WK1

CHAPTER12.doc

CHAPTER 12 SMALL GROUPS: ONGOING INTERACTIONS

Two old school chums have a chance encounter as adults, and one—English, but raised at first in China—is left thinking about the other English fellow's casual remark that he had been “such an odd bird at school.”

In fact, it has always been a puzzle to me that Osbourne should have said such a thing of me that morning, since my own memory is that I blended perfectly into English school life. I do not believe that I did anything to cause myself embarrassment. On my very first day, for instance, I recall observing a mannerism many of the boys adopted when standing and talking—of tucking the right hand into a waistcoat pocket and moving the left shoulder up and down in a kind of shrug to underline certain of their remarks. I distinctly remember reproducing this mannerism on that same first day with sufficient expertise that not a single of my fellows noticed anything odd or thought to make fun.

In much the same bold spirit, I rapidly absorbed other gestures, turns of phrase and exclamations popular among my peers, as well as grasping the deeper mores and etiquettes prevailing in my new surroundings. (Ishiguro, 2000, p. 7) 1

Who has not, as the new kid in school, or newcomer at work, struggled to grasp the group's ways of talking and behaving, in order to fit in? And no wonder, because being branded as the odd duck bodes ill for one's continuing place in the group.

The group is the lynchpin of social situations. We began this book by describing people's core social motivation to belong as underlying all the other social motives. At the outset, we saw the evidence for people's well-being and even survival as enhanced by belonging to at least one group. The book so far has toured various forms of social being in that context, from within the individual to between pairs of people. The previous chapter started to move to the group level of analysis, examining how people's division of themselves and others into ingroups and outgroups can cause a variety of intergroup biases, even when only two people are involved, because they are seen as representing distinct groups. In this chapter, we turn to processes that operate within just one group, processes that motivate many intergroup, interpersonal, and intrapersonal processes we have seen before. We address both the individual in the group and the group itself. We begin by defining groups and some salient social motives and then move to the development of group membership—actual belonging—including the role of social identity, cohesion, diversity, and socialization. A focus on socially shared cognition—shared understanding—will frame the discussion of group structure, including norms, roles, leadership, and subgroups. Group performance—understanding group goals and controlling productivity—brings us to decision making and performance. Finally, intragroup conflict results from people's attempt to control their own and other people's outcomes, when these are not identical, comprising a discussion of social dilemmas and negotiation. This chapter revisits some theories and concepts seen earlier in the book, but cast in a new light by the specifically group-oriented context and the special salience of belonging to a group, so as not to be an odd duck, a turkey, or any other kind of “odd bird.”

WHAT IS A GROUP?

Naturally interacting groups typically include two to six members, according to observational studies. People prefer these small groups when actual interaction is integral to the purpose of the group, whether accomplishing a task or enjoying each other socially (Levine & Moreland, 1998; Mullen, 1991). As total group size increases, people get unhappy; participation rate not only decreases but becomes more variable (some people dominate), and various indicators of commitment show dissatisfaction: Attrition, turnover, and conflict increase; cooperation and performance decrease. This chapter focuses on these small, face-to-face interacting groups.

Conceptual Definitions

Social psychologists converge on three approaches to defining a group. Some definitions view a group as comprising individuals whose combined behavior simply summates into a group phenomenon. Others consider a group to have unique properties that go beyond merely interpersonal processes (for example, developing a social identity; Chapter 11 ). Others view groups as a major evolutionary transition within the biological hierarchy, like the transition whereby cells form organisms; groups represent the next level after individuals, when between-group selection dominates within-group selection (Wilson, Van Vugt, & O'Gorman, 2008). Still others abandon the effort at definition (Hogg, 1995a; Levine & Moreland, 1998). Based on what social psychologists have discovered about them, we argue that groups have unique, emergent properties that differentiate them from a mere aggregate of individuals on three counts: perceived entitativity, perceived volition, and actual behavior.

PERCEIVED ENTITATIVITY

For our purposes, we consider a psychological group—whom we can initially define as interacting people considered by themselves or others to belong together. What tends to make people define an aggregate of people as a group? Viewing the quality of groupiness as a continuum, not a dichotomy, will help. Half a century ago, Donald Campbell (1958) relied on the Gestalt properties of perceiving stimuli to be a single unit, an entity having real existence. Recall from Chapter 11 that entitativity entails being perceived as a coherent whole, based on similarity, common fate (interdependence), and perhaps proximity.

Several people in an elevator are proximate but do not fit any other definition of a group. They perceive themselves as more groupy, more of a psychological unit or entity, if the elevator stalls, because they then develop a common fate. Safety depends on some degree of cooperation. Having a common goal creates interdependence, because they need each other to accomplish it. If they are similar in age, gender, ethnicity, or class, they may become still more of a group. Ongoing interaction (after the initial event, for example in a posttrauma therapy group) would make them even more of a group because it increases the sense of a common fate. Interdependence toward a common purpose is key in many definitions of groups (e.g., McGrath, Arrow, & Berdahl, 2000). All these factors—proximity, similarity, interdependence (common fate), and ongoing interaction—encourage cohesion and social integration, which foster among the individuals a sense of group membership, the topic of a later section. Essentially, cohesion results in developing a shared understanding of their situation and an emotional bond with each other. Psychological groupiness from the perspective of group members seems to follow Gestalt principles.

From the perspective of outsiders, what makes a group distinctive, as compared to an aggregate of individuals? Gestalt principles again apply (Asch, 1952). Observers expect groups to be more unitary and coherent than mere aggregates, although not as unified and coherent as an individual person (Hamilton & Sherman, 1996; Hamilton, Sherman, & Castelli, 2002). This makes sense: We expect individuals to have a coherent personality without too many internal contradictions, but we are not surprised when various members of a group show various attributes, including some that are inconsistent between members. Still, in accord with Gestalt principles, we expect the group members to be more similar (and less contradictory) than a random assortment of people.

To test this idea, researchers have to compare the same set of behaviors attributed to a single individual, a meaningful group, or a loose aggregate and see whether people's impression-formation processes differ with the degree of expected entitativity. In one such study (Susskind, Maurer, Thakkar, Hamilton, & Sherman, 1999), participants viewed clusters of four behaviors (for example, “did 100 sit-ups and 50 push-ups before bed”; “attended two parties with friends over the weekend”; “wrote a letter to congressman about the pending bill”; and “won a chess tournament against strong competition”). Each cluster represented either an individual doing four things, a group of four close friends each doing one thing, or an aggregate of four randomly selected people each doing one thing. People saw the very same four behaviors, when attributed to an individual, as fitting a pattern and being an organized, integrated unit (i.e., an entity), more so than when the same behaviors were attributed to four group members or a random sample of four students. Their confidence about and memory for an individual were stronger as well, and here the friendship group fared better than the mere aggregate, suggesting that the group was more of an entity than the aggregate. Although people do expect individuals to be more predictable and coherent than groups, when they encounter a group explicitly described as having members who know each other well and do a lot of things together, they form impressions of the group in ways that closely parallel their impressions of an individual and differ from an aggregate randomly selected from dorms at a large state university (McConnell, Sherman, & Hamilton, 1997).

In earlier studies (Srull, 1981; Srull, Lichtenstein, & Rothbart, 1985), people forming an impression from a series of behaviors had later recalled more when the behaviors were attributed to individuals and meaningful groups (e.g., a political caucus) than nonmeaningful groups (various people who all happen to be more conscientious than average). What's more, the pattern of memory shows that people puzzle over inconsistencies more for individuals and meaningful groups than for mere aggregates. Again, people expect more coherence from individuals and meaningful groups than from aggregates, so they struggle to resolve incongruence. This too is consistent with the idea that entitativity—being perceived as a meaningful unit—forms a continuum from an individual to a psychologically meaningful group to an aggregate.

Among groups, some specific types seem more entitative than others (Lickel et al., 2000): Intimacy groups (family, close friends) receive the highest marks on entitativity, followed by task groups (committee, project team), both of which differ from broad social categories (blacks, women), and loose associations (people in the same neighborhood, classical music lovers). Groups naturally perceived as having more entitativity are also those seen as having common goals, common outcomes, and similarity (all Gestalt principles), as well as being high in personal importance and degree of interaction.

People view highly entitative groups as having characteristic kinds (as well as amounts) of interaction. Recall from Chapter 8 the four relational stylesidentified in A. P. Fiske's (1992) relational models theory: communal sharing, equality matching, market pricing, and authority ranking. People expect intimacy groups to have communal-sharing relational styles, task groups to have authority ranking, loose associations to use market pricing, and social categories to have no particular interaction styles (Lickel, Hamilton, & Sherman, 2001). Moreover, people value their own groups the more they perceive them to be a distinct entity. The groups highest on entitativity (family and friends) also link most to people's social identity (Lickel et al., 2000).

The differences among intimacy groups, task groups, and social categories are validated by a study of memory confusion. Recall from Chapter 11 that people categorize each other by gender, race, and age, confusing people within category (i.e., confusing two black people with each other) more than between categories (confusing a black person and a white person). This who-said-what technique (Taylor et al., 1978) shows people's propensity to treat category members as interchangeable but distinct from other categories. In a study validating perceived differences among types of groups (e.g., intimacy versus task groups), people confused people within type of group. That is, they confused people within different intimacy groups, so they might confuse a family member with a friend, or they confused people within different task groups (a coworker with a jury member), but they did not confuse people across type of group (they did not confuse an intimacy group member with a task group member). Specifically (Sherman, Castelli, & Hamilton, 2002), participants saw 60 male faces, each paired with a group label (family, friend, coworker, juror, French, Presbyterian). In a subsequent recognition test ( Figure 12.1 ), people made more errors within category than between categories—showing people's propensity to treat certain categories of groups (i.e., intimacy groups, such as friends and family) as interchangeable but distinct from others (social categories, such as French and Presbyterian). This further supports the idea of different types of groups, which vary in degree of entitativity.

Figure 12.1 Memory Confusions within Types of Groups, More Than between Types

Individuals do differ in the propensity to see groups (and other stimuli) as fixed entities; to the extent they do view groups as entities, they also tend to see groups as homogeneous (Levy et al., 2001). Similarly, cultural differences in the tendency to rely on group-level (versus individual) understandings are paralleled by memory for events that best fit the cultural schema (Ng & Zhu, 2001). Not only do groups vary in perceived entitativity, but individuals and cultures vary in the extent to which they tend to view groups that way.

PERCEIVED VOLITION: GROUPS' HOSTILE INTENT

Besides perceived entitativity as defining a group, people perceive groups to differ from each other and from individuals in degree of volition or intent. That is, people assume that individuals are causal agents, origins of actions, as Chapter 3 demonstrated in discussing attributions of causality to individuals. If people view groups as they do individuals, people should view groups also as causal agents. If they view groups differently from individuals, they should not. As the entitativity research indicates, groups vary in how unitary and cohesive they seem, that is, in how much they resemble individuals who are unitary and cohesive. To the extent that people view the group as an entity, then, they might also view the group as a causal agent.

Research indeed supports this idea. Situational manipulations that increase the perceived similarity and proximity of group members (factors that increase entitativity) also increase perceived responsibility and intent. Groups that seem more cohesive also seem more responsible, and people attribute a mind to the group (Waytz & Young, 2012). The more the group has a mind, the less its individuals members seem to have.

The catch is that people attribute not only intent to the group but also specifically hostile intent (Abelson, Dasgupta, Park, & Banaji, 1998). For example, priming the word they before an ambiguous story then makes collective actors more threatening, just as priming the word he makes a single actor more threatening. In both cases, a focus on the causal agent (protagonist) presumably increases perceived intent and (in this instance) perceived hostility. Similarly, manipulating the similarity and proximity of imaginary group members—making them more entitative—leads to more negative judgments about their likely intergroup behavior. When groups compete, of course, intergroup hostility deepens, sparking neural reward centers' pleasure at “their” losses, even to third parties (Cikara, Botvinick, & Fiske, 2011).

Besides manipulating the degree to which a group is an entity, researchers can draw on cultural tendencies to focus on individual or collective agency. That is, collectivist cultures see groups as having more impact than individualist cultures do. In East Asian newspapers, compared with American newspapers, accounts of trading scandals focus more on the organization than on the individuals (Menon et al., 1999). Similar results occur for team versus individual responsibility in a scenario study involving a maladjusted team member who is not well integrated into the group. Although the bad situation could be the fault of the individual or the group, East Asians tend to view it as more of a group responsibility, whereas Americans blame the individual.

Why do groups perceived to be entitative elicit not only attributed intent but specifically hostile intent? Part of the answer lies in perceptions of what makes a group a group. As noted in the previous chapter, people expect different groups to have incompatible goals. The same is not so markedly true of individuals. If part of what defines a group as a group is shared goals (as noted), then different groups will have different goals. To the extent they have differing goals, the goals are likely to be incompatible or competing. Hence, merely in pursuing its own goals, one group will interfere with another group and provoke negative emotions (Fiske & Ruscher, 1993). People seem to expect this, given the just-cited finding that perceived groupiness elicits perceived hostile intent (Abelson et al., 1998). What's more, people preferentially remember instances of intergroup competition, compared with interpersonal (interindividual) competition (Pemberton, Insko, & Schopler, 1996), thereby confirming their expectations that groups allegedly have hostile intent more than individuals do.

ACTUAL BEHAVIOR: INTERPERSONAL-INTERGROUP DISCONTINUITY IN HOSTILITY

People apparently are not wrong to expect hostile behavior from other people in groups. The interpersonal-intergroup discontinuity effect (Schopler & Insko, 1992) shows that people behave differently in interpersonal dyads than they do in intergroup interactions. The experimental paradigm for this work is a mixed-motive game, in which self-interest conflicts with the other side's self-interest and yields lower outcomes than mutual cooperation. In Chapters 8 and 9 , we saw matrices like the one in Figure 12.2 , in discussing relationship interdependence and cooperative helping. In the mixed-motive case depicted here, each side individually profits by making a competitive choice, but only if the other side simultaneously cooperates. Both sides lose if they both make a competitive choice. Thus, each side's outcomes are negatively correlated—noncorrespondent—on the whole; their interests are not the same. However, the negative correlation is not complete; people can make choices (both cooperate) that together serve them better than other choices (both compete).

