• unlimited access with print and download
    $ 37 00
  • read full document, no print or download, expires after 72 hours
    $ 4 99
More info
Unlimited access including download and printing, plus availability for reading and annotating in your in your Udini library.
  • Access to this article in your Udini library for 72 hours from purchase.
  • The article will not be available for download or print.
  • Upgrade to the full version of this document at a reduced price.
  • Your trial access payment is credited when purchasing the full version.
Buy
Continue searching

The influence of social factors on drug use among drug-experienced adults

ProQuest Dissertations and Theses, 2009
Dissertation
Author: Amy S Buchanan
Abstract:
Background. Drug use poses a significant public health concern, and the social context of drug use warrants attention. Prior research has shown that there is a relationship between drug use and social network drug use, but theories differ as to which precedes the other. Specifically, proponents of differential association theory posit that the individual's drug use is influenced by the drug use of his or her social network members. Meanwhile, social control theory posits that an individual with a lessened bond to conventional society chooses to use drugs and consequently spends time with others who also use drugs. Interactional and life-course theories propose that the relationship is bidirectional. This theoretical conflict has also been conceptualized under the rubric of social causation/social influence vs. social selection. The present study examined the theories above by modeling bi-directional influences between drug use and network drug use among drug-experienced, following methods that have been developed to simultaneously test social influence and social selection theories. Additional analysis was conducted to determine if change in network drug use behavior is due to social network turnover, consistent with social control theory, or change in the behavior of network members over time. Follow-up models were estimated to determine if drug use by key network members explained later index drug use beyond the total network drug use, consistent with differential association theory. Lastly, the present study took advantage of data from a thorough social network inventory to conduct the above analyses, as well as to describe and report on the make-up of the social networks, social support available from the social network, and changes in the social network over time. Results. At baseline, 41% of all network members identified were current heroin/cocaine users. Social networks were primarily made up of individuals identified as "friends" or nuclear family members. While only 7% of the sample reported being married, 61% reported a main partner. Participants reported high levels of trust overall for their network, and the majority had at least one person who provided or could provide each type of support queried. In a factor analysis the number of supporters with whom the index shared meals and housing loaded onto one factor, while the number of supporters who the index could rely on for lending money, to give up time, for trusting with the index's money, for having fun, for talking about private matters, or for advice loaded onto another factor. The mean size of the social network at baseline was 9.1 individuals. The size of the social networks decreased at assessment 2 to 7.4, but increased at assessment 4 to 8.4, which was statistically greater than the second assessment, but also statistically lower than baseline. A similar pattern of change in the support network and drug network size over the follow-up was observed. In a SEM, bidirectional influences between index drug use and social network drug use were observed. In recursive models it was found that the social network drug use influenced index drug use through long-term processes, while index drug use influenced network drug use through short-term processes. Evidence of relationships in both directions was stronger between times 1 to 2 than times 2 to 4, the later two assessments having a greater time lag. Evidence of a relationship in the direction of index drug use influencing later network drug use was more consistent across assessments than evidence for a relationship in the opposite direction. In Microsocial Analysis it was found that the majority of change in social network drug use behavior was accounted for by turnover in the social network, rather than change in the behavior of individual network members over time. In SEMs that examined key network members, the relationship between key network member drug use and index drug use was null or negative in association, in models that also included total network drug use. All SEMs exhibited good fit. Conclusions. Drug use in the social networks of this sample was high, and drug treatment utilization low. There was a fair degree of overlap of the drug network with the support and sex networks (those individuals with whom the index has sexual relations). However, the sample also reported a substantial degree of social support from their networks. The results of the bidirectional model were consistent with what would be expected under both social control and differential association theories in drug use among drug-experienced adults. Findings of support in both directions also provide some support life-course and Interactional theories. The Microsocial Analysis provided further support for the process suggested by Social Control theory to explain the correlation of an individual's drug use with that of their network. The results of SEMs that analyzed key network members contradicted the premise that the closest network members exert greater influence over drug use behavior. Cumulatively, the findings provide support for a variety of intervention strategies, including engaging supporters in treatment, training participants to conduct peer outreach in order to reduce drug use in the community, and programs such as job training and relationship counseling to increase investment in traditional society.

