Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2024 Feb 23.
Published in final edited form as: Alcohol Treat Q. 2023 Feb 23;41(2):237–249. doi: 10.1080/07347324.2023.2181120

Complex Contexts within Oxford Houses: Psychiatrically Comorbid Social Networks

Ted J Bobak 1, John M Majer 2, Leonard A Jason 3
PMCID: PMC10237632  NIHMSID: NIHMS1875429  PMID: 37274149

Abstract

This study aimed to explore whether there are differences between Oxford House recovery home residents with psychiatric comorbidity in their ability to form, maintain, and dissolve loaning ties and seek advice, when compared to Oxford House residents without comorbidity, and if differences do exist, are those ties mono- or bi-directional. Findings indicated unique interdependencies among individuals with psychiatric comorbidity for advice seeking, loaning, and recovery factor scores. The results of this investigation are consistent with the dynamic systems theory conceptions of community-based recovery. Recovery homes provide access to social capital, via the residents’ social network, by facilitating recovery-oriented social exchanges, which can lead to changes to the recovery home social dynamics. Upon interpreting the results of this study, components from a dynamic systems theory emerged (e.g., explaining the processes that preserve or undermine the development, maintenance, and dissolution of a network); and provided a framework for interpreting the loaning, advice-seeking, and the latent recovery factor networks and their relationship with psychiatric comorbidity. A deeper understanding of the interplay among these dynamics is described providing an understanding of how Oxford House recovery homes promote long-term recovery in a shared community setting for those with high psychiatric comorbidity.

Keywords: Recovery Homes, Oxford Houses, Social Networks, Psychiatric Comorbidity

Recovery homes are the most utilized form of substance use disorder (SUD) post-treatment aftercare in the United States (Polcin et al., 2010). It is estimated that there are over 17,000 recovery homes in the United States that serve about 270,000 individuals over the course of a calendar year (Jason, Wiedbusch, Bobak, & Taullahu, 2020). Recovery homes provide transitional, cost effective, recovery supportive housing (Jason & Ferrari, 2010), and serve to bridge the gap between inpatient treatment, or an institutional setting, and the full reentry into mainstream society. Currently, there are nearly 3,000 self-help run Oxford Houses located throughout the United States and several other countries (Oxford House, 2021). The average number of residents who co-inhabit an Oxford House ranges anywhere between six and twelve individuals. Each house is responsible for conducting weekly business meetings, which are facilitated by the democratically elected house president. Current residents are expected to attend these business meetings, contribute an equal portion toward house expenses (i.e., rent, fines), refrain from disruptive behaviors, maintain house cleanliness through assigned chores, and abstain from alcohol and illicit substance use (Jason & Ferrari, 2010).

Individuals with a psychiatric comorbidity had poorer post-treatment outcomes compared to those had a SUD and no co-occurring mental disorder (Aase, Jason, Ferrari, Li, & Scott, 2014); moreover, individuals with psychiatric comorbidity have lower quality of life scores when compared to the general population and individuals with minor health problems (Fei, Yee, & Habil, 2016). Additionally, substance misuse has shown to be a risk factor (among others) for the onset of psychiatric disorders (Enez-Darcin, Nurmedov, Noyan, Yilmaz, & Dilbaz, 2015). Individuals with SUDs experience various impairments that include cognitive, behavioral, and physiological symptoms which may contribute to psychiatric comorbidity patterns and prevalence rates for co-occurring psychiatric disorders (American Psychiatric Association, 2022). Psychiatric comorbidity prevalence rates were highest for anxiety disorders, affective disorders, and antisocial personality disorder among Oxford House recovery home residents (Majer et al., 2002a).

Dynamic systems theory (DST) is commonly used to explain a system in which a large network of factors, absent of a central control, with simple rules of operation that give rise to complex collective behavior patterns, sophisticated advice processes, and adaptation via information sharing and learning (De Bot, Lowie, & Verspoor, 2007). DST can be used to describe the complex changing behaviors of a social network that emerge from the collective actions of many interacting components (Mitchell et al., 2009). DST addresses principles of change and development over time, without constraints on any specific endpoint (Thelen & Ulrich, 1991); examining the processes of change, over time, instead of specific outcomes. DST allows researchers to examine cycles of change, from one timepoint to the next, of stabilization and destabilization (Thelen & Ulrich, 1991), providing the groundwork for measuring the dimensions of a dynamic system. Applying DST can help us understand the ways that components, or actors of a network are interconnected and how individuals interact and behave (Houchin & MacLean 2005; Stoebenau & Valente 2003).

To analyze the trajectory of a dynamic system, the system must be simulated through iterations. The current study seeks to build on previous social network literature from Bobak, Majer, and Jason (2021) - that demonstrated those with psychiatric comorbidity who lived in houses with others with psychiatric comorbidity had better outcomes. The present research uses DST as a foundation, coupled with a stochastic actor-oriented modelling framework (described below), to determine if loaning and advice seeking ties are predictive recovery dimensions for persons with psychiatric comorbidity who are living in an Oxford House.