Figure 12.2 Experimental Game Matrix of Outcomes, Given Each Side's Choice, in a Mixed-Motive Case

In this case, individuals typically learn to cooperate over time (as we will see in a later section), but groups do not. Groups make more competitive choices on average than individuals do, as the top row of Table 12.1 indicates (Insko, Schopler, Hoyle, Dardis, & Graetz, 1990). Groups choose competition over cooperation roughly 10 out of 20 times, whereas individuals do so only 1 in 20 times (left two columns). Given the additional choice to withdraw, groups still choose to compete more than twice as often as individuals do (right two columns). The effect is reliable and substantial over studies and appears even in domains that do not use an explicit payoff matrix (Pemberton, Insko, & Schopler, 1996; Schopler et al., 2001). The more negatively correlated the outcomes of the two sides, the more groups (but not individuals) behave competitively. Counter to gender stereotypes, groups of women show this escalation under increased competition more than groups of men do (Schopler et al., 2001). One explanation is that they react with more anger to the unfairness of the other side's behavior (Mikolic, Parker, & Pruitt, 1997), perhaps because they tend to have a more interdependent self-construal ( Chapter 5 ).

TABLE 12.1 Competitive Choices, Cooperative Choices, and Withdrawal Choices as a Function of Type or Number of Choices and Groups versus Individuals

	Choices
	Cooperative or Competitive Only	Cooperative, Competitive, or Withdrawal
Responses	Groups	Individuals	Groups	Individuals
Competitive	10.14	1.00	4.50	1.75
Cooperative	9.86	19.00	10.43	17.36
Withdrawal	–	–	5.07	0.89

Each of the three choice totals has a possible range from 0 to 20.

The interpersonal-intergroup discontinuity is driven by two processes: fear and greed. Fear appears in people's negative expectations that outgroups will be competitive and self-serving, which (some would argue) is just a rational response in experimental games of the kind shown in the figure. Evidence for the role of fear comes from studies showing a correlation between fewer cooperative choices and how much the group on one side discusses their distrust of the other side (Insko et al., 1990). Groups are less trusting than individuals (Kugler, Bornstein, Kocher, & Sutter, 2007). Their fear fits the threat theories of intergroup bias (such as realistic group conflict, social dominance orientation, and right-wing authoritarianism, covered in Chapter 11 ). Fear thus fits our core social motive of controlling one's outcomes. It might also fit our understanding motive as a way to make sense of intergroup rivalry.

Greed, on the other hand, fits control and perhaps also self-enhancement. Even if the situation neutralizes fear (and competitive expectancies), apparent greed remains in the willingness to exploit the other side. Suppose one side has the opportunity to withdraw from competition (the middle choice in Figure 12.3 ). Groups do use this option (more than individuals do), and their cooperation increases when this self-protective choice is available, but they still cooperate less often than individuals do (see right side of Table 12.1 ; Insko et al., 1990; Schopler, et al., 1993). Greed increases with ingroup social support, in the form of group members arguing for the exploitation of the other side (Schopler et al., 1993; Wildschut, Insko, & Gaertner, 2002). Groups try to maximize their relative advantage over other groups, as predicted by social identity theory, introduced in Chapter 11 . The locus of the effect also appears in maximizing own group outcomes, which fits the literature showing that much discrimination operates via ingroup favoritism more than outgroup derogation. Groups' greed apparently focuses on short-term advantage, because under long-term interactions, or even the anticipation of same, the effect dissipates (Insko et al., 2001; Insko et al., 1998).

Figure 12.3 Experimental Game Matrix of Outcomes, with Addition of a Safe Withdrawal Option

The interpersonal-intergroup discontinuity effect emphasizes the intrinsic differences between individual and group behavior. Not only do people expect groups to be more competitive and self-serving than individuals, as the previous section indicated, but also they are apparently right to do so, as this section indicates. Although entitativity helps define a group psychologically, competition forms an integral part of people's understanding of groups.

Operational Definitions

As implied by some of the research just described, the groups studied by social psychologists most often (about three-quarters of the time) are short-term experimental groups formed by recruiting participants from psychology department subject pools (Moreland, Hogg, & Hains, 1994). These groups tend to meet for a short time, often only one session, to work on a task assigned by the experimenter.

Some of the typical experimental methods have already appeared in our discussion of the interpersonal-intergroup discontinuity; participants play experimental games (with tangible monetary outcomes) in the laboratory, to simulate groups in the outside world. We will see more examples of this technique in the section on conflict in groups. Other methods include assigning the group a problem-solving or decision task.

Problems with the laboratory approach include isolation from a larger context (e.g., a team exists within an organization), lack of a past or a future, assumptions that the behaviors of student participants generalize to a variety of other adults, and narrowly focusing on a particular cause-effect relationship (McGrath, Arrow, & Berdahl, 2000). These issues are not unique to the literature on groups (e.g., Chapters 9 – 11 ). To be fair, experiments are designed precisely to isolate and imply a particular cause-effect relationship, and they cannot necessarily claim generalizability to other participant populations and contexts. That is why social psychology typically replicates its most important findings in field studies, outside the laboratory. Contrary to the expectations of skeptics, most group phenomena are actually stronger in the field (real groups) than in the laboratory (artificial groups), according to meta-analyses of various group effects (Mullen et al., 1992; Mullen & Copper, 1994; Mullen & Hu, 1989; Ostrom & Sedikides, 1992).

About a tenth of group studies use surveys, presumably yielding correlational data on groups outside the laboratory (Moreland et al., 1994). This overall pattern reflects the methods of the rest of social psychology, so it is not surprising, except that many studies of groups reflect on phenomena such as organizational behavior. Indeed, compared with group studies appearing in social psychology journals, those appearing in organizational journals tend to report experiments less often (50%), field studies and field experiments more often (28%), and about the same number of surveys (10%) (Sanna & Parks, 1997). Field studies on “real” groups tend to take more account of the complexities of the groups' organizational context, for example, in examining airline cockpit crews, management teams, and quality improvement groups (McGrath et al., 2000).

Regardless of experimental or correlational technique, laboratory or field, studying groups is an inherently daunting enterprise, which has at times discouraged researchers from this topic (Hogg, 1995a; Levine & Moreland, 1998; Steiner, 1974). The biggest problem is that the unit of analysis becomes the group, instead of the individuals in the group. That is, to draw conclusions about group behavior, researchers need to observe enough separate, independent groups to be able to find statistically reliable results. So, for example, a phenomenon that might require 30 people in a study of individuals would require 3 to 5 times that number of people to study as a group phenomenon (assuming a minimum of 3 people to make a group makes the total 90; assuming a more complex group of 5 makes the total 150 individuals). The group level of analysis makes finding enough participants many times more inconvenient and expensive.

What's more, group data are inherently complex. Examining a single individual's reactions entails one single set of observations. Examining a dyad entails three kinds of observation: each person's perspective, as well as any interactions that are unique to particular combinations of people (e.g., individual males might react one way and individual females another, but the particular gender combination in a dyad might create still a third pattern). Now consider a three-person minimalist group. The researcher would have data from three individuals, as well as data from each of them reacting specifically to the other two (six more kinds of observations), as well as data resulting from the group as a whole, which might vary as a function of group composition. That's ten types of observation. Besides, interaction patterns may create coalitions. You see the problem.

Whatever the number and pattern of interactions, their content then has to be coded. One long-running system for coding the content of group interactions is Interaction Process Analysis, also known as SYMLOG (simultaneous multilevel analysis for groups; Bales, 1950, 1970, 1999). The coding scheme involves three dimensions. The first dimension, dominance-submission, is labeled Upward-Downward and is most reliably indicated by sheer talking time in the group. The second dimension, friendly-unfriendly, is labeled Positive-Negative and indicates the pleasantness of people's interaction. The third dimension, on-task–off-task, is labeled Forward-Backward and indicates the degree of instrumental control versus emotional expression. To give a sense about the combinations, consider how some famous people would appear in a group, along these dimensions (Isenberg & Ennis, 1981). Groucho Marx would be Upward (talkative), Positive (friendly), and Backward (expressive and off-task); Adolph Hitler would also be Upward (dominant) but Negative (unpleasant) and Forward (instrumental); Abraham Lincoln would be Upward, Positive, and Forward. Famous people are less likely to be submissive, but Charlie Chaplin would be Positive and Backward (like Groucho Marx, but more Downward than Upward). And Shakespeare's agonizing but inactive Hamlet would be Downward, Negative, and Backward (passive, gloomy, expressive, and off-task). In a real group, each comment can be coded, online, in real time, by observers (such as your author, in graduate school). The pattern of interactions across individuals reveals a three-dimensional space that describes the group structure. Content analysis techniques yield an abundance of rich data, challenging but ultimately satisfying. Indeed, “completing a major project on groups, especially one that involves extensive data on interaction patterns, seems to serve as a ‘rite of passage’” for researchers new to the small-groups area (Levine & Moreland, 1998, p. 419).

Finally, the statistical challenges of analyzing group interactions are nontrivial, because all the members of a group influence each other, violating some basic statistical assumptions about independence of observations. Solutions often require the researcher to move outside the normal statistical procedures (Forsyth, 1998; Kenny, Mannetti, Pierro, Livi, & Kashy, 2002; Sadler & Judd, 2001). No wonder one commentator observed “the data from group research are rarely neat, tidy, and easily explained” (Hogg, 1995a, p. 270). Nevertheless, the efforts are extremely important and all the more amazing when the data come out.

One recent reaction to these methodological issues has been computational models of isolated features of group interactions. For example, modeling the mutual influences among interacting individuals, in dynamical evolutionary psychology, focuses on competition and cooperation (especially mating) within a web of mutual contingencies (Kenrick, Li, & Butner, 2003). This allows quantitative specification of individual decisions in a social context. Emergent properties characterize the group, norms being one example, and they are not predictable from simple rules about individual beliefs or even dyadic interaction, so agent-based computational models (i.e., comprising autonomous individuals) (Goldstone, Roberts, & Gureckis, 2008) represent an alternative method.

Core Social Motives

Going back to the social psychology of the group itself, three main motives—belonging, understanding, and controlling—have formed the core of small-group research and seem to capture people's main motives for participating in group interaction.

BELONGING

Earlier, in the context of definitions, we said that entitativity was one of the main features of a psychologically meaningful group. Now, in the context of motives, consider how the entitativity of the group allows it to meet people's need to belong (Yzerbyt, Castano, Leyens, & Paladino, 2000). Ingroup entitativity predicts degree of social identification with the group, it being easier to identify with a group that has clear definitions and boundaries. Seeing the ingroup as coherent enables people to see the group positively, as a whole, thereby increasing identification.

But more than that, high identifiers emphasize ingroup homogeneity (entitativity) especially when the group is threatened, for example by its having low status, unfavorable stereotypes, unfair outcomes, or even mortality salience (Castano, Yzerbyt, Paldini, & Sacchi, 2002; Yzerbyt et al., 2000). (Recall group homogeneity from Chapter 11 and mortality salience from Chapters 9 – 11 .) When group members identify with their group, its coherence matters to them because it is motivationally easier to protect loyally a well-defined entity than a loosely defined aggregate. Similarly, threat from an entitative outgroup, compared with a mere aggregate, increases identification with the ingroup (Dépret & Fiske, 1999). Consider the typical American emphasis on the diversity of the nation, but consider also how much Americans seemed to pull together as a cohesive unit after September 11, 2001, when the threat by a more clearly defined outgroup became unavoidable. Belonging together became a central motivation for most Americans under intergroup threat.

Two other lines of evidence implicate the motivational primacy of ingroup belonging. A phenomenon dubbed the black sheep effect shows that ingroup members firmly reject an ingroup deviant, often even more than an outgroup member who deviates from ingroup values (Marques, Abrams, Páez, & Hogg, 2001; Marques, Yzerbyt, & Leyens, 1988). Compared with ratings of outgroup members, judgments of ingroup members are more extremely positive if the person is likeable (an extension of ingroup favoritism) but also more extremely negative if the person is unlikable (the black sheep effect). Reactions to outgroup members are not as extreme, either positively or negatively, as reactions to ingroup members. In comparisons of ingroup and outgroup, correlations among perceived traits of the ingroup are higher, suggesting that people's implicit theory of the ingroup is more coherent, making it seem more entitative. Thus, ingroup deviants stand out more than outgroup deviants.

People seem to maintain ingroup boundaries with some care, rejecting those negative deviants unacceptable to social identity. The role of identity is specifically implicated because the black sheep effect occurs only on identity-relevant traits. Also, high identifiers show the effect more clearly than low identifiers and are willing to invest considerable cognitive resources to concentrate attention on the threatening negative ingroup member (Coull, Yzerbyt, Castano, Paladino, & Leemans, 2001). All this suggests the motivational importance of ingroup belonging.

If the black sheep effect reflects group members curing the already-contaminated ingroup, another phenomenon may be viewed as preventing potential contamination (Yzerbyt et al., 2000). The ingroup overexclusion effect refers to ingroup members' carefully guarding their own group's boundaries, more than other groups' (Leyens & Yzerbyt, 1992; Yzerbyt, Leyens, & Bellour, 1995). That is, group members are more cautious about inferring ingroup membership, compared with outgroup membership, for someone whose group membership is ambiguous. For example, people demand more information to admit the potential ingroup member, and especially when the cues are positive or stereotype-consistent (because negative or stereotype-inconsistent cues would allow them to reject the potential ingroup member immediately). Ingroup members must be judged more carefully than outgroup members because the stakes of inclusion are higher for one's own group than the stakes of allowing someone to be classified as belonging (or not) to someone else's group. Overall, the clear psychological entitativity and boundaries of ingroups support group belonging motives.

A complementary line of evidence even more dramatically indicates the motivational importance of belonging by showing what people do when they are excluded (e.g., rejected as an ingroup member). Two theories address the problem. A model of ostracism suggests that being socially rejected threatens people's need to belong. In one study, people reported their experiences of receiving the silent treatment, that is, other people avoiding eye contact and verbal communication. People specifically mentioned their damaged feelings of belonging (as well as self-esteem, control, and meaning; Williams et al., 1998). In another study, two confederate participants excluded a third real participant simply by tossing a ball back and forth to each other and leaving out the other person (Williams & Sommer, 1997). Excluded female participants worked especially hard on an ensuing group task, as apparent social compensation for their prior exclusion (males did not; we will come back to this). More direct evidence related to the belonging motive comes from a study that simulated the ball-tossing game and subsequent exclusion on an Internet website (Williams et al., 2000). Even with this minimal form of ostracism (a symbolic ball and excluders who were never encountered directly as real people), 1,486 participants from 62 countries reported losing a sense of belonging and self-esteem, which in turn made them feel bad. In a second study, participants ostracized in this way showed more conformity on a subsequent task, again an apparent social compensation for their exclusion. People apparently try harder at first, but high and persistent levels of ostracism eventually make them withdraw effort. Regardless of whether computer-mediated or face-to-face, ostracism has broadly similar negative effects (Williams, 2007; Williams et al., 2002).

