1 Social Psychology Quarterly 1995, Vol. 58, No. 2, Identities and Self-Verification in the Small Group* ANNA RILEY PETER J. BURKE Washington State University This research examines the relationship between the meanings contained in one's identity and the meanings attributed to one's behavior by both oneself and others in small-group interaction. The goal is to provide an empirical test of expectations derived from identity theory and from the structural symbolic interaction perspective concerning the link between persons' identities, their behaviors, their own interpretation of their behaviors, and others' interpretations of their behaviors. Of interest are three issues: whether others attribute the same meanings to one's role performance as does the self, whether the meanings attributed both by the self and by others verify (correspond to) the meanings contained in one's identity, and the consequences when these meanings fail to correspond. The resultsuggest that a shared meaning structure does develop among actors in a small group and allows all membersimilarly to interpret each other's behavior, and that this shared interpretation tends to verify the group members' identities. In addition, it was found that when discrepancies exist between the meanings of a group member's role performance and the meanings of his or her identity, the group member is less satisfied with his or her role performance in the group. The implications of these results for identity theory are discussed. Both motivation and reflexivity are central with the internal self-standards (Burke 1980). components of the identity model as outlined The striving for consistency between one's selfin identity theory (Burke 1991). These two relevant feedback and one's internal selfcomponents become even more significant in standards (also known as self-verification; see applying identity theory to individuals in a Swann 1983) is the motivational component of group, because it is through them that a num- the identity model (Burke 1991). Thus persons ber of important processes take place. In the can observe their own role performance as well group we must account not only for the link as others' reactions to it, and can continuously between a person's identity and his or her be- use both of those perceptions to modify their havior, but also for the maintenance of that role performance so that it supports and is conlink in the presence of other demands on the sistent with their identity standard. person's behavior; these other demands take In this way, identity theory explains the the form of others' behavior and expectations, relationship between identities and perforas well as the situational demands of the group mances. Most of the research that examines in regard to attaining its goals. Reflexivity and this relationship, however, has been from the motivation are keys to this account. perspective of one person (the actor) and the Identity theory views reflexivity in terms of a behavioral choices made by the actor (Burke control system (Powers 1973) which takes ac- and Hoelter 1988; Burke and Reitzes 1981, count not only of feedback about the self from 1991; Burke and Tully 1977; Swann 1987). the social environment, but also of self-views Strictly speaking, the identity model is already incorporated into the identity standard. concerned only with the actor's identity, From a control system perspective, reflexivity perceptions (feedback), and behavior, and is the self's way of taking account of both inthus is fairly psychological in its orientation; ternal self-standards external self-relevant we must expand our model to consider sets of feedback from one's current role performance interacting persons or groups in order to deal to influence that role performance in ways that with sociologically interesting issues. make the new self-relevant feedback consistent This expansion becomes possible by incorporating additional ideas from the symbolic in- * This research was partially supported by grants from teraction framework. In that framework it is the Division of Social Sciences, National Science assumed that people share meanings and com- Foundation (NSF BNS ), and from the National Institutes of Health (MH 46828). We wish to thank Jan municate with significant (shared) symbols. Thus Stets and three anonymous reviewers for their comments the meanings of one's behavior to oneself should on earlier versions of this manuscript. be "the same" as the meanings of that behavior 61
2 62 SOCIAL PSYCHOLOGY QUARTERLY to other interactants the situation. Therefore tradition. They have always been considered it is of interest to examine the process by which critical for understanding the individual in a people maintain role performances that both ver- social context. Although the process has been ify or supportheir own identities and contrib- conceptualized in different ways, identities ute to the overall group processes of achieving are thoughto motivate role performances. In the group's goal. Do others perceive the mean- each of these conceptions, however, meaning ings of one's behavior as does oneself? Do those is a critical component. Foote (1951) argues perceptions by others play a role in the feed- that identitieserve as a source of motivation back process? If self-verification is to occur in not only by calling up particular activity groups, the answer to both of these question (relevanto a particularole identity), but also must be yes. by giving that activity meaning and purpose The line of research most