Abstract
Those who study treatment and recovery from alcohol use disorder (AUD) and substance use disorder (SUD) generally agree that an individual’s social context impacts his or her success (or failure) in recovery. Recently, as the use of social network analysis has increased, studies on SUD recovery and treatment have adopted ego networks as a research tool. This review aims to tie together a thread of research for an efficient and effective summary. We selected peer-reviewed articles on individuals receiving treatment an intervention for SUD or AUD that used ego network measures of individual social networks. Ego networks have been studied as treatment outcomes, predictors of treatment outcomes in general, and how an individual’s ego network might be used to predict what specific treatment is most likely to succeed. We discuss relevant findings of studies using ego networks, the strengths and weaknesses of ego network approaches, and how future studies may benefit from the use of ego networks.
Keywords: Ego networks, alcohol use disorder, substance use disorder, treatment outcomes, review
Researchers have often examined social context when studying treatment and recovery from alcohol use disorder (AUD) and substance use disorder (SUD; in this review, SUD may refer to drug misuse or to simultaneous drug and alcohol misuse; AUD will be used only when alcohol is discussed in exclusion of other substances). Frequently, these studies have focused on abstract concepts such as social support or involvement in recovery-oriented social settings (e.g., Alcoholics Anonymous Affiliation; Majer et al., 2013). While these studies are meant to describe phenomena associated with an individual’s relation to a social network, they often collect data in a way that treats the network as a single entity, functioning as a characteristic of an individual respondent.
Although these studies may uncover important information, this approach to data collection does not allow one to account for the heterogeneity of an individual’s social network and also prevents one from adequately describing how an individual’s social context changes over time. These limitations obscure the social processes of SUD recovery.
In recent years, there has been a paradigm shift in how we study social context. Network analysis is one such method, and is unique in allowing the researcher to break down the amorphous summary of an ego’s social context, revealing some of the detailed structure of relationships between an ego and his or her alters. One way that the social network can be represented is as an “ego network” (sometimes referred to as a “personal network”). An ego network differs from a “whole network” by its being centered on an individual, and usually identified entirely by that individual. Whole network studies, in contrast, usually describe relationships within more or less bounded groups, where all members of the group are individually surveyed, and all members report on relationships with other members. This review aims to describe ways in which ego networks have been used to further our understanding of SUD treatment and recovery.
For the purposes of this review, we limit our discussion to ego networks. Ego networks are regularly used to study SUD and have been used in many studies of addiction treatment, in large part because the study population normally does not share the same context, making whole network methods inappropriate. Researchers might use inventories or name generators in interviews to map respondents’ ego networks. The respondent then provides information on each person in his ego network. The result is a data set containing multiple ties (alters) for the same respondent (ego), with each item for each alter as a separate variable. This approach offers some of the same detail as a whole network design, differing from whole network studies mainly in the lack of independent information from alters. Researchers have found many ways to use this range of available data to create models related to SUD recovery.
In this review, we examine three ways in which ego networks have been used in research on SUD recovery. Specifically, we look at studies where the network is used to predict treatment outcome, studies examining how treatment and recovery predict change in the network, and studies exploring how an individual’s pre-treatment network might indicate which treatment approach will be most effective. We identify the most frequently studied variables, summarize findings, consider variation across studies, and discuss directions for future research.
Our hope is that in compiling this review of literature, we can tie together a thread of research for an efficient and effective summary. While past literature reviews have examined relationships between treatment outcomes and social relationships, our review focuses specifically on instrumentation using ego networks. Additionally, we hope that focusing on the similarities and differences in various ego network studies may provide researchers with insights on future research.
Materials and Methods
For this review we selected peer-reviewed articles on individuals receiving treatment an intervention for SUD or AUD that used ego network measures of individual social networks. Google Scholar was used to locate relevant articles. Search terms included combinations of “personal network,” “ego network,” or “social network,” with “recovery,” “treatment,” or “intervention,” and “addiction,” “alcoholism,” “substance use disorder,” etc. To ensure that we exhausted all relevant literature, we also examined articles that either cited or were cited by those identified in these initial searches. We decided to include studies of recovery residences, because, despite the absence of professional staff and concrete treatment plans, these residences exist for the specific goal of maintaining sobriety.