Table of Contents Chapter 1: Introduction 1 Chapter 2: Background and Significance 4 2.1 Background 4 2.1.1 The Problems of Drug Use and Abuse 4 2.1.2 Drugs of Abuse in Baltimore, MD 5 2.1.3 Theories of Deviant Behavior and Social Relationships 6 2.1.4 Additional Theoretical Considerations for the Relationship of Peers and Drug Use 11 2.1.5 Social Network Methodology 12 2.1.6 Social Support and Drug Use 14 2.1.7 Gaps in the Study of Adult Drug Use and Social Factors 15 2.1.8 Neighborhood Factors and Drug Use 19 2.1.9 The Present Study 19 2.2 Limitations and Off-Setting Strengths 21 2.2.1 Self-Report of Drug Use Data 21 2.2.2 Aim-Specific Limitations and Strengths 22 2.3 Significance 23 2.3.1 Application to Public Mental Health 24 2.4 Generalizability 25 Chapter 3: Methods and Materials 26 3.1 Design 26 ix

3.1.1 Recruitment and Eligibility 26 3.1.2 Waves of Assessment 27 3.2 Data Collection, Entry, and Quality Control 29 3.3 Protection of Confidentiality and Human Subjects Policies 29 3.4 Sample 29 3.5 Study Variables 30 3.5.1 Primary Measures 30 3.6 Analytic Plan 32 3.6.1 Factor Analysis 33 3.6.2 Structural Equation Modeling 33 3.6.3 Recursive Models 35 3.6.4 Microsocial Analysis 36 3.6.5 Overview of Analyses 37 3.7 Attrition and Missing Data 38 Chapter 4: A Description of the Social Networks of a Sample of Drug- Experienced Adults in Low-Income Neighborhoods from the SHIELD Cohort in Baltimore, MD 39 4.1 Introduction 39 4.2 Background 39 4.2.1 Social Networks 39 4.2.2 Social Networks and the Study of Drug Use 40 4.2.3 Changes in Social Networks Over Time 41 4.2.4 Social Support and Drug Use 42 x

4.2.5 Overlap of Support, Sex, and Drug Networks 42 4.2.6 Significance 43 4.3 Methods 43 4.3.1 Study Design 43 4.3.2 The Sample 44 4.3.3 Instruments 45 4.3.4 Analysis 46 4.3.5 Human Subjects 48 4.4 Results 48 4.4.1 The Baseline Index Sample 48 4.4.2 The Social Networks at Baseline 52 4.4.3 Support Offered by Network Members at Baseline 55 4.4.4 Longitudinal Changes in Social Networks 61 4.5 Discussion 65 4.5.1 Drug Use Status in the SHIELD Study 65 4.5.2 Social Networks at Baseline 65 4.5.3 Social Support at Baseline 66 4.5.4 Social Networks Over Time 68 4.5.5 Conclusions 68 Chapter 5: The Role of Social Control and Differential Association Processes in the Relationship Between Drug Use and Social Network Drug Use Among Drug- Experienced Adults 70 5.1 Introduction 70 xi

5.2 Background 70 5.2.1 Theories of Deviant Behavior and Social Relationships 70 5.2.2 Parallel Theoretical Conflict: Social Selection and Social Influence 74 5.2.3 Gaps in the Study of Adult Drug Use and Social Network Drug Use 74 5.2.4 The Use of Social Network Data 76 5.2.5 The Present Study 77 5.3 Methods 78 5.3.1 Recruitment and Sampling 78 5.3.2 Measurement 80 5.3.3 Analysis 91 5.4 Results 83 5.4.1 Structural Equation Model (SEM) 87 5.4.2 SEM: Model Fit 90 5.4.3 Recursive Models of Significant Paths 90 5.5 Discussion 93 5.5.1 Limitations and Off-Setting Strengths 95 5.5.2 Conclusions and Future Directions 97 Chapter 6: Further Exploration of the Mechanisms of Social Control and Differential Association Processes on Adult Heroin/Cocaine Use 98 6.1 Introduction 98 6.2 Background 98 6.2.1 Theories on the Association of Substance Use and Peer Substance Use 98 xn

6.2.2 The Nature of Changes in Network Substance Use 100 6.2.3 The Role of Key Network Members 101 6.2.4 The Present Study 102 6.3 Methods 103 6.3.1 Recruitment and Sampling 103 6.3.2 Measurement 104 6.3.3 Analysis 106 6.4 Results 108 6.4.1 Microsocial Analysis 108 6.4.2 Key Peer Analysis 109 6.5 Discussion 113 6.5.1 Conclusions 114 Chapter 7: Conclusions and Future Directions 115 7.1 Summary of Findings 117 7.2 Conclusions 117 7.2.1 Implications for Intervention 118 7.3 Limitations and Strengths 119 7.4 Future Directions 124 7.5 Final Conclusions 125 Appendices 127 Bibliography 133 Curriculum Vita 141 xiii