Although evidence suggests that Oxford Houses are effective in helping (e.g., facilitate long-term recovery) residents with psychiatric comorbidities, there is a need to understand how social dynamics create therapeutic outcomes for this vulnerable population. An Oxford House resident’s selection and endorsement of monetary loaning (trust) and advice-seeking behaviors describe elements of social capital within the home (Jason, Guerrero, Lynch, Stevens, Salomon-Amend, & Light, 2020). Although there is a body of research that examines SUD, recovery homes (i.e., Oxford Houses), and comorbid psychiatric severity (Abou-Saleh & Janca, 2004; Grant et al., 2004; Majer, Payne, & Jason, 2014; Regier et al., 1990), the literature lacks a clear description of the linkages between psychiatric severity and social network composition, recovery capital, and the dynamic processes that exist among house residents.

Through the DST lens of exploration, it would be useful to examine psychiatric comorbid Oxford House residents’ interactions with others inside of their house-specific social network, in terms of their willingness to loan money - a catalyst for trust - to other residents. These behavior processes are not static, as they tend to change over time. DST is an excellent framework for explaining these changes. The current study explored possible differences between high and no Psychiatric Severity Index (PSI) groups. In addition, an analysis of social network ties among Oxford House residents can help understand cohesion, influence, and selection differences between these two groups. Lastly, this study explores the microcosms of individuals with high versus low PSI scores at the loaning and advice-seeking levels to understand how these domains contribute to their overall recovery factor scores.

Method

Participants and Procedures.

Self-report survey data were collected every four months, over a two-year period, from Oxford House residents in Texas, Oregon, and North Carolina for a total of seven (7) waves – including baseline; the three geographical sites were selected to amplify the generalizability of our results. There were 714 Oxford House residents throughout the 2-year study period; 93% (n = 666) agreed to participate in the study. Of those, 74% (n = 497) left the Oxford House at some time during the 6-wave study period.

The survey included questions regarding sociodemographic information: age, race/ethnicity, sex, marital status, drug of choice, length of substance misuse and abstinence, length of stay in an Oxford House, level of education, and employment status. The sample percentages, broken down by race/ethnicity: White (78.8%), Latinx (10.1%), Black (8.6%), and Other (2.5%). The small number of cases within ethnicity categories in the present study limited a full analysis of ethnicity groups. However, a recent investigation found unique effects among African American recovery home residents (Jason, Guerrero, Bobak, Light, & Stoolmiller, 2021), so to extend these findings in relation to social dynamics among residents with psychiatric comorbidity, race/ethnicity was categorized into a dichotomous variable for analyses by comparing participants who reported their ethnicity as African American (n = 62) to a collapsed group of all other reported ethnicities (“other,” n = 565). Participants were relatively equal with respect to their sex (51.7% male & 48.3% female). Participants were evenly split on education level for high school or less = 43.9% vs. some college = 43.7%, with 12.4% reporting college degree or higher. The majority of residents were employed full-time (59%). In addition, participants reported an average length of stay in an Oxford House of 6.13 months with a SD of 9.36.

Participants were recruited with the assistance of individual Oxford House presidents, who provided a synopsis of the research project from a script that the research team constructed, recruited the study participants, during their monthly house meeting. Trained recruiters, through individual face-to-face interactions, conducted the survey interviews. They began with a brief overview of the project; acceptance criteria included that the house president and all, or all minus one resident agreed to participate in the study. The survey questionnaires were de-identified to ensure participant confidentiality, and the DePaul University Institutional Review Board granted permission to conduct the study. Each participant was compensated $20 for completing the survey, at every data collection wave.

Measures

Addiction Severity Index-Lite.

The Addiction Severity Index-Lite (ASI-Lite; McLellan et al., 1997) is used to assess problematic drug and alcohol use over the past 30 days. The ASI-Lite is shown to have good validity and reliability (Cacciola, et al., 2007). The Psychiatric Severity Index (PSI) – a subscale of the ASI – is used for assessing psychiatric problem severity. The composite scores are calculated using a weighted formula that generates scores ranging from .00 to 1.00, with higher scores indicating greater psychiatric severity (McLellan et al., 1992). The PSI has excellent test-retest reliability (≥.83) and has been used in substance abuse research for over nearly four decades (McLellan, Luborsky, Woody, O’Brien, & Druley, 1983). The PSI is a measure of overall psychiatric severity that was used as a proxy to indicate psychiatric comorbidity, by dichotomizing PSI scores into two groups (i.e., high vs low/zero) where the high PSI group indicates psychiatric comorbidity, an approach consistent with assessing psychiatric comorbidity in previous studies (Ball et al., 2004; Cridland et al., 2012; Majer et al., 2008, 2016). The high PSI group is defined by PSI scores that are ≥ +1 SD above the mean (McLellan et al., 1983). The present sample had a PSI mean of 0.146 and a SD of 0.185. Participants (n = 82) with PSI scores ≥ 0.331 (0.146 + 0.185) comprised the high PSI group, whereas other participants (n = 472) represented the low/zero PSI group.

Latent Recovery Factor.

The latent recovery factor (RF)– where higher scores indicate more positive recovery outcomes - was calculated from a confirmatory factor analysis across several recovery capital indicators (Jason et al., 2020). This measure was constructed from the following instruments:

Wages.

Self-report data for wages, up to 30 days before survey completion, were square root transformed to reduce right skew and treated as a continuous variable.

Quality of Life.