relevan to this as belonging to the self. Symbolic interactionissue is the work on reflected appraisals by ists indicate that identities motivate role Felson and others (e.g., Felson 1980, 1985; performances because they classify (give Felson and Reed 1986; Ichiyama 1993). meaning to) social objects including the self, Much of Felson's work, using survey tech- others, and items of performances (Burke niques, tends to show that there is little 1980; Burke and Reitzes 1981; Stryker 1980). relationship between others' appraisals of Identities motivate role performances that one's identity and one's own identity. sustain and verify the meanings contained in Ichiyama, however, has shown a much closer the identity. Thus identities enable people to correspondence in small-group interaction predict and control the nature of social reality, studied directly. A number of other studies which in turn is necessary for survival (Swann have examined others' assessments of the et al. 1987). In sum, identities motivate role meaning of one's role performance and how performances because those role perforthat corresponds to the meaning of one's mances are meaningful. They are meaningful identity (Alexander and Rudd 1984; Alex- to the actor by providing self-verification, and ander and Wiley 1981; Heise and Thomas to others by providing them with ways to 1989; Smith-Lovin and Heise 1988). Few identify and categorize the actor. In addition, studies, however, have analyzed the relation- the self-verification motive is particularly ship between identities and role performances strong because the failure of self-verification in situations where multiple actors may leads to dissatisfaction, discomfort, and disturb any one person's identity/behavior distress (Burke 1991). Hence the actor is relationship. The work of Alexander cited motivated in part by the desire to avoid the above, for example, is confined primarily to dissatisfaction and distress. laboratory studies of an actor's ability to The idea that people are motivated to convey identifying meanings through behav- sustain their self-views and that they do so by ior aimed toward an audience; the work of thinking and behaving in ways that reinforce Heise and his associates is limited primarily those self-views leads to the second important to paper-and-pencil studies of responses to aspect of the self, its reflexivity (Burke 1980; written vignettes and stimulus sentences to Swann 1987). Reflexivity is often described describe the influence of various (written) as individuals' ability to see themselves as behaviors on perceptions of meanings attrib- objects, which allows them to "self-conuted to actors' identities. To further our sciously" take themselves into account in understanding of the relationship between formulating action alternatives. Reflexivity identity and role performance, this research allows the self-concepto first develop and considers the roles of reflexivity and motiva- then sustain itself through self-verification. tion in extending identity theory to examine The feedback process of the identity model how both one's own and others' assessments (described below) is based on the reflexive of one's role performance correspond to one's nature of the self and explains the relationship identity, given multiple actors in the situation. between identities and role performances. We describe the identity model below and then THEORETICAL BACKGROUND return to the concepts of motivation and reflexivity to understand identity in a group. Identity Feedback Model In identity theory, an identity is viewed as a The concepts of motivation and reflexivity control system (Powers 1973) composed of are not new to the symbolic interaction four parts (see Figure 1). Input from the
3 IDENTITY IN THE SMALL GROUP 63 Identity Standard Co parator Input Person\ Output..., I... * Environment \ Symbol * Flows Figure 1. Model of the Identity Process (Adapted from Burke 1991) environment, consisting of self-relevant input perceptions are the controlled quantity, meanings, is broughto the comparator along and the "goal" of the system is to verify and with self-defining meanings from the identity support the self-defining meanings of the standard. Perception of the self-relevant identity standard. Identities do this by inducaspects of the situation is the reflexive aspect ing behavior that changes social situationso of the self. The comparatorelates the two that the input perceptions conform to the sets of meanings. Insofar as they differ, error self-defining meanings.' This is the motivais present and is felt as a form of discomfort tional aspect of the self. ranging from relatively mild dissatisfaction to The negative feedback cycle of the normal severe distress. Output, or meaningful behav- operation of identities is a continuous proior, varies according to the magnitude of the cess. The feedback that takes place during error. This behavior in turn modifies the social interaction is part of a continuous loop situation and creates new perceptions of input. The system operates on the principle of ' Identity theory recognizes that persistent failure to negative feedback to minimize the error achieve congruity by modifying the situation may first result in attempting to leave the situation (Swann 1990). between the input perceptions (self-relevant Failure in that attempt may ultimately lead to change in meanings) and the self-defining meanings the identity standard that defines who one is (Burke from the identity standard. In this sense, the 1991).
