the formation of group affect and team effectiveness: the moderating role of identification

19
The Formation of Group Affect and Team Effectiveness: The Moderating Role of Identification Jacqueline Tanghe, 1,2 Barbara Wisse 3 and Henk van der Flier 1 1 Department of Work and Organizational Psychology, VU University, The Netherlands, 2 Institute of Integration and Social Efficacy, Faculty of Behavioral and Social Sciences, University of Groningen, Grote Kruisstraat 2/1, 9712 TS Groningen, The Netherlands, and 3 Department of Social and Organizational Psychology, University of Groningen, The Netherlands Corresponding author email: [email protected] In the current research we use the social identity perspective to enhance our understanding of group affect (i.e. a collectively shared pattern of affective states among group members). Because higher identification (i.e. the extent to which group members define themselves in terms of their group membership) is related to higher attentiveness to fellow group members, we expected that group identification would foster affective convergence, and that the effects of group affective tone on team effectiveness would be stronger for higher identifying groups. A survey of teams (n 5 71 teams) confirmed our expectations. A scenario experiment (n 5 121 participants) added to our findings by showing that identification does indeed lead group members to affectively converge to their fellow group members and that this affective convergence, in turn, explains subsequent team-oriented attitudes. Our study testifies to the notion that team managers may want to take notice of and manage affect in work groups, because, especially in higher identifying teams, affect may spread among team members and influence the team’s effectiveness. Recent years have witnessed a growing interest in the role of affect (i.e. emotions and moods) in organizations and work groups. Research find- ings, showing that affect influences work-related attitudes, values, judgements and behaviour, justify this increased awareness of the significance of affect (e.g. Ashkanasy, Zerbe and Ha¨ rtel, 2005; Barsade, 2002; Forgas and George, 2001; George, 1990; Ha¨rtel, Zerbe and Ashkanasy, 2005; Kelly and Barsade, 2001; Lord, Klimosky and Kanfer, 2002). However, contemporary studies on affect focus almost exclusively on either intrapersonal processes (e.g. how do a persons’ affective states influence his or her own behaviours, attitudes etc.) or on dyadic interaction processes (e.g. the role of affect in negotiations, leadership processes etc.) (e.g. van Knippenberg, B., van Knippenberg and De Cremer, 2007). Research that takes into account that groups are the building blocks of organizations (West, Borrill and Unsworth, 1998) is much scarcer. Fortunately, a line of research that takes on a group perspective is beginning to emerge. Several researchers argue that affect may at times be regarded as a group level phenomenon, and refer to it as group affect (e.g. Barsade, 2002; George, 1990). Group affect is formed as a result of emotional convergence among group members. Various studies show that affect transmitted among members of groups may indeed establish affective states that are sufficiently homogeneous among group members and recognizable by independent observers to be treated as a collective property of the group (e.g. Bartel and Saavedra, 2000; George, 1990; Kelly and Barsade, 2001; Totterdell et al., 1998). Group affect may be British Journal of Management, Vol. 21, 340–358 (2010) DOI: 10.1111/j.1467-8551.2009.00656.x r 2009 British Academy of Management. Published by Blackwell Publishing Ltd, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA, 02148, USA.

Upload: independent

Post on 15-Nov-2023

0 views

Category:

Documents


0 download

TRANSCRIPT

The Formation of Group Affect and TeamEffectiveness: The Moderating Role of

Identification

Jacqueline Tanghe,1,2 Barbara Wisse3 and Henk van der Flier11Department of Work and Organizational Psychology, VU University, The Netherlands, 2Institute of

Integration and Social Efficacy, Faculty of Behavioral and Social Sciences, University of Groningen, Grote

Kruisstraat 2/1, 9712 TS Groningen, The Netherlands, and 3Department of Social and Organizational

Psychology, University of Groningen, The Netherlands

Corresponding author email: [email protected]

In the current research we use the social identity perspective to enhance our

understanding of group affect (i.e. a collectively shared pattern of affective statesamong group members). Because higher identification (i.e. the extent to which group

members define themselves in terms of their group membership) is related to higher

attentiveness to fellow group members, we expected that group identification would

foster affective convergence, and that the effects of group affective tone on teameffectiveness would be stronger for higher identifying groups. A survey of teams (n5 71

teams) confirmed our expectations. A scenario experiment (n5 121 participants) added

to our findings by showing that identification does indeed lead group members to

affectively converge to their fellow group members and that this affective convergence,in turn, explains subsequent team-oriented attitudes. Our study testifies to the notion

that team managers may want to take notice of and manage affect in work groups,

because, especially in higher identifying teams, affect may spread among team membersand influence the team’s effectiveness.

Recent years have witnessed a growing interest inthe role of affect (i.e. emotions and moods) inorganizations and work groups. Research find-ings, showing that affect influences work-relatedattitudes, values, judgements and behaviour,justify this increased awareness of the significanceof affect (e.g. Ashkanasy, Zerbe and Hartel, 2005;Barsade, 2002; Forgas and George, 2001; George,1990; Hartel, Zerbe and Ashkanasy, 2005; Kellyand Barsade, 2001; Lord, Klimosky and Kanfer,2002).However, contemporary studies on affect focus

almost exclusively on either intrapersonal processes(e.g. how do a persons’ affective states influence hisor her own behaviours, attitudes etc.) or on dyadicinteraction processes (e.g. the role of affect innegotiations, leadership processes etc.) (e.g. vanKnippenberg, B., van Knippenberg and De

Cremer, 2007). Research that takes into accountthat groups are the building blocks of organizations(West, Borrill and Unsworth, 1998) is muchscarcer. Fortunately, a line of research that takeson a group perspective is beginning to emerge.Several researchers argue that affect may at

times be regarded as a group level phenomenon,and refer to it as group affect (e.g. Barsade, 2002;George, 1990). Group affect is formed as a resultof emotional convergence among group members.Various studies show that affect transmittedamong members of groups may indeed establishaffective states that are sufficiently homogeneousamong group members and recognizable byindependent observers to be treated as a collectiveproperty of the group (e.g. Bartel and Saavedra,2000; George, 1990; Kelly and Barsade, 2001;Totterdell et al., 1998). Group affect may be

British Journal of Management, Vol. 21, 340–358 (2010)DOI: 10.1111/j.1467-8551.2009.00656.x

r 2009 British Academy of Management. Published by Blackwell Publishing Ltd, 9600 Garsington Road, OxfordOX4 2DQ, UK and 350 Main Street, Malden, MA, 02148, USA.

either relatively positive or negative in nature.The relative positivity or negativity of group affectspecifies the group’s affective tone. Importantly,this group affective tone may influence subsequentgroup functioning and effectiveness (e.g. Barsade,2002; George, 1990, 1995; Totterdell, 2000).Despite the apparent relevance of the topic,

there is little insight into the conditions thatfoster the development of group affect (Forgasand George, 2001). Therefore, the present studyaims to shed some light on this issue. Specifically,we integrate insights from the social identityperspective (e.g. Hogg and Abrams, 1988; Maeland Ashforth, 1992; Tajfel, 1974; Tajfel andTurner, 1986; Turner, 1985) with theory providedby research on the role of affect in organizationsand work groups. We assert that identification,i.e. the extent to which people define themselvesin terms of their group membership (Ashforthand Mael, 1989), may be regarded as a keydeterminant of emotional convergence amonggroup members. Moreover, we argue that identi-fication also moderates the effects of the group’saffective tone on team effectiveness. Our studytherefore contributes to research on the role ofaffect in the workplace by presenting a frame-work that explains when and why affect mayconverge among group members and result in agroup affective tone that influences work groupfunctioning and effectiveness.

Group affect

The term affect refers to the experience of feeling.While it could refer to sentiments (affectiveevaluations of whether something is liked ordisliked) (Kelly and Barsade, 2001), the term ismost often used to refer to specific emotions andmoods. Compared to emotions, moods lastlonger, are less intense, are more diffuse by lackof a (specific) object, and may be evoked byrelatively insignificant events (Forgas, 1992;Frijda, 1993). Moods and emotions may bedescribed in terms of the extent to which a personfeels good or bad (i.e. the valence dimension),and the extent to which a person feels activated(i.e. the activation dimension) (Russell, 1980;Russell and Feldman Barrett, 1999; Watson,Clark and Tellegen, 1988). Thus, an individualmay experience numerous moods and emotions,varying in both valence and activation.

Importantly, a groupmay experience numerousaffective states as well. Thus, a group, as a whole,may be characterized as being composed ofmembers who uniformly feel more or less positiveor negative, activated or unactivated. Similar toindividual affective states, these group affectivestates may refer to shared emotions or sharedmoods (George, 2002). Group affect, a sharedpattern of affective states, has been shown to existin groups of students (Barsade, 2002), sales people(George, 1990, 1995), nurses, accountants (Tot-terdell et al., 1998) and in sports teams (Totterdell,2000). The question that arises, however, is howthese homogeneous affective states come about.Group affect has been argued to arise as a

result of emotional convergence. During groupmember interaction a continuous, but verysubtle, transfer of affective states among teammembers may take place (Bartel and Saavedra,2000). Emotional convergence thus refers to theprocess in which people are affectively influencedby others and become more similar with regard totheir affective states. Emotional convergence hasso far been shown to occur in dyads (e.g.Anderson, Keltner and John, 2003; Levensonand Gottman, 1983), but, more importantly,group members may also converge emotionally(e.g. Totterdell et al., 1998). Indeed, a group maybe an excellent breeding ground for the transferof affective states from one person to another (i.e.one grouchy colleague may spoil the mood ofmany co-workers).

