Chapter 2 -- continued
Development of social network analysis 27
Networks: Total and Partial
It was in the work of a small group of active fieldworkers associated with the Department of Social Anthropology at Manchester University - most notably John Barnes, Clyde Mitchell and Elizabeth Bott 16 _ that the framework of social network analysis took a novel turn. Those whom I have loosely described as the 'Manchester' anthropologists were even more strongly influenced by RadcliffeBrown than were their Harvard counterparts, and they sought to develop his ideas in a novel direction. Instead of emphasizing integration and cohesion, they emphasized conflict and change. A central figure at Manchester was Max Gluckman, who combined an interest in complex African societies with a concern to develop a structural approach which recognized the important part played by conflict and power in both the maintenance and the transformation of social structures. For Gluckman, conflict and power were integral elements of any social structure, and his analyses stressed the everpresent activities of negotiation, bargaining and coercion in the production of social integration. Gluckman actively encouraged his colleagues and students who were undertaking investigations of small-scale interpersonal communities to pursue these themes.
The dominance of the Parsonian approach to sociology and of cultural approaches in anthropology during the 1950s was an important factor in directing the work of the Manchester school as a distinctly critical tradition. Where classical sociologists had emphasized that actions were to be understood in terms of their location in a structure of social relations, Parsons (1951) held that actions must be explained as expressions of internalized value orientations. The work of the Manchester anthropologists, with its emphasis on seeing structures as 'networks' of relations, combined the formal techniques of network analysis with substantive sociological concepts ' This proved an impressive and powerful mixture, which brought it close to the emerging framework of 'conflict theory' in sociology, but their emphasis on interpersonal relations meant that it did not appear as a full-blown alternative to Parsonian theory. For this reason, social network analysis could not help but be seen as a specialized area of study rather than a critical alternative to conventional sociology.
The Manchester researchers, then, paid less attention to the formally institutionalized norms and institutions of a society and rather more to the actual configuration of relations which arose from the exercise of conflict and power. The theoretical ideas inherited from the past, geared to the understanding of simple, kinship-based societies, were unable to handle these phenomena,
28 Social network analysis
and it was in recognition of this inadequacy that they began to try to systematize such metaphorical notions as the 'web' and 'network' of social relations to which such writers as Radcliffe-Brown had pointed.
Initially, these researchers began to employ the idea of a social network simply in its metaphorical sense, but Barnes, in the early 1950s, took a lead in applying this idea in a more rigorous and analytical way. His approach had a considerable influence on the work of Bott, and the two began to explore more closely the work that had been undertaken in the sociometric tradition. Their various papers (Barnes, 1954; Bott, 1955, 1956) received a broad welcome among social anthropologists, the concept of the social network seeming to meet a need for appropriate concepts to use in understanding complex societies. Siegfried Nadel espoused this approach in a set of lectures and an associated book (Nadel, 1957) which became a programmatic charter statement from a leading figure in the discipline. But it was Clyde Mitchell who undertook the tasks outlined by Nadel and laid the basis for a systematic framework of social network analysis. Mitchell turned to the mathematics of graph theory which had emerged from the early sociometric concerns, and he reformulated these ideas as the basis of a distinctly sociological framework. Summarizing the ideas that had begun to crystallize during the 1950s in his own work and that of his colleagues (Mitchell, 1969), he set out a body of sociological concepts which, he believed, could adequately grasp the structural properties of social organization. Intriguingly, Mitchell's translation of graph theory and sociometry into a sociological framework led him to a concentration on exactly those features of 'informal' and interpersonal organization that had been highlighted by Mayo, Warner and Homans.
Barnes began his academic career at the Rhodes-Livingstone Institute in Central Africa, a major research centre for many of the Manchester anthropologists. After joining the Manchester department in 1949 he decided to undertake some fieldwork in the rather unusual setting of a fishing village in south-west Norway. Although it was a small village community, Bremnes was not an isolated locale structured exclusively through the kinship relations of its members. It was an integral part of a complex and socially differentiated national society, but it had its own economic, political and other institutions, which were only imperfectly coordinated into an integrated system. Barnes was strongly drawn to the part played bv kinship, friendship and neighbouring in the production of community integration. These primordial relations were not directly tied to territorial locales or to formal economic or political struc-
Development of social network analysis 29
tures. Instead, they formed a distinct and relatively integrated sphere of informal, interpersonal relations. Barnes claimed that 'the whole of social life' could be seen as 'a set of points some of which are joined by lines' to form a 'total network' of relations. The informal sphere of interpersonal relations was to be seen as one part, a 'partial network', of this total network (Barnes, 1954: 43).
