con- included information on medical and psycho- ducted with different participants, this design logical consequences of bulimia, goal setting, demonstrates some generality of intervention functional analysis of the binge-purge cycle, effects across subjects. There were four exam- selecting an appropriate weight for oneself, cop- ples of this design in the literature (see Table 1). ing strategies, combating irrational cognitions Connors, Johnson, and Stuckey (1984) im- about thinness, and ways to maintain progress. plemented an NCMBD in a study of brief psy- This treatment was applied after 2, 6, or 10 choeducational group therapy for the treatment weeks of stable baselines with three groups of of bulimia. Treatment consisted of education, clients, the length of baseline for each group self-monitoring, goal setting, assertion training, being randomly determined. The weekly rate of relaxation, and cognitive restructuring, which binging and purging markedly declined for all was applied with two groups of 10 normal- three groups after the onset of treatment and, weight bulimic women, one group 3 weeks be- with the exception of one client, lower rates of fore the other. The investigators reported a 70% eating disorders were mostly maintained during reduction in binge/purge episodes and signifi- follow-up assessments. cant improvements in self-esteem, depression, assertiveness, and attitudes about eating that Alternating Treatments Design (ATD) coincided with the introduction of treatment for each group. Group workers may want to know which Edleson, Miller, Stone, and Chapman (1985) intervention is most efficacious for their mem- used a “modified” MBD that resembled a NCMBD bers, and the ATD (Barlow & Hayes, 1979)isa This document is copyrighted by the American Psychological Association or one of its allied publishers. to evaluate a 12-week, small-group treatment design specifically for the purpose of comparing This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. for men who battered their partners. Groups two or more interventions. The ATD makes were guided through topics of analyzing chains comparisons between interventions by rapidly of events leading to violent situations, progres- alternating (e.g., from one group session to the sive relaxation, cognitive restructuring, problem next) between two or more treatment conditions solving, taking the partner’s view of the prob- and baseline conditions on a random schedule. lem, and negotiation skills. Three groups began Sometimes a sign or cue is used to identify the treatment at three different points in time. Edle- session (e.g., “Positive Interactions Session”) so son and colleagues reported substantial declines that participants can distinguish which condi- in the mean incidence of self-reported spouse tion is currently in effect. The resulting data are abuse in two of the three groups as they began graphed and data points from the same condi- 150 MACGOWAN AND WONG

tion are connected by lines to form curves rep- age intervention, verbal and written announce- resenting group members’ performance during ments of daily activities plus individual feed- the different conditions. The distance separating back and praise, on patients’ attendance at the curves represents the differential effects group activities on a postacute psychiatric tran- of the various conditions. The ATD is an effi- sition unit. On a weekly basis for the 20 weeks cient design that can reveal differences between of the study, these procedures to encourage at- conditions after a small number of sessions. tendance alternated randomly with the unit’s Potential problems of the ATD are contrast ef- customary procedure of having nursing staff fects and multiple-treatment interference, which verbally remind patients that they were ex- are experimental artifacts brought on by close pected to attend all groups as part of their treat- juxtaposition of intervention conditions or ment. Attendance steadily increased in both cross-contamination and interaction of the dif- conditions during the length of the study, but ferent conditions, respectively. Given that attendance was significantly higher in the groups develop and mature over time, there is a prompting and feedback condition. good likelihood of interdependency between Kastner and May (2009) utilized an ATD to phases, producing such interaction effects. gauge the effects of action-oriented techniques These threats can be reduced by ending the (AO) on group climate (using the Group Cli- ATD with a final phase in which the interven- mate QuestionnaireÐShort Version, MacKenzie, tion with the best outcome is presented alone 1983) and disruptive out-of-group behavior on a consecutively for several sessions. If perfor- group of seven middle-school students referred mance within this final phase is similar to that for behavioral or emotional difficulties. Group recorded in the earlier alternating treatments sessions consisted of discussion on topics se- phase of the study, then the results were not due lected by group members and alternated ran- to contrast effects or multiple-treatment inter- domly between sessions with AO (i.e., the ference. double, the empty chair, role reversal, and psy- As seen in Table 1, the first publications that chodramatic and behavioral role play) and ses- included the ATD in group work appeared in sions without AO. There were no significant the 1980s (k ϭ 2), with only two more over the differences between the two conditions in the next two decades. Wong and colleagues (1988) frequency of disruptive behaviors, ratings of used an ATD to assess the effects of two types group engagement, or ratings of avoidance. of structured group activities (e.g., team sports, However, ratings of engagement increased and art projects) on bizarre psychotic and appropri- ratings of avoidance decreased significantly ate behaviors in 10 male chronic psychiatric over time, whereas ratings of conflict were un- patients on a hospital ward. On a random sched- changed. ule, the group of patients participated in one of three different types of sessions: (a) structured Mixed and Combined Designs group activities with positive reinforcement (to- kens earned through active participation), (b) Although articles and textbooks on SCDs structured group activities with noncontingent present these formats as relatively fixed struc- reinforcement (tokens given regardless of the tures, the essence of all of these designs is the level of participation), or (c) baseline sessions systematic manipulation of the independent (control sessions in which no structured activity variable to produce a corresponding pattern of This document is copyrighted by the American Psychological Association or one of its allied publishers. was offered). Psychotic behavior was signifi- changes in the dependent variable that is un- This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. cantly lower (70%) and appropriate behavior likely to have been due to random fluctuation or was substantially higher (approximately 500%) extraneous events. Researchers who appreciate in the recreational sessions as opposed to the the underlying logic of SCDs can fully exploit baseline sessions. Data for the two recreational their flexibility by mixing or combining the conditions were roughly equivalent indicating manipulations characteristic of the different de- that the beneficial effect was due to the recre- signs (e.g., reversals, successive application of ational activities and not contingent reinforce- treatment across behaviors) to fit the needs of ment. the particular study or to investigate additional Blake, Owens, and Keane (1990) used an questions of interest (Kazdin, 2011; Kennedy, ATD to evaluate the efficacy of a single pack- 2005; Smith, 2012; Wong & Liberman, 1981). SINGLE-CASE DESIGNS GROUP WORK 151

The major drawback to mixed and combined tion, and positive (monetary) reinforcement. designs is that they are more intricate and re- Covert generalization assessments conducted quire more sophisticated understanding of SCD on the hospital ward revealed that despite sub- methodology. stantial increases in trained responses in the Kohler and Fowler (1985) used a combina- game setting, these skills were rarely exhibited tion of concurrent MBDs and ABA designs in a on the ward where patients resided. To promote study teaching three young children to invite generalization of the skills beyond the game peers to play and to use social amenities (e.g., situation, minimal in-situ training involving in- “thank you,” “please”). The study evaluated termittent verbal prompts and tangible rein- effects of group training and the termination of forcement (C) were administered when the training (during the reversal phase) on the chil- skills were used during social interactions on dren’s behavior and their peers’ responses out- the ward. Minimal in-situ training was success- side of group in a classroom. For two of the fully applied with each of the three adolescents three children, the MBD showed that training immediately producing substantial improve- increased invitations and amenities performed ment in skill usage on the ward in each patient, by the children and those responses were recip- thereby replicating treatment effects in the pat- rocated by their peers. Termination of training tern of a concurrent MBD across subjects. Thus, resulted in a reduction of invitations and ame- this study utilized an ABC design to show that nities, but not to the near zero levels observed in game training (B) failed to produce changes that the initial baseline. For the third child (who was generalized to the more naturalistic ward set- only taught invitations), group training in- ting, but that minimal in-situ training (C) did creased the rate of invitations in the target child engender those changes, which was proven by but these responses were not reciprocated by her replicating those improvements across subjects peers, and termination of training resulted in a in the concurrent MBD. return to the low levels of invitations observed in baseline. For this child, additional training of Discussion and Recommendations the target child and her peers along with group rewards was implemented to raise the number Although group work researchers have not of invitations. This condition increased both the used SCDs frequently, they have used a range number of invitations given by the child and of designs to evaluate diverse interventions made by her peers, and when these procedures aimed at changing a broad spectrum of depen- were terminated, the level of invitations re- dent variables. The number of published reports mained above that recorded in previous baseline using SCDs in group work found in this review phases. The complicated questions about lasting (k ϭ 51) is substantially larger than the two effects of training and the testing of procedures reported in Tolman and Molidor’s paper (1994). to increase invitations for the third child could After a decrease in publications in the 1990s, not have been answered by using a MBD or an there has been growth in the past decade. Per- ABA design alone. Additional experimental haps this is an encouraging sign. manipulations were needed for these purposes. SCDs have been used to investigate both Wong, Morgan, Crowley and Baker (1996) processes and outcomes of group therapy using used a mixed ABC design and a concurrent both individual and group levels of measure- MBD across subjects to evaluate minimal in- ment. Most studies focused on group outcomes, This document is copyrighted by the American Psychological Association or one of its allied publishers. situ training to promote generalization of social across a broad range of target areas, including This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. skills taught to three adolescent psychiatric in- social skills, eating disorders, anxiety, self- patients. After collecting baseline data (A), the expression, warmth and hostility, and self- “Stacking the Deck” table game (Foxx, McMor- efficacy. The variables focusing on the group row, & Schloss, 1983) (B) was used to teach itself (e.g., processes) included attendance and reinforce appropriate interactive skills to (Blake et al., 1990; Helmus, Saules, Schoener, the three adolescents. The game involved re- & Roll, 2003; Johnson et al., 2001; Kirby, Ker- sponding to social scenarios described on game win, Carpenedo, Rosenwasser, & Gardner, playing cards, modeling of correct responses, 2008; Martinez & Wong, 2009; Petry et al., giving praise and the opportunity to move one’s 2001), verbal participation (Gajar, Schloss, game piece after correct responses to the situa- Schloss, & Thompson, 1984; Hauserman, Zwe- 152 MACGOWAN AND WONG

back, & Plotkin, 1972; Johnson et al., 2001; reconstructed baseline and interpreting the result- Kohler & Fowler, 1985; Linsk et al., 1975; ing data in a questionable manner. The doubtful Tracey, Briddell, & Wilson, 1974), group inter- accuracy of reconstructed baseline data was a action (Toseland, et al., 1978), language pat- weakness in itself, but that combined with desired terns (Fontao & Mergenthaler, 2008), group changes occurring prior to the introduction of ther- climate (Kastner & May, 2009), and satisfaction apy should have discouraged the authors from with group (O’Brien, et al., 1999; Patterson & drawing conclusions about treatment efficacy. Basham, 2002). Most of the other studies cited here, however, Most studies used the individual as the level applied SCDs in a technically sound manner and of measurement and then derived a group mean their results were more definitive. score for analytical purposes. This method of analysis, though convenient and appropriate in Recommendations for Research many cases, obscures individual and subgroup differences. Patterson and Basham (2002) ex- Use of SCDs in group work appears to be at amined satisfaction with group at both the indi- an intermediate level of development where a vidual and group levels using an innovative number of designs have been used and some mapping procedure, showing the importance of fairly often, but there are opportunities for more examining both simultaneously. Rather than us- varied and refined applications. Group work ing individual scores or the mean score for the researchers have not always distinguished be- group as the unit of analysis, this approach tween uncontrolled case study SCDs (e.g., B, showed variability in both individual and group AB, ABC) and controlled experimental SCDs outcomes. Researchers should be aware of the (e.g., ABAB, MBDs), only the latter of which benefits and limitations of individual and group replicate treatment effects and thereby permit levels of measurement and analysis, and select logical inference of treatment efficacy. There the most appropriate method. are efforts underway to raise the standards for The quality and rigor of group work studies SCD research. Guidelines have been developed using SCD methodology varied to a large de- for rating SCDs based on their level of evidence gree. SCDs are merely evaluation tools and they for demonstrating a causal relationship (Smith, cannot rise above other key components of the 2012). According to such guidelines (Romeiser clinical study. For example, in the case of Rubin Logan, Hickman, Harris, & Heriza, 2008; What (1991), SCDs were unable to produce consistent Works Clearinghouse, 2013), a MBD with three or interpretable results for interventions that the baselines could provide a convincing demon- author himself admitted were unsystematic and stration of treatment efficacy, but a MBD with vacillated from session to session. It is unclear less than three baselines could not. This is note- how such a study could yield reliable results, or worthy, as the proposed standard is more rigor- if such results emerged how they would be ous and stringent than that previously presented understood. This also raises the question about in published research and textbooks covering what treatments lend themselves to SCDs. As SCDs (Bailey & Burch, 2002; Cooper, Heron, indicated in Table 1, most interventions incor- & Heward, 2007). porated forms of behavior therapy, but others Future research could apply SCDs not used in utilized solution focused (Nelson & Kelley, previous group work studies. One example is 2001), experiential (Patterson & Basham, the Changing Criterion Design, which is appli- This document is copyrighted by the American Psychological Association or one of its allied publishers. 2002), couples group therapy (Nelson & Kelley, cable in situations where a dependent variable This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. 2001), and other well-defined techniques (John- can be tightly controlled (Hall & Fox, 1977; son et al., 2001). This shows that SCDs can Hartmann & Hall, 1976). This design requires accommodate many different types of therapy. one group member or group and one indepen- However, whatever theoretical treatment ap- dent variable that produces stepwise changes in proach is being evaluated, it must be applied the dependent variable. For example, it may be systematically and consistently to produce in- unreasonable to expect that a group member terpretable results. This is true of SCDs as well with limited social competence be fully en- as between-groups designs. gaged in the group within a few sessions as a The study by Edleson and colleagues (1985) result of an engagement-building intervention. presented a different set of concerns by using a Instead, incremental engagement objectives SINGLE-CASE DESIGNS GROUP WORK 153

may be set over a number of group sessions more rigorous designs with stronger evidence (e.g., 25% more engaged per set of sessions), for causality could have a suppressive effect on leading to full engagement as the ultimate goal. the usage of these designs in evaluating prac- Other novel designs and experimental manipu- tice. Standards that devalue weaker SCDs might lations that could be used to explore group work result in even fewer SCDs being applied and questions or to strengthen the internal validity less attempts to empirically evaluate group of a SCD study involve randomization. Points work. Individual evaluation, even if imperfect, in time that interventions are started or the order is preferable to automatically assuming that a in which group members or groups are exposed treatment validated in a prior between-groups to treatment are randomly determined, and such study will be equally effective with the group of randomization can be incorporated into almost clients one is currently working. It is also any SCD (Edgington, 1975; Kratochwill & clearly superior to practice relying on tradition, Levin, 2010). Randomization is likely to require popularity of techniques, personal belief, or SCDs with longer baseline phases, but this may other approaches not based on client data. be acceptable in some group work research (see, There has been much discussion about the for example, Toseland, et al., 1978). importance of developing and/or adapting exist- An important question that group workers ing treatments to maximize efficacy with par- could explore using SCDs is whether group ticular client populations (Benish, Quintana, & processes directly impact group outcomes. This Wampold, 2011; Castro, Barrera, & Steiker, research could be conducted with two-stage 2010; Chen, Kakkad, & Balzano, 2008). La SCDs where both processes and outcomes are Roche and Christopher (2008) noted that no simultaneously monitored, and data are ana- empirically supported treatment has met all of lyzed to see if changes in relevant processes the criteria set forth by the APA Task Force for reliably precede changes in desired outcomes. treatment efficacy (American Psychological As- This type of investigation would be difficult to sociation, 2006) for any minority group. In conduct using conventional between-groups re- adapting existing group treatments, components search designs, but would be manageable using are added that attend to culture, while retaining within-groups SCDs. There are many measures the “essential” ingredients of the established of group processes, such as group climate group treatment (Macgowan & Hanbidge, (MacKenzie, 1983, as used in Kastner & May, 2014). A MBD across different groups or set- 2009) and group engagement (Macgowan & tings could be used to test the relative effective- Newman, 2005), but the most frequently stud- ness of a group intervention that is progres- ied variable has been cohesion, which has been sively adapted to include important cultural reliably associated with outcomes in many components. If the SCD is done rigorously group studies (Burlingame, McClendon, & (Chambless et al., 1998; Kratochwill & Levin, Alonso, 2011; Burlingame et al., 2006; Strauss, 2010), these data can contribute to the develop- Burlingame, & Bormann, 2008). As Fike (1980) ment of an empirically supported group inter- argued, it is important to capture not only indi- vention appropriate for that population. vidual outcomes, but also factors that reflect the Uncontrolled case study designs such as the group experience on individuals and on the B, AB, and ABC designs should be used in group, such as satisfaction and the nature of clinical practice, but with full awareness of their interactions. limited inferential power. Such designs do not This document is copyrighted by the American Psychological Association or one of its allied publishers. provide the controls or replications necessary to This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. Recommendations for Practice allow clinicians to logically conclude that their interventions caused whatever changes might The goals, resources, and constraints for have occurred. Nevertheless, these designs are practice differ substantially from that for re- valuable because “even if the therapist and cli- search, and recommendations for SCDs for ents are not certain that the intervention was the practice should reflect those differences (Wong, sole cause of the changes, it still makes it pos- 2010). Recall that our review of group research sible to ascertain the degree to which goals are from 1972 to 2012 found an average of only being achieved” (Rose, 1977, p. 67). slightly more than one SCD study published per Group work practitioners need not be con- year. Guidelines for SCD research calling for strained by earlier group work evaluative meth- 154 MACGOWAN AND WONG

ods. They can apply more refined variants of the dential Task Force on Evidence-Based Practice. B design. Examples include the Periodic Treat- American Psychologist, 61, 271Ð285. doi:10.1037/ ments Design (Hayes, 1981; Hayes, Barlow, & 0003-066X.61.4.271 Nelson-Gray, 1999) and the Repeated Pretest- ASGW. (2000). Professional standards for the train- Posttest Design (Thyer & Curtis, 1983), which ing of group workers. Retrieved September 6, 2013, from http://www.asgw.org/training_ are controlled designs that can evaluate an in- standards.htm tervention during a single intervention phase ASGW. (2007). Association for Specialists in Group that is neither preceded nor followed by a base- Work: Best practice guidelines 2007 revisions. Re- line phase. These designs can demonstrate ex- trieved August 6, 2013, from http://www.asgw.org/ perimental control by replicating improvements PDF/Best_Practices.pdf occurring soon after periodic exposure to inter- Bailey, J. S., & Burch, M. R. (2002). Research meth- vention (e.g., a technique to improve group co- ods in applied behavior analysis. Thousand Oaks, hesion) while showing the absence of such im- CA: Sage. provements at other times during the phase. In Barlow, A., Burlingame, G. M., & Fuhriman, A. addition, the internal validity of other case study (2000). Therapeutic application of groups from designs, such as the AB or ABC, can be Pratt’s “thought control classes” to modern group psychotherapy. Group Dynamics: Theory, Re- strengthened with the earlier mentioned proce- search, and Practice, 4, 115Ð134. doi:10.1037/ dure of randomly selecting the starting point for 1089-2699.4.1.115 interventions (Edgington, 1975; Kratochwill & Barlow, D. H., & Hayes, S. C. (1979). Alternating treat- Levin, 2010; Toseland, et al., 1978). ments design: One strategy for comparing the effects of two treatments in a single subject. Journal of Conclusions Applied Behavior Analysis, 12, 199Ð210. doi:10 .1901/jaba.1979.12-199 Barlow, S. H., Fuhriman, A., & Burlingame, G. M. This paper has elucidated how SCDs have been (2004). The history of group counseling and psy- used in group work by reviewing the designs in chotherapy. In J. L. DeLucia-Waack, D. A. Ger- published studies, discussing their respective rity, C. R. Kalodner, & M. Riva (Eds.), Handbook strengths and limitations, and examining proce- of group counseling and psychotherapy (pp. 3Ð22). dural variations producing data that were rela- Thousand Oaks, CA: Sage. doi:10.4135/ tively clear or confounded. Most of the group 9781452229683.n1 work studies reviewed here successfully used Bates, P. (1980). The effectiveness of interpersonal SCDs to accomplish their purpose—to systemat- skills training on the social skill acquisition of ically evaluate the effects of their interventions on moderately and mildly retarded adults. Journal of processes and outcomes. Group researchers can Applied Behavior Analysis, 13, 237Ð248. doi: learn from the mistakes and build upon the 10.1901/jaba.1980.13-237 Benish, S. G., Quintana, S., & Wampold, B. E. strengths of these earlier studies to become more (2011). Culturally adapted psychotherapy and the facile with this research methodology. Although legitimacy of myth: A direct-comparison meta- SCD technology has been used in group work analysis. Journal of Counseling Psychology, 58, since the early 1970s, there are many important 279Ð289. doi:10.1037/a0023626 questions that still need to be investigated and Blake, D. D., Owens, M. D., & Keane, T. M. (1990). group researchers have only begun to tap the po- Increasing group attendance on a psychiatric unit: tential of these evaluative formats. An alternating treatments design comparison. Journal of Behavior Therapy and Experimental This document is copyrighted by the American Psychological Association or one of its allied publishers. Psychiatry, 21, 15Ð20. doi:10.1016/0005- This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. References 7916(90)90044-L Bloom, M., Fischer, J., & Orme, J. G. (2009). Eval- AASWG. (2006). Standards for social work practice uating practice: Guidelines for the accountable with groups (2nd ed.). Retrieved from http://www professional (6th ed.). Boston, MA: Pearson Allyn .aaswg.org/files/AASWG_Standards_for_Social_ & Bacon. Work_Practice_with_Groups.pdf Burlingame, G. M., Strauss, B., & Joyce, A. (2013). AGPA. (2007). Practice guidelines for group psy- Change mechanisms and effectiveness of small chotherapy. Retrieved September 5, 2013, from group treatments. In M. J. Lambert (Ed.), Bergin http://www.agpa.org/guidelines/index.html and Garfield’s handbook of psychotherapy and American Psychological Association. (2006). Evi- behavior change (6th ed., pp. 640Ð689). New dence-based practice in psychology: APA Presi- York: Wiley. SINGLE-CASE DESIGNS GROUP WORK 155

