Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Networks in Social Psychology, Beginning with Kurt Lewin Affiliation Patrick Doreian [email protected] Department of Sociology, University of Pittsburgh, Pittsburgh, PA, USA Faculty of Social Sciences, University of Ljubljana, Ljubljana, Slovenia Synonyms Human Group: Group, human social groups Glossary A human group is composed of two or more interdependent individuals who interact and influence each other through their interactions. A Social Network is made up of social actors (possibly of multiple types) linked by dyadic ties (from one or more social relations) and multiple memberships in collective arrangements. The simplest social network is a small group of humans interacting with each other in a specific context. Social Network Analysis: This is a program of research focused on the ties linking social actors that: is grounded in systematic empirical data; utilizes graphic imagery for displaying structural information; and uses mathematical and/or computational models to analyze, and understand, structural social phenomena. Social Psychology is the systematic study how people are influenced by the presence of others in social contexts with regard to behaviors, perceptions, cognitions, motivations, thoughts, and feelings, and how these attributes influence interaction. Group Dynamics: This is system of human behaviors and psychological processes occurring within a human group (intra-group dynamics) and between human groups (inter-group dynamics). Definition Social network analysis (SNA) and Social Psychology are research fields within which attempts are made to understand human social behavior. Despite seeming communalities between these research programs, considering them together is particularly interesting because, historically, there have been periods of 1 inherent tension between them. Robins and Kashima (2008: 1) describe this tension succinctly. “All too often, the research disciplines of social psychology and social networks talk by each other, often unaware of each other’s contributions, each with a singular perspective on how to research and understand social phenomena.” This essay uses some of the contributions of Kurt Lewin to reflect on this tension and point towards future research integrating these research domains for understanding structural social phenomena. Introduction Historical accounts of academic disciplines are varied, incomplete, and often controversial. They are attempts to identify the intrinsic features of disciplines and so identify boundaries between them. As such, they are limited because disciplines are sprawling collections of ideas that spread over these boundaries and involve some people who actively stretch boundaries when pursuing research ideas. This essay goes beyond a simple historical account by tracing some older ideas to examine their contemporary relevance. Also, the work of Kurt Lewin (1890-1947) cannot be contained within a single discipline. Disciplines also change, sometimes dramatically, over time when new data are obtained, new methods are developed, new paradigms are adopted, and new theories are developed. Despite their differences, all historical accounts feature people and places as well as content. Lewin is one of the creators of group dynamics as a set of methods for studying human groups. He also played a role in the development of modern social network analysis. Forsythe (1990) provides a detailed treatment of group dynamics and Freeman (2004) presents a comprehensive account of the emergence of social network analysis, one that is focused on ideas. Rightly, both books consider contributions from Lewin. Historical Background Lewin was a geographical and intellectual migrant. In terms of Lewin’s fields at the university level, he first studied medicine but switched to biology before moving to philosophy and psychology. Within psychology he moved between behavioral psychology and Gestalt psychology. He taught first in Germany. His academic stays in the US included Stanford University (in 1930, before emigrating) and then, having emigrated to the US in 1933, Cornell University, the University of Iowa, and MIT. Lewin’s contributions to Social Psychology and Social Network Analysis His contributions take two forms: i) developing ideas, theories, and approaches for studying human behavior; and ii) building disciplinary social infrastructure (programs) within which researchers can work on significant problems. Both have been consequential for the study of social networks. (Some of) Lewin’s ideas Lewin formulated the notion of a field as a set of interdependent features integral to all social contexts containing humans. More importantly, he saw the field as having forces acting on people that drove, in 2 part, their behaviors. Among the features of a field were the motives of people, their goals, their needs, their anxieties and beliefs. These are all concepts operating at the individual level and are ‘psychological’. However for Lewin, the features in a field, crucially, included other individuals in human groups and the interactions of these individuals. These all are parts of the social system of a group and are clearly ‘social’. When Lewin proposed the concept of group dynamics it was explicitly formulated to include the ways in which humans, and the groups to which they belong, respond to their environments. He saw individual and group attributes and processes were a part of the dynamics of (human) social groups. In describing groups, Forsyth (1990: 12) formulates a question: “And what word adequately summarizes 1) the group’s capacity . . . as an arena for social interaction, 2) the powerful impact of group structure on members’ actions, 3) the diversity of groups in terms of size, 4) their usefulness . . . for accomplishing goals, 5) the ways groups become cohesive, and 6) their ability to change over time?” He provides Lewin’s answer: dynamic. Lewin also proposed two additional ideas as crucial for understanding behavior in human groups. One was interdependence of fate to capture the realization of individuals that their fates are linked to each other’s fates and to the fate of the group. The second was task interdependence to emphasize that the goals of group members are coupled as are their actions in the group. Lewin introduced a formula, consistent with his belief that mathematics could be useful for understanding individual and group behavior. Denoting behavior by b, a person by p, and using e to denote the environment, it stated b = f(p, e). Apart from using a short set of symbols to represent the notion that the behavior of a person can be coupled to the environment within which the person is located, this equation is pure nonsense because the variables are not operationalized and the form of the function is not specified. Even so, it has some relevance for our discussion. Social infrastructure for research Freeman (2004: 67) describes Lewin as a gifted teacher and, at Iowa, “he trained a new generation of social psychologists (who) were imbued with respect for empirical – particularly experimental – research.” This training provided a set of rich resources for social psychology. When he moved from Iowa to MIT he formed the Research Center for Group Dynamics (RCGD) in 1945. The design features of this research group included an emphasis on the importance of theory, studying the interdependence of humans and their environments, and the idea that theory is useful for practice. Two years later, he died too early of a heart attack at age 57. In the view of its administrators, with RCGD lacking such a charismatic leader, MIT was not willing to fund it further. The Center was moved to the University of Michigan in 1948. The members of this research group, at both MIT and Michigan, were extraordinarily productive and the authors of the papers and books they produced in the late 1940s and early 1950s read 3 as a ‘who’s who’ of prominent social psychologists after WWII. This continued well into the 1960s and beyond. The tension between social psychology and social network analysis A radical psychological (individualist) stance argues that to understand the perceptions, motivations, cognitions, and moods of people all that are needed are the people, sets of instruments, and experimental designs to control everything else. There is also a radical structural stance that claims that all we need to understand social behavior of humans in networks is the structure of those networks. Indeed, some social network presentations emphasize the difference between ‘attribute social science’ where the covariance structure of variables measuring human attributes are analyzed to understand social phenomena and ‘structural social science’ where explanations come from characterizing structures and locating actors in these social structures. Of course, the latter were deemed superior in these accounts. Robins and Kashima (2008: 1) describe the difference between typical social psychological and social network analytic explanations. On the study of perceptions, they point out that “. . . with notable exceptions, social psychological research is not usually concerned about how . . . individual behaviors cumulate to a social system that may feed back on to affect the processes of social perception being studied.” And in commenting on SNA research, they write “social network studies typically pay little attention to the motivated social cognition of individuals, and thereby risk a seriously under-theorized account of a system of human social actors.” While they are correct with both of their arguments, I focus more on the SNA ‘deficit’ with regard to what is studied. A huge part of Lewin’s legacy was in the infrastructure for research that he created, especially RCGD. As noted above, he recruited many talented individuals and created for them a research arena for their work. Two examples are pertinent for this discussion. One was created by Alex Bavelas. He conducted an innovative communication experiment where research subjects were placed into groups to solve a collective problem but with constraints on who could communicate with whom. Bavelas (1951) showed that the permitted communication network structure was predictive of the speed with which the collective task was completed. There was evidence that the reaction of the subjects to the experiment depended on where they were located in these networks. From this study, two fundamental social network paradigmatic beliefs emerged: i) the network structure of the group affects the collective outcome of the group; and ii) where actors are located in a network affects their individual outcomes. When stated in this way, it is clear that individual attributes can be seen as irrelevant. Indeed, some of the subsequent communication/centrality experiments demonstrated that placing individuals with different attributes in different positions made no material differences to the outcomes. The experimental results were convincing but the general social network ‘tribal belief’ about the overriding impact of network structure was over-generalized because the severe constraints imposed by the experimental design were ignored. 