SUBSTANCE USE & MISUSE Vol. 39, Nos. 10–12, pp. 1685–1712, 2004
Using Social Networks to Understand and Prevent Substance Use: A Transdisciplinary Perspective Thomas W. Valente, Ph.D.,* Peggy Gallaher, Ph.D., and Michele Mouttapa, Ph.D. Department of Preventive Medicine, Keck School of Medicine, University of Southern California, Alhambra, California, USA
ABSTRACT We review findings from research on smoking, alcohol, and other drug use, which show that the network approach is instructive for understanding social influences on substance use. A hypothetical network is used throughout to illustrate different network findings and provide a short glossary of terms. We then describe how network analysis can be used to design more effective prevention programs and to monitor and evaluate these programs. The article closes with a discussion of the inherent transdisciplinarity of social network analysis.
*Correspondence: Thomas W. Valente, Ph.D., Department of Preventive Medicine, Keck School of Medicine, University of Southern California, 1000 S. Fremont Ave., Building A-5133, Alhambra, CA 91803, USA; E-mail:
[email protected]. 1685 DOI: 10.1081/LSUM-200033210 Copyright & 2004 by Marcel Dekker, Inc.
1082-6084 (Print); 1532-2491 (Online) www.dekker.com
ORDER
1686
REPRINTS
Valente, Gallaher, and Mouttapa Key Words: Contextual factors; Social networks; Social network analysis; Peer influence.
INTRODUCTION Studies of human behavior have, by and large, focused on how individual attributes correlate and sometimes cause certain outcomes. For example, one’s level of sensation seeking might be correlated with criminal behavior. Increasingly, however, scientists have begun to realize that contextual factors (e.g., the social and physical environment) contribute significantly to variation in outcomes (Gorman et al., in press; Mason et al., in press). This is particularly true with substance use related behaviors. Social network analysis has emerged as an important perspective that provides a way to study the social context of substance use in a transdisciplinary manner.
WHAT IS SOCIAL NETWORKS ANALYSIS? Social network analysis is a set of theories, methods, and techniques used to understand social relationships and how these relationships might influence individual and group behavior. The theories used are typically embedded within other disciplines such as Anthropology, Communication, Economics, Psychology, Sociology, and many others. The common basis for these theories is that individuals are influenced by the people they have contact with, and that individual positions within larger social structures can determine behavior (either through constraint or influence). The methods of social network analysis consist of established procedures for measuring characteristics and dimensions of these relationships, and the specific mathematical algorithms (and associated software) used to operationalize network constructs (INSNA, 2003). There are two primary types of network data collection techniques: .
Egocentric techniques provide measures of a person’s local social network. For example, a researcher can ask respondents to provide the first names of their closest friends; and then ask the respondent to state whether each of these friends engages in certain behaviors; and whether the respondent engaged in these behaviors with each friend (Valente and Vlahov, 2001). These techniques are amenable to random sampling approaches,
ORDER
Social Networks
.
REPRINTS
1687
but do not generate a complete network that can be manipulated mathematically to understand the network’s structure. Sociometric techniques provide a measure of the entire social network by interviewing all members of the network. Sociometric techniques are most often used in small communities, schools, and organizations where the boundary of the network can be defined. Sociometric techniques are more powerful in the sense that they provide a global view of the network and indicators for individual positions in that network. Sociometric data are an aggregate composition of egocentric data. Sociometric techniques also require the use of specialty social network software such as UCINET or specific mathematical programming to calculate network indicators (Analytic Technologies, 2002). The ability to generate a complete social network map of social relations provides considerable explanatory power.
Sociometric social network measures are generated primarily at two ecological levels: individual and network. There are hundreds of individual measures that can be generated from one social network question. For example, asking students for the names of their friends at school can generate hundreds of indicators measuring, for each student, their relative centrality in the friendship network, their membership in groups, reciprocity of ties, and so on. The same data can be used to generate network level indicators such as the degree to which the network is dense or sparse, its centralization, the number of cliques, their overlap, and so on. Several good introductions to social network analysis are readily available (Burt, 1980; Marsden, 1990; Monge and Contractor, 2002; Rogers and Kincaid, 1981; Scott, 2000; Wasserman and Faust, 1994). This article reviews and discusses studies conducted to understand the application of social network analysis to substance use. To date, studies have primarily collected sociometric data, a (near) complete enumeration of the study population. This review will be presented in five parts. First, this article presents key social network definitions. Second, the article discusses how social networks contribute to understanding the initiation and use of substances. For example, the degree to which peer substance use is associated with an individual’s own substance use will be examined. Third, we discuss how to use social network analysis for implementing interventions designed to prevent or reduce substance misuse. Although in its infancy, this topic offers considerable promise for optimizing the interactional process in substance misuse prevention programs. Fourth, the article discusses the use of social network analysis
ORDER
REPRINTS
1688
Valente, Gallaher, and Mouttapa
for program monitoring and evaluation. Finally, the article summarizes the transdisciplinary nature of social network analysis and closes with a summary and recommendations for future directions. Figure 1 provides a graph, a sociogram, of a hypothetical social network. This graph will be used to illustrate key points in the review that follows. The network could be friendship ties among students in a school or communication among coworkers in an organization. Although this network is small, 20 nodes, network analysis can be conducted on large networks (e.g., thousands of nodes). Note that these network relations could also include weights indicating the strength of associations, for example, how close two people are. And it can also include valences, i.e., liking and disliking.
SOCIAL NETWORK DEFINITIONS Social network influences may operate at several levels, including the individual level (e.g., one’s positioning in the social network) and the group level (e.g., network density). This section describes the various
10
9
11
12
3
4
1
14
13
2
5 15
8
16
6 18
17
7
19
20
Figure 1. Hypothetical social network, each circle represents a person and the links between connections corresponding to friendship, communication, or other network definitions. In the language of sociometrics, Fig. 1 is referred to as a ‘‘graph,’’ each circle a node, and the lines connecting them are referred to as ‘‘edges.’’
ORDER
REPRINTS
Social Networks
1689
social network variables that can be generated to explain adolescent substance use (see Table 1 and Glossary). Many studies have been conducted in schools since the boundary is readily drawn, though many contexts could be chosen. The school will be used here as the context, for illustration. One of the unique aspects of social network analysis is that a question that asks students to name their friends in a school can be used to generate hundreds of different variables. One class of variables consists of measures of social position, such as centrality, bridges, and isolates.