Antecedents of emotional convergence

To date, little research attention has been devotedto the conditions that foster affective conver-gence. Yet, some antecedents have been reported.Bartel and Saavedra (2000) found that affectconverges more easily when group membership isstable, and when individuals are dependent onone another, both on the interpersonal level andin terms of task performance. Another finding isthat the convergence of affect among teammembers is easier when group members takepart in collective activities (Totterdell, 2000) andwhen people share (strong) work ties in organiza-tional networks (Totterdell et al., 2004). It hasalso been suggested that organizational emotionnorms, group norms and emotional history mightpromote affective convergence between people(Kelly and Barsade, 2001). However, we argue

Group Affect and Identification 341

r 2009 British Academy of Management.

that the field may benefit from an overarchingtheoretical framework explaining when and howgroup affect is formed.The social identity perspective (e.g. Ashforth

and Mael, 1989; Hogg and Abrams, 1988; Tajfel,1974; Tajfel and Turner, 1986) may provide sucha theoretical framework. The social identityperspective has already been used to explainaffective processes in intergroup contexts (e.g.Giner-Sorolla, Mackie and Smith, 2007; Iyer,Leach and Crosby, 2003; Smith, Seger andMackie, 2007). Indeed, people’s reaction to socialgroups and their members may be seen from anaffective perspective. For instance, terrorist at-tacks may give rise to feelings of fear, and theremembrance of a colonial past may engenderfeelings of guilt or shame, especially in those whoidentify with their country. Because these inter-group emotions may be socially shared within agroup, all members of the (in)group mayexperience the same affective state because oftheir dealings with the same outgroup.We posit that the social identity perspective

may be helpful in explaining affective intragroupprocesses as well. Not only may group-basedaffect arise because of similar reaction to con-frontation with other groups and their members,but affective states may also converge to othermembers of the same group because of the extentto which group members feel psychologicallyattached to, or identify with, their group. Indeed,as we will elaborate on below, we argue thatidentification is conducive to emotional conver-gence within groups.

Identification

Social identity and self-categorization perspec-tives posit that an individual’s self-conception isaffected not only by his or her unique personalcharacteristics, but also by his or her membershipin social groups, such as work groups ororganizations (e.g. Hogg and Abrams, 1988;Tajfel and Turner, 1986; Turner, 1985). Identifi-cation reflects the perception of oneness with orbelongingness to a group, where the individualdefines him or herself in terms of the group ofwhich he or she is a member (e.g. Hogg andAbrams, 1988; Mael and Ashforth, 1992; Tajfel,1974; Tajfel and Turner, 1986; Turner, 1985).Group members may differ in the extent to whichgroup membership defines their self-image and

thus in the extent to which they identify with thegroup (Ashforth and Mael, 1989; Hogg andTerry, 2000).Interestingly, identification is assumed to

largely depend on a sense of shared social identity(Haslam, 2001; Turner, 2001). Thus, it is moredifficult to identify with a collective whenmembers, belonging to the same group, do notshare similar perceptions of what is representa-tive for the group’s identity. Therefore, groupmembers’ identification levels may converge overtime (van Knippenberg, B., and van Knippen-berg, 2005; Reicher, Haslam and Hopkins, 2005).Identification leads to activities that are con-

gruent with the group’s identity (van Knippen-berg, D., and Hogg, 2003). Higher identificationmay lead group members to conform more togroup norms, values and attitudes. People whoidentify strongly with their organization are morewilling to expend effort on behalf of the organiza-tion, are more interested in organizational pro-cesses, offer more support to and are more loyalto the organization, are less likely to turn-overand are more satisfied with their job than thosewho identify less strongly (Ashforth and Mael,1989; van Knippenberg, B., Martin and Tyler,2006; Mael and Ashforth, 1992; Van Dick et al.,2004a). In addition, high identification may leadto increased organizational citizenship behaviour(OCB) (Van Dick et al., 2006) or the voluntarydemonstration of pro-social behaviour that goesbeyond the employee’s normal task performanceand task description (cf. Podsakoff et al., 2000) aswell as to higher work motivation (Riketta andNienaber, 2007). In sum, organizational and teameffectiveness may benefit from higher levels ofgroup identification (see Riketta, 2005).

Identification and emotional convergence

Identification is considered to enhance the like-lihood that group members are open to theinfluence of other group members. Haslam (2001)for instance contends that people who identifyhighly with the group are more motivated to pickup on signals from other group members, and aretherefore more attentive to the other groupmembers’ behaviours, feelings and attitudes.Identification, thus, may enhance the likelihoodthat group members (consciously or uncon-sciously) detect other group members’ affectivestates. We argue that this heightened motivation

342 J. Tanghe, B. Wisse and H. van der Flier

r 2009 British Academy of Management.

and attention to others’ affective signals may setseveral emotion-regulation processes (George,2002) in motion, which in turn may result inemotional convergence.First, primitive emotional contagion processes

(i.e. the non-conscious mimicking and synchro-nizing of facial expressions, vocal expressions andposture of other people) (Hatfield, Cacioppo andRapson, 1994) may be responsible for affectiveconvergence in groups. Primitive emotional con-tagion, however, is more likely to occur whenmore attention is being allocated to the otherperson (Barsade, 2002). Contagion is thereforemore likely to occur with increasing identifica-tion, because identification is positively related toattentiveness. Second, vicarious processes (Ban-dura, 1986; Lazarus, 1991) and emotional em-pathy (Davis, 1983) engender the transfer ofaffect, because the self is psychologically placedin another person’s situation. This may beparticularly easy for high identifiers, because forthem the psychological boundaries between selfand other group members are already relativelyhazy (cf. Norton et al., 2003). Third, higheridentification leads group members to see fellowgroup members as more similar to the self and asmore relevant sources of information (vanKnippenberg, D., 2000; van Knippenberg, D.,and Hogg, 2003). This may foster affective socialcomparison processes (cf. Festinger, 1954), inwhich the appropriateness of one’s own feelings isassessed by comparing them to those of othergroup members and, if incongruity is detected,adapting one’s own feelings. Finally, adherenceto emotion-management strategies (i.e. layingdown how one should feel and express oneself)(Hochschild, 1990) is also more likely for highidentifiers, because stronger identification leadsgroup members to be more willing to conform togroup norms, values and attitudes.Therefore, based on the above, our first general

hypothesis is:

H1: The higher the group identification is, thestronger the affective convergence among teammembers.

Group affective tone, team effectiveness andidentification

Emotional convergence may lead to group affectthat has a (specific) affective tone. Groups have

been found to be more positive in their affectivestates (groups that, as a collective, feel pleasant)or more negative (groups that, as a collective, feelless pleasant) (Barsade, 2002; George, 1990;Totterdell, 2000; Totterdell et al., 1998). Thisgroup affective tone may have a profoundinfluence on work group functioning.Several studies focusing explicitly on the effects

of group affect indicate that positive groupaffective tone may promote team effectiveness.Positive group affective tone has been found to benegatively related to absenteeism (George, 1990;Mason and Griffin, 2003) and positively relatedto pro-social or helping behaviours towards co-workers, customer-service behaviours and altruis-tic behaviours in teams of sales people (George,1990, 1995). Also, collective happy moods ofprofessional cricket teams were positively relatedto their subjective performance (Totterdell,2000). A laboratory study with students hasshown that positive group affective tone (i.e. theincrease of positive affect after being subjected toa confederate’s positive affective displays) leadsto an increase in (perceived) task performance, animprovement in cooperation and a decrease ofconflict (Barsade, 2002). Theoretical explanationsfor these findings often focus on the idea thatpositive affect is considered to broaden people’sattention and create an environment in whichresources and skills can be built, subsequentlypromoting individual and team effectiveness(Fredrickson, 1998). Likewise, positive affect inindividuals is often considered to advance crea-tivity, organizational commitment, pro-socialbehaviours (e.g. Forgas and George, 2001;Frederickson, 1998, 2000; George, 1991; Isen,2004; Isen, Daubman and Nowicki, 1987),problem solving and decision making (for anoverview see Isen, 2000, 2004).There seems to be less clarity in the literature

about the effects of negative affect on teameffectiveness. Some have argued that negativeaffect is less important than positive affect when itcomes to social interactions and the effects ofaffect congruency (e.g. Barsade et al., 2000;Damen, van Knippenberg and van Knippenberg,2008; Watson et al., 1992). Most frequently,though, negative affect is assumed to haveharmful effects in work group settings (e.g.Fredrickson, 1998; Lord, Klimosky and Kanfer,2002). Empirical studies on the effects of negativegroup affective tone are relatively scarce, but

Group Affect and Identification 343

r 2009 British Academy of Management.

suggest a negative link between negative groupaffective tone and team effectiveness. George(1990), for instance, showed that negative groupaffective tone may lead to a lessened display ofcustomer-service behaviour in teams of salespeople.Importantly, we propose that group identifica-

tion is likely to moderate the effects of groupaffective tone on team effectiveness. As we haveargued in the above, higher identification maylead group members to behave more in accor-dance with the group’s identity (i.e. group norms,values and attitudes). Identification, thus, isconsidered to strengthen the effects of groupdefining characteristics. Although most researchattention to the role of identification typically hasgone to more cognitive group defining character-istics (i.e. norms, attitudes), the process under-lying the effects of identification may be similarfor those matters that make a group affectivelydistinctive. Indeed, the group’s collective affectivestate may also be regarded as a group definingcharacteristic. Thus, identification may strength-en the effects of group affective tone. Morespecifically, and mirroring earlier findings show-ing that positive group affective tone has apositive link with team effectiveness and negativegroup affective tone has a negative link with teameffectiveness, our second and third generalhypotheses are as follows.

H2: Positive group affective tone will bepositively associated with team effectivenessand this effect will be stronger for higher levelsof group identification.

H3: Negative group affective tone will benegatively associated with team effectivenessand this effect will be stronger for higher levelsof group identification.

These general hypotheses are tested in twostudies. First, we use a cross-sectional survey ofwork groups (Study 1) to determine whether thepredicted relationships may be observed inorganizational settings. Second, we employ ascenario study (Study 2), which holds theadvantage that it is a controlled experimentalset-up yielding results with high internal validitymaking conclusions about causality possible. Asindicators of team effectiveness we opted forOCB (Study 1; Study 2) and perceived team

performance (Study 1). Both OCB (Organ, 1988)and perceived (or self-reported) team perfor-mance may be considered valuable indicators ofteam effectiveness (cf. George, 1991; West, Borrilland Unsworth, 1998).