Still another theoretical approach centers on the belonging motive and its role in social exclusion. Recall from Chapter 1 the Baumeister-Leary (1995) evidence for the centrality of the belonging motive. In general, people seek social attachments, benefit physically and psychologically from them, and resist their dissolution. Specifically with regard to social exclusion, recall from Chapter 5 that people's self-esteem depends on their assessment of their degree of personal acceptance by others, a phenomenon studied as the sociometer hypothesis (Leary et al., 1995). Personal exclusion from important social groups correlates with anxiety and depression (Baumeister & Tice, 1990; Leary, 1990; Nezlek, Kowalski, Leary, Blevins, & Holgate, 1997). To illustrate people's efforts to belong, one might consider a study of 22 separate task groups, in which positive, pleasant acts outnumbered negative, unpleasant acts by a ratio of 3.4 to 1 (Bales & Strodtbeck, 1951). The baseline positivity of interaction suggests people's efforts to cooperate and get along in the service of continued belonging.

Finally, attachment theory (recall from Chapter 8 )—which began by explaining infant-parent bonds and later moved into adult close relationships—also can apply to an individual's attachment to social groups (Smith, Murphy, & Coats, 1999). Just as people have mental working models of self and others in close relationships, so they have working models of self (as a group member) and the group (as a source of social support and identity). Just as with attachment styles in close relationships, so too some people are relatively avoidant, anxious, or secure in their attachment to groups, and this predicts their style of belonging to the group.

Other individual difference and person variables relate to group belonging, including possible gender differences. To the extent that women have more interdependent self-construals on average, compared with men, they may operate more in terms of relationships than in terms of autonomy (Cross & Madson, 1997). Consequently, women and men may view their group memberships differently (Gabriel & Gardner, 1999). In particular, women might focus more on the relational aspects of interdependence within groups, whereas men might focus relatively more on collective aspects of interdependence within groups. That is, women operate relatively more in terms of intimate relationships and interpersonal harmony within dyads and small groups and less often in terms of larger collectives, such as teams and social groups. Men within groups operate in relatively more collective and less relational ways than women. Relevant data include self-construals, important emotional events, memory for others, and behavior in experimental games. In one study, women were more attentive during the interaction, particularly in committed relationships, and men were less attentive (Stiles et al., 1997). Thus, the meaning of belonging to groups may differ on average by gender.

Cultural variation also shows that the meaning of belonging differs by such dimensions as collectivism-individualism (see previous chapters) and power distance (which differentiates more hierarchical cultures from more egalitarian ones) (M. H. Bond & Smith, 1996). For example, people in collectivist cultures are likely to define themselves and be motivated more in terms of their group memberships than are people in individualist cultures. On the other hand, people in individualist cultures belong to more groups, moving in and out of them more rapidly and easily, and are less attached to any one of them.

As another example of cultural variation, power distance strengthens the role expectations for leaders and followers; that is, it increases the difference between leaders and followers as a function of their respective roles. For example, high power distance demands more assertive and dominant behavior by leaders, whereas low power distance demands more ingratiation and exchange from leaders. Regardless of cultural and individual variation, however, belonging forms an important motivation for human group membership. Attempts to fortify belonging, in the face of exclusion, tend to make people think and behave prosocially (Williams, 2007), which helps maintain the group (Levine & Kerr, 2007).

UNDERSTANDING

So far in this book, we have seen that motives for a socially shared understanding matter for people making sense of others, the self, and various attitude objects, as well as entering into close relationships, helping, aggression, and intergroup bias. So, too, do groups serve an understanding motive. People have a stronger trust in a shared reality with a group than with an individual (Hardin & Higgins, 1996; Higgins, Echterhoff, Crespillo, & Kopietz, 2007).

A clear example of the understanding motive in groups comes from a motivational theory of social identity processes. The subjective uncertainty reduction hypothesis (Hogg, 2000, 2007b; Hogg & Abrams, 1993) argues that, when people feel unsure about self-relevant matters, self-categorization as a group member reduces their uncertainty. For example, a student new on campus—full of self-doubt and ignorance about local norms—often gladly identifies with the first-year class, certain clubs or teams, and some areas of study rather than others. When people identify with a group, they depersonalize—become less oriented to their individual identity—and orient more toward being a prototypic member of the group. Assimilating self to the group's prescriptive prototype, that is, the group ideals, reduces feelings of uncertainty by providing guides for thoughts, feelings, and actions. Moreover, the group consensus validates individual group members' reactions when they assimilate to the group prototype. People are especially likely to identify with highly entitative groups when their self-certainty is low (Hogg, Sherman, Dierselhuis, Maitner, & Moffitt, 2007). The theory argues that people are motivated to join groups, precisely in order to reduce their own uncertainties: “People need to feel certain about their world and their place within it; certainty renders experience meaningful and gives one confidence in how to behave, and what to expect from one's physical and social environment” (Hogg, 2000, p. 227). The Ishiguro example that opened this chapter characterizes this experience.

Evidence for the uncertainty reduction hypothesis comes from several studies using the minimal group paradigm (recall from last chapter), in which people are divided into arbitrary groups and then allocate rewards to ingroup and outgroup. The typical finding is that people advantage the ingroup, which is interpreted as identifying with their own group. What is, of course, most surprising is how easily people do this, even with the most minimal of explicitly arbitrary groups. In this paradigm, Hogg argues, people show ingroup favoritism only when they are in a state of temporary uncertainty. For example, when people first encounter the unfamiliar, complex matrices they have to use to allocate rewards in this paradigm, they identify with the ingroup more than when they have already practiced with the matrices (Grieve & Hogg, 1999). The same occurs when people are uncertain about important attitudes (Mullin & Hogg, 1999). What's more, reminiscent of the importance of entitativity, people prefer homogeneous (entitative) groups for uncertainty reduction (Jetten, Hogg, & Mullin, 2000). All this evidence supports the importance of group membership as validating one's own responses under uncertainty. An entitative group provides a sense of consensus, socially shared understanding.

Socially shared cognition in ongoing groups results from common experience, social interaction, coordinated communication, and social comparison (Tindale, Meisenhelder, Dykema-Engblade, & Hogg, 2001), as a later section elaborates. And Chapter 13 elaborates on some classic studies noted in Chapter 1 (Asch's and Sherif's studies), which show how groups allow people to develop norms for appropriate judgments. Another classic example is people's tendency to affiliate with others when they are afraid, in order to make social comparisons of their ambiguous experience (Schachter, 1959). For present, the point is that motives for a socially shared understanding are satisfied by group identity.

When a sense of shared social understanding fails, people become uncomfortable (as argued in previous chapters). In the group context, people may become paranoid (irrationally distrustful and suspicious), when they do not share the group's understanding (Kramer, 1998). Uncertainty about one's social standing can result from being new or different, for example. These situational uncertainties can make people hypervigilant and make them ruminate, which together lead to perceived conspiracies, personal insults, and evil intent. These maladaptive forms of understanding occur only when people have not fulfilled their motive for shared understanding (and belonging) by fully secure group membership. Issues of socially shared understanding occur at early stages in the life history of a group, when many members are new and are all getting oriented to each other (Bales & Strodtbeck, 1951). At those stages, giving or requesting information, clarification, or confirmation each illustrates an effort to develop shared understanding.

CONTROLLING

Because groups, by definition, comprise people who are interdependent, motives for control matter in groups. Precisely because members' outcomes depend on each other, groups arouse, and often meet, people's needs for control. Commitment to the group results from its mutual reward value, as newcomers become socialized (Levine, Moreland, & Choi, 2001). Group negotiations (Thompson, Medvec, Seiden, & Kopelman, 2001), power relations (Lord, Brown, & Harvey, 2001), and social dilemmas, which pit individual against group interest (Kerr & Park, 2001), all especially raise issues of control, which later sections revisit. In addition, individuals and the group as a whole aim to control the group's outcomes and reach common goals. Group-based, vicarious control can substitute for a felt lack of personal control (Fritsche et al., 2013; Fritsche et al., 2011).

Some people and some situations elicit a particularly strong need for control. When people are ostracized, if their control motives also are endangered, then they may respond in antisocial ways, further undermining the group and their place in it (Williams, 2007). But this is an extreme reaction to ostracism; in an ongoing group, control needs can facilitate certain kinds of group process. When people are dispositionally or situationally high in one kind of control motive, need for closure, the group becomes more business-like. Interaction focuses toward the task at hand and away from positive social interactions that do not directly advance the task (DeGrada, Kruglanski, Mannetti, & Pierro, 1999). At such junctures, group interactions are less egalitarian and demand more conformity from their members. Groups similarly may focus on issues of control at later stages in group development, when executing their tasks becomes critical (Bales & Strodtbeck, 1951). Giving and requesting suggestions, directions, or plans illustrate behaviors that reflect mutual control over each other and the group's outcomes. Clearly, this also implies a role for the core motive of trust, but the research has been framed more in terms of mutual control than mutual trust.

Summary of Definitions and Motives

Psychological groups comprise interacting people who consider themselves or whom others consider to belong together. Groups differ in entitativity, the degree to which they make a coherent whole, through Gestalt principles of similarity, proximity, interdependence, and interaction. Intimacy groups, such as family and friends, are experienced as more coherent and cohesive than loose associations (such as neighborhoods) and social categories (such as women), with task groups falling in between. People value their own groups to the extent they view them as cohesive, coherent entities. In contrast, to the extent people view an outgroup as being an entitative group, they see it as having hostile intent. Because groups are typified by shared goals that operate in the group's own interest, and entitative groups appear as causal agents—that is, as originating action—outgroups are expected to act on their own interests, which will be hostile to other groups. And, indeed, groups are more competitive than individuals. Both fear of losing control over one's outcomes and greed to enhance self contribute to group competitiveness.

Social psychologists most often study groups by creating short-term, artificial, experimental groups that often play experimental games, all of which might seem to lack generalizability to the real world. Nonetheless, real-world effects are often stronger than laboratory demonstrations of the same phenomena. Researchers use survey and observational methods when they study real-world groups. People in both artificial and real groups satisfy a number of core social motives, including most obviously the need to belong but also to validate a shared social understanding and a sense of control. Each of the main sections of the chapter addresses one or two of the core motives, to discuss group membership (belonging), socially shared cognition (understanding and controlling), group performance (understanding and controlling), and intragroup conflict (controlling).

GROUP MEMBERSHIP: BELONGING

This section addresses what happens when people belong to a group: how their identity changes, how their attraction to group members creates cohesion, and how they become socialized into the group. People readily affiliate with similar and proximate others, forming interdependent bonds that foster identity and attraction; groups deal with diversity in composition via socialization processes. All these processes occur because people are fundamentally motivated to belong to groups.

Social Identity Operates in Context

In the last chapter, social identity theory (SIT) and self-categorization theory (SCT) appeared in their intergroup context as sources of bias. Self-definition in a given social context creates a useful and feasible particular group identity for a particular setting. Social identity is that part of the self-concept that derives from group membership. In the last chapter, we saw that merely categorizing self and others as group members increases each group's perceived homogeneity (and entitativity). Self-categorization and social (as opposed to interpersonal) identity increase ingroup favoritism and consequent outgroup disadvantage. This kind of intergroup differentiation increases at least short-term self-esteem, linking the belonging motive to self-enhancement.

In the current context of examining intra-group reactions, the experience of the individual becoming a group member is paramount. Why should people move from interpersonal to intergroup self-categorization? In some contexts, the individual self motivationally dominates the collective self (Gaertner, Sedikides, & Graetz, 1999). Relative to collective threat, Westerners consider individual threat more severe, react more negatively, and more severely derogate the source of the threat. These results hold up despite individual variations in individualism and collectivism. This might suggest the motivational primacy of the individual self, who might then not bother to join groups. Nevertheless, people demonstrably do want to join groups, and the very nature of their individual self depends on the groups that they actually join. The individual self (however important it may be) still varies as a function of context (as Chapter 5 showed). Social identity, whether individual or collective, is context-specific; that is, it depends on one's relevant groups and situations.

In one study, for example, targets were rated on five personality traits by people from three different groups, all of whom knew them well: family, friends, and coworkers (Malloy, Albright, Kenny, Agatstein, & Winquist, 1997). Each person was judged consensually within any given group but differently in differentgroups. Within-group consensus accounted for fully 30% of the variance in ratings, and consensus was statistically reliable in all 15 tests (5 traits in each of 3 groups). In contrast, between-group consensus accounted for much less variance (less than 10%) and was statistically reliable in only 5 of 15 tests (5 traits in each series of 3 paired comparisons). This difference in perceptions across groups (but not within groups) supports the idea that one's personality and identity depend on the context provided by a particular group. You might be one person to your family and quite another to your coworkers or friends. Contextual differences in your perceived personality could reflect either different groups' different perceptions or differences in your actual behavior, but both bear on group-specific identity.

Given the importance of social identity, it is critical to know when social identity matters more and when personal identity matters more. A taxonomy of situations predicts that group commitment and type of threat combine to raise distinct identity concerns and motives (Ellemers, Spears, & Doosje, 2002). Individual concerns and motives come to the fore when group commitment is low, as shown in Table 12.2 , and as follows:

· Under no threat to self, low group commitment (cell 1) allows a focus on efficiency in relating to others, as in dual-process models of social cognition (Fiske et al., 1999; Brewer & Harasty Feinstein, 1999; see Chapter 4 ).

· However, when the individual self is threatened, and group commitment remains low (cell 3), individuals are concerned about self-affirmation (Steele, 1988; see Chapter 11 ). Cases of stereotype threat (Steele et al., 2002; see Chapter 11 ) and outgroup derogation to protect damage to personal self-esteem (Fein & Spencer, 1997) exemplify this self-threat, low group-commitment context. Here, individuals' primary concern is having themselves categorized negatively.

· In the last low-commitment situation, but now under a group threat (cell 5), individuals orient toward individual mobility out of the group, being more concerned about restoring the value of their own personal identity. Individuals become dissatisfied, distance themselves from the group, identify less with their group, see it as more heterogeneous, self-stereotype less as a group member, and are less willing to help the group, unless they can gain personally (Branscombe, Spears, Ellemers, & Doosje, 2002; Doosje, Spears, & Ellemers, 2002; Ellemers, 1993; Ellemers et al., 1997; Jackson, Sullivan, Harnish, & Hodge, 1996; Spears, Doosje, & Ellemers, 1997).