The World Health Organization Quality of Life Assessment-Brief (WHOQOL Group, 1998) is a 24-item questionnaire that assesses participant quality of life across social, environmental, physical, and psychosocial dimensions. This scale has been validated in substance-misusing populations (Garcia-Rea & LePage, 2010). The subscales varied in their reliability (α = .89 for social relationships, .84 for environment, .83 for physical, and .83 for psychological). The alpha for the entire measure, in our sample, was .89.

Abstinence Coping Self-efficacy.

The brief Drug Taking Confidence Questionnaire (DTCQ-8, Sklar, Annis, & Turner., 1999) is an 8-item survey, derived from the 50-item Drug Taking Confidence Questionnaire (DTCQ-50), that measures abstinence self-efficacy (Stevens, Jason, Ferrari, & Hunter, 2010). The DTCQ-8 accounts for 95% of the total variance from the DTCQ-50 and correlated with 0.97 of the total DTCQ-50 scores (Skylar, Annis, & Turner, 1999). The survey includes questions that prompt participants to consider themselves in eight, theoretically high-risk situations and indicates how confident they are in their abilities to resist the temptation to use alcohol, or illicit substance given the hypothetical circumstances. This measure, for our sample, has good reliability (α =.95).

Self-esteem.

The Rosenberg’s Self-Esteem Scale (Rosenberg, 1965) measures participants’ positive and negative perspectives about themselves. The Self-Esteem Scale (SES) is a 10-item measure that utilizes a 4-point Likert Scale that ranges from “strongly agree” to “strongly disagree.” The internal reliability (α =.92) of the SES is good, for our sample.

Stress.

The Perceived Stress Scale (PSS, Cohen et al., 1983) measures the degree in which participants perceive situations in their lives to be stressful, in the last 30 days. The PSS consists of 4-items measured on a 5-point Likert scale ranging from “never” to “very often”. The internal reliability of the PSS, for our sample, was .73.

Social support.

The Interpersonal Support Evaluation List (ISEL, Cohen & Wills, 1985; Cohen, Mermelstein et al., 1985) measures three (3) types of perceived social support (tangible, appraisal, and belonging). Tangible support refers to instrumental aid and monetary assistance; appraisal support refers to having someone to talk to about one’s problems; and belonging support refers to the availability of people with whom one might engage in activities. The ISEL consists of 12-items measured on a 4-point Likert scale ranging from definitely false to definitely true. The internal reliability of the support scale, for our sample, was .88.

Sense of Community.

The Psychological Sense of Community (SOC) is a 9-item scale utilized to measure participant’s sense of community (Jason et al., 2015). The three subscales are Entity, Membership, and Self, and for our sample, they have Cronbach alphas of .67, .92, and .91, respectively. The SOC scale was used as a whole measure (α = .91) (Stevens, Jason, Ferrari, & Hunter, 2010; Graham, Jason, & Ferrari, 2009).

Hope.

The State Hope Scale (Snyder et al., 1996) consists of 6 items that measure participants’ current state of hope. The Hope measure contains two sub-scales Agency (α = .94) and Pathways (α = .81). We included a 3-item subscale of hope that measures Environmental Context (Stevens et al., 2014) (α = .97). This 9-item scale was analyzed as a whole measure, and for our sample the α = .90.

Social Network Analysis

Social Network Instrument.

The Social Network Instrument (SNI; Jason & Stevens, 2017) was utilized to capture the social dynamics within each Oxford House. This instrument has been used in several investigations on the social networks of recovery home residents (Jason & Stevens, 2017; Jason et al., 2018). This type of network measure is a reliable instrument (Hlebec & Ferligoj, 2002). The SNI has a Cronbach’s alpha of .81 and all items contribute positively. The SNI is used to measure multiple relationship characteristics, where Oxford House residents rated each member of their house on the network relationships of money loaning and advice-seeking; data were also collected on frequency and strength of these network ties. Each social network relationship type was measured with a 5-point Likert scale. Participant ratings were represented by an adjacency matrix with each row representing the ratings provided by an individual and each column representing the ratings received by an individual. The SAOM framework requires that all rating values be dichotomized (0 = no relationship present; 1 = relationship present) and entered as a corresponding element of the matrix. An advice-seeking relationship was present if the respondent reported seeking advice from another resident “very often” or “quite often”, but not present otherwise (e.g., regularly, rarely, never). A money loaning relationship (i.e., a willingness to lend resources) was present if the respondent endorsed a willingness to loan either $500 or $100 to another house resident but was not considered present if the respondent reported lesser amounts (i.e., $0, $10, $50).

A social network analysis was conducted of 42 Oxford Houses using R (R Core Team, 2022) – a free, open-source statistical software environment and programming language that can be used to wrangle, analyze, and graph data. Additionally, the (R - Simulation Investigation for Empirical Network Analysis) (RSIENA) (Ripley et al., 2020) - that is generally used to analyze the dynamics in a social network - was utilized to implement a stochastic actor-oriented model (SAOM) that examined the endogenous co-evolution of behavior and social relationships (e.g., selection and influence). These statistical methodologies move beyond the individual-level focus; instead, they seek to explore the transactions between recovery home residents and their environments by illuminating the mechanisms that coalesce individual factors with their environment (Jason & Glenwick, 2016; Parkin, 2015).