Burlingame, G. M., McClendon, D. T., & Alonso, J. Fike, D. F. (1980). Evaluating group intervention. (2011). Cohesion in group therapy. Psychother- Social Work with Groups, 3, 41Ð51. doi:10.1300/ apy, 48, 34Ð42. doi:10.1037/a0022063 J009v03n02_04 Burlingame, G. M., Strauss, B., Joyce, A., MacNair- Fontao, M. I., & Mergenthaler, E. (2008). Therapeutic Semands, R., MacKenzie, K. R., Ogrodniczuk, J., factors and language patterns in group therapy appli- & Taylor, S. (2006). CORE BatteryÐRevised: An cation of computer-assisted text analysis to the ex- assessment toolkit for promoting optimal group amination of microprocesses in group therapy: Pre- selection, process, and outcome. New York, NY: liminary findings. Psychotherapy Research, 18, 345Ð American Group Psychotherapy Association. 354. doi:10.1080/10503300701576352 Campbell, D. T., & Stanley, J. C. (1966). Experimen- Foster, L. H., Watson, T. S., & Young, J. S. (2002). tal and quasi-experimental designs for research. Single-subject research design for school counsel- Boston, MA: Houghton Mifflin Company. ors: Becoming an applied researcher. Professional Castro, F. G., Barrera, M., & Steiker, L. K. H. (2010). School Counseling, 6, 146Ð154. Issues and challenges in the design of culturally Foxx, R. M., McMorrow, M. J., Bittle, R. G., & Ness, adapted evidence-based interventions. Annual Re- J. (1986). An analysis of social skills generaliza- view of Clinical Psychology, 6, 213Ð239. doi: tion in two natural settings. Journal of Applied 10.1146/annurev-clinpsy-033109-132032 Behavior Analysis, 19, 299Ð305. doi:10.1901/jaba Cetingok, M., & Hirayama, H. (1983). Evaluating .1986.19-299 the effects of group work with the elderly: An Foxx, R. M., McMorrow, M. J., & Schloss, C. N. experiment using a single-subject design. Small (1983). Stacking the deck: Teaching social skills to Group Behavior, 14, 327Ð335. doi:10.1177/ retarded adults with a modified table game. Jour- 104649648301400304 nal of Applied Behavior Analysis, 16, 157Ð170. Chambless, D. L., Baker, M. J., Baucom, D. H., doi:10.1901/jaba.1983.16-157 Beutler, L. E., Calhoun, K. S., Crits-Christoph, P., Frank, J. D. (1975). Group psychotherapy research . . . Woody, S. R. (1998). Update on empirically 25 years later. International Journal of Group Psy- validated therapies, II. The Clinical Psychologist, chotherapy, 25, 159Ð162. 51, 3Ð16. Fry, J. P., Botterill, W. M., & Pring, T. R. (2009). Chen, E. C., Kakkad, D., & Balzano, J. (2008). The effect of an intensive group therapy pro- Multicultural competence and evidence-based gram for young adults who stutter: A single practice in group therapy. Journal of Clinical Psy- subject study. International Journal of Speech- chology, 64, 1261Ð1278. doi:10.1002/jclp.20533 Language Pathology, 11, 12Ð19. doi:10.1080/ Clark, M. L., Cunningham, L. J., & Cunningham, 17549500802600990 C. E. (1989). Improving the social behavior of Gajar, A., Schloss, P. J., Schloss, C. N., & Thomp- siblings of autistic children using a group problem son, C. K. (1984). Effects of feedback and self- solving approach. Child & Family Behavior Ther- monitoring on head trauma youths’ conversation apy, 11, 19Ð33. doi:10.1300/J019v11n01_02 skills. Journal of Applied Behavior Analysis, 17, Connors, M. E., Johnson, C. L., & Stuckey, M. K. 353Ð358. doi:10.1901/jaba.1984.17-353 (1984). Treatment of bulimia with brief psychoe- Galassi, J. P., & Gersh, T. L. (1993). Myths, miscon- ducational group therapy. The American Journal ceptions, and missed opportunity: Single-case de- of Psychiatry, 141, 1512Ð1516. signs and counseling psychology. Journal of Cook, T. D., & Campbell, D. T. (1979). Quasi- Counseling Psychology, 40, 525Ð531. doi: experimentation: Design & analysis issues for field 10.1037/0022-0167.40.4.525 settings. Chicago, IL: Rand McNally College Pub. Garrett, K. J. (2005). School social workers’ evalu- Co. ation of group work practices. Children & Schools, Cooper, J. O., Heron, T. E., & Heward, W. L. (2007). 27, 247Ð252. doi:10.1093/cs/27.4.247 Applied behavior analysis (2nd ed.). Upper Saddle Gronna, S. S., Serna, L. A., Kennedy, C. H., & Prater, This document is copyrighted by the American Psychological Association or one of its allied publishers. River, N. J.: Pearson/Merrill-Prentice Hall. M. A. (1999). Promoting generalized social inter- This article is intended solely for the personal use of the individualEdgington, user and is not to be disseminated broadly. E. S. (1975). Randomization tests for actions using puppets and script training in an one-subject operant experiments. The Journal of integrated preschool. A single-case study using Psychology: Interdisciplinary and Applied, 90, multiple baseline design. Behavior Modification, 57Ð68. doi:10.1080/00223980.1975.9923926 23, 419Ð440. doi:10.1177/0145445599233005 Edleson, J. L., Miller, D. M., Stone, G. W., & Chap- Hall, J. A., Dineen, J. P., Schlesinger, D. J., & Stan- man, D. G. (1985). Group treatment for men who ton, R. (2000). Advanced group treatment for batter. Social Work Research & Abstracts, 21, developmentally disabled adults with social skill 18Ð21. doi:10.1093/swra/21.3.18 deficits. Research on Social Work Practice, 10, Feldman, R. A. (1987). Group work knowledge and 301Ð326. research: A two-decade comparison. Social Work Hall, J. A., Schlesinger, D. J., & Dineen, J. P. (1997). with Groups, 9, 7Ð14. doi:10.1300/J009v09n03_03 Social skills training in groups with developmen- 156 MACGOWAN AND WONG

tally disabled adults. Research on Social Work Prac- subject research to identify evidence-based prac- tice, 7, 187Ð201. tice in special education. Exceptional Children, 71, Hall, K. R. (2006). Using problem-based learning 165Ð179. with victims of bullying behavior. Professional Johnson, P., Beckerman, A., & Auerbach, C. (2001). School Counseling, 9, 231Ð237. Researching our own practice: Single system de- Hall, R. V., & Fox, R. G. (1977). Changing-criterion sign for groupwork. Groupwork, 13, 57Ð72. designs: An alternate applied behavior analysis Kastner, J. W., & May, W. (2009). Action-oriented procedure. In B. Etzel, J. Leblanc, & D. M. Baer techniques in adolescent group therapy. Group, 33, (Eds.), New developments in behavioral research: 315Ð327. Theory, method, and application (pp. 151Ð166). Kazdin, A. E. (2003). Research design in clinical New York, NY: Wiley. psychology (4th ed.). Boston, MA: Allyn & Bacon. Hansen, D. J., St. Lawrence, J. S., & Christoff, K. A. Kazdin, A. E. (2011). Single-case research designs: (1985). Effects of interpersonal problem-solving Methods for clinical and applied settings (2nd ed.). training with chronic aftercare patients on prob- New York, NY: Oxford University Press. lem-solving component skills and effectiveness of Kelly, J. A., Laughlin, C., Claiborne, M., & Patter- solutions. Journal of Consulting and Clinical Psy- son, J. (1979). A group procedure for teaching job chology, 53, 167Ð174. doi:10.1037/0022-006X.53 interviewing skills to formerly hospitalized psychi- .2.167 atric patients. Behavior Therapy, 10, 299Ð310. Hansen, D. J., St. Lawrence, J. S., & Christoff, K. A. doi:10.1016/S0005-7894(79)80020-9 (1989). Group conversational-skills training with in- Kelly, J. A., Wildman, B. G., Urey, J. R., & Thur- patient children and adolescents: Social validation, man, C. (1980). Group skills training to increase generalization, and maintenance. Behavior Modifica- the conversational repertoire of retarded adoles- tion, 13, 4Ð31. doi:10.1177/01454455890131001 cents. Child & Family Behavior Therapy, 1, 323Ð Hartmann, D. P., & Hall, R. V. (1976). The changing 336. doi:10.1300/J473v01n04_02 criterion design. Journal of Applied Behavior Kennedy, C. H. (2005). Single-case designs for edu- Analysis, 9, 527Ð532. doi:10.1901/jaba.1976.9- cational research. Boston, MA: Pearson/A & B. 527 Kettlewell, P. W., Mizes, J. S., & Wasylyshyn, N. A. Hauserman, N., Zweback, S., & Plotkin, A. (1972). (1992). A cognitive-behavioral group treatment of Use of concrete reinforcement to facilitate verbal bulimia. Behavior Therapy, 23, 657Ð670. doi: initiations in adolescent group therapy. Journal of 10.1016/S0005-7894(05)80227-8 Consulting and Clinical Psychology, 38, 90Ð96. Kirby, K. C., Kerwin, M. E., Carpenedo, C. M., doi:10.1037/h0032410 Rosenwasser, B. J., & Gardner, R. S. (2008). In- Hayes, S. C. (1981). Single case experimental design terdependent group contingency management for and empirical clinical practice. Journal of Consult- cocaine-dependent methadone maintenance pa- ing and Clinical Psychology, 49, 193Ð211. doi: tients. Journal of Applied Behavior Analysis, 41, 10.1037/0022-006X.49.2.193 579Ð595. doi:10.1901/jaba.2008.41-579 Hayes, S. C., Barlow, D. H., & Nelson-Gray, R. O. Kohler, F. W., & Fowler, S. A. (1985). Training (1999). The scientist practitioner: Research and prosocial behaviors to young children: An analysis accountability in the age of managed care (2nd of reciprocity with untrained peers. Journal of ed.). Boston, MA: Allyn & Bacon. Applied Behavior Analysis, 18, 187Ð200. doi: Helmus, T. C., Saules, K. K., Schoener, E. P., & Roll, 10.1901/jaba.1985.18-187 J. M. (2003). Reinforcement of counseling atten- Kratochwill, T. R., & Levin, J. R. (2010). Enhancing dance and alcohol abstinence in a community- the scientific credibility of single-case intervention based dual-diagnosis treatment program: A feasi- research: Randomization to the rescue. Psycholog- bility study. Psychology of Addictive Behaviors, ical Methods, 15, 124Ð144. doi:10.1037/a0017736 17, 249Ð251. doi:10.1037/0893-164X.17.3.249 La Roche, M., & Christopher, M. S. (2008). Culture This document is copyrighted by the American Psychological Association or one of its allied publishers. Heppner, P. P., Kivlighan, D. M., & Wampold, B. E. and empirically supported treatments: On the road This article is intended solely for the personal use of the individual user and is not(2008). to be disseminated broadly. Research design in counseling (3rd ed.). to a collision? Culture & Psychology, 14, 333Ð Belmont, CA: Thomson Brooks/Cole. 356. doi:10.1177/1354067X08092637 Hersen, M., & Barlow, D. H. (1976). Single case Ledgerwood, D. M., Alessi, S. M., Hanson, T., God- experimental designs: Strategies for studying be- ley, M. D., & Petry, N. M. (2008). Contingency havior change. New York, NY: Pergamon Press. management for attendance to group substance Horne, A. M., & Rosenthal, R. N. (1997). Research abuse treatment administered by clinicians in com- in group work: How did we get where we are? munity clinics. Journal of Applied Behavior Anal- Journal for Specialists in Group Work, 22, 228Ð ysis, 41, 517Ð526. doi:10.1901/jaba.2008.41-517 240. doi:10.1080/01933929708415527 Lees, D. G., & Ronan, K. R. (2008). Engagement and Horner, R. H., Carr, E. G., Halle, J., McGee, G., effectiveness of parent management training (in- Odom, S., & Wolery, M. (2005). The use of single- credible years) for solo high-risk mothers: A mul- SINGLE-CASE DESIGNS GROUP WORK 157

tiple baseline evaluation. Behaviour Change, 25, Mouton, S. G., & Stanley, M. A. (1996). Habit re- 109Ð128. doi:10.1375/bech.25.2.109 versal training for trichotillomania: A group ap- Linsk, N., Howe, M. W., & Pinkston, E. M. (1975). proach. Cognitive and Behavioral Practice, 3, Behavioral group work in a home for the aged. 159Ð182. doi:10.1016/S1077-7229(96)80036-8 Social Work, 20, 454Ð463. Nelson, T. S., & Kelley, L. (2001). Solution-focused Lumpkin, P. W., Silverman, W. K., Weems, C. F., couples group. Journal of Systemic Therapies, 20, Markham, M. R., & Kurtines, W. M. (2002). 47Ð66. doi:10.1521/jsyt.20.4.47.23085 Treating a heterogeneous set of anxiety disorders Newman, F. L., & Wong, S. E. (2004). Progress and in youths with group cognitive behavioral therapy: outcome assessment of individual patient data: A partially nonconcurrent multiple-baseline eval- Selecting single subject design and statistical pro- uation. Behavior Therapy, 33, 163Ð177. doi: cedures. In M. E. Maruish (Ed.), Use of psycho- 10.1016/S0005-7894(02)80011-9 logical testing for treatment planning & outcome Lundervold, D. A., & Belwood, M. F. (2000). The assessment (3rd ed., Vol. 1, pp. 273Ð289). Mah- best kept secret in counseling: Single-case (Nϭ1) wah, NJ: Erlbaum. experimental designs. Journal of Counseling & O’Brien, W. H., Korchynsky, R., Fabrizio, J., Development, 78, 92Ð102. doi:10.1002/j.1556- McGrath, J., & Swank, A. (1999). Evaluating 6676.2000.tb02565.x group process in a stress management intervention: Macgowan, M. J. (2008). A guide to evidence-based Relationships between perceived process and car- group work. New York, NY: Oxford University diovascular reactivity to stress. Research on Social Press. doi:10.1093/acprof:oso/9780195183450 Work Practice, 9, 608Ð630. doi:10.1177/ .001.0001 104973159900900506 Macgowan, M. J., & Hanbidge, A. S. (2014). Ad- Patterson, D. A., & Basham, R. E. (2002). A data vancing evidence-based group work in community visualization procedure for the evaluation of group settings: Methods, opportunities, and challenges. treatment outcomes across units of analysis. Small In J. L. DeLucia-Waack, C. R. Kalodner, & M. Group Research, 33, 209Ð232. doi:10.1177/ Riva (Eds.), The handbook of group counseling 104649640203300203 and psychotherapy (2nd ed., pp. 303Ð317). Thou- Petry, N. M., Martin, B., & Finocche, C. (2001). sand Oaks, CA: Sage. Contingency management in group treatment: A Macgowan, M. J., & Newman, F. L. (2005). The demonstration project in an HIV drop-in center. factor structure of the Group Engagement Mea- Journal of Substance Abuse Treatment, 21, 89Ð96. sure. Social Work Research, 29, 107Ð118. doi: doi:10.1016/S0740-5472(01)00184-2 10.1093/swr/29.2.107 Petry, N. M., Weinstock, J., & Alessi, S. M. (2011). MacKenzie, K. R. (1983). The clinical application of A randomized trial of contingency management a group climate measure. In R. R. Dies & K. R. delivered in the context of group counseling. Jour- MacKenzie (Eds.), Advances in group psychother- nal of Consulting and Clinical Psychology, 79, apy: Integrating research and practice (pp. 159Ð 686Ð696. doi:10.1037/a0024813 170). New York, NY: International Universities Phaneuf, L., & McIntyre, L. L. (2007). Effects of Press. individualized video feedback combined with Marr, D. D., & Fairchild, T. N. (1993). A problem- group parent training on maternal inappropriate solving strategy and self-esteem in recovering behavior. Journal of Applied Behavior Analysis, chemically dependent women. Alcoholism Treat- 40, 737Ð741. ment Quarterly, 10, 171Ð186. doi:10.1300/ Phaneuf, L., & McIntyre, L. L. (2011). The appli- J020V10N01_11 cation of a three-tier model of intervention to Martinez, K. K., & Wong, S. E. (2009). Using parent training. Journal of Positive Behavior prompts to increase attendance at groups for sur- Interventions, 13, 198Ð207. doi:10.1177/ vivors of domestic violence. Research on Social 1098300711405337 This document is copyrighted by the American Psychological Association or one of its allied publishers. Work Practice, 19, 460Ð463. doi:10.1177/ Robison, F. F., Morran, D. K., & Hulse-Killacky, D. This article is intended solely for the personal use of the individual user and is not1049731508329384 to be disseminated broadly. (1989). Single-subject research designs for group Maughan, D. R., Christiansen, E., Jenson, W. R., counselors studying their own groups. Journal for Olympia, D., & Clark, E. (2005). Behavioral par- Specialists in Group Work, 14, 93Ð97. doi: ent training as a treatment for externalizing behav- 10.1080/01933928908411892 iors and disruptive behavior disorders: A meta- Roessler, R. T., & Lewis, F. D. (1984). Conversation analysis. School Psychology Review, 34, 267Ð286. skill training with mentally retarded and learning McGrath, J. E. (1997). Small group research: That disabled sheltered workshop clients. Rehabilita- once and future field: An interpretation of the past tion Counseling Bulletin, 27, 161Ð171. with an eye to the future. Group Dynamics: The- Romeiser Logan, L., Hickman, R. R., Harris, S. R., & ory, Research, and Practice, 1, 7Ð27. doi:10.1037/ Heriza, C. B. (2008). Single-subject research de- 1089-2699.1.1.7 sign: Recommendations for levels of evidence and 158 MACGOWAN AND WONG