Social Positions One unique aspect of social network analysis is its ability to categorize people in terms of their position in the network. By nature of their social relations, some people are more central or popular in a school than others. The most frequent and possibly useful position identified by social network analysis is centrality; the degree a person occupies a central position in the network. Centrality can be measured at least a dozen ways in the same network (Costenbader and Valente, 2003; Freeman, 1979; Wasserman and Faust, 1994). The most intuitive measure of centrality is a count of the number of times a person is named in response to a network question. Persons 3, 5, and 15 in Fig. 1 would be considered the most central since they received the most nominations, four. For example, in a school-based friendship network, students who are named as friends most frequently would be considered the most central students in the school and also considered the most popular. There are several other key centrality measures. For example, betweenness is defined as the degree to which a person lies on the shortest paths connecting others in the network. Closeness is the inverse of the average distance to others in the network. Integration is defined as the reverse distance to others in the network. Eigenvector is the first eigenvector of the matrix of ties. Finally, power is defined as the weighted centrality scores of a person’s ties. All of these measures are standardized by the size of the network. Popular students are often selected as peer educators in prevention programs, regardless of whether popularity in the classroom is related to tobacco or alcohol use (Terre et al., 1992). Alexander and others (2001) showed that popular students, those receiving the most ties, were more likely to smoke in schools with high smoking prevalence, and less likely to smoke in schools with low prevalence. This finding indicates that peer leaders often exemplify the norms for their communities and are likely to be earlier and, perhaps, heavier users of substances in communities where
ORDER
1690
REPRINTS
Valente, Gallaher, and Mouttapa
substance use is accepted. Since leaders are often seen as influential and harbingers of behavior, they are often used to implement programs, a point we return to later. Another position identified by social network analysis is liaison. A liaison is a person who connects otherwise disconnected or weakly connected groups (Granovetter, 1973). Liaisons are paradoxical in that they are weakly connected to groups, yet these weak connections give them strength. Granovetter (1973) referred to this as the strength of weak ties, because the ties were to people and groups which the person does not see often or is not connected to through multiple channels. Yet this weakness is a strength because it gives the liaison access to information and resources that the rest of the group do not have. In Fig. 1, Person 11 acts as a liaison between two otherwise disconnected groups. Liaisons may be at risk for substance use because they become exposed to the norms of two different groups, either of which may support misuse (Ennett and Bauman, 1993). Further, liaisons often try to fit in with a group that they are not strongly connected to and may initiate substance use in order to conform to the group’s norms. On the other hand, since liaisons are not embedded within a group, they may be more resistant to group peer pressure. Thus, liaisons represent a position that might or might not put them at risk depending on their individual characteristics and the group dynamics within the network. The lack of adequate, generalizable empirical findings in this area leaves it as an area needing more research, particularly transdisciplinarity (TD) research. Isolates, people connected to no one, can also be identified by network analysis. In diffusion of innovations research, isolates have been shown to be later adopters of innovation because their position puts them outside the flow of information about new ideas (Rogers, 1995; Valente, 1995). Similarly, in a high-risk setting where substance use is high, being an isolate may offer protection because the individual is metaphorically quarantined from negative influences in the group. On the other hand, isolates are often strongly connected to another group and this other group may put them at risk. For example, a middle school student reporting no in-school friends, and receiving no friendship nominations, may have their friends outside the school or in another school and these ties may influence the student to use and/or misuse substances. Isolates in Fig. 1 are Persons 7, 19, and 20. The increasing presence of electronic communications, cell phones, internet, and pagers has expanded social networks immeasurably by removing spatial and temporaral syncronicity as a necessary condition for communication.
ORDER
REPRINTS
Social Networks
1691
Table 1. A short glossary of network terms (see Scott (2000), Valente (1995), and Wasserman and Faust (1994) for more complete glossaries). Term Nominations
Centrality Reciprocity
Network exposure Bridges or liaisons Group
Isolate Density Transitivity Distance
Personal network density Centralization
Density
Betweenness Closeness Integration
Definition The choices a person makes in response to a network question. For example, the people that a respondent names in response to the question, who are your friends are that respondent’s nominations The degree a person is centrally located in the network. There are at least a dozen different centrality measures The degree a person’s nominations also nominate him/her. Reciprocity can be direct or indirect (A nominates B who nominates C who nominates A) The degree a person’s network engages in the behavior. For example, the number or proportion of substance-using friends is a person’s network exposure to substance use A person who links two or more otherwise disconnected groups is a bridge A set of people who are connected to one another at a rate greater than to others in the network. There are dozens of ways to measure groups A person not connected to anyone in the network The volume of connections in the network. Sparse networks are those with low density Refers to whether connections between two people imply a third connection The number of steps (links between two people) connecting two people in the network. For example, the friend of a friend is two steps away A measure of how many of one’s friends are connected The degree ties in a network are clustered around one or a small group of people. For example, in a highly centralized network most people nominate the same person or the same few people as friends The number of ties in a network as a proportion of those possible. Dense networks have many ties (links) while sparse ones have few The degree to which a person lies on the shortest paths connecting others in the network The inverse of the average distance to others in the network The reverse distance to others in the network
ORDER
1692
REPRINTS
Valente, Gallaher, and Mouttapa
Network Measures In addition to measures of peer influence and network position, network analysis can also be used to create network measures that characterize a person’s interpersonal environment. For example, reciprocity is the degree that the people a person names also name them. A high degree of reciprocity in a friendship network indicates a shared understanding of who likes whom and, hence, a flat structure. Low levels of reciprocity, an asymmetric network, could possibly indicate a hierarchical structure. Person 1 has three of his/her nominations reciprocated (Persons 2, 3, and 5). A second measure is personal network density, the degree a person’s ties are connected to one another. A dense personal network indicates that a person’s friends know and like one another. Dense personal networks can reinforce behavioral norms since once a behavior is accepted by a majority of the group it is reinforced. Person 1’s personal network density is 25% (three links, Persons 2 to 3, 4 to 5, and 5 to 4; of twelve possible; see Fig. 1 for clarification).
IMPLICATIONS FOR ADOLESCENT SUBSTANCE USE Network level measures (centralization, density, transitivity) may also directly or indirectly influence substance use. Centralization is the degree network ties are concentrated on one or few people. More centralized networks may increase the effects of opinion leader behavior or more dense networks may increase the effects of prevalence overestimates. Density is the number of ties in a network divided by the number possible. Dense networks may accelerate behavioral diffusion since there might be more peer modeling and peer influence. Transitivity refers to whether connections between two people imply a third. For example, Person 1 nominated 4 and Person 4 nominated 5, so we expect, in a transitive network, that Person 5 will nominate 1. It is unclear how transitivity may affect behavioral diffusion, but possibly its effects are similar to those of density. Youth networks exist at school, at home (including the neighborhood), and at other public places (church, religious institutions, clubs, the park). An individual’s network connections may be different in different contexts and may possibly influence drug use as well as nonuse differentially. Sociometric methods, which require assessments of (nearly) all members of a network, are not feasible in all contexts; however, network analysis of high-risk contexts may be particularly
ORDER
REPRINTS
Social Networks
1693
informative. Sussman and others (1998) found that the single most frequent location of drug use among high-risk adolescents was in the youth’s bedroom, with friends. Neighborhood networks may be measurable and very informative regarding the etiology and topography of teen drug use (Mason et al., in press). In his examination of multiple contexts, Cook and others (Cook et al., 2002) found that contexts tended to be correlated in terms of their ability to cause or support a range of adolescent behaviors. Thus, context itself may act as a mediator or moderator of drug use behavior. Research on the consistency of an individual’s network position across contexts is an unexplored area of research, although it is likely that network positions tend to become more stable across contexts with age. Similarly, multilevel social network modeling is an unexplored, but potentially fruitful method of understanding the determinants of adolescent drug use. Many studies have demonstrated that social network variables and measures influence substance use (Neiagus et al., 2001).