4 64 SOCIAL PSYCHOLOGY QUARTERLY from the input of the actor's perception of as to themselves. Because each can know the identity-relevant meanings to the output of the other as he or she knows his or her own actor's meaningful role performance in the identity, coordinated interaction is possible; situation, and back again to the input. this simultaneously accomplishes the group's Depending on several factors such as the goals and sustains the individuals' identities. salience of the identity, the degree of Achieving this coordination may well require commitment to the identity, or the importance negotiation and compromise (McCall and of the identity, interruption or failure of this Simmons 1978). Hence we cannot expect a process of maintaining congruence between perfect correspondence. Sustaining and verithe perceptions of self-relevant meanings in fying one's identity in a group requires not the situation and the self-meanings the only behavior on the part of the actor that identity standard will result in outcomes confirms his or her identity, but also that the ranging from dissatisfaction with the role to behavior be interpreted and accepted by psychological distress (Burke 1991). others, and that the behavior of those others confirm the actor's identity. Thus the identity Group Processes, Identities, and Satisfaction system operates on the basis of one person's activity and perceptions, whereas social When we move from studying the relation- interaction requires shared meanings (Mead ship between identities and behavior in 1934). isolation to studying that relationship in a As we have seen, however, compromise in group, additional considerations arise. In a the self-verification process leads to distress and group it becomes more problematic to control dissatisfaction. In considering this issue with perceptions of self-relevant meanings by respec to the two sources of feedback in the altering performances until one achieves some self-verification process, McCall and Simmons degree of correspondence between those (1978:88-89) suggesthat in most cases, indimeanings and the meanings in the identity viduals attach less importance to others' expecstandard. One reason for this difficulty is that tations about theirole performance than to their in the group, some of the self-relevant own expectations. The authors point out that meanings monitored by the actor emanate others' evaluation often is built into one's own from the actions of others, each of whom has his or her own goals (in part to sustain their self-expectations. Thus, in regard to the idenown identities). In a group, every person must tity feedback process, when a gap exists beappraise everyonelse. This situation places tween the meanings of one's identity and the important sources of self-relevant meanings meanings contained feedback about one's perbeyond the actor's direct control and creates formance, actors may respond by becoming dismore ways in which actors may fail to keep tressed and thus dissatisfied with their perforself-relevant meanings in alignment with their mances or with the situation itself. identity standard. Thus it produces dissatis- This idea is explicated further by Stryker faction, discomfort, or distress. Another and Statham (1984: ), who suggest reason for the difficulty of keeping self- that satisfaction depends on the extent to relevant feedback consistent with one's iden- which role performance and self-image are tity in the group is that each person, to some integrated into the interaction process. In extent, must subordinate the self to the goals other words, the more successfully an actor of the group. The coordinated activity of the can cause others to assess his or her role group members needed to achieve the group's performance as representative of the actor's goals places additional constraints on the own self-image, the more likely the actor will actors, who are trying to maintain correspon- be satisfied. Therefore our research hypothedence between their inputs and their identity sizes that correspondence between the actor's standards. identity and others' assessmentshould give Within the symbolic interactionist frame- the actor a feeling of satisfaction with role work, it is understood that people share and performance, as well as with the situation communicate with significant symbols; thus itself. Conversely, dissatisfaction should arise group members can share an understanding of from the lack of correspondence. the meanings of behavior. These shared Investigating the relationship between idenmeanings help to define or identify all of the tities and role performances in the group, members of the group to one another as well then, leads to the following hypotheses:
5 IDENTITY IN THE SMALL GROUP Because social interaction requires shared examine the relationship between specific meanings, a positive relationship should exist identities and the performances of activities between an actor's perceptions of his or her role associated with those identities, and 3) a performance and others' perceptions of that role method for examining others' assessments of performance. each actor's performance in an interactive 2. Because actors attempt to keep perceptions of the meaning of their role performances consissituation. On the negative side, the task tent with their identities, a positive relationship leadership role in small laboratory groups is should exist between the meanings of their probably not very important; this may reduce behavior and their identities. the effects we wish to test. 3. Because coordinated interaction the group is based on the use of significant symbols, a positive relationship should exist between the Sample meanings of an actor's identity and the meanings of his or her role performance as perceived The sample analyzed for this research by others in the group. consists of 48 four-person laboratory groups, 4. Because actors' failure to keep their percep- each composed of two males and two tions of the meanings of thei role performances females. To form the groups we randomly consistent with their identities leads to discom- sampled undergraduates from the whole fort and distress, a negative relationship should student population at a large midwestern exist between performance-identity discrepanuniversity and invited them to participate in a cies and satisfaction with the performance. study of communication in small groups. The students attended a general meeting (of 50 to 60 students at a time), at which we explained PROCEDURE the project in general terms as a study of Context communication in groups and the factors that influence communication. The students were This research uses a small group setting to told tha they would be paid $10 for filling out test these hypotheses. Although the task a background questionnaire at the leadership role evolves in general response to the meeting and for participating in a discussion solution of certain group problems, Bales and group at some point during the next two his coworkers have shown that the person weeks. who plays that emergent role, while not Next the students filled out a schedule of elected in the context studied, nevertheless times when they would be tends to persist in available; then they that position over a series of completed the background questionnaire, sessions (Slater 1955). Thus role perforwhich took about 20 minutes. Meanwhile the mances seem to express particular characterinvestigator constructed groups randomly istics of persons who occupy a given position from the persons who were available at each (a situation leading to their persistence in the specific time, with the added constraint that role), as well as the position itself. Those each group contain two males and two performances represent how persons come to females. After the questionnaire was comterms with expectations (both their own and pleted, group assignment times were given to others). These individual dispositions to such each person along with a reminder slip. All performances of leadership roles have not subjects were called on the been studied fully. Yet day preceding a leadership role their scheduled meeting to remind them of identity, though not investigated until now, that meeting. would be a strong candidate for such an The group discussion sessions were held individual characteristic. Even so, in hypothover the two weeks esizing leadership role identity as an following the general impormeeting. Each group of four persons tant particideterminant of leadership role behavior, pated in four different discussions based on we are not suggesting that it is the only group polarization or choice dilemma protodeterminant. cols2 (two that Studying the task usually showed a shifto risk leadership role (identity and performance) in the small, task-oriented 2 discussion group that Bales popularized The choice dilemmas represent fictitious life circumstances in which a person must choose between a risky provides an excellent context for testing our alternative (e.g., an attractive job in a high-risk company hypotheses. We are provided with 1) a way of that may fail, or a delicate but risky operation to relieve a studying face-to-face interaction, 2) a way to condition preventing pregnancy) with potentially high
6 66 SOCIAL PSYCHOLOGY QUARTERLY and two that usually showed a shift to task leadership role identity scores ranging conservatism). We used the choice dilemma from 5 to 25. problems to provide the groups with a task in We created the variable perceived task which they had to reach a consensus. The leadership role performance in a similar four discussions were held during the one fashion, based on items on the postsession in which the group met. Each session discussion questionnaire which were designed lasted about an hour and a half; each to measure task leadershiperformance. After discussion lasted 10 to 20 minutes. each discussion (four times in all), the Each of the discussions followed the same participants were asked to rank each other and format. Before the discussion, the individual themselves on four task leadership performembers read the choice dilemma,3 and mance items (shown in Table 1). These have wrote down their personal recommendations. been used in prio research (Burke 1971). The Then the members were instructed to discuss items formed a single factor with an omega the problem and to reach a consensus