TABLE 12.2 Primary Concerns and Motives of the Social Self: A Taxonomy

	Group Commitment
	Low	High
No threat	1.	2.
Concern	Efficiency	Social meaning
Motive	Noninvolvement	Identity expression
Individual-directed threat	3.	4.
Concern	Negative categorization	Exclusion
Motive	Self-affirmation	Acceptance
Group-directed threat	5.	6.
Concern	Own value	Group distinctiveness, value
Motive	Individual mobility	Group-affirmation

In contrast, this taxonomy proposes that high group commitment brings other identity concerns to the fore, given varying degrees of threat.

· Under no threat but high commitment (cell 2), group members' intragroup and intergroup reactions should follow social identity and self-categorization theory (Tajfel, 1974; Tajfel & Turner, 1979; Turner, 1985), expressing their identity as a member of a positive, distinct group. As the previous chapter indicates, high commitment and identification enhance perceived group homogeneity and ingroup favoritism.

· When self-threat intervenes in a high-commitment context (cell 4), group members become concerned about being excluded by the group and are motivated toward acceptance. This illustrates situations in which someone wants to be a member of a group that is reluctant to admit the person. Marginal group members become even more critical of other marginal group members (Marques & Paez, 1994) and of outgroup members, in public, where it might raise their status in the ingroup (Branscombe, Wann, Noel, & Coleman, 1993; Noel, Wann, & Branscombe, 1995).

· Finally, in the case of high group commitment but group threat (cell 6), group members are concerned about the value and distinctiveness of their group and move to affirm it, by self-stereotyping (Spears et al., 1997), emphasizing ingroup homogeneity and cohesiveness (Branscombe, Schmitt, & Harvey, 1999; Doosje, Ellemers, & Spears, 1995; Doosje et al., 2002).

Determining individual commitment to the group is crucial to this taxonomy. Also, commitment overlaps conceptually and operationally with identification, and social identification with the group is both an independent variable and a dependent variable in this research. For example, people may apparently belong to one group but personally identify with another; this will affect their degree of identification, loyalty, and commitment (Barreto & Ellemers, 2002). Regardless of the challenge of specifying degrees of commitment, the distinction clearly matters to social identity, in interaction with threat. The context provided by group commitment, together with group threat or self-threat, determines loyalty to the ingroup.

Attraction to the Group Fosters Cohesion

People join groups in part because they become attracted to groups at a social level that differs from attraction to the individual group members. The social attraction hypothesis (Hogg, 1992, 1993) proposes that attraction to the group is not simply the sum of individual members' attraction to each other. Rather, it reflects the forces that keep the group together. Theoretically, social attraction results from interdependence in the satisfaction or expected satisfaction of one's goals, and social attraction turns people into a psychologically meaningful group.

Empirically, social attraction results from identification, which itself reflects not only identifying with the group, but also having a clear sense of the group's prototype and seeing self as prototypical. Social attraction does not result from interpersonal attraction, which relates to similarity and personal friendship (as we saw in Chapter 7 ). In an intergroup context, depersonalized social attraction hinges on self-categorization as a group member, not interpersonal relations (Hogg & Hains, 1996). Perceived prototypicality as a group member itself is associated with social attraction, not personal attraction, in both enduring organizational groups and ad hoc student groups (Hogg, Cooper-Shaw, & Holzworth, 1993).

Social attraction predicts conformity to the group (Hogg, 2001; Hogg & Hains, 1998; Hogg & Hardie, 1992), so it is useful for group cohesion. When group members are attracted to the ideal of the group, its prototype, they adhere to its norms regardless of individual friendships. Social attraction is the glue that builds group cohesion.

Social attraction has theoretical origins in interdependence, that is, people being attracted to groups that meet goals they cannot meet as individuals. However, the attraction-interdependence link is less than clear from the social-identity/self-categorization theory (SIT/SCT) approaches, which have not pursued interdependence in detail. Indeed, a narrow interpretation of SIT/SCT would say that categorization, not interdependence, suffices for group membership and attraction. However, group-level interdependence by itself, rather than similarity (an interpersonal variable), does form the ingroup boundaries (Flippen et al., 1996). In a study of pedestrians encountering an allegedly lost envelope, neighborhood residents helped unknown other residents more when their common fate was stressed (their neighborhood was perceived to be under threat). Identification with the group goes hand-in-hand with interdependence in the service of group goals (Caporael & Dawes, 1991).

Diversity Both Challenges and Facilitates the Group

Although we have just seen that interpersonal similarity does not affect attraction to the group as a whole, variability in composition of the group does matter to group identity. Diversity can reflect variety in visible demographic variables such as ethnicity, gender, and age, or it can reflect variety in nonobservable attributes such as values, skills, or experience (Mannix & Neale, 2005; Milliken & Martins, 1996; Williams & O'Reilly, 1998). Identifying with a diverse group is harder, perhaps because the group is a less clearly defined entity. As we saw, depersonalized attraction to a group fosters cohesion, so the diversity of group composition should matter to group members because it might undercut cohesion. As we noted at the outset (Levine & Moreland, 1998), naturally interacting groups typically comprise 2–6 members, and in practice they indeed are often homogeneous. Groups are more homogeneous on personal attributes than would be expected by chance. One common mechanism—recruitment from within the larger organization—encourages the trend toward homogeneity (e.g., Jackson et al., 1991). People are attracted to similar others, as we know from Chapter 7 , which makes difficult people's interactions with people who differ from themselves, as we know from Chapter 11 .

Primarily focusing on field settings, diversity's effects differ in the affective, cognitive, and behavioral domains. Affectively, as the previous chapter might suggest, many group members are uncomfortable with diversity (e.g., Pelled, Eisenhardt, & Xin, 1999). Members of the contextually minority category within any particular group are the most unhappy, perhaps because they stand out (Mullen, 1991). Indeed, dissatisfaction with being the minority particularly occurs for men and for whites, who may not be used to being in that position (Williams & O'Reilly, 1998). The minority subcategory often expresses lower satisfaction, commitment, and identification, as well as perceiving discrimination (a tendency that occurs regardless of the minority's own gender or ethnicity). The group as a whole experiences some coordination costs—friction and conflict—leading to lower social integration (e.g., Mannix & Neale, 2005; O'Reilly, Caldwell, & Barnett, 1989).

The behavioral results of these uncomfortable feelings (particularly for people who are different—the minority subgroups in a given setting) are absenteeism, deviant behavior, and increased turnover (e.g., Jackson et al., 1991; Wagner, Pfeffer, & O'Reilly, 1984). All these factors would tend to increase the homogeneity of the group, if attrition selects out the people who are different. Many of the conflict and turnover effects can improve over time, however, as people accept the diversity as inevitable and learn to work together (Pelled et al., 1999; Watson et al., 1993). Also, if people feel affirmed as individuals, then they may feel free to identify with the group, and productivity improves (Swann et al., 2003). And group diversity can measurably increase morale and cooperation when groups are proud of their own diversity (Cox, Lobel, & McLeod, 1991; Jehn, Northcraft, & Neale, 1999; van Knippenberg & Schippers, 2007).

On the cognitive side, the group clearly benefits from the range of perspectives, innovation, and quality of ideas introduced by diversity (e.g., Hoffman & Maier, 1961). The behavioral results of these creative cognitions are improved performance (Jehn et al., 1999) and communication with outsiders, probably because the group thinks in more complex ways and has more external resources (Milliken & Martins, 1996). Diversity of demographic or nonobservable attributes has enormous symbolic value for the organization, which may indirectly improve identification, commitment, and performance. Overall, some research supports the value in diversity hypothesis (Cox et al., 1991), but the road is rocky and paved with conflict. Team-building activities may help (e.g., stressing interdependence and common goals), as may the group endorsing the intrinsic value of its diversity, and simply waiting for time to heal the tensions, as people get used to each other (van Knippenberg & Schippers, 2007).

Joining a Group Occurs in Stages

Given a diversity of members, as well as variance in identity and attraction, how do a variety of people become members of the group? Consider the last group you joined (your academic concentration, a club). As Figure 12.4 indicates, joining a group is a process, not a single outcome (Levine & Moreland, 1994; Levine, Moreland, & Choi, 2001; Moreland & Levine, 1982; see also Bales & Strodtbeck, 1951). Even in formal groups with crisp calendars for initiating new recruits (e.g., colleges, the military), people go through a process of deciding to apply and being accepted, and then after joining, they spend time learning the ropes. For informal groups, joining is less cut-and-dried, but it still constitutes a process with certain typical characteristics. Individuals and groups move through as many as five potential stages that unfold over time and vary in degree of commitment:

· Investigation, when individual reconnaissance identifies groups that might meet that person's needs and, reciprocally, group recruitment identifies individuals who might meet the group's goals.

· Upon the individual's entry into the group, socialization, when the individual assimilates to the group and the group accommodates the individual.

· Upon mutual acceptance, maintenance, when both individual and group negotiate the person's role. If role negotiation succeeds, meeting both individual needs and group goals, mutual commitment remains high and maintains the membership.

· If role negotiation fails, interests diverge, and commitment falls, resocialization can attempt to accommodate the group and to assimilate the individual, which can prompt a return to maintenance.

· If interests continue to diverge, and commitment falls further, then the individual may exit the group, creating remembrance, when individuals reminiscence about the group and the individual becomes part of the group's history.

Figure 12.4 Phases of Group Membership

Research on the group socialization model has focused on the first, investigation stage (for some work on other stages, see Levine & Moreland, 1994). For example, prospective group members (college freshmen scouting for activities to join) were more optimistic about their own future experiences as group members than about the average other person (suggesting a self-enhancement motive for belonging; Brinthaupt, Moreland, & Levine, 1991). In general, college freshmen with good experiences in high school groups made more effort identifying and pursuing relevant groups (Pavelchak, Moreland, & Levine, 1986). Investigation is mutual, of course, and groups are particularly likely to be open to new members when their numbers (staffing levels) fall below what they need (Cini, Moreland, & Levine, 1993). But new members raise issues of trust, toward them and among existing group members (Moreland & Levine, 2002). Nevertheless, as with other kinds of diversity, new members bring the possibility of innovation (Levine et al., 2001).

Summary of Group Belonging

Belonging to a group, becoming a member, revolves around one's identity as a group member, which depends on context, as social identity theory and self-categorization theory predict. Individual concerns matter most when group commitment is low, and in that case, threats to self or group simply motivate individual solutions (self-affirmation or individual mobility out of the group). When group commitment is high, group-level concerns matter more, especially under threat of personal exclusion or threat to group cohesion. Depersonalized social attraction to the group fosters cohesion, as members identify with the group prototype, based on their interdependence as group members. Diversity both challenges and benefits the group. Minuses include potential discomfort, friction, and turnover, but pluses include creativity, innovation, and performance. Joining a group is a process, not an outcome, as groups and group members move through mutual investigation, socialization, maintenance, resocialization, and remembrance.

SOCIALLY SHARED COGNITION: UNDERSTANDING GROUP STRUCTURE

Having joined a group, members acquire an understanding of its norms, culture, roles, leadership, rankings, and coalitions. For groups to function, and especially for groups to function effectively, people have to have an understanding of what's in other people's minds: their positions, relations, norms, and values (Asch, 1952). Moreover, those representations must overlap if the group is to function. As a variant on the Chapter 4 comment derived from William James, namely, that thinking is for doing (S. T. Fiske, 1992), collective thinking is for collective doing (Levine, 1999). As noted in this chapter and last chapter's discussion of social identity and self-categorization, psychological membership in a group derives from linking oneself to the cognitive representation of the group as a whole, its prototype (Turner, Hogg, Oakes, Reicher, & Wetherell, 1987; Hogg, 2000, 2007). These must be socially shared cognitions for the group to function. That is, people's mutual understandings of the group prototype must be good enough for the group to operate effectively.

Social sharing constitutes one criterion for good-enough accuracy. Recall (from Chapter 4 ) that social consensus sometimes operationalizes accuracy. People's knowledge of each other's minds entails at least some accuracy in judging dispositions such as traits and attitudes, as well as emotions ( Chapters 3 and 4 ). In ongoing interactions, another kind of psychological state also matters, namely people's transitory thoughts and feelings. Ability to judge another person's short-term state constitutes empathic accuracy, as distinct from judging more dispositional qualities of the individual (Ickes, 1993). This kind of socially shared understanding—of the other person's goals, thoughts, and feelings—emerges from a history of interaction and an expectation of future interaction, part of any ongoing, psychologically meaningful group.

What's more, this kind of shared reality, social verification through joint experience, acknowledged by others, enables people to feel that their subjective experiences (otherwise transitory and random) are objective (fixed and systematic). People create and maintain their experiences through a social process of understanding, in this view (Hardin & Higgins, 1996). SIT/SCT approaches argue that social reality is objective reality for group members (as we will see in Chapter 13 ).

At the group level, operationally, shared understanding comprises people's knowledge of each other's respective contributions (expertise), their representation of systematic links (who works with whom), and subordination (who's under whom) (Weick & Roberts, 1993). For example, as flight operation crews on aircraft carriers heed these collective understandings, errors decrease. Operational reliability results from reasonably accurate, workable, socially shared understandings.

People are encouraged to develop a socially shared reality by various motivations, namely, to define the situation, reduce uncertainty, reach common ground, and communicate effectively (Thompson & Fine, 1999). As Figure 12.5 shows, these motivations for understanding drive goals for social interaction and group interdependence, which result in shared meaning. Although socially shared meaning includes affect-laden outcomes, such as group identification and self-enhancement through group membership, these issues were already discussed in this and the previous chapter. Socially shared meaning that results in more overt behavior (such as performance effects) appear in a later section.

Figure 12.5 Developing Socially Shared Meaning in Groups

A third class of shared meaning is cognitive, including shared mental representations, transactive memory, and shared mental models. These cognitive forms of social meaning lie at the heart of this section. Shared mental representations include social representations (Moscovici, 1988), which are everyday, collective ways of explaining strange, new, cultural trends as something more familiar (a current dictator as Hitler, Internet messages as a form of mail, portable computers as notebooks). Social representations emerge from group processes to create commonsense knowledge, framed by the social position of the group vis-à-vis other groups. For example, some types of religious groups, observing that AIDS was at first overrepresented among gay men, spontaneously framed AIDS as divine retribution, whereas anti-American groups framed it as a CIA plot, and racist groups framed it as an African problem (Lorenzi-Cioldi & Clémence, 2001).

Shared mental representations also include transactive memory (Wegner, 1986), in which group members delegate different domains of memory expertise to different people and perform better as a result (Hollingshead, 1998; Wegner, Erber, & Raymond, 1991). For example, one group member might become the informal historian, remembering precedents; one might become the informal ambassador, remembering outside contacts; another might become the organizer, remembering deadlines. Also relevant to group performance are shared mental models, which reflect a group's need to hold common representations of task requirements (goals, roles, processes; Klimoski & Mohammed, 1994). A mental model is essentially equivalent to an expectation ( Chapter 4 ). The remainder of this section focuses on various kinds of mental models, focusing in turn on how people develop a shared understanding of group norms, roles including leadership, and structure including subgroups.