REVIEW OF SOCIAL NETWORKS AND ADOLESCENT SUBSTANCE USE This section briefly reviews previous studies that have utilized social network analysis to explain substance use among adolescents. Theories that are pertinent to these studies include a simple ‘‘birds of a feather flock together’’ notion, differential association theory, and the theory of reasoned action.
The Influence of Peers’ Behaviors: ‘‘Birds of a Feather Flock Together’’ This simple conception is not much more than an observation that youth tend to cluster together based on shared activities. Indeed, most substance-use researchers would agree that people who misuse substances are often surrounded by friends, family members, and associates that also misuse substances or tacitly approve of doing so. There are numerous empirical examples. Studies have shown that an individual adolescent’s substance use is associated with, and perhaps causally linked with, substance use by their friends. In the case of smoking, for example, having a best friend who smokes (Urberg et al., 1997) and having friends who smoke (Alexander et al., 2001; Aloise-Young et al., 1994; Bauman and Ennett, 1994; Botvin et al., 1993; Flay et al., 1994; Unger and
ORDER
1694
REPRINTS
Valente, Gallaher, and Mouttapa
Chen, 1999; Urberg et al., 1997) are associated with smoking. Unger and Chen (1999) provide longitudinal evidence that suggests that individuals who have friends who smoke are more likely to start smoking themselves. There is cross-sectional evidence which suggests that among high school adolescents, peer involvement in illicit drug use (Rai et al., 2003; Windle, 2000) and alcohol use (Windle, 2000) are associated with one’s own involvement in those behaviors. Longitudinal evidence linking peer use and adolescent use also exists (Rice et al., 2003). Other studies have examined the association between the number of friends who use substances and the individual’s substance use (Donato et al., 1994; Meijer et al., 1994; Wang et al., 1997). There is evidence that the number of friends who use illicit drugs (Jenkins and Zunguze, 1998) and smoke cigarettes (Wang et al., 1997) is positively associated with one’s own illicit drug use and smoking, respectively. Alexander and others (2001) found that adolescents with a majority of friends who smoke were almost twice as likely to smoke themselves. Studies have also found a positive association between smoking and the proportion of friends who smoke (Botvin et al., 1993; Urberg et al., 1997). Referring to Fig. 1, these results would suggest that if Persons 2, 3, and 4 misuse substances, then Person 1 would also likely misuse substances.
Social Learning and Differential Association Theories The birds of a feather flock together notion is simplistic. One would want to understand the mechanisms underlying this network clustering. One theory that might be invoked to provide an explanation of this clustering is social learning theory. Social learning theory (Bandura, 1986) posits that involvement in substance use is the result of modeling significant others’ substance use and social reinforcement for initiating substance use. Social learning theory suggests that youth may gain an interest in using drugs merely from watching others apparently receiving rewards of use. This vicarious exposure and reward might consist of contact with the media (movies, magazines), youth one observes from a distance in one’s neighborhood, or one’s close friends. Use is observed or instructed and one might try a drug. Another, slightly different theory is one that suggests that exposure to drug use is a function of differential association. Differential association theory (Sutherland and Cressey, 1974) posits that adolescents learn delinquent behavior such as substance use from close friends and family who also use substances themselves and/or have
ORDER
REPRINTS
Social Networks
1695
favorable attitudes towards substance use. Hence, differential association theory suggests that associations with substance-using friends precedes actual substance use. Youth may not tend to model such risky behaviors from strangers or from impersonal influences, according to a strict interpretation of this model. Ennett and Bauman (1994) provide support for this assumption. They found that membership in a friendship group was associated with smoking. Nonsmokers who associated with people in cliques comprised of smokers were more likely to become smokers than were those who associated with people in nonsmoking cliques. Friedman and others (1997) showed that being connected to a large group of people who use drugs is associated with drug use. Sieving and others (2000) examined adolescent friendships longitudinally for three years and demonstrated that, over time, higher levels of friends’ drug use led to increased alcohol use.
Friendship Grouping—Antecedent or Consequence of Substance Use? Both differential association theory and social learning theory assume that adolescents use substances because they are influenced by their peers’ substance-using behaviors. However, the opposite may be true; students may select peers as friends based on similar patterns of delinquent behavior. Donohew, Clayton, Skinner, and Colon (1999) posit that individuals who are high on sensation seeking tend to select friends who are also high on sensation seeking, and are more likely to experiment with alcohol, marijuana, and other substances. Consistent with this assertion, Pearson and West (2003) used Markov models to show that people who became substance users transitioned from belonging to nonrisk-taking groups to risktaking-groups. Analyses by Ennett and Baumann (1994) were conducted to determine whether the homogeneity of smoking behavior within groups was caused by social influence or friendship selection based on smoking behavior. This study was motivated, in part, by the need to understand how much of the correlation between adolescent smoking and that of their peers is due to influence and how much due to selection. They conclude ‘‘Although our findings contradict the popular wisdom that peer group influence is largely responsible for adolescent smoking, they substantiate previous research that found that both influence and selection processes contribute to smoking homogeneity among peers’’ (Ennett and Bauman, 1994). Kandel’s (1985) longitudinal study of high school students also found that models that
ORDER
1696
REPRINTS
Valente, Gallaher, and Mouttapa
included both selection and peer influence explained initiation into marijuana use more fully than either factor alone. Other studies also have attempted to disentangle influence from selection (Engels et al., 1997; Fisher and Bauman, 1988). Fisher and Bauman (1988) looked at stable dyadic friendships and showed that the two became similar in their smoking behavior, suggesting influence. On the other hand, dynamic friendship dyads also showed similarity, indicating an effect of selection. Engels and others (1997) also found support for both influence and selection. In a more recent study, Urberg and others (1997) include assessments of the peer group behavior and friend’s smoking reports. They found that ‘‘ . . . the amount of influence over the school year was very modest in magnitude and came from the closest friend for initiation of use’’ (Urberg et al., 1997). A recent study conducted in England by Michell and Amos (1997) showed that girls who belonged to groups where smoking was common were more likely to smoke. Indeed, the authors showed that girls began smoking in order to enhance their prestige in the community. Many girls initiated smoking early in order to demonstrate their status as being more mature. Referring to Fig. 1, these results would indicate that Persons 1 through 10 are more likely to smoke if smoking is prevalent and/or accepted in this group. In sum, both influence and selection are responsible for the similarity of smoking behavior among friends, and each suggests different causal mechanisms for how peers contribute to smoking onset and maintenance.