quality rating. Developmental Medicine & Child Vaccaro, F. J. (1990). Application of social skills Neurology, 50, 99Ð103. doi:10.1111/j.1469-8749 training in a group of institutionalized aggressive .2007.02005.x elderly subjects. Psychology and Aging, 5, 369Ð Rose, S. D. (1977). Group therapy: A behavioral 378. doi:10.1037/0882-7974.5.3.369 approach. Englewood Cliffs, NJ: Prentice Hall. Ventimiglia, J. A., Marschke, J., Carmichael, P., & Rubin, A. (1991). The effectiveness of outreach Loew, R. (2000). How do clinicians evaluate their counseling and support groups for battered wom- practice effectiveness? A survey of clinical social en: A preliminary evaluation. Research on Social workers. Smith College Studies in Social Work, 70, Work Practice, 1, 332Ð357. doi:10.1177/ 287Ð306. doi:10.1080/00377310009517593 104973159100100401 Vonk, M. E., Tripodi, T., & Epstein, I. (2007). Re- Serna, L. A., Schumaker, J. B., Sherman, J. A., & Sheldon, search techniques for clinical social workers (2nd J. B. (1991). In-home generalization of social interac- ed.). New York, NY: Columbia University Press. tions in families of adolescents with behavior prob- Wade, C. M., Ortiz, C., & Gorman, B. S. (2007). lems. Journal of Applied Behavior Analysis, 24, 733Ð Two-session group parent training for bedtime 746. doi:10.1901/jaba.1991.24-733 noncompliance in Head Start preschoolers. Child Sidman, M. (1960). Tactics of scientific research: & Family Behavior Therapy, 29, 23Ð55. doi: Evaluating experimental data in psychology. New 10.1300/J019v29n03_03 York, NY: Basic Books. Watson, P. J., & Workman, E. A. (1981). The non- Smith, J. D. (2012). Single-case experimental de- concurrent multiple baseline across-individuals signs: A systematic review of published research design: An extension of the traditional multiple and current standards. Psychological Methods, 17, baseline design. Journal of Behavior Therapy and 510Ð550. doi:10.1037/a0029312 Experimental Psychiatry, 12, 257Ð259. doi: Stevens, R. E. S. (2006). Research ‘Alive’ in the 10.1016/0005-7916(81)90055-0 social work classroom: Teaching graduate social Weber, L. A. (1980). The effect of videotape and work students to evaluate practice effectiveness. playback on an in-patient adolescent group. Inter- national Journal of Group Psychotherapy, 30, Arete, 29, 62Ð68. 213Ð227. Strauss, B., Burlingame, G. M., & Bormann, B. What Works Clearinghouse. (2013). Procedures and (2008). Using the CORE-R battery in group psy- standards handbook (Version 3.0). Retrieved from chotherapy. Journal of Clinical Psychology, 64, http://ies.Ed.gov/ncee/wwc/pdf/reference_resources/ 1225Ð1237. doi:10.1002/jclp.20535 wwc_procedures_v3_0_draft_standards_handbook Thyer, B. A., & Curtis, G. C. (1983). The repeated .pdf pretest-posttest single-subject experiment: A new Wong, S. E. (2010). Single-case evaluation designs for design for empirical clinical practice. Journal of practitioners. Journal of Social Service Research, 36, Behavior Therapy and Experimental Psychiatry, 248Ð259. doi:10.1080/01488371003707654 14, 311Ð315. doi:10.1016/0005-7916(83)90073-3 Wong, S. E., & Liberman, R. P. (1981). Mixed single- Tolman, R. M., & Molidor, C. E. (1994). A decade of subject designs in clinical research: Variations of the social group work research: Trends in methodol- multiple-baseline. Behavioral Assessment, 3, 297Ð ogy, theory, and program development. Research 306. on Social Work Practice, 4, 142Ð159. doi:10.1177/ Wong, S. E., Morgan, C., Crowley, R., & Baker, J. N. 104973159400400202 (1996). Using a table game to teach social skills to Toseland, R., Krebs, A., & Vahsen, J. (1978). Chang- adolescent psychiatric inpatients: Do the skills ing group interaction patterns. Journal of Social generalize? Child & Family Behavior Therapy, 18, Service Research, 2, 219Ð232. 1Ð17. doi:10.1300/J019v18n04_01 Toseland, R. W., & Rivas, R. F. (2012). An introduc- Wong, S. E., Wright, J., Terranova, M. D., Bowen, tion to group work practice (7th ed.). Boston, MA: L., Zarate, R., & Zarate, R. (1988). Effects of This document is copyrighted by the American Psychological Association or one of its allied publishers. Allyn & Bacon. structured ward activities on appropriate and psy- This article is intended solely for the personal use of the individualTracey, user and is not to be disseminated broadly. D. A., Briddell, D. W., & Wilson, G. T. chotic behavior of chronic psychiatric patients. (1974). Generalization of verbal conditioning to Behavioral Residential Treatment, 3, 41Ð50. doi: verbal and nonverbal behavior: Group therapy 10.1002/bin.2360030104 with chronic psychiatric patients. Journal of Ap- Zastrow, C. (2006). Social work with groups: A com- plied Behavior Analysis, 7, 391Ð402. doi:10.1901/ prehensive workbook (6th ed.). Belmont, CA: jaba.1974.7-391 Thomson. Tymchuk, A. J., Andron, L., & Rahbar, B. (1988). Effec- tive decision-making/problem-solving training with Received February 15, 2013 mothers who have mental retardation. American Revision received November 4, 2013 Journal on Mental Retardation, 92, 510Ð516. Accepted November 5, 2013 Ⅲ Group Dynamics: Theory, Research, and Practice © 2014 American Psychological Association 2014, Vol. 18, No. 2, 159Ð173 1089-2699/14/$12.00 DOI: 10.1037/gdn0000004 Team Task Conflict Resolution: An Examination of Its Linkages to Team Personality Composition and Team Effectiveness Outcomes

Thomas A. O’Neill Natalie J. Allen University of Calgary Western University

By drawing attention to task conflict resolution and reporting on its empirical relations with key team variables, the present study offers a new perspective on team effective- ness. Specifically, we examined how three “dark” personality traits (Manipulativeness, Narcissism, and Secondary Psychopathy) relate to team conflict resolution, team innovation, and team task performance in student engineering design project teams. Interestingly, results indicate that mean team levels of Secondary Psychopathy is the most important predictor, and task conflict resolution is a mediator linking mean Secondary Psychopathy to team task performance. Furthermore, the prediction of team performance by Secondary Psychopathy is as strong as is any of the “Big 5” personality variables investigated in meta-analyses.

Keywords: team conflict, task conflict, task conflict resolution, team effectiveness, Dark Triad, personality, psychopathy

Work teams form the essential building place emphasis on task conflict, which is the blocks of modern organizations (Hackman, extent to which team members have different 2002). As noted by Kozlowski and Bell (2003), opinions, perspectives, and views of the task the adoption of teams emerged, in large part, (Jehn, 1995). Task conflict is of particular in- from global competition and an emphasis on terest because, unlike most types of conflict, it innovation, which created a need for organiza- has been argued to be conducive to team per- tions to be increasingly flexible and adaptable. formance (Amason, 1996). Meta-analyses, Identifying the determinants of work team pro- however, indicate that task conflict is weakly cesses, innovation, and performance, therefore, related to team performance, on average (e.g., is fundamental to our understanding of organi- O’Neill, Allen, & Hastings, 2013). zational effectiveness (see Sundstrom, McIn- A potential limitation of task conflict re- tyre, Halfhill, & Richards, 2000). search is that it ignores whether resolutions of Perhaps not surprisingly, team conflict is one those conflicts were actually reached. This is an of the most researched team processes (see important issue, because a team with unresolved Korsgaard, Jeong, Mahony, & Pitariu, 2008). conflicts may experience impasses, deadlocks, Consider the recent meta-analysis by de Wit, and barriers to forward movement, whereas a Greer, and Jehn (2012); in that research, the team with resolved conflicts likely integrates, authors report on data from 116 studies involv- ing almost 9,000 teams. Many of the studies combines, and synthesizes divergent views and This document is copyrighted by the American Psychological Association or one of its allied publishers. achieves consensus on how best to proceed (see

This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. Robey, Farrow, & Franz, 1989). Our review of the literature presented below, however, sug- gests that empirical research examining both Thomas A. O’Neill, Department of Psychology, Univer- antecedents and consequences of task conflict sity of Calgary, Calgary, Alberta, Canada; Natalie J. Allen, Department of Psychology, Western University, London, resolution is lacking. In light of the potential Ontario, Canada. importance of task conflict resolution, we con- Correspondence concerning this article should be ad- sider both innovation and task performance as dressed to Thomas A. O’Neill, Department of Psychology, possible team outcomes in a sample of engi- University of Calgary, 2500 University Drive, N.W., Cal- gary, AB Canada T2N 1N4, P: 403-220-5207. E-mail: neering design project teams. We also investi- [email protected] gate potential antecedents (i.e., input variables)

159 160 O’NEILL AND ALLEN

that could be useful for pinpointing the condi- al., 2013). Resolving task conflicts, on the other tions in which task conflicts are most likely to hand, could be critical to team outcomes be- be resolved. Specifically, in this research, we cause team members have encountered and ad- focus on team-level Manipulativeness, Narcis- equately addressed divergent viewpoints re- sism, and Secondary Psychopathy. These traits garding the direction in which the team ought to are similar, although not identical, to the “Dark proceed. Compared with research on conflict Triad” (Paulhus & Williams, 2002), which in- types, there is scant consideration of the extent to cludes a common core of selfishness, remorse- which a team actually resolves conflicts, the type lessness, disagreeableness, and aggressiveness of conflicts resolved, and whether resolutions (Jonason, Webster, Schmitt, Li, & Crysel, predict team outcomes (e.g., De Dreu, 2011). 2012). Although Judge and LePine (2007) We also note that there is research on the strat- called for team research on dark traits, current egies, processes, and styles of negotiating con- literature has primarily focused on the Five- flicts, resolving disputes, and allocating scarce Factor Model (Bell, 2007; Driskell, Goodwin, resources (e.g., Behfar, Peterson, Mannix, & Salas, & O’Shea, 2006). Trochim, 2008; Katz & Kahn, 1978; Rubin, Given our examination of team inputs (per- Pruitt, & Kim, 1994), but there is little empirical sonality composition), processes (task conflict research informing our understanding of the resolution), and outcomes (innovation and task antecedents and consequences of task conflict performance), the current research design pres- resolution. ents the opportunity to investigate the role of Theoretical work on conflict resolution by task conflict resolution as a process variable that Robey and colleagues (i.e., Robey & Farrow, mediates the effects of team personality com- 1982; Robey et al., 1989) deals with factors position on team outcomes. As LePine, Buck- leading to satisfactory conflict resolution (see man, Crawford, and Methot (2011) noted in also Barki & Hartwick, 1994). Notable findings their recent review, the intervening processes from that research suggest participation leads to and mechanisms linking team personality inputs power and influence, which in turn is positively to team outcomes are understudied and largely associated with satisfactory resolution of con- unknown. This study aims to address that gap. flict. However, those studies were conducted at the individual level and, therefore, they ignore Theoretical Development whether the team, as a collective, was satisfied with resolutions. Furthermore, the studies con- Task Conflict Resolution found the different types of conflict (e.g., task, In the current research we draw attention to relationship), and conflict resolution is treated the issue of task conflict resolution and investi- as the final outcome rather than as a process gate its associations with team inputs and out- affecting team innovation and task perfor- comes. As described by Robey et al. (1989), mance. In a notable exception, Jehn, Greer, “conflict resolution is apparent when members Levine, and Szulanksi (2008) offer a three-item have fewer objections and indicate more agree- measure of task conflict resolution efficacy, ment with each other...itindicates a solution which was related positively to team viability that is acceptable to the entire group” (p. 1174). and emergent states. However, the construct Wall and Callister (1995) noted that conflict pertains to “the belief that conflict can be easily This document is copyrighted by the American Psychological Association or one of its allied publishers. resolutions “may be in the form of explicit or resolved” (p. 465) rather than actual conflict This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. tacit agreement” (p. 525), and Kuhn and Poole resolutions. Greer, Jehn, and Mannix (2008) (2000) viewed conflict resolution as “coming to used single-item scales to measure each of task, a mutually agreeable solution” (p. 570). We relationship (i.e., interpersonal incompatibili- define task conflict resolution as the extent to ties), and process (e.g., work-structuring incom- which different opinions, viewpoints, and per- patibilities) conflict resolutions. Arguably, mul- spectives were resolved. tiitem scales might be more reliable and could As mentioned, recent meta-analyses summa- provide better coverage of the content domains, rizing the substantial body of research involving and consideration of team outcomes (e.g., team task conflict and team performance report small task performance) might shed new light on the effect sizes (e.g., de Wit et al., 2012; O’Neill et importance of task conflict resolution. TEAM TASK CONFLICT RESOLUTION 161

Task Conflict Resolution and H1b: Task conflict resolution will be posi- Team Outcomes tively correlated with team task performance.

Team innovation involves developing a cre- Input-Process-Output (IPO) Model and ative or novel solution that is also practical and feasible in its development and application Team Personality Composition (e.g., Amabile, 1983; Drach-Zahavy & Somech, In the current research we applied the well- 2001a, 2001b), whereas team task performance known IPO model of team effectiveness (see refers to the overall quality of the team’s task- McGrath, 1984). The IPO model is best consid- related output (e.g., Hackman, 1987). This dis- ered a heuristic that organizes theorizing about tinction suited the context of the current study. how team inputs (e.g., personality) affect team Specifically, the objective of the engineering processes (e.g., task conflict resolution) and, in project teams involved was to develop a novel turn, team outcomes (e.g., innovation, task per- and effective approach to demonstrating how an formance). In the context of the present re- environmentally friendly technology creates re- search, task conflict resolution is theorized to be newable energy (e.g., team innovation), and to a mechanism by which personality composition deliver an extensive report documenting the impacts team outcomes. team’s design and the prototype’s capabilities Variables organized under the IPO frame- (team task performance). work are conceptualized at the team level and, Teams capable of resolving task conflicts are therefore, personality variables must be aggre- likely to perform well and reach high levels of gated prior to hypothesis testing. Operationaliz- innovation. Drawing from social interdepen- ing traits using the within-team mean, as we do dence theory, compatible team-member goals here, invokes an additive theory of aggregation appear to facilitate effective conflict handling, (see Chan, 1998). Based on the additive theory, which in turn results in identifying novel and the arithmetic mean of team member scores is high-quality solutions (see Deutsch, 2006). Ad- most relevant because it is the overall “abun- ditionally, based on information processing the- dance” of the trait, within the team, that is ory, we suggest unresolved conflicts may arouse expected to be most consequential to the team’s the autonomic nervous system, thereby inhibit- work (e.g., LePine, Hollenbeck, Ilgen, & Hed- ing critical analysis of the task and creative lund, 1997; LePine, 2003; O’Neill & Allen, thinking (e.g., Carnevale & Probst, 1998). In- 2011). deed, teams with positive interpersonal relation- ships exhibit greater creativity than do teams in Dark Traits as Antecedents of Task conflict (e.g., Craig & Kelly, 1999). Conflict Resolution In project teams of the nature studied here, we anticipated this would be particularly impor- One potential approach is the “Dark Triad,” tant because meeting milestones and keeping which encompasses Narcissism, Machiavellian- forward momentum was essential to team inno- ism, and Psychopathy (Paulhus & Williams, vation and team task performance (see Chioc- 2002). Narcissism describes people who hold chio, 2007). Behfar et al.’s (2008) qualitative feelings of superiority, grandiosity, and entitle- study of project teams found that high- ment (e.g., Ames, Rose, & Anderson, 2006; This document is copyrighted by the American Psychological Association or one of its allied publishers. performing teams used pluralistic strategies in- Raskin & Terry, 1988). Rather than focusing on This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. volving either a preemptive focus on addressing Machiavellianism, we chose a central dimen- potential conflicts, or quickly and efficiently sion of Machiavellianism referred to as Manip- containing conflicts by invoking rules to resolve ulativeness (see Paunonen & Jackson, 2000). conflicts (e.g., majority rules). Teams failing to Manipulativeness involves the willingness to address and to resolve different viewpoints are use of flattery, ingratiation, and deception to likely to experience stalls or deadlocks that in- advance personal agendas (see Paunonen, Had- hibit smooth progress. In sum, we predict the dock, Forsterling, & Keinonen, 2003). We view following: the content of the Manipulativeness dimension H1a: Task conflict resolution will be posi- of Machiavellianism as highly relevant to the tively correlated with team innovation. teams’ capacities to resolve conflicts. Although 162 O’NEILL AND ALLEN

individuals high on Manipulativeness use im- and sequential time periods. Although we do pression management techniques to advance not use an experimental design and, as a con- their own agenda, they do not necessarily pos- sequence, causality cannot be unequivocally sess the charm, appeal, and magnetism of Ma- determined, the ordering of measurement co- chiavellians (cf. Paunonen & Jackson, 2000). incides with the likely temporal ordering of Finally, we considered Secondary Psychopathy, these variables. Specifically, we theorize that which involves impulsivity, lack of goal com- personality would most likely be an anteced- mitment, delinquency, irresponsibility, and dif- ent (not a mediator or outcome), and that task ficulties establishing relationships (e.g., Leven- conflict resolution would most likely precede son, Kiehl, & Fitzpatrick, 1995; McHoskey, team outcomes. Worzel, & Szyarto, 1998). Although the current H3: Task conflict resolution will mediate the traits do not overlap perfectly with the Dark relationships between team personality vari- Triad, the Dark Triad literature is relevant to ables (Manipulativeness, Narcissism, and Sec- formulating our hypotheses. ondary Psychopathy) and team outcome vari- Affective events theory (AET; Weiss & Cro- ables (innovation and performance). panzano, 1996) is a framework for understand- ing how Manipulativenss, Narcissism, and Sec- ondary Psychopathy may interfere with the Dark Traits and Team Outcomes resolution of task conflict. One proposition that is consistent with AET is that personality affects Although we argued for an indirect linkage the occurrence of events at work and how those involving the dark traits and team outcomes, events translate into affective experiences, theory also supports direct effects. Manipulative which in turn have implications for behavior individuals engage in impression management (see Wegge, van Dick, Fisher, West, & Daw- strategies involving ingratiation and self- son, 2006). Common tendencies of individuals promotion (see Paunonen et al., 2003), which, high on dark traits similar to those considered in when viewed as insincere, tend to be associated this study include the use of adversarial and with distrust and skepticism (e.g., Baron, 1986). disruptive behaviors, and engagement in malev- As several team members attempt to manipulate olent interpersonal dealings (e.g., Jonason & each another, the team atmosphere would likely Webster, 2012). As such, these individuals be rife with suspicion, which would not likely would likely exacerbate the challenge of effec- promote team collaboration, innovation, or per- tively resolving task conflicts by arousing neg- formance. Team members high on Narcissism ative emotions. This could occur through their might have excessively optimistic views about use of “hard tactics” such as threats, punish- the quality of their work or their own contribu- ments, and manipulations (e.g., Jonason, Slom- tions even in light of evidence to the contrary. ski, & Partyka, 2012), lack of self-control and Furthermore, narcissistic individuals are seen understanding of behavioral consequences (e.g., Jonason & Tost, 2010), emotional intelligence by their group members as domineering, ingen- deficits (e.g., Jordan & Troth, 2004; Petrides, uous, and lacking in interpersonal intelligence Vernon, Schermer, & Veselka, 2011), and em- (e.g., Rauthmann, 2012), and a team of such pathy limitations (e.g., Jonason & Krause, individuals would not likely be effective. Fi- 2013; Wai & Tiliopoulos, 2012). Negative emo- nally, teams with members high on Secondary This document is copyrighted by the American Psychological Association or one of its allied publishers. tions and arousal in teams high on dark traits Psychopathy would likely derail the team’s This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. would interfere with effective information pro- progress because of their members’ impulsivity, cessing (e.g., Carnevale & Probst, 1998), which aggressiveness, and emotional instability (cf. is likely to detract from problem-focused dis- McHoskey et al., 1998; Paulhus & Williams, cussions aimed at resolving task conflicts. Ac- 2002). In sum, we propose the following: cordingly, we offer the following predictions: H4a: Mean Manipulativeness, Narcissism, H2: Mean Manipulativeness, Narcissism, and and Secondary Psychopathy will be negatively Secondary Psychopathy will be negatively cor- correlated with team innovation. related with task conflict resolution. H4b: Mean Manipulativeness, Narcissism, In the current study we measured dark and Secondary Psychopathy will be negatively traits, conflict, and outcomes at three distinct correlated with team task performance. TEAM TASK CONFLICT RESOLUTION 163

