Annu. Rev. Sociol. 1998. 24:265���90 Copyright ' 1998 by Annual Reviews. All rights reserved DIFFUSION IN ORGANIZATIONS AND SOCIAL MOVEMENTS: From Hybrid Corn to Poison Pills David Strang Department of Sociology, Cornell University, Ithaca, New York 14853 e-mail: ds20@cornell.edu Sarah A. Soule Department of Sociology, University of Arizona, Tucson, Arizona 85719 e-mail: soule@U.arizona.edu KEY WORDS: contagion, network analysis, discourse, protest, interorganizational relations ABSTRACT There has been rapid growth in the study of diffusion across organizations and social movements in recent years, fueled by interest in institutional argu- ments and in network and dynamic analysis. This research develops a socio- logically grounded account of change emphasizing the channels along which practices flow. Our review focuses on characteristic lines of argument, em- phasizing the structural and cultural logic of diffusion processes. We argue for closer theoretical attention to why practices diffuse at different rates and via different pathways in different settings. Three strategies for further de- velopment are proposed: broader comparative research designs, closer in- spection of the content of social relations between collective actors, and more attention to diffusion industries run by the media and communities of experts. What we really need is some new heroes in Engineering. I took that word from Deal���s culture book, and I���m trying to identify the Engineering heroes. Divisional Manager (Kunda 1992, p. 100) They are making more out of this culture stuff than it���s worth...I never read that stuff, maybe see it in passing. It���s the same nauseating stuff they print in Business Week. Group Manager (Kunda 1992, p. 180) 0360-0572/98/0815-0265$08.00 265

INTRODUCTION As the above quotations suggest, skillful players in business and other arenas display a keen sense of fashions and movements within their spheres of action. Much as academics are aware of intellectual currents and exemplars in their fields, we may be confident that executives know what new developments are hot and which are not, and that political activists are attuned to successes and disappointments elsewhere. And as the quotes emphasize, individuals counter as well as endorse and employ the cultural materials provided by a larger sys- tem of discourse. Diffusion studies work with this awareness and its consequences by exam- ining how practices spread. They provide an opportunity to locate and docu- ment social structure, where we consider how patterns of apparent influence reflect durable social relations. And they provide an opportunity to observe the cultural construction of meaning, where we learn how practices are locally and globally interpreted, and ask why some practices flow while others languish. This review treats contemporary uses of diffusion arguments within the fields of organizations and social movements. Diffusion imagery, models, and explanations are on the rise in both fields and with clearly productive effect. We seek to map the logic of these developments, emphasizing characteristic lines of argument, methods, and research designs. At the same time, we strike a cautionary note, arguing that theoretical advance requires closer attention to both structural and cultural bases of diffusion. CONCEPTUAL OVERVIEW Diffusion refers to the spread of something within a social system. The key term here is ���spread,��� and it should be taken viscerally (as far as one���s con- structionism permits) to denote flow or movement from a source to an adopter, paradigmatically via communication and influence. We use the term ���prac- tice��� to denote the diffusing item, which might be a behavior, strategy, belief, technology, or structure. Diffusion is the most general and abstract term we have for this sort of process, embracing contagion, mimicry, social learning, organized dissemination, and other family members. The term ���diffusion��� is sometimes used in an alternative sense to denote in- creasing incidence: Something diffuses when more and more people do it. But treatment of diffusion as an outcome makes it uninteresting, since practices rise and fall in frequency for every possible reason. We thus focus on diffusion as a kind of causal process and seek to map some major lines of argument and important findings. Diffusion arguments cannot be segregated easily from other causal dynam- ics. They verge on the one hand toward models of individual choice, since dif- 266 STRANG & SOULE

fusion models often treat the adopter as a reflective decision-maker. They verge on the other hand toward a broader class of contextual and environ- mental processes, where conditions outside the actor shape behavior. While it is easy to see when one has strayed much too far (analyses of the diffusion of puberty or the diffusion effects of gender composition on job satisfaction), useful hard and fast rules are not readily apparent. Rather than patrol the boundaries, we focus attention on lines of research with affinities to the core notions underlying diffusion. These include models that attend explicitly to flows of material along social relations, efforts of ex- ternal change agents to promote adoption, and interpretive work aligning sources and adopters. The emphasis is on processes treated as involving mean- ingful behavior on the part of both source and adopter.1 Classical Diffusion Studies All lines of argument have empirical fields of application to which they are particularly suited. The home territory of diffusion is the innovation. Innova- tions are novel (at least to the adopting community), making communication a necessary condition for adoption. Innovations are also culturally understood as progressive, strengthening the hand of change agents. And since innovations are risky and uncertain, adopters carefully weigh the experience of others bef- ore acting. The elective affinity between diffusion and innovation is so strong that we sometimes think of diffusion as the only causal process underlying the adoption pattern of innovations. Diffusion studies thus generally investigate the introduction and adoption of an innovation. Classic studies include Ryan & Gross���s (1943) analysis of the diffusion of hybrid corn, Hagerstrand���s (1967) investigation of the diffu- sion of innovations such as the telephone and tests for tuberculosis involving the destruction of cattle in rural Sweden, and Coleman et al���s (1966) analysis of the diffusion of a prescription drug in four Midwestern cities. These studies focused directly on communication processes and channels, tracing the role of the mass media, professional change agents, and interper- sonal interaction within the adopting community. Adoption patterns and self- reports pointed to the impact of external sources in introducing the innovation to cosmopolitans, and the cascading of adoption via relational networks within communities [most famously in Katz & Lazarsfeld���s (1944) two-step flow of influence]. Relative innovativeness was explained largely by modern values and institutional markers of this orientation such as educational background, probably because the acceptance of modern, scientific practices was at issue. Rogers (1995) authoritatively reviews this literature. DIFFUSION 267 1 1A quite different theoretical orientation would be ���practice-centric,��� attending to the flow of resourceful practices across a landscape of carriers.