Social network analysis in RSIENA (Ripley et al., 2020) examined bidirectional relationship patterns (i.e., reciprocity, density). Reciprocity measures the symbiosis of a friendship tie. Density is the number of existing ties, divided by the total number of possible ties. This statistical method examined relations between the latent recovery factor (social capital) and network ties for loaning and advice seeking.

RSIENA simulates longitudinal data (i.e., decisions by individual actors) based on the cumulating effects of network change mechanisms by deducing from the observed networks (Veenstra & Steglich, 2012). Several indicators of the data were considered, prior to running a SOAM (e.g., test of normality). RSIENA can be used to model network mechanisms of change through a method of moments estimation (Ripley et al., 2020), where a series of steps are utilized to respectfully measure selection, creation, maintenance, and dissolvement phenomena. Additionally, longitudinal composition changes, accounting for actors who leave or join the network in-between waves/observations, also utilizes the structural zero approach which signifies the introduction (1) or absence (0) of an actor into the network at any data collection wave (Ripley et al., 2020). RSIENA allows for a maximum of 20% missingness in data per wave; there is less than 20% of missing data (e.g., network, covariate, behavioral) per wave in the current dataset. The default method for treating missing data in RSIENA were examined by Zandberg et al. (2019) and Huisman et al. (2008) and found to provide the best performance when compared to other methods of handling missing data. The model estimation reliability is determined via convergence statistics such as t-ratios (simulated vs. observed), instead of R2 statistics, for each predictor. A good model convergence is determined when t-ratio values are ≤ .10, where lesser values demonstrate better convergence, with a maximum convergence ratio threshold of 0.25 (Ripley et al., 2020).

The relationship types that were examined in this study, via structural network effects (i.e., density, reciprocity, similarity, indegree and outdegree, were a willingness to loan and advice-seeking; along with behavioral effects (i.e., latent recovery factor scores), and individual attributes (age, sex, and race/ethnicity); descriptive statistics for density (number of ties) and mutual dyads (reciprocity) across the loaning and advice-seeking networks.

Results

A SAOM framework was utilized via the RSIENA package (Ripley et al., 2020). Table 1 includes statistics for two network types (e.g., loaning, and advice-seeking), over the six data collection waves; the number of possible ties vary between waves, as participants entered or exited the recovery homes throughout the duration of the study. Table 1 also shows parameter estimates, standard errors, p values, confidence intervals (b = estimate, [95% CI Lower, Upper]), and t ratio statistics. The overall maximum convergence ratio (a summary measure across effects) was .18, indicating that this model convergence is very good; each individual parameter convergence t ratio (an autocorrelation between successive iterative estimates) was ≤ .06 (scores closer to zero are ideal).

Table 1:

Stochastic Actor-Oriented Model Results–maximum likelihood estimation

Parameter Estimate SE p-Value 95% Confidence Interval Convergence t-ratio
Network Dynamics
 1. Advice rate (period 1) 1.99 0.27 <.001 (1.5, 2.5) −0.02
 2. Advice rate (period 2) 2.21 0.44 <.001 (1.4, 3.1) −0.03
 3. Advice rate (period 3) 2.03 0.38 <.001 (1.3, 2.8) 0.05
 4. Advice rate (period 4) 1.86 0.35 <.001 (1.2, 2.5) 0.03
 5. Advice rate (period 5) 5.62 1.20 <.001 (3.3, 7.9) 0.05
 6. Advice: outdegree (density) −0.26 0.09 .003 (−0.4, −0.1) −0.01
 7. Advice: reciprocity 0.88 0.14 <.001 (0.6, 1.2) 0.01
 8. Advice: PSI similarity 0.33 0.15 .002 (0.0, 0.6) 0.01
 9. Loan rate (period 1) 3.08 0.61 <.001 (1.9, 4.3) 0.03
 10. Loan rate (period 2) 2.81 0.72 <.001 (1.4, 4.2) −0.04
 11. Loan rate (period 3) 2.46 0.42 <.001 (1.6, 3.3) 0.02
 12. Loan rate (period 4) 2.54 0.52 <.001 (1.5, 3.6) 0.02
 13. Loan rate (period 5) 3.47 0.68 <.001 (2.1, 4.8) 0.01
 14. Loan: outdegree (density) −0.50 0.08 <.001 (−0.7, −0.3) −0.04
 15. Loan: reciprocity 0.81 0.13 <.001 (0.6, 1.1) −0.01
 16. Loan: PSI alter 0.38 0.18 <.001 (0.1, 0.7) −0.04
Behavior Dynamics
 17. Rate RF (period 1) 1.46 0.30 <.001 (0.9, 2.0) 0.04
 18. Rate RF (period 2) 1.22 0.29 <.001 (0.6, 1.8) −0.01
 19. Rate RF (period 3) 1.56 0.39 <.001 (0.8, 2.3) 0.01
 20. Rate RF (period 4) 1.41 0.29 <.001 (0.8, 2.0) 0.01
 21. Rate RF (period 5) 1.83 0.48 <.001 (0.9, 2.8) −0.05
 22. RF linear shape 0.53 0.11 <.001 (0.3, 0.8) 0.02
 23. RF quadratic shape −0.49 0.0 <.001 (−0.7, −0.3) −0.01
 24. RF: effect from Sex −0.47 0.15 .002 (−0.8, −0.2) 0.01
 25. RF: effect from Race 0.57 0.25 .020 (0.1, 1.1) 0.01
 26. RF: effect from PSI −0.66 0.25 <.001 (−1.1, −0.2) 0.06
Open in a new tab