List of Tables Table 3.1: Sampling Diagram of the SHIELD study, Baltimore, MD 28 Table 4.1: Baseline sample characteristics, by baseline drug use status (defined as any route and form of opiates or cocaine) in the SHIELD study, Baltimore, MD, 1997-1999 50 Table 4.2: Baseline demographic variables of the total network (n = 13,072 from n = 1,431 networks) 54 Table 4.3: Baseline frequencies of relationship types among the total network (n = 13,072 fromn= 1,431 index networks) 54 Table 4.4: Baseline frequencies of types of support offered among total network (n = 13,072 from n = 1,431 index networks) and frequencies of having at least one network member who could provide each type of support among the indexes in the SHIELD study, Baltimore, MD 58 Table 4.5: Mean and standard deviations of counts (by index, n = 1,431) of network members who can provide each type of support in the SHIELD baseline study 59 Table 4.6: Factor loadings and uniqueness of counts of number of network members who provide each type of support for index participants at baseline in the SHIELD study.... 60 xiv

Table 4.7: Baseline network characteristics of indexes (n = 1,431) who were and were not lost to follow-up (did not have a wave 2 or wave 4 assessment) in the SHIELD study, 1997-1999 62 Table 4.8: Descriptive statistics of the size of index networks at waves 1,2, and 4 for SHIELD participants with a known history of heroin and/or cocaine use 64 Table 5.1: Drug use (heroin and/or cocaine) and network drug use at assessments 1,2, and 4 for 1,108 participants in the SHIELD study, Baltimore, MD 84 Table 5.2: Baseline sample characteristics of those included in the final SEM model, by baseline drug use status (defined as any route or form of heroin or cocaine) in the SHIELD study, Baltimore, MD, 1997-1999 85 Table 5.3: Recursive models of significant longitudinal paths in the larger Structural Equation Model, using data from waves 1,2, and 4 from the SHIELD study in Baltimore, MD. "Net Drug Use" refers to the total number of network members who are current drug (cocaine and/or heroin) users 92 xv

List of Figures Figure 2.1: Theoretical models of the relationship between peers and drug use 10 Figure 5.1: Theoretical models of the relationship between peers and drug use 73 Figure 5.2: Standardized estimated coefficients from a Structural Equation Model, using data from waves 1, 2, and 4 of the SHIELD study, Baltimore. MD. "D. U." refers to the drug use defined as heroin and/or cocaine use in the prior 6 months 89 Figure 6.1: SEM results for index drug use predicted by both total network drug use and if there are any drug users among those friends rated at the highest level of trust, from the SHIELD study in Baltimore, MD Il l Figure 6.2: SEM results for index drug use predicted by both total network drug use and if there are any drug users among those friends who the index sees daily, from the SHIELD study in Baltimore, MD 112 xvi

List of Appendixes Appendix 1: Model results for the Structural Equation Model shown in figure 5.2, demonstrating the longitudinal relationship between index and network drug use in the SHIELD study in Baltimore, MD 128 Appendix 2: Model results for the Structural Equation Model shown in figure 6.1, demonstrating the longitudinal relationship between index, highly trusted peers, and network drug use in the SHIELD study in Baltimore, MD 129 Appendix 3: Model results for the Structural Equation Model shown in figure 6.2, demonstrating the longitudinal relationship between index, peers seen daily, and network drug use in the SHIELD study in Baltimore, MD 130 Appendix 4: Model results for the Structural Equation Model demonstrating the longitudinal relationship between index DAILY DRUG USE and network drug use in the SHIELD study in Baltimore, MD 131 Appendix 5: Model results for the Structural Equation Model demonstrating the longitudinal relationship between index HEROIN USE (in the prior 6 months) and total network drug use (heroin and/or cocaine in the prior 6 months) in the SHIELD study in Baltimore, MD 132 xvn