4 At RCGD, Dorwin Cartwright (the initial director of this center after it moved to Michigan) together with Frank Harary provided a mathematical formalization of Heider’s (1946) ideas about the dynamics of signed relations. While Heider was not a disciple of Lewin, he paid close attention to the cognitions and perceptions of the actors in signed triples with two actors (p and o) and a social object (x) in pox-triads in ways that were fully consistent with Lewin’s ideas about group dynamics. He focused also on triples of social actors (poq-triples). His theory was about the dynamics induced by triples that were imbalanced (having an odd number of signed ties) and so created stress for individuals located in them. Heider argued that actors responded by trying to reduce this stress. His conception the dynamics of a group was fully consistent with the notion of group dynamics envisioned by Lewin. Also, Cartwright and Harary’s (1956) formalization of part of Heider’s theory was consistent with the Lewinian belief that mathematics can be used fruitfully to understand the results of human behavior. Their formalized version of Heider’s theory led to a structure theorem that stated that if a signed network (having at least one negative tie) had only balanced signed triples then the actors could be partitioned into two clusters such that all of the positive ties were between actors in the same cluster and all of the negative ties were between actors in different clusters. This result couples micro-level individual dynamics to the resulting macro structure of the group and remains a brilliant example of resolving the ‘macro-micro problem’. This was consistent with Lewin’s conception of fields and forces acting on humans. Yet it came at an enormous cost. In this formalization, the distinction between the pox-triples involving Heider’s unit formation relations linking social actors, p and o to social objects, x, and poq-triples involving three human actors was ignored by discarding the former (Doreian, 2004). Worse, the cognitive representations/perceptions of the network ties in the minds of actors that were integral to Heider’s theory were discarded also. Finally, all sense of the structural dynamics leading to changes in network structures was dropped in favor of establishing an existence theorem for the state of the network once these dynamics were finished. As Robins and Kashima (2008: 7) put it, “This was nice mathematics but, by stripping away the dissonance and stress aspects of the explanation, the explicit psychological aspects were disregarded.” Consistent with this, the empirical evidence for Heider’s theory about signed network structures tending towards balance was, at best, modest and mixed. There have been at least two responses to this. Hummon and Doreian (2003) presented a simulation study of ‘Heider actors’ that allowed for actors having different images of the signed ties that were around them. If their perceptions were inaccurate, these errors are part of the features of a Lewinian field within which actors are located. The parameters that were varied in this simulation were group size, the ways that changes in ties were reported to others, and the initial level of contentiousness (proportion of negative ties) in groups. The outcome measures were the number of actor choices needed to reach equilibrium, the 5 level of imbalance (if any) at equilibrium, the number of actors whose cognitive image of the networks was balanced, and the number of clusters that existed at equilibrium. The structural trajectories of these simulated groups varied greatly and made it clear that the details of balancing processes are very complex. The overall results contain a strong suggestion that Heider’s theory about change will be very difficult to test empirically without the cognitions of the actors involved. Lewin would not have been surprised by this implication. When structural balance was studied, the focus was on structural balance alone. A second response tackled the issue that structural balance in not the only active process in small group dynamics. The presence of multiple processes affecting change in network structures is also consistent with Lewin’s ideas of what can operate in fields. Relaxed structural balance (Doreian and Mrvar, 2009) allows for the operation of additional processes affecting the ties among social actors. Methodologically, it is an approach that allows the consideration of signed two-mode networks that contain both social actors and social objects as two sets of units (the two modes). As such, this operationalizes explicitly Heider’s idea of unit formation relations in pox-triads and extends it to permit the analysis of signed two-mode networks. The examples of the communication/centrality experiments and the structural balance literature are instructive regarding the influence of Lewin: This influence is inconsistent because the general ideas of Lewin contain so much that it is hard to hold all of the ideas together in a single study. The Bavelas (1951) study was consistent with Lewin’s sense of experimental rigor and that network structure mattered for group outcomes. Some of the lessons learned from the experimental results ignored the implicit scope conditions of the experiment and the claims about the primacy of network structure for group outcomes were part of a radical structuralist position. The Cartwright and Harary (1956) generalization of Heider (1946) was a brilliant demonstration of the power of mathematics for studying human groups, as Lewin would have expected, but it discarded features of Heider’s theory – actor perceptions, actor stress and actions to reduce stress, and changing network structures over time - that seem far more important given Lewin’s formulation of group dynamics. It is not clear if relaxed structural balance goes far enough in the direction of embracing the group dynamics approach advocated by Lewin