Network Dynamics

Rate parameters for advice and loaning are classified as inter-wave-specific estimates that show the amount of change in each endogenous variable – changed or determined by its relationship with other variables within the model – and sufficiently confirms variation for the model to explain (Jason, Lynch, Bobak, Light, and Doogan, 2021). Network closure parameters (e.g., outdegree [density] & reciprocity) are used to index structural tendencies that predict network ties and provide descriptive information (see Table 1). The outdegree (density) parameters for advice (b = −0.26, [−0.4, −0.1]) and loaning (b = −0.50, [−0.7, −0.3]) were significant. The negative outdegree effects for density show that the loan networks are relatively sparse and that the more ties one has, the less likely they are to add more. Reciprocity for advice (b = 0.88, [0.6, 1.2]) and loaning (b = 0.81, [0.6, 1.1]) were also significant and positive for all networks, suggesting a tendency for these relationships to be bidirectional. For advice networks, the more similar the PSI level between the ego and alter, the more likely they are to seek advice from one another (b = 0.33, [0.0, 0.6], p = .002). For loan networks, the alter reported being willing to loan to those who high PSI scores and this effect was significant (b = 0.31, [0.1, 0.7]).

Behavior Dynamics

The model also examined predictors of latent RF scores, specifically sex, race, and PSI. The RF quadratic and linear shape effects, that are included in the model, represent the shape and location of the RF distribution for the three predictors (i.e., sex, race, PSI), when the other model terms are set to zero; beyond this function, the quadratic and linear shape effects are primarily irrelevant (Jason, Lynch, Bobak, Light, & Doogan, 2021). For the RF, a negative significant parameter for sex effects show that male residents were less likely to improve their RF than females (b = −0.39, [−0.8, −0.2]). Additionally, a positive significant parameter for African American ethnicity (race) shows that African Americans had better RF scores than the other ethnicities in the sample (b = 0.63, [0.1, 1.1]). Lastly, the model included effects of PSI on the RF; we found a negative significant effect of high PSI as a risk factor for lower RF scores (b = −0.34, [−1.1, −0.2]).

Discussion

Our study found that residents with psychiatric comorbidity sought advice from those who also have psychiatric comorbidity. The DST framework helps explain the behavioral changes of a social network that emerge from the collective actions of many interacting components (Mitchell et al., 2009). In this case, the collective actions refer to the tendency of residents with psychiatric comorbidity to seek advice from those who also have psychiatric comorbidity. This phenomenon is often referred to as a homophily - the tendency of those who are socially connected to display preferences towards others who have similarities across demographics (i.e., values, beliefs, experienced stigma) (Bobak, Majer, & Jason, 2021). Although advice seeking has been found to be related to higher stress and lower positive recovery outcomes (Jason et al., 2020), seeking advice from individuals who are more “recovered” (i.e., who have higher RF scores) has been related to beneficial recovery outcomes (Jason, Lynch, Bobak, Light, & Doogan, 2021). Findings in the present investigation draw attention to the importance of homophily effects when examining social dynamics; extending our understanding of social networks in terms of homophily with respect to abstinence social support (Majer et al., 2002).

Another finding involved the loaning network via the PSI alter effect. The significant “alter” effect - a sum of the loaning scores from all the alters that the ego can utilize for a loan at any given time – suggests that alters are more likely to loan money to a housemate with psychiatric comorbidity, giving rise to complex collective behavior patterns (loaning). It is possible that those with comorbid conditions are seen as needing more support or resources, and the fact that other residents are willing to share funds with the residents with a comorbid status might reflect just wanting to reach out to those who are vulnerable, which is what the ethos of Oxford House recovery homes embodies.

Another finding involves psychiatric comorbidity groups and RF scores. This negative relationship indicates that as PSI scores trend higher, RF scores tend to go lower. These results are consistent with other research that demonstrated inverted relationships between increases in PSI scores and decreases in quality-of-life scores (Bobak, Majer, & Jason, 2020), along with lower levels of drug taking confidence, hope, abstinence self-efficacy, and self-esteem (Abbinanti, Bobak, & Jason, 2022).

Lastly, findings of interest that are replicated from previous studies include significant RF score differences with respect to race (African American/Black) and sex (men vs. women). Jason et al. (2021) found that African American residents showed greater improvements on their RF scores when compared to the other races (Caucasian, Asian, American Indian, LatinX, Other). These findings are consistent with Harvey (2014), who found that African Americans, who were living in a recovery home at the time of data collection, had lower relapse rates than Non-Hispanic Caucasian counterparts. Additionally, men showed improved RF score outcomes that were significantly greater than their female counterparts. However, Davis and Jason (2005) found that social support characteristics are different for women than for men; where women tend to have higher levels of social support from unrelated friends, while men were more likely to endorse family members as their primary source of social support (Robles et al., 1998). In addition, Porcaro et al. (2020) found a significant negative relationship between PSI scores and coping resources among male Oxford House residents, but this relationship was positive among the female Oxford Houses. This could be explained by the previously mentioned tendencies of men to seek social support from their family members, which could make it more difficult for them to seek that support from their recovery home roommates.