After collecting these articles, we classified them as either using network variables to predict treatment outcome, using treatment and behavior change variables to predict changes in participants’ social networks, or examining the interaction between pre-treatment network and treatment type in predicting outcomes. Several articles fit into multiple categories. We did not conduct a meta-analysis, as there were not a sufficient number of articles on the topic.
Results
Given the selection criteria, we collected a total of 14 articles. All but one of these studies (Tracy et al., 2012) were longitudinal. Most used adaptations of the Important Persons and Activity Inventory (IP; Clifford & Longabaugh, 1991), while two used EgoNet (McCarty, 2003), and one used a unique set of questions (Broome et al., 2002). The treatments received by participants included short-term inpatient, various outpatient programs, and residential treatment. Retention rates were relatively high in these studies, with only two studies reporting follow up rates below 75% (71%, Broome et al., 2002; 51%, Beattie & Longabaugh, 1999). More information on the studies included is available in Table 1.
Table 1.
Sample characteristics, design, and outcomes.
| Authors (year) | Approach type | Sample | Treatment type | Design | Assessment tool | Predictors evaluated | Conclusions |
|---|---|---|---|---|---|---|---|
| Beattie & Longabaugh (1999) | Network predicting outcome | 152 AUD | Outpatient | Longitudinal | IPA Procidano and Heller’s (1983) 40-item Perceived Social Support self-report instrumenta |
|
Both general and alcohol-specific social support predicted higher PDA shortly after Tx Alcohol-specific support continued to predict PDA in the long term |
| Broome, Simpson, & Joe (2002) | Network predicting outcome | 748 AUD/SUD | Short term inpatient | Longitudinal | Questions formulated for this study specifically |
|
Deviant network and living with user/drinker predicted use of cocaine and frequent drinking Support for abstinence at home reduced odds of cocaine use |
| Groh et. al (2007) | Network predicting outcome | 897 AUD/SUD | Residential (OH) | Longitudinal | IP |
|
General support from friends interacted with length of stay to predict less alcohol use 90 days before follow up |
| Groh et. al (2011) | Network predicting outcome | 897 AUD/SUD | Residential (OH) | Longitudinal | IP |
|
Only drinking Bx of network predicted drinking at 4 months |
| Litt et. al (2009) | Tx predicting network change | 210 AUD | Outpatient | Longitudinal | IPA |
|
Bx Support for drinking decreased in all Tx Groups Bx support for abstinence increased for both NS interventions at 15 months; dropped for NS+C but rose for CM at 27 months Attitudinal support for abstinence higher over time for both NS and NS+C NS –number of abstinent friends increased then leveled at 15 months; NS+C declined at 15 |
| Longabaugh et. al (1995) | [network indicating effective Tx approach] | 229 AUD | Outpatient | Longitudinal | IPA |
|
Participants with low support for abstinence or low social investment, but not both had greatest PDA with ERE BBS treatment was found to be more effective for patients with either low investment in an unsupportive network or high investment in a supportive network. |
| Longabaugh et. al (2010) | Network predicting outcome | 1,373 AUD | Outpatient | Longitudinal | IPA |
|
Alcohol specific support predicted more days abstinent during and following Tx and fewer HDD following Tx Network opposition to drinking was increasingly predictive of higher PDA and lower PHDD Daily support was increasingly related to more HDD and fewer days abstinent following Tx |
| Min et. al (2013) | Tx predicting network change | 377 women with SUD | Residential (n=119) Outpatient (n=258) |
Longitudinal | EgoNet |
|
Women in residential had more substance using alters in network Only residential showed within group time effect, with significant increase in the number of Tx-related alters and decrease in substance using alters |
| Mohr et. al (2001) | Network predicting outcome Tx predicting network change | 1,726 AUD | Outpatient | Longitudinal, | IPA |
|
Participants who maintained high proportion drinking friends had more DDD at follow up than those who decreased proportion of drinking friends; Increase in proportion of non-drinking friends predicted fewer DDD at follow up Higher proportion of important drinking friends at baseline predicted more DDD at follow up Increased importance of non-drinking friends predicted PDA at follow up Proportion of drinkers decreased while proportion of non-drinkers increased post-Tx Contact with both drinking and non-drinking friends decreased Liking and importance of both drinking and non-drinking friends increased |
| Stout et. al (2012) | Network predicting outcome | 1,726 AUD | Outpatient | Longitudinal, | IPA |
|
At months 15 and 39: No effect for propensity stratification Number of pro-drinkers predicted poorer outcomes for PDA and DDD Number of pro-abstainers predicted higher PDA |