for reliability of.95. The rankings were reversed making a group recommendation. After each so that a high number corresponded to a high discussion was completed, the subjects each ranking, and were summed across items. filled out a questionnaire, in which they We derived two different measures from evaluated the discussion and rated each other the performance items. First was the set of on a series of items measuring the degree to self-rankings: each person received the averwhich they performed various activities dur- age of their own rankings of themselves ing the discussion. across the four questions. The second set consisted of the rankings by others (not including the self). Each person was assigned Measures the average of the 12 rankings applied to him Task leadership role identity is measured or her by the others in the group (three other from participants' responses to five self- members multiplied by four questions). In descriptive statements about task-oriented case of a tied ranking, we used the mean rank activities; these were included in the back- of the tied participants. ground questionnaire filled out at the general We employed two measures of satisfaction! meeting. The content of the items, shown in dissatisfaction. Each was based on an 11- Table 1, is consistent with descriptions of the category Likert-type post discussion questioncharacteristics of task-oriented individuals naire item that dealt directly with satisfaction. described in the literature (Bales 1950; Burke The first of these was a general satisfaction 1967, 1968, 1971; Slater 1955). Response item pertaining to the discussion as a whole categories consisted of five-point Likert (To what extent do you feel satisfied with this scales ranging from "strongly agree" to last discussion?); the second was directed "strongly disagree" on some statements, and more specifically to the respondent's satisfacfrom "usually" to "never" on others. The tion with his or her role in the discussion (To responses were scored from 1 to 5: low task what extent were you satisfied with the role orientation was scored 1. As shown in Table you played in this last discussion?). The 11 1, these items factored into a single value categories were anchored at the ends (very with an omega reliability (Heise and Bohrn- little and very much) and in the middle stedt 1970) of.79. We summed the responses (moderate). The average correlation between on these ratings for each person, and obtained the two items over the four discussions was.68. benefits, or a conservative alternative (e.g., a merely We used two measures of discrepancy "okay" job with a very stable company, or no operation, between identity and role performance. Each but no threato life). Subjects must indicate the highest was based on the magnitude of the absolute level of risk they would tolerate while still recommending difference between one of the two perceived the risky alternative (odds of failure are 1, 3, 5, 7, or 9 out of 10). The "shifto risk" (or to conservatism) occurs task leadership role performances (as meawhen the average of the individual preferences before sured above) and the role performance group discussion is less (or more) risky then the group's expected, given the respondent's task leaderdecision. An example protocol is found in Brown ship role identity. The expected role perfor- (1965:657). 3 The four choice dilemma problems were presented in mance was measured as the predicted role a randomized, balanced order to remove possible effects performance based on OLS regression of of order of presentation. perceived role performance on task leader
7 IDENTITY IN THE SMALL GROUP 67 Table 1. Items, Factor Loadings, and Reliabilities for Task Leadership Identity and Task Leadership Performance Task Leadership Identity Loadings (1) When I work on a committee I like to take charge of things..71 (2) I am able to keep at a job longer than most people..55 (3) I try to influence strongly other people's actions..62 (4) I am a hard worker..70 (5) I try to be a dorninant person when I am with people..72 Reliability (fq).79 Task Leadership Performance Loadings (1) Providing fuel for the discussion by introducing ideas and opinions for the rest of the group of discussions.96 (2) Guiding the discussion and kept it moving effectively.90 (3) Attempting to influence the group's opinion.83 (4) Standing out as a leader of the discussion.96 Reliability (fq).95 a Iterated principal-factor analysis identity.4 Other discrepancy the magnitude perceptions of his or her task leadership role of the absolute difference between the role performance is correlated highly with others' performance expected on the basis of identity perceptions of the respondent's performance and the actual role performance as perceived (.6 to.7), indicating (in accordance with our by others in the group. Self-discrepancy the first hypothesis) a high degree of shared absolute difference between the role perfor- understanding of the meanings of the role mance expected on the basis of identity and behavior perceived. Each actor perceives his the actual role performance as perceived by or her own role performance very si'milarly to the actor.5 the way it is perceived