Akkademia di Psicopolis
Interaction in Small Groups
by
Peter J. Burke University of California, Riverside
1 | 2

INTRODUCTION

In the early 70s, the questioned was raised of whatever happened to research on the group in social psychology (Steiner, 1974). A year earlier small group research was declared dead in a chapter on small groups subtitled "the light that failed" (Mullins & Mullins, 1973). In the early 80s, Rosenberg and Turner's coverage of the field of social psychology included a chapter on small groups by Kurt Back (1981), but almost all of the research cited was done before the mid to late 1950s. The more recent coverage of the field of social psychology did not include a chapter on group processes (Cook, Fine, & House, 1995). However, it did include a section under the rubric of social relationships and group processes in which seven chapters were placed. Small group research has not disappeared; rather, it has become ubiquitous, spread among a number of research issues (e.g., networks, exchange, bargaining, justice, group decision making, intergroup relations, jury studies, expectation states, minority influence, leadership, cohesion, therapy and self-analytic processes, and power and status) and disciplines (e.g., sociology, psychology, communications, organizational research) (Davis, 1996). In fact, research in all of these areas is active, though the outlets for such research are varied, and it is likely that no one is completely aware of the full range of activity. On the other hand, research on groups has diminished in sociology as a result of the way in which much research on group processes is conducted. Following the insights of Zelditch (1969), laboratory practice in sociology has shifted from the earlier study of freely interacting persons in a group context to the study of particular processes, perceptions, and reactions that can often be studied on individuals within real or simulated social settings. This approach was often used in psychology from the early studies of Sherif on norms and the autokinetic effect (Sherif, 1936) and the Asch studies of conformity to group pressures (Asch, 1960), as well as the work on "groups" by Thibaut and Kelley (1959). As sociologists began to focus experimentally more on particular processes such as status or exchange, studies of the group qua group declined, but did not disappear. In the present volume, two of the most active areas of group research have been elevated to theoretical orientations (expectation states and social exchange theory), and two other areas have their own chapters (intergroup relationships and interaction in networks). Still, the area of small group interaction contains a wide and rich history and set of empirical works that I attempt to summarize in this chapter. This chapter is broken down into three sections. In the first, I review some of the historical foundations of small group research. I then cover selected research on three issues, again examining some important historical landmarks as well as more current theory and research. These three issue areas are status, power, and leadership; group integration and cohesion; and interaction. EARLY BEGINNINGS Among the earliest writings on the small group is the work of Georg Simmel who, in the late 1800s and early 1900s, was concerned with general principles of groups and group formation (Wolff, 1950). At one end of the size continuum, he focused on how two person groups (dyads) differed from individuals in isolation and how groups of three (triads) differed from dyads (Wolff, 1950). At a more general level, he analyzed how people affiliate into groups of all sizes and how those multiple group affiliations influence the individual (Simmel, 1955). He also analyzed small groups, large groups, issues of divisions in groups, of authority and prestige as well as of superordination and subordination (Wolff, 1950) all matters that still concern researchers in small groups. Another writer in the early 1900s was Charles H. Cooley with interests in the nature of the social order. His work on conceptualizing primary groups reflected a general concern about changes in society, and how what are now called primary relationships (person to person) were Interaction in Small Groups - 3 - giving way to more impersonal role to role relationships, what are now called secondary relationships (Cooley, 1909). Thrasher's (1927) study of gangs in Chicago in the early twenties focused on groups and group processes in a natural habitat. With discussions of status and leadership, the structure of and roles in the gang, social control of members, Thrasher examined many of the same group processes that continue to occupy researchers (cf. Short & Strodtbeck, 1965). The rise of group therapy in the military during WWII to handle the large numbers of battle stressed soldiers, who could not be accommodated in traditional individual therapy, gave rise to the study of what came to be known as T-groups (for therapy groups and (leadership) training groups). The study of therapy groups produced a plethora of research on group processes and the relationship between group processes and therapeutic processes (Bion, 1961; Scheflen, 1974; Whitaker & Lieberman, 1967). Much of this work had psychoanalytic underpinnings, often focusing on member leader/therapist relations growing out of Freud's discussion of group psychology (Freud, 1959). The National Training Laboratory in Bethel, Maine, started by Kurt Lewin's Research Center for Group Dynamics, became the center for research on training groups (cf. Bennis, Benne, & Chin, 1961). This latter work also influenced Bales as he was working out the observational method of Interaction Process Analysis (Bales, 1950), though it had more influence in his later work examining the