Method Measures Time 1: Personality. Manipulativeness Participants and Teamwork Context was measured using Paunonen’s (2002) scale Data were collected from 344 team members from the Supernumerary Personality Inventory working in 81 engineering project design teams (SPI). The scale contains 15 items with 5-point Likert-type response options (␣ϭ.70). Exam- for 6.5 months. The participants were students ple items are as follows: By flattering someone in an engineering design course at a large Ca- I can make him or her more apt to agree with nadian university. Teams had four (n ϭ 61) or me and I will pretend to be extremely interested five (n ϭ 20) members, and 85% of participants in what a person is saying in order to get were male (M age ϭ 18.6 years, SD ϭ 2.4). something. Narcissism was measured using the Teams were formed pseudorandomly. Specifi- 16-item forced-choice form of the Narcissistic cally, individuals were asked to rank order their Personality Inventory (␣ϭ.67; Ames et al., proficiency on “skill sets” (i.e., communication, 2006). Specifically, respondents were asked to computer, planning, hands on), and then teams choose which of two statements are most like were built such that they had one individual them, such as I really like to be the center of with a ranking of “1” on each dimension. The attention and It makes me uncomfortable to be course instructors preferred this method to full the center of attention. Another item set was random assignment. People sometimes believe what I tell them and I Our observations indicate teams took great can make anybody believe anything I want them pride in designing their devices and presenting to. Secondary Psychopathy was measured using them at a high-profile design showcase in which 10 items from the Levenson Self-Report Psy- the city mayor, university president, engineer- chopathy Scale (␣ϭ.63; Levenson et al., 1995) ing dean and faculty, and the public were in with 5-point Likert-type response options. Ex- attendance. This “major design project” was the ample items are I find myself in the same kinds focus of team innovation and team task perfor- of trouble, time after time and I have been in a mance in the current research. The projects lot of shouting matches with other people. were complicated and inherently high on Time 2: Task conflict resolution and task knowledge interdependence because team conflict. To measure task conflict resolution, members tended to become experts in particular we observed teams in their engineering design design areas critical to the prototype’s innova- laboratories over a 6 month period. Extensive tion, functionality, and documentation. Thus, diary notes were taken during the approximate the teams in this sample fit well within the 100 hours we spent, overall, observing various accepted definition of work teams: highly inter- teams receiving instructions, training, and feed- dependent members working on consequential back, and during brainstorming, decision- tasks and having a collective purpose (e.g., Al- making exercises, and design work. Specifi- len & West, 2005; Hackman, 1987). cally, the first author attended 4 of the 8 weekly design labs on a rotating basis during which Procedure and Timeline teams were engaged with various group activi- ties including the focal project. Through this Participants were invited to complete person- informal team task analysis, we identified four This document is copyrighted by the American Psychological Association or one of its allied publishers. ality and conflict surveys during weekly design key project milestones: a) problem definition, b) This article is intended solely for the personal use of the individualclasses. user and is not to be disseminated broadly. Personality data were collected in the deciding on the team’s design concept, c) de- first design class before participants were as- veloping the team’s prototype, and d) preparing signed to their teams; conflict processes were the team presentation. Next, we adopted the assessed 6 months later, 2 weeks before com- following general form across four items: the pleting the major design project. Team innova- extent to which different opinions, viewpoints, tion ratings were based on prototype photos and and perspectives were resolved before settling written descriptions collected a week before on the team’s ______. For each item, final project reports were due. Team task per- the blank contained problem definition, design formance was assessed after the major design concept, prototype, and presentation (␣ϭ.89), project was submitted at the 6.5-month mark. respectively. Jehn’s (1995) four-item measure 164 O’NEILL AND ALLEN

of task conflict was also administered (␣ϭ.89). cept, CAD sketches, mathematical modeling, Scale anchors for both measures ranged from 1 and implications. The assessment of team task (A very small amount)to5(A lot). performance consisted of ratings provided by Time 3: Team innovation. Team innova- experienced course instructors. Because raters tion was assessed using Amabile’s (1983) con- did not rate the same teams, we controlled for sensual assessment technique, which involves the possibility of raters using different perfor- collecting innovation judgments from domain- mance distributions by z-standardizing within relevant experts. Five expert judges who were raters (which is common practice in similar engineering graduate students rated the innova- situations; O’Neill & Allen, 2012; Wageman & tion of each design prototype. The targets of Gordon, 2005). experts’ ratings were short, written descriptions and 4 ϫ 6 in. color photographs of the team prototypes. The relative percentile method Results (RPM) formed the basis of the rating scale employed to assess team innovation (see Goffin, Aggregation Gellatly, Paunonen, Jackson, & Meyer, 1996). The current application of the RPM required Task conflict resolution and task conflict are judges to rate each team’s level of innovation on “shared-unit” constructs; therefore, within-team a scale ranging from zero to 100. Team inno- agreement and between-team heterogeneity is vation was scored as the average of each judge’s required to support construct validity prior to rating of each team prototype. An intraclass aggregation (Kozlowski & Klein, 2000). Ac- correlation (ICC) was computed for assessing cording to Bliese (2000), ICCs indicate the pro- the reliability of multiple raters rating team in- portion of variance explained by group mem- novation on a continuous scale [ICC(2) ϭ .71; bership [that is, ICC(1)] and the reliability of Shrout & Fleiss, 1979]. We note that the experts the group mean [that is, ICC(2)]. Table 1 con- were course teaching assistants leading differ- tains ICCs for the shared-unit constructs of task ent design studios, and they were each familiar with a unique set of approximately 10 of the 81 conflict and task conflict resolution, which were projects. However, all judges rated all projects all significantly greater than zero. Furthermore, in a fully crossed design. Therefore, any possi- Jehn et al. (2008) interpreted ICCs of this mag- ble bias introduced by familiarity with the proj- nitude (and lower) as sufficient for justifying ects would likely be limited after averaging aggregation. The personality variables were ag- across all five judge ratings, and the strong gregated using the mean but were not expected interrater reliability supports this view. to demonstrate within-group agreement because Time 4: Team task performance. Teams teams were formed pseudorandomly, traits are were required to submit reports of approxi- configural unit properties (i.e., not shared; Koz- mately 100 pages containing background re- lowski & Klein, 2000), and their function is search, documentation of the theoretical and theorized to be additive across team members observed capabilities of the team’s design con- (LePine, 2003; O’Neill & Allen, 2011).

Table 1 This document is copyrighted by the American Psychological Association or one of its allied publishers. Means, Standard Deviations, ICCs, and Correlations Involving Team-Level Study Variables This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. Variable ICC(1) ICC(2) MSD 123456 1. Manipulativeness 2.91 0.21 ءNarcissism 4.95 1.38 .40 .2 ءSecondary psychopathy 2.34 0.23 .17 .29 .3 ء26. 21. 17. 0.51 2.13 ء54. ءTask conflict .24 .4 ءϪ.39 ءϪ.02 Ϫ.12 Ϫ.30 0.55 3.91 ء40. ءTask conflict resolution .14 .5 6. Team innovation 53.96 17.38 .05 Ϫ.06 Ϫ.21 Ϫ.11 .14 ء37. ءϪ.18 .30 ءTeam task performance 0.00 0.96 Ϫ.06 .08 Ϫ.30 .7 Note. n ϭ 81. .p Ͻ .05 ء TEAM TASK CONFLICT RESOLUTION 165

Tests of Hypotheses variables (see Table 1). One procedure for com- paring the importance of predictors is relative We predicted task conflict resolution would importance analysis (e.g., Johnson, 2000), be positively related to team innovation (H1a). which overcomes the limitations of both zero- Although in the predicted direction, this relation order correlations as well as betas from standard did not reach significance (see Table 1). Task multiple regression (e.g., Johnson & LeBreton, conflict resolution was expected to positively 2004). Table 2 contains the results of the rela- correlate with team task performance (H1b). As tive importance analysis as well as overall 2 2 this was the case, H1b was supported. model R values. Model R values indicate that We anticipated that Manipulativeness, Narcis- the personality traits, as a set, accounted for sism, and Secondary Psychopathy would be neg- 12% of the variance in team task performance atively related to task conflict resolution (H2). and 10% of the variance in task conflict resolution. Although mean Manipulativeness and Narcissism Overall prediction of task conflict and team inno- were not related to task conflict resolution, there vation was nonsignificant. In terms of the relative was a significant negative relation involving Sec- importance of the three personality traits, the ondary Psychopathy (see Table 1). We also ex- results clearly indicate that Secondary Psychop- pected that the effect of the dark personality traits athy was dominant. Of the proportion of vari- on team outcomes would be mediated by task ance explained by the personality variables as a conflict resolution (H3). Preacher and Hayes set, Secondary Psychopathy accounted for 57% (2004) recommended the bootstrapping approach (task conflict), 78% (team task performance), for testing the significance of the indirect effect of 82% (team innovation), and 91% (task conflict the predictor on the outcome through the interven- resolution). Thus, Secondary Psychopathy was ing variable. Specifically, bias-corrected upper the most important predictor of the team process and lower 95% confidence intervals were calcu- and outcome variables studied. lated using 1000 bootstrapped samples. Analyses supported the significance of the indirect effect for Discussion one model (see Figure 1): Secondary Psychopathy ¡ task conflict resolution ¡ team task perfor- The overarching purpose of this study was to mance (unstandardized indirect effect ϭϪ.28, investigate the role of task conflict resolution as ϭϪ Ϫ a team process linking dark personality traits to CI95% .68 to .06). This indicates a one-unit increase in mean Secondary Psychopathy was re- team outcomes. This is notable because the ac- lated to a .28 decrease in team task performance tual resolution of task conflict has received through task conflict resolution. Furthermore, the much less consideration than has the occurrence direct effect of Secondary Psychopathy on team of task conflict, yet resolution could be at least task performance remained significant after taking as important. Taken together, the set of hypoth- into account the indirect effects, suggesting Sec- eses we forwarded involving these relations re- ondary Psychopathy has both direct and indirect ceived partial support. In our view, the contri- paths to team task performance.1 bution of the current study is threefold. First, Regarding team outcomes, Manipulativeness, compared with task conflict, findings for the Narcissism, and Secondary Psychopathy should resolution of task conflict appear more intrigu- ing and theoretically informative. Task conflict be related negatively to team innovation (H4a). However, none of the relations reached conven- resolution was significantly related to team task This document is copyrighted by the American Psychological Association or one of its allied publishers. tional levels of significance (see Table 1). Turn- performance, and it was a mediator between This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. Secondary Psychopathy and team task perfor- ing to team task performance (H4b), of the three traits examined, a significant negative associa- mance. This is novel given the dearth of past tion was found only for Secondary Psychopa- thy. 1 For completeness, we considered task conflict as a me- diator and as a moderator in moderated mediation analyses. Supplemental Analyses The moderated mediation analyses examined the role of task conflict moderating either the effect of task conflict resolution on team outcomes or personality traits on task Given the pattern of results involving person- conflict resolution (Hayes, 2013). We did not find any ality, Secondary Psychopathy appeared to be significant effects for the function of task conflict in any of the most important predictor of the teamwork these roles. 166 O’NEILL AND ALLEN

Task Conflict -.72* Resolution .39*

Secondary Team Task

Psychopathy Performance -.94*

Figure 1. Mediation model with unstandardized path coefficients. All paths in the model are significant including the direct path from Psychopathy to team task performance after controlling for the indirect effect through task conflict resolution. The indirect effect was .p Ͻ .05 ء .Ϫ.28

research on task conflict resolution and the ex- 2003; Tjosvold, 1998). It is also consistent with isting focus on task conflict levels. Indeed, the the project management literature (e.g., Chioc- authors of recent meta-analyses found 83 chio & Lafrenière, 2009) given that resolved (O’Neill et al., 2013) and 116 (de Wit et al., task conflict is needed to progress smoothly and 2012) articles investigating conflict types and efficiently through project stages. Thus, it may team performance. Consistent with information be advantageous for researchers and practitio- processing theory, results from those meta- ners to emphasize strategies that help teams analyses suggested, with few exceptions, that work through task conflicts rather than to en- task conflict is either unrelated to or harmful for courage teams to engage in task conflict without team performance (as it was in the present re- addressing its resolution. search). Accordingly, a continued focus on task A second novel contribution of the current conflict itself may be limiting (see also research involves the examination of team per- DeChurch, Mesmer-Magnus, & Doty, 2013). sonality composition from the perspective of Our findings are important for theory and prac- dark traits (cf. Judge & LePine, 2007). Relative tice because they imply that the resolution of to Manipulativeness and Narcissism, Secondary task conflict may be a critical factor. This fits Psychopathy dominated the prediction of task well within social interdependence theory, as conflict resolution, team innovation, and team interdependence often leads to conflict and, task performance. Interestingly, Secondary Psy- therefore, the manner in which the conflict is chopathy exhibited stronger relations with team managed may be most important (see Johnson, task performance than did any of the Big Five,

Table 2 Relative Importance Analysis of Team-Level Personality Variables

Criterion and statistic Manipulativeness Narcissism Secondary psychopathy R2 Task conflict ␤ .08 .12 .21† .09 Raw relative weight .02 .03 .05 Regression weight as a percentage of R2 16 28 57 Task conflict resolution ء ء Ϫ Ϫ ␤ This document is copyrighted by the American Psychological Association or one of its allied publishers. .05 .05 .30 .10 Raw relative weight .00 .01 .09 This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. Regression weight as a percentage of R2 18 91 Team innovation ␤ .11 Ϫ.04 Ϫ.22† .06 Raw relative weight .01 .00 .05 Regression weight as a percentage of R2 11 7 82 Team task performance ء12. ءϪ.09 .21 Ϫ.32␤ Raw relative weight .01 .02 .10 Regression weight as a percentage of R2 51778 Note. n ϭ 81. .p Ͻ .05 ء TEAM TASK CONFLICT RESOLUTION 167

in terms of uncorrected correlations, reported in als high on Secondary Psychopathy will tend to past meta-analyses (e.g., Bell, 2007). Character- experience failures in resolving task conflicts, istic patterns of behavior of individuals high on which, in turn, results in difficulties moving the Secondary Psychopathy include difficulties re- project forward and achieving high task perfor- lating to others, emotional outbursts, impulsiv- mance. Thus, although meta-analyses highlight ity, and lack of goal commitment (Levenson et the association between personality and perfor- al., 1995; McHoskey et al., 1998). Consistent mance in teams (e.g., Bell, 2007), in the current with theories such as affective events theory, study we report on one of the few empirical teams of individuals with high Secondary Psy- findings supporting a linking mechanism chopathy likely experienced frequent and re- through which team personality composition in- peated cycles of negative affective interactions, fluences team performance. As mentioned potentially resulting in reinforcing downward above, teams high on mean Secondary Psychop- spirals (cf. Peterson & Behfar, 2003). For ex- athy contain individuals who are likely hostile, ample, individuals high on Secondary Psychop- aggressive, impulsive, and unable to remain athy often use social influence tactics involving committed to goals. It appears teams of such coercion, debasement, and threats (Jonason et individuals experience barriers to resolving task al., 2012); tend to be dishonest and lack humil- conflicts and to finding agreement regarding ity (Lee & Ashton, 2005); exhibit self-control avenues toward high-quality task work. In turn, difficulties (Jones & Paulhus, 2011; Jonason & this was associated with reductions in team ef- Tost, 2010); appear to engage in various forms fectiveness. Importantly, both indirect and di- of aggression, cheating, and bullying (Boddy, rect connections to team task performance were 2010; Coyne & Thomas, 2008); and tend to supported in the model, suggesting mean Sec- hold social dominance beliefs (Hodson, Hogg, ondary Psychopathy has at least two distinct & MacInnis, 2009). paths through which it impacts team task per- Given the above, placing a number of indi- formance. Investigating additional team pro- viduals high on Secondary Psychopathy in a cesses, such as psychological safety (Edmond- team will likely lead to numerous negative af- fective interactions, disputes, and arguments. It son, 1999), communication quality and quantity may be advisable, therefore, to build teams such (Barrick, Stewart, Neubert, & Mount, 1998), that overall levels of Secondary Psychopathy and other potentially relevant processes are are as low as possible. One way to accomplish promising avenues for future research. this involves the “seeding” method outlined by Humphrey, Hollenbeck, Meyer, and Ilgen On the Role of Mixed Motives (2007). When a trait is known to combine ad- ditively (such that increasing or decreasing the It is worth considering whether a lack of team mean on that trait is helpful for perfor- mixed motives and hierarchical goal structures mance), they would suggest assessing the trait might explain why Manipulativeness and Nar- and distributing the highest scorers across teams cissism were unrelated to team conflict pro- rather than concentrating them in any one team. cesses and team outcomes. The underlying ra- With criterion validity evidence and an analysis tionale for the relevance of Manipulativeness of the personality-based work requirements and Narcissism is that they should be related to suggesting the trait is job relevant (see Arthur, individuals advancing their own objectives, This document is copyrighted by the American Psychological Association or one of its allied publishers. Edwards, Bell, Villado, & Bennett, 2005), the agendas, and goals at the expense of pursuing This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. case for the legal defensibility of using the test the team’s collective purpose. Findings from to make personnel decisions with seeding meth- studies examining social interdependence the- ods could be strengthened (cf. Society for In- ory indicate that this situation leads to intrateam dustrial-Organizational Psychology Principles, competition, thereby contributing to selective 2003, and local employment law). information sharing, an argumentative approach A third contribution is that we identify task to differences of opinion, distrust, and a closed- conflict resolution as a mechanism linking team minded interpretation of new information (see personality composition, specifically mean lev- reviews by Johnson, 2003; Johnson & Johnson, els of Secondary Psychopathy, to team task 1989). But whether team members had proself performance. This suggests a team of individu- motives is somewhat of an open question. 168 O’NEILL AND ALLEN

Our observations and familiarity with the which supports the construct validity of the task teamwork context indicate that members quite conflict resolution measure. Thus, future re- likely experienced varying degrees of diver- search on task conflict resolution would seem gence in individual- and team-level motives. fruitful. First, participants had not yet specialized in a A second direction for future research in- particular subfield of engineering, and therefore volves studying how conflict is resolved. There the team was faced with identifying one or two is a voluminous literature on conflict manage- subfields on which their prototype would be ment approaches (e.g., Deutsch, 1973) and based (e.g., mechanical, electrical, software, ur- styles (e.g., Rahim, 1983), providing evidence ban, robotics). Given that team members often that cooperative goals and integrative styles are had different interests, this decision likely cre- associated with resolved conflicts (see Lewicki, ated some variance between individual and Weiss, & Lewin, 1992). But there is a need for team preferences. In turn, we expect that this research on how to promote effective ap- creates the conditions for manipulativeness, proaches to conflict management and resolu- conceit, and selfishness to inhibit conflict reso- tion, and what specific strategies and behaviors lution and joint effort toward a collective pur- lead to effective resolutions. For example, pose. Second, team members likely had diver- would it matter whether disagreements are re- gence on the level of quality and detail they felt solved through voting, autocracy, or avoidance? was needed, how many hours and meetings to Recently, Behfar et al., (2008) identified spe- invest, the nature of member roles and time- cific strategies used by teams to resolve con- lines, and so forth. Here again, manipulative flicts. Characteristic of high-performing and and egotistical behaviors would hinder integra- highly satisfied teams was equity-focused reso- tion, compromise, and consensus, thereby con- lution (e.g., work assignments based on skill). tributing to conflict and poor team outcomes. To Characteristic of low-performing and unsatis- the extent that team members had different pref- fied teams was adhocracy-focused resolution erences and goals such as those described here, (e.g., avoidance of conflict and group members it is almost certain that some individuals’ ob- in general). Tactics such as voting and majority jectives would be at odds with team objectives. rules were associated with high performance As a result, we speculated that teams high on but, interestingly, low satisfaction. On the other Manipulativeness or Narcissism would suffer. hand, equality rules that promoted cohesiveness The extent to which individual and group mo- were associated with low performance and high tives and goals were distinct, however, is diffi- satisfaction. What might drive this research for- cult to pinpoint, and future studies investigating ward is scale development and empirical exam- Manipulativeness, Narcissism, and related traits ination of how these different resolution strate- in teams may benefit from considering goal gies impact resolution itself, satisfaction with hierarchies and goal incongruences (e.g., De- the resolution, and team outcomes. This could Shon, Kozlowski, Schmidt, Milner, & Wiech- deepen our understanding of the specific behav- mann, 2004). iors that are needed for effective conflict reso- lution. Future Research In the way of a third avenue for future re- search, it may be valuable to consider relation- In light of the current study, we view team ship conflict resolution and process conflict res- This document is copyrighted by the American Psychological Association or one of its allied publishers. task conflict resolution as a promising avenue olution. The current research focused on task This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. for advancing team conflict research. In addi- conflict resolution as a first step, as this type of tion to the positive association with team task conflict has been studied extensively and has performance, team task conflict resolution ex- been described as the “productive” form of con- hibited a negative relation with task conflict. flict (e.g., De Dreu, 1997). It seemed to us that This suggests a number of teams experienced task conflict resolution may be promising, and difficulties with persistent and unresolved task our research provides cause for optimism in this conflicts, which may well have reduced their regard. Greer et al. (2008) found process con- ability to function effectively. Correlations be- flict resolution moderated the relation between tween task conflict resolution and the dark per- early process conflict and later process, task, sonality traits were in the negative direction, and relationship conflict such that resolutions TEAM TASK CONFLICT RESOLUTION 169