All Norms Are Local

Why do concertgoers sometimes get crushed? Why do victorious sports fans sometimes rampage through downtown, breaking windows and overturning cars? Why do race riots occur? Older ideas had explained the sometimes violent and destructive behavior of mobs as a loss of individual responsibility and autonomy. One of the original social psychology theories, LeBon's (1895/1995) theory of crowd behavior, held that the collective mind took over when people in a crowd become anonymous, highly suggestible (easily influenced), and subject to contagion (mindless adoption of others' behavior). In more scientific terms, Festinger, Pepitone, and Newcomb (1952) described a state of deindividuation in which people do not act as individuals, and inner restraints against counternormative behavior fall away, leading to classic mob behavior. A later version of the theory (Zimbardo, 1969) proposed that even individuals could become deindividuated (disinhibited), given such conditions as anonymity, low individual responsibility, arousal, mind-altering substances, sensory overload, or ill-defined situations. Another account focused on lack of self-awareness (recall from Chapter 6 ), leading to deindividuation (Diener, 1980). Typical studies manipulated group size, anonymity, or self-awareness to induce deindividuation. They then measured disinhibition, stealing, cheating, aggression, or failure to help (Postmes & Spears, 1998). Although these are classic social psychology accounts, they may be wrong according to more recent evidence, which suggests groups developing a shared social understanding of local norms that guide behavior. Instead of disinhibited behavior in response to deindividuation, crowd behavior may better fit conformity to perceived group norms.

Norms are behaviors of group members that act as implicit rules, considered to be both descriptive of what group members are and prescriptive of how they should be (Levine & Moreland, 1998; Miller & Prentice, 1996). Norms emerge as groups become coherent, developing socially shared understandings. In one view, mere aggregates of people—crowds—become coherent groups as people mill around, exchanging rumors, information, interpretations, and norms, which become the basis for action. Crowds become groups by information exchange, according to emergent norm theory (Turner & Killian, 1987). But this theory does not account for rapid changes in understanding, those that occur too quickly to have followed a process of milling around, all people communicating with all others.

Self-categorization and social identity theory (SCT/SIT, cited earlier) do better. If people identify as members of a particular group, then its norms follow. For example, a speaker at a rally will often evoke particular group identities (American citizenship, oppressed ethnicity, neighborhood residence) in order to evoke particular norms (thus, shared ideas about what the present group should do) and therefore shared actions. Meta-analysis of 60 studies fails to support predictions from the older deindividuation theories but instead indicates that situation-specific norms predict collective behavior (Postmes & Spears, 1998). For example, although a crowd's behavior may be counternormative with respect to society at large, the crowd as a group expresses its own norms for appropriate behavior in that particular situation, which may include smashing store windows or even attacking certain (outgroup) categories of people.

In particular, one SIT model of crowds, the social identity model of deindividuation (Reicher, Spears, & Postmes, 1995), argues that as a particular group identity becomes salient, people strategically express that identity by contextually evoked behavior. Under an elaborated social identity model (Reicher, 2001), people's identity in crowds develops as part of the intergroup interaction. An observational research example: Crowds in various protest movements often initially represent both moderate and extreme factions, but an external group (the police or an opposing protest movement) may treat them all alike as equally extreme and illegitimate. The outgroup (police) defining the ingroup (protesters) as a homogeneous category causes the outgroup to move against the ingroup crowd, unifying the heterogeneous factions in the face of such treatment. As political activists know, police action can radicalize the moderates.

Group behavior may look irrational from the outside but may instead be driven by the contingencies of the particular situation, which include the behavior of powerful outgroups. Shared social norms define what behavior the ingroup members view as contextually appropriate, according to their potentially changing definition of their own identities in the intergroup context. Deindividuation, in this view, depersonalizes the individual and makes ingroup identity cognitively salient, enhancing the importance of ingroup norms. At the same time, deindividuation makes it strategically possible for group members to express their identity if they are identifiable to the ingroup but not to the outgroup, thereby freeing them to act on their own group's norms with fewer concerns for consequences from an opposing outgroup. Clearly, this view of norms is highly sensitive to the context.

More generally, defining norms as local is far more effective than expecting group members to act on abstract societal norms. For every abstract norm, some subgroup defines itself in part by acting against it. (Try thinking of a societal norm that does not get contradicted by at least one social group, as we have defined it.) Like politics, all norms are local. According to Miller and Prentice (1996), norms are a function of context, which is inherently local. For example, research shows that people's global self-esteem and judgments of their own ability depend entirely on their standard of comparison, which tends to be local. That is, students in a high-caliber school actually view themselves less favorably than do students of the same ability level in a low-caliber school (Marsh & Parker, 1984), because of their frame of reference. Also, recall from Chapter 7 the importance of physical attractiveness to initial liking: Even standards for attractiveness can change, depending on the attractiveness of others in context—presumably a kind of social norm—changing evaluations of one's potential romantic partner (Kenrick & Gutierres, 1980; Kenrick, Gutierres, & Goldberg, 1989). Local norms and standards guide perceptions and behavior.

As implicit rules, norms are not given. People must infer norms (Miller & Prentice, 1996), and they do so by direct communication (“this is how we do things here”), observing behavior (everyone walking around in orange and black jackets), and extrapolating from self-knowledge (if I am willing to do it, then everyone in my ingroup must be). Direct communication might seem the least ambiguous method of socializing people to norms, but in practice, socialization involves both what is said and what is unsaid. What is said depends on a host of factors, all reflecting the norm's apparent support, on the basis of social desirability, perceived trends, and known traditions. What is unsaid depends paradoxically on what people assume is so obvious or normal that it does not have to be said. For example, people explain the gender gap in voting behavior (e.g., whereby women tend to show less support for war) by explaining the nonnormative group (in this case, women, as I did here; Miller, Taylor, & Buck, 1991). This implies that the standard, normative voter is male, so women need to be explained, just as the standard, normative parent is female, so the fathers' behavior has to be explained.

Inferring norms is not the only process in constructing them. People also pick certain others to observe. As noted in Chapter 5 , social comparison serves functions of understanding and self-enhancement by evaluating self in comparison to norms represented by particular others one chooses to heed. Several principles apply. In making comparisons, first, people heavily weight others who are similar on characteristics relevant to the judgment, according to related attributes theory (Goethals & Darley, 1977). Second, when people do construct comparisons by relevant attributes, those are relatively universalistic assessments, but people often prefer to weight ingroup others, those relevant according to shared identity, and make particularistic evaluations, rather than general ones (Miller, Turnbull, & McFarland, 1988). People may specifically seek comparisons on attributes relevant to performance in that particular group, because esteem derives from group feedback (Darley, 2001). For example, the relevant standards differ considerably in task groups and affinity (social) groups. Third, people may compare to groups that are not their own membership groups, that is, reference groups they admire and want to emulate (or derogate and want to avoid), even if those are not their ingroups. Not all membership groups are reference groups and vice versa. Finally, as noted in Chapter 5 , people sometimes prefer esteem-enhancing downward comparisons that promote comfort and reduce threat, but sometimes they do seek self-improving upward comparisons. All forms of comparison appraise the self vis-à-vis norms perceived to be relevant in a given context, so those norms often are quite specific.

Moreover, the salience of norms will differ, as will their importance in guiding behavior. Both the salience of norms and accountability to the ingroup will increase the importance of enforcing ingroup norms (Marques, Abrams, Paez, & Martinez-Taboada, 2001). Norms matter because people who do not conform to the local group norms pay a substantial price, if the group matters to people and if the deviance occurs on a dimension relevant to the group's purpose. Group members first try persuading ingroup deviants to change and then reject deviants, as classic social psychology has shown (Schachter, 1951). Deviants who consistently maintain an opinion that differs from modal group opinion are rejected, especially if their subgroup size is substantial, according to meta-analysis (Tanford & Penrod, 1984; Tata et al., 1996).

Group members enforce ingroup norms by several techniques. One way is by derogating deviant ingroup members and modal outgroup members (both of whom challenge ingroup norms and support outgroup norms) (Marques et al., 2001). Another way is by enhancing those who conform (ingroup modal and outgroup deviant). Group members bolster their social identity by enforcing ingroup norms, as we saw earlier in arguing for the primacy of the ingroup, and the resulting black sheep effect that penalizes ingroup deviants even more than typical outgroup members. Another piece of evidence argues for the importance of defending ingroup norms, as a route to a legitimate and distinctive ingroup identity. When ingroup norms are especially uncertain, group members particularly derogate an ingroup deviant (Marques, Abrams, & Serodio, 2001). Finally, it is not deviance from the ingroup average that matters, so much as counternormative deviance; ingroup members who deviate in favor of the group norm—but more extremely than the modal ingroup member—are valued. Pronorm deviance is preferred (Abrams, Marques, Bown, & Henson, 2000). It's not the deviance the group rejects, it's the norm that the group protects.

According to a theory of subjective group dynamics (Marques, Abrams, Paez, & Hogg, 2001; see Figure 12.6 ), descriptive ingroup norms often operate automatically, as in categorizing “us versus them” by salient physical features (see top intergroup half of figure). The automaticity of social categorization (e.g., by gender, race, and age), as well as social categorization theory's meta-contrast principle (both described in Chapter 11 ), point to this type of rapidly assessed descriptive ingroup norm. People automatically classify self and others, making ingroup deviants salient.

Figure 12.6 A Model of Subjective Group Dynamics, Separating Automatic Intergroup and Deliberative Intragroup Levels

However, ingroup deviants make salient the prescriptive, intragroup norms (lower half of figure): Why doesn't this person fit in? This question matters, as noted, more for ingroup than outgroup, due to needs for positive identity. This two-process model is analogous to others we have seen, whereby the default is rapid and automatic, whereas the troubleshooting deliberative mode intervenes in the event of problems. The subjective group dynamics bring to people's minds their interdependence with their ingroup, as well as renewing awareness of and commitment to ingroup prescriptive norms; this results in derogating outgroup deviants, as in the black sheep effect. The model thus shows both inter- and intragroup dynamics.

When ingroup deviants make prescriptive norms salient, people try to explain the deviant, reasoning backward to consider the standard the person has violated. Norm theory (Kahneman & Miller, 1986) describes this kind of backward processing. When people encounter an unexpected event, they reason backward to generate a context-specific frame of reference (i.e., a norm) to understand why the event feels counterintuitive. The reasoning process creates a counterfactual (what could have been); Chapter 4 described this process of mentally undoing an unfortunate event, but the theory applies to any unexpected event, such as an ingroup member who deviates from ingroup norms. Constructing what-should-have-been makes people aware (in retrospect) of the norm that has been violated, fixing the prescriptive norm that is otherwise taken for granted (Forsyth, 1995).

Norms overall depend on the immediate group context and one's social identity. Norms matter because they express group members' social identity. Norms matter because group members influence each other (Tanford & Penrod, 1984). Norms matter because they predict attitudes (Cooper, Kelly, & Weaver, 2001) and behavior (Ajzen, 1996; Terry & Hogg, 1996) that basically conform to the group. Norms also matter because they show how motivated people are to belong to the group and understand its implicit rules, to avoid rejection, as well as to express social identity.

Roles Include Leadership and Much More

People typically conform to group norms, for reasons of both expressing group identity and avoiding group censure. Nevertheless, people also need to balance the self and the group. Social roles serve this function. Social roles—expectations for the behavior of a person in a particular position—differ from norms because they pertain to a single person or set of people occupying that position (e.g., Levine & Moreland, 1998). Because of this mix of individual and group, social roles meet needs both for connectedness to the group (belonging) and for autonomy of the self (controlling), as viewed through the lens of need satisfaction (Bettencourt & Sheldon, 2001). Students in both ongoing and ad hoc social groups rated roles that fostered their own authentic self-expression as also being those that best met their needs for group relatedness. Autonomy and relatedness each separately predicted subjective well-being. This principle also supports Brewer's optimal distinctiveness theory ( Chapter 11 ). Thus, understanding and fulfilling one's social role can combine what might otherwise seem to be conflicting motives for group belonging and individual controlling.

Roles also function well for the group (Levine & Moreland, 1998), developing as the group needs them, rewarding those who fulfill those group functions, and improving group dynamics. Group roles that come immediately to laypeople's minds are leader and follower, but the range of potential roles is far more complex (consider a club you know well): task facilitator, socioemotional facilitator, idea person, recorder, scapegoat, devil's advocate, entertainer, ambassador, realist, and more. Ironically, the functional contribution of roles has been studied less than their dysfunctional qualities. Role dysfunctions (again, think of your club) include conflict over who should have what role, failed consensus over what the role entails, ambiguity for enacting the role, strain in playing the role, conflict between simultaneous roles, and transitions into and out of roles.

One reason most roles may escape research attention could be the difficulty of feasibly operationalizing them. Some work manipulates roles by defining and then assigning them to ad hoc groups (Bettencourt & Sheldon, 2001). Other work defines roles by their place in the SYMLOG dimensions introduced earlier: upward/downward, forward/backward, and positive/negative (Isenberg & Ennis, 1981). Using only the upward/downward dimension, a specific operationalization of the leadership role could be sheer talking time (Mullen, Salas, & Driskell, 1989; Stein & Heller, 1979); according to meta-analysis, this holds regardless of whether group members or observers identify the leader and whether the group is real or artificial.

Nevertheless, leaders are something more than the biggest talkers. Leaders do tend to be extraverted, open to experience, and conscientious (Hogg, 2007a; 2010). According to meta-analyses, they are close to others, central to group communication, satisfied with the group, and esteemed (Bass, 1954; Mullen, 1991). How does this happen? Leadership depends less on individual personality characteristics than people—including social psychologists—think at first (Chemers, 2000; Hogg, 2007a; 2010; Hollander, 1985; Levine & Moreland, 1998). People who are leaders in one group are not necessarily leaders in another group, in part because group tasks may require a variety of expertise (motor, artistic, mathematical, literary, social, spatial; Barlund, 1962). Nevertheless, consistent leaders do emerge, with ratings that average as much as 64% due to something about the individual, perhaps adaptability to different contexts (Kenny & Zaccaro, 1983). Four major types of theory address contextual factors, stressing the leaders' responsiveness to a particular group.