| Tracy et. al (2012) | Tx predicting network change | 242 women with SUD | Outpatient | Cross-sectional | EgoNet |
|
Networks in active treatment stage included more network members from Tx programs or 12-Step Neither the type nor amount of social support differed across Tx stages Networks of women in active Tx were less connected, as indicated by a higher number of components Networks of women in the persuasion stage had a higher degree of centralization, as indicated by networks dominated by people with the most ties |
| Wu & Witkiewitz (2008) | Network predicting outcome [Network indicating effective Tx approach] | 952 AUD | Outpatient | Longitudinal | IPA |
|
More network support for drinking predicted more drinking consequences, but only for participants assigned to MET or CBT |
| Zywiak, Longabaugh, & Wirtz (2002) | Network predicting outcome [Network indicating effective Tx approach] | 952 AUD | Outpatient | Longitudinal | IPA |
|
Larger daily networks and more abstainers/recovering alcoholics in networks predicted higher PDA and lower monthly volume of alcohol consumed Patients with a higher network drinking frequency did better (in terms of DDD and PDA) in Twelve Step Facilitation than in Motivational Enhancement Therapy |
| Zywiak et. al (2009) | Network predicting outcome Bx change predicting network change | 141 cocaine dependent | Residential | Longitudinal | IPDA |
|
Size of daily network predicted less drinking, less drug use, and less problem severity post-Tx Network substance involvement decreased for those who stayed abstinent |
Notes. Network variables predicting outcome listed by numerals, Bx change or Tx variables predicting network change listed by letters. PDA = percentage of days abstinent; DDD= drinks per drinking day, HDD= heavy drinking days, PHDD= percentage of heavy drinking days.
Procidano and Heller’s (1983) 40-item Perceived Social Support self-report instrument not collected as an ego network measure.
Network Predicting Outcome
Ten articles examined network variables as potential predictors of treatment outcome. Some of these variables were structural, describing more superficial aspects of the social network (e.g., network size), while others were more abstract, relating more to the roles and actions of network members (e.g., network support for drinking). Both structural and other variables were found to be relevant, albeit not consistently across studies.
General support
General support refers to social support that is unrelated to drinking, use, or abstinence (e.g., frequency of contact). Of the five studies that evaluated the effects of general support, two found higher levels of general support to be associated with better treatment outcomes, such as higher percentage of days abstinent (PDA; Beattie & Longabaugh, 1999) and less overall drinking and drug use (Groh et al., 2007).
Network size, daily network size
Of the three studies that examined the effects of network size and daily network size (number of alters with daily contact with participant), none found overall network size to predict treatment outcome. On the other hand, all three studies found associations between size of the daily network (the number of network members with daily contact with the ego) and treatment outcomes. Two of these studies detailed positive associations between daily network size and outcome, such as higher PDA and lower monthly volume (MV) of alcohol consumed (Zywiak et al., 2002) and less overall drinking and drug use and lower problem severity (Zywiak et al., 2009). Contradicting these findings, however, another study found daily network size to be associated with more heavy drinking days (HDD) and fewer days abstinent (Longabaugh et al., 2010).
Alcohol-specific support and support for abstinence and treatment
We conceptualized alcohol-specific support and support for abstinence and treatment as equivalent. The variables discussed in this section all concern the network’s perceived encouragement of seeking treatment or abstinence. Of the six articles that tested the relationship between treatment outcomes and alcoholic-specific support/support for treatment and abstinence, four found associations between network support for abstinence and treatment and better treatment outcomes. Improved outcomes included more days abstinent (Longabaugh et al., 2010), higher PDA (Stout et al., 2012; Beattie, 1999), fewer HDD (Longabaugh et al., 2010), and lower odds of drug use (Broome et al., 2002).
Support for drinking
Of the four articles that had network support for drinking as a variable of interest, two found associations between support for drinking and treatment outcomes. More support for drinking was related to worse recovery outcomes in these studies, such as lower PDA (Stout et al., 2012), more drinks per drinking day (DDD; Stout et al., 2012), and more drinking consequences (Wu & Witkiewitz, 2008).