Study 1

The aim of this first study, a cross-sectionalsurvey of work groups, was to observe whetherthe hypothesized relationships do occur in reality.The questionnaires allowed us to calculate

team scores of emotional convergence, groupaffective tone and team effectiveness (i.e. collec-tive willingness to display OCB and collectiveperceived team performance). Additionally, wefirst tested whether identification may indeed betreated as a group level variable, as theorysuggests (cf. Haslam, 2001), and we then calcu-lated identification team scores.With these measures Study 1 was used to

specifically test whether (1) group identification ispositively associated with affective convergenceamong group members, (2) positive groupaffective tone has a stronger positive associationwith collective willingness to display OCB andcollective perceived team performance for highergroup identification, and (3) negative groupaffective tone has a stronger negative associationwith collective willingness to display OCB andcollective perceived team performance for highergroup identification.

Method

Sample. Respondents were 171 employees (Mage

5 28.6, SD5 9.73, 43% male) of 71 teams (two tofour members; Mwork-experience59.21 years,SD5 8.02; Mworking-period-current-team5 3.06 years,SD5 4.78). Teams were mainly derived fromcommercially oriented service organizations, suchas shops, bars, restaurants and physiotherapists’offices, but we also surveyed teams working atpromotion stands.

Procedure. Respondents were approached atwork. Participation was restricted to teams thathad three or more members and that had at leasttwo members willing to fill in the questionnaire.Participation was anonymous and voluntary. If

344 J. Tanghe, B. Wisse and H. van der Flier

r 2009 British Academy of Management.

teams agreed to participate, group members filledout the questionnaire on the spot and returned itto the researcher upon completion. The instruc-tion read that the questionnaire focused on howpeople experience their work and on how theyfeel during work, and that the items about workteam referred to the team of people they werecurrently working with. Because people wereapproached at work we kept the survey shortand to the point.

Measures. All responses were assessed on sev-en-point scales (15 completely disagree to75 completely agree).Identification was measured using four items

(e.g. ‘The successes of this team are mysuccesses’). We used the Dutch translation (vanKnippenberg, A., et al., 2000) of the Mael andAshforth (1992) scale.Although affect is often measured with the

PANAS (i.e. Positive Affect Negative AffectSchedule; Watson, Clark and Tellegen, 1988) itsdisadvantage is that it only uses affect terms thatfall on the high end of the arousal scale (Russelland Carroll, 1999). We opted for a measure thatis more sensitive to the fact that moods andemotions may not only differ in hedonic valencebut may also differ in the extent to which theyrepresent high and low arousal (activation)affective states. We decided for a scale that isderived more directly from the circumplex modelof affect (Larsen and Diener, 1992). As aconsequence, positive affect (PA) and negativeaffect (NA) are measured by assessing highactivation moods and low activation moods. PAwas measured with six items. Three itemsassessed high activation moods (e.g. ‘I feelenthusiastic’) and three items assessed lowactivation moods (e.g. ‘I feel relaxed’). NA wasalso measured with six items. Again, three itemsassessed high activation moods (e.g. ‘I feelirritated’) and three items assessed low activationmoods (e.g. ‘I feel bored’). Participants wereinstructed to answer all items regarding theirfeelings by reporting how they felt at thatparticular moment. Although usually presentand past mood states are related, current moodstates may be more accurately and reliablyreported than recalled moods.Willingness to engage in OCB was measured

with five items from the individual initiative scale

(Moorman and Blakely, 1995) (e.g. ‘I frequentlyoffer suggestions to co-workers on how thegroup can improve’). Perceived team performancewas measured with four items (e.g. ‘This workteam is performing better than can be expectedof us’).

Results

Preliminary analysis. See Table 1 for means,standard deviations, Cronbach’s alphas andintercorrelations of all variables for individualscores. Note that group members’ identificationis positively associated with own positive affectand negatively associated with own negativeaffect (cf. Van Dick et al., 2004b).

We conducted principal components analyseswith OBLIMIN rotation of all items, yielding theexpected solutions and explaining at least 41%variance. Factor loadings were higher than 0.60for identification, higher than 0.036 for PA,higher than 0.57 for NA, higher than 0.64 forwillingness to engage in OCB, and higher than0.68 for perceived performance.

To justify aggregation we first calculated intra-class correlation (ICC) coefficients (Shrout andFleiss, 1979). Note that because group affect ishypothesized to be dependent on identification,we expect moods to converge more (i.e. be morein agreement) for groups with higher identifica-tion levels. As a consequence, relatively lowestimates of agreement may be obtained forgroup affect measures.

The ICC(1) values, estimating the extent towhich members of the same group respondedsimilarly, were 0.36 for identification, 0.22 forwillingness to engage in OCB, 0.22 for perceivedteam performance, 0.24 for NA and 0.09 for PA.

Table 1. Means, standard deviations and intercorrelations for the

individual variables of Study 1

M SD 1 2 3 4 5

1. Identification 5.84 0.88 (0.68)

2. PA 5.18 0.89 0.36* (0.71)

3. NA 2.23 1.00 � 0.23* � 0.24* (0.80)

4. OCB 5.25 1.09 0.40* 0.20* � 0.02 (0.84)

5. Performance 5.45 0.87 0.30* 0.28* � 0.07 0.42* (0.71)

Notes: Cronbach’s alphas are displayed on the diagonal.

n5 171 (listwise).*po0.01.

Group Affect and Identification 345

r 2009 British Academy of Management.

These values are well above the median value of0.12 reported in the organizational literature(James, 1982) except for PA, which (barely) doesnot meet this criterion. In addition, the ICC(2)values, reflecting the interrater agreement andreliability of the average rating, were 0.58 foridentification, 0.40 for willingness to engage inOCB, 0.40 for perceived team performance, 0.44for NA and 0.19 for PA. Moreover, the one-wayanalyses revealed significant effects for the ran-dom between-subjects factor team for identifica-tion, F(1, 70)5 2.36, po0.001, Z2 5 0.62, forwillingness to engage in OCB, F(1, 70)5 1.67,po0.01, Z2 5 0.54, for perceived team perfor-mance, F(1, 70)5 1.67, po0.01, Z2 5 0.54, andfor NA, F(1, 70)5 1.78, po0.01, Z2 5 0.55.Because for PA the one-way analyses failed tomeet significance, F(1, 70)5 1.23, p5 0.17,Z2 5 0.46, we also calculated the average within-group reliability (Rwg) (James, Demaree andWolf, 1984). The values are 0.87 for identifica-tion, 0.82 for willingness to engage in OCB, 0.89for perceived team performance and 0.85 for NA.Importantly, the Rwg for PA is high as well (0.89).These values are well above the suggested 0.70cut-off value for aggregating ratings from theindividual level to the group level (James,Demaree and Wolf, 1984). Clearly, taking allthese values (i.e. ICC(1), ICC(2) and Rwg) intoconsideration, aggregation is justified.Means, standard deviations and intercorrela-

tions of all variables for the aggregated scores aredisplayed in Table 2.

Affective convergence. We aggregated the moodmeasures for PA and NA and obtained averagedeviation indices (AD) (Burke and Dunlap,2002). With an AD index one determines the

extent to which a person’s rating differs from thegroup (mean) rating, by summing up the absolutevalues of these deviations and then dividing thisscore by the number of deviations. This measurethus indicates the extent to which group membersare in agreement with other group membersregarding their affective states.We found that identification correlates nega-

tively with the AD indices for both PA (r5 �0.28,po0.02)1 and NA (r5 �0.24, po0.05).1 Thesenegative correlations indicate that as identificationgets stronger members deviate less from their teammembers on their mood ratings. Our first hypoth-esis, stating that group identification is associatedwith stronger emotional convergence among groupmembers, is thus supported.

Table 2. Means, standard deviations and intercorrelations for the aggregated variables of Study 1

M SD 1 2 3 4 5 6 7

1. Identification 5.85 0.72 –

2. PA (AD) 0.47 0.39 � 0.28* –

3. NA (AD) 0.48 0.45 � 0.24* 0.13 –

4. PA (Mean) 5.16 0.65 0.34** � 0.13 � 0.18 –

5. NA (Mean) 2.20 0.76 � 0.27* 0.01 0.53** � 0.19 –

6. OCB 5.27 0.84 0.48** � 0.25* 0.02 0.17 0.07 –

7. Performance 5.48 0.67 0.33** � 0.10 � 0.06 0.19 � 0.00 0.40** –

Note: n5 71 (listwise).*po0.05; **po0.01.

1Adding the control variables gender, age, work timewith current team and the mean score of PA and NA didnot translate into more variance being explained. Also,the patterns of results for all analyses that included thecontrol variables led to the same conclusions as the onesreported in the paper. In one of the analyses we enteredthe mean NA score and the group identification score aspredictors of the AD measure of NA. This analysisshowed that the main effect of the mean score for NAwas significant (b5 0.47, t5 4.44, po0.001), while themain effect of identification was not (b5 –0.06, t5 0.58,ns). Yet, simple slope analyses for the interaction termidentification � mean NA (b5 –0.22, t5 2.12,po0.04) show that for higher values of NA identifica-tion does correlate negatively with the AD of NA (b5–0.24, p5 0.05) but not for smaller values of NA(b5 0.12, ns). This relationship between mean NA andAD NA therefore seems to be due to a bottom effect.(That is, when groups experience very low levels ofnegative group affect, average deviation between groupmembers between groups is automatically limited. As aconsequence, correlations between the AD index of NAand identification are less likely.)

346 J. Tanghe, B. Wisse and H. van der Flier

r 2009 British Academy of Management.

Team effectivenessWillingness to engage in OCB. We conducteda hierarchical regression analysis2,3,4 with theaggregated mean score of OCB intentions asdependent variable and the aggregated meanscores of PA, identification and the PA � identi-fication interaction term as predictors. FollowingAiken and West (1991) we used centred scores tocalculate main effects and the interaction term(see Table 3).We found a main effect of identification.

Identification was positively related to willingnessto engage in OCB. We also found an interactionbetween PA and identification. This interactionrevealed that, as expected, willingness to engagein OCB was contingent on PA when groupidentification was higher. Simple slope analyses(Aiken and West, 1991) further revealed thatpositive group affect was marginally significantlyrelated to willingness to engage in OCB whengroup identification was high (1 SD above themean), b5 0.27, p5 0.09, but was not related towillingness to engage in OCB when group

identification was low (1 SD below the mean),b5 �0.06, ns (see Figure 1). Apparently, thepositive effects of positive group affective tone onwillingness to engage in OCB are more pro-nounced for higher identifying teams.We performed a similar hierarchical regression

analysis with the aggregated mean scores of NA.We again found the significant main effect ofidentification, showing that identification waspositively related to willingness to engage inOCB. No significant effect of the interactionbetween NA and identification was obtained (seeTable 3).

