SOCIAL MOVEMENTS AND COLLECTIVE BEHAVIOR:
SOCIAL PSYCHOLOGICAL DIMENSIONS AND CONSIDERATIONS
David A. Snow
University of Arizona, Tucson
Pamela E. Oliver
University of Wisconsin, Madison
NOTE: THIS DRAFT IS NOT THE FINAL PUBLISHED VERSION. I AM MAKING IT AVAILABLE AS A RESOURCE FOR MY STUDENTS. DO NOT QUOTE OR CITE. ALL QUOTATIONS AND CITATIONS SHOULD BE TO THE FINAL PUBLISHED VERSION.
Written for inclusion in Sociological Perspectives on Social Psychology, edited by Karen Cook, Gary Fine, and James House, published in 1995 by Allyn & Bacon.
SOCIAL MOVEMENTS AND COLLECTIVE BEHAVIOR:
SOCIAL PSYCHOLOGICAL DIMENSIONS AND CONSIDERATIONS
This paper concerns the social psychological aspects of social movements and the collective behaviors that occur in relation to them. Social movements have historically been treated as variants of the concept of collective behavior. Broadly conceived, collective behavior refers to extrainstitutional, group problem-solving behavior that encompasses an array of collective actions, ranging from protest demonstrations, to behavior in disasters, to mass or diffuse phenomena, such as fads and crazes, to social movements and even revolution.1 Within the last twenty years, most scholars have concluded that the terrain of "collective behavior" traditionally conceived is too broad to permit meaningful empirical or theoretical generalizations, and the field has largely split into those who focus on social movements, and those who focus on more transitory or ephemeral events such as disasters, emergency evacuations, crowd actions, and fads and crazes. Because attempting to review both these traditions within the page limits of this article would require too superficial a treatment, we have opted for depth rather than breadth, and have focused this review on the social movements tradition.2 However, some parts of the collective behavior tradition are significant for social movements, because movements include transitory or ephemeral crowd events such as demonstrations, sit-ins, boycotts, and riots. We include some findings form the collective behavior tradition in our review.
As with most concepts in social science, there is no consensus about the concepts "crowd" and "social movements," and different theoretical traditions define the terms differently. However, the following discussions would generally be considered acceptable (although not entirely correct) by most scholars from diverse theoretical orientations.3
Most theoretical definitions of social movement include the following elements: (1) change-oriented goals; (2) some degree of organization; (3) some degree of temporal continuity; and (4) some extrainstitutional collective action, or at least a mixture of extrainstitutional (e.g., protesting in the streets) and institutional (political lobbying) activity. All traditions conceptually distinguish a social movement organization (SMO) from a whole social movement -- which transcends any particular organization -- but there is less agreement about what kind of "thing" the social movement is. Most usage of the term implies that a social movement is a set of people who share goals, even though the most commonly cited definition is "a set of opinions and beliefs [emphasis added] in a population which represents preferences for changing some elements of the social structure or reward distribution of a society" (McCarthy and Zald 1977, pp. 1217-1218). Others argue that social movements are best thought of as sets of behaviors oriented toward goals or issues (Marwell and Oliver 1984; Oliver 1989). We can ignore these conceptual distinctions and keep in mind, instead, that social movements are marked by collective actions that occur with some degree of organization and continuity outside of institutional channels with the purpose of promoting or resisting change in the group, society, or world order of which they are a part (see Benford 1992, p. 1880; Turner and Killian 1987, p. 223; Wilson 1973, p. 8; Zurcher and Snow 1981, p. 447).