self-analytic group (Bales, 1970; 1999). In the late 1940s and 1950s, there was a surge of work on small groups in psychology and sociology, such as William F. Whyte's (1955) study of a street corner gang, Moreno's (e.g.,1951) research on sociometry (which began much of the current work on networks), and the work by Roethlisberger and Dixon (1970) on group processes in the bank-wiring room at the General Electric plant. Homans (1950) used several of these studies to generate principles of group interaction. Today, much of the work in sociology can be traced back to the work of Robert F. Bales and his students in the Laboratory for Social Relations at Harvard, especially as this theoretical Interaction in Small Groups - 4 - work was influence by the social systems approach of Talcott Parsons (e.g., Parsons, Bales, & Shils, 1953). Much of the work in psychology was built upon the work of Kurt Lewin and his students at the Research Center for Group Dynamics, first at MIT and then at the University of Michigan. Below, I briefly review earlier work within the framework of each of these "schools" and more current work that directly or indirectly has built on them. In addition to the two locations, each of these schools has had a number of distinctive features. The Harvard school tended to study intact groups freely interacting to solve a common problem. The Michigan school tended to study individuals in contrived social settings or groups that were constrained in some way to prevent free interaction. The Harvard school was interested in the development of social structure within the group. The Michigan school was interested in testing theoretical principles with controlled experiments. The Harvard school was made up primarily of sociologists. The Michigan school was made up primarily of psychologists. The Harvard School Research in the Harvard school was spear-headed in 1950 by the publication of Interaction Process Analysis (Bales, 1950). This book described a procedure for scientifically coding group interaction so that the objective study of group processes and structures could be conducted. This book, together with a series of publications that used the methodology, provided a new framework for systematically studying "whole" groups. The interaction process analysis (IPA) coding system was developed over several years of studying groups. Behavior was broken down into acts, each defined as a simple sentence or its nonverbal equivalent. A person's turn at talk received one or more codes for each act, with a notation of who acted, to whom it was directed, and the sequence order of the acts. Each act was coded into one of 12 categories (see Figure 1). These were arranged into four symmetric groups: positive reactions and negative reactions (both representing socioemotional activity), and problem-solving attempts and questions (both representing instrumental activity). Interaction in Small Groups - 5 - (Figure 1 about here) The coding conventions called for every act to be classified into one of the categories, with ambiguous acts classified into the more extreme (toward categories 1 or 12) of the categories for which it might be relevant. This latter convention was to counter a bias in most coders that was less sensitive to the more emotional and extreme categories of action. With training, coders could achieve a high degree of reliability and agreement (Borgatta & Bales, 1953a). As in many fields, the presence of a new methodology opens up a new line of research, and that was true in this case, with a significant increase in the amount of small group research published. It also opened the field of group research to several other systems for coding interaction that developed over the next several years (e.g., Borgatta & Crowther, 1965; Gottman, Markman, & Notarius, 1977; Mills, 1964). Many of the issues that were to occupy researchers in the following years were first explored using the IPA scoring system and post-discussion questionnaires on groups in the Harvard laboratory. These issues included the development of leadership status orderings (Bales, 1956; 1958; Borgatta & Bales, 1953b; 1956), leadership role differentiation (Bales, 1956; Borgatta, Bales, & Couch, 1954; Slater, 1955), and the phases in group development (Bales, 1953; Heinicke & Bales, 1953). The Michigan School The Center for the Study of Group Dynamics was formed under the guidance of Kurt Lewin at MIT. Later it was moved to Michigan, where its work began to receive attention with the publication of an edited collection of theory and research. Much of this collection grew out of research within the framework of the Michigan school, but it also drew on work that was being done in a number of places (Cartwright & Zander, 1953b). 2 This collection was characterized by a strong theoretical focus and a commitment to careful experimental design to test hypotheses rather than to discover or observe and document group phenomenon. Issues were often couched Interaction in Small Groups - 6 - in the field theoretic approach of Lewin and included group cohesiveness, group pressures and standards, group goals and locomotion, the structural properties of groups, and group leadership. The field theoretic approach, with it's view of groups as interdependences among individuals that are mediated by cognitions and perceptions (life-space), dominated this line of research (Lindenberg, 1997). Now classic studies collected into this volume include, among many others, selections from Festinger, Schachter, and Back's (1950) study of social pressures in the Westgate and Westgate West communities, Schachter's (1953) study of reactions to deviance in groups, Bavelas' (1953) study the effect of different communication structures on problem- solving ability in groups, and White