buffered continuing conflict occurrences. These mensional construct (Paunonen & Jackson, findings are encouraging despite being based on 2000) and Secondary Psychopathy is a narrower single-item conflict resolution scales; therefore, measure of a broad Psychopathy construct development of multiitem scales would likely (Ross, Bye, Wrobel, & Horton, 2008), suggest- be advantageous. ing the issue may not be as relevant for these latter traits. Limitations and Conclusion The current research emphasizes the role of task conflict resolution and provides evidence First, although the dark traits investigated in of a positive relation with team task perfor- the current research are similar to the Dark mance. Furthermore, task conflict resolution Triad, and we draw from the Dark Triad litera- was found to operate as an indirect linking ture, there are some important differences. We mechanism between mean Secondary Psychop- focused on Manipulativeness, which is a feature athy and team task performance. Thus, it seems of Machiavellianism but is not identical. Machi- new research focusing on the resolution of con- avellianism is a broader trait including charm, flict, rather than its levels, could be useful for charismatic tendencies (see Christie, Geis, & enriching our understanding of the role of con- Berge, 1970), and self-control (see Jones & flict processes in organizational work teams. Paulhus, 2011), and the fact that Manipulative- Finally, in addition to influencing both task con- ness was not predictive of teamwork variables flict resolution and overall task conflict levels, does not rule out the potential predictiveness of mean Secondary Psychopathy exhibited nega- Machiavellianism. We also considered Second- tive relations with team task performance, and ary Psychopathy because we expected that it at a marginally significant level, team innova- would have little overlap with Manipulativeness tion. Thus, further investigations of Secondary relative to Primary Psychopathy. However, the Psychopathy may be of considerable value for Dark Triad contains coverage of Primary Psy- advancing theory on team personality composi- chopathy, such as cynicism and callousness (Jo- tion. nason & Webster, 2010), which is strongly cor- related with Machiavellianism (e.g., McHoskey et al., 1998). Therefore, we had a theoretical References interest in considering traits that are slightly Allen, N. J., & West, M. A. (2005). Selecting for different from the Dark Triad, and our findings teamwork. In A. Evers, N. Anderson, & O. should not be assumed to fully capture the ef- Voskuijl (Eds.), The Blackwell handbook of per- fects of the Dark Triad on team conflict pro- sonnel selection (pp. 476Ð494). Oxford, UK: cesses or team outcomes. Second, despite the Blackwell Press. use of a cross-sectional design with three mea- Amabile, T. M. (1983). The social psychology of surement periods that corresponded to our the- creativity. New York, NY: Springer-Verlag. doi: oretical causal ordering of the relevant vari- 10.1007/978-1-4612-5533-8 ables, the mediation analyses do not allow Amason, A. C. (1996). Distinguishing the effects of functional and dysfunctional conflict on strategic definitive “cause-and-effect” conclusions (Judd decision making: Resolving a paradox for top & Kenny, 2010). Thus, experiments with full management teams. Academy of Management randomization to conditions should be con- Journal, 39, 123Ð148. doi:10.2307/256633 ducted. Third, replication of the current research Ames, D. R., Rose, P., & Anderson, C. P. (2006). The This document is copyrighted by the American Psychological Association or one of its allied publishers. is necessary to shed light on the robustness and NPI-16 as a short measure of narcissism. Journal This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. generalizability of the findings. Fourth, al- of Research in Personality, 40, 440Ð450. doi: though we used a unidimensional measure of 10.1016/j.jrp.2005.03.002 Narcissism, this trait consists of subdimensions Arthur, W., Edwards, B. D., Bell, S. T., Villado, concerning entitlement, leadership, self-absorp- A. J., & Bennett, W. (2005). Team task analysis: tion, and superiority (Emmons, 1987). These Identifying tasks and jobs that are team based. Human Factors, 47, 654Ð669. doi:10.1518/ subdimensions may have differential relations 001872005774860087 with teamwork variables, and therefore the cur- Barki, H., & Hartwick, J. (1994). User participation, rent study may not have detected nontrivial conflict, and conflict resolution: The mediating facet-level findings (cf. O’Neill & Allen, 2011). roles of influence. Information Systems Research, On the other hand, Manipulativeness is a unidi- 5, 422Ð438. doi:10.1287/isre.5.4.422 170 O’NEILL AND ALLEN

Baron, R. A. (1986). Self Presentation in Job Inter- ual Differences, 44, 1105Ð1115. doi:10.1016/j views: When there can be “too much of a good .paid.2007.11.002 thing”. Journal of Applied Social Psychology, 16, Craig, T. Y., & Kelly, J. R. (1999). Group cohesive- 16Ð28. doi:10.1111/j.1559-1816.1986.tb02275.x ness and creative performance. Group Dynamics: Barrick, M. R., Stewart, G. L., Neubert, M. J., & Theory, Research, and Practice, 3, 243Ð256. doi: Mount, M. K. (1998). Relating member ability and 10.1037/1089-2699.3.4.243 personality to work-team processes and team ef- DeChurch, L. A., Mesmer-Magnus, J. R., & Doty, D. fectiveness. Journal of Applied Psychology, 83, (2013). Moving beyond relationship and task con- 377Ð391. doi:10.1037/0021-9010.83.3.377 flict: Toward a process-state perspective. Journal Behfar, K. J., Perterson, R. S., Mannix, E. A., & of Applied Psychology, 98, 559Ð578. doi:10.1037/ Trochim, M. K. (2008). The critical role of conflict a0032896 resolution in teams: A close look at the links De Dreu, C. K. W. (1997). Productive conflict: The between conflict type, conflict management strat- importance of conflict management and conflict egies, and team outcomes. Journal of Applied Psy- issue. In C. K. W. De Dreu & E. Van De Vliert chology, 93, 170Ð188. doi:10.1037/0021-9010.93 (Eds.), Using conflict in organizations (pp. 10Ð .1.170 22). Thousand Oaks, CA: Sage. doi:10.4135/ Bell, S. T. (2007). Deep-level composition variables 9781446217016.n2 as predictors of team performance: A metaanaly- De Dreu, C. K. W. (2011). Conflict at work: Basic sis. Journal of Applied Psychology, 92, 595Ð615. principles and applied issues. In S. Zedeck (Ed.), doi:10.1037/0021-9010.92.3.595 APA Handbook of industrial and organizational Bliese, P. D. (2000). Within-group agreement, non- psychology (Vol. 3, pp. 461Ð493). Washington, independence, and reliability: Implications for data DC: APA. doi:10.1037/12171-013 aggregation and analysis. In K. J. Klein & S. W. J. DeShon, R. P., Kozlowski, S. W. J., Schmidt, A. M., Kozlowski (Eds.), Multilevel theory, research, and Milner, K. R., & Wiechmann, D. (2004). A mul- methods in organizations: Foundations, exten- tiple-goal, multilevel model of feedback effects on the regulation of individual and team performance. sions, and new directions (pp. 349Ð381). San Journal of Applied Psychology, 89, 1035Ð1056. Francisco, CA: Jossey-Bass. doi:10.1037/0021-9010.89.6.1035 Boddy, C. R. (2010). Corporate psychopaths, bully- Deutsch, M. (1973). The resolution of conflict: Con- ing and unfair supervision in the workplace. Jour- structive and destructive processes. New Haven, nal of Business Ethics, 100, 367Ð379. doi:10.1007/ CT: Yale University Press. doi:10.1177/ s10551-010-0689-5 000276427301700206 Carnevale, P. J., & Probst, T. M. (1998). Social Deutsch, M. (2006). Cooperation and competition. In values and social conflict in creative problem solv- M. Deutsch, P. T. Coleman, & E. C. Marcus ing and categorization. Journal of Personality and (Eds.), The handbook of conflict resolution: The- Social Psychology, 74, 1300Ð1309. doi:10.1037/ ory and practice (2nd ed., pp. 23Ð42). San Fran- 0022-3514.74.5.1300 cisco, CA: Wiley Chan, D. (1998). Functional relations among con- de Wit, F. R. C., Greer, L. L., & Jehn, K. A. (2012). structs in the same content domain at different The paradox of intragroup conflict: A meta- levels of analysis: A typology of composition analysis. Journal of Applied Psychology, 97, 360Ð models. Journal of Applied Psychology, 83, 234Ð 390. doi:10.1037/a0024844 246. doi:10.1037/0021-9010.83.2.234 Drach-Zahavy, A., & Somech, A. (2001a). Under- Chiocchio, F. (2007). Project team performance: A standing team innovation: The role of team pro- study of electronic task and coordination commu- cesses and structures. Group Dynamics: Theory, nication. Project Management Journal, 38, 97Ð Research and Practice, 5, 111Ð123. doi:10.1037/ 109. 1089-2699.5.2.111 This document is copyrighted by the American Psychological Association or one of its allied publishers. Chiocchio, F., & Lafrenière, A. (2009). A project Drach-Zahavy, A., & Somech, A. (2001b). Can we This article is intended solely for the personal use of the individual user and is notmanagement to be disseminated broadly. perspective on student’s declarative win them all? Benefits and costs of structure and commitments to goals established within asyn- flexible innovation implementations. Group Dy- chronous communication. Journal of Computer namics: Theory, Research and Practice, 25, 217Ð Assisted Learning, 25, 294Ð305. doi:10.1111/j 234. doi:10.1002/job.239 .1365-2729.2009.00310.x Driskell, J. E., Goodwin, G. F., Salas, E., & O’Shea, Christie, R., Geis, F. L., & Berger, D. (1970). Studies P. G. (2006). What makes a good team player? in Machiavellianism (p. ). New York, NY: Aca- Personality and team effectiveness. Group Dynam- demic Press. ics: Theory, Research, and Practice, 10, 249Ð271. Coyne, S. M., & Thomas, T. J. (2008). Psychopathy, doi:10.1037/1089-2699.10.4.249 aggression, and cheating behavior: A test of the Edmondson, A. C. (1999). Psychological safety and cheater-hawk hypothesis. Personality and Individ- learning behavior in work teams. Administrative TEAM TASK CONFLICT RESOLUTION 171

Science Quarterly, 44, 350Ð383. doi:10.2307/ Johnson, D. W., & Johnson, R. T. (1989). Coopera- 2666999 tion and competition: Theory and research. Edina, Emmons, R. A. (1987). Narcissism: Theory and mea- MN: Interaction Book. surement. Journal of Personality and Social Psy- Johnson, J. W. (2000). A heuristic method for estimat- chology, 52, 11Ð17. doi:10.1037/0022-3514.52 ing the relative weight of predictor variables in mul- .1.11 tiple regression. Multivariate Behavioral Research, Goffin, R. D., Gellatly, I. R., Paunonen, S. V., Jack- 35, 1Ð19. doi:10.1207/S15327906MBR3501_1 son, D. N., & Meyer, J. P. (1996). Criterion vali- Johnson, J., & LeBreton, J. M. (2004). History and dation of two approaches to performance ap- use of relative importance indices in organizational praisal: The behavioral observation scale and the research. Organizational Research Methods, 7, relative percentile method. Journal of Business 238Ð257. doi:10.1177/1094428104266510 and Psychology, 11, 23Ð33. doi:10.1007/ Jonason, P. K., & Krause, L. (2013). The emotional BF02278252 deficits associated with the Dark Triad traits: Cog- Greer, L. L., Jehn, K. A., & Mannix, E. A. (2008). nitive empathy, affective empathy, and alexithy- Conflict transformation: A longitudinal investiga- mia. Personality and Individual Differences, 55, tion of the relationships between different types of 532Ð537. doi:10.1016/j.paid.2013.04.027 intragroup conflict and the moderating role of con- Jonason, P. K., Slomski, S., & Partyka, J. (2012). The flict resolution. Small Group Research, 39, 278Ð dark triad at work: How toxic employees get their 302. doi:10.1177/1046496408317793 way. Personality and Individual Differences, 52, Hackman, J. R. (1987). The design of work teams. In 449Ð453. doi:10.1016/j.paid.2011.11.008 J. W. Lorsch (Ed.), Handbook of organizational Jonason, P. K., & Tost, J. (2010). I just cannot control behavior (pp. 315Ð342). Englewood Cliffs, NJ: myself: The dark triad and self-control. Personal- Prentice Hall. ity and Individual Differences, 49, 611Ð615. doi: Hackman, J. R. (2002). Leading teams: Setting the 10.1016/j.paid.2010.05.031 stage for great performances. Boston, MA: Har- Jonason, P. K., & Webster, G. D. (2010). The dirty dozen: A concise measure of the dark triad. Psy- vard Business School Press. chological Assessment, 22, 420Ð432. doi:10.1037/ Hayes, A. R. (2013). Introduction to mediation, mod- a0019265 eration, and conditional process analysis: A re- Jonason, P. K., & Webster, G. D. (2012). A protean gression-based approach. New York, NY: Guil- approach to social influence: Dark Triad personal- ford Press. ities and social influence tactics. Personality and Hodson, G., Hogg, S. M., & MacInnis, C. C. (2009). Individual Differences, 52, 521Ð526. doi:10.1016/ The role of “dark personalities” (Narcissism, Ma- j.paid.2011.11.023 chiavellianism, Psychopathy), big five personality Jonason, P. K., Webster, G. D., Schmitt, D. P., Li, factors and ideology in explaining prejudice. Jour- N. P., & Crysel, L. (2012). The antihero in popular nal of Research in Personality, 43, 686Ð690. doi: culture: Life history theory and the dark personal- 10.1016/j.jrp.2009.02.005 ity traits. Review of General Psychology, 16, 192Ð Humphrey, S. E., Hollenbeck, J. R., Meyer, C. J., & 199. doi:10.1037/a0027914 Ilgen, D. R. (2007). Trait configurations in self- Jones, D. N., & Paulhus, D. L. (2011). The role of managed teams: A conceptual examination of the impulsivity in the dark triad of personality. Per- use of seeding for maximizing and minimizing sonality and Individual Differences, 51, 679Ð682. trait variance in teams. Journal of Applied Psy- doi:10.1016/j.paid.2011.04.011 chology, 92, 885Ð892. doi:10.1037/0021-9010.92 Jordan, P. J., & Troth, A. C. (2004). Managing emo- .3.885 tions during team problem solving: Emotional intel- Jehn, K. A. (1995). A multimethod examination of ligence and conflict resolution. Human Performance, This document is copyrighted by the American Psychological Association or one of its allied publishers. the benefits and detriments of intragroup conflict. 17, 195Ð218. doi:10.1207/s15327043hup1702_4 This article is intended solely for the personal use of the individual user and is notAdministrative to be disseminated broadly. Science Quarterly, 40, 256Ð282. Judd, C. M., & Kenny, D. A. (2010). Data analysis in doi:10.2307/2393638 social psychology: Recent and recurring issues. In Jehn, K. A., Greer, L., Levine, S., & Szulanski, G. S. T. Fiske, D. T. Gilbert, & L. Gardner (Eds.), (2008). The effects of conflict types, dimensions, and Handbook of social psychology (5th ed., Vol. 1, emergent states on group outcomes. Group Decision pp. 115Ð139). Mahwah, NJ: Wiley. and Negotiation, 17, 465Ð495. doi:10.1007/s10726- Judge, T. A., & LePine, J. A. (2007). The bright and 008-9107-0 dark sides of personality: Implications for person- Johnson, D. W. (2003). Social interdependence: Inter- nel selection in individual and team contexts. In J. relationships among theory, research and practice. Langen-Fox, C. L. Cooper, & R. J. Klimoski American Psychologist, 58, 934Ð945. doi:10.1037/ (Eds.), Research companion to the dysfunctional 0003-066X.58.11.934 workplace: Management challenges and symptoms 172 O’NEILL AND ALLEN

(pp. 332Ð355). London, UK: Edward Elgar Pub- Organizational Behavior, 13, 209Ð252. doi: lishing Ltd. doi:10.4337/9781847207081.00028 10.1002/job.4030130303 Katz, D., & Kahn, R. L. (1978). The social psychol- McGrath, J. E. (1984). Groups: Interaction and per- ogy of organizations (2nd ed.). New York, NY: formance. Englewood Cliffs, NJ: Prentice Hall. Wiley and Sons. McHoskey, J. W., Worzel, W., & Szyarto, C. (1998). Korsgaard, M. A., Jeong, S. S., Mahony, D. M., & Machiavellianism and psychopathy. Journal of Pitariu, A. H. (2008). A multilevel view of intra- Personality and Social Psychology, 74, 192Ð210. group conflict. Journal of Management, 34, 1222Ð doi:10.1037/0022-3514.74.1.192 1252. doi:10.1177/0149206308325124 O’Neill, T. A., & Allen, N. A. (2011). Personality Kozlowski, S. W. J., & Bell, B. S. (2003). Work and the prediction of team performance. European groups and teams in organizations. In W. C. Bor- Journal of Personality, 25, 31Ð42. doi:10.1002/per man, D. R. Ilgen, & R. J. Klimoski (Eds.), Hand- .769 book of psychology (Vol. 12): Industrial and orga- O’Neill, T. A., & Allen, N. A. (2012). Team meeting nizational psychology (pp. 333Ð375). New York, attitudes: Conceptualization and investigation of a NY: Wiley. doi:10.1002/9781118133880 new construct. Small Group Research, 43, 186Ð .hop212017 210. doi:10.1177/1046496411426485 Kozlowski, S. W. J., & Klein, K. J. (2000). A mul- tilevel approach to theory and research in organi- O’Neill, T. A., Allen, N. A., & Hastings, S. E. zations: Contextual, temporal, and emergent pro- (2013). Examining the “pros” and “cons” of team cesses. In K. J. Klein & S. W. J. Kozlowski (Eds.), conflict: A team-level meta-analysis of task, rela- Multilevel theory, research, and methods in orga- tionship, and process conflict. Human Perfor- nizations: Foundations, extensions, and new direc- mance, 26, 236Ð260. doi:10.1080/08959285.2013 tions (pp. 3Ð90). San Francisco, CA: Jossey-Bass. .795573 Kuhn, T., & Poole, M. S. (2000). Do conflict manage- Paulhus, D. L., & Williams, K. M. (2002). The dark ment styles affect group decision making? Evidence triad of personality: Narcissism, Machiavellian- from a longitudinal field study. Human Communica- ism, and psychopathy. Journal of Research in Per- tion Research, 26, 558Ð590. doi:10.1111/j.1468- sonality, 36, 556Ð563. doi:10.1016/S0092- 2958.2000.tb00769.x 6566(02)00505-6 Lee, K., & Ashton, M. C. (2005). Psychopathy, Ma- Paunonen, S. V. (2002). Design and construction of chiavellianism, and Narcissism in the five-factor the Supernumerary Personality Inventory. Re- model and the HEXACO model of personality search Bulletin (Vol. 763). London, Ontario: Uni- structure. Personality and Individual Differences, versity of Western Ontario. 38, 1571Ð1582. doi:10.1016/j.paid.2004.09.016 Paunonen, S. V., Haddock, G., Forsterling, F., & LePine, J. A. (2003). Team adaptation and Keinonen, M. (2003). Broad versus narrow per- postchange performance: Effects of team compo- sonality measures and the prediction of behavior sition in terms of members’ cognitive ability and across cultures. European Journal of Personality, personality. Journal of Applied Psychology, 88, 17, 413Ð433. doi:10.1002/per.496 27Ð39. doi:10.1037/0021-9010.88.1.27 Paunonen, S. V., & Jackson, D. N. (2000). What is LePine, J. A., Buckman, B. R., Crawford, E. R., & beyond the Big Five? Plenty! Journal of Research Methot, J. R. (2011). A review of research on in Personality, 68, 821Ð835. doi:10.1111/1467- personality in teams: Accounting for pathways 6494.00117 spanning levels of theory and analysis. Human Peterson, R. S., & Behfar, K. J. (2003). The dynamic Resource Management Review, 21, 311Ð330. doi: relationship between performance feedback, trust, 10.1016/j.hrmr.2010.10.004 and conflict in groups: A longitudinal study. Or- LePine, J. A., Hollenbeck, J. R., Ilgen, D. R., & Hedlund, J. (1997). Effects of individual differ- ganizational Behaviour and Human Decision Pro- This document is copyrighted by the American Psychological Association or one of its allied publishers. ences on the performance of hierarchical decision- cesses, 92, 102Ð112. doi:10.1016/S0749- This article is intended solely for the personal use of the individual user and is notmaking to be disseminated broadly. teams: Much more than g. Journal of Ap- 5978(03)00090-6 plied Psychology, 82, 803Ð811. doi:10.1037/0021- Petrides, K. V., Vernon, P. A., Schermer, J. A., & 9010.82.5.803 Veselka, L. (2011). Trait emotional intelligence and Levenson, M. R., Kiehl, K., & Fitzpatrick, C. M. the dark triad traits of personality. Twin Research and (1995). Assessing psychopathic attributes in a non- Human Genetics, 14, 35Ð41. doi:10.1375/twin.14 institutionalised population. Journal of Personality .1.35 and Social Psychology, 68, 151Ð158. doi:10.1037/ Preacher, K. J., & Hayes, A. F. (2004). SPSS and 0022-3514.68.1.151 SAS procedures for estimating indirect effects in Lewicki, R. J., Weiss, S. E., & Lewin, D. (1992). simple mediation models. Behavior Research Models of conflict, negotiation, and third party Methods, Instruments & Computers, 36, 717Ð731. intervention: A review and synthesis. Journal of doi:10.3758/BF03206553 TEAM TASK CONFLICT RESOLUTION 173