Perceptions of Peers: Theory of Reasoned Action The Theory of Reasoned Action in its simplest form posits that one’s perceptions of (peer) social norms, one’s willingness to comply with those norms, and one’s expectations regarding the cost and benefits of engaging in the behavior, will influence one’s own intentions to act. These intentions often lead to behaviors (Fishbein and Ajzen, 1981). This constellation of normative beliefs is strongly influenced by interaction with peers. Like social learning theory, norms are learned from observing peers’ behaviors, and/or hearing what their friends tell them. In some cases, the pervasive influence of peers leads to gross inaccuracies in one’s social perceptions of what is normative for different reference groups. For example, a person whose friends use certain substances may believe it is normative for everyone to use these substances even though only a small percentage do.
ORDER
REPRINTS
Social Networks
1697
Peer Norms Theory of reasoned action states that behavior is influenced in part by perceived peer norms (Fishbein and Ajzen, 1981). One finding from research on adolescents has been that youth often misuse substances because they incorrectly believe that it is normative to use them. Adolescents have vastly overestimated the prevalence of smoking, drinking, and drug consumption in their schools. For example, Sussman, and others (1988) found that regardless of smoking status, eighth and ninth graders overestimated weekly use of cigarettes among youth their age. However, nonsmokers made gross underestimations and regular smokers made overestimations of smoking among their age group. This finding has been used in substance use prevention programs to ‘‘re-norm’’ these perceptions. In their intervention study, MacKinnon and others (1991) found that changes in perceptions of friends’ tolerance of drug use was the most substantial mediator of program effects on drug use. In their review, McCaul and Glasgow (1985) posit that smoking prevention programs should communicate social consequences information to peer leaders so they can change subjective norms and reduce the smoking behavior of others. A research question one might ponder is whether youth overstate drug use in general, or whether youth base their general inaccurate judgments of use on accurate perceptions of use by significant others. If youth overestimate drug use in general, then one might confront youth with actual use rates among persons they know or others in their physical environment to instruct and correct the overestimate (and hence, reduce perceptions that one should use drugs because everybody else does). However, if overestimates of friends’ drug use are accurate, one would need to expose these youth to wider social use norms and perhaps counteract deviant influences among friends. Iannotti and Bush (1992) asked students to name their three closest friends and to state whether they thought each smoked, drank alcohol, or used marijuana. They found that respondents’ reports of their friends use did not correlate well with those friends’ self-reports. Iannotti and Bush (1992) found that perceptions of friends’ use was more highly associated with self-reports of use. Similarly, Rice and others (2003) found that high school student responses to a general question regarding friends’ use did not correlate with the self-reports of their friends. Perceptions of friends’ use may reflect projection and implicit cognitions (Stacy et al., in this issue) rather than environmental influence. This distortion can be corrected by a general overestimates correction followed by confrontation with positive,
ORDER
1698
REPRINTS
Valente, Gallaher, and Mouttapa
nonusing social influence from friends and other peers (Sussman et al., 1988). In a related study, Valente and others (1997) found that women in voluntary organizations in Cameroon misjudged their friends’ contraceptive use. They found, however, that perceived friends’ use was associated with one’s own use, regardless of those perceptions’ accuracy. There is a need to determine factors that are associated with discrepancies between peer reports and self-reports of health-risk behaviors. One’s social networks, thus, may be created in part based on one’s misperceptions of one’s clique members. Alternatively, one may also share prosocial activities with one’s clique members that might be identified through assessment of multiple measures that might underlie clustering patterns.
Perceived Social Consequences There is also a need to distinguish between peer use and perceptions of peer approval of use. Many adolescents believe that desirable social consequences, such as peer acceptance or peer support, will occur as a result of using substances, and are, therefore, motivated to initiate substance use. For example, Jenkins (2001) found that among nonsubstance-using high school students, peer pressure was the most frequently cited reason why refusing beer, marijuana, and drug use offers was difficult. Among continuation high school students, Sussman, and others (1995) also found that perceived peer pressure was a reason why at-risk students use drugs other than tobacco. Furthermore, other studies have shown that among high school boys, drug use (Luthar and D’Avanzo, 1999) and cigarette smoking (Alexander et al., 1999; Vega et al., 1996) are associated with peer acceptance of such use.
Changes in Peer Networks as a Function of Age It should be noted that the effects of peers’ behaviors, perceived peer norms, and perceived social consequences may differ among adolescents of different ages due to changes in the structure and importance of peer relationships during adolescence. In a longitudinal study, Feiring and Lewis (1991) found that peer networks were larger by age 13, and the number of same-sex friends increased. By high school age, youth spend more time away from adults and their lives become less dominated by social interactions with small groups of same-sexed peer groups.
ORDER
REPRINTS
Social Networks
1699
Concomitantly, they become exposed to a wider range of unsupervised social gatherings, which consist more of interactions with crowds, more dyadic relationships such as dating, and more weak ties (liaisons) than in earlier years (Dunphy, 1963; Gavin and Furman, 1989; Shrum and Cheek, 1987). Of course, social-situational factors, including same-sex peer group relationships, still remain quite important predictors of tobacco use across both junior high school and high school (Sussman et al., 1995). However, other findings suggest that the influence of specific friends may become attenuated during adolescence, due the development of one’s self-concept. For example, Clark-Lempers, Lempers, and Ho (1991) suggest that the importance of various people, including one’s same-sex best friend, decreases from early to late adolescence. Hence, intervention studies on adolescents should also target individual characteristics such as motivation to resist substance use and social skills to resist substance use offers [see Sussman et al. (in this issue)]. The next section describes how social network analysis can be used to design or augment substanceuse prevention and treatment interventions.