and Lippitt's (1953) study of group members reactions to democratic, laissez-faire, and autocratic leaders. In additional to its experimental approach to testing theory, the Michigan school gave the evolving field of small group research an important approach to concepts such as cohesion and group structure in terms of interdependencies among individuals, and a cognitive focus that dominates much research today. Three Focal Issues The critical issues that have influenced much of the work in the area of small groups within sociological social psychology are status and power, integration and cohesion, and interaction. The most influential issue in sociology has been research concerned with status and power, or as some prefer to label it, social inequality. Work on cohesion and interaction processes diminished, but in recent years has begun to increase. These three areas will be explored in the remaining parts of this chapter. STATUS, POWER AND LEADERSHIP Since much of the work in sociology on status and power in groups can be traced back to the work of Bales, I begin this section with some background. Among the early work by the Bales group at Harvard were two papers that outlined interests in the development of structure and Interaction in Small Groups - 7 - process in problem-solving groups. The first paper examined the phases task oriented groups went through in solving task problems (Bales & Strodtbeck, 1951). A second paper incorporated many of the results of the first paper and focused on the equilibrium problem in small groups (Bales, 1953). The equilibrium problem, from the functional perspective of Parsons and Bales (1953:123), is the problem of establishing cyclic patterns of interaction that move the group forward to accomplish the task, and patterns of interaction that restore the internal socioemotional balance disturbed by the pursuit of the task. Using data obtained through application of the IPA coding system, a number of empirical regularities were documented as evidence of the types of equilibria that a group maintained (Bales, 1953). There was a balancing of proaction (that initiated a new line of activity) and reaction (the first response to another actor). Among the reactions, positive reactions were seen to outnumber negative reactions. There was unequal participation of members. The most active members talked more to the group as a whole, and less active persons talked more to those ranked above them in participation than below them. Thus, persons who participated more also received proportionately more positive reactions. These patterns of participation produced a "fountain effect" with contributions going up the hierarchy and then sprinkling out on the group as a whole (Bales, Strodtbeck, Mills, & Roseborough, 1951). It was also noted that there were phases in the type of activity that occurred over time. Activity in the problem-solving sequence moved from orientation to evaluation to control. Simultaneously, both positive and negative reactions built up over time with a final surge of positive reactions and joking toward the end. It was also observed that there was a differentiation of activity across persons, with some persons being more proactive and others being more reactive. The most active person was less well liked (and more disliked) than the next most active member. This led to ideas of a more active instrumental/adaptive specialist and a less active integrative/expressive specialist, each of whom fulfilled important functions in the group. Interaction in Small Groups - 8 - Role Differentiation. Bales and Slater (1955; Slater, 1955) formalized many of the above ideas in a study outlining this theory of leadership specialization or leadership role differentiation. This was an interesting issue that combined work on the status/power issue with work on the integration/cohesion issue. 3 Bales and Slater studied small, task-oriented, decision-making groups composed of male undergraduate students at Harvard. They gave members of each group a five page written summary of an administrative case problem and told them to consider themselves members of an administrative team and return a report to the central authority. The report was to contain their opinion as to why the persons involved in the case were behaving as they were, and their recommendation as to what the central authority should do about it. Bales and Slater coded the interaction in these groups using the IPA coding system described earlier. In addition, after the discussion, they gave forms to the members to rate each other in terms of liking and on the leadership activities of providing the best ideas and guiding the discussion. Bales and Slater conceptualized the observed actions with their various qualities as emerging from a latent "social interaction system" that was differentiated in a number of ways. Proactions (initiation of new lines of activity) tended to be concentrated in the instrumental categories of give suggestion, opinion, or information, while reactions tended to be concentrated in the expressive categories of showing agreement, disagreement, or tension release (e.g., laughter). Additionally, reactions, while often coming after proactions by another person, also tended to be differentiated in time. A larger proportion appeared toward the end of the meeting during a final period of laughing and joking, suggesting that the "latent state of the total system" varied over time. Another type of differentiation was discovered in the data, which Bales and Slater (1955) described as a "separation [over time] of the rankings on likes from the rankings on other measured characteristics [task contributions]." Accompanying this separation of the best liked person from the person making the largest task contributions was a