Rahim, M. A. (1983). A measure of styles of han- personnel selection procedures (4th ed.). Bowling dling interpersonal conflict. Academy of Manage- Green, OH: SIOP. ment Journal, 26, 368Ð376. doi:10.2307/255985 Sundstrom, E., McIntyre, M., Halfhill, T., & Rich- Raskin, R., & Terry, H. (1988). A principle compo- ards, H. (2000). Work groups: From the Haw- nents analysis of the narcissistic personality inven- thorne studies to work teams of the 1990s and tory and further evidence of its construct validity. beyond. Group Dynamics: Theory, Research, and Journal of Personality and Social Psychology, 54, Practice, 4, 44Ð67. doi:10.1037//1089-2699.4 890Ð902. doi:10.1037/0022-3514.54.5.890 .1.44 Rauthmann, J. F. (2012). The Dark Triad and inter- Tjosvold, D. (1998). Cooperative and competitive personal perception: Similarities and differences in goal approach to conflict: Accomplishments and the social consequences of Narcissism, Machiavel- challenges. Applied Psychology, 47, 285Ð342. doi: lianism, and Psychopathy. Social Psychological 10.1111/j.1464-0597.1998.tb00025.x and Personality Science, 3, 487Ð496. doi:10.1177/ Wageman, R., & Gordon, F. M. (2005). As the twig 1948550611427608 is bent: How group values shape emergent task Robey, D., & Farrow, D. L. (1982). User involve- interdependence in groups. Organization Science, ment in information system development: A con- 16, 687Ð700. doi:10.1287/orsc.1050.0146 Wai, M., & Tiliopoulos, N. (2012). The affective and flict model and empirical test. Management Sci- cognitive empathic nature of the dark triad of ence, 28, 73Ð85. doi:10.1287/mnsc.28.1.73 personality. Personality and Individual Differ- Robey, D., Farrow, D. L., & Franz, C. R. (1989). Group ences, 52, 794Ð799. doi:10.1016/j.paid.2012.01 process and conflict in system development. Man- .008 agement Science, 35, 1172Ð1191. doi:10.1287/mnsc Wall, J. A., & Callister, R. R. (1995). Conflict and its .35.10.1172 management. Journal of Management, 21, 515Ð Ross, S. R., Bye, K., Wrobel, T. A., & Horton, R. S. 558. doi:10.1177/014920639502100306 (2008). Primary and secondary psychopathic char- Wegge, J., van Dick, R., Fisher, G. K., West, M. A., acteristics and the schedule for non-adaptive and & Dawson, J. F. (2006). A test of basic assump- adaptive personality (SNAP). Personality and In- tions of affective events theory (AET) in call cen- dividual Differences, 45, 249Ð254. doi:10.1016/j tre work. British Journal of Management, 17, 237Ð .paid.2008.04.007 254. doi:10.1111/j.1467-8551.2006.00489.x Rubin, J. Z., Pruitt, D. G., & Kim, S. H. (1994). Weiss, H. M., & Cropanzano, R. (1996). Affective Social conflict: Escalation, stalemate, and settle- events theory: A theoretical discussion of the ment (2nd ed.). New York, NY: McGraw-Hill. structure, causes and consequences of affective Shrout, P. E., & Fleiss, J. L. (1979). Intraclass cor- experiences at work. Research in Organizational relations: Uses in assessing rater reliability. Psy- Behavior, 18, 1Ð74. chological Bulletin, 86, 420Ð428. doi:10.1037/ 0033-2909.86.2.420 Received May 7, 2013 Society for Industrial-Organizational Psychology. Revision received November 14, 2013 (2003). Principles for the validation and use of Accepted November 20, 2013 Ⅲ This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. Group Dynamics: Theory, Research, and Practice © 2014 American Psychological Association 2014, Vol. 18, No. 2, 174–188 1089-2699/14/$12.00 DOI: 10.1037/a0035797 The Impact of Group Norms and Behavioral Congruence on the Internalization of an Illegal Downloading Behavior

Sophie Sansfaçon and Catherine E. Amiot Université du Québec a` Montréal

This experimental study (n ϭ 115) investigates how group norms and individuals’ congruence with these norms predicted internalization (i.e., self-determination) of an illegal behavior. We manipulated ingroup norms in favor of versus against illegal downloading of software, and assessed participants’ behavioral intentions, their moti- vations for emitting this specific behavior, and how they subjectively experienced the experimental situation (compartmentalizing their illegal behavior; feeling conflicted). Participants were more likely to endorse the behavior that was normative in their ingroup. Also, participants were more self-determined to engage in the behavior when they intended to behave congruently with a norm against illegal downloading, and when they intended to behave incongruently with a norm in favor of illegal download- ing. Participants who did not intend to engage in illegal downloading felt less com- partmentalized and conflicted than participants who intended to engage in illegal downloading. Results are discussed in light of self-determination theory (Deci & Ryan, 2000) and social identity theory (Tajfel & Turner, 1986).

Keywords: dissent, group norms, illegal behavior, norm and behavior congruence, self- determination

What drives people to behave illegally, such software companies are already making a lot of as by stealing or cheating? One could say that money. Still, we can wonder about the variables the reasons why people act illegally are instru- that facilitate the internalization of such illegal mental, and serve to access valued resources behavior within the self. Previous studies have (Lee & Allen, 2002) or gain social approval demonstrated that group norms influence not (Baumeister & Vohs, 2004). For example, stu- only the emission of negative behaviors (Frone dents may have an external motivation to ille- & Brown, 2010; Kura, Shamsudin, & Chauhan, gally download a computer software from the 2013; Robinson & O’leary-Kelly, 1998), but Internet because they need that program to suc- also the internalization of those behaviors ceed in their studies. Other students may engage (Amiot, Sansfaçon, & Louis, 2013; Amiot, in the illegal behavior in a more internalized Sansfaçon, Louis & Yelle, 2012). When per- manner, because they feel that it is right for ceiving one’s group norm as encouraging a be- students not to pay for computer software be- havior, it is possible that those behaviors be- cause they don’t have a salary yet, and that the come internalized by individual group members, such that they are part of their self (Haslam & Reicher, 2007) and are engaged in

This document is copyrighted by the American Psychological Association or one of its allied publishers. out of pleasure or because they align with one’s This article is intended solely for the personal use of the individual user and isSophie not to be disseminated broadly. Sansfaçon and Catherine E. Amiot, Département values (Amiot, Sansfaçon, & Louis, 2013; in de Psychologie, Université du Québec a` Montréal, Mon- press). In the current research, we aim to test, in tréal, Québec, Canada. an experimental context, whether a situationally This research was facilitated thanks to grants from the Social Sciences and Humanities Research Council of Canada salient norm influences the immediate internal- (SSHRC) and from the Fond pour la recherche en santé– ization of the normative behaviors. Québec (FRSQ) to Catherine E. Amiot. In fact, past studies have showed that when Correspondence concerning this article should be ad- an ingroup is important to the individual, the dressed to Sophie Sansfaçon, Département de psychologie, norms of this group can become internalized Université du Québec a` Montréal, C.P. 8888, Succ. Centre- Ville, Montréal, PQ, H3C 3P8. E-mail: cpf.sansfaçon@ and impact significantly on the behaviors of gmail.com individual group members (e.g., Fielding, Mc-

174 NORM INTERNALISATION AND COMPARTMENTALISATION 175

Donald, & Louis, 2008; Terry & Hogg, 1996). social identity perspective, when an individual Relying on self-determination theory (Deci & identifies with a group, the group identity be- Ryan, 2000) and intergroup theories based on comes a part of the self, and so do the norms of the social identity approach (Tajfel & Turner, this group (Turner, 1991). Whether norms are 1986), the first goal of the current study is to followed or not also depends on the social con- investigate the internalization of a particular text: when a group membership is salient in a illegal behavior, namely, the illegal download- social context and this membership is important ing of computer software. This behavior is pri- to the person’s self-definition, group norms will vate, anonymous, and relatively common (Bel- become activated and will guide one’s behavior gian Anti-Piracy Federation, 2009); it was (Smith & Louis, 2008; Terry & Hogg, 1996). considered relevant to study in relation to social This internalization is also facilitated by intra- identification and internalization processes group processes, including communication that given that social influence and norms operate takes place within the ingroup (Kerr & Kauf- strongly under conditions of isolation and ano- man-Gilliland, 1994) and consensualization nymity (Spears, Postmes, Lea, & Wolbert, processes (Turner, 1991). 2002). We expected this illegal behavior to be For example, at a hockey game, if a hockey internalized when social norms and cues are fan identifies strongly as a fan of the team and salient. perceives that his or her team group is encour- Our second goal is to investigate how partic- aging him/her in engaging in a harmful behav- ipants subjectively feel when they behave in iors against fans of other team, s/he will tend to line with group norms or not. To this aim, we engage in the normative behavior and internal- assessed how engaging in an illegal behavior is ize this behavior to a greater extent (Amiot, compartmentalized and to what extent it can Sansfaçon, & Louis, in press). Importantly, the cause conflict within the self. We also explore intergroup perspectives propose that the inter- how participants subjectively perceive how dif- nalization process should take place regardless ferent sources of influence guided their behavior of whether the normative group behaviors are in the experimental setting. Doing so opens the prosocial or potentially harmful (Turner, 1991) door to a new subjective perspective on group – such as whether behaviors are illegal or not. norms that departs from traditional norms re- search, which has typically looked at objective Self-Determination Theory (e.g., manipulated) sources of normative influ- ence that operate outside of individuals’ aware- Previous studies from the organizational lit- ness (Vorauer & Miller, 1997). erature have confirmed that motivations to en- gage in illegal behaviors in the workplace can Social Norms Promote Internalization be diverse, from thrill seeking to role modeling, and to feelings that such behaviors allow to Social norms are useful guides for what is restore justice (Robinson & Greenberg, 1998). considered appropriate behavior in group set- This suggests that negative behavior can be tings. Norms represent what is done and pre- emitted out of different motives. Herein we scribed by the other group members (Jacobson, focus on motives that range from being driven Mortensen, & Cialdini, 2011). Despite being by external sources, to behaviors that are fol- potentially coercive (Turner, 1991), norms can lowed for more internal reasons (Amiot, Sans- This document is copyrighted by the American Psychological Association or one of its allied publishers. also be internalized by group members and façon, & Louis, 2013). To concretely capture This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. come to be followed freely and out of choice the possible motivations for people’s behaviors (Amiot, Sansfaçon, & Louis, 2013). Internaliza- and the extent to which these behaviors— tion refers to the process by which an individual including normative behaviors—are internal- is taking in the values and behaviors of a group ized, the present research relies on self- and integrating them within their sense of self, determination theory (SDT; Deci & Ryan, by making them their own (Kelman, 1961). 2000). SDT is a motivational theory that has This process operates beyond acting for per- been applied mostly to understanding the moti- sonal gain, or to avoid the negative conse- vations that underlie constructive social behav- quences that derive from violating the norm iors (e.g., volunteering; Millette & Gagné, (Kerr, Garst, Kiehle, & Harris, 1997). From a 2008). SDT proposes that when psychological 176 SANSFAÇON AND AMIOT

needs (autonomy, relatedness, competence) are cial manner (Deci & Ryan, 2000). Empirically, satisfied by the social environment, behaviors studies have found that abusive supervision in tend to be emitted out of self-determined mo- the workplace context is linked to organiza- tives (internalized), and individuals report tional deviance, and that this link is mediated by higher well-being (Lynch, La Guardia, & Ryan, need satisfaction (Lian, Ferris, & Brown, 2012). 2009). For SDT theorists, motivations vary In the current research, we expected, based on along a continuum ranging from reasons that are SDT, that harmful behaviors—such as illegal not internalized and are not considered as part behaviors—should hence trigger compartmen- of the self (non–self-determined), to reasons talization and intraindividual conflict and un- that are considered internalized and part of the ease given that such behaviors clash with the self (self-determined). More precisely, the non– person’s true goals and aims. self-determined motivations include amotiva- Prior studies confirmed that prosocial (rather tion, external regulation, and introjected regula- than harmful) norms are easier to internalize tion. Amotivation represents a total absence of and that group norms and individuals’ behav- motivation. External regulation is present when ioral congruence with the norms do have an an individual acts out of external contingencies, impact on the internalization of a specific nor- such as to obtain rewards or to flee punishment. mative behavior (Amiot et al., 2012). These Introjected regulation occurs when the individ- studies tested competing hypotheses based on ual is acting out of internal pressure, so as not to SIT versus SDT about the possibility of group feel bad or guilty. norms to be internalized and endorsed out of The more internalized and self-determined self-determination. The findings tended to sup- motivations include identified regulation, inte- port SDT’s perspective: When an important in- grated regulation, and intrinsic motivation. group was encouraging a pro-social behavior, Identified regulation is present when individuals such as distributing resources equally between feel that engaging in the behavior is important groups (prosocial parity norm), and individuals for them and allows them to reach valued goals. were behaving congruently with this norm, this Integrated regulation is present when individu- behavior was more strongly internalized and als act because the behavior is coherent with self-determined (Amiot et al., 2012). The re- their personal objectives and values. Finally, verse pattern was observed for harmful norms in intrinsic motivation occurs when individuals favor of discrimination: When one’s ingroup behave out of pleasure and for the satisfaction encouraged discrimination, it is individuals who of engaging in the activity itself. behaved incongruently with this norm (by en- These motivations have been explored in var- gaging instead in parity) who were more self- ious contexts (Vallerand, 1997), and are begin- determined. These results suggest that ingroup ning to be applied to group settings (e.g., Amiot norms do have an impact on the internalization & Sansfaçon, 2011). In the current research, we of the behavior, and that engaging in a poten- employ the SDT framework not only to capture tially harmful social behavior tends to be more internalization, but also because this framework difficult to internalize in the self, even if this proposes that socially harmful behaviors will be behavior is proscribed by ingroup norms. As more difficult to internalize in the self (Deci & well, and in line with Packer (2008), partici- Ryan, 2000). Whereas the intergroup ap- pants who went against a socially harmful norm proaches do not make distinctions between so- in favor of discrimination by engaging in a This document is copyrighted by the American Psychological Association or one of its allied publishers. cially harmful and beneficial group norms with pro-social behavior instead may have drawn This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. respect to their capacity to become internalized their motivation from particularly internalized in the self (Turner, 1991), SDT clearly puts values. Risking to go against the ingroup norms limits to this capacity for internalization by sug- can be costly for the individual with regard of gesting that potentially harmful social behaviors his or her membership within a group, which are fuelled by a lack of autonomy and self- suggests that deviant behaviors are engaged in determination, and that such behaviors occur out of deeply held values and beliefs. when one’s basic psychological needs have The current study will test whether the been thwarted (Deci & Ryan, 2000). In fact, same results are observable when we manip- SDT is a humanistic theory that assumes that ulate norms that concern an illegal behavior, humans strive to act effectively and in a proso- namely the illegal downloading and use of NORM INTERNALISATION AND COMPARTMENTALISATION 177

computer software. This type of behavior is Compartmentalization considered to be socially harmful from a legal and a moral point of view. The law considers Compartmentalization is a term used when pirated software downloading as theft (Copy- referring to distinct parts of the self that are kept right Board of Canada, 2009). Downloading separated from each other (Ryan & Deci, 2003). software illegally does not recognize the work Compartmentalization of behaviors occurs conducted by software designers and com- when individuals behave in a way that fits with puter companies. Moreover, not paying for a the social identity and the social norms that are product does not encourage software devel- salient in one specific social context, but with- opment and increases the costs of software for out linking back this behavior to other life con- buyers (BAF, 2009). We expected that this texts or to their own personal values. For exam- type of behavior will be possible to change ple, a prisoner can act in an illegal manner when s/he is in the prison context and his or her via the manipulation of situational norms prisoner identity is salient, but can follow laws given that it is an anonymous behavior and regulations when s/he is in another group, (Spears et al., 2002) that is considered quite her/his family for instance (Tripp, 2009). In this benign in terms of its consequences (Aguiar case, the illegal behavior does not represent the & Martens, 2013). Furthermore, different individual as a whole, but represents the person studies conducted among university studies only in one particular context. Compartmental- report diverging statistics regarding the fre- ization should be experienced to a greater extent quency of illegal downloading behaviors: For when individuals intend to engage in illegal example, whereas a Belgian survey estab- behaviors, because these kinds of behaviors are lished that 90% of students engage in the generally disapproved of by society, and need to illegal downloading of software (BAF, 2009), be kept in a distinct and hermetic part of the self a French survey revealed that about 40% of so as not to contaminate their entire sense of self students engage in such behavior (JDN, (Ryan & Deci, 2003). We expect that individu- 2004). This divergence suggests that norms als who intend to act illegally (who intend to regarding the prevalence of illegal download- behave congruently with a downloading norm ing are somewhat vague, and that the norms or incongruently with a no-downloading norm) manipulated in this context may hence be will report higher compartmentalization than relied on more strongly as a means to resolve those who intend to act legally. this ambiguity and vagueness (Smith, Hogg, Martin, & Terry, 2007). This allows us to Intraindividual Conflict expect that the norm manipulation used in the current study will have a potent impact on Similarly, conflict arises when an individual participants’ downloading behaviors and on feels that there is a discrepancy between differ- their internalization of these behaviors. ent parts of him or herself, such as between diverging norms or values (Benet-Martínez & Haritatos, 2005). Conflict may arise when the Subjective Processes individual acts in a way that does not represent his or her own personal values and attitudes, or The experiment will also explore three new when following group norms that are pro- This document is copyrighted by the American Psychological Association or one of its allied publishers. variables that capture individual group mem- scribed by another of his or her social groups. This article is intended solely for the personal use of the individualbers’ user and is not to be disseminated broadly. subjective experiences of the norm inter- Because acting illegally is considered to be a nalization process. More specifically, we inves- socially harmful behavior and also based on tigate how individual group members SDT postulates, we hypothesize that intraindi- compartmentalize their illegal behaviors within vidual conflict will be positively related to com- themselves, the level of internal pressure or partmentalization. Moreover, we expect that in- conflict they subjectively feel when they intend dividuals who intend to act illegally (who to engage or not in a normative illegal behavior, intend to behave congruently with a download- and which specific sources of normative influ- ing norm or incongruently with a no-download- ence they believe are influencing them when ing norm) will feel more conflict than those who intending to engage (or not) in illegal behaviors. act legally. 178 SANSFAÇON AND AMIOT