First, contingency theories (e.g., Fiedler, 1964, 1978) hold that effective leaders emerge as a function of the interaction between person and situation. In this theory, the leader's own personal style (relative task or social emphasis) interacts with the group's degree of situational control (degree of authority, task clarity, quality of relationships). When situational control is moderate, a social style should work well. But a task emphasis should work well when control is either very low (structure is needed) or very high (efficiency is fine). Leadership style is operationalized by a test of the leader's regard for the least preferred coworker(less dislike for one's least favorite person indicates greater tolerance and more social orientation). The original datasets analyzed to produce this curvilinear theory supported it better than subsequent studies did, but the overall support is not bad (Graen, Alvares, Orris, & Martella, 1970; Peters, Hartke, & Pohlmann, 1985; Strube & Garcia, 1981). Subsequent theories picked up the idea of leadership interacting with the group's situation (Chemers, 2000; Howell, Dorfman, & Kerr, 1986). This type of theory echoes core social motives of group effectiveness and control over outcomes, as well as shared understandings of who's in charge and what that means.

A second wave of leadership theory, transactional leadership theories emphasize leader and follower exchange of valued resources (praise, esteem, loyalty, as well as tangible rewards). In this view (Burns, 1978), ordinary people as followers need not reach consensus with each other or their political and ideological leaders. Instead, conflicts create a dynamic reciprocal exchange that in turn creates leaders. That is, the leader is the one who can distribute resources the most effectively and receive resources in turn from each other group member. This theory, which emphasizes motives of control over resources and self-enhancement, also relies on shared understanding of and compliance with group norms.

Leadership theory based on contingency or transaction did not, however, sufficiently explain cultural and demographic differences, so it returned to some of the personality ideas that started this line of work. A third wave of leadership research, cognitive theories emphasizing leadership prototypes, makes some inroads here, clearly demonstrating the motive for shared understanding constrained by a particular context (Lord, Brown, & Harvey, 2001; Lord & Maher, 1991). Leader prototypes specify intelligence, masculinity, and dominance, according to meta-analysis (Lord, de Vader, & Alliger, 1986). Moreover, across cultures (62 nations; Den Hartog et al., 1999), people's implicit theory of leadership tends to agree on ideal leaders being visionary, inspirational, team-integrating, trustworthy, diplomatic, and not self-centered, malevolent, inexplicit, or autocratic. Cultures differ the most on whether leaders are supposed to be harmony-oriented (conflict-avoidant, subdued, sensitive). A leader can be both dominant and insist on avoiding conflict (some collectivist cultures) or dominant and tolerate conflict (some individualist cultures). Whatever the cultural expectation, people unconsciously evaluate potential leaders on their fit to salient leadership prototypes (Emrich, 1999), make attributions on that basis (Phillips & Lord, 1981), and remember accordingly (Lord, Foti, & de Vader, 1984; Phillips & Lord, 1982).

To the extent that cultures expect leaders to be masculine, people view women as being poor candidates for leadership. According to this lack-of-fit model(Heilman, 1983, 2001), people's expectations for good managers and for typical women do not mesh—indeed, prototypical managers have masculine characteristics—so people tend to be biased against the idea of women as leaders. Meta-analyses indicate a small overall tendency to evaluate female leaders less favorably, but this is particularly true when the female leader behaves in a stereotypically masculine style, that is, outside the gender role, though inside the leader role (Eagly, Makhijani, & Klonsky, 1992). This is unfortunate, given that actual leaders in organizations do not differ by gender in their leadership styles (Eagly & Johnson, 1990), so quite a few task-oriented women leaders are likely to end up being devalued. Men and women leaders are equally effective, according to meta-analysis, but not surprisingly, men fare better when leadership is defined in more masculine terms (Eagly, Karau, & Makhijani, 1995). All these results tend to support a social role theory of leadership, emphasizing perceived lack of fit for women in certain kinds of leadership roles. Bias in evaluating women leaders is global (Schein, 2001), resting on the assumption that men will be more legitimate and competent leaders, according to expectation states theory (Ridgeway, 2001). Certain ascribed status characteristics (e.g., being male, white, older) lead to assumptions about who will (and will not) top the hierarchy. Women do disproportionately hold roles incompatible with enhancing hierarchies and themselves leading them. Women instead tend to hold hierarchy-attenuating roles (helpers and caretakers, rather than enforcers and competitors), as predicted by social dominance theory (Pratto, Stallworth, Sidanius, & Siers, 1997; introduced in Chapter 11 ). The dearth of women in hierarchy-enhancing positions seems to be due not only to hiring biases, as lack-of-fit would predict, but also to self-selection (women opt out) and value-matching (hierarchy-enhancing values tend to favor men). However, women's leadership style is actually effective for both genders (Eagly et al., 1995).

Social identity theories, the fourth and new wave of leadership theories, take the concept of leader prototype one step further (Hogg, 2001, 2007a, 2010). In this view, group members differ in how much they fit the prototype of the group. Recall that depersonalized social attraction to the group encourages cohesion, because group members identify with the group prototype, based on their interdependence in the service of shared goals. Because of members' attraction to the prototype, the person who best fits the group prototype will have the most influence and therefore be the leader. Consensual social attraction also confers apparent status, which leads to power relations in which followers depend on leaders and attend to them as unique individuals (Fiske, 1993; Fiske & Dépret, 1996). Correspondence bias or the fundamental attribution error ( Chapter 3 ) assigns the leader's influence to allegedly special features of that individual's personality. All this results in a leader being seen as the one who best represents the group's identity. Group membership has to be salient for this process to occur, creating a frame of reference in which group members see themselves as similar to each other (Hogg, Hains, & Mason, 1998). When people do identify with the group (i.e., it is salient to them), they endorse leaders prototypical of the ingroup (Platow & van Knippenberg, 2001), as the social identity theory of leadership would suggest.

Regardless of which leadership theory one endorses, leaders' behavior affects people in the reciprocal role, that is, followers. Declaring someone a leader can make others work less hard, for example (Kerr & Stanfel, 1993), but generally, people like to have a leader. To the extent the leader leads (by initiating structure or considering others), followers are more satisfied, according to meta-analyses (Lowe, Kroeck, & Sivisubramaniam, 1996; Mullen, Symons, Hu, & Salas, 1989; Wofford & Liska, 1993). Groups need leaders to coordinate effectively, but leader-follower ambivalence is perhaps inevitable, due to possible exploitation (Van Vugt, Hogan, & Kaiser, 2008; also see next chapter's discussion of power).

According to meta-analysis, people prefer democratic leadership, especially when it is authentically democratic, and this is more true as group size increases (presumably because smaller groups have more cohesion and less need of explicit democracy) (Foels, Driskell, Mullen, & Salas, 2000; Gastil, 1994). Legitimate authority in groups rests on procedural justice—that is, on having fair processes—not just on people's own personal outcomes (Tyler & Lind, 1992). People want to feel that, in being treated fairly, they are respected by the group and can feel proud of the group (Tyler, Degoey, & Smith, 1996); in other words, their belonging is affirmed. Democracy doubtless ranks high as a fair process because it respects everyone's opinions as an equally valued member of the group.

Many of the leadership theories address a shared understanding motive, specifying who and what the leader is to be. In addition (Chemers, 2000, 2001), leaders function to meet people's other core social motives by establishing credibility (trusting), relationship (belonging), and effectiveness (controlling). The leadership role is, by definition, the most influential role, but all roles embody the shared understanding of group structure.

Subgroups: Minorities and Majorities

So far, we have examined mutually understood group structure in the form of norms as implicit rules and individual roles emphasizing leadership. Now we turn to structures of member relationships within groups. The patterns of member relations represent various networks of liking, status, influence, and communication. As Figure 12.7 indicates, people may like each other reciprocally or not, as revealed by various networks of sociometric choice (Moreno, 1953; introduced in Chapter 7 ). People's communication patterns (frequency of who-to-whom) constitute part of the SYMLOG record already mentioned. And influence hierarchies appear in members' judging each other and self on (e.g.) influence, contributions, and popularity (Arrow, 1997). By these measures, small groups show a fair degree of equilibrium in structure over time, readjusting after disruptions such as member turnover. Status hierarchies are communicated by repeated enactment of deference (hesitant, unconfident, unassertive behavior); they quickly result in consensus judgments of members' status and competence (Ridgeway & Erickson, 2000).

Figure 12.7 Patterns of Sociometric Choice Typical in an Early Field Study

Source: Moreno, 1953.

Subgroups inevitably result from patterns of influence, status, communication, and liking. Social impact theory (Latané, 1981; to be addressed in more detail next chapter), a dynamical social computational model, argues that four features of group dynamics result from spatially adjacent individuals influencing each other. These features are the relevant point here: (a) consolidation of the group, (b) clustering of subgroup opinions, (c) correlations of opinions within subgroup clusters, and (d) continuing diversity by protecting minorities within subgroups (Latané & Bourgeois, 2001). Because people most often interact with and influence those adjacent to themselves, even if the group started out with randomly distributed opinions, subgroup opinion clusters would eventually result, as shown via computer models simulating social impact on attitude change (Latané & Bourgeois, 2001). For current purposes, the point is that the existence of subgroups within an overall group is likely. And recognizing this inevitability may even maintain group harmony (Hornsey & Hogg, 2000; Kameda & Sugimori, 1995; see Chapter 11 on subgroups).

Whenever subgroups occur, some will be larger and some smaller. As our discussion of group norms has implied (and as the topic of conformity in the next chapter will develop), people typically conform to the larger group. The 1940 Asch line-perception conformity study, introduced in Chapter 1 , nicely illustrates that idea and has had substantial impact. In reaction to all that emphasis on conformity to majority group norms, Moscovici (1976, 1980, 1985) argued that minority influence is important to innovation. It results from the conflict that minorities create (Levine & Russo, 1987; Martin & Hewstone, 2001). In this view, minorities are a force for innovation because they advocate their positions with consistency, confidence, and conviction, so they disrupt the existing norm, creating uncertainty. Because they will not budge, they demand attention for their alternative position, and the majority is forced to accommodate and consider their understanding of the issues. As originally conceived, minorities provide informational influence, pressuring group members to validate their position and convert by virtue of the merit of the information available (thus, the understanding motive). In contrast, people conform to majorities by normative influence, whereby they compare themselves to other group members and do what they do (thus, the belonging motive). As the next chapter will show, these two forms of influence are applied more broadly than minority and minority influence (but appear here for the first time in this chapter). A related two-process idea proposed different kinds of thinking engendered by majorities—convergent thinking (unreflective acceptance of conventional wisdom)—and minorities—divergent thinking (attending to a wider range of information and reexamining premises to come to more creative decisions) (Nemeth, 1986).

Research supports some, but not all, aspects of the theory (Wood, 2000). Minority impact results from a consistent but flexible ingroup minority whose deviant opinions otherwise fit the group's values (Mugny, Perez, & Lamongie, 1991). Ingroup minorities are influential when their stands do not threaten the group's existence (Crano, 2000). The unexpected distinctiveness of the ingroup minority's message encourages careful consideration of their viewpoint, with a leniency contract (benefit of the doubt) accorded to them because of their ingroup status (Alvaro & Crano, 1997; Crano & Chen, 1998). Minority influence can be persistent precisely because it is information-based, especially if the topic is not central to the group and the minority is active (Kerr, 2002). Individualistic cultures are especially likely to show higher decision quality under minority influence, but collectivistic cultures do also, if the minority influence agent is high status (Ng & Van Dyne, 2001).

In an odd twist, minority influence is most often indirect, according to meta-analysis (Wood, Lungren, Ouellette, Busceme, & Blackstone, 1994). That is, minorities have an influence on topics that are related to, but not equivalent to, their focal arguments, as if majority members do not want to admit that the minority changed the majority's mind. Consistent with this form of indirect influence is a stronger minority effect on measures that are private, concealed from the source. People's self-esteem suffers when they find themselves aligned with derogated minorities (Pool, Wood, & Leck, 1998), so they shift accordingly (Wood, Pool, Leck, & Purvis, 1996).

Identification with the ingroup is a necessary condition for minority influence (Leonardelli, 2001). This fits much of the other work we have seen that emphasizes the importance of social identity and the primacy of the ingroup. Outgroup minorities have little influence.

Some of the dynamics of subgroups depend on sheer impact of numbers on cognition. The proportionate sizes of different groups affect who captures attention; small numbers attract attention because they stand out (Taylor, 1981; Taylor & Fiske, 1978). As noted in Chapter 4 , salience has social cognitive consequences. The smaller perceptual unit (minority) appears as a figure against the larger group as a background. A series of meta-analyses focused on salience, as a function of groups' proportionate size (Mullen, 1991), indicated that people in the minority have more self-focused attention, overestimate their own consensus, exaggerate their own ingroup bias, see themselves as more alike, and see group differences as larger. Reciprocally, majorities do overattend to minorities, overestimate their consensus, and view them as all alike. So the cognitive effects of proportionate subgroup sizes are a two-way street.

What are the consequences for members of smaller subgroups to have more self-focused attention (Mullen, 1983)? Solos or tokens (who are the only members of their social categories) can be distracted by the experience of other people attending to them and their own self-consciousness as a result. Consequently, they may not remember as much as uninvolved observers do (Lord & Saenz, 1985); the memory deficit appears to result from being worried about the scrutiny of others and the impression one is making (Lord, Saenz, & Godfrey, 1987). Indeed, if tokens mentally restructure their role so that they are the ones judging others, instead of focusing on being judged, the effect completely evaporates (Saenz & Lord, 1989).

Whether single individuals or small subgroups, minorities within groups are critical to understanding the group's structure, even if much of the influence they exert is indirect. Some of the consistency and conviction of opinion minorities may result from the cognitive factors that increase their ingroup focus.

Summary of Group Cognition

Group structure results from socially shared understandings of the group. Norms specify implicit local rules for group members' behavior. Specific individuals fill roles that specify the behavior expected of them, and leaders fill a particularly important role. Subgroups within the group emerge as a function of differing opinions, and ingroup minorities can elicit thoughtful but indirect forms of opinion change.

PERFORMANCE: UNDERSTANDING AND CONTROLLING

Shared-group understanding determines group performance. To the extent that cognitions, identities, and preferences of group members are shared, those socially shared ideas will have a greater impact on group outcomes (Tindale & Kameda, 2000). Social sharedness is crucial to group decision making and performance, in keeping with the core social motive of understanding. Not only are group members motivated to have socially shared understandings, but they also are motivated to obtain favorable outcomes; that is, they show the core motive for controlling what happens to them. Typically following both the understanding and controlling motives, social psychologists' research on group performance focuses on decision making and productivity.