This study may help explain the stigma surrounding psychiatrically comorbid Oxford House residents, which continues to be a barrier in how these individuals are provided treatment and opportunities to be re-integrated into community settings. Based on the literature reviewed in the introduction, those with SUDs and comorbid conditions are often not provided the types of community re-entry experiences that they need. Yet, what is often lacking is inexpensive but widely available support to help these at-risk individuals to have an ecologically positive setting to maintain their current abstinence. Our study found that psychiatrically comorbid residents’ recovery scores have a lower trajectory than others, but their trajectories are still positive over time. Based on prior work, we know psychiatrically comorbid Oxford House residents are more likely to go to other psychiatrically comorbid housemates when seeking advice. We also found that these higher risk residents are actually provided more loans from their roommates, an unexpected outcome. Thus, Oxford Houses appear to be a safe and inexpensive setting in which referrals can be made, even to those with comorbid conditions.

Findings from this study of Oxford House recovery homes may not generalize to other types of recovery homes that are run by (or have part-time professional) staff. Regarding psychiatric comorbidity, there exists a need to measure it more accurately with specialized diagnostic instruments. The current study used the psychiatric subscale of the Addiction Severity Index (McLellan et al., 1992), which only measures whether a person might have a psychiatric comorbidity but does not capture the specific mental disorder associated with the comorbidity. In other words, a score on the PSI does not indicate which mental disorder is comorbid with the SUD.

Overall, this investigation presents a comprehensive dynamic model of recovery home social networks that include interrelationships with advice seeking, loaning networks, latent recovery factor, and psychiatric severity ties. This study demonstrates that social embeddedness (i.e., an ego’s position in the relationship network) influences recovery related outcomes and has important implications for future studies to focus on the mechanisms that predict the formation of such ties. Specifically, this study demonstrates how social embeddedness (i.e., one’s position in a relationship network) affects recovery outcomes, loaning tendencies, and advice seeking behaviors for this population. Examining these complex processes is instrumental in understanding how long-term recovery is facilitated within these residential, community-based settings.

Acknowledgments

The authors appreciate the financial support from the National Institute on Alcohol Abuse and Alcoholism (grant number AA022763).