PREVENTION INTERVENTIONS This section describes how previous studies have used the social networks approach to implement substance-use prevention and intervention programs. Early research on the diffusion of innovations showed that opinion leaders can be effective health promoters. Rogers and Kincaid (1981) studied the adoption of family planning by rural Korean women in the 1960s. They discovered that the opinion leaders were early adopters of specific contraceptive methods, while the rest of the village would later adopt the same contraceptive method. Latkin (1998) recruited street opinion leaders to communicate safe injecting practices and showed that these opinion leaders adopted the safe injecting messages themselves and effectively communicated it to others. Opinion leaders tend to be similar in most respects to the people they lead. For example, Moore and others (2004) recruited opinion leaders in substance-user treatment facilities to train other counselors in the adoption of a new treatment protocol. They found that opinion leaders were similar to other counselors in terms of age, education, and experience, but were more knowledgeable about the treatment option being adopted (‘‘dual diagnosis’’ in this case). Broadhead and others (1998) developed a network method for recruiting clients into treatment facilities to be educated on safe injection
ORDER
1700
REPRINTS
Valente, Gallaher, and Mouttapa
practices. In their system, existing clients are given vouchers to go out and recruit new clients to come to the clinic, and these new clients are in turn used to recruit new clients from their network. In this manner the pool of clients grows through naturally occurring networks that are larger and more diverse than could be obtained with traditional outreach. In school-based studies, there has been a long tradition of using peer leaders to assist in program delivery. Peer leaders are usually chosen by asking students to write the names of other students who they consider to be good leaders (Johnson et al., 1986; Perry et al., 2002). The teacher collects the data, and selects as peer leaders those students who received the most nominations. Valente and others (1999) expanded this methodology to allow students to be assigned to a leader they nominated, or were closest to sociometrically. Their study showed that use of a sociometric method of selection was more effective than leaders popularly chosen but placed in groups defined randomly. These studies show that network data can be used to improve program implementation, yet there are many other possibilities. For example, network data can be used to define subsets, cliques, or groups in a network that can be targeted for intervention. Often getting a group to change behavior can be easier than trying to change people individually, since the group can reinforce the message and provide social support for maintaining the new behavior (e.g., abstinence). The optimality algorithm used by Valente and others (2004) can be improved by borrowing ideas from location science (the branch of operations research devoted to understanding optimal locations for emergency services, warehouses, etc.). Social networks, however, are not transportation networks and so algorithms will not be directly transferable. Still, considerable research can and should be conducted to determine optimal group formation techniques in the implementation of health promotion and substance-use prevention programming. These experiences have led to the creation of a substance-use prevention program using social network analysis to structure work groups. The central research question is whether social network analysis can be used to tailor an existing evidence-based prevention program to increase its effectiveness. The Transdisciplinary Drug Abuse Prevention Research Center (TPRC) at USC has launched a program which adapts Toward No Drug Abuse (TND), an evidence-based substance abuse prevention program (Sussman, in this issue) using social network analysis. The adaptation will include more group activities and social network analysis will be used to define these groups. Outcomes from this
ORDER
REPRINTS
Social Networks
1701
study will affect how programs should be created and, hence, disseminated (Pentz et al., in this issue).
POTENTIAL USES OF NETWORK DATA FOR MONITORING AND EVALUATION Few studies have been conducted that make explicit use of network data for program monitoring and evaluation because social network analysis tools have been readily available only recently. This section summarizes what has been done so far and what could be done in future studies. Social network analysis reminds us that relations matter, and positions derived from those relations matter. As such, monitoring and evaluation plans may be augmented by assessing who is affected by an intervention. For example, a school-based program found to have minimal impact may still be judged a success if key opinion leaders were influenced. In such a scenario, the absolute (quantitative) change measured in the evaluation may be modest, the qualitative change in terms subsequent affects community norms through these leaders, however, could be substantial. A second use of network analysis for evaluation is to show that a program changed the dynamics of peer influence. For example, in an earlier study, it was shown that a media campaign was associated with changed perceptions of network members’ use of contraception (Valente and Saba, 2001). Similarly, substance-use prevention programs can collect network data on the perceived use of substances by specific peers and monitor whether these perceptions change. Finally, network analysis can be used to disseminate evidence-based programs. For example, Lomas and others (1991) used network analysis to identify peer opinion leaders to institute a practice change in selected health care settings. Network analysis can be used to target dissemination activities so that evidence-based and proven programs can be implemented more readily. Network analysis provides a technique to map specifically who has adopted evidence-based programs and where they are in the network. This then provides information on where and to whom to devote additional matched resources to accelerate this diffusion. The network map provides important monitoring information indicating how well the practice is spreading (Stoebenau and Valente, 2003). Change agents can be recruited to accelerate diffusion in social network groups lacking existing adopters.
ORDER
1702
REPRINTS
Valente, Gallaher, and Mouttapa
Transdisciplinarity Network analysis is maturing as a field of inquiry and is now making substantial contributions to research on health issues. The computational tools for managing relational data have matured to the point where a community of users (the International Network for Social Network Analysis) has formed to use network analysis to research a broad range of topics. Empirical evidence indicates that network concepts and variables are important explanations for human behavior. Network analysis offers a theoretically rich set of explanatory variables and procedures useful for understanding the etiology of substance use and means to prevent or treat it. For example, rather than seeing youth as solely a product of their familial upbringing, network analysis shows that there is a complex interaction of adult and peer role models that influence behavior. The field of social network analysis seemed to foment independently in Anthropology, Sociology, and Psychology (Freeman, 2000) and is now inherently transdisciplinary. Many network analysts have some mathematical training necessary for understanding the concepts and programming the measures. This mathematical training is not always present with most social scientists and it often limits network application to those with this interest or training. Network analysis can be described as mathematical ethnography, providing a deep description of a community, in mathematically tractable ways. Recent developments in graph theory and computer science have aided the development of the network field considerably by providing tools needed to display and analyze large networks. Network analysis, then, has no disciplinary home, and relies on scanning journals from numerous fields to find application of the method. Interestingly, network analysis is probably the best tool for studying transdisciplinarity (Fuqua et al., in this issue). The collaboration behavior and citation behavior of individuals, groups, and fields can be studied to measure the degree they interact with one another. The tradition in the social sciences has been to collect data on attributes of individuals and study associations between those attributes and outcomes. Statistical programs such as SAS, SPSS, and STATA, were designed to treat individuals as atoms moving independently in their universe. Social network analysis turns these notions on their head. It assumes individuals are connected in molecular and organic structures whose parameters have implications for human behavior. How an individual attaches to that structure and how their actions may change that structure provides a new means to understand why certain individuals misuse substances and ways to develop programs to prevent and treat it.