Perceived team performance. We conducted ahierarchical regression analysis5 with the aggre-gated mean score of perceived team performanceas our dependent variable and the centredaggregated mean scores of PA and identificationand the PA � identification interaction term aspredictors (Aiken and West, 1991) (see Table 4).We found a main effect of identification.

Identification was positively related to perceivedteam performance. We again obtained an inter-action between PA and identification. Thisinteraction revealed that indeed, as expected,perceived team performance was contingent onPA when group identification was higher. Simpleslope analyses (Aiken and West, 1991) furtherrevealed that positive group affect was signifi-cantly related to perceived team performancewhen group identification was high (1 SD abovethe mean), b5 0.35, po0.05, but was not relatedto perceived team performance when group

Table 3. Results of hierarchical regression analyses for willingness

to engage in OCB

Predictor b t p DR2

Positive affect 0.11 0.91 ns 0.01

Identification 0.52 4.66 0.000 0.23

PA � identification 0.26 2.28 0.026 0.06

Negative affect 0.20 1.84 ns 0.04

Identification 0.55 5.05 0.000 0.27

NA � identification � 0.07 � 0.63 ns 0.004

Notes: n5 71 (listwise). DR2 is the variance explained by each

predictor after the other predictors are entered into the

equation.

4

5

6

High Identification Low Identification

Will

ingn

ess

to e

ngag

e in

OC

B

High Positive Group AffectLow Positive Group Affect

Figure 1. Interaction between positive group affect and identifica-

tion on willingness to engage in OCB for Study 1

2For all analyses adding the control variables gender,age, work time with current team and group size did nottranslate into more variance being explained. Also, thepatterns of results for all analyses that included thecontrol variables led to the same conclusions as the onesreported in the paper.3Alternative models were tested to see whether otherrelationships between team effectiveness, identificationand PA existed. None of these results was significant.4The model under investigation in this study does notwarrant analysing the data by means of hierarchicallinear modelling (HLM). HLM is specifically used foranalysing models which investigate relationships be-tween both individual level variables and group levelvariables. All of the investigated relationships in ourstudy refer to group level variables. 5See footnotes 2 and 3.

Group Affect and Identification 347

r 2009 British Academy of Management.

identification was low (1 SD below the mean),b5 0.03, ns (see Figure 2). Apparently, thepositive effects of positive group affective toneon perceived team performance are more pro-nounced for higher identifying teams.We performed a similar hierarchical regression

analysis with the aggregated mean scores of NA.We again found the significant main effect ofidentification, showing that identification waspositively related to perceived team performance.No significant effect of the interaction betweenNA and identification was found (see Table 4).

Discussion

This first study showed that group members’affective states are more homogeneous whengroup identification is higher. Indeed, as ex-pected, the data suggest that affective conver-gence was stronger for both positive affect andnegative affect with higher levels of group

identification. Apparently our notion that identi-fication plays an underlying role in the formationof group affect is corroborated.We also found that positive group affective

tone has a stronger positive influence on collec-tive willingness to display OCB and collectiveperceived team performance with higher groupidentification. These results show that the inter-play between positive group affect and groupidentification indeed affects team effectiveness.However, we did not find that the effects ofnegative group affective tone on willingness toengage in OCB and on perceived team perfor-mance were stronger with higher group identifi-cation. How may we explain this lack ofsignificant findings for negative group affect?There is evidence suggesting that positive affect

may be more important for social interactionthan negative affect (cf. Watson et al., 1992).Barsade et al. (2000) found that similarity inpositive affect predicted work team processes andperformance, but also that similarity in negativeaffect did not. Similarly, Damen, van Knippen-berg and van Knippenberg (2008) found that theextent to which leaders and followers matchaffectively (i.e. congruency between followeraffective state and leader affective displays) onlypredicts leadership effectiveness when positiveaffect is investigated, and not when negativeaffect is investigated. The prevalence of positiveaffect has also been found in intra-individualaffect research. Studies that focused on mood–memory congruence found stronger effects whenthere is congruency in PA than when there iscongruency in NA (Blaney, 1986; Singer andSalovey, 1988). Moreover, it may well be thatexpressing and acting upon negative moods in theworkplace is constrained by norms prescribingproper organizational conduct. Negative groupaffect and its interplay with identification maythus translate less easily into variations in teameffectiveness than positive affect does.Our first study also extends some earlier

findings on identification. We found that boththe collective willingness to engage in OCB andthe collective perceived team performance werehigher for higher group identification. Earlierresearch on the relationship between identifica-tion and OCB or performance has yielded similarresults (e.g. Christ et al., 2003; Riketta, 2005).However, the research typically focuses on theeffects of individual level identification. This

4

5

6

High Identification Low Identification

Per

ceiv

ed T

eam

Per

form

ance

High Positive Group AffectLow Positive Group Affect

Figure 2. Interaction between positive group affect and identifica-

tion on perceived team performance for Study 1

Table 4. Results of hierarchical regression analyses for perceived

team performance

Predictor b t p DR2

Positive affect 0.19 1.47 ns 0.03

Identification 0.33 2.76 0.007 0.10

PA � identification 0.26 2.11 0.039 0.06

Negative affect 0.08 0.65 ns 0.006

Identification 0.37 3.07 0.003 0.12

NA � identification � 0.10 � 0.82 ns 0.009

Notes: n5 71 (listwise). DR2 is the variance explained by each

predictor after the other predictors are entered into the

equation.

348 J. Tanghe, B. Wisse and H. van der Flier

r 2009 British Academy of Management.

study, in contrast, focused on group identifica-tion. Our decision to treat identification as agroup level construct therefore seems justified notonly from a theoretical perspective but also froma more empirical perspective. Indeed, the resultsof this study both with regard to the obtainedICC scores for identification as well as withregard to the effects of group identification onteam effectiveness measures substantiate ourreasoning.However, this first study leaves some impor-

tant issues unresolved. We argued that identifica-tion renders people more open to the affectiveinfluence of fellow group members, makingtransfer of affective states more likely. Ourcorrelational survey may show that identificationis associated with stronger affective convergence,but it does not indicate whether identification infact leads people to converge to the affectivestates of their fellow group members. The surveyalso does not inform us on whether the conver-gence process may help to explain the combinedeffects of group affect and identification on teameffectiveness. Therefore, the aim of our secondstudy is to replicate the main findings of the firststudy, to gain more insight into the underlyingprocesses, and to provide causal evidence for theproposed relationships.A scenario experiment was therefore designed.

Participants were to imagine being part of groupswith team members displaying either positive ornegative group affect. The moods used in thisstudy are relaxedness and boredom. Both moodsare low in activation (Larsen and Diener, 1992),but one is a pleasant, positive affective state, andthe other an unpleasant, negative affective state.We chose these two categories because they aremore difficult to detect than the more activatedmoods enthusiasm or anger, making their detec-tion more dependent on information searchingbehaviours, attention and interest. In addition,these two moods may be more frequentlydisplayed in work settings. Furthermore, partici-pants were assigned to either the high or lowidentification condition. Team effectiveness wasassessed by participants’ own willingness toengage in OCB. We did not measure (perceived)performance, because this is more difficult toassess in a hypothetical situation.Study 2 thus investigates whether (1) high

identification leads people to converge affectivelyto fellow group members, (2) positive group

affective tone will have more beneficial effects onan individual’s willingness to engage in OCB thannegative affect, especially if this individualidentifies highly, and (3) high identifiers’ affectiveconvergence will explain the combined effects ofidentification and group affective tone on will-ingness to engage in OCB.

Study 2

Method

Participants and design. Participants were 121students (Mage 5 22.01, SD5 2.47, 27.3% male)at the VU University of Amsterdam, whoparticipated voluntarily and were randomlyassigned to the conditions of a 2 (group affect:relaxed versus bored) � 2 (identification: lowversus high) design.

Procedure. Participants read a business scenarioin which they had to imagine that they werechosen, together with three colleagues, to comeup with a plan to make their organization moreefficient. They were given some informationabout the company, including a picture of thecompany building.

Identification manipulation. In the identificationmanipulation participants were to imagine theirbond with the organization. Similar toB. van Knippenberg, Martin and Tyler (2006)we used key aspects of the construct but refrainedfrom using jargon. More specifically, participantsin the high identification condition had to imaginethat they fit well in the team, felt completely athome in their team, and that there was a goodmatch between themselves and the other teammembers. In the low identification conditionparticipants had to imagine that they did not fitin the team at all, did not feel at home in theirteam, and that the match between themselves andthe other team members was not good. Next,participants were told that the team had beenworking together for several weeks, and that theywere currently developing concrete plans to maketheir organization more efficient.

Group affect manipulation. Participants weresubsequently shown four photographs. Onephotograph displayed all three team members(two men and a woman) sitting at a table,

Group Affect and Identification 349

r 2009 British Academy of Management.

apparently in the middle of plan development.The other three photographs were headshots ofevery team member. In the relaxed condition theposture of the team members was calm andcomfortable and team members had an engagedorientation towards each other. Facial expres-sions were relaxed with the corners of the mouthturned slightly upwards. In the bored conditionthe postures were slouching and oriented awayfrom the other group members. Facial expres-sions were bored and expressed disinterest. Eyeswere partially closed and group members had afixed stare away from the others (cf. Bartel andSaavedra, 2000).