Contributor Information

Ted J. Bobak, Saint Xavier University

John M. Majer, Harry S. Truman College

Leonard A. Jason, DePaul University

References

  1. Aase DM, Jason LA, Ferrari JR, Li Y, & Scott G (2014). Comorbid mental health and substance abuse issues among individuals in recovery homes: Prospective environmental mediators. Mental health and substance use: dual diagnosis, 7(2), 170–183. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Abou-Saleh MT and Janca A (2004), The epidemiology of substance misuse and comorbid psychiatric disorders. Acta Neuropsychiatrica, 16: 3–8. 10.1111/j.1601-5215.2004.0075.x [DOI] [PubMed] [Google Scholar]
  3. American Psychiatric Association. (2022). Diagnostic and statistical manual of mental disorders (5th ed. text revision). Arlington, VA: American Psychiatric Association. [Google Scholar]
  4. Bates D, & Maechler M (2021). Matrix: Sparse and Dense Matrix Classes and Methods. R package version 1.3–4. https://CRAN.R-project.org/package=Matrix
  5. Bobak TJ, Majer JM, & Jason LA (2021). An examination of psychiatric severity and social cohesion outcomes within Oxford Houses. Community Mental Health Journal, DOI: 10.1007/s10597-021-00825-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Borgatti S, & Halgin D (2011). On Network Theory. DOI: 10.2139/ssrn.2260993 [DOI]
  7. Burns L, Teesson M and O’Neill K (2005), The impact of comorbid anxiety and depression on alcohol treatment outcomes. Addiction, 100: 787–796. 10.1111/j.1360-0443.2005.001069.x [DOI] [PubMed] [Google Scholar]
  8. Butts C,T (2020). sna: Tools for Social Network Analysis. R package version 2.6. https://CRAN.R-project.org/package=sna
  9. Butts C (2015). _network: Classes for Relational Data_. The Statnet Project (<URL: http://www.statnet.org>). R package version 1.13.0.1, <URL: https://CRAN.R-project.org/package=network>.
  10. Butts C (2008). “network: a Package for Managing Relational Data in R.” _Journal of Statistical Software_, *24*(2). <URL: https://www.jstatsoft.org/v24/i02/paper>. [Google Scholar]
  11. Cacciola JS, Alterman AI, McLellan AT, Lin Y-T, & Lynch KG. (2007). Initial evidence for the reliability and validity of a “Lite” version of the Addiction Severity Index. Drug and Alcohol Dependence, 87:297–302. [DOI] [PubMed] [Google Scholar]
  12. Cloud W, & Granfield R. (2008). Conceptualizing recovery capital: expansion of a theoretical construct. Subst Use Misuse, 43(12–13):1971–86. doi: 10.1080/10826080802289762. [DOI] [PubMed] [Google Scholar]
  13. Cohen S, Kamarck T, & Mermelstein R (1983). A global measure of perceived stress. Journal of Health and Social Behavior, 24, 385–396. [PubMed] [Google Scholar]
  14. Cohen S, & McKay G (1984). Social support, stress and the buffering hypothesis: A theoretical analysis. In Baum A, Taylor SE, & Singer JE (Eds.), Handbook of psychology and health (pp. 253–267). Hillsdale, NJ: Erlbaum. [Google Scholar]
  15. Cohen S, & Wills TA (1985). Stress, social support, and the buffering hypothesis. Psychological Bulletin, 98, 310–357. [PubMed] [Google Scholar]
  16. Coleman JS (1988). Social Capital in the Creation of Human Capital American Journal of Sociology, 94:, S95–S120 [Google Scholar]
  17. Cook KS, Cheshire C, Rice ERW, Nakagawa S (2013) Social Exchange Theory. In: DeLamater J, Ward A (eds) Handbook of Social Psychology. Handbooks of Sociology and Social Research. Springer, Dordrecht. 10.1007/978-94-007-6772-0_3 [DOI] [Google Scholar]
  18. Cook SH, Bauermeister JA, Gordon-Messer D, Zimmerman MA (2013). Online network influences on emerging adults’ alcohol and drug use. Journal of Youth and Adolescence, 42(11):1674–1686 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Csárdi G (2021). keyring: Access the System Credential Store from R. R package version 1.2.0. https://CRAN.R-project.org/package=keyring
  20. Daigre C, Grau-López L, Rodríguez-Cintas L, Ros-Cucurull E, Sorribes-Puertas M, Esculies O, Bones-Rocha K, & Roncero C (2017). The role of dual diagnosis in health-related quality of life among treatment-seeking patients in Spain. Quality of Life Research: An International Journal of Quality of Life Aspects of Treatment, Care & Rehabilitation, 26(12), 3201–3209. https://doi-org.ezproxy.depaul.edu/10.1007/s11136-017-1668-4 [DOI] [PubMed] [Google Scholar]
  21. De Bot K, Lowie W, & Verspoor M (2007). A Dynamic Systems Theory approach to second language acquisition. Bilingualism: Language and Cognition, 10(1), 7–21. doi: 10.1017/S1366728906002732 [DOI] [Google Scholar]
  22. Drake RE, Mueser KT, Brunette MF, & McHugo GJ (2004). A review of treatments for people with severe mental illnesses and co-occurring substance use disorders. Psychiatric Rehabilitation Journal, 27(4), 360–374. 10.2975/27.2004.360.374 [DOI] [PubMed] [Google Scholar]
  23. Dowle M, & Srinivasan A (2021). data.table: Extension of ‘data.frame’. R package version 1.14.0. https://CRAN.R-project.org/package=data.table
  24. DSM-5. (2013). Diagnostic and statistical manual of mental disorders. American Psychiatric Association; Washington, D.C. [Google Scholar]
  25. Enez Darcin A, Nurmedov S, Noyan CO, Yilmaz O, & Dilbaz N (2015). Psychiatric comorbidity among inpatients in an addiction clinic and its association with the process of addiction. Düşünen Adam: Journal of Psychiatry and Neurological Sciences, 28(3), 196–203. [Google Scholar]
  26. Fei, Teoh Bing. J, Yee A, & Habil M,H (2016). Psychiatric comorbidity among patients on methadone maintenance therapy and its influence on quality of life. American Journal on Addiction, 25(1):49–55. [DOI] [PubMed] [Google Scholar]
  27. Garcia-Rea EA, & LePage JP (2010). Reliability and validity of the World Health Organization Quality of Life: Brief version (WHOQOL-BREF) in a homeless substance dependent veteran population. Social Indicators Research, 99(2), 333–340. 10.1007/s11205-010-9583-x [DOI] [PubMed] [Google Scholar]