For crowds, the following dimensions have been emphasized: (1) joint action in the sense that some number of individuals are "engaged in one or more behaviors (e.g., orientation, locomotion, gesticulation, tactile manipulation, and/or vocalization) that can be judged common or convergent on one or more dimensions (e.g., direction, velocity, tempo, and/or substantive content)" (McPhail and Wohlstein 1983, p. 580-581; McPhail 1991); (2) close physical proximity, such that the participants can monitor each other by being visible to or within earshot of one another (Lofland 1981, p. 416; Snow and Paulsen 1992); (3) unconventional or extrainstitutional occurrences in the sense that they are neither temporally nor spatially routinized but involve, instead, the appropriation and use of spatial areas (street, part, mall) or physical structures (office buildings, lunch counters, theaters) for purposes other than those for which they were designed and intended (Snow, Zurcher, and Peters 1981, p. 38; Snow and Paulsen 1992); (4) normative regulation in the sense that the various behaviors are coordinated rather than random and disconnected (Turner and Killian 1972, 1987); and (5) ephemerality in the sense that they are relatively fleeting or "temporary gatherings" (McPhail 1991, p. 153). These defining characteristics distinguish the crowds associated with social movements from more diffuse or mass behavior, such as fads, deviant epidemics, and mass hysteria, and from more conventional crowds that are routinely on the community calendar and that are sponsored and orchestrated by the state or community, such as sporting events, holiday parades and electoral political rallies (see Aguirre 1984). Thus, when we refer to crowds in this chapter, we have in mind those gatherings that share the above defining characteristics, such as protest marches and rallies, victory celebrations, and riots.
The study of crowds and social movements has deep roots in both political sociology and social psychology, and a major trend in current scholarship is to integrate these traditions by focusing on the linkages between macro and micro processes (see McAdam, McCarthy, and Zald 1988). In this chapter, we will focus on the social psychological dimensions of crowds and social movements, and give only passing attention to the ways these micro processes are linked to macro processes. More specifically, then, our aim is to identify the key social psychological dimensions of crowds and social movements, and to elaborate how research and theorizing pertinent to these dimensions have informed understanding them. We begin with a brief overview of the historical association between social psychology and the study of crowds and social movements, and then turn to a discussion of their key social psychological dimensions and the pertinent literature.
The Historical Linkage
The association between social psychology and the study of crowd and social movement phenomena is particularly long and intimate, dating at least from the 1895 publication of Le Bon's The Crowd, what Gordon W. Allport called "perhaps the most influential book ever written in social psychology" (1954). While the merit of this claim is certainly a matter of dispute, there is no question but that this "little book," as Merton referred to it in his introduction to the 1960 edition, has exercised considerable influence on the study of crowds and even collective behavior more generally, particularly through the 1950s (see McPhail 1991; Moscovici 1985). Other early influential works by psychologists treating collective behavior and social movements as a subfield of social psychology include Freud's Group Psychology and the Analysis of the Ego (1921), Floyd Allport's Social Psychology (1924), Dollard et al's Frustration and Aggression (1939), Miller and Dollard's Social Learning and Imitation (1941), and Adorno, Frenkl-Brunswik, Levinson, and Sanford's The Authoritarian Personality (1950). Through the 1960s, sociologists also viewed collective behavior as an important subfield of social psychology. Work rooted theoretically in symbolic interactionism was particularly important (see, for example, Blumer 1939; Lang and Lang 1961; Turner and Killian 1959, 1972, 1987).4
Social psychological perspectives on the study of crowds and social movements have not always been dominant. As the protest-ridden 1960s faded into the 1970s, the reigning social psychological perspectives on collective behavior were largely jettisoned in favor of the "resource mobilization paradigm" grounded in political sociology and the study of organizations (see, for example, Gamson 1968, 1975; McCarthy and Zald 1973, 1977; Oberschall 1973; Tilly 1978). One early advocate of resource mobilization (who has since recanted) even went so far as suggest that the social psychological collective behavior perspective was "stultifying" and constituted a "straightjacket" on the study of protest-oriented collective action (Gamson 1975).
This does not mean that social psychology had ceased to matter, but rather that resource mobilization theory was grounded in a different psychology, i.e. rational decision making (e.g. McCarthy and Zald 1973, 1977; Oberschall 1973; Tilly 1978; Gamson 1975). Early resource mobilization theorists also stressed the importance of social networks and preexisting organization as preconditions for mobilization. Contrary to earlier theories which used hidden motives or exceptional group processes to explain protest, resource mobilization theorists assumed that blacks participated in the civil rights movement because they wanted to end discrimination, and that students participated in anti-war protests because they wanted to end the war. They argued that people were motivated by social change goals and constrained by resources, costs, network ties, and organizational capacities. The macro forces of politics and organization created the structures that led people to have goals and resources, and the the link between objective conditions and subjective perceptions was seen as unproblematic. Resource mobilization and its rationalist assumptions were largely hegemonic in the 1970s. Several scholars from the previous generation argued that they had been misunderstood, and tried to call attention to other perspectives, but they were largely unheard, except by those who already agreed with them, until after 1980.
