Chapter 4 - continued

Community Structure and Density

The power and utility of density analysis can be illustrated through some concrete studies. Barry Wellman (1979, 1982), a member of Harrison White's original cohort of network analysts at Harvard,

80 Social network analysis

has supervised a large study of community structure, in which density plays a key role. He took as his starting point the longstanding tradition of community studies, in which writers on 'community' were generally concerned to investigate whether the communal solidarities associated with small-scale, rural villages had been able to withstand the modernizing forces of industrialization and urbanization. Wellman wanted to use social network analysis to see whether the development of modern society had resulted in the disappearance of community and the emergence of urban anomie. It had been pointed out by some critics of community studies that social relations of all kinds had become detached from specific localities, with relations having an increasingly national or international scope (see the discussion in Bulmer, 1985). Wellman's research aimed to investigate this issue for a particular urban area in Toronto, East York, and, like Fischer (1977, 1982), he focused on the question of whether 'personal communities' had stretched beyond the bounds of the local neighbourhood itself.

East York is an inner city suburb of private houses and apartment blocks and was, at the time of the research in 1968, occupied mainly by skilled manual workers and routine white collar workers. The fieldwork involved interviews with a random sample of 845 adults, and a central question in the interview asked people to name their six closest associates. They were then asked to say whether those named were themselves close to one another (see also McCallister and Fischer, 1978). The responses to these questions could be used to construct ego-centric networks of intimate association for each respondent. By asking about the connections among the persons who were named by each respondent, Wellman was able to measure the density of each personal network. The calculation of density followed the strategy outlined earlier, and ignored links between respondents and their intimates. That is, data were collected on ego and his or her six intimate associates, but the densities of the egocentric networks were calculated for the links among the six associates only.

Wellman discovered that many of the intimate associates (about a half) were relatives of the respondents, but kin and non-kin associates were all to be found across a wide geographical area. The majority of all links were with people who lived in the city itself, though very few of these links were based in the immediate locality of East York. A quarter of all the intimate associates who were named lived outside the city, some living overseas. Having made a number of these summary statements about the broad framework of people's social networks, Wellman turned to the densities of these networks. The mean density of the ego-centric personal networks of

Lines, direction and density 81

the respondents was 0.33, 1 3 only one-fifth of networks having a density greater than 0.50 (Wellman, 1979: 1215). A density of 0.33 meant that five out of the 15 possible links among intimate associates were actually present. 14 Wellman discovered that the densest networks tended to be those that were composed mainly of kin, owing to the fact that it was more likely that the kin of the respondents would maintain mutual contacts. Where kinship obligations were absent, such contacts were less likely to be maintained.

Figure 4.6 Density of personal networks

Wellman's principal findings on personal networks are summarized in Figure 4.6. He interprets these data as indicating that people were involved in networks which were 'sparsely knit' - i.e., networks of low density. 'Communal' links were neither solidaristic nor localized. People had others that they could rely on, but the low density of their personal networks, their lack of mutual crosslinkages, meant that such help was limited. These personal networks were, nevertheless, important sources of help and support, on both an everyday basis and in emergencies: 'East Yorkers can almost always count on help from at least one of their intimates, but they cannot count on such help from most of them' (Wellman, 1979: 1217). Those intimate associates who were less likely to give help and support were more likely to be significant for sociability. Helpers were more likely to be kin, while those who were most important for sociability were more likely to be co-residents or co-workers.

To pursue some further issues, a follow-up study was undertaken in which in-depth interviews were carried out, during 1977-8, with 34 of the original respondents. The aim was to get more 'qualitative' contextual data for the structural data of the earlier study. Although the detailed results of this stage of the inquiry go beyond the immediate concerns of this chapter, some of the directions pursued can usefully be outlined. Wellman discovered that the interpersonal