Dependent measures. All dependent measureswere answered on a scale from 1 (completelydisagree) to 5 (completely agree).Participants’ own affect was measured with

four items derived from the circumplex model(Larsen and Diener, 1992). We distinguished twomood scales (i.e. relaxed and bored), bothmeasured with two items similar to the ones usedin the first study. An example is ‘I feel relaxed’.Willingness to engage in OCB was measured withthree items. An example is ‘I am willing to domore than reasonably can be expected of me’.To check the successfulness of the identifica-

tion manipulation, we again used the Dutchtranslation (van Knippenberg, A., et al., 2000) ofthe Mael and Ashforth (1992) scale (four items,a5 0.82, M5 3.29, SD5 1.06). An example itemis ‘When someone criticizes the team this feels asa personal insult’. Note that we measuredidentification before measuring any of the depen-dent variables. To check the successfulness of thegroup affect manipulation, participants wereasked to first take a close look at the picturesagain and then fill in the questions that com-prised our observed group affect scale. Ourobserved group affect scale contained two sub-scales (i.e. observed relaxedness and observedboredom). Both sub-scales consisted of two items(respectively, a5 0.85, M5 3.36, SD5 0.87, anda5 0.92, M5 3.28, SD5 1.39). An example itemis ‘They feel bored’.

Results

Preliminary analysis. Descriptive statistics andintercorrelations for all dependent variables aredisplayed in Table 5.

We first performed a principal componentsanalysis with OBLIMIN rotation of the itemscomprising the manipulation checks (i.e. identifi-cation, observed relaxedness, observed boredom).This analysis yielded a three-factor solution,accounting for 78.01% of the variance, with allitems loading above 0.75 on the intended compo-nent and cross-loadings not higher than 0.16.Next, we performed a principal components

analysis with OBLIMIN rotation of the itemscomprising the main dependent variables (i.e. ownrelaxedness, own boredom, willingness to engagein OCB). This analysis yielded a three-factorsolution, accounting for 79.4% of the variance,with all items loading above 0.77 on the intendedcomponent and cross-loadings not higher than0.17.

Manipulation checks. In all ANOVAs groupaffect (relaxed/bored) and identification (high/low) were factors in the design.An ANOVA on the identification scale re-

vealed that participants in the low identificationcondition identified less than participants in thehigh identification condition (F(1, 117)5 148.52,po0.001, Z2 5 0.56; M5 2.52, SD5 0.70 versusM5 4.08, SD5 0.72).An ANOVA on observed boredom revealed

that participants in the bored group affectcondition perceived their team members to bemore bored than participants in the relaxedgroup affect condition, F(1, 117)5 177.13,po0.001, Z2 5 0.60 (M5 4.34, SD5 0.85 versusM5 2.20, SD5 0.93). An ANOVA on observedrelaxedness revealed that participants in therelaxed group affect condition perceived theirteam members to be more relaxed than partici-pants in the bored group affect condition,F(1, 117)5 63.34, po0.001, Z2 5 0.35 (M5

3.88, SD5 0.71 versus M5 2.85, SD5 0.70).

Table 5. Means, standard deviations and intercorrelations for

Study 2

M SD 1 2 3 4

1. Identification 3.29 1.06 (0.82)

2. Own relaxed 3.00 0.94 0.25** (0.81)

3. Own bored 3.11 1.21 � 0.23* � 0.37** (0.92)

4. OCB 3.24 0.85 0.51** 0.35** � 0.42** (0.73)

Notes: Cronbach’s alphas are displayed on the diagonal.

n5 121 (listwise).*po0.05; **po0.01.

350 J. Tanghe, B. Wisse and H. van der Flier

r 2009 British Academy of Management.

No other effects were obtained. We concludethat the manipulations were successful.

Affective convergence. An ANOVA with groupaffect and identification as between-subjectsvariables and own affect as within-subjectsvariable was conducted.We found a group affect main effect,

F(1, 117)5 13.30, po0.001, Z2 5 0.10, showingthat both low activation level moods were lessstrongly experienced when participants wereconfronted with relaxed team members(M5 2.87, SD5 0.45) than when they wereconfronted with bored team members(M5 3.24, SD5 0.63).This main effect was qualified by a group affect

� own affect interaction, F(1, 117)5 63.63,po0.001, Z2 5 0.35. Participants with relaxedteam members felt more relaxed (M5 3.28,SD5 0.90) than participants with bored teammembers (M5 2.72, SD5 0.91), and participantswith bored team members felt more bored(M5 3.76, SD5 0.92) than participants withrelaxed team members (M5 2.45, SD5 1.10).A group affect � identification interaction,

F(1, 117)5 9.20, p5 0.003, Z2 5 0.076, showedthat participants were influenced less by theaffective state of their team members when theyidentified weakly (Mrelaxed 5 3.03, SD5 0.52 ver-sus Mbored 5 3.10, SD5 0.46) than when theyidentified strongly (Mrelaxed 5 2.70, SD5 0.51versus Mbored 5 3.39, SD5 0.74).As expected, we found a group affect �

identification � own affect interaction, F(1, 117)5 14.08, po0.001, Z2 5 0.11. High identifiers feltmore bored when confronted with bored teammembers than with relaxed team members(M5 3.78, SD5 1.07 versus M5 1.72, SD5

0.75), F(1, 119)569.20, po0.001, Z2 50.37. Highidentifiers also felt more relaxed when confrontedwith relaxed team members than with bored teammembers (M5 3.68, SD5 0.80 versus M53.00,SD5 0.96), F(1, 119)5 8.34, p5 0.005, Z2 50.07.Importantly, for low identifiers no such differenceswere obtained (F(1, 119)5 3.48, ns, andF(1, 119)5 3.42, ns).Finally, we found an identification � own

affect interaction (F(1, 117)5 34.99, po0.001,Z25 0.23). High identifiers felt more relaxed(M53.34, SD5 0.94 versus M52.66, SD50.83)and less bored (M5 2.75, SD5 1.39 versusM5 3.47, SD5 0.87) than low identifiers.

Willingness to engage in OCB. An ANOVA onthe willingness to engage in OCB scale revealedmain effects for group affect, F(1, 117)5 8.99,p5 0.003, Z2 5 0.07, and identification,F(1, 117)5 37.79, po0.001, Z2 5 0.24. Partici-pants with relaxed team members were morewilling to engage in OCB (M5 3.44, SD5 0.86)than participants with bored team members(M5 3.05, SD5 0.82). Also, high identifiers weremore willing to engage in OCB (M5 3.65,SD5 0.78) than low identifiers (M5 2.85,SD5 0.73). The ANOVA also revealed theexpected group affect � identification interac-tion, F(1, 117)5 6.75, po0.02, Z2 5 0.06 (seeFigure 3). High identifiers were more willing toengage in OCB with relaxed team members thanwith bored team members (M5 4.02, SD5 0.56versus M5 3.28, SD5 0.81), F(1, 119)5 12.39,p5 0.001, Z2 5 0.09. Importantly, no difference(M5 2.87, SD5 0.70 versus M5 2.83, SD5

0.77) in willingness to engage in OCB was foundfor low identifiers, F(1, 119)5 0.06, ns.

Mediational analysis. According to Baron andKenny (1986) a variable can be called a mediatorwhen (1) variations in levels of the independentvariable account for variations in the presumedmediator, (2) variations in the mediator accountfor variations in the dependent variable, and (3)when the effect of the presumed mediator iscontrolled for, a previously significant effect ofthe independent variable on the dependentvariable disappears or decreases.

In this study we expected participants’ ownaffective states to mediate the effects of group

2

2.5

3

3.5

4

4.5

High Identification Low Identification

Will

ingn

ess

to e

ngag

e in

OC

B

Group Affect RelaxedGroup Affect Bored

Figure 3. Interaction between team member affect and identifica-

tion on willingness to engage in OCB for Study 2

Group Affect and Identification 351

r 2009 British Academy of Management.

affect on willingness to engage in OCB. However,this mediation was only expected for people whoidentify highly (they are, after all, the people whoare expected to converge to the moods of theirteam members).We conducted a 2 (group affect) � 2 (identifi-

cation) ANCOVA with willingness to engage inOCB as the dependent variable and participants’boredom as a covariate (F(1, 118)5 11.45,b5 � 0.25, p5 0.001, for the regression, whichtestifies to the existence of a negative link betweenown affect and willingness to engage in OCB).The analysis showed that the formerly significanteffect (F(1, 119)5 12.39, p5 0.001, Z2 5 0.09) ofgroup affect within the high identification condi-tion (i.e. high identifiers were more willing toengage in OCB when they had relaxed teammembers than when they had bored teammembers) disappeared (F(1, 118)5 0.75, p5

0.39). In addition, the Sobel test was significant(z5 2.65, p5 0.008).We then conducted a 2 (group affect) � 2 (identi-

fication) ANCOVA with willingness to engage inOCB as the dependent variable and participants’relaxedness as a covariate (F(1, 118)511.73,b5 0.27, p50.001, for the regression, whichtestifies to the existence of a positive link betweenown affect and willingness to engage in OCB). Theanalysis showed that the formerly significant effect(F(1, 119)512.39, p50.001, Z250.09) of groupaffect within the high identification condition (i.e.high identifiers were more willing to engage in OCBwhen they had relaxed team members than whenthey had bored team members) became lesssignificant (F(1, 118)57.16, p50.009). In addi-tion, the Sobel test was significant (z52.21,po0.03).As has been described in the above, the simple

main analyses showed no effects in the relationshipbetween group affect and the willingness to engagein OCB for low identification. The ANCOVAswith willingness to engage in OCB as dependentvariable and participants’ boredom and relaxednessas covariates did not reveal any changes in theeffects of group affect within the low identificationcondition, nor did the corresponding Sobel tests.We conclude that only high identifiers converge

to the affective state of fellow group members andthat this explains why identification moderates theeffects of group affect on willingness to engage inOCB (all preconditions, as proposed by Baron andKenny (1986), were met).

Discussion

With this second study we were able to replicatethe main findings of our first study, to gain moreinsight into the underlying processes, and toprovide causal evidence for our proposed rela-tionships.First, replicating the findings of Study 1, this

study showed that high identifiers convergedaffectively to their team members’ affective statewhile low identifiers did not. High identifiers’own mood was more relaxed when the existingmood in the group was relaxed and more boredwhen the existing mood in the group was bored.The moods of low identifiers were not dependenton the affective states of their team members.Clearly, high identification leads people to con-verge affectively with fellow group members.Note that the interaction between identificationand group members’ affect makes it less likelythat demand characteristics may have played arole here. If participants felt that affectivecongruence between themselves and their teammembers may have been expected of them (andtherefore reported but not truly felt), the moder-ating role of identification would have been lesslikely to emerge.Second, we found evidence for our hypothesis

that positive group affective tone will have morefavourable effects on team effectiveness thannegative group affective tone, especially for highidentifiers. In Study 1 we found that the interplaybetween positive group affective tone and identi-fication explained team effectiveness, but that theinterplay between negative group affective toneand identification did not. The results of thisstudy, however, indicate that negative groupaffect has a less positive impact on willingnessto engage in OCB than positive affect, and thatthis is especially true for high identifiers.Third, the effects of team members’ affective

states on the willingness to engage in OCB weremediated by high identifiers’ own affective states.This testifies to the notion that high identifiersconverged to the affective states of their teammembers, which in turn made them more willingto engage in OCB when the existing groupaffective state was positive compared to negative.These results support our proposed underlyingrole of identification in the formation of groupaffect and its subsequent influence on teameffectiveness.