The tide turned around 1980. Several articles were widely read and cited which acknowledged the gains of the resource mobilization paradigm, but called for a reconsideration of symbolic interactionism, attribution theory and other perspectives which had been tossed out indiscriminately along with the "authoritarian personality" and the "conflict of generations" (Killian 1980; Turner 1981; Zurcher and Snow 1981; Ferree and Miller 19855). Social psychological processes were once again an object of widely-cited studies. Gamson and his colleagues (1982) watched how small groups decided whether to resist an unjust situation. Klandermans (1984) stressed the subjective nature of the terms in expected utility models, and called for attention to the social construction processes that lead to these subjective perceptions. Snow, Rochford, Worden and Benford (1986) used Goffman's (1974) framing concepts to theorize the ways movements and their constituencies define and interpret their issues and concerns.
By the second half of the 1980s, then, students of social movements were rediscovering the relevance of social psychological perspectives for understanding aspects of the dynamics of social movements, and thereby reestablishing the longstanding association between social psychology and the study of collective behavior. The social psychological perspectives that were now being invoked were clearly not identical with those that had currency in earlier times, but social psychology was once again part of the mainstream.6
SOCIAL PSYCHOLOGICAL DIMENSIONS OF CROWDS AND SOCIAL MOVEMENTS The reasons for the linkage between social psychology and collective behavior phenomena are not difficult to fathom, at least not from our vantage point. Stated boldly, there are aspects of the empirical phenomena of crowds and social movements that are impossible to grasp or understand in the absence of social psychological and microlevel theorizing and research. This is because there are essentially five basic social psychological dimensions or aspects of crowds and social movements. They include: (1) microstructural and social relational dimensions, (2) personality dimensions and related psychological processes, (3) socialization dimensions, (4) cognitive dimensions, and (5) affective dimensions. In the remainder of the chapter, we elaborate these dimensions and the research relevant to them.
MICROSTRUCTURAL AND SOCIAL RELATIONAL DIMENSIONS
The collective decisions and actions constitutive of social movement actions, including crowd events, have long been seen as the product of dynamic interaction. There are at least two different social psychologies that provide contrasting points of departure for thinking about these interactions.
One approach hinges on the perception of participants as solitary disconnected individuals who are particularly receptive to the influence of others. Most of this work has been oriented to emergent crowds in which, these scholars assume, indivdiuals are anonymous and thus freed from their usual social constraints. Variants of this view were advanced by such early writers as Tarde (1890), Le Bon (1895), and Freud (1921), and later proponents of the "deindividuation" thesis (Zimbardo 1969; Diener 1977, 1980). The concept of "circular reaction," initially coined by Park and Burgess (1921) and developed more fully by Blumer (1939), also emphasizes the uncritical acceptance and mechanical reproduction of whatever influences or suggestions are encountered. Coleman's (1990) recent emphasis on the "unilateral transfer of control" among crowd participants is a continuation of this tradition. Mass society theorists of the 1950s (Adorno et al. 1950; Kornhauser 1959) advance a similar argument for social movements, arguing that they draw from isolated individuals who lack memberships in groups and associations which would constrain their behavior. This general perspective has been dubbed "contagion theory" by Turner and Killian (1972), the "transformation hypothesis" by McPhail (1991), and is a major part of what Tilly, Tilly, and Tilly (1975) call "breakdown theory."
Standing in sharp contrast is an alternative perspective that emphasizes the group-based nature of behavior in both emergent crowds and social movements. For emergent crowds, research indicates that participants attend or join in intact social groups, either small units of family and friends or, in the case of large social movement crowd events, in organizational units (see McPhail 1991 for a comprehensive review). For social movements more generally, research indicates that participants in social movement organizations and events have extensive organizational ties and are often recruited through those ties (see Oberschall 1973 for an early reviewof this evidence). It is this latter view which dominates current work and which is the focus of our discussion.