Contemporary ���Macro��� Diffusion Research Diffusion arguments go in and out of style in sociology as in other disciplines. There is the greatest continuity in interpersonal studies of contagion and influ- ence, but even here their fortunes are tied to relevance to empirical problems. For example, efforts to model the spread of HIV/AIDS has generated much important diffusion research (see the 1995 special issue of Social Networks). Interest in diffusion processes is also a function of broader intellectual move- ments, such as the role of social science in supporting the spread of modernizing innovation. In this review we treat not the rich contemporary literature on interpersonal influence but instead the recent development of more ���macro��� diffusion analy- sis in two fields: social movements and organizations. In the study of social movements, views of contagion as the irrational, spontaneous transmission of antisocial behavior (LeBon 1897, Tarde 1903, Kornhauser 1959) have given way to nuanced studies of diffusion as reflecting ���normal learning and influ- ence processes as mediated by the network structures of everyday life��� (Mc- Adam 1995, p. 231). Diffusion processes play a central role in contemporary explanations of the incidence of collective action and the spread of protest symbols and tactics.2 Diffusion arguments also flourish when there is theoretical attention to the larger environment, to the way cultural models condition behavior, and to his- torical context and change rather than comparative statics. The new institu- tionalism (Powell & DiMaggio 1991) has precisely these emphases, and much diffusion research emerges in organizational studies where this school is most influential. Institutional lines of argument also appear in the social movement literature, as does network imagery in organizational research, so that diffu- sion studies in the two fields are fairly strongly connected. Diffusion research in these fields differs in obvious ways both from the classics of the genre and from current work on interpersonal diffusion. Con- temporary work on organizations and social movements typically examines the spread of behavioral strategies and structures rather than technical innova- tions, emphasizes adoptions by social collectivities more than individuals within those collectivities, works with a much larger historical and spatial can- vas, and incorporates diffusion as one sort of explanation rather than as the overarching framework. As one example, Fligstein (1985) evaluates five theo- ries of the rise of the multidivisional form across the nation���s largest firms over the twentieth century, one of which involves imitation. 268 STRANG & SOULE 2 2The literature on recruitment to activism also emphasizes the effects of network ties. See Curtis & Zurcher (1973), Snow et al (1980), McAdam (1982, 1988), Morris (1984), McAdam & Paulsen (1993), and McCarthy (1996).

Given this context, contemporary diffusion research on social movements and organizations can learn from the classics but should not blindly copy them. INITIAL ELEMENTS OF A DIFFUSION ARGUMENT We briefly flag two important concerns that play a role in all kinds of diffusion arguments but that for present purposes are treated contextually rather than within our main story line. What Is Observed? While most diffusion research emphasizes that adopters are influenced by im- mediate or second-hand observation of the diffusing practice, there is often much ambiguity about what is actually observed. Sometimes we treat the po- tential adopter as exposed to the practice itself. This involves discovering that something is possible, witnessing it in action, or hearing secondhand about its objectives, rationale, and operation. For example, executives may come into contact with poison pills when they sit on the boards of other firms that have instituted them (Davis 1991), managers may learn which markets leading firms enter (Haveman 1993), and activists in Switzerland may hear about pro- tests in the Netherlands (Kriesi et al 1995, Chapter 8). A potential adopter may also observe the consequences of a practice. To continue the above examples, one might measure contact with companies that had successfully warded off takeovers by wielding the pill, or calculate rates of return for firms that enter various markets, or contrast situations in which pro- tester demands were met to those in which they were not. The contrast between observing practices and observing their outcomes is tied only loosely to a contrast between diffusion as mimicry and diffusion as social learning. One can readily motivate diffusion in rational choice-theoretic terms even when no information about consequences is provided (Banerjee 1992). And consequences may be implicit in descriptions of the practice or un- interpretable without close local knowledge or a good theory. Research that directly measures the consequences of adoption elsewhere suggests that both are salient. Conell & Cohn (1995) find that French coal min- ing strikes were stimulated by other strikes in the same department but most strongly by victorious ones. And Holden (1986) shows that hijacking attempts were stimulated by prior hijackings, especially when a ransom was paid. In most studies, however, these distinctions are not or cannot be made. We typically know that potential adopters are brought into contact with the diffus- ing practice but do not know quite what they see, particularly whether they ob- serve results. This inability to specify what is observed produces some theo- retical fuzziness about the microprocesses involved in diffusion. DIFFUSION 269