Network drinking/drug use behavior
Of the six articles discussing the relationship between the behavior of the network and treatment outcomes, three found associations. The drinking/drug using behavior of network members was associated with higher likelihood of drinking at four months (Groh et al., 2011), more cocaine use and more frequent drinking (Broome et al., 2002), and more DDD (Mohr et al., 2001).
Treatment and Behavioral Change Predicting Network Change
Five of the examined articles aimed to study how abstinence and participation in treatment related to changes in the personal networks of participants. Of these, four found that treatment (Litt et al., 2009; Min et al., 2013; Mohr et al., 2001) and abstinence (Zywiak et al., 2009) could be related to a decrease in the number or proportion of substance use disorder in the network, although Min et al. (2013) only saw this relationship among participants in residential and not outpatient treatment.
Additionally, four of the articles detailed findings relating to change in the amount of support for abstinence (Litt et al., 2009; Tracy et al., 2012) or the number of abstinent/treatment related alters in the network (Litt et al., 2009; Min et al., 2013; Mohr et al., 2001; Tracy et al., 2012). While Tracy et al. (2012) found no difference in the amount or type of support across treatment stages, Lit et al. (2009) found that all of the three treatment approaches examined led to increases in behavioral and attitudinal support for abstinence, although the degree of change varied over time following treatment. All four of the articles that examined changes in the number of abstinent or treatment-related network members found increases with treatment, although Min et al. (2001) only observed this in participants in residential and not outpatient treatment and Litt et al. (2009) only observed this for participants who received treatment that specifically targeted network support.
Interactions between Network Characteristics and Treatment Approach
While only three of the articles we examined addressed the interaction between participants’ networks at treatment entry and the type of treatment received, all found that, in certain network conditions, some treatment approaches led to better outcomes than others. Treatment approaches that targeted network change, such as extended relationship enhancement (Longabaugh et al., 1995) or twelve-step facilitation (an intervention that focuses on increasing an individual’s social support for abstinence by encouraging participation in Alcoholics Anonymous; Wu & Witkiewitz, 2008; Zywiak et al., 2002) were related to better outcomes for participants with low investment in abstinent-supportive networks or high investment in unsupportive networks and for participants with higher network drinking frequency. For participants whose networks had higher drinking frequency, twelve-step facilitation was more effective at reducing the number of consequences experienced (Wu & Witkiewitz, 2008), reducing DDD (Zywiak et al., 2002), and increasing PDA (Zywiak et al., 2002). Similarly, while broad spectrum therapy was most effective when participants were less involved with unsupportive networks or more involved in more supportive networks, extended relationship enhancement was more effective in increasing PDA for patients who had low investment or unsupportive networks, but not both (Longabaugh, 1995).
Conclusions
Reviewing the literature, it is clear that an individuals’ social network is related to treatment and recovery outcomes. All of the studies reviewed found relationships between at least one of the tested variables and outcomes of interest. Still, results were inconsistent for most variables, with some studies finding relationships while others did not, and in one case producing relationships in opposite directions. One consistent finding was that completing treatment was related to increases of the number of abstinent and treatment-related alters in the network (Litt et al., 2009; Min et al., 2013; Mohr et al., 2001; Tracy et al., 2012), although, in some cases, the relationship was found to be contingent on the type of treatment received (Litt et al., 2009; Min et al., 2001).
Nearly all treatment-related changes in social networks found in these studies were positive, i.e., increases in abstinent alters, as well as with decreases in substance/alcohol use of alters (Litt et al., 2009; Min et al., 2013; Mohr et al., 2001; Zywiak et al., 2009). While Mohr et al. (2001) found that the amount of contact with non-drinking friends decreased following treatment, the same effect was observed for contact with drinking friends, and the proportion of abstinent friends still increased along with importance and liking. Based on these findings, we can conclude that treatment interventions often support positive restructuring of one’s social network.