ORDER
REPRINTS
Social Networks
1703
CONCLUSIONS Network analysis has its limitations. The sociometric technique is amenable to studies with populations in settings with natural boundaries (organizations, schools, small communities) and lacks some of the strengths of random sample designs. It also tends to quantify, reducing to a number, the complexity of interpersonal communication and relationships (not all ties are the same). Finally, there are issues regarding appropriate statistical tests for nonindependent data as often used in network studies. These limitations aside, network analysis offers a promising new area of the causes, consequences, and potential treatments for substance-use behavior. The contributions of network analysis have been delayed for decades because social science disciplines focused on individual attributes [Socio– Economic Status (SES), ethnicity, gender] as key determinants of behavior. While these factors are certainly important, increasingly we recognize that people’s social circles and their social capital (or lack thereof) are also key determinants of behavior. The social network approach provides a rigorous yet versatile approach to studying these peer influences. This article reviewed studies that show that individual substance use and misuse is strongly associated with, and perhaps influenced by, use in one’s social network. By taking a social network approach, several research questions on the role of social and peer influences on substance use are suggested. Further, it was shown that positions in a social network (popularity, liaisons, isolates) are also associated with substance use, and that they may interact with group and cultural level properties (Unger et al., in this issue). Although it was omitted from the article, we should note that much of the effects described in this article would also apply to nonuse or abstinence. Finally, this article suggests that networklevel properties such as density or centralization might influence individual use within those networks. The social network field has begun to mature as evidenced by this rich body of research findings and its expanded application to intervention design, program monitoring, and evaluation, and use as a means to disseminate evidence-based programs. The maturation, however, is not without costs. Increasingly it is difficult for those outside the field to learn the specialized jargon, software, and culture of social network analysis. Nonetheless, the field is open to new discovery and approaches. Social network analysts are united by a common method (the focus on relationships rather than attributes) that transcends disciplinary boundaries. The goal of this transdisciplinary team is to
ORDER
REPRINTS
1704
Valente, Gallaher, and Mouttapa
blend these perspectives, and develop a more complete understanding of the factors that influence nonsubstance use/abstinence, substance use, and ways to deter it.
GLOSSARY Differential association theory: The theory posits that adolescents learn delinquent behavior such as substance use from close friends and family who also use substances themselves and/or have favorable attitudes towards substance use; associations with substance-using friends precede actual substance use. Personal network density: The degree a person’s ties are connected to one another. Social network analysis: A set of theories, methods, and techniques used to understand social relationships and how these relationships might influence individual and group behavior. Social learning theory: This theory posits that involvement in substance use is the result of modeling significant others’ substance use and social reinforcement for initiating substance use. This theory suggests that youth may gain an interest in using drugs merely from watching others apparently receiving rewards of use. Theory of reasoned action: This theory posits that one’s perceptions of (peer) social norms, one’s willingness to comply with those norms, and one’s expectations regarding the cost and benefits of engaging in the behavior, will influence one’s own intentions to act. These intentions often lead to behaviors.
ACKNOWLEDGMENTS Support for this project was provided by NIDA grant P50-DA16094.
ORDER
REPRINTS
Social Networks
1705
THE AUTHORS Thomas W. Valente is an Associate Professor in the Department of Preventive Medicine, Keck School of Medicine, and Director of the Master of Public Health Program at the University of Southern California. He received his Ph.D. in Communication from the Annenberg School for Communication at USC in 1991 and then spent nine years at the Johns Hopkins University Bloomberg School of Public Health. He is author of Evaluating Health Promotion Programs (2002, Oxford University Press); Network Models of the Diffusion of Innovations (1995, Hampton Press); and over 50 articles and chapters on public health, social networks, behavior change, and program evaluation. Valente uses social network analysis, health communication, and mathematical models to implement and evaluate health promotion programs, primarily aimed at preventing substance abuse, tobacco use, unwanted pregnancies, and STD/HIV infections. Peggy E. Gallaher, Ph.D., is a Research Associate in Preventive Medicine at the University of Southern California’s Keck School of Medicine. Her research interests include adapting psychosocial measures written for adults for use by children, defining and assessing acculturation and ethnic identity, and establishing the cultural equivalence of psychological tests. She received her doctorate in Psychology from the University of Texas at Austin in 1988 and a Master’s degree in Biostatistics from Columbia University in 1994.
ORDER
1706
REPRINTS
Valente, Gallaher, and Mouttapa
Michele Mouttapa, Ph.D., received her Ph.D. in Health Behavior Research at the Institute for Health Promotion and Disease Prevention Research at the University of Southern California. Her research interests include examining the effects of sociocultural influences on adolescent smoking. In addition, she is examining the influence of individualism and collectivism on numerous mediators related to health outcomes, including peer susceptibility and family values. Mouttapa received her M.A. in Psychology (with an emphasis in social-cognitive psychology) from California State University, Fullerton in 1999.
REFERENCES Alexander, C. S., Allen, P., Crawford, M. A., McCormick, L. K. (1999). Taking a first puff: cigarette smoking experiences among ethnically diverse adolescents. Ethnicity and Health 4:245–257. Alexander, C., Piazza, M., Mekos, D., Valente, T. W. (2001). Peer networks and adolescent cigarette smoking: an analysis of the national longitudinal study of adolescent health. Journal of Adolescent Health 29:22–30. Aloise-Young, P., Graham, J. W., Hansen, W. B. (1994). Peer influence on smoking initiation during early adolescence: a comparison of group members and group outsiders. Journal of Applied Psychology 79:281–287. Analytic Technologies (2000). UCINET Version 5. Boston, MA. Bandura, A. (1986). Social Foundations of Thought and Action: A Social Cognitive Theory. New Jersey: Prentice Hall. Bauman, K. E., Ennett, S. T. (1994). Peer influence on adolescent drug use. American Psychologist 49(9):820–822. Botvin, G. J., Baker, E., Botvin, E. M., Dusenbury, L., Cardwell, J., Diaz, T. (1993). Factors promoting cigarette smoking among black youth: a causal modeling approach. Addictive Behaviors 18:397–405. Broadhead, R. S., Hechathorn, D. D., Weakliem, D. L., Anthony, D. L., Madray, H., Mills, R. J., Hughes, J. (1998). Harnessing peer
ORDER
Social Networks
REPRINTS
1707
networks as an instrument for AIDS prevention: results from a peer-driven intervention. Public Health Reports 113(S1):42–57. Burt, R. S. (1980). Models of network structure. Annual Review of Sociology 6:79–141. Clark-Lempers, D. S., Lempers, J. D., Ho, C. (1991). Early, middle, and late adolescents’ perceptions of their relationships with significant others. Journal of Adolescent Research 6:296–315. Cook, T., Herman, M., Phillips, M., Setterston, R. (2002). How neighborhoods, families, peer groups and schools jointly affect changes in early adolescent development. Child Development 73:1283–1309. Costenbader, E., Valente, T. W. (2003). The stability of centrality when networks are sampled. Social Networks 25:283–307. Donato, F., Monarca, S., Chiesa, R., Feretti, D., et al. (1994). Smoking among high school students in 10 Italian towns: patterns and covariates. International Journal of the Addictions 29:1537–1557. Donohew, L., Clayton, R. R., Skinner, W. F., Colon, S. (1999). Peer networks and sensation seeking: some implications for primary socialization theory. Substance Use and Misuse 34:1013–1023. Dunphy, D. C. (1963). The social structure of urban adolescent peer groups. Sociometry 26:230–246. Engels, R. C. M. E., Knibbe, R. A., Drop, M. J., de Haan, Y. T. (1997). Homogeneity of cigarette smoking within peer groups: influence or selection? Health Education Behavior 24:801–811. Ennett, S. T., Bauman, K. E. (1993). Peer group structure and adolescent cigarette smoking: a social network analysis. Journal of Health and Social Behavior 34:226–236. Ennett, S. T., Bauman, K. E. (1994). The contribution of influence and seletion to adolescent peer group homogeneity: the case of adolescent cigarette smoking. Journal of Personality and Social Psychology 67:653–663. Feiring, C., Lewis, M. (1991). The transition from middle-childhood to early adolescence: sex differences in the social network and perceived self-competence. Sex Roles 24:489–509. Fishbein, M., Ajzen, I. (1981). On construct validity: a critique of miniard and cohen’s paper. Journal of Experimental Social Psychology 17:340–350. Fisher, L. A., Bauman, K. E. (1988). Influence and selection in the friend-adolescent relationship: findings from studies of adolescent smoking and drinking. Journal of Applied Social Psychology 18:289–314.