CHAPTER 1: INTRODUCTION Drug use and abuse pose a significant public health concern. Consequently, scientific inquiry is required to understand modifiable factors that influence drug use in order to develop appropriate interventions. There has been considerable interest in social factors that affect drug taking behavior, given the social nature of drug use. Additionally, the social context of drug use is important to understand and address in improve the effectiveness of any intervention targeting drug use. Research has consistently shown a relationship between an individuals' substance use and the substance use of his or her social network members. Competing theories posit that an individual's level of deviant behavior is influenced by the deviant behavior of his or her peers (differential association), poor bonding to conventional society and consequently seeking out or drifting towards deviant-behaving friends (social control), or that the relationship is bidirectional (interactional/ life-course). The direction of the causal relationship between drug use and network drug use has also been conceptualized and studied as a conflict between social selection and social causation/social influence. All of these theories have been primarily examined in the context of adolescent deviant behavior and drug use, and rarely in the context of adults. These theories seek to explain why individuals transition in and out of drug use. Perhaps the most important aspect of transitions in drug use among adults is the cessation of drug use, as individuals are significantly less likely to initiate drug use after the age of 29 as before. However, relatively little attention has been given to social factors related to drug use maintenance, cessation, and relapse compared to initiation. Furthermore, 1

many (if not most) individuals quit using drugs outside of treatment settings, a phenomenon referred to as "natural cessation." Nonetheless, relatively little research has addresses psychosocial factors among drug-experienced adults who are not in drug treatment. The present study examined the theories described above by modeling drug use and social network drug use using longitudinal data from current and former adult heroin/cocaine users in Baltimore, Maryland. Analyses included modeling relationships in both directions, as well as follow-up analysis that explored some of the hypothesized processes underlying these theories. Specifically, the nature of changes in network drug use is studied in detail, as social control theory would predict that changes in network drug use are due to changes in who is in the network, either from seeking out or drifting towards like peers, rather than changes in the behavior of friends over time. Additionally, the importance of key network ties in predicting later index drug use is explored, as differential association theory places a particular emphasis on the influence of closer relations. First, however, the present study will take advantage of social network inventory data to describe the social networks and social support of this sample of inner-city, drug-experienced adults. Study findings could inform important avenues for intervention among adult drug users. 2

Research Aims The aims of the present study were as follows: 1. To describe the structure of and social support from the social networks of drug- experienced adults in Baltimore, Maryland, and to describe changes in their social networks over time. 2. To examine how well the observed relationship between drug use and social network drug use among drug-experienced adults fits with the expected relationships under differential association and social control theories. 3. To examine whether changes in social network drug use are due to changes in behavior of network members over time or changes in the composition of the network. 4. To examine if drug use among closest network members or network members who have daily contact with the index has an influence on index drug use beyond the total network drug use. 3

CHAPTER 2: BACKGROUND AND SIGNIFICANCE 2.1 Background 2.1.1 The Problems of Drug Use and Abuse The use and abuse of narcotics pose a critical public health problem. The years of potential life lost due to overdose, chronic liver disease, heart disease, cancer, homicide, suicide and accidents (among other causes) are significantly greater for chronic users of opiates than the general population (Smyth et al., 2007). Substance use results in a large economic burden through decreased productivity, crime, and increased medical utilization (Mark et al., 2001; Rice 1999). Beyond direct health consequences caused by the ingestion of psychoactive substances, drug use is associated with the acquisition of communicable disease. Behavioral risks that occur as a result of sharing injection equipment (e.g., Gyarmathy and Neaigus, 2006) or decisions while intoxicated to engage in unprotected sex (e.g. Beadnell et al., 2006; Brien et al, 1994) result in an increased probability of acquiring HIV, one of the leading causes of loss of disability-adjusted life years (Murray and Lopez, 1997). Sharing injection equipment also results in high rates of Hepatitis C transmission, with prevalence of infection estimated to be as high as 80% for injection drug users (IDUs; Garfein et al., 1998; Gerlich et al., 2006). Because of the many health and social consequences of drug use and abuse, it is important to develop effective interventions to reduce use. Consequently, scientific research on longitudinal data is needed to understand what factors influence use, rather than factors that co-occur with changes in drug use behavior. As drug taking is often a 4