Perceived Influence experiences in the norm internalization process, and more specifically their feelings of compart- According to some researchers, the social mentalization and intraindividual conflict, along influence process is an implicit, unconscious with the sources of influence that they perceived mechanism (Vorauer & Miller, 1997), which as guiding their behavior in the experiment. By also implies that the subjective perception of doing so, the present study goes a step further in norms can be biased (Baer, Stacy & Larimer, a line of research bridging SDT and intergroup 1991). Still, we can wonder how individuals theories. By testing these hypotheses with a perceive what has influenced them in a specific potentially malleable illegal behavior, and in- situation, such as group norms or personal val- vestigating these additional subjective pro- ues (Van Bavel, Packer, & Cunningham, 2008). cesses, this study will allow a broader look at This study will hence explore how participants the consequences of engaging in and following subjectively perceive different possible sources socially “harmful” norms. Moreover, whereas of influence—namely the situation, their own other studies have explored the internalization personal values and sense of morality, and in- of negative social norms within correlational group norms—and how they consider that these design in real life contexts (e.g., discrimination sources guided their behavior in the experimen- among hockey fans: Amiot, Sansfaçon, & tal context. Exploring these subjective percep- Louis, in press), the present study will test tions is important given that multiple sources of whether a situational group norm will influence influence have the potential to guide people’s the immediate internalization of such behav- behaviors in a specific context (i.e., attitudes iors, in an experimental context. and social norm; Ajzen, 2001) and that we wish To reach these goals, we present an experi- to capture what influences people’s illegal ac- mental study in which we manipulated group tions, from their own point of view. norms and assessed participants’ congruence The perception of what influenced one’s own with these norms. In line with past studies on behavior is probably a process that relates to the norm internalization (Amiot et al., 2012), we compartmentalization process: A person who decided to manipulate the descriptive norm here compartmentalizes his or her behavior associ- (individual’s perception of what other group ates and restricts this behavior to a particular members do), instead of injunctive norms (per- life context. In this sense, the more a person ception of what group members encouraged in- compartmentalizes his or her behavior, the more dividual to do; Reno, Cialdini, & Källgren, likely we expect that he or she will perceive that 1993: Smith & Louis, 2008; Smith et al., 2008). it is a specific group or the situation itself (rather Based on prior research, the following hy- than his or her own morality or personal values) potheses were made: (H1) Ingroup norms will that influenced the behavior. We hence expect influence participants’ behavioral intentions: that the different sources of influence will relate Participants exposed to a prodownloading in- differently with compartmentalization. group norm will themselves report stronger in- tentions to engage in illegal prodownloading in The Present Research comparison with participants exposed to a no- downloading ingroup norm. In line with prior Following past studies (Amiot et al., 2012), studies (Amiot et al., 2012), H2a predicts that the first objective of this research is to test participants who act congruently with the no- This document is copyrighted by the American Psychological Association or one of its allied publishers. whether a “prodownloading norm” (i.e., a norm downloading norm (and hence who do not in- This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. in favor of illegal downloading) versus a “no- tend to behave illegally themselves) will be downloading norm” (i.e., a norm that is not more self-determined than those who behave against illegal downloading) influences partici- incongruently with this norm. Among partici- pants’ behavioral intentions to engage in illegal pants in the prodownloading condition, those downloading behaviors. The second objective is who intend to behave incongruently with the to investigate how individuals’ congruence/ norm should report higher self-determination incongruence with these norms influences their than those who intend to behave in line with this self-determination (as a proxy for internaliza- norm (H2b). H3 and H4 predict that participants tion) for engaging in this behavior. A third will report higher compartmentalization and objective is to capture participants’ subjective more conflict when they intend to act illegally, NORM INTERNALISATION AND COMPARTMENTALISATION 179

that is, both when they plan to act congruently when asking students the following question: with a prodownloading norm or incongruently Do you personally illegally download and use with a no-downloading norm. Based on a sub- computer software? For participants in the jective approach to group norms, H5 predicts downloading norm condition, the graphs that compartmentalization will be positively re- showed that 76% of students said that they were lated to intraindividual conflict. We also antic- engaging in illegal downloading while 24% ipate that the more a person compartmentalizes were not. For participants in the no-download- his or her behavior, the more likely we expect ing norm condition, these percentages were re- that he or she will perceive that it is a specific versed. group or the situation itself (rather than his or To strengthen the norm manipulation further, her own morality or personal values) that influ- participants read a series of five opinion state- enced his or her behavior (H6). Finally, given ments about the target issue of computer soft- the lack of prior studies on how individuals ware downloading, ostensibly from five stu- subjectively perceive different sources of influ- dents from their program who had participated ence, we will explore how these sources of in the survey before them. The goal with this influence are perceived across experimental manipulation was to provide participants with conditions. further ideological justifications for endorsing the norm (Amiot & Bourhis, 2003). These state- Method ments indicated that four of five students from their program either were engaging in the be- Participants and Design havior themselves (or not, depending on the experimental condition), and that one of five Participants were 123 students enrolled at a was neutral. As a manipulation check, partici- French-speaking university in Montréal. The pants were asked whether the graphs and com- study employed a between-participants design ments showed that the majority of the students where we manipulated social identity salience at their university were engaging in the behavior (personal vs. student identity; Haslam, Oakes, that involves illegally downloading computer Reynolds, & Turner, 1999)1 and ingroup norm- software or not. s(prodownloading vs. no-downloading of com- puter software). No participant had suspicions 1 about what the goal of the study was, but 8 Identity Salience Manipulation. In line with past re- search (Haslam et al., 1999) and to test whether identity participants did not understand the ingroup salience accentuates the internalization of social norms norm manipulation correctly and were excluded (Smith & Louis, 2008), a social identity manipulation was from the sample. The final sample (N ϭ 115) first presented. This identity salience manipulation was included 90 females and 24 males (1 participant based on Haslam et al. (1999): half of the participants were did not report their gender), who were on aver- primed with their personal identity, and the other half with ϭ their social identity as a student of their university. Partic- age 22.50 years old (SD 4.74). ipants in the social identity condition were asked to answer four questions pertaining to their student identity. The same Procedure questions were asked to participants in the personal identity condition, but concerning themselves as individuals. Fol- Participants were presented with information lowing the manipulation, two manipulation check questions about the general context of illegal software were asked: To what extend it is easy for you to describe This document is copyrighted by the American Psychological Association or one of its allied publishers. yourself as a UQAM student? (high identification condition: downloading among university students. They ϭ ϭ ϭ

This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. M 4.25; ST 1.50; low identification condition: M were told that studying at University requires 4.55; SD ϭ 1.35); To what extent your UQAM student buying and downloading costly computer pro- identity is important for you? (high identification condition: ϭ ϭ ϭ ϭ grams, and that some of the students may ille- M 5.09; ST 1.35; low identification: M 5.20; SD 1.26), and answered ona1to7Likert scale, ranging from gally download those programs. In line with Not at all to Extremely. To assess the impact of the identity past research (Smith & Louis, 2008), partici- salience manipulation, ANOVAs were conducted on the pants were then informed that a survey had been two manipulation check questions. No significant difference conducted among students of their University emerged between the two conditions on the two questions. The identity salience manipulation did not show any main (UQAM) to gather their opinions and behaviors or interactive effects with the norms manipulation on self- regarding illegal downloading. Bar graphs were determination index, t(112) ϭ 1.127, p ϭ .262; t(112 ϭ presented and illustrated the results observed .497, p ϭ .633). 180 SANSFAÇON AND AMIOT

Measures not represent who I am as a person; My downloading behaviors are only emitted in Behavioral intentions. To assess partici- the current context; My downloading behav- pants’ behavioral intentions, participants iors are specific to a particular context of my were asked whether they would illegally life; My downloading behaviors represent download and use software if their program of only a small part of who I am as a person My study requires it (yes/no). Answers to this ; downloading behaviors are associated with a question were coded to indicate that partici- ␣ϭ pants’ behavior was either congruent or in- particular situation in my life; .84). congruent with their ingroup norm. Intraindividual conflict. To assess the Motivation for emitting the behavior conflict between the behavior participants in- displayed. Participants were then asked to tend to emit and their personal values, five think of the reasons why they intend to en- items were adapted from the Benet-Martínez gage in this behavior. They were then pro- and Haritatos’ (2005) cultural conflict scale vided with 24 items based on previously val- (e.g., My behaviors pertaining to computer idated SDT scales (Guay, Mageau, & downloading are in conflict with my personal Vallerand, 2003). These items measured amo- values; I feel torn between my downloading tivation (I am not really sure that this behav- behaviors and my personal values; ␣ϭ.86). ior is really worth it; ␣ϭ.74), external Perceived influence. To assess the per- regulation (because its allows me to do what ceived influence of different possible sources, is expected from me; ␣ϭ.71), introjected we asked participants to what extent five regulation (because I would have felt sources influenced their downloading behav- ashamed if I had behave in another way; ␣ϭ ior in the experiment: The situation, their .90), identified regulation (because I believe personal values, their morality, the software that it is important to behave in that way; ␣ϭ companies, and other UQAM students. .65), integrated regulation (because this man- ner of behaving is in line with my values; ␣ϭ .92), and intrinsic motivation (because I de- Results rived satisfaction from behaving this way; ␣ϭ.93). Each of the six motivation types in- Manipulations Checks: Norm and cluded four items. An index of self-determined Behaviors Displayed motivation was created by using the following ϫ ϩ ϫ formula: (3 intrinsic motivation 2 inte- To assess the impact of the ingroup norm ϩ ϫ Ϫ grated regulation 1 identified regulation) manipulation on participants’ behavioral inten- (1ϫ introjected regulation ϩ 2ϫ external regu- ϩ ϫ tions, chi-square analyses were conducted. A lation 3 amotivation; see Vallerand & main effect of the ingroup norm emerged, Bissonnette, 1992). It is common for SDT ␹2(df ϭ 1, n ϭ 114) ϭ 10.10, p ϭ .001: Par- researchers to compute an overall self- ticipants exposed to the prodownloading norm determination index (Ryan & Connell, 1989; displayed more downloading intentions in com- Vallerand, 1997; Vallerand & Bissonnette, 1992). Greater scores on this index represent parison with participants exposed to the no- higher self-determination motives relative to downloading norm. More specifically, among This document is copyrighted by the American Psychological Association or one of its allied publishers. non–self-determination motives. For this participants whose ingroup endorses the norm This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. ϭ measure and the following ones, 1 (not at all) in favor of illegal downloading, 79% (n 44) to7(extremely) Likert scales were used (the displayed downloading intentions and 21% ϭ index ranges from Ϫ36 up to 36). (n 12) did not display downloading inten- Compartmentalization. Based on Ryan tions. Among participants whose ingroup en- and Deci’s (2003) conceptualization of com- dorsed a no-downloading norm, 50% (n ϭ 29) partmentalization, a 5-item scale was created displayed downloading intentions and 50% to measure the extent to which participants’ (n ϭ 29) did not display downloading inten- illegal downloading behavior in the context of tions. These results confirm H1: Ingroup norms this study was compartmentalized in their significantly influenced participants’ behavioral sense of self (My downloading behaviors do intentions. NORM INTERNALISATION AND COMPARTMENTALISATION 181

Descriptive Statistics and Correlations 2 (ingroup norm: prodownloading/no-down- Among All Variables loading norm) ϫ 2 (congruence: participant’s behavioral intentions are congruent/incongru- Table 1 displays the correlations between ent with the ingroup norm) ANOVA was con- the variables for participants in all conditions. ducted on the self-determination index. The As could be expected because self-determina- main effect of the ingroup norm was signifi- tion reflects greater internalization and engag- cant, F(1, 112) ϭ 10.67, p Ͻ .001, ␩2 ϭ .09: ing in behaviors because they reflect one’s Participants in the no-downloading norm con- choices and values, the self-determination in- dition were more self-determined (M ϭ 12. dex correlates positively with perceiving that 04; SD ϭ 10.90) than participants in the what influenced one in the experiment are prodownloading norm condition (M ϭ 3.80; their personal values (r ϭ .44, p Ͻ .01) and SD ϭ 9.16). The main effect of congruence morality (r ϭ .41, p Ͻ .01), but negatively was not significant, F(1, 112) ϭ 0.01, p ϭ with perception of situational influence (r ϭ .923, ␩2 ϭ .000. A significant interaction was Ϫ Ͻ .20, p .05). Moreover, supporting H5, observed between ingroup norm and congru- compartmentalization correlates with intrain- ence (Figure 1): F(1, 112) ϭ 17.295, p Ͻ ϭ Ͻ dividual conflict (r .40, p .01) and with .001, ␩2 ϭ .14; participants whose behavioral ϭ Ͻ situational influence (r .41, p .01), which intentions were congruent with the no- is also coherent with the definition of com- downloading norm condition (and who did partmentalization as taking place when dis- not engage in illegal downloading) reported tinct behaviors are kept separated from each greater self-determination (M ϭ 8.54; SD ϭ other in the self and associated only with 7.81) than participants whose behavioral in- specific situations (Ryan & Deci, 2003). tentions were incongruent with this norm (and Compartmentalization also correlates nega- who engaged in illegal downloading instead; tively with the perception that morality influ- ϭ ϭ ϭ ϭ ϭϪ Ͻ M 1.69; SD 9.32; F(1, 57) 8.86, p enced one’s behavior (r .22, p .05), .004, ␩2 ϭ .14). In contrast, participants which is also coherent with the concept and whose behavioral intentions were incongruent with H6. Finally, and in line with social iden- with a prodownloading norm (such that they tity theory, perceiving the students as influ- did not engage in illegal downloading) were ential is related to identifying with the student more self-determined (M ϭ 3.13; SD ϭ 9.80) ingroup (r ϭ .21, p Ͻ .05). than participants whose behavioral intentions Motivation for Emitting the Behavior were congruent with this norm (and who hence engaged in illegal downloading; M ϭ To assess the impact of the ingroup norm Ϫ3.40; SD ϭ 4.88, F(1, 55) ϭ 10.02, p ϭ and the congruence of participants’ intentions .003, ␩2 ϭ .16). These findings support H2a relative to this norm on self-determination, a and H2b.

Table 1 Descriptive Statistics and Correlations Among All Variables

This document is copyrighted by the American Psychological Association or one of its allied publishers. Variables MSD 1234567 This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. 1. Index 7.71 8.32 ءءCompartmentalization 3.49 1.76 Ϫ.40 .2 ءء40. ءءIntraindividual conflict 2.69 1.36 Ϫ.50 .3 17. ءء41. ءInfluence: Situation 5.15 1.57 Ϫ.20 .4 Ϫ.15 Ϫ.17 Ϫ.08 ءءInfluence: Personal values 4.85 1.78 .44 .5 ءءءϪ.08 .00 .87 ءϪ.22 ءءInfluence: Morality 4.56 2.03 .41 .6 12. 12. ءInfluence: Software companies 3.23 2.01 .11 Ϫ.05 Ϫ.12 .23 .7 ءءInfluence: Students 2.52 1.67 Ϫ.04 .07 .11 .16 .14 .42 .8 ءStudent identification 4.78 1.17 Ϫ.06 .15 .16 Ϫ.02 Ϫ.01 .09 .21 .9 Note. M ϭ mean; SD ϭ standard deviation. .p Ͻ .001 ءءء .p Ͻ .01 ءء .p Ͻ .05 ء 182 SANSFAÇON AND AMIOT

Figure 1. Self-determination as a function of ingroup norm and congruence.

Compartmentalization were not significant. However, a significant in- teraction was observed between ingroup norm To assess the impact of the ingroup norm and and congruence, F(1, 108) ϭ 16,03, p Ͻ .001, the congruence of participants’ intentions rela- ␩2 ϭ .129. Interpretation of this interaction (see tive to this norm on their perceptions of com- Figure 2) revealed that when the ingroup norm partmentalization, a 2 (ingroup norm: prodown- was a no-downloading one, participants whose loading/no-downloading) ϫ 2 (congruence: behavioral intentions were congruent with this congruent/incongruent) ANOVA was con- norm (those who did not engage in the illegal ducted on the measure of compartmentalization. behavior) reported lower compartmentalization The main effects of the ingroup norm, F(1, (M ϭ 2.44; SD ϭ 1.25) than participants whose 108) ϭ 0.92 p ϭ .338, ␩2 ϭ .008, and congru- behavioral intentions were incongruent with ence, F(1, 108) ϭ 0.01, p ϭ .948, ␩2 ϭ .000, this norm (and who did engage in illegal down- This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Figure 2. Compartmentalization as a function of ingroup norm and congruence. NORM INTERNALISATION AND COMPARTMENTALISATION 183

loading; M ϭ 3.81; SD ϭ 1.92; F(1, 55) ϭ (those who did not engage in the illegal behav- 10,13, p ϭ .002, ␩2 ϭ .156). In contrast, when ior) reported lower conflict (M ϭ 1.95; SD ϭ the ingroup norm was a prodownloading one, 1.04) than participants whose behavioral inten- participants whose behavioral intentions were tions were incongruent with this norm (and who congruent with this norm (and who engaged in did engage in the illegal behavior; M ϭ 3.18; the illegal behavior) reported greater compart- SD ϭ 1.68; F(1, 54) ϭ 10.66, p ϭ .002, ␩2 ϭ mentalization (M ϭ 4.16; SD ϭ 1.62) than .165). In contrast, when the ingroup norm was a participants whose behavioral intentions were prodownloading one, participants whose behav- incongruent with this norm (and who did not ioral intentions were congruent with this norm engage in illegal downloading; M ϭ 2.75; SD ϭ (and who did engage in the illegal behavior) 1.59, F(1, 55) ϭ 6,72, p ϭ .012, ␩2 ϭ .112). reported greater conflict (M ϭ 3.01; SD ϭ 1.22) These findings support H3. than participants whose behavioral intentions were incongruent with this norm (and who did Intraindividual Conflict not engage in illegal behavior; M ϭ 2.09; SD ϭ 0.67, F(1, 53) ϭ 5.64, p ϭ .021, ␩2 ϭ .096), To assess the impact of the ingroup norm and supporting H4. the congruence of participants’ intentions rela- tive to this norm on their feelings of intraindi- Perceived Influence of Different Sources vidual conflict, a 2 (ingroup norm: prodown- loading/no-downloading) ϫ 2 (congruence: To assess the impact of group norms and the congruent/incongruent) ANOVA was con- congruence of participants’ intentions relative ducted on the conflict variable. The main effects to the ingroup norm on their perceptions of the of the ingroup norm, F(1, 107) ϭ .01 p ϭ .942, influence of different sources, we conducted a 2 ␩2 ϭ .000, and congruence, F(1, 107) ϭ .31, (ingroup norm: prodownloading/no-download- p ϭ .577, ␩2 ϭ .003, were not significant. ing) ϫ 2 (congruence: congruent/incongruent) However, a significant interaction was observed ANOVA with the five sources of normative between ingroup norm and congruence, F(1, influence as a repeated-measures factor. A sig- 107) ϭ 15.07, p Ͻ .001, ␩2 ϭ .12. Interpreta- nificant main effect of the ingroup norm tion of this interaction (see Figure 3) revealed emerged, F(1, 104) ϭ 5.02, p ϭ .027, ␩2 ϭ that participants whose behavioral intentions .046, showing that participants generally per- were congruent with the no-downloading norm ceived these sources of influence to be stronger This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Figure 3. Intraindividual conflict as a function of ingroup norm and congruence. 184 SANSFAÇON AND AMIOT

when the ingroup norm was not supporting il- by congruence interactions were not significant, legal downloading (M ϭ 4.26; SD ϭ 0.17) than Fs ϭ .75 to 1.34, ps ϭ .25 to .39, ␩2 ϭ .007 to when this norm was supporting illegal down- .01. For the 3 other sources (i.e., personal val- loading (M ϭ 3.77; SD ϭ 0.17). A significant ues, morality, situation), significant norm by main effect for the source of influence emerged, congruence interactions emerged (Figure 4; F(5, 100) ϭ 48.38, p Ͻ .001, ␩2 ϭ .708. Fs ϭ 10.117 to 13.28, ps ϭ .000 to .026, ␩2 ϭ Planned contrasts conducted to locate the dif- .04 to .11). Participants in the prodownloading ferences among the sources showed that partic- norm condition who intended to behave congru- ipants perceived the sources of influence per- ently with this norm perceived that it was the taining to the situation (M ϭ 5.04; SD ϭ 0.16) situation (M ϭ 5.46; SD ϭ 1.50) that guided and to their personal values (M ϭ 4.97; SD ϭ them more than their personal values (M ϭ 0.17) to be stronger than their morality (M ϭ 3.93; SD ϭ 1.79; F(1, 40) ϭ 22.98, p Ͻ .001, 4.77; SD ϭ 0.19; F(1, 10) ϭ 3.49, p ϭ .029, ␩2 ϭ .37), followed by their morality (M ϭ ␩2 ϭ .045). Moreover, morality was seen as a 3.66; SD ϭ 1.93; F(1, 40) ϭ 4.91, p ϭ .032, more influential source than software compa- ␩2 ϭ .10). In contrast, participants who in- nies (M ϭ 3.19; SD ϭ 0.22; F(1, 10) ϭ 22.16, tended to behave congruently with a no- p ϭ .043, ␩2 ϭ .039). Finally, students from downloading norm perceived that their personal participants’ university were perceived as hav- values (M ϭ 5.89; SD ϭ 1.45) and their moral- ing the least influence on their own behavior ity (M ϭ 5.89; SD ϭ 1.44; F(1, 27) ϭ 00 p ϭ (M ϭ 2.41; SD ϭ 0.18; F(1, 104) ϭ 52.34, p ϭ 1.00, ␩2 ϭ .00) guided them more in their .006, ␩2 ϭ .070). behavior than the situation (M ϭ 4.79; SD ϭ A three-way interaction emerged between in- 1.55; F(1, 27) ϭ 10.45, p ϭ .003, ␩2 ϭ .28). group norm, congruence, and source of norma- There is no significant difference for partici- tive influence, F(5, 100) ϭ 4.36, p ϭ .001, ␩2 ϭ pants who intended to behave either incongru- .179. Simple effect tests revealed that for the ently with a no-downloading or a prodownload- sources of normative influence pertaining to ing norm (Fs ϭ 2.33 to 2.40, ps ϭ .10 to .17, software companies and to students, the norm ␩2 ϭ .08 to .13).