  28. Graham BC, Jason LA, & Ferrari JF (2009). Sense of community within recovery housing: Impact of resident age and income. In Jason LA, & Ferrari JR (Eds.). Recovery from addiction in communal living settings: The Oxford House model [Special Issue]. Journal of Groups in Addiction & Recovery, 4, 62–70. 10.1080/15560350802712405 [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Grant BF, Stinson FS, Dawson DA, et al. (2004). Prevalence and co-occurrence of substance use disorders and independent mood and anxiety disorders: Results from the National Epidemiologic Survey on Alcohol and Related Conditions. Arch Gen Psychiatry, 61(8):807–816. doi: 10.1001/archpsyc.61.8.807 [DOI] [PubMed] [Google Scholar]
  30. Hahm HC, Kolaczyk E, Jang J, Swenson T, & Bhindarwala AM (2012). Binge drinking trajectories from adolescence to young adulthood: The effects of peer social network. Substance Use & Misuse, 47(6), 745–756. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Harris PA, Taylor R, Thielke R, Payne J, Gonzalez N, Conde JG (2009). Research electronic data capture (REDCap) – A metadata-driven methodology and workflow process for providing translational research informatics support, J Biomed Inform;42(2):377–81. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Haverfield MC, Ilgen M, Schmidt E, Shelley A, Timko C (2019). Social support networks and symptom severity among patients with co-occurring mental health and substance use disorders. Community Ment. Health J, 55 (2019), pp. 768–776, 10.1007/s10597-019-00396-7 [DOI] [PubMed] [Google Scholar]
  33. Hlebec V, Ferligoj A. Reliability of Social Network Measurement Instruments. Field Methods. 2002;14(3):288–306. doi: 10.1177/15222X014003003 [DOI] [Google Scholar]
  34. Homans GC (1958). Social behavior as exchange. American Journal of Sociology, 63, 597–606. 10.1086/222355 [DOI] [Google Scholar]
  35. Houchin K and MacLean D (2005), Complexity Theory and strategic change: An empirically informed critique*. British Journal of Management, 16: 149–166. 10.1111/j.1467-8551.2005.00427.x [DOI] [Google Scholar]
  36. Iannone R, & Cheng J (2020). blastula: Easily Send HTML Email Messages. R package version 0.3.2. https://CRAN.R-project.org/package=blastula
  37. Jason LA, Wiedbusch E, Bobak T, & Taullahu D (2020). Estimating the number of substance use disorder recovery homes in the United States. Alcoholism Treatment Quarterly, 38(4), 506 514. DOI: 10.1080/07347324.2020.1760756 [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Jason LA, Davis MI, Ferrari JR, & Anderson E (2007). Longitudinal analysis of Oxford House. Addictive Behaviors, 32(4), 803–818. [DOI] [PubMed] [Google Scholar]
  39. Jason LA, & Ferrari JR (2010). Oxford House recovery homes: Characteristics and effectiveness. Psychological Services, 7, 92–102. DOI: 10.1037/a0017932 [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Jason LA, Stevens E, Ferrari JR, Thompson E, & Legler R (2012). Social networks among residents in recovery homes. Advances in Psychology Study, 1, 4–12. [PMC free article] [PubMed] [Google Scholar]
  41. Jason LA, Guerrero M, Lynch G, Stevens E, Salomon-Amend M, & Light JN (2020). Recovery home networks as social capital. Journal of Community Psychology, 48, 645–657. 10.1002/jcop.22277 [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Jason LA, Guerrero M, Bobak T, Light JM, & Stoolmiller M (2021). Reducing health disparities among Black individuals in the post-treatment environment. Journal of Ethnicity in Substance Abuse. 10.1080/15332640.2020.1861497 [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Jason LA, Guerrero M, Salomon-Amend M, Stevens E, Light JN & Stoolmiller M (2021). Context matters: Home-level but not individual-level recovery social capital predict residents’ relapse. American Journal of Community Psychology, 67(3–4), 392–404. 10.1002/ajcp.12481 [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Jason LA, Guerrero M, Salomon-Amend M, Lynch G, Stevens EB, Light JN, Stoolmiller M, & Doogan NJ (2021). Network measures of advice-seeking and resource sharing are related to well-being in recovery homes. International Journal of Drug Policy, 92, 10.1016/j.drugpo.2020.102970 [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Jason LA, & Stevens E (2017). The reliability and reciprocity of a social network measure. Alcoholism Treatment Quarterly, 35, 317–327. DOI: 10.1080/07347324.2017.1355220 [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Jason LA, Stevens E & Ram D (2015) Development of a three-factor psychological sense of community scale. Journal of Community Psychology, 43 973–985. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Kassanits J, Bobak T, Stevens E, Guerrero M, Light J, & Jason LA (2020). The relationship of Oxford Houses across heterogeneous house and setting characteristics. American Journal of Orthopsychiatry 90(3), 324–327. DOI: 10.1037/ort0000437 [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Kelly JF, Hoeppner B, Stout RL, Pagano M. (2012). Determining the relative importance of the mechanisms of behavior change within Alcoholics Anonymous: a multiple mediator analysis. Addiction, 107(2):289–99. doi: 10.1111/j.1360-0443.2011.03593.x. Epub 2011 Sep 19. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Kelly TM, & Daley DC, (2013). Integrated treatment of substance use and psychiatric disorders, Social Work in Public Health, 28:3–4, 388–406, DOI: 10.1080/19371918.2013.774673 [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Kushner MG, Abrams K, Thuras P, Hanson KL, Brekke M and Sletten S (2005). Follow-up Study of Anxiety Disorder and Alcohol Dependence in Comorbid Alcoholism Treatment Patients. Alcoholism: Clinical and Experimental Research, 29: 1432–1443. 10.1097/01.alc.0000175072.17623.f8 [DOI] [PubMed] [Google Scholar]
  51. Laudet AB, Magura S, Vogel HS, & Knight E (2000). Recovery challenges among dually diagnosed individuals. Journal of Substance Abuse Treatment, 18(4), 321–329. 10.1016/S0740-5472(99)00077-X [DOI] [PubMed] [Google Scholar]
  52. Laudet AB, & White W,L (2008). Recovery capital as prospective predictor of sustained recovery, life satisfaction, and stress among former polysubstance users. Substance Use Misuse, 43:27–54. [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Light J, & Doogan N (NA). RHNetTools: Data Management Tools for RH-Systems Project. R package version 2.1.3.
  54. Longabaugh R, Wirtz PW, Zywiak WH, O’Malley SS. (2010). Network support as a prognostic indicator of