social behavior, potentially intervenable social factors should be explored. Additionally, it has been argued by Link and Phelan (1995) that social factors and the social context of any condition must be addressed for health interventions to have the desired effect. The present analyses will examine the role of exposure to drug use in the social network in order to inform intervention development, and will increase scientific understanding on the social context of drug use. The present study will do so using longitudinal data derived from a community-based cohort in Baltimore city, Maryland. 2.1.2 Drugs of Abuse in Baltimore city, MD Heroin is pharmacologically similar to morphine, the primary active ingredient in opium (Meyer and Quenzer, 2005). Heroin is more potent when injected, and is also more rapidly absorbed when snorted. Treatment of heroin addiction is complex and multidimensional. Specifically, physical/biological (withdrawal), psychological, and environmental aspects must be addressed. After detoxification, treatment may be given on an inpatient or outpatient basis. The most effective treatment with widespread implementation is methadone maintenance, particularly for individuals with a long history of heroin use. In this treatment, individuals are stabilized on methadone, a synthetic, long-acting opiate that is taken orally, frequently under supervision, and which reduces cravings for heroin. LAAM (Levo-Alpha Acetyl Methadol) and buprenorphine are used in a similar manner in treatment, and have a longer duration of effect. In addition, psychological counseling and twelve-step programs are commonly used in combination with these medications, or in isolation. 5

Cocaine is derived from coca plants, and when converted into hydrochloride salt it can be taken orally, intranasally, or by injection (Meyer and Quenzer, 2005). The hydrochloride salt can be converted back and smoked, either through a process called freebasing or through combining it with baking soda and then heating and drying it. This latter process is less volatile and produces the substance commonly known as "crack." Ingestion via injection and smoking lead to the fastest absorption of the drug, and consequently are thought of as more addictive. Unlike in the treatment of heroin, a substitution treatment is not available for cocaine. Consequently, out- and inpatient behavioral and psychosocial therapies are all that is available to individuals addicted to cocaine. Use of twelve-step treatment programs is also common in cocaine treatment. Baltimore city, Maryland, provides a unique location with which to examine adult drug use. Based on reports from the Drug Enforcement Agency, it has been claimed that Baltimore has the worst heroin problem and one of the worst crack cocaine problems in the United States, given the rates of per-capita use (Craig, 2000). For this particular urban location and others, devising interventions that reduce drug use are critical to the health and economy of the city. In order to develop appropriate interventions to an urban population, we must first understand modifiable factors that influence substance use within this particular population. As demonstrated in the next section, one such potential factor is exposure to social network substance use. 2.1.3 Theories of Deviant Behavior and Social Relationships Research has consistently shown a relationship between drug use and the drug use by one's social network members (Williams and Latkin, 2007; Best et al., 2005; Latkin et 6

al., 1999; Latkin et al., 1995; Kandel et al., 1978). Due to the complex causal relationships that determine behavior, the direction of this relationship is not well understood. There are several competing theories to explain the direction of influence between peers and deviant behavior. One key difference among the theories is directionality, i.e. whether drug-using peers lead to drug use or drug use leads to having drug-using peers. Understanding the direction of influence between peer drug use and an individual's substance use is crucial to understanding where opportunities for intervention occur. A number of theories have been developed that propose variations on the idea that peers influence the choice to engage in deviant behaviors such as drug use. The earliest was Sutherland's theory of differential association (Sutherland and Cressey, 1978; described by Thornberry and Krohn, 1997). Sutherland proposes that peers influence the creation of beliefs that are favorable or unfavorable to drug use and other deviant behaviors, which then influences decisions by the individual to engage in these behaviors. Sutherland placed particular emphasis on an emotionally close peer network. Similarly, social cognition/learning theory suggests that deviant behavior is learned through processes of operant conditioning (Akers and Jensen, 2003; Patterson et al., 1989). Deviance training theory also builds upon differential association theory, proposing that peers may reinforce deviant behaviors through attention, laughter, and other positive cues as a response to conversation and behavior (Dishion et al., 1999), which may account for the often iatrogenic effects of drug and alcohol treatment for adolescents. Lastly, social network theory is also an extension of differential association, and posits that individuals 7

