Introduction
The large issues which this book addresses may be posed as a series of simple questions about young peopleâs social worlds. What is the nature of the childâs social networkâhow large is it, what is its composition and structure, how does it change over time? How do the characteristics of the childâs social network affect social, cognitive, and emotional development? And, critically, what are the key variables one must consider in approaching these questions?
These questions have both theoretical and practical significance. On theoretical grounds, one would expect that from the very beginning, the childâs social contacts are essential components in the acquisition of social modes and orientations. Despite the expansion over several millennia of the ways in which human beings influence each other without having direct social contactâgossip, writing, and more recently, such mass media as radio and televisionânetworks of direct and closely mediated indirect ties remain the most powerful sources of cultural information, social feedback and reinforcement, and values. This is true throughout life but must be even more strongly true early in life, when all of the childâs social knowledge and skills are being developed, in partial imitation of and in direct response to the people immediately around the child. These direct links are initially the only routes of social transmission; the structure of these sets of links, and the way they fit into the larger networks in which they in turn are embedded, must affect the nature and adequacy of the process of social transmission.
On the practical side, research on the impact of social networks on health may serve as an example. A number of studies have found characteristics of peopleâs social networks to be strongly associated with health outcomes for a range of physical and psychiatric conditions, and even with mortality rates (see, e.g., Berkman, 1984; Leavy, 1983). These studies have largely been limited to adults, but one may expect that social contacts would also prove relevant for childrenâs health. More broadly, knowledge of childrenâs sets of social relationships can offer insight into such areas as moral and cognitive developmental processes and may suggest possible intervention strategies for children at risk for delinquency, learning problems, and various kinds of pathology.
Although it is a truism that social contact is essential to normal development in social animals, the processes are complex and require many different kinds of empirical research to be understood. We know a good deal, for example, from experimental work on young primates (e.g., Harlow, H. F., Dodsworth, & Harlow, M. K., 1965; Mitchell, 1970), about the effects of isolation, diminished social contact, and modified patterns of social contact. These effects include poor social development, pathological affective states, inappropriate sexual behavior, and inadequate development of parenting skillsâthe latter particularly important because it results in the transmission of deleterious behavioral and emotional effects from one generation to the next. From such studies with nonhuman primates, and more recently from human work as well, there has emerged the recognition that parent and peer relationships have strong, but different, impacts on development (Bemdt, 1979; Kandel, 1973; Suomi & Harlow, 1975), so that an adequate understanding of the developmental process and its adult outcomes must include attention to both. Some of the more recent primate field studies have added âauntsâ and adult male contacts to mothers and peers as having important influences on development (e.g., Rosenblum, Coe, & Bromley, 1975). The early primate studies showed dramatic effects of severe isolation, not researchable in human subjects in analogous experimental situations; however, some of the more recent field research with primates has a closer parallel with studies of childrenâs social networks.
Research on childrenâs social networks emerges from a large body of social research on children, which has been fruitful but somewhat fragmented. It has considered separately the effect of broad demographic variables such as ethnicity, race, and economic status (Harrison, Serafica, & McAdoo, 1984); it has examined the differential effects of rural and urban environments (Klineberg, 1938); it has considered the effects on the child of dyadic relationships with some specific family membersâmothers, fathers (Lamb, 1976), or siblings (Sutton-Smith & Rosenberg, 1970); it has examined the effects of some family typesâsingle or two-parent families (Herzog & Sudia, 1973), intact or divorced families (Hetherington & Camara, 1984), and working or nonworking mother households (Gold & Andreas, 1978; Hoffman, 1984); it has analyzed some of the effects of a childâs place in the family, such as birth order (Abramovitch, Corter, & Lando, 1979); it has examined peer status and found repeatedly that childrenâs social status among peers is a very strong predictor of such outcomes as physical and psychiatric health, behavior problems, learning difficulties, and delinquency (Cowen, Pederson, Babigian, Izzo, & Trost, 1973; Hartup, 1983; Robins, 1966; Roff, Sells, & Golden, 1972).
There are other areas, however, that have been neglected. Researchers have seldom quantified and examined the effects of the extended family, even though its significance for the child has presumably always been important and may even increase with the rise in divorce rate and single-parent households. Studies of peer relations and childrenâs friendshipsâwhich have recently become the subject of renewed interest and extensive research (Asher & Gottman, 1981; Coie & Kupersmidt, 1983; Dodge, 1983; Gottman & Parkhurst, 1980; Hartup, 1983)âhave all too often been entirely separated from the study of family relationships, as if the child exists in two entirely distinct social worlds. And finally, in the literature on child development, there are virtually no studies of extrafamilial significant adults.
Social network analysis provides a more comprehensive view of the childâs social environment. Social networks involve, in principle, the most elementary level of social analysis. Conceptually, social network analysis entails no cultural or psychological prejudgments beyond the minimal ones that there are individuals who have repeated contact with each other, and that there is some cumulative nature to these contacts such that a link may be said to exist between pairs of individuals when these repeated contacts occur. The distributions of these links around an individual or within a population then constitute the networks of social linkages.