by others in the group. Agreement is not perfect, however, and there RESULTS is room for misunderstandings. To test our second hypothesis about the Table 2 presents basic information about link between an actor's identity and the each of the measures used in the study, actor's own assessment of his or her role including means, standard deviations, and performance, we regressed measures of role correlations for each of the four discussions. performance assessment on the identity mea- As the table shows, a respondent's own sures in each of the discussions, using seemingly unrelated regressions (Hanushek 4 From the regression formula y = a + bx, where x is and Jackson 1977). Table 3 presents the the task leadership identity, y is the predicted task standardized regression coefficients for the leadership performance, and a and b respectively are OLS-based estimates of the intercept and the slope, one relationship between perceived task leadercan see that using the predicted scores amounts to a ship performance and task leadership identity rescaling of the identity measure into units of the measure in each discussion. The effect of the actor's of perceived role performance. Thus, when a difference task leadership identity on his or her own is calculated (the discrepancy), we are dealing with the same units of measure. As can be seen from this assessment of task leadership role perforprocedure, discrepancy equivalento the absolute value mance strongly supports our second hypotheof the residual (error) derived by regressing perceived sis. The standardized regression coefficients role performance on leadership identity. show a significant correspondence between 5 The use of a composite (difference) score here is dictated by the theoretical construct that is the actor's task leadership identity, as meabeing measured, namely a discrepancy. Although it is true that sured before the discussions, and his or her a difference score is less reliable than either of the two own assessed task leadershi performance in parts that make it up (when the two parts are correlated each of the discussions. These results confirm positively), this means only that the measure of our the expectation that group members will theoretical construct may not have as high a reliability as we might like. It does not mean that we should avoid the maintain consistency between their role peruse of the measure, or that the measure has no reliability. formances and their identities (as measured This lower reliability can work against us because it before the discussions), even in the presence implies greater difficulty in testing hypotheses: the power of others who are trying to do the same thing. of the tests would be less. On the other hand, if results obtained with the measure are significant, they are Although it is clear that this process is going significant in spite of the somewhat lower reliability. on, the coefficients are not so high as to
8 68 SOCIAL PSYCHOLOGY QUARTERLY Table 2. Means, Standard Deviations, and Correlations among Measures Discussion 1 Correlations Mean sd (1) (2) (3) (4) (50 (6) (1) Other ratings (2) Self-ratings (3) Other discrepancy (4) Self-discrepancy (5) Discussion satisfaction (6) Role satisfaction (7) Task leadership identity Discussion 2 Correlations Mean sd (1) (2) (3) (4) (50 (6) (1) Other ratings (2) Self-ratings (3) Other discrepancy (4) Self-discrepancy (5) Discussion satisfaction (6) Role satisfaction (7) Task leadership identity Discussion 3 Correlations Mean sd (1) (2) (3) (4) (5) (6) (1) Other ratings (2) Self-ratings (3) Other discrepancy (4) Self-discrepancy (5) Discussion satisfaction (6) Role satisfaction (7) Task leadership identity Discussion 4 Correlations Mean sd (1) (2) (3) (4) (5) (6) (1) Other ratings (2) Self-ratings (3) Other discrepancy (4) Self-discrepancy (5) Discussion satisfaction (6) Role satisfaction (7) Task leadership identity preclude other determinants of task leadership role performance as perceived by others in the behavior. As a result, we find a variable group. Thus group members not only share amount of discrepancy between the actual meanings of the role performances; in addition, (perceived) role performance and the role because an actor's performances are tied to that performancexpected on the basis of the actor's identity, others can infer correctly (with participant's identity. Table 3 also presents results relevanto a test Table 3. Standardized Regression Coefficients from Seemof Hypothesis 3. These are the results of an ingly Unrelated Regressions of Perceptions of Task Leadership Role Performance on Task analysisimilar to that used for testing Hypoth- Leadership Identity esis 2. In this case, however, rather than using an actor's own perceptions of his or her task Task leadership behavior, we use other group mem- Performance Ratings bers' perceptions of the actor's task leadership Discussion By Self By Others performance. In this case we see a very similar ** 0.240** pattern of results, thus confirming Hypothesis ** 0.206** 3. The resultshow the expected consistency ** 0.243** between the meanings of the actor's task lead * 0.136* ership identity and the meanings of the * actor's p c.05; ** p c.01.