82 Social network analysis

networks of households were differentiated by gender divisions and by the involvement of household members in paid work. The research discovered, for example, a number of differences between households where women were involved in paid employment and those where they were involved only in domestic work. He discovered that the social relations of a household and their access to interpersonal support from their kin, friends, neighbours and coworkers were most likely to be maintained by women rather than by men. This was, in particular, true of households where women were engaged solely in domestic work. Households where women were involved in both domestic work and in paid employment had far less dense networks of relations and were, therefore, able to obtain less support and fewer services from their contacts (Wellman, 1985). 15

Wellman's investigations used survey analysis to generate the relational data which he used in the study, but similar ideas can be used on other forms of relational data. Smith (1979), for example, used historical data derived from documentary sources to investigate communal patterns in an English village in the thirteenth century. Smith's data came from the records of the manor court of Redgrave in Suffolk, these records showing such things as patterns of landholding, property transactions and financial disputes among the villagers. In total, he considered 13,592 relations among 575 individuals over the period 1259 to 1293. Initially, he analysed the different types of relations and their frequency, which showed that about two-thirds of the relations were 'pledging' relations. These were relations in which one person gave a specific legal commitment in support of another person in relation to debt re-payments and other financial arrangements.

Smith's concern was with the role of kinship and other local ties in organising these relations and in structuring communal relations. Homans (1941) had previously undertaken a similar historical study of communal solidarity, but had not applied any social network concepts in his study. By contrast, Smith used the idea of the egocentric network as his principal orientating concept. The 425 Redgrave landholders of the year 1289 were divided into four categories according to the size of their landholdings, and equalsized random samples were drawn from each category. This gave 112 individuals for analysis, and their documented relations with all other people over the ten-year period from 1283 to 1292 were extracted from the database. The personal, ego-centric networks of the 112 people, taking account of the distance 1 relations, were then analysed in terms of their social bases and geographical spread. The distribution of the densities of the personal networks showed a curvilinear relation to landholding. Density increased steadily with

Lines, direction and density 83

size of landholding among those with four acres of land or less, and it decreased steadily with the size of holding for those with more than four acres. Those with three or four acres, therefore, had the densest personal networks, median density among these households being between 0.2 and 0.4. They were also the most involved in multiplex relations. It was, therefore, the middling landholders who were best integrated into their village community. In the light of the earlier discussion of the relation between network size and density, it is interesting to note that Smith discovered a correlation of just 0.012 between the two measures. He concluded, therefore, that the variations in network density which were observed were not a mere artefact of network size, but reflected real variations in the quality of interpersonal relations.

Taking a-.count of all his network data, Smith rejects the idea of a tightly knit organic community organized around kin and neighhours. The network structure of the medieval village, at least so far as Redgrave was concerned, was much looser than this image. Neither were distant kin an especially important source of social support:

those individuals who interacted most frequently with near neighbours also interacted most frequently with kin, although probably on most occasions residing apart from them. These kin, however, tended to be close: siblings, uncles, nephews, nieces, fathers and mothers, sons and daughters. (Smith, 1979: 244)

Wellman recognized that the ego-centric networks that he studied in East York were linked into chains of connection through overlapping associations: there was, he held, a 'concatenation of networks' with personal networks being 'strands in the larger metropolitan web' (Wellman, 1979: 1227). But he does not directly investigate these global features of the socio-centric networks of East York. Some pointers to this 'concatenation' are provided in Grieco's (1987) extension of the work of White (1970) and Granovetter (1974). Grieco's research concerned the giving and receiving of information about job opportunities, and she showed that the flow of help from particular individuals to their network contacts produces an alteration in the global structure of the network. Where information is received indirectly, from contacts at a distance of 2 or more, there is a tendency for a new direct link, albeit a weak one, to be established between the originator of the information and those who received it (Grieco, 1987: 108ff.). The overall density of the network, therefore, increases, and some of these links may be solidified and strengthened through feelings of solidarity and obligation. Thus, some of the initial increase in density will persist.

84 Social network analysis

When others in the network acquire the ability to reciprocate for the help that they have received they will, in turn, tend to create new direct links and a further alteration in the density of the network. In this way, changes at the individual level of ego-centric contacts result in a continual transformation of the density and the other socio-centric, global features of the network.