Some findings were less clear. That network variables were rarely found to be consistently related to outcomes interfered with our ability to draw specific conclusions of which network characteristics were reliable predictors of outcomes. Although the direction of effects was consistent for studies identifying relationships between outcome and general support (Groh et al., 2007; Beattie & Longabaugh, 1999), alcohol-specific support (Longabaugh et al., 2010; Stout et al., 2012; Beattie, 2001; Broome et al., 2002), support for drinking (Stout et al., 2012; Wu & Witkiewitz, 2008), or network drinking/using behavior (Broome et al., 2002; Groh et al., 2011; Mohr et al., 2001), each variable had at least one study where no such relationship was found. Additionally, studies reported contradictory findings on the effect of daily network size. It is possible that these inconsistencies could be explained by differences in how studies measured network variables or outcomes. There is also the possibility that these relationships depend on as yet-unidentified conditions, e.g. treatment experience specifics, characteristics of individuals, and so on. Additional research might be able to clarify the meaning of these findings.
While these studies illustrate ways in which ego networks can be used in researching treatment for SUD, they also enable us to identify unexamined questions. While it is useful for us to be able to describe changes in personal networks following treatment or ways in which network variables predict treatment outcomes, understanding the mechanisms behind these interactions is equally important. Only with such understanding are treatment and support providers able to improve programs to encourage and support the type of re-engineering of substance abusers’ social relationships in a way that is consistent with sustained recovery. This would include developing models where personal change (treatment gains, SUD behaviors) and social network change are mutually interactive, rather than one being exogenous and the other endogenous.
Given the proportion of these studies that used some iteration of the IP, it is evident that this measure has become the standard in examining the role of social networks in SUD recovery. Unfortunately, there is inconsistency in the way investigators use the IP. Some choose to analyze individual items (e.g., number of drinking network members; Zywiak et al., 2002), while others choose to calculate composite or summary variables for analysis (e.g., Support for Drinking from Network Members; Groh et al., 2011). Additionally complicating matters, there is disagreement among researchers as to how many and which composite factors create the best model. The inconsistency of variables’ predictive value may be a reflection of this non-uniformity.
The generalizability of results found may be limited, as nine of the studies considered only alcohol-related outcomes without collecting information on use of other substances (Groh et al., 2011; Groh et al., 2007; Litt et al., 2009; Mohr et al., 2001; Zywiak et al., 2002; Stout et al., 2012; Wu & Witkiewitz, 2008; Longabaugh et al., 2010; Beattie & Longabaugh, 1999). While most of said studies accepted only participants with AUD, two of these studies examined only alcohol-related outcomes even though their sample was composed mainly of poly-substance use disorders (Groh et al., 2007; Groh et al., 2011). Two additional studies limited their sample and outcome measures to cocaine (Zywiak et al., 2009) or cocaine and alcohol (Broome et al., 2002).
The recruitment methods utilized and the demographics of participants may also limit generalizability. While some studies recruited throughout communities (e.g., Litt et al., 2009), most studies recruited all (Zywiak et al., 2009; Broome et al., 2002; Beattie & Longabaugh, 1999; Longabaugh et al., 1995) or some (Mohr et al., 2001; Zywiak et al., 2002; Stout et al., 2012; Wu & Witkiewitz, 2008) participants from clinical or treatment settings. One study (Longabaugh et al., 2010) relied entirely on a population of individuals receiving medical and behavioral interventions as part of a clinical trial. There is in general no reason to expect results from these different populations to match.
The studies discussed demonstrate that ego networks can be a valuable tool in studying recovery from SUD and AUD. Familiarity with a patient’s social network composition may be used to inform treatment. With further research, practitioners may be able to identify risk factors within the social networks, select treatments that have been found to alter these risk factors, and, ultimately, improve treatment matching and outcome. To facilitate this, future studies should continue to clarify what social network variables affect which outcomes and examine active ingredients in interventions that are found to affect social network composition. Such studies will provide valuable contributions to the field of addiction psychology.
Acknowledgments
The authors appreciate the financial support from the National Institute on Drug Abuse (grant numbers DA13231 and DA19935).
Contributor Information
Ariel Stone, DePaul University, Chicago, IL, USA.