ORDER
1708
REPRINTS
Valente, Gallaher, and Mouttapa
Flay, B. R., Hu, F. B., Siddiqui, O., Day, L. E., et al. (1994). Differential influence of parental smoking and friends’ smoking on adolescent initiation and escalation of smoking. Journal of Health and Social Behavior 35:248–265. Freeman, L. (1979). Centrality in social networks: conceptual clarification. Social Networks 1:215–239. Freeman, L. (2000). Visualizing social networks. Journal of Social Structure 1, February 4. www.cmu.edu/joss. Friedman, S. R., Neaius, A., Jose, B., et al. (1997). Sociometric risk networks and risk for HIV infection. American Journal of Public Health 87:1289–1296. Fuqua, J., Stokols, D., Gress, J., Phillips, K., Harvey, Y. (2004). Transdisciplinary collaboration as a basis for enhancing the science and prevention of substance use and misuse. Substance Use and Misuse. Gavin, L. A., Furman, W. (1989). Age differences in adolescents’ perceptions of their peer groups. Developmental Psychology 25:827–834. Gorman, D. M., Gruenewald, P. J., Mezic, I., Waller, L. A., CastilloChavez, C., Bradley, E., Mezic, J. (2004). Implications of systems dynamic models and control theory for environmental approaches to the prevention of alcohol and other drug-related problems. Substance Use and Misuse. Granovetter, M. (1973). The strength of weak ties. American Journal of Sociology 78:1360–1380. Iannotti, R. J., Bush, P. J. (1992). Perceived vs. actual friends’ use of alcohol, cigarettes, marijuana, and cocaine: which has the most influence? Journal of Youth and Adolescence 21:375–389. INSNA (2003). International Network for Social Network Analysis. http://www.sfu.ca/insna/ Jenkins, J. E. (2001). Rural adolescent perceptions of alcohol and other drug resistance. Child Study Journal 31:211–224. Jenkins, J. E., Zunguze, S. T. (1998). The relationship of family structure to adolescent drug use, peer affiliation, and perception of peer acceptance of drug use. Adolescence 33:811–822. Johnson, C. A., Hansen, W. B., Collins, L. M., Graham, J. W. (1986). High school smoking prevention: results of a three-year longitudinal study. Journal of Behavioral Medicine 9:439–452. Kandel, D. (1985). On processes of peer influences in adolescent drug use: a developmental perspective. Advances in Alcohol and Substance Abuse 4:139–163.
ORDER
Social Networks
REPRINTS
1709
Latkin, C. (1998). Outreach in natural setting: the use of peer leaders for HIV prevention among injecting drug users’ networks. Public Health Reports 113(S1):151–159. Lomas, J., Enkin, M., Anderson, G. M., Hanna, W. J., Vayda, E., Singer, J. (1991). Opinion leaders vs. audit feedback to implement practice guidelines: delivery after previous cesarean section. Journal of American Medical Association 265:2202–2207. Luthar, S. S., D’Avanzo, K. (1999). Contextual factors in substance use: a study of suburban and inner-city adolescents. Development and Psychopathology 11:845–867. MacKinnon, D. P., Johnson, C. A., Pentz, M. A., Dwyer, J. H., Hansen, W. B., Flay, B. R., Wang, E. Y. (1991). Mediating effects in a school-based drug prevention program: first-year effects of the midwestern prevention project. Health Psychology 10:164–172. Marsden, P. V. (1990). Network data and measurement. Annual Review of Sociology 16:435–463. Mason, M. J., Cheung, I., Walker, L. (2004). Substance use, social networks and the geography of risk and protection of urban adolescents. Substance Use and Misuse. McCaul, K. D., Glasgow, R. E. (1985). Preventing adolescent smoking: what have we learned about treatment construct validity? Health Psychology 4:361–387. Meijer, R. R., Muijtjens, A.M.M., van der Vleuten, C.P.M. (1994). Nonparametric person-fit research: some theoretical issues and an empirical example. Applied Measurement in Education 9:77–89. Michell, L., Amos, A. (1997). Girls, pecking order and smoking. Social Science Medicine 44:1861–1869. Monge, P. R., Contractor, N. S. (2002). Theories of Communication Network. New York: Oxford University Press. Moore, K. A., Peters, R. H., Hills, H. A., LeVasseur, J. B., Rich, A. R., Hunt, W. M., Young, M. S., Valente, T. W. (in press). Characteristics of opinion leaders in substance abuse treatment agencies. Journal of Drug and Alcohol Dependence. Neiagus, A., Friedman, S. R., Kottiri, B. J., Des Jarlais, D. C. (2001). HIV risk networks and HIV transmission among injecting drug users. Evaluation and Program Planning 24:221–226. Pearson, M., West, P. (in press). Drifting smoke rings: social network analysis and markov processes in a longitudinal study of friendship groups and risk-taking. Connections, 26. Pentz, M. A., Mares, D., Schinke, S., Rohrbach, L. (2004). Policital science, public policy, and drug abuse prevention. Substance Use and Misuse.