Influence of Personal Values Influence of One's Morality 6 Influence of Situation 5 4 3 Source 2 This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Perceived Influence of Influence Perceived 1 0 No- Pro- No- Pro- downloading downloading downloading downloading Norm- Norm- Norm- Norm- Congruence Congruence IncongruenceIncongruence

Figure 4. Perceived influence as a function of ingroup norm, congruence, and three sources of influence (personal values, morality, situation). NORM INTERNALISATION AND COMPARTMENTALISATION 185

Discussion showing that prosocial behaviors tend to be more self-determined, even if the individual is The first objective of this experiment was to acting against his or her group norm. Moreover, investigate the motives that underlie an illegal these prior studies directly measured the behav- normative behavior (i.e., illegally downloading ior, whereas the current study assessed behav- computer software or not) and to measure the ioral intentions. It is reassuring to observe sim- extent to which this behavior can be situation- ilar results across the studies. ally internalized in the self and endorsed for The current results are also consistent with self-determined motivations. As a second objec- literature that states that an individual may de- tive, this experiment tested the impact of a viate from his or her group norms if s/he con- manipulated norm and of participants’ inten- siders that it is better for the group to do so; tions to behave in congruence with this norm on such as because doing so may improve the variables that captured participants’ subjective group situation (Packer, 2008), or if the ingroup perceptions in the internalization process, norms go against one’s own moral values namely compartmentalization, intraindividual (Hornsey, Smith, & Begg, 2007). Still, it would conflict, and perceived sources of influence. be interesting to further investigate whether the This study used an established methodology norm that encourage illegally downloading from the social norms literature (Smith & Louis, computer software is perceived as being accept- 2008). Our findings confirmed the role played able for the individual and as being hurtful by behavioral congruence in the internalization versus good for the group. Even if it is illegal, of a potentially socially harmful behavior, while accessing the latest software can be perceived as providing new information on participants’ sub- good when it help the group to maintain high jective perceptions in this process. Investigating standards which may then allow its members to these ideas allows a better comprehension of the gain social recognition and status. The percep- motivations that drive some illegal behaviors. In tion of how the norm helps (vs. hurts) the group addition, identifying the subjective processes may influence the internalization process by that are activated when people plan to act ille- facilitating the acceptation or rejection of the gally allows to capture what the perpetrators of norm within his or her self. Furthermore, devi- these actions experience when engaging in such ating from ingroup norms when these norms are behaviors, and to develop more targeted inter- perceived as harmful may represent a strategy vention programs aimed at reducing illegal be- that group members use to realign their group havior such as theft. toward more prosocial actions and norms The ingroup norms we manipulated did in- (Packer, 2008). fluence participants’ intentions to engage in Moreover, participants who intended to act prodownloading versus no-downloading be- illegally—both by behaving incongruently with haviors, confirming H1. Also, participants a no-downloading norm and by failing to follow who intended to behave congruently with a no-downloading norm—felt more compart- their ingroup’s no-downloading norm were mentalization and intraindividual conflict than more self-determined compared to partici- participants who did not intend to engage in the pants whose behavioral intentions were not illegal downloading behavior (supporting H3 congruent with this norm (i.e., and who and H4). These results confirm that compart- planned to behave illegally instead), which mentalization and intraindividual conflict are This document is copyrighted by the American Psychological Association or one of its allied publishers. confirms H2a. In addition, participants whose most strongly activated when participants in- This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. behavioral intentions were incongruent with tend to behave illegally, regardless of ingroup the prodownloading norm reported higher norms: Behaving incongruently with the norm self-determination compared to participants of an important ingroup per se did not induce whose behavioral intentions were congruent such high levels of compartmentalization or with this prodownloading ingroup norm, sup- conflict. These variables may help to protect the porting H2b. Hence, above and beyond the nor- self from the negative impact of behaving ille- mative pressure, individuals internalize more gally. While the current data do not provide easily a prolegal than an illegal behavior. These exact information about which specific mecha- results correspond with SDT and replicate nisms may be blocking the internalization of Amiot and colleagues’ findings (2012), by illegal norms and also making these norms 186 SANSFAÇON AND AMIOT

harder to feel comfortable about (e.g., intuitive served in different life contexts and over the knowledge about what’s right vs. wrong), this is longer-term. It is important to note that the an exciting direction for future research which current research investigated the immediate im- align with SDT postulates. pact of a situational norm in an experimental Participants’ perceptions of the five sources context. Still, we can wonder about the process of influence also differed as a function of norms of internalization over time in real life contexts and congruence. When participants were in- and of how repeating behaviors over time may tending to behave legally, they were more likely lead to the internalization of these behaviors. to perceive that it was their personal values/ Further studies should investigate these longer- morality that guided their behavior. When they term processes by using longitudinal design. intended to behave illegally, they stated that the For example, we could follow over a 1-year situational sources influenced them the most. period a cohort of new university students as These findings could be attributable to a classic they are in the process of endorsing new social self-serving bias (Miller & Ross, 1975), norms, including potentially harmful norms whereby people attribute their positive behav- (e.g., cheating in exams). By assessing repeat- iors to internal causes (their values) and attri- edly over this period of time variables such as bute their negative behaviors to external causes perceived norms, the frequency of the norma- (the situation). These findings could also relate tive behavior and self-determination to engage to the notion of compartmentalization; when in this behavior, and degree of social identifi- participants behave illegally, they may become cation with the group, such research will allow more prone to perceive the situation as influ- to paint a more complete picture of the inter- encing them and to dissociate themselves per- nalization process over time. Moreover, the rate sonally from their action by putting a distance of internalization of group norms is possibly between their self (i.e., their own values and linked to changes in identification with the morality) and their behavior. group, such that a normative behavior becomes Interestingly, and despite the fact that our more congruent and internalized in the person’s ingroup norm manipulation had a strong effect self when this person develops a stronger sense on participants’ intentions in the experiment, of identification with the group (Terry & Hogg, participants were less likely to perceive that 1996). Understanding the temporal and longer- students from their university had influenced term processes involved in norm internalization their behavioral intentions. This pattern of find- could be useful to eventually create programs or ings is also congruent with literature suggesting campaigns that aim to reduce illegal behaviors that explicit and implicit processes are not al- in the workplace or school context. Considering ways congruent (Briñol, Petty, & Wheeler, the immediate versus the long-term impact of a 2006) and that what is really influencing a per- group norm on the behaviors and their internal- son’s behavior is often different from what the ization will also provide a more complete ac- person’s subjectively reported as influencing count for why individuals engage in those neg- him/her (Nisbett & Wilson, 1977). Still, the ative behaviors and how such behaviors correlation between identifying with the student develop. ingroup and perceiving the group as a source of influence suggest that the more participants References identify with this ingroup, the more they con- This document is copyrighted by the American Psychological Association or one of its allied publishers. sider the student group as influencing them, Aguiar, L., & Martens, B. (2013). Digital music This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. which is coherent with social identity theory consumption on the Internet: Evidence from Click- and its approach to group norms (e.g., Terry & stream Data. European Commission: Institute for Hogg, 1996; Turner, 1991). Future studies prospective technological studies. doi:10.2791/ should further investigate these subjective per- 83798 ceptions, how they shape and predict different Ajzen, I. (2001). Nature and operation of attitudes. Annual Review of Psychology, 52, 27–58. types of normative behaviors, and what are the Amiot, C. E., & Bourhis, R. Y. (2003). Discrimina- potential effects of this mismatch between per- tion and the positive-negative asymmetry effect: ceptions and behaviors. Ideological and normative processes. Personality It will also be interesting to further test and Social Psychology Bulletin, 29, 597–608. doi: whether the results obtained herein can be ob- 10.1177/0146167203029005005 NORM INTERNALISATION AND COMPARTMENTALISATION 187

Amiot, C. E., & Sansfaçon, S. (2011). Motivations to workers. Journal of Studies on Alcohol and Drugs, identify with social groups: Positive and negative 71, 526–534. consequences. Group Dynamics: Theory, Re- Guay, F., Mageau, G. A., & Vallerand, R. J. (2003). search, and Practice, 15, 105–127. doi:10.1037/ On the hierarchical structure of self determined a0023158 motivation: A test of top-down, bottom-up, recip- Amiot, C. E., Sansfaçon, S., & Louis, W. R. (2013). rocal, and horizontal effects. Personality and So- Investigating the motivations underlying harmful cial Psychology Bulletin, 29, 992–1004. doi: social behaviors and the motivational nature of 10.1177/0146167203253297 social norms. Journal of Applied Social Psychol- Haslam, S. A., Oakes, P. J., Reynolds, K. J., & ogy, 43, 2146–2157. doi:10.1111/jasp.12167 Turner, J. C. (1999). Social identity salience and Amiot, C. E., Sansfaçon, S., & Louis, W. R. (in the emergence of stereotype consensus. Personal- press). How normative and social identification ity and Social Psychology Bulletin, 25, 809–818. processes predict self-determination to engage in doi:10.1177/0146167299025007004 derogatory behaviours against outgroup hockey Haslam, S. A., & Reicher, S. (2007). Beyond the fans. European Journal of Social Psychology. banality of evil: Three dynamics of an integration- Amiot, C. E., Sansfaçon, S., Louis, W. R., & Yelle, ist social psychology of tyranny. Personality and M. (2012). Testing the role of group norms and Social Psychology Bulletin, 33, 615–622. doi: behavioral congruence in the internalization of dis- 10.1177/0146167206298570 criminatory and parity behaviors. Personality and Hornsey, M. J., Smith, J. R., & Begg, D. (2007). Social Psychology Bulletin, 38, 63–76. doi: Effects of norms among those with moral convic- 10.1177/0146167211429804 tion: Counter-conformity emerges on intentions Baer, J. S., Stacy, A., & Larimer, M. (1991). Biases but not behaviors. Social Influence, 2, 244–268. in the perception of drinking norms among college doi:10.1080/15534510701476500 students. Journal of Studies on Alcohol, 52, 580– Jacobson, R. J., Mortensen, C. R., & Cialdini, R. 586. (2011). Bodies obliged and unbound: Differenti- Baumeister, R. F., & Vohs, K. D. (2004). Four roots ated response tendencies for injunctive and de- of evil. In M. G. Arthur (Ed.), The social psychol- scriptive social norms. Journal of Personality and ogy of good and evil, (pp. 87–101). New York, Social Psychology, 100, 433–448. NY: Guilford Press. JDN. (2004). Téléchargement illégal: On continue Belgian Anti-Piracy Federation. (2009). Étude sur le malgré les risques. (Illegal downloading: We con- téléchargement auprès des jeunes en Belgique tinue regardless of the risks.) Retrieved on Sep- (Study on illegal downloading in Belgium Youth). tember 20, 2013, from http://www.journaldunet Press Release. .com/0405/040514sondage.shtml Benet-Martínez, V., & Haritatos’, J. (2005). Bicul- Kelman, H. C. (1961). Processes of opinion change. tural Identity Integration (BII): Components and Public Opinion Quarterly, 25, 57–78. doi:10.1086/ socio-personality antecedents. Journal of Person- 266996 ality, 73, 1015–1050. doi:10.1111/j.1467-6494 Kerr, N. L., Garst, J., Kiehle, D., & Harris, S. (1997). .2005.00337.x That still, small voice: Commitment to cooperate Briñol, P., Petty, R. E., & Wheeler, S. C. (2006). as an internalized vs. a social norm. Personality Discrepancies between explicit and implicit self and Social Psychology Bulletin, 23, 1300–1311. concepts: Consequences for information process- doi:10.1177/01461672972312007 ing. Journal of Personality and Social Psychology, Kerr, N. L., & Kaufman-Gilliland, C. M. (1994). 91, 154–170. doi:10.1037/0022-3514.91.1.154 Communication, commitment, and cooperation in Copyright Board of Canada. (2009). Copying for social dilemmas. Journal of Personality and Social Private Use. Copyright Art of Canada, subsection, Psychology, 66, 513–529. doi:10.1037/0022-3514 83(8). .66.3.513 This document is copyrighted by the American Psychological Association or one of its allied publishers. Deci, E. L., & Ryan, R. M. (2000). The “what” and Kura, K. M., Shamsudin, F. M., & Chauhan, A. This article is intended solely for the personal use of the individual user and is not“why” to be disseminated broadly. of goal pursuits: Human needs and the self- (2013). Modeling the influence of group norms determination of behavior. Psychological Inquiry, and self-regulatory efficacy on workplace deviant 11, 227–268. doi:10.1207/S15327965PLI1104_01 behaviour. Asian Social Science, 9, 113–122. Fielding, K. S., McDonald, R., & Louis, W. R. Lee, K., & Allen, N. J. (2002). Organizational citi- (2008). Theory of planned behavior, identity, and zenship behavior and workplace deviance: The intentions to engage in environmental activism. role of affect and cognition. Journal of Applied Journal of Environmental Psychology, 28, 318– Psychology, 87, 131–142. doi:10.1037/0021-9010 326. doi:10.1016/j.jenvp.2008.03.003 .87.1.131 Frone, M. R., & Brown, A. L. (2010). Workplace Lian, H., Ferris, L. D., & Brown, D. J. (2012). Does substance-use norms as predictors of employee taking the good with the bad make things worse? substance use and impairment: A survey of US How abusive supervision and leader–member ex- 188 SANSFAÇON AND AMIOT

change interact to impact need satisfaction and ish Journal of Social Psychology, 46, 769–792. organizational deviance. Organizational Behavior doi:10.1348/014466606X164439 and Human Decision Processes, 117, 41–52. doi: Smith, J. R., & Louis, W. R. (2008). Do as we say 10.1016/j.obhdp.2011.10.003 and as we do: The interplay of descriptive and Lynch, M. F., La Guardia, J. G., & Ryan, R. M. injunctive group norms in the attitude-behavior (2009). On being yourself in different cultures: relationship. British Journal of Social Psychology, Ideal and actual self-concept, autonomy support, 47, 647–666. doi:10.1348/014466607X269748 and well-being in China, Russia, and the United Smith, J. R., Terry, D. J., Manstead, A. S. R. Louis, States. The Journal of Positive Psychology, 4, W. R., Kotterman, D., & Wolfs, J. (2008). The 290–304. doi:10.1080/17439760902933765 attitude-behavior relationship in consumer con- Miller, D. T., & Ross, M. (1975). Sell-serving biases duct: The role of norms, past behavior, and self- in the attribution of causality: Fact or fiction? identity. The Journal of Social Psychology, 148, Psychological Bulletin, 82, 213–225. doi:10.1037/ 311–333. doi:10.3200/SOCP.148.3.311-334 h0076486 Spears, R., Postmes, T., Lea, M., & Wolbert, A. Millette, V., & Gagné, M. (2008). Designing volun- (2002). When are net effects of gross products? teers’ tasks to maximize motivation, satisfaction The power of influence and the influence of power and performance: The impact of job characteristics in computer-mediated communication. Journal of on volunteer engagement. Motivation and Emo- Social Issues, 58(1), 91–107. doi:10.1111/1540- tion, 32, 11–22. doi:10.1007/s11031-007-9079-4 4560.00250 Nisbett, R. E., & Wilson, T. D. (1977). Telling more Tajfel, H., & Turner, J. C. (1986). The social identity than we can know: Verbal reports on mental pro- theory of inter-group behavior. In S. Worchel & cesses. Psychological Review, 84, 1–259. L. W. Austin (Eds.), Psychology of intergroup Packer, D. J. (2008). On being both with us and relations. Chicago, IL: Nelson-Hall. against us: A normative conflict model of dissent Terry, D. J., & Hogg, M. A. (1996). Group norms in social groups. Personality and Social Psy- and the attitude-behavior relationship: A role for group identification. Personality and Social Psy- chology Review, 12, 50–72. doi:10.1177/ chology Bulletin, 22, 776–793. doi:10.1177/ 1088868307309606 0146167296228002 Reno, R. R., Cialdini, R. B., & Kallgren, C. A. Tripp, B. (2009). Fathers in jail: Managing dual iden- (1993). The transsituational influence of social tities. Applied Psychology in Criminal Justice, 5, norms. Journal of Personality and Social Psychol- 26–56. ogy, 64, 104–112. doi:10.1037/0022-3514.64.1 Turner, J. C. (1991). Social influence. Belmont, CA: .104 Thomson Brooks/Cole Publishing Co. Robinson, S. L., & Greenberg, J. (1998). Employees Vallerand, R. J. (1997). Toward a hierarchical behaving badly: Dimensions, determinants and di- model of intrinsic and extrinsic motivation. In lemmas in the study of workplace deviance. In M. P. Zanna (Ed.), Advances in experimental C. L. Cooper & D. M. Rousseau (Ed.), Trends in social psychology (Vol. 29, pp. 271–360). San organizational behavior (Vol. 5, pp. 1–30). New Diego, CA: Academic Press. doi:10.1016/S0065- York, NY: Wiley. 2601(08)60019-2 Robinson, S. L., & O’Leary-Kelly, A. M. (1998). Vallerand, R. J., & Bissonnette, R. (1992). Intrinsic, Monkey see, monkey do: The influence of work extrinsic, and amotivational styles as predictors of groups on the antisocial behavior of employees. behavior: A prospective study. Journal of Person- Academy of Management Journal, 41, 658–672. ality, 60, 599–620. doi:10.1111/j.1467-6494.1992 doi:10.2307/256963 .tb00922.x Ryan, R. M., & Connell, J. P. (1989). Perceived locus Van Bavel, J. J., Packer, D. J., & Cunningham, W. A. of causality and internalization: Examining rea- (2008). The neural substrates of in-group bias a This document is copyrighted by the American Psychological Association or one of its allied publishers. sons for acting in two domains. Journal of Person- functional magnetic resonance imaging investiga- This article is intended solely for the personal use of the individual user and is notality to be disseminated broadly. and Social Psychology, 57, 749–761. doi: tion. Psychological Science, 19, 1131–1139. doi: 10.1037/0022-3514.57.5.749 10.1111/j.1467-9280.2008.02214.x Ryan, R. M., & Deci, E. L. (2003). On assimilating Vorauer, J. D., & Miller, D. T. (1997). Failure to identities to the self: A self-determination theory recognize the effect of implicit social influence on perspective on internalization and integrity within the presentation of self. Journal of Personality and cultures. In M. R. Ryan & E. L. Deci (Eds.), Social Psychology, 73, 281–295. doi:10.1037/ Handbook on self & identity (pp. 253–272). New 0022-3514.73.2.281 York, NY: The Guilford Press. Smith, J. R., Hogg, M. A., Martin, R., & Terry, D. J. Received September 24, 2012 (2007). Uncertainty and the influence o group Revision received December 14, 2013 norms in the attitude-behaviour relationship. Brit- Accepted December 17, 2013 Ⅲ