Bott, a Canadian psychologist, had studied anthropology under Lloyd Warner at Chicago, and it may be assumed that, like Barnes, she had some familiarity with the Yankee City studies. She joined the Tavistock Institute in 1950 and soon began some fieldwork on the lives of a number of British families. Bott was principally concerned with their kinship relations, and she employed the concept of a 'network' as an analytical device for investigating the varying forms taken by these kinship relations. This work was published in two influential articles and a book (Bott, 1955, 1956, 1957), and it was the basis of the PhD that Bott received from the London School of Economics in 1956.
The evolving theoretical framework of her study was undoubtedly influenced by her colleagues at the Tavistock Institute which had, in 1947, joined with the Research Center for Group Dynamics at Ann Arbor to publish the journal Human Relations. As a psychologist with an interest in psychotherapy, she was aware of the work that had been undertaken by Moreno. Indeed, both she and Barnes cited Moreno in their own papers. The more immediate influence on Bott's work, however, was Lewin's field theory, and even Barnes wrote of the existence of distinct 'fields' of activity in Bremnes society. Human Relations published articles by Lewin, Festinger, Newcomb, Cartwright and other American leaders of small-group research, and it was there that both Bott and Barnes published their work on social networks.
Barnes had presented his initial ideas in seminars at Manchester and Oxford during 1953. It was in 1954 that Bott learned of Barnes's work and adopted the term 'network' as the basis of her own theoretical interpretations. By the time that Barnes's article was published, he was working under Raymond Firth at the London School of Economics, where Bott was already registered for her PhD - she presented drafts of her own paper that year at the LSE and at Manchester. These biographical details are not given for purely antiquarian reasons, nor are they given simply as illustrations of the importance of academic networks. My concern is to show how a small number of key individuals were responsible, in a very short space of time, for constructing the basis of a major theoretical innovation in British social anthropology. Once Barnes and Bott had made their breakthrough, the way was open for further
30 Social network analysis
developments which would consolidate their advances with further lessons from the American researchers. A key voice in legitimating this direction of theoretical advance was Siegfried Nadel. An Austrian psychologist, influenced by K6hler and Lewin, Nadel had transferred to anthropological studies in the early 1930s. In 1955 he presented a series of lectures on social structure at the LSE. Bames and Bott had been important influences on the development of his work, and they were mentioned as both commentators and friends in the Preface to the published version of these lectures (Nadel, 1957). Nadel's starting point was a definition of 'structure' as the articulation or arrangement of elements to form a whole. By separating the forms of relations from their contents, he argues, the general features of structures can be described and they can be investigated through a comparative method. To pursue the aim of the construction of formal models, he advocated a mathematical approach to structure.
Social structure, according to Nadel, is 'an overall system, network or pattern' of relations (1957: 12), which the analyst abstracts from the concretely observable actions of individuals. By 'network' he means 'the interlocking of relationships whereby the interactions implicit in one determine those occurring in others, (Nadel, 1957: 16). A particular claim of Nadel was the idea that ,role' should be seen as the central concept in sociological theory. Social structures are structures of roles, and roles, together with their role sets, are defined through networks of interdependent activities. Nadel argued that algebraic and matrix methods should be applied to role analysis, but apart from one or two brief illustrations, he gave little indication of how this was to be done. His early death, in 1956, prevented him from contributing further to the advances that he had signposted.
Mitchell and others associated with Manchester and the RhodesLivingstone Institute attempted to systematize this view during the 1950s and 1960s. Indeed, Mitchell can be seen as the true inheritor of Nadel's aspirations. Mitchell's codification of social network analysis in 1969 generalized Barnes's conception of the sphere of interpersonal relations into that of the 'personal order'. 17 The personal order is the pattern of 'personal links individuals have with a set of people and the links these people have in turn among themselves' (Mitchell, 1969: 10). These patterns of interaction are, for Mitchell, the sphere of network analysis. Such interpersonal networks, he added, are built from two different ideal types of action, which combine in varying ways to form concrete interaction networks. There is, first of all, 'communication', which involves the transfer of information between individuals, the establishment of
Development of social network analysis 31
social norms and the creation of a degree of consensus. On the other hand, there is the 'instrumental' or purposive type of action, which involves the transfer of material goods and services between people (1969: 36-9). 18 Any particular action will combine elements of both of these ideal types, and so particular social networks will embody both a flow of information and a transfer of resources and services.