difference in the activities of Interaction in Small Groups - 9 - these two persons. The best ideas person had an activity profile across the 12 IPA categories that was similar to the proactive profile, while the best liked person had an activity profile that was similar to the reactive profile. Bales and Slater (1955) theorized that the differentiation of task and expressive leadership functions between two different group members was the result of several factors. First, the different types of activity reflected responses to the different demands on the group for solving both the instrumental problems relating the group to its environment and task conditions, and the socioemotional problems of maintaining interpersonal relationships to keep the group intact. Second, these different activities were performed by different persons since the task specialist "tends to arouse a certain amount of hostility because his prestige is rising relative to other members, because he talks a large proportion of the time, and because his suggestions constitute proposed new elements to be added to the common culture, to which all members will be committed if they agree" (1955, p. 297). Liking thus becomes centered on a person who is less active and who can reciprocate the positive affect. After these initial findings were reported, there was a flurry of publications in which the theory was both criticized and elaborated. 4 Some suggested that instrumental and expressive leadership may be more likely to reside in the same person in non-laboratory groups, thus indicating that leadership role differentiation may be conditional (Leik, 1963; Mann, 1961). Verba (1961) suggested that the conditionality depended upon the legitimacy of the task leader. His argument, elaborating on the suggestion of Bales and Slater, was that the negative reactions of the group members toward the task leader were brought about by non-legitimate task leadership. If the task leader were legitimate, such negative reactions would be less likely to occur. In a series of experiments Burke (1967; 1968; 1971) tested this idea and suggested an elaboration of the theory (Burke, 1974a). Using better measures of socioemotional leadership activity and role-differentiation, strong experimental support for this theory about the effects of legitimation was found. Role Interaction in Small Groups - 10 - differentiation did not tend to occur when the task leader was given positional legitimation by being appointed by the experimenter (Burke, 1968), nor did it occur when task activity was legitimated by providing strong motivation for the group members to accomplish the task (Burke, 1967). The incompatibility of the two types of activity was demonstrated, under conditions of low task legitimation, by a strong negative correlation between task performance and expressive performance for the task leader (Burke, 1968). Because role differentiation tends to occur only under conditions of low legitimation, it is not often observed in non-laboratory groups where legitimation tends to be higher. Status Structures The study of the emergence of leadership structures out of freely interacting task-oriented groups described above, was taken up by other researchers who were interested in how such (task) status structures emerged in the first place and the impact that they had on group processes. With a systems understanding of the nature of groups and group interaction, Bales (1953) suggested that the differentiation was the result of both the task and socioemotional domains as well as their relationship. Others were interested in the mechanisms by which some individuals claimed and were granted more status and interaction time. The study of these status organizing processes showed that individuals over time came to have expectations about the future performances of group members (including themselves) based on perceptions of inequalities and differences in the characteristics upon which perceptions of status were based (Berger, Fisek, Norman, & Zelditch, 1977). Once formed, these expectations came to determine subsequent task related interactions among the group members. The task related behaviors that were influenced by these status expectations (both for self and other) were the performance outputs (problem-solving attempts), action opportunities (questions), communicated evaluations (positive and negative reactions), and influence (acceptance or rejection of suggestions given disagreement) (Berger et al., 1977). Note that these Interaction in Small Groups - 11 - categories of task related behaviors are (with the exception of the last) the categories of Bales' IPA coding system, which form a single cluster or correlated activity. The last category, influence or agreement and disagreement, was moved from the socioemotional areas (A and D in Bales' IPA) to the task area and came to play a significant role in the experimental procedures that developed to build and test the newly developing expectation states theory and status characteristics theory. The probability that one person deferred to another (accepted or agreed to the other's suggestion even when one privately disagreed with it) became the experimental (and theoretical) definition of status ordering; the more one deferred to another, the lower was one's status relative to the other. This probability of not deferring, called the probability of staying (with one's own opinion), was termed the P(s). In some ways, this was an unfortunate choice because, without knowing the reasons for the compliance, it confounded power and status (or prestige), which are only now beginning to be experimentally disentangled. 