Decision Making

Conventional wisdom promotes group decisions as more considered and balanced than individual decisions, resulting in better-quality decisions overall. However, fifty years ago, two group phenomena proved conventional wisdom to be wrong. First, groups make more polarized (extreme) decisions than the aggregation of individual decisions. Second, they pressure group members to agree and not to consider deviant viewpoints. These two decision phenomena have shaped the study of group decision making ever since.

GROUP POLARIZATION

Early research indicated that group discussion surprisingly leads to more risky decisions than the original preferences of any of the group members as individuals (Wallach, Kogan, & Bem, 1964; for an early review, see Myers & Lamm, 1976). Although originally dubbed the risky shift, it turns out that groups go to extremes in either a more risky or a more cautious direction, so the effect more accurately became group polarization. Meta-analyses (Isenberg, 1986) initially supported either of two processes: (a) Merely knowing other people's views creates a normative influence process of social comparison, whereby group discussion reveals cultural values more extreme than those of individuals, so members shift to gain approval. (b) Hearing other people's arguments creates an informational influence process of persuasive arguments, whereby group discussion highlights novel arguments that favor its initial tendency, causing group members to become more extreme (Burnstein & Sentis, 1981). Later work also supported (c) a social-identity/self-categorization theory approach, whereby group members construct a group prototype, to which they conform (Turner, Wetherell, & Hogg, 1989). This ideal prototype may serve identity by being more extreme than the average group member's opinion, so conformity leads to polarization (McGarty, Turner, Hogg, Davidson, & Wetherell, 1992). As a variant on this process, priming the ingroup leads to attitude assimilation, whereas priming the outgroup leads to contrast, creating attitude polarization depending on context and group identity (Ledgerwood & Chaiken, 2007). Still another possibility does not depend on the individual feeling connected to the group: (d) people become more convinced, the more they repeat their own arguments (Brauer, Judd, & Gliner, 1995), consistent with mere exposure, availability, or self-enhancement.

GROUPTHINK

Groupthink illustrates the failure of groups to consider all available and relevant information in making a decision (Janis, 1972). First observed in policy-making groups responsible for some incredible international fiascos (e.g., the United States 1961 Bay of Pigs invasion of Cuba), groupthink theoretically results from a “psychological drive for consensus at any cost that suppresses dissent and appraisal of alternatives in cohesive decision-making groups.” Groupthink theoretically occurs when the group is cohesive and isolated, the leader is biased, procedures are unclear, and members are homogeneous. Provocative situations include high stress from external threats with no good solution, and a temporary drop in self-esteem because of recent failures, current difficulties, and moral dilemmas.

Although an appealing theory that deserves to be true, and although evidence from descriptive studies seemed supportive, nonetheless groupthink experiments failed to pin it down clearly (Aldag & Fuller, 1993; Kerr & Tindale, 2004; Mullen, Anthony, Salas, & Driskell, 1994). General models of group problem solving currently seem more useful.

GROUP PROBLEM SOLVING

Decision-making groups are trying to solve a problem, and clearly both task and social dimensions matter. For example, information sharing is key to task performance. Individuals often prefer information that supports their initial attitudes over information that conflicts, as Chapter 6 indicated. Groups prefer supportive information as well (Schulz-Hardt, Frey, Luethgens, & Moscovici, 2000; Stasser & Titus, 1985). Group members also prefer to discuss information that is already shared, rather than unshared, so shared information has more influence on group decisions (Kerr & Tindale, 2004; Stasser & Titus, 1985; introduced in Chapter 1 , Table 1). The common knowledge effect—the influence of previously shared information—could result either from sheer likelihood of being brought up in the discussion, from people's preference for shared information, from carrying more weight in individual judgments, or from premature closure, the sheer press to decide (Gigone & Hastie, 1997; Kerr & Tindale, 2004; Stasser & Titus, 1987). Groups do show this less-than-optimal tendency to neglect unshared information, especially under cognitive load and time pressure, consistent with groupthink predictions. They do, however, get to unshared information eventually, over time (Larson, Foster-Fishman, & Keys, 1994; Stasser & Titus, 1987). And knowing who knows what—that is, assigning certain people to be experts—facilitates the mentioning of unshared information (Stewart & Stasser, 1995).

Social cohesiveness is also key, in keeping with social identity approaches. Depersonalized social attraction promotes groupthink, whereas personal friendships undermine it (Hogg & Hains, 1998). When other antecedents (e.g., overly directive leadership) promote groupthink, cohesiveness exaggerates the detrimental effect on group decision making (Mullen, Anthony, Salas, & Driskell, 1994). This holds true especially for cohesiveness operationalized as attraction.

A series of such results shows how complicated group decision making really is. No single broad-brush model will make sense of all the important factors. Parts of the process are indicated in Figure 12.8 (Aldag & Fuller, 1993). Various antecedent conditions include characteristics of the decision (e.g., its importance, time pressure), the group structure (e.g., cohesiveness, homogeneity), and the context (goals, political realities). Emergent group characteristics result from this combination of a particular group making a particular decision in a particular context. For example, a group's perceived vulnerability might affect its openness to processing negative feedback. The decision process itself then entails a variety of information processing and decision stages, resulting in not only the decision, but also political implications for leaders and group, as well as their satisfaction. All these factors influence how groups go about solving the problem of making their decision.

Figure 12.8 General Group Problem-Solving Model

PREDICTING GROUP DECISIONS

If groups' bias equals individuals' bias, how do group decisions result from those individuals (Kerr, MacCoun, & Kramer, 1996)? Social decision rules(consistent patterns in the group's rules for its decisions) use less cognitive effort than weighing all the pros and cons of each individual person's preferences. For example, many groups use “majorities or pluralities win” to make binary choices (Davis, 1973), especially when no clearly correct alternative emerges. This decision rule may be explicit in group decision procedures; even if not, the final result fits a formal model (mathematical equation) in which the group behaves as if using this rule. “Majority rules” is an adaptive policy (Kerr & Tindale, 2004). If choices require decisions along a continuum, then the rule is “median wins” (Crott, Szilvas, & Zuber, 1991; Davis, Hulbert, Au, Chen, & Zarnoth, 1997); note that the group mean (arithmetic average) does less well than the median (middle value). Again, this is a formal mathematical representation of how to predict the outcome, not the psychological process of group members' decision. If choices require an active consensus building for a decision along a continuum, then the group follows a weighted linear combination, where members' opinions receive more weight if their position is closer to the group's center (Davis et al., 1997).

This last model fits the social sharedness perspective, whereby group members are influential in direct proportion to their representing the group. As the person becomes more cognitively central, the person's influence increases because of perceived expertise resulting from the group agreement (Kameda, Ohtsubo, & Takezawa, 1997). This would contrast with a social-identity/self-categorization model, in which a more extreme (but idealized prototype) group member might carry more weight. All three of the formal models predict the idea of group polarization, in the sense of exaggerating the majority position among individual preferences because members closer to the group median carry more weight (Tindale & Kameda, 2000). The potentially moderating influence of people who disagree is apparently lost in most group decisions, consistent with the findings of group polarization.

Other kinds of decision rules operate when the group must decide on a demonstrably correct answer. In that case, “truth wins,” if members share beliefs about how to establish what is true (Laughlin & Ellis, 1986). Groups develop mental models of the problem, just as they develop the previously introduced mental models of norms, roles, and subgroups, functioning better when those understandings are shared (Tindale, 1993).

Productivity

Groups make decisions in order to be productive, to further their goals, and to obtain desired outcomes. All these goals entail the core social motive to control individual and group outcomes. Depending on its type of work, a group's productivity concretely may include various records of output: revenues, publications, recordings, performances, demand met, and work remaining unfinished. Records of production quality also may include inspector- or manager-assigned evaluations (Sundstrom, McIntyre, Halfhill, & Richards, 2000). To the extent that group productivity involves processing information, group motivations explicitly involve the core motive of understanding. Here, cognitive commonality matters to productivity: shared information, as noted earlier; overlap of ideas; depth of cognitive processes; and experience of belonging (Hinsz, Tindale, & Vollrath, 1997). Both understanding and controlling run through productivity, along with the core motive of belonging.

In attempting to understand information, control outcomes, and belong, people often think they will be more productive in groups than alone. Indeed, people do work faster in the presence of others when they have a well-practiced or dominant response, a classic phenomenon termed social facilitation (Zajonc, 1965). Even cockroaches show social facilitation, and people show social facilitation to their favorite TV characters (Gardner & Knowles, 2008). Nevertheless, people assume they will be more productive in groups, regardless of task and practice. For example, people think they will generate more ideas brainstorming in groups than alone (Paulus, Dzindolet, Poletes, & Camacho, 1993). In fact, they do not, according to meta-analysis (Mullen, Johnson, & Salas, 1991). But the illusion results from people forgetting who thought of what idea and (through self-enhancement) taking credit for other people's ideas (Paulus et al., 1993; Stroebe, Diehl, & Abakoumkin, 1992). The principle of getting less done in a genuinely interactive group than in a merely nominal group (performance aggregated across individuals) is termed productivity loss, the most frequent group performance outcome (Kerr & Tindale, 2004), as compared with the few occasions of productivity gain.

PRODUCTIVITY LOSS

In brainstorming groups, productivity loss partly results from people matching the performance of the least productive member (Paulus & Dzindolet, 1993); people who are individually inhibited about expressing their ideas thus tend to dampen the group's overall production of ideas (Camacho & Paulus, 1995). Inhibited people particularly worry about what others will think of them (evaluation apprehension), so social anxiety undermines their motivation to contribute to the group, according to a variety of group studies (Geen, 1991). This fits the pattern of brainstorming groups, where productivity loss is greater under those circumstances especially likely to inhibit shy people: larger groups, experimenter presence, tape-recording oral contributions, and the physical presence of fellow group members (Mullen et al., 1991). Even without shy group members, another explanation for productivity loss in brainstorming groups is sheer access to talking time, whereby more productive members block less productive ones (Paulus, Dugosh, Dzindolet, Coskun, & Putman, 2002; Stroebe & Diehl, 1994).

Beyond brainstorming, research on group productivity indicates the pervasive problem variously called social loafing (loss of motivation and effort in group tasks; Latané, Williams, & Harkins, 1979), free riding (enjoying group benefits without having to make the effort to contribute; Kerr, 1983; Olson, 1965), and the sucker effect (wanting to avoid the risk of being the only one to invest in contributing to benefit the group; Kerr, 1983; Orbell & Dawes, 1981). In keeping with the core social motive of controlling one's outcomes, a key mechanism in all these cases of productivity loss is decreased motivation, when people individually (a) lack incentive to contribute, (b) see contributing as costly, and (c) view their own perspective as dispensable (Shepperd, 1993). In a meta-analysis, social loafing fits this kind of individual incentives perspective: Social loafing is more likely when people are not accountable for their individual inputs to the group, when their inputs are redundant, when expectations of coworker performance are high or moderate, and when the task or group is unpleasant (Karau & Williams, 1993). Also in line with individual incentives, Westerners and men (more individualist, on average) show stronger social loafing effects than Easterners and women.

PRODUCTIVITY GAIN

Groups do work better than individuals when they counteract all the processes that lead to productivity loss. For example, providing people with higher standards, comparable to individual performance, eliminates productivity loss that otherwise results from members conforming to the least productive member (Paulus & Dzindolet, 1993). To the extent that the group makes people accountable for their individual inputs to the group, and to the extent that people compete to contribute unique perspectives, rather than feeling dispensable and redundant, they may contribute more (Paulus et al., 2002).

Group cohesiveness also correlates with group performance, according to meta-analyses (Evans & Dion, 1991; Gully, Devine, & Whitney, 1995; Mullen & Copper, 1994). Performance increases cohesiveness more than vice versa, although both are true. The causal factor may be commitment to the task. When people are committed to the task, they attend more to each other's input and stimulate each other (Dugosh, Paulus, Roland, & Yang, 2000; Paulus et al., 2002). Such groups are both more cohesive and more productive. Similarly, when group members are able to get others to see them as they see themselves (i.e., attend to their self-presentations), members feel more connected to their groups and group grades improve on creative tasks (Swann, Milton, & Polzer, 2000). Friendship groups perform better than acquaintance groups (Jehn & Shah, 1997). And Latinos, who prefer groups with a strong interpersonal orientation, may be especially sensitive to the relationships between cohesion and performance (Sanchez-Burks, Nisbett, & Ybarra, 2000). When people feel good in organizations, they are creative, helpful, and persistent (Brief & Weiss, 2002). As noted, they help coworkers, protect the organization, make constructive suggestions, and spread goodwill (George & Brief, 1992). People work better when they are rewarded in various ways (Shepperd, 1995). All this seems obvious, but it is surprising how much these basic ideas are neglected in many organizations. In all these settings, an appreciation for one's individual role in the group enhances productivity and cohesiveness.

General principles of group productivity include more than the factors already mentioned: individual accountability, individual standards, cohesiveness, feeling appreciated, and commitment to the task. A feeling of group efficacy (“we think we can”) also helps (e.g., Hecht, Allen, Klammer, & Kelly, 2002). As mentioned earlier, diversity in composition can also boost performance. And a variety of other factors matter: average ability, training, tenure, interdependence, rewards, and external communication (Guzzo & Dickson, 1996; Sundstrom et al., 2000). As one example, in a wide variety of college science, math, and technology courses, small-group learning reliably enhances performance (achievement, persistence, and attitudes) (Springer, Stanne, & Donovan, 1999). And cultural variations in motives, goals, feedback, rewards, and satisfaction all moderate group performance (Gelfand, Erez, & Aycan, 2007). Overall, although groups do not outperform individuals as much as people think they do, under the right circumstances groups can be highly productive.

Summary of Group Performance

Group performance draws on motivations to form a shared understanding and to control individual and group outcomes. Group decision making can polarize individual attitudes, as people learn about other people's attitudes, as well as express their own, and conform to the group prototype. Although groupthink per se occurs less than first thought, groups do tend to focus on already shared common knowledge, and cohesiveness can undermine optimal decision making. Groups actually make decisions using a variety of decision rules that predict group judgment better than any single principle. Groups can be less productive than individuals when incentives are low, costs are high, and individuals feel dispensable. One salient cost is evaluation apprehension, which can drag group contributions down to the least productive group member. Social loafing is a salient example of this. On the other hand, groups can be highly productive when people are individually accountable to appropriate standards, when the group is performing well or thinks it can, and when the group is cohesive.