352 J. Tanghe, B. Wisse and H. van der Flier

r 2009 British Academy of Management.

Furthermore, this second study showed thatwillingness to engage in OCB was higher for highidentifiers than for low identifiers, therebyreplicating Study 1 and earlier findings (e.g.Christ et al., 2003; Riketta, 2005). Also, mirror-ing the effects of Study 1 and earlier findingsshowing that people feel more positive and atease in a group or organization they can identifywith (cf. Van Dick et al., 2004b), we found that ingeneral low identifiers were more bored and highidentifiers were more relaxed. These replicationsof our own survey findings and other fieldresearch further strengthen our confidence inthe results and validity of this scenario study.

General discussion

Group affect may be deemed an important aspectof work group functioning, and, as such, theincreasing research attention to the topic is welldeserved. Yet, so far little is known about theconditions under which group affect may arise.The present research identified identification, akey aspect of group membership, to be animportant factor in the development of groupaffect. The first study revealed that affectiveconvergence was stronger for both positive andnegative affect with higher levels of groupidentification. Our second study showed thathigh identification indeed leads people to con-verge affectively with the positive or negativeaffective states of fellow group members. As weexplained, identification is considered to enhancethe likelihood that group members are open tothe influence of other group members. Higheridentification generally increases the motivationto pick up on signals from other group members,and leads to greater attentiveness to other groupmembers’ behaviours, feelings and attitudes.Identification thus may enhance the likelihoodthat group members (consciously or uncon-sciously) detect and conform to other groupmembers’ affective states.The social identity perspective may also help to

understand the results of other research on affectin groups. For instance, emotional convergence isfound to develop more easily when groupmembership is stable, when people are dependenton one another (Bartel and Saavedra, 2000),when group members take part in collectiveactivities (Totterdell, 2000) and when people

share (strong) work ties in organizational net-works (Totterdell et al., 2004). However, theseconstructs or variables may all be regarded asproxies or correlates of identification. Indeed, webelieve that the results of previous research maybe interpreted as pointing to the idea that theextent to which group members feel tied orattached to the group and its members may fosterthe development of group affect. Perhaps, thus,the social identity perspective may aid furthertheory development in these specific researchlines as well.The present study extends earlier research on

the role of identification in the formation ofgroup-based affect (e.g. Giner-Sorolla, Mackieand Smith, 2007; Iyer, Leach and Crosby, 2003;Smith, Seger and Mackie, 2007). Indeed, itreveals that identification is important not onlyfor explaining emotions that may arise inintergroup contexts, but also for understandingwhen members belonging to the same group maybe more likely to converge affectively as a resultof the transfer of affect between group members.Thus, taken together, our findings and earlierfindings indicate that identification may play animportant role in affective intergroup and in-tragroup processes.The present research showed that the interplay

between identification and group affective tonemay influence work group effectiveness. The firststudy revealed that positive group affective tonehad a stronger positive influence on collectivewillingness to display OCB and collective per-ceived team performance for higher groupidentification. Our second study, contrastingpositive group affective tone with negative groupaffective tone, showed that positive group affecthad more favourable effects on willingness toengage in OCB than negative affect, especially forhigh identifiers. Moreover, the second studyshowed that these effects may be explained byhigh identifiers’ greater convergence to theaffective states of fellow group members. Overall,these results indicate that the effects of groupaffective tone may be particularly strong whengroup members identify highly with their group.Identification thus strengthens the effects ofaffective group characteristics on perceived groupmember behaviour.All in all, we have found compelling evidence

for the role of identification in group affectiveprocesses. We found evidence for the assertions

Group Affect and Identification 353

r 2009 British Academy of Management.

that (1) identification fosters affective conver-gence, (2) positive group affect has a positiveinfluence on team effectiveness especially withhigher levels of identification, and (3) highidentifiers’ affective convergence explains themoderating effects of identification on the influ-ence of group affect on team effectiveness.Confidence in our findings is bolstered by thefact that our conclusions were based on twostudies employing different methodologies (asurvey of teams and a scenario experiment), bythe fact that replication over studies was found,and that the studies used different samples (i.e.teams in organizations, students).Yet, multiple aspects of the relationship

between social identity and group affect processescould benefit from further research. The value ofthe social identity perspective for the study ofgroup affect lies in the fact that it offers atheoretical framework for the further study offactors that may shape group affective processes.For instance, according to the social identityapproach, people cognitively represent groups asprototypes, fuzzy sets of attributes (perceptions,attitudes, feelings, behaviours) that in a particu-lar context capture the essence of the ingroupand differentiate the ingroup from relevantoutgroups (Hogg and van Knippenberg, 2003).Group members may differ in the extent towhich they match this prototype. The moreprototypical the group member, the more he orshe represents the group’s standards, values andnorms: prototypical group members thus exem-plify group normative behaviour and reflectwhat members of the group have in common(Turner et al., 1987). As outlined in self-categorization theory (Turner et al., 1987), groupmember prototypicality plays an importantrole in group-based influence. As group member-ship becomes more salient, group memberinfluence becomes more contingent on the extentto which the group member is prototypical ofthe group (Hogg, 2001). Indeed, prototypicalgroup members are generally more influentialthan peripheral members (Hogg and van Knip-penberg, 2003; van Knippenberg, B., and vanKnippenberg, 2005). Likewise, we may expect theaffective states of prototypical members totransfer more easily to fellow group membersthan affective states of more peripheral members.We would also like to point out that it is

possible that the relationship between identifica-

tion and group affect may be such that positiveaffect leads to increased identification. Peoplewho exhibit a positive mood are seen as moreattractive (Staw and Barsade, 1993), which, inturn, may enhance identification processes.Moreover, sharing affect is considered to bindpeople and groups together (Barsade and Gibson,1998), and it may thus lead to feelings ofbelongingness and therefore strengthen the col-lective identity (George, 2002; Klep, van Knip-penberg and van der Flier, 2007). Furtherresearch is necessary to substantiate these linesof reasoning.The literature, including the present study,

mostly ignores the affective construct senti-ments (i.e. affective evaluations of whethersomething is liked or disliked). This is unfortu-nate, because sentiments may also substantiallyaffect group processes. For instance, Kelly andBarsade (2001) discuss one specific group levelsentiment, i.e. the sentiment of group cohesion(i.e. the liking of or attraction to one’s group).Group cohesion is considered to lead to moreeffective group performance (cf. Mullen andCopper, 1994). Theoretical analysis of the socialidentity perspective suggests that cohesion is aresult of identification (Hogg and Terry, 2000).Therefore, for a more complete picture, futureresearch could focus not only on the role ofidentification in the development of group moodsand group emotions, but also on that of groupsentiments in general and group cohesion inparticular.A potential weakness of our research is that

both studies relied on subjective effectivenessmeasures. Although this is common practice inresearch on affect in groups (e.g. Barsade, 2002;George, 1990, 1995; Totterdell, 2000), studyingmore objective team performance criteria wouldhave added extra value. Moreover, we focused onthe willingness to display OCB and on perceivedteam performance as indicators of team effec-tiveness. We may expect, however, that identifi-cation would also strengthen the establishedeffects of group affective tone on other outcomevariables, such as performance, pro-social beha-viour, creativity and commitment (George, 1990,1991; Lord, Klimosky and Kanfer, 2002). How-ever, in order to become certain of the role ofidentification in influencing group affective pro-cesses on other outcome variables, more researchis warranted.

354 J. Tanghe, B. Wisse and H. van der Flier

r 2009 British Academy of Management.

A limitation of our research is that, althoughwe did consider whether longevity of teaminteraction affected the relationships under in-vestigation, we reflected neither on other aspectsof the teams’ common history nor on theirfuture. Longevity, or the time that teams havebeen working together, did not explain anyvariance in our analyses. This does not mean,however, that other aspects of their commonhistory, like events that took place, do not havean impact (George, 2002). Future research mayinvestigate the possibility that the teams’ commonhistory and future may affect the relationshipsunder investigation.Additionally, the literature, including the pre-

sent study, seems to suggest that negative affecthas either no effects or less positive effects thanpositive affect on the functioning of groups andteams. Although negative affect may generally beless influential than positive affect in situationsthat require social interaction (Barsade et al.,2000; Watson et al., 1992), we do not want toimply that negative affect may not impact groupfunctioning in certain circumstances. For in-stance, it may be that negative affect asserts moreinfluence on the behaviour of group memberswhen the affective state is felt for a prolongedperiod of time. The current studies both assessedpresent mood state and, as such, they did not tapinto more chronic negative group affect. Futureresearch may focus on the explanatory power oftemporal aspects in the relationship betweennegative group affect and group member beha-viour and attitudes. Moreover, both the situa-tional context in which negative affect is felt andthe dependent variables that are being investi-gated may be of importance here. George andZhou (2002) have shown that when perceivedrecognition and rewards for creative performancewere high, negative moods were positively relatedto creative performance. Other research suggeststhat negative moods may engender more carefuland systematic processing styles (Bless, 2000;Schwarz, 2001), enhancing cognitive task perfor-mance. The results of the present researchprovide limited insights regarding the conditionsunder which negative group affect might influ-ence team effectiveness.Finally, a practical ramification of the present

study may be that team managers may takenotice of the fact that the moods and emotions oftheir employees may transfer to fellow group