5 Additionally, by focusing exclusively on task status, omitting socioemotional considerations, the full interaction structure studied by Bales was neglected. A second consequence of using P(s) as the outcome to be studied was that the study of a group process became the study of an individual perception/action. This meant that the experiments studying the impact of various factors on status required only individuals to be put into a situation in which their probability of deference, [1-P(s)], could be determined, and this was often, especially in more recent work, to synthetic or computer others with no group or interaction processes. As it developed, this line of work took the group out of group processes, 6 but it also set the precedent in sociology for the way in which laboratory work and theorizing was to be done. Because this work on status characteristics and expectation states is more fully described in another chapter in this volume, I will not discuss it further. However, a number of other theories about groups and group processes have evolved from the expectation states and status characteristics theories and traditions that are worth discussing more fully. Interaction in Small Groups - 12 - Theories of Legitimation As already mentioned, the issue of legitimation came up early in the work on leadership and was instrumental in understanding the conditions under which task and socioemotional leadership role differentiation occurred. In the work on expectation states and status characteristics theory, legitimation was taken for granted. Legitimation was one of the three bases of social power initially described by French and Raven (1960) (the others were reward power and coercive power). They defined legitimate power as the power that stems from internalized values in person A that dictates that person B has a legitimate right to influence person A, and that person A has an obligation to accept this influence. However, more recent research sees legitimacy as a property that can be applied to acts as well as persons and positions (Michener & Burt, 1975; Walker, Thomas, & Zelditch, 1986). Three sources of legitimation are distinguished: endorsement (from peers, or "validity" in the terminology of Dornbusch and Scott (1975)), authorization (from more powerful persons), and propriety (from the focal actor). Walker and his colleagues (Walker et al., 1986) showed that the effect of legitimation in the form of endorsement acts to stabilize a system of positions in a group (Berger, Ridgeway, Fisek, & Norman, 1998; Zelditch, 2001), a fact also reflected in Hollander's (1993) discussion of the importance of follower endorsement in understanding the relational nature of leadership (i.e., that leadership is a relationship not a personal characteristic). Building upon this work on legitimacy, Ridgeway and Berger (1986) turned the question around to understand the way in which informal status structures come to be legitimated in groups. This was done by extending expectation states theory and viewing status (and the status order) as a reward, about which members come to have expectations. These expectations were derived from ideas in the general culture (referential structures) about the way in which rewards, including status, are normally distributed. Three types of referential structures were posited from expectation states theory: categorical beliefs (such as males having higher status than females), ability structures (suggesting that those with the highest ability have higher status), and outcome Interaction in Small Groups - 13 - beliefs (suggesting that those who are successful have higher status). The theory went on to argue that legitimation would occur to the extent that the expectations based on the referential structures were consistent across dimensions, more differentiated, and shared and similarly responded to by others, thus validating them in the eyes of the focal person (cf. Ridgeway, Johnson, & Diekema, 1994). These ideas provided the seed for the development of status construction theory (Ridgeway, 1991; 2001). Here the question was how do status characteristics (such as race and sex) come to have status value in the first place. The logic of the argument is that it occurred through much the same process that status structures come to be legitimated, only now with the focus on the status characteristic. The full argument is presented in the chapter on expectation states theory in this volume. In most of the above research, status and power were not clearly separated. Recent research, however, is beginning more clearly to make that separation and to ask about the relationship between power and status (Lovaglia, 1995b; Thye, 2000; Willer, Lovaglia, & Markovsky, 1997). By bringing together two theoretical paradigms and experimental procedures (power as investigated in network exchange theory and status as investigated by expectation states theory) these two concepts are theoretically and experimentally related (Willer et al., 1997). Lovaglia (1995b) created power differences based on structural dependence and observed that those with more structural power were accorded more status in the sense that participants held expectations of higher ability for persons in the powerful positions. However, these expectations did not translate to increased behavioral influence. As pointed out by Willer and his associates (Willer et al., 1997), emotion played a role in the translation of power to status. If negative emotional responses to power occur, these can prevent the attribution of status to the powerful person. Thye (2000), in his status value theory of power, examined the reverse effect of status on power and found that persons with high status had more power in an exchange setting, and that Interaction in Small Groups - 14 - this power resulted from the increase in attributed value of the resources held by a higher status person.