CONFLICT WITHIN GROUPS: CONTROLLING

We have seen that group membership provides people with a sense of belonging (identity and attraction), even in diverse groups and over the stages of joining a group. Group members develop socially shared understandings of group structures such as norms, roles, and subgroups. And we have just seen that they can be productive under the right circumstances. But groups are composed of individuals whose agendas sometimes contradict each other. For example, one evolutionary viewpoint would hold that people individually want to protect self, form coalitions, gain status, choose mates, maintain relationships, and care for offspring (Kenrick et al., 2002); these goals predict certain trade-offs between cooperation and competition, which computer simulation can model (Kenrick et al., 2003). Our core social motives framework suggests that people's individual goals include enhancing self and controlling own outcomes, which could introduce conflict within a group, whereas the others (belonging, understanding, and trusting) would tend to reduce conflict within the group.

Regardless of the source, conflict within groups highlights two topics we will cover here: both the social dilemma of when to strive for individual or group gain, as well as more explicit negotiation and bargaining within the group (e.g., Pruitt, 1998; Levine & Thompson, 1996).

Social Dilemmas

Group members by definition are interdependent, that is, their outcomes depend on each other. In this circumstance, people may discover the advantages to cooperation over repeated interactions (Kelley et al., 1962). But in a social dilemma, self-interest and collective interest conflict: Each person can receive a higher individual outcome for defecting than cooperating, but everyone loses if everyone else defects (Pruitt, 1998). All are better off if all cooperate consistently. One kind of social dilemma is a commons dilemma, in which people all harvest from a shared resource, but if everyone partakes too much, it will be depleted. Originally named for the practice of grazing private flocks on the shared village green, it persists today in roommates sharing from a common larder, colleagues sharing laboratory space, or people using public parks. Another kind is a public goods dilemma, in which people contribute resources to keep something going that they (and others) can use. Everyday examples include public radio and club dues. People can benefit individually from overharvesting or failing to contribute, but if everyone does so, then the commons are destroyed, the public goods are no longer available, and no one benefits. In general, social dilemmas often appear as mixed–motive (self vs. collective) experimental games that use the payoff matrix format described earlier.

Given the structure of payoff matrices in cases where self-interest and collective interest conflict, and given some (mis)interpretations of individualistically oriented evolutionary survival motives, it might seem surprising that people ever cooperate at all. Indeed, conflict among group members for individual outcomes is assumed in mixed-motive settings (Deutsch, 1949, 1973). Rational economic models classically assume individual self-interest. As one test of a completely rational actor, computer simulations can vary strategies, modeling the behavior of individuals who play tit-for-tat (reciprocating the partner's previous move); win-stay and lose-change (repeat a winning strategy, change a losing one); or win-cooperate and lose-defect (self-explanatory). Computer simulations that vary strategies, incentives, and social comparisons often produce patterns that mimic data from laboratory experiments (Messick & Liebrand, 1995, 1997).

People's tendency to follow narrow self-interest may be less frequent than many observers assume. First, groups can handle conflicts over outcome distributions in a variety of ways, not all of which pit people against each other: (a) conflict avoidance (inaction, withdrawal, or preemptive solutions), (b) conflict reduction (unilateralism, voting, negotiation), or (c) conflict exacerbation (highlighting or exaggerating differences) (Levine & Thompson, 1996). Second, when conflict does occur, many people demonstrably do sacrifice their own preferences for others, accommodate instead of retaliate, and forgive betrayals (Rusbult & Van Lange, 2003).

Third, individual differences in social value orientation, covered in Chapter 9 , demonstrate that only some people take chronically individualist orientations (max own), whereas others hold orientations that are competitive (max difference), cooperative (max joint), egalitarian (max equality), or even generous (max other) (Van Lange, 2000). The last three kinds of motivation often operate in concert, displayed by the same individuals. Prosocial orientations result from experiences growing up (e.g., having siblings, especially sisters) and in adult life (e.g., the long-term gain in outcomes from cooperation). Even tit-for-tat, the typical strategy designed to elicit cooperation over repeated encounters ( Chapter 9 ) may yield to a more flexible process that allows for noise (random error that interferes with communication). When generosity occurs along with reciprocity, joint outcomes are higher (Van Lange, Ouwerkerk, & Tazelaar, 2002). That is, a partner who does not just exactly reciprocate good for good and bad for bad but reciprocates and adds a little extra indeed does elicit more cooperation—even in a noisy situation that better approximates real life ( Figure 12.9 ).

Figure 12.9 Cooperation, as a Function of Tit-for-Tat (TFT) or the More Generous Tit-for-Tat Plus a Bonus (TFT+1), as Well as Kind of Noise

Finally, people who have a choice about playing the game may be more cooperatively inclined than people without choice (Orbell & Dawes, 1993). People who intend to cooperate may be more likely to enter willingly into interdependent relationships. Cooperatively oriented participants are sensitive to issues of resource depletion (Kramer, McClintock, & Messick, 1986). Perhaps trust is underrated in comparing individual and collective interest.

At another level, though, people are less generous outside the ingroup. As Chapter 11 shows, intergroup trust is low, and ingroup identity is key here. As one indicator of the importance of identity, dilemmas between groups (i.e., outgroups) represent the biggest challenge (Dawes & Messick, 2000). Moreover, although large groups typically elicit less cooperation than small groups, perhaps because of identity issues, people might cooperate with smaller factions within larger groups. A computer simulation can model this as a function not necessarily of identity, but of sheer payoff matrix to the individual (Parks & Komorita, 1997) or perhaps to the subgroup (de Heus, 2000). Besides identity and sheer payoff issues, cooperation under conflicting motives may be a function of self-efficacy or collective efficacy (Kerr, 1996). That is, feelings of efficacy or personal control (does my contribution really matter?) can enhance cooperation either directly or as a mediator for other antecedents such as group size. Moreover, communication among all concerned can dramatically increase cooperation. Consistent with the importance of collective communication, when people identify with the group, they have a collective sense of self, so their goals transform such that individual goals become group goals (De Cremer, Van Knippenberg, Van Dijk, & Van Leeuwen, 2008). Thus, a variety of factors determine when people compromise individual short-term self-interest for the collective good.

Negotiation

When group members have conflicting individual interests, sometimes they can act unilaterally and sometimes they use another type of social decision rule, such as voting. But sometimes interdependent group members must problem-solve to make joint decisions about allocating scarce resources, that is, engage in negotiation (Bazerman, Curhan, & Moore, 2001; Bazerman, Mannix, & Thompson, 1988; Levine & Thompson, 1996; Thompson, 1990). Negotiation reduces conflict when it is binding: All must abide by the consensus. To reduce conflict, consensus must be possible and better than alternatives outside the group. Just as in close relationships ( Chapter 8 ), comparison level for alternatives can undermine people's motivation for negotiating.

On the bright side, integrative negotiation can uncover ways in which all participants' interests can mesh. That is, additional resources may improve everyone's outcomes, or some people may care more about some issues, while others care about others, and trade-offs can leave everyone satisfied. People often fail to understand the ways that their interests can be compatible, partly because people fail to provide the necessary information (Thompson, 1991). Participants often think they know what other people want, tending to be biased toward perceiving the possible outcomes as a fixed pie (diametrically opposed) and believing that differences are irreconcilable. Indeed, opposing parties often exaggerate the conflict, the extremity of their opponent's position, and their opponent's biases (Keltner & Robinson, 1996).

Revealing actual priorities is more constructive than making assumptions and often results in better outcomes (Thompson, 1991). Accuracy motivation helps negotiators revise their fixed-pie perceptions to reflect the possibilities of integrative solutions (De Dreu, Koole, & Steinel, 2000). In collectivist cultures, negotiators may be more sensitive to the types of relationship they have with the other party, trying to preserve the relationships (Carnevale & Leung, 2001). Concerns with relationships can enhance accuracy in perceiving the position of the other party in collectivist settings, undermining the fixed-pie error (Gelfand & Christakopoulou, 1999).

Accuracy can be undermined or promoted by individual differences. According to a classic theory of cooperation and competition (Deutsch, 1949), egoistic (self-interested, individualistic) negotiators show distrust, hostility, and negative perceptions; as a result, they argue, bluff, threaten, and manipulate to get their way. Prosocial negotiators show trust, openness, and positive perceptions, and consequently they listen, inform, and understand, in order to reach integrative solutions, which they often do. According to meta-analysis, participants with a prosocial orientation do indeed contend less, problem-solve more, and achieve higher joint outcomes (De Dreu, Weingart, & Kwon, 2000), but only under one additional condition, high resistance to yielding (i.e., negotiators' intransigence).

To understand the role of resistance, consider another theory of integrative negotiation. The dual concerns theory (Pruitt, 1998; Pruitt & Rubin, 1986) also holds that negotiators vary in their social motives, but it separates self-concern (low to high) from other-concern (low to high). For example, individualist motives reflect high self-concern and low other-concern, whereas prosocial motives reflect high concern for both; thus, other-concern best differentiates the two.

Egoistic and prosocial motives relate to valued outcomes. In contrast, resistance relates to feasible means (behaviors), that is, willingness to make concessions. Independent of other-concern is negotiators' resistance to yielding, that is, their intransigence and commitment to their own position. Resistance is operationalized as lack of external boundaries: low time pressure, high bottom-line limits, high accountability to third parties, low importance of reaching agreement, and high concern for self (De Dreu, Weingart, & Kwon, 2000). Prosocial motives facilitate constructive motivation only when resistance is high. That is, negotiators have to respect their own constraints and concerns, as well as the other person's concerns, for negotiations to be successful.

CHAPTER SUMMARY

Psychological groups comprise interacting people who consider themselves or whom others consider to belong together. Groups differ in entitativity, the degree to which they make a coherent whole, through Gestalt principles of similarity, proximity, interdependence, and interaction. Intimacy groups, such as family and friends, are experienced as more coherent and cohesive than loose associations (e.g., neighborhoods) and social categories (e.g., women), with task groups falling in between. People value their own groups to the extent they view them as cohesive, coherent entities. In contrast, to the extent people view an outgroup as being an entitative group, they see it as having hostile intent. Because groups are typified by shared goals that operate in the group's own interest, and entitative groups are seen as causal agents, that is, as originating action, outgroups are expected to act on their own interests, which will be hostile to other groups. And indeed groups are more competitive than individuals. Both fear of losing control over one's outcomes and greed to enhance self contribute to group competitiveness.

Social psychologists most often study groups by creating short-term, artificial, experimental groups, often playing experimental games, all which might seem to lack generalizability to the real world. Nonetheless, real-world effects are often stronger than laboratory demonstrations of the same phenomena. Researchers use survey and observational methods when they study real-world groups. People in both artificial and real groups satisfy a number of core social motives, including most obviously the need to belong, but also to validate a shared social understanding and a sense of control. Each of the main sections of the chapter addresses one or two of the core motives, to discuss group membership (belonging), socially shared cognition (understanding and controlling), group performance (understanding and controlling), and intragroup conflict (controlling).

Belonging to a group, becoming a member, revolves around one's identity as a group member, which depends on context, as social identity theory and self-categorization theory predict. Individual concerns matter most when group commitment is low, and threats to self or group simply motivate individual solutions (self-affirmation or individual mobility out of the group). When group commitment is high, group-level concerns matter more, especially under threat of personal exclusion or threat to group cohesion. Depersonalized social attraction to the group fosters cohesion, as members identify with the group prototype, based on their interdependence as group members. Diversity both challenges and benefits the group. Minuses include potential discomfort, friction, and turnover, but pluses include creativity, innovation, and performance. Joining a group is a process, not an outcome, as groups and group members move through mutual investigation, socialization, maintenance, resocialization, and remembrance.

Group performance results from decision making and productivity. Group decision making is less rational than originally expected. Groups tend to polarize in their decisions, compared with the starting positions of members as individuals. Groupthink, conformity to the opinions of a leader with a particular agenda within a highly cohesive and isolated group, occurs only under particular circumstances. More useful are general models that acknowledge the range of antecedents, processes, and consequences in group decision making, as well as social decision schemes that predict the outcome of group decisions. Group productivity occurs when people are individually accountable to appropriate standards, the group is already performing well (or thinks it can), and people are committed to the task. Individual incentives matter.

Group members conflict over controlling resources. In social dilemmas, individual self-interest conflicts with collective interest, creating mixed motives. In a commons dilemma, the conflict occurs over taking resources from a common pool, and in a public goods dilemma, conflict occurs over contributing to the maintenance of a shared benefit. Sometimes groups avoid, reduce, or accommodate conflict, but sometimes they exacerbate it. Individual differences in social value orientation matter a lot here. Generosity creates higher joint payoffs, and it is especially likely when people have a choice, interact with their ingroup, or feel efficacious.

When cooperation fails and conflicting interests endure, group members must negotiate over scarce resources. Negotiation works well when it is binding and when people reach integrative solutions that go beyond a fixed pie. Accurate information about the others' preferences and priorities, as well as a prosocial orientation and high concern for the others, all facilitate negotiations. But successful negotiators must also stand firm with their own concerns and resist too-rapid yielding to the influence of others. Chapter 13 takes us into our final exploration in this book, namely, how people do and do not influence each other.

SUGGESTIONS FOR FURTHER READING

1. Brown, R. (2000). Group processes (2nd ed). Oxford, UK: Blackwell.

2. De Dreu, C. K. W. (2010). Social conflict: The emergence and consequences of struggle and negotiation. In S. T. Fiske, D. T. Gilbert, & G. Lindzey (Eds.), Handbook of social psychology (5th ed., Vol. 2, pp. 983–1023). New York: Wiley.

3. Gruenfeld, D. H., & Tiedens, L. Z. (2010). On the social psychology of organizing. In S. T. Fiske, D. T. Gilbert, & G. Lindzey (Eds.), Handbook of social psychology (5th ed., Vol. 2, pp. 1252–1287). New York: Wiley.

4. Hackman, R., & Katz, N. (2010). Group performance and behavior. In S. T. Fiske, D. T. Gilbert, & G. Lindzey (Eds.), Handbook of social psychology (5th ed., Vol. 2, pp. 1208–1251). New York: Wiley.

5. Hogg, M. A. (2010). Influence and leadership. In S. T. Fiske, D. T. Gilbert, & G. Lindzey (Eds.), Handbook of social psychology (5th ed., Vol. 2, pp. 1166–1207). New York: Wiley.

6. Kerr, N. L. & Tindale, R. S. (2004). Group performance and decision making. Annual Review of Psychology, 55, 623–655.

7. Levine, J. M., & Kerr, N. L. (2007). Inclusion and exclusion: Consequences for group process. In A. W. Kruglanski & E. T. Higgins (Eds.), Social psychology: Handbook of basic principles (2nd ed., pp. 759–784). New York: Guilford.

8. Williams, K. D. (2007). Ostracism. Annual Review of Psychology, 58, 425–452.

Notes