members. The way that an employee feels duringworking hours is therefore not something thatmay only affect that particular individual; insteadit may affect the entire team. Notably, thesecollective affective states can be both positive andnegative. Thus, the transfer of affective statesamong group members may result in teams that,as a collective, feel generally good or bad. Ifmanagers wish to avoid the development ofnegative group affect, due attention to groupcomposition may be of importance. For instance,if a team hosts many employees who score highon negative trait affect and are therefore morelikely to experience negative moods (Barsadeet al., 2000; George, 1991), their affectiveinfluence on others may tip the balance towardsa negative affective tone. Because high identifiersare especially sensitive to the mood of groupmembers, attention to the (affective) compositionof the group may be especially warranted for highidentifying teams. Of course, it may be difficultfor managers to adjust group composition. Inthat case, the organization could make use ofstrategies that influence affective expressions andemotion-related behaviours in the workplace (Ash-forth and Humphrey, 1995; Hochschild, 1990).Another, related, ramification is one that may bederived from a combination of our work with thatof Bono and Ilies (2006). Bono and Ilies focused onthe effects of leader affective states. Their workshows that leaders’ positive affective states maytransfer to their subordinates. Evidently, managersshould consider that not only can the affective stateof one group member transfer to the other groupmembers, but also that their own affective statecould transfer to other group members andsubsequently influence the team’s effectiveness.This may be particularly likely when they superviseteams with a high level of identification.To conclude, we assert that more research is

needed to fully comprehend all aspects of theeffects of identification on group affective pro-cesses and its subsequent influence on teameffectiveness and behaviour in organizations.We also suggest that a further integration of thesocial identity perspective with theory on (group)affect may be helpful in these endeavours. Mostimportantly, however, we argue that (group)affect has considerable consequences for pro-cesses in groups, teams and organizations, andtherefore deserves ample attention, in bothpractice and research.

Group Affect and Identification 355

r 2009 British Academy of Management.

References

Aiken, L. S. and S. G. West (1991). Multiple Regression:

Testing and Interpreting Interactions. Thousand Oaks, CA:

Sage.

Anderson, C., D. Keltner and O. P. John (2003). ‘Emotional

convergence between people over time’, Journal of Person-

ality and Social Psychology, 84, pp. 1054–1068.

Ashforth, B. E. and R. H. Humphrey (1995). ‘Emotion in the

workplace: a reappraisal’, Human Relations, 48, pp. 97–125.

Ashforth, B. E. and F. Mael (1989). ‘The social identity theory

and the organization’, Academy of Management Review, 14,

pp. 20–39.

Ashkanasy, N. M., W. J. Zerbe and C. E. J. Hartel (2005). The

Effect of Affect in Organizational Settings. Cambridge, MA:

Elsevier.

Bandura, A. (1986). Social Foundations of Thought and Action:

A Social Cognitive Theory. Englewood Cliffs, NJ: Prentice

Hall.

Baron, R. M. and D. A. Kenny (1986). ‘The moderator–

mediator variable distinction in social psychology research:

conceptual, strategic, and statistical considerations’, Journal

of Personality and Social Psychology, 51, pp. 1173–1182.

Barsade, S. G. (2002). ‘The ripple effect: emotional contagion

and its influence on group behaviour’, Administrative Science

Quarterly, 47, pp. 644–675.

Barsade, S. G. and D. E. Gibson (1998). ‘Group emotion: a

view from top and bottom’. In D. Gruenfeld, E. Mannix and

M. Neale (eds), Composition: Research in Managing Groups

and Teams, Vol. 1, pp. 81–102. Stamford, CT: JAI Press.

Barsade, S. G., A. J. Ward, J. D. F. Turner and J. A.

Sonnenfeld (2000). ‘To your heart’s content: a model of

affective diversity in top management teams’, Administrative

Science Quarterly, 45, pp. 802–836.

Bartel, C. A. and R. Saavedra (2000). ‘The collective construc-

tion of work group moods’, Administrative Science Quarterly,

45, pp. 197–231.

Blaney, P. H. (1986). ‘Affect and memory: a review’,

Psychological Bulletin, 99, pp. 229–246.

Bless, H. (2000). ‘The interplay of affect and cognition: the

mediating role of general knowledge structures’. In J. P.

Forgas (ed.), Feeling and Thinking: The Role of Affect in

Social Cognition, pp. 201–222. New York: Cambridge

University Press.

Bono, J. C. and R. Ilies (2006). ‘Charisma, positive emotions

and mood contagion’, Leadership Quarterly, 17, pp. 317–334.

Burke, M. J. and W. P. Dunlap (2002). ‘Estimating interrater

agreement with the average deviation index: a user’s guide’,

Organizational Research Methods, 5, pp. 159–172.

Christ, O., R. Van Dick, U. Wagner and J. Stellmacher (2003).

‘When teachers go the extra mile: foci of organizational

identification as determinants of different forms of organiza-

tional citizenship behaviour among schoolteachers’, British

Journal of Educational Psychology, 73, pp. 329–341.

Damen, F., B. van Knippenberg and D. van Knippenberg

(2008). ‘Affective match: leader emotional displays, follower

positive affect, and follower performance’, Journal of Applied

Social Psychology, 38, pp. 868–902.

Davis, M. H. (1983). ‘Measuring individual differences in

empathy: evidence for a multidimensional approach’, Journal

of Personality and Social Psychology, 44, pp. 113–126.

Festinger, L. A. (1954). ‘Theory of social comparison pro-

cesses’, Human Relations, 7, pp. 117–140.

Forgas, J. P. (1992). ‘Affect in social judgements and decisions:

a multi process model’. In M. P. Zanna (ed.), Advances in

Experimental Social Psychology, pp. 227–275. Orlando, FL:

Academic Press.

Forgas, J. P. and J. M. George (2001). ‘Affective influences on

judgements and behaviour in organizations: an information

processing perspective’, Organizational Behaviour and Human

Decision Processes, 86, pp. 3–34.

Fredrickson, B. L. (1998). ‘What good are emotions’, Review of

General Psychology, 2, pp. 300–319.

Fredrickson, B. L. (2000). ‘Why positive emotions matter in

organizations: lessons from the broaden-and-build model’,

The Psychologist-Manager Journal, 4, pp. 131–142.

Frijda, N. H. (1993). ‘Moods, emotion episodes, and emotions’.

In M. Lewis and J. M. Haviland (eds), Handbook of

Emotions, pp. 381–403. New York: Guilford Press.

George, J. M. (1990). ‘Personality, affect, and behaviour in

groups’, Journal of Applied Psychology, 75, pp. 107–116.

George, J. M. (1991). ‘State or trait: effects of positive mood on

prosocial behaviours at work’, Journal of Applied Psychology,

76, pp. 299–307.

George, J. M. (1995). ‘Leader positive mood and group

performance: the case of customer service’, Journal of Applied

Social Psychology, 25, pp. 778–794.

George, J. M. (2002). ‘Affect regulation in groups and teams’.

In R. G. Lord, R. J. Klimosky and R. Kanfer (eds), Emotions

in the Workplace: Understanding the Structure and Role of

Emotions in Organizational Behaviour, pp. 183–218. San

Francisco, CA: Jossey-Bass, Wiley.

George, J. M. and J. Zhou (2002). ‘Understanding when bad

moods foster creativity and good ones don’t: the role of

context and clarity of feelings’, Journal of Applied Psychol-

ogy, 87, pp. 687–697.

Giner-Sorolla, R., D. M. Mackie and E. R. Smith (2007).

‘Special issue on intergroup emotions: Introduction’, Group

Processes and Intergroup Relations, 10, pp. 5–8.

Hartel, C. E. J., W. J. Zerbe and N. M. Ashkanasy (2005).

Emotions in Organizational Behaviour. Mahwah, NJ: Erl-

baum.

Haslam, S. A. (2001). Psychology in Organizations: The Social

Identity Approach. London: Sage.

Hatfield, E., J. Cacioppo and R. L. Rapson (1994). Emotional

Contagion. Cambridge: Cambridge University Press.

Hochschild, A. R. (1990). ‘Ideology and emotion manage-

ment: a perspective and path for future research’. In T. D.

Kemper (ed.), Research Agendas in the Sociology of Emotions,

pp. 117–142. Albany, NY: State University of New York

Press.

Hogg, M. A. (2001). ‘A social identity theory of leadership’,

Personality and Social Psychology Review, 5, pp. 184–200.

Hogg, M. A. and D. Abrams (1988). Social Identifications: A

Social Psychology of Intergroup Relations and Group Pro-

cesses. London: Routledge.

Hogg, M. A. and D. van Knippenberg (2003). ‘Social identity

and leadership processes in groups’. In M. P. Zanna (ed.),

Advances in Experimental Social Psychology, Vol. 35, pp. 1–

52. San Diego, CA: Academic Press.

Hogg, M. A. and D. J. Terry (2000). ‘Social identity and self

categorization processes in organizational contexts’, Acad-

emy of Management Review, 25, pp. 121–141.

356 J. Tanghe, B. Wisse and H. van der Flier

r 2009 British Academy of Management.

Isen, A. M. (2000). ‘Positive affect and decision making’. In M.

Lewis and J. M. Haviland-Jones (eds), Handbook of

Emotions, pp. 417–435. New York: Guilford Press.

Isen, A. M. (2004). ‘Some perspectives on positive feelings and

emotions: positive affect facilitates thinking and problem

solving’. In A. S. R. Manstead, N. Frijda and A. Fischer

(eds), Feelings and Emotion: The Amsterdam Symposium, pp.

263–281. Cambridge: Cambridge University Press.

Isen, A. M., K. A. Daubman and G. P. Nowicki (1987).

‘Positive affect facilitates creative problem solving’, Journal

of Personality and Social Psychology, 52, pp. 1122–1131.

Iyer, A., C. W. Leach and F. J. Crosby (2003). ‘White guilt and

racial compensation: the benefits and limits of self-focus’,

Personality and Social Psychology Bulletin, 29, pp. 117–129.

James, L. R. (1982). ‘Aggregation bias in estimates of

perceptual agreement’, Journal of Applied Psychology, 67,

pp. 219–229.

James, L. R., R. G. Demaree and G. Wolf (1984). ‘Estimating

within-group interrater reliability with and without response

bias’, Journal of Applied Psychology, 69, pp. 85–98.