ORDER
1710
REPRINTS
Valente, Gallaher, and Mouttapa
Perry, C., Williams, C. L., Komro, K. A., Veblen-Mortenson, S., et al. (2002). Project northland: long-term outcomes of community action to reduce adolescent alcohol use. Health Education Research 17:117–132. Rai, A. A., Stanton, B., Wu, Y., Li, X., Galbraith, J., Cottrell, L., Pack, R., Harris, C., D’Alessandri, D., Burns, J. (2003). Relative influences of perceived parental monitoring and perceived peer involvement on adolescent risk behaviors: an analysis of six cross-sectional data sets. Journal of Adolescent Health 33:108–118. Rice, R. E., Donohew, L., Clayton, R. (in press). Peer network, sensation seeking, and drug use among junior and senior high school students. Connections 26(2):32–58. Rogers, E. M. (1995). Diffusion of Innovations. 4th ed. New York: The Free Press. Rogers, E. M., Kincaid, D. L. (1981). Communication Networks: A New Paradigm for Research. New York: The Free Press. Scott, J. (2000). Social Network Analysis: A Handbook. 2nd ed. Newberry Park, CA: Sage. Shrum, W., Cheek, N. H. (1987). Social structure during the school years: onset of the degrouping process. American Sociological Review 52:218–223. Sieving, R., Perry, C., Williams, C. (2000). Do friendships change behaviors, or do behaviors change friendships? examining paths of influence in young adolescents’ alcohol use. Journal of Adolescent Health 26:27–35. Stacy, A. W., Ames, S., Knowlton, B. (2004). Neurally plausible distinction in cognition: implications for drug abuse prevention. Substance Use and Misuse. Stoebenau, K., Valente, T. W. (2003). The role of network analysis in community-based program evaluation: a case study from highland madagascar. International Family Planning Perspectives. Sussman, S. (2004). A drug abuse theoretical integration: a transdisciplinary speculation. Substance Use and Misuse. Sussman, S., Dent, C. W., Mestel-Rauch, J., Johnson, C. A., Hansen, W. B., Flay, B. R. (1988). Adolescent nonsmokers, triers, and regular smokers’ estimates of cigarette smoking prevalence: when do overestimations occur and by whom? Journal of Applied Social Psychology 18:537–551. Sussman, S., Stacy, A. W., Dent, C. W., Simon, T. R., Galaif, E. R., Moss, M. A., Craig, S., Johnson, C. A. (1995). Continuation high schools: youth at risk for drug abuse. Journal of Drug Education 25:191–209.
ORDER
Social Networks
REPRINTS
1711
Sussman, S., Stacy, A., Ames, S., Freedman, L. (1998). Self-reported high-risk locations of adolescent drug use. Addictive Behaviors 23:405–411. Sutherland, E. H., Cressey, D. R. (1974). Criminology. 9th ed. Philadelphia, PA: Lippincott. Terre, L., Drabman, R., Meydrech, E., Hsu, H. (1992). Relationship between peer status and health behaviors. Adolescence 27:595–602. Unger, J. B., Chen, X. (1999). The role of social networks and media receptivity in predicting age of smoking initiation: a proportional hazards model of risk and protective factors. Addictive Behaviors 24:371–381. Unger, J. B., Baezconde-Garbanati, L., Shakib, S., Palmer, P. H., Nezami, E., Mora, J. (2004). What are the implications of structural-cultural theory for drug abuse prevention. Substance Use and Misuse. Urberg, K. A., Degirmencioglu, S. M., Pilgrim, C. (1997). Close friend and group influence on adolescent cigarette smoking and alcohol use. Developmental Psychology 33:834–844. Valente, T. W. (1995). Network Models of the Diffusion of Innovations. Cresskill, NJ: Hampton Press. Valente, T. W., Davis, R. L. (1999). Accelerating the diffusion of innovations using opinion leaders. The Annals of the American Academy of the Political and Social Sciences 566:55–67. Valente, T. W., Hoffman, B. R., Ritt-Olson, A., Lichtman, K., Johnson, C. A. (2003). The effects of a social network method for group assignment strategies on peer led tobacco prevention programs in schools. American Journal of Public Health. Valente, T. W., Jato, M. N., Van der Straten, A., Tsitol, L. M. (1997). Social network influences on contraceptive use among cameroonian women in voluntary associations. Social Science and Medicine 45:677–687. Valente, T. W., Saba, W. (2001). Campaign recognition and interpersonal communication as factors in contraceptive use in bolivia. Journal of Health Communication 6:1–20. Valente, T. W., Vlahov, D. (2001). Selective risk taking among needle exchange participants in baltimore: implications for supplemental interventions. American Journal of Public Health 91:406–411. Vega, W. A., Apospori, E., Gil, A. G., Zimmerman, R.S., et al. (1996). A replication and elaboration of the self esteem-enhancement model. Psychiatry: Interpersonal Biological Processes 59:128–144. Wang, M. Q., Fitzhugh, E. C., Eddy, J. M., Fu, Q., et al. (1997). Social influences on adolescents’ smoking progress: a longitudinal analysis. American Journal of Health Behavior 21:111–117.
ORDER
1712
REPRINTS
Valente, Gallaher, and Mouttapa
Wasserman, S., Faust, K. (1994). Social Networks Analysis: Methods and Applications. Cambridge, UK: Cambridge University Press. Windle, M. (2000). Parental, sibling, and peer influences on adolescent substance use and alcohol problems. Applied Developmental Science 4:98–110.
Request Permission or Order Reprints Instantly! Interested in copying and sharing this article? In most cases, U.S. Copyright Law requires that you get permission from the article’s rightsholder before using copyrighted content. All information and materials found in this article, including but not limited to text, trademarks, patents, logos, graphics and images (the "Materials"), are the copyrighted works and other forms of intellectual property of Marcel Dekker, Inc., or its licensors. All rights not expressly granted are reserved. Get permission to lawfully reproduce and distribute the Materials or order reprints quickly and painlessly. Simply click on the "Request Permission/ Order Reprints" link below and follow the instructions. Visit the U.S. Copyright Office for information on Fair Use limitations of U.S. copyright law. Please refer to The Association of American Publishers’ (AAP) website for guidelines on Fair Use in the Classroom. The Materials are for your personal use only and cannot be reformatted, reposted, resold or distributed by electronic means or otherwise without permission from Marcel Dekker, Inc. Marcel Dekker, Inc. grants you the limited right to display the Materials only on your personal computer or personal wireless device, and to copy and download single copies of such Materials provided that any copyright, trademark or other notice appearing on such Materials is also retained by, displayed, copied or downloaded as part of the Materials and is not removed or obscured, and provided you do not edit, modify, alter or enhance the Materials. Please refer to our Website User Agreement for more details.
Request Permission/Order Reprints Reprints of this article can also be ordered at http://www.dekker.com/servlet/product/DOI/101081JA200033210