The study of childrenâs networks constitutes a new field of endeavor, and we need the most basic kind of information: What constitutes normal, or adequate, social contact for children, at what ages, and does this vary with context? Even more fundamentally, what are the dimensions of networks that are essential to the establishment of basic normative data? If oneâs social space were in effect a homogeneous aggregate of all the individuals with whom one has social contact, then network size alone would constitute the critical dimension. What we so far know argues against such a stark simplification, although the effects of isolation and of excessive crowding indicate that there might be extreme conditions in which sheer amount of contact has implications for pathological behavior. Even here, however, âcrowdingâ seems to be more a function of social organization than of sheer numbers (e.g., Cassel, 1976). Assuming, then, that social space is not homogeneous, what are its significant dimensions? And how does one discover them? The answer, we expect, will be complex: Although there will probably be some invariants, we will probably also find many functional relationships in which different dimensions of social space are critical for the development, or breakdown, of different aspects of human behavior. Thus, for example, extremes of social isolation seem quite generally detrimental for social animals (e.g., Harlow et al., 1965), whereas strength of contacts or density of interconnectedness seem differentially beneficial or harmful under different conditions (e.g., Hirsch & Reischl, 1985; House & Kahn, 1985).
What constitutes ânormalâ social contact in humans is further complicated by the wide range of cultural variants in normal social behavior. It will be necessary, as we study social networks and social network effects in our own culture, to check our findings on other cultural groups to see whether the relationship between network characteristics and behavior varies with the ecological or cultural niche in which we study these effects.
A developmental perspective dictates additionally that our examination of social networks take account of the fact that normal patterns of social contact vary with age. As children grow, become more biologically mature, more self-reliant, and more socially skilled, we will expect changes to occur in their networks in response to both their needs and competence. One of the more fascinating issues that the studies in this book bear on is, what are the continuities and discontinuities, across the life span, in network processes? The infant is born into a complex social network that is, from the very beginning, not identical to anyone else s, though determined by the networks of the infantâs caretakers. How soon, to what extent, and in what ways does the infant begin to make his or her own social distinctions among available contacts, which contribute to an emerging personal autonomous network; how does this process evolve over the course of childhood and into adulthood; and what variations in the social world around the child have an impact on the process?
The analysis of childrenâs social networks raises some special methodological issues, in terms of selecting strategic network parameters for study, devising appropriate measures, and eliciting the relevant data.
Although systematic research on childrenâs social networks has barely begun, different studies have tended to utilize at least some of the same parameters, presumably drawing on the child development literature for a sense of what may be the most important variables to consider. This fortunately enables us to some degree to compare studies, to begin to verify findings and establish age norms for social contact. Thus, personal social space for children has most often been conceptualized in terms of a set of contrasts, the most important of which are kin versus nonkin, adults versus peers, and same versus opposite gender contactsâimplicitly supporting the idea that social space should not be considered to be homogeneous.
Network sizeâthat is, the number of people with whom the child has contactâis probably the most commonly used measure, as is also the case in the adult network literature. Important variants include a weighted measure of network size, which takes account of frequency of contact; and measures of the size or relative size of different network components (e.g., kin versus nonkin). Other measures appearing in the social network literature on adults, such as reciprocity, density of connections, and clustering patterns (e.g., Granovetter, 1973; Hammer, 1980; S. Salzinger, Kaplan, & Artemyeff, 1983), have only rarely been utilized in research on young people (see e.g., L. Salzinger, 1982). The sociometric literature on children is an exception to this; but it is typically limited to classmates rather than open networks (see e.g., Coie, Dodge, & Coppotelli, 1982; Dodge, 1983; Gronlund, 1959). In this book, Ladd and his colleaguesâ use of ânumber of play groupsâ shows a conceptual relationship with clustering, and Antrobus and his colleagues have been able to use a density measure on some of their data.
Collecting accurate network data on children also presents some special difficulties. Self-report has been the most commonly used data source in network studies among adults, though there is considerable variation in the form in which that report is elicited. Such self-reports are sometimes verified by complementary reports from some of those named, and the level of agreement on such things as how often the informant and the person named see each other has been found to be quite high (Hammer, 1984). For adolescents as well as adults, self-report is also the most commonly used data source. It is, however, unavailable as a data source for the youngest children, and may not be very reliable till midchildhood. Direct interviewing of school-age children has nevertheless increasingly come to be used to collect data on various aspects of childrenâs lives, supplemented by a parallel format that elicits information from a parent as well. McHale and Gamble (chapter 7), for example, interviewed both child and mother about the childâs contacts in the past 6 weeks and supplemented these data with telephone interviews of the child with respect to his or her contacts on that day.
Most of the information on young childrenâs networks has come from their mothersâ reports. For numerous reasons, maternal reports might be expected to be somewhat flawed: They are not comprehensive because mothers are not privy to all their childrenâs contacts; mothers are not necessarily sensitive to the significance of all the childâs different contacts and may underreport them; and the reports are typically retrospective and subject to the usual inaccuracies of such data. However, if data on the social contacts of babies and very young children are to be obtained, the use of mothersâ reports is inescapable. Some ingenious attempts have therefore been made to improve the quality of such data, such as S. Salzinger and Hampsonâs (see chapters 2 and 3) use of cumulative logging, which requires the mother to list the childrenâs contacts for the previous 24 hours, sampled repeatedly over a number of occasions. They have also, in some cases, supplemented maternal reports with reports from fathers and baby-sitters.
Studies with children also provide us with a methodological opportunity that is somewhat less available with adults. Because the normal settings in which children spend their time include small, relatively bounded groups (e.g., classrooms, play groups) to an important degree, we are able to observe childrenâs interactions with others directly.1 We have already begun to profit from the use of such observational methodology. Ellis and her colleagues (1981) were able, for example, to describe the changes that occur with age in the composition of peer groups; and Ladd and his colleagues (Ladd, 1983; Ladd, Price, Price, & Hart, in press), by meticulously making repeated observations of young children in playgrounds, have been able to describe not only the groups themselves but the relationship of individual children to the groups with respect to such variables as centrality and participation in cliques.