Mitchell goes on to conceptualize the 'total network' of a society as 'the general ever-ramifying, ever-reticulating set of linkages that stretches within and beyond the confines of any community or Organisation' (Mitchell, 1969: 12). In actual research, he argues, it is always necessary to select particular aspects of the total network for attention, and these aspects he conceptualizes as 'partial networks' ' There are two bases on which such abstraction can proceed, though Mitchell concentrates his own attention almost exclusively on one of these. First, there is abstraction which is 'anchored' around a particular individual so as to generate 'ego-centred' networks of social relations of all kinds. Second is abstraction of the overall ' 'global' features of networks in relation to a particular aspect of social activity: political ties, kinship obligations, friendship or work relations etc. For Mitchell and for most of the Manchester researchers, it was individually anchored partial networks that were to be the focus of attention. In this kind of research, individuals are identified and their direct and indirect links to others are traced. Such research generates a collection of ego-centred networks, one for each of the individuals studied. A similar approach was taken in Bott's earlier investigation of the ego-centred networks of husbands and wives, where she measured the 'connectedness' of these networks and the degree of overlap between marital partners' networks.
Mitchell recognizes the importance of the second mode of abstraction identified above - that which defines partial networks by the 'content' or meaning of the relations involved - but he sees this also as needing to be anchored around particular individuals. The 'partial networks' studied by sociologists and social anthropologists are always ego-centred networks focused around particular types of social relationship. Most such networks, Mitchell argues, are 'multistranded' or 'multiplex': they involve the combination of a number of meaningfully distinct relations. Thus, Barnes's original notion of the network, and that taken up by Bott, was a partial network in which kinship, friendship and neighbourliness were combined into a single, multi-stranded relationship which it was inappropriate to break down into its constituent elements.
Interpersonal networks, Mitchell claimed, can be analysed through a number of concepts which describe the quality of the
32 Social network analysis
relations involved. These are the 'reciprocity', the 'intensity' and the 'durability' of the relations (Mitchell, 1969: 24-9), concepts which echo Homans's distinctions between direction, frequency and intensity. Some, but not all, relationships involve a transaction or exchange, and so can be considered as 'directed' from one person to another. An important measure of such relations, therefore, is the degree to which the transaction or orientation is reciprocated. One person may, for example, choose another as a friend, but this choice may not be returned: the chooser may be ignored or spurned. Multistranded relationships can involve a complex balance of compensating reciprocated and unreciprocated relations, in which, for example, financial aid flows in one direction and political support in the other. 19 'Durability' is a measure of how enduring are the underlying relations and obligations which are activated in particular transactions (Mitchell refers to Katz, 1966). Those which are constantly being activated in interaction are highly durable, while those which persist only for one or two activities are highly transient. While kinship obligations, for example, are very durable - they last for the whole of one's life - those that arise for a particularly limited purpose are more likely to be transient. 'Intensity' refers to the strength of the obligations involved in a relation. This reflects either the strength of the commitment to these obligations or the multiplexity of the relationship: multi-stranded relationships tend to be more intense because they are more diffuse in character. 20
Mitchell adds a further set of concepts, derived from a translation of graph theory into sociological language, which can be used to describe the texture of social networks. 'Density', for example, he sees as the completeness of the network: the extent to which all possible relations are actually present. This is what Barnes and Bott tried to describe with their notions of the 'mesh' and 'connectedness' of networks. 'Reachability' refers to how easy it is for all people to contact one another through a limited number of steps: how easy is it, for example, for gossip, ideas or resources to be diffused through the network. To these concepts, Barnes (1969) has added 'cliques' and 'clusters' as terms for identifying social groupings within networks, but these were not taken up in the empirical studies collected together by Mitchell (1969).
Institutionalized roles and statuses are the framework within which interpersonal networks are constructed, but they exist only in and through the reproduction of interpersonal networks. But Mitchell and the Manchester tradition equivocated about whether the institutional structure of roles is itself a part of network analysis or is separate from it. While some of the Manchester school saw the
Development of social network analysis 33
institutional role structure as a network of relations which exists alongside the interpersonal network, Mitchell often distinguished networks of interpersonal relations from structures of institutional relations. Mitchell's discussion, therefore, tended towards a 'residual' definition of the social network: network analysis concerns only the interpersonal sphere that is left behind after formal economic, political and other roles are extracted (Whitten and Wolfe, 1973). This proved to be highly significant for the subsequent development of social network analysis in Britain. To the extent that he sees social network analysis as a special method for the analysis of interpersonal relations, Mitchell departs from Nadel's aspiration for a general framework of structural sociology rooted in formal network analysis. This equivocation proved fateful for the development of social network analysis in Britain, which largely failed to attract adherents from outside the area of community studies.
Return to Table of Contents