because, thus far, actor attributes are left implicit. It may be that Lewin’s very comprehensive conception of features of fields is so extensive that applications including some of these features can be seen as inconsistent with Lewin’s vision by leaving out other features. “The invasion of the physicists” Lewin’s advocacy of using mathematics and his formulation of ‘fields’, ‘forces’ and ‘force fields’ acting on humans through complex dynamics can be viewed as ideas drawn from physics and so form an apparent open invitation for physicists to enter the social network field. Even his nonsense equation could 6 be seen as a welcoming sign, one to be improved upon. While it is unlikely that those entering from physics have read any of the work of Lewin, they have entered or, in their view, have invented a ‘new’ field called ‘social networks’, later relabeled as ‘network science’. Their entry appears to be very much in the radical structuralist framework where networks are thought to take generic topological forms and the vertices have few, if any, interesting properties. Many networks are said to be, in essence, self organizing systems that generate, for example, ‘scale free’ networks (Barabási and Albert, 1999) or ‘small world’ networks (Watts and Strogatz, 1998) with mechanisms like ‘preferential attachment’ for the former and ‘rewiring’ (or creating bridges between dense clumps of ties) for the latter. While these approaches merit (selective) attention from social network analysts (Bonacich, 2004), the focus on topology alone is disturbing. If social network analysts were to attend a convention of physicists and ask if they could interview some atoms or molecules of physical systems, they would be laughed out of the room. Yet within the physics-oriented social network paradigm it seems appropriate to treat social networks as if their parts are atoms devoid of relevant properties. Self-organization of networks may be an outcome resulting from forces operating - but so much of Lewin’s conception of group dynamics is left out that, when studying human social groups, the level of understanding that can be achieved will be stunted. To avoid this, a critical empirical task for social network analysts is to couple ideas of social network selforganization with actor agency rather than choosing to focus on one of these ideas or the other. Key Points and Future Directions Kurt Lewin had a profound, but not exclusive, impact on the study of social network analysis and on social psychology. Yet not much of his writing is relevant beyond providing inspiration for those who work in these fields now. He had a huge immediate impact through the generation he trained. However as the two fields have moved so far since then, especially in the creation of many new tools, the relevance of his specific prescriptions have faded. Yet, his general claims retain their importance. Social network analysts ignoring them do so at their peril. Much useful knowledge has been, and will continue to be, created within social psychology (while ignoring social ties of actors and contexts) and within social network analysis (while ignoring actor attributes and, too often, contexts). But unless some scholars work across this divide then our overall collective achievements will remain too limited. Robins and Kashima (2008: 10) put it well in urging that the two fields “at some appropriate level . . . integrate their findings and knowledge, and thereby to permit a richer understanding of how humans operate within the complexities of the modern world.” They provide some examples where this has been done - but more of us in the social network analytic field need to put our shoulders to this particular wheel. Embracing the generality of Lewin’s vision for social scientific research will provide both inspiration and motivation for such an effort. 7 Cross-References Evolution of Social Networks, 00318 Futures of Social Networks: Where Are Trends Heading?, 00064 Human Behavior and Social Networks, 00235 Modeling Social Preferences Based on Social Interactions, 00016 Models for Group Formation, 00181 Origins of Social Network Analysis, 00362 Signed Graphs, 00251 Structural Models, 00264 References Books Freeman, L. C., (2004) The Development of Social Network Analysis: A Study in the sociology of Science, Vancouver BC: ΣΡ Empirical Press. Forsythe, D. R., (1990) Group Dynamics (Second Edition), Pacific Grove, CA: Brooks/Cole Publishing Company. Articles Barabási, A-L and Albert, R (1999): “Emergence of scaling in random networks, Science, 286: 509-512. Bavelas, A. (1951): “Communication patterns in task oriented groups”, Journal of the Acoustical Society of America, 22: 725-730. Bonacich, P. (2004) “The invasion of the physicists”, Social Networks, 26: 285-287. Cartwright, D. and Harary, F. (1956): “Structural balance: A generalization of Heider's theory.” Psychological Review, 63: 277-292. Doreian, P. (2004): “Evolution of human signed networks” Metodološki Zvezki, 1: 277-293 Doreian, P. and Mrvar, A. (2009): “Partitioning signed social networks”, Social Networks. 31: 1-11. Heider, F. (1946): “Attitudes and cognitive organization”, Journal of Psychology. 21: 107-112. Hummon, N. P. and Doreian, P. (2003) “Some dynamics of social balance processes: Bringing Heider back into balance theory”, Social Networks. 25: 17-49. Mrvar, A., & Doreian, P. (2009): Partitioning signed two-mode networks. Journal of Mathematical Sociology, 33, 196-221. Robins, G. and Kashima, Y. (2008): “Social psychology and social networks: Individuals and social systems”, Asian Journal of Social Psychology, 11: 1-12. Watts, D.J. and Strogatz, S.H (1998): “Collective dynamics of ‘small world’ networks”, Nature, 393: 440-442. 8