Kelly, J. R. and S. G. Barsade (2001). ‘Mood and emotions in

small groups and work teams’, Organizational Behaviour and

Human Decision Processes, 86, pp. 99–130.

Klep, A. H. M., B. van Knippenberg and H. van der Flier

(2007) ‘Beyond individual affect: group affect and task

performance’. Paper presented at the 22nd Annual SIOP

Conference, New York.

van Knippenberg, A., B. van Knippenberg, C. van Knippen-

berg and D. van Knippenberg (2000). ‘Identificatie met de

organisatie: een meetinstrument’ [Organizational identifica-

tion: a Dutch measurement], Gedrag en Organisatie, 14, pp.

67–73.

van Knippenberg, B. and D. van Knippenberg (2005). ‘Leader

self-sacrifice and leadership effectiveness: the moderating role

of leader prototypicality’, Journal of Applied Psychology, 90,

pp. 25–37.

van Knippenberg, B., D. van Knippenberg and D. De Cremer

(2007). ‘The interplay between justice and affect in groups’. In

D. De Cremer (ed.), Advances in the Psychology of Justice and

Affect, pp. 191–204. Charlotte, NC: Information Age

Publishing.

van Knippenberg, B., L. Martin and T. Tyler (2006). ‘Process-

orientation versus outcome-orientation during organizational

change: the role of organizational identification’, Journal of

Organizational Behaviour, 27, pp. 685–704.

van Knippenberg, D. (2000). ‘Work motivation and perfor-

mance: a social identity perspective’, Applied Psychology: An

International Review, 49, pp. 357–371.

van Knippenberg, D. and M. A. Hogg (2003). ‘A social identity

model of leadership effectiveness in organizations’, Research

in Organizational Behaviour, 25, pp. 243–296.

Larsen, R. J. and E. Diener (1992). ‘Promises and problems

with the circumplex model of emotion’. In M. S. Clark (ed.),

Emotion: Review of Personality and Social Psychology, Vol.

13, pp. 25–59. Newbury Park, CA: Sage.

Lazarus, R. S. (1991). Emotion and Adaptation. New York:

Oxford University Press.

Levenson, R. W. and J. M. Gottman (1983). ‘Marital

interaction: physiological linkage and affective exchange’,

Journal of Personality and Social Psychology, 45, pp. 587–

597.

Lord, R. G., R. J. Klimosky and R. Kanfer (eds) (2002).

Emotions in the Workplace: Understanding the Structure and

Role of Emotions in Organizational Behaviour, 1st edn. San

Francisco, CA: Jossey-Bass, Wiley.

Mael, F. and B. E. Ashforth (1992). ‘Alumni and their alma

mater: a partial test of the reformulated model of organiza-

tional identification’, Journal of Organizational Behaviour, 13,

pp. 103–123.

Mason, C. M. and M. A. Griffin (2003). ‘Group absenteeism

and positive affective tone: a longitudinal study’, Journal of

Organizational Behaviour, 24, pp. 667–687.

Moorman, R. H. and G. L. Blakely (1995). ‘Individualism–

collectivism as an individual difference predictor in organiza-

tional citizenship behaviour’, Journal of Organizational

Behaviour, 16, pp. 127–142.

Mullen, B. and C. Copper (1994). ‘The relation between group

cohesiveness and performance: an integration’, Psychological

Bulletin, 115, pp. 210–227.

Norton, M. I., B. Monin, J. Cooper and M. A. Hogg (2003).

‘Vicarious dissonance: attitude change from inconsistency of

others’, Journal of Personality and Social Psychology, 85, pp.

47–62.

Organ, D. W. (1988). Organizational Citizenship Behaviour: The

Good Soldier Syndrome. Lexington, MA: Lexington Books.

Podsakoff, P. M., S. B. MacKenzie, J. B. Paine and D. G.

Bachrach (2000). ‘Organizational citizenship behaviours: a

critical review of the theoretical and empirical literature and

suggestions for future research’, Journal of Management, 26,

pp. 513–563.

Reicher, S., S. A. Haslam and N. Hopkins (2005). ‘Social

identity and the dynamics of leadership: leaders and followers

as collaborative agents in the transformation of social

reality’, Leadership Quarterly, 16, pp. 547–568.

Riketta, M. (2005). ‘Organizational identification: a meta-

analysis’, Journal of Vocational Psychology, 66, pp. 358–384.

Riketta, M. and S. Nienaber (2007). ‘Multiple identities and

work motivation: the role of perceived compatibility between

nested organizational units’, British Journal of Management,

18, pp. 61–77.

Russell, J. A. (1980). ‘A circumplex model of affect’, Journal of

Personality and Social Psychology, 39, pp. 1161–1178.

Russell, J. A. and J. M. Carroll (1999). ‘On the bipolarity of

positive and negative affect’, Psychological Bulletin, 125, pp.

3–30.

Russell, J. A. and L. Feldman Barrett (1999). ‘Core affect,

prototypical emotional episodes, and other things called

emotion: dissecting the elephant’, Journal of Personality and

Social Psychology, 76, pp. 805–819.

Shrout, P. E. and J. L. Fleiss (1979). ‘Intraclass correlations:

uses in assessing rater reliability’, Psychological Bulletin, 86,

pp. 420–428.

Schwarz, N. (2001). ‘Feelings as information: implications for

affective influences on information processing’. In L. L.

Martin and G. L. Clore (eds), Theories of Mood and

Cognition: A User’s Guidebook, pp. 159–176. Mahwah, NJ:

Lawrence Erlbaum.

Singer, J. A. and P. Salovey (1988). ‘Mood and memory:

evaluating the network theory of affect’, Clinical Psychology

Review, 8, pp. 211–251.

Smith, E. R., C. H. Seger and D. M. Mackie (2007). ‘Can

emotions be truly group level? Evidence regarding four

Group Affect and Identification 357

r 2009 British Academy of Management.

conceptual criteria’, Journal of Personality and Social

Psychology, 93, pp. 431–446.

Staw, B. M. and S. G. Barsade (1993). ‘Affect and managerial

performance: a test of the sadder-but-wiser vs. happier-and-

smarter hypotheses’, Administrative Science Quarterly, 38,

pp. 304–331.

Tajfel, H. (1974). ‘Social identity and intergroup behaviour’,

Social Science Information, 13, pp. 65–93.

Tajfel, H. and J. C. Turner (1986). ‘The social identity theory of

intergroup behaviour’. In S. Worchel and W. G. Austin (eds),

Psychology of Intergroup Relations, pp. 7–24. Chicago, IL:

Nelson-Hall.

Totterdell, P. (2000). ‘Catching moods and hitting runs: mood

linkage and subjective performance in professional sport

teams’, Journal of Applied Psychology, 85, pp. 848–859.

Totterdell, P., S. Kellett, R. B. Briner and K. Teuchmann

(1998). ‘Evidence of mood linkage in work groups’, Journal

of Personality and Social Psychology, 74, pp. 1504–1515.

Totterdell, P., T. Wall, D. Holman, H. Diamond and O.

Epitropaki (2004). ‘Affect networks: a structural analysis of

the relationship between work ties and job-related affect’,

Journal of Applied Psychology, 89, pp. 854–867.

Turner, J. C. (1985). ‘Social categorization and the self-concept:

a social cognitive theory of group behaviour’, Advances in

Group Processes, 2, pp. 77–121.

Turner, J. C. (2001). ‘Foreword: What the social identity

approach is and why it matters’. In S. A. Haslam (ed.),

Psychology in Organizations: The Social Identity Approach.

London: Sage.

Turner, J. C., M. A. Hogg, P. J. Oakes, S. Reicher and M.

Wetherell (1987). Rediscovering the Social Group: A Self-

Categorization Theory. Oxford: Blackwell.

Van Dick, R., O. Christ, J. Stellmacher, U. Wagner, O.

Ahlswede, C. Grubba, M. Hauptmeier, C. Hohfeld, K.

Moltzen and P. A. Tissington. (2004a). ‘Should I stay or

should I go? Explaining turnover intentions with organiza-

tional identification and job satisfaction’, British Journal of

Management, 15, pp. 351–360.

Van Dick, R., U. Wagner, J. Stellmacher and O. Christ (2004b).

‘The utility of a broader conceptualization of organizational

identification: which aspects really matter?’, Journal of Occu-

pational and Organizational Psychology, 77, pp. 171–191.

Van Dick, R., M. W. Grojean, O. Christ and J. Wieseke (2006).

‘Identity and the extra mile: relationships between organiza-

tional identification and organizational citizenship beha-

viour’, British Journal of Management, 17, pp. 283–301.

Watson, D., L. A. Clark and A. Tellegen (1988). ‘Development

and validation of brief measures of positive and negative

affect: the PANAS scales’, Journal of Personality and Social

Psychology, 54, pp. 1063–1070.

Watson, D., L. A. Clark, C. W. McIntyre and S. Hamaker

(1992). ‘Affect, personality, and social activity’, Journal of

Personality and Social Psychology, 63, pp. 1011–1025.

West, M. A., C. S. Borrill and K. L. Unsworth (1998). ‘Team

effectiveness in organizations’. In C. L. Cooper and I. T.

Robertson (eds), International Review of Industrial and

Organizational Psychology, Vol. 13, pp. 1–48. Chichester:

Wiley.

Jacqueline Tanghe MSc working at VU University, Amsterdam, The Netherlands. Currently sheworks as a Postdoc at the University of Groningen, The Netherlands, at the Institute of Integrationand Social Efficacy. Research interests include (group) affect, the role of self and identity inorganizational contexts, trust and cooperation.

Barbara Wisse (also known as Barbara van Knippenberg) is affiliated to the University ofGroningen, The Netherlands. She obtained her PhD at Leiden University, The Netherlands. Hercurrent research interests include the role of self and identity in organizational contexts, leadership,(group) emotions, organizational change, and power and influence.

Henk van der Flier was Head of the Department of Work and Organizational Psychology at VUUniversity, Amsterdam, The Netherlands. He obtained his PhD at VU University. His researchinterests are in the fields of working conditions, safety, personnel selection, psychometrics and cross-cultural psychology.

358 J. Tanghe, B. Wisse and H. van der Flier

r 2009 British Academy of Management.