are influenced both by the individuals with whom they interact, as well as the structure of their social environment (Valente et al., 2004; Hall and Wellman, 1985). The primary theory in conflict with differential association (and its related theories) is social control theory. According to this theory, deviant behaviors, such as substance use, result from a weakened or broken bond to society (Hirschi, 1974). The theory proposes that the cost of deviance is lessened when there are fewer commitments, relationships, or responsibilities, and that a person with fewer ties is more likely to violate laws and social norms. Under this theory, a person with weakened bonds to society seeks out the company of similarly disaffiliated peers. Consequently, according to social control theory, the trajectory that leads to substance use (and other deviant behavior) occurs before affiliation with substance-using/deviant peers. The theory further hypothesizes that an individual who has increased their bond to society, often through marriage or a new job, stops their drug use as a result, and consequently decreases their association with drug-using individuals. In cross-sectional studies the association between substance use and peer use would appear the same under social control theory as it would under differential association theory. Others have argued that these two primary perspectives are not necessarily in competition or mutually exclusive (Erickson et al., 2000). Interactional (Thornberry and Krohn, 1997) and life-course (Sampson and Laub, 1990) theories suggest more complex, bi-directional modes of causation. Like social control theory, these theories view the individual in the context of a lifelong trajectory, but see both peers influencing deviant behavior and deviant behavior influencing choice of relationships. The life-course theory in particular suggests that transitions that occur over the lifetime, such as entering 8

Full document contains 163 pages
Abstract: Background. Drug use poses a significant public health concern, and the social context of drug use warrants attention. Prior research has shown that there is a relationship between drug use and social network drug use, but theories differ as to which precedes the other. Specifically, proponents of differential association theory posit that the individual's drug use is influenced by the drug use of his or her social network members. Meanwhile, social control theory posits that an individual with a lessened bond to conventional society chooses to use drugs and consequently spends time with others who also use drugs. Interactional and life-course theories propose that the relationship is bidirectional. This theoretical conflict has also been conceptualized under the rubric of social causation/social influence vs. social selection. The present study examined the theories above by modeling bi-directional influences between drug use and network drug use among drug-experienced, following methods that have been developed to simultaneously test social influence and social selection theories. Additional analysis was conducted to determine if change in network drug use behavior is due to social network turnover, consistent with social control theory, or change in the behavior of network members over time. Follow-up models were estimated to determine if drug use by key network members explained later index drug use beyond the total network drug use, consistent with differential association theory. Lastly, the present study took advantage of data from a thorough social network inventory to conduct the above analyses, as well as to describe and report on the make-up of the social networks, social support available from the social network, and changes in the social network over time. Results. At baseline, 41% of all network members identified were current heroin/cocaine users. Social networks were primarily made up of individuals identified as "friends" or nuclear family members. While only 7% of the sample reported being married, 61% reported a main partner. Participants reported high levels of trust overall for their network, and the majority had at least one person who provided or could provide each type of support queried. In a factor analysis the number of supporters with whom the index shared meals and housing loaded onto one factor, while the number of supporters who the index could rely on for lending money, to give up time, for trusting with the index's money, for having fun, for talking about private matters, or for advice loaded onto another factor. The mean size of the social network at baseline was 9.1 individuals. The size of the social networks decreased at assessment 2 to 7.4, but increased at assessment 4 to 8.4, which was statistically greater than the second assessment, but also statistically lower than baseline. A similar pattern of change in the support network and drug network size over the follow-up was observed. In a SEM, bidirectional influences between index drug use and social network drug use were observed. In recursive models it was found that the social network drug use influenced index drug use through long-term processes, while index drug use influenced network drug use through short-term processes. Evidence of relationships in both directions was stronger between times 1 to 2 than times 2 to 4, the later two assessments having a greater time lag. Evidence of a relationship in the direction of index drug use influencing later network drug use was more consistent across assessments than evidence for a relationship in the opposite direction. In Microsocial Analysis it was found that the majority of change in social network drug use behavior was accounted for by turnover in the social network, rather than change in the behavior of individual network members over time. In SEMs that examined key network members, the relationship between key network member drug use and index drug use was null or negative in association, in models that also included total network drug use. All SEMs exhibited good fit. Conclusions. Drug use in the social networks of this sample was high, and drug treatment utilization low. There was a fair degree of overlap of the drug network with the support and sex networks (those individuals with whom the index has sexual relations). However, the sample also reported a substantial degree of social support from their networks. The results of the bidirectional model were consistent with what would be expected under both social control and differential association theories in drug use among drug-experienced adults. Findings of support in both directions also provide some support life-course and Interactional theories. The Microsocial Analysis provided further support for the process suggested by Social Control theory to explain the correlation of an individual's drug use with that of their network. The results of SEMs that analyzed key network members contradicted the premise that the closest network members exert greater influence over drug use behavior. Cumulatively, the findings provide support for a variety of intervention strategies, including engaging supporters in treatment, training participants to conduct peer outreach in order to reduce drug use in the community, and programs such as job training and relationship counseling to increase investment in traditional society.