Leonard A. Jason, DePaul University, Chicago, IL, USA
John M. Light, Oregon Research Institute
Edward B. Stevens, DePaul University, Chicago, IL, USA
References
- Beattie MC. Meta-analysis of social relationships and posttreatment drinking outcomes: Comparison of relationship structure, function, and quality. J Stud Alcohol Drugs. 2001;62:518–527. doi: 10.15288/jsa.2001.62.518. [DOI] [PubMed] [Google Scholar]
- Beattie MC, Longabaugh R. General and alcohol-specific social support following treatment. Addict Behav. 1999;24(5):593–606. doi: 10.1016/s0306-4603(98)00120-8. [DOI] [PubMed] [Google Scholar]
- Broome KM, Simpson D, Joe GW. The role of social support following short-term inpatient treatment. Am J Addiction. 2002;11(1):57–65. doi: 10.1080/10550490252801648. [DOI] [PubMed] [Google Scholar]
- Clifford PR, Longabaugh R. Environmental Treatment of Alcohol Abusers. Richard Longabaugh, Principal Investigator; 1991. Manual for the Administration of the Important People and Activities Instrument. Adapted for use by Project MATCH for NIAAA 5 R01AA06698-05. [Google Scholar]
- Groh DR, Jason LA, Davis MI, Olson BD, Ferrari JR. Friends, family, and alcohol abuse: An examination of general and alcohol-specific social support. Am J Addiction. 2007;16:49–55. doi: 10.1080/10550490601080084. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Groh D, Jason LA, Ferrari J, Halpert J. A longitudinal investigation of the predictability of the three-factor model of the Important People Inventory. Am J Drug Alcohol Ab. 2011;37(4):259–263. doi: 10.3109/00952990.2011.591017. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Litt MD, Kadden RM, Kabela-Cormier E, Petry NM. Changing network support for drinking: Network Support Project 2-year follow up. J Consult Clin Psych. 2009;77(2):229–242. doi: 10.1037/a0015252. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Longabaugh R, Wirtz PW, Zywiak WH, O’malley SS. Network support as a prognostic indicator of drinking outcomes: The COMBINE study. J Stud Alcohol Drugs. 2010;71(6):837. doi: 10.15288/jsad.2010.71.837. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Longabaugh R, Wirtz PW, Beattie MC, Noel N, Stout R. Matching treatment focus to patient social investment and support: 18-month follow-up results. J Consult Clin Psych. 1995;63(2):296. doi: 10.1037//0022-006x.63.2.296. [DOI] [PubMed] [Google Scholar]
- Majer JM, Jason LA, Aase DM, Droege JR, Ferrari JR. Categorical 12-step involvement and continuous abstinence at 2 years. J Subst Abuse Treat. 2013;44(1):46–51. doi: 10.1016/j.jsat.2012.03.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- McCarty C. Personal Network Software. 2003. Egonet. [Google Scholar]
- Min MO, Tracy EM, Kim H, Park H, Jun M, Brown S, McCarty C, Laudet A. Changes in personal networks of women in residential and outpatient substance abuse treatment. J Subst Abuse Treat. 2013;45:325–334. doi: 10.1016/j.jsat.2013.04.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Mohr CD, Averna S, Kenny DA, Del Boca FK. Getting by (or getting high) with a little help from my friends: An examination of adult alcoholics’ friendships. J Stud Alcohol. 2001;62:637–645. doi: 10.15288/jsa.2001.62.637. [DOI] [PubMed] [Google Scholar]
- Stout RL, Kelly JF, Magill M, Pagano ME. Association between social influences and drinking outcomes across three years. J Stud Alcohol Drugs. 2012;73:489–497. doi: 10.15288/jsad.2012.73.489. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Tracy EM, Kim H, Brown S, Min MO, Jun M, McCarty C. Substance abuse treatment stage and personal networks of women in substance abuse treatment. J Soc Social Work Res. 2012;3(2):65–79. doi: 10.5243/jsswr.2012.5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Zywiak WH, Neighbors CJ, Martin RA, Johnson JE, Eaton CA, Rohsenow DJ. The Important People Drug and Alcohol interview: Psychometric properties, predictive validity, and implications for treatment. J Subst Abuse Treat. 2009;36(3):321–330. doi: 10.1016/j.jsat.2008.08.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Zywiak WH, Longabaugh R, Wirtz PW. Decomposing the relationships between pretreatment social network characteristics and alcohol treatment outcomes. J Stud Alcohol. 2002;63(1):114–121. [PubMed] [Google Scholar]
- Wu J, Witkiewitz K. Network support for drinking: An application of multiple groups growth mixture modeling to examine client-treatment matching. J Stud Alcohol Drugs. 2008;69(1):21–29. doi: 10.15288/jsad.2008.69.21. [DOI] [PMC free article] [PubMed] [Google Scholar]