SOCIAL NETWORKS

Social network theory assumes that social interactions have the potential to influence attitudes and behavior. In the sphere of population this assumption has begun to be confirmed by demographers. Network models bear some similarity to diffusion models, but they offer a more structured approach to social interaction by focusing on the specific links that connect individuals and groups. In the 1980s interest in social networks by sociologists led to the collection of network data, the elaboration of theory, and the development of new analytic methods. Demographers have borrowed heavily from this theoretical and methodological work to guide their empirical analyses. Most have concentrated on fertility, investigating the fertility transitions of individuals in local communities, but network approaches have also been used in the study of migration and mortality, clusters of villages, population elites, and organizations.

Evidence that people do indeed talk with friends, relatives, and neighbors about fertility control can be found in historical sources, interviews with elderly people, and surveys conducted in developing countries in the 1960s and 1970s. In these surveys, the content and context of the conversations are missing. When researchers have collected qualitative data, they have found that talk about subjects such as family planning may be open, casual, and quite specific rather than private, formal, and vague. Qualitative data also provide insight into how network partners are selected, an issue relevant for determining appropriate statistical techniques. Both theory and data suggest that social interactions concerning fertility control are especially likely in situations of uncertainty.

Network theory postulates a variety of effects of networks, but in the study of fertility transitions the focus has been on new information transmitted and evaluated in networks and on the influence that networks exert on its members to adopt or resist innovations. Links among network partners are characterized in a variety of ways. An important distinction is that between strong and weak links: strong links are hypothesized to constrain the flow of new information and to exert more social influence than weak links. A variety of indicators of the strength of ties has been proposed, such as whether the network partner is a confidant, a friend, or an acquaintance and the duration of the relationship. Networks have also been characterized by the degree of homogeneity of members with respect to characteristics such as age, education, gender, and ethnicity. Heterogeneous networks may facilitate the spread of new information, whereas homogeneous networks may be more effective in exerting social influence.

Studies using cross-sectional data provide convincing evidence of an association between the attitudes and behavior of individuals and the characteristics of their networks. For example, the probability that a woman is using contraception is typically higher if her network partners are also using contraception, and the method she uses is likely to be the same as the methods used by those with whom she interacts. This probability has been found to depend also on the characteristics of the individual and her networks, as well as on the particular context; networks may also impede contraceptive use. Because of the dearth of network data, some analysts have used aggregate data to represent local or transnational networks, again finding associations that suggest the importance of networks.

While there is a clear empirical association between network characteristics and attitudes and behaviors, establishing a causal effect of networks on attitudes and behaviors has been difficult. Because actors may select those with whom they discuss topics such as family size and family planning or whether and when to migrate, determining the direction of causality requires longitudinal data or analytic techniques that take the selectivity of networks into account.

In summary, empirical work by demographers has established the potential significance of network approaches for examining processes of demographic change. By implication, analyses that treat individual actors in isolation are not sufficient. Fully realizing this potential may require further theoretical development; it will certainly require new efforts at data collection, including qualitative and survey data on the links among network partners, as well as further development of analytic techniques.

See also: Diffusion in Population Theory; Fertility Transition, Socioeconomic Determinants of; Social Capital.

bibliography

Bongaarts, John, and Susan C. Watkins. 1996. "Social Interactions and Contemporary Fertility Transitions." Population and Development Review 22: 639–682.

Casterline, John B. 2001. "Diffusion Processes and Fertility Transition: Introduction." In Diffusion Processes and Fertility Transition, ed. John B. Casterline. Washington, D.C.: National Academy Press.

Entwisle, Barbara., Ronald R. Rindfuss, David K. Guilkey, Apichat Chamratrithirong, Sara R. Curran, and Yothin Sawangdee. 1997. "Community and Contraceptive Choice in Thailand: A Case Study of Nan Rong." Demography 33:1–11.

Kohler, Hans-Peter. 2001. Fertility and Social Interaction. Oxford: Oxford University Press.

Kohler, Hans-Peter, Jere R. Behrman, and Susan C. Watkins. 2001. "The Density of Social Networks and Fertility Decisions: Evidence from South Nyanza District, Kenya." Demography 38: 43–58.

Montgomery, Mark R. and John B. Casterline. 1996. "Social Influence, Social Learning, and New Models of Fertility." In Fertility in the United States: New Patterns, New Theories, ed. John B. Casterline, R. D. Lee, and K. A. Foote. Supplement to Population and Development Review. New York: Population Council.

Valente, Thomas W. 1995. Network Models in the Diffusion of Innovations. Cresskill, NJ: Hampton Press.

Susan Cotts Watkins