Preexisting Groupings and Affiliations
By preexisting groupings and affiliations, we refer to structures of social relation that exist apart from and prior to the crowd and social movement activities in question. These include cliques, friendships and acquaintanceships, small groups, and secondary associations. Their relationship to is twofold: they can function as conduits for communication, and they can provide the facilitative contexts for the emergence of new ideas and actions.
Social Networks as Information Conduits and Bridges. Research demonstrating the importance social networks for communication is abundant. Whether the research site or object is religious cults and movements (Snow, Zurcher, and Ekland-Olson, 1980; Rochford 1982; Stark and Bainbridge 1980), more politically oriented movements, such as the civil rights movement (McAdam 1986; Morris 1984), the women's movement (Freeman 1973; Rosenthal, Fingrutd, Ethier, Karant, and McDonald 1985), and the Dutch peace movement (Klandermans and Oegema 1987), underground terrorist organizations (Della Porta, 1988), crowd assembling processes ( McPhail and Miller 1973; Shelly, Anderson, Mattley 1992), victory celebrations (Aveni 1977; Snow et al. 1981), or looting and rioting (Quarantelli and Dynes 1968; Singer 1970; Berk and Aldrich 1972), the research findings almost invariably point to the same conclusion: that preexisting social ties or network linkages not only facilitate participation in most crowd and social movement activity, but also channel the diffusion of that activity. Even in the most elementary forms of collective behavior, social networks are found to perform these facilitative and channelling functions. In their now classic study of an incident of so-called hysterical contagion in a southern clothing manufacturing plant, for example, Kerckhoff, Back, and Miller found that the contagion spread through sociometric channels (1965, p. 394). In a similar vein, Knopf's (1975) extensive examination of the relation among rumors and race riots revealed that "these rumors were essentially social phenomenon" and that the participants in the process "neither related nor responded as isolated or independent units" (1975, pp. 65-66).
In most of the research examining the relationship between social networks and crowd and movement dynamics, there has been a narrow focus on the presense or absence and number of preexisting ties. More recently, however, as the study of this relationship has become more sharply focused and technically sophisticated, attention has shifted from simply counting network ties to assessing there structure and multiplexity. Thus, in their research on recruitment to the Freedom Summer campaign at the University of Wisconsin, Fernandez and McAdam (1989) found that an individual's network prominence or centrality within the campus's multiorganizational field figured significantly in predicting participation in the campaign. In a similar vein, Marwell, Oliver and Prahl (1988), in a computer simulation of the contributions of network density, centralization, and organizing costs to the prospects of collective action, found that centralized network structures are more facilitative of mobilization than less centralized network structures and that it is not so much the relative strength of ties that matter as it is their centralization. Regarding this latter finding, they write:
Residents who are all bridged by the same weak tie -- to a parish priest, for example -- are more likely to be mobilized than those linked by the same number of weak ties distributed more widely through cross-cutting associational memberships (1988, p. 531).
Moving even further beyond the earlier tendency to assess network effects by simply counting participants' preexisting social ties, Gould has focused attention on the role of network multiplexity in the mobilization process. In his study of insurgency in the Paris Commune of 1871, he found "that successful mobilization depended not on the sheer number of ties, but on the interplay between social ties created by insurgent organizations and pre-existing social networks rooted in Parisian neighborhoods" (Gould 1991, p. 716).
Such findings clearly underscore the importance of network ties, strength, density, centralization and multiplexity in relation to collective action mobilization processes. Indeed, given the pervasiveness of one or more of these network dimensions across nearly all forms collective behavior, coupled with the parallel findings of research on participation in voluntary associations (see, for example, McPherson, Popielarz, and Drobnic 1992), it is tempting to conclude that little else matters in the determination of patterns of recruitment to crowd and social movement activities. Yet, such a conclusion is not fully warranted. It glosses the role of personality, socialization, cognitive, and affective processes in relation to mobilization, the occasional finding that so-called structural isolates sometimes figure significantly in the development of various forms of collective behavior (see Fernandez and McAdam 1989; and Kerckhoff et al. 1965), and the fact that the relative influence of pre-existing ties tends to vary with differences in the risks and costs associated with different crowd and movement activities (McAdam 1986; Wiltfang and McAdam 1991). Additionally, to weight network factors too heavily runs the risk of overlooking the extent to which the salience of those factors can vary with differences in sociocultural and historical context.
Facilitative Social Contexts. That some social contexts are especially facilitative of collective action has been suggested by the coinage of such concepts as "the youth ghetto" (Lofland 1968), indigenous or "internal organization" (Morris 1981, 1984), "free spaces" (Evans and Boyte 1986) and "micromobilization contexts" (McAdam et al. 1988). Undergirding these concepts is the historical fact that movement activity not only clusters temporally (Tarrow 1989a, 1989b), but in those periods when it does occur, it is often contextually pocketed or is more robust in some localities than others. Haimson's (1964) examination of the nature and sources of social instability in urban Russia several years before the revolution is particularly illustrative. Focusing on the growth of the strike movement, Haimson notes that it was most heavily concentrated in the Petersburg area, particularly among the workers in the metalworking industry. Why? The answer, he suggests, resided in the "peculiar combination of skilled and unskilled, experienced and inexperienced, workers" in the Petersburg metalworking industry, with "the older and more skilled workers contributing in their contacts with the young and unskilled a long-standing exposure to revolutionary, and specifically Bolshevik indoctrination" (Haimson 1974, p. 637). In short, the metalworking industry in Petersburg constituted a more militant and explosive political context than that found in other industries at the time.
Petras and Zeitlin's (1967) examination of the diffusion of radical political consciousness among the peasantry in Chile in the early 1960s yielded a strikingly similar set of observations regarding the mobilizing role of unusually nurturant contexts, in this case mining municipalities located in the countryside. The main determinant of peasant radicalism, they found, was proximity to these highly organized and politically radical mining centers. Similarly, Portes (1971) found that lower class urban Chileans' radicalism was predicted by their tenure in radical neighborhoods.
In a similar vein, Kriesi's (1988a) research on the Dutch peace movement revealed that support for the movement and its campaigns has tended to vary with proximity to and integration into more leftist, locality-based communities or neighborhoods he terms "countercultural networks." "In such localities," he notes, it is difficult, if not well-nigh impossible, to escape contact with the movement because it tends to be "integrated into everyday activities" (1988a, p. 69). Because of this and other such considerations, Kriesi concluded that:
When people are not embedded in larger countercultural networks, they are less likely to live in an environment supportive of NSMs (new social movements) and are therefore less likely to develop corresponding attitudes and sympathies" (1988a, p. 73).
Although using somewhat different language, Broadbent's (1986) study of the mobilization efforts of twelve environmental movements in Japan produced a similar conclusion: that "the local social fabric forms an important mediating variable between social forces coming from the external dominant institution, and the way local people mobilize into movements in response" (p. 231). In particular, Broadbent found that the character of mobilization varied with whether the "local social fabric" was communal or associational, with the former contexts mobilizing more quickly and pervasively because of greater solidarity.
Finally, it is worth noting that the extension of McAdam's research on recruitment to Freedom Summer from the University of Wisconsin to Berkeley yielded findings consistent with the foregoing. At Berkeley, not only were there a greater number of both applicants and participants, but network centrality or prominence was not found to be a significant predictor of either. The reason, Fernandez and McAdam surmised (1988; see also McAdam and Fernandez 1989), was that Berkeley constituted much more of an "activist context," a context that not only affected rates of participation but also affected all variables, including network position, that typically impinge on decisions to participate.
Taken together, the above studies clearly underscore the relevance of facilitative contexts to mobilization. Indeed, each of the above cases provide compelling examples of social contexts that appear to be so nurturant of movement activity that information diffusion and recruitment are overdetermined, thus rendering network ties and structure largely superfluous as determinants of these mobilization processes. This is not to say that network ties were not operative, but they appeared to be so dense and overlapping in each context that nearly everyone was interconnected. And for those who were not, it appeared as though it was difficult to escape contact with movement activity because of its intensity and pervasiveness within the context.
Such contexts thus appear to provide particularly fertile soil for movement mobilization activities. The reason is not only because of extensive network density, but also because of the confluence of residential proximity, demographic and/or socioeconomic homopholy, and a tendency toward ideological congruence, coupled with sufficient shared grievances and the opportunity to act collectively. These contexts, as we have conceptualized them, are not a necessary condition for social movement activity, but their existence clearly increases the probability of successful mobilization. More importantly from the vantage point of this chapter, the existence of such nurturant contexts reminds us again of the extent to which crowd and movement activities, and the social psychological processes associated with them, are embedded in preexisting groupings and webs of association.
Emergent Structures of Relation
While all types of crowd and social movement activities entail some level of joint action, these coordinated interactions are not all rooted in preexisting structures of relation. Some are emergent or peculiar to the particular collective behavior episode or event itself. In other words, they grow out of rather than precede some crowd episode or social movement activity. It is this fact that undergirds Weller and Quarantelli's (1973) contention that the social organizational basis of collective behavior is both normative and social relational, and that collective behavior can therefore be predicated on either enduring or emergent norms and enduring or emerging social relations.
Evidence of the link between collective behavior and emergent social relationships is particularly abundant in research on organizational and community responses to disasters (Dynes 1970; Ross 1978; Zurcher 1968), but that linkage is not confined to disasters. It has also been found in instances of rioting and looting associated with civil disturbances, however short-lived (Kerner 1969; Quarantelli and Dynes 1970), as well as in various kinds of social movements. In his previously mentioned research on mobilization in the Paris Commune of 1871, for instance, Gould (1991) found two bases of social relation: preexisting neighborhood ties and emergent insurgent organizational ties. Moreover, both sets of ties functioned to build and maintain solidarity, thus prompting Gould to argue that "mobilization does not just depend on social ties; it also creates them" (1992, p. 719). Snow also found this to be the case in his research on the Nichiren Shoshu Buddhist movement: commitment and solidarity were not only based on the preexisting ties that facilitated recruitment, but also on a horizontal structure of emergent peer group associations within the movement. Together, these two sets of overlapping and interlocking relationships functioned to generate "a more cohesive and highly integrated movement, and a more highly committed and mobilizable constituency" (Snow 1987, p. 159).
Such findings, coming from research on a variety of crowd phenomena and social movements, suffice to highlight the existence of emergent structures of relation and their relevance to mobilization processes. Although preestablished associations are more fundamental to the assemblage process for crowd phenomena and to the recruitment process for social movements, it seems equally clear that emergent relations are often critical for the accomplishment of specific tasks in crowd contexts and can contribute significantly to the development and maintenance of commitment and solidarity in the case of social movements. Whatever the exact functions and relative influence of these two structures of relation, the foregoing observations indicate that they are complementary, fundamental to understanding processes of mobilization, and that they are an appropriate point of departure for understanding much about the social psychology of crowds and social movements.
Whether the structure of relations among collective actors is based on preexisting or emergent relations, the interacting units are typically groups rather than individuals. The groups may vary in terms of size, membership requirements, boundary permeability, life span, and the like, but it is almost always some set of individuals working together to accomplish a particular task vis-a-vis some other group or target of influence that constitutes the primary unit of action within the collective behavior arena. Thus, insofar as one is interested in the dynamics of crowds and social movements, analytic attention should be focused in part on groups and the interaction among them rather than just on the individuals that comprise them.
To suggest such a focus must seem like an axiomatic assertion from a sociological standpoint; yet the research and writing on crowds and social movements varies considerably in terms of level of analysis. Group-level processes and dynamics have always figured more prominently in the analysis of social movements than crowds, largely because of the realization that much social movement activity is highly organized. But the group focus has moved even more center stage in the study of social movements over the past twenty years with the ascendance of the resource mobilization perspective and its cornerstone concept of "social movement organizations" (SMOs) (Zald and Ash 1966; McCarthy and Zald 1977; Zald and McCarthy 1987) and the use of such cognate terms as "challenging groups," "target" groups, and "affinity groups" (Barkan 1979; Gamson 1990). Correctly noting that many movements grow out of small groups, that such groups are critical to the operation of most social movements, and that they often develop their own small group cultures or "idiocultures," Fine and Stoecker have argued that the study of social movements could benefit even further by examining how the course and character of movements involve "the working out of small group processes" (1985, p. 2).
The same argument can be made with respect to the study of crowds, perhaps even more so. Group processes and dynamics have been matters of concern to some students of crowd behavior, to be sure. The empirical inspiration for Turner and Killian's emergent norm perspective, for example, comes largely from a series of well-known small group experiments (Asch 1952; French 1944; Sherif and Harvey 1952). Still, the bulk of research and theorizing on crowd phenomena has been at the individual level of analysis, as evidenced by the broad range of research that can be subsumed under either the "convergence" and "gaming" or rational decision perspectives.
This individualistic focus notwithstanding, there are a number of empirical investigations of behavior in crowd contexts that suggest the analytic utility of a group level focus. Based on a comparative study of 146 protest demonstrations, MacCannell argued that "the natural sub-parts of demonstrations...are not...the individual participants but...groups, groups of demonstrators, bystanders, press, police, and others" (1973 pp. 1-2, emphasis original). He acknowledged that "some demonstrations dissolve into individualistic behavior," but, emphasized,
no demonstration starts this way.... Under "normal" conditions a demonstration runs its course as a delicate interplay between the subgroups relating to each other as totalities: "the police" did X then the "demonstrators" did Y (1973, p.2).
McPhail's (1991; McPhail and Wohlstein 1983) systematic empirical examination of behaviors in crowd contexts for more than a decade also sheds light on the group nature and embeddedness of much of what transpires in those contexts. His data indicate that people in crowds usually spend most of their time interacting with the people closest to them, and that all coordinated crowd actions require attention to a common focus of attention. Buttressed by his (1989) reading of Mead that a common focus of attention cannot be treated as unimportant or unproblematic, McPhail has shown that coordinating instructions determine the form of collective locomation (McPhail and Wohlstein 1986) and has developed cybernetic control theory models of the ways collective action is coordinated by individuals adjusting their behavior to bring their perceptual signals in line with a reference signal ( e.g. McPhail McPhail, Power and Tucker 1992).
For those behaviors that fall into one of these two categories, it seems clear that the preponderance of them would be coordinated and group-based, whether the group be preestablished or emergent. That is clearly what MacCannell's research suggests, and it is what Snow and his colleagues argue based on a team, field study of a variety of crowd episodes in Austin, Texas in the first half of the 1980s (Snow, et al., 1981; Snow and Anderson 1985; Snow and Paulsen 1992). Drawing on Wright's (1978) distinction between crowd activities (redundant behaviors seemingly common to all crowd episodes, such as assemblage, milling, and divergence) and task activities (joint activities and behaviors that are context specific and necessary for the attainment of a specific goal or the resolution of a specific problem, such as parading, picketing, and looting), they found that all of the individuals in the crowd episodes they examined fell into one of three generic groupings: task performers, spectators or bystanders, and social control agents. The task performers included the individuals performing both the main tasks and subordinate tasks. In the case of a Ku Klux Klan rally, for example, the main task performers were the marching Klansman, with counterdemonstraters and the media constituting subordinate task performers. The spectators varied in terms of their proximity to the episodes and their aggressiveness and relative influence on the activity of the main task performers, as did the behaviors of the social control agents, consisting primarily of police and military personnel. Without going into further detail, the central finding of this research thus far has been that the course and character of each crowd episode was largely a function of the interaction among the various groups of actors -- the task performers, spectators and police. In some cases the nature of that interaction was negotiated prior to the episode; in other cases it was more emergent. But in all cases, the moving dynamic was group interaction.
There is mounting evidence, then, that insofar as one is interested in understanding the dynamics of crowd behavior, the focus of analysis ought to be at the group level. This makes good sense sociologically, but what about social psychologically? We think it makes good sense social psychologically, too -- not only because it is consistent with the previous observations regarding the microstructural and social relational basis of recruitment and participation patterns, but also because a social psychology that fails to anchor itself in social context, whether it be small groups or society writ large, is a social psychology that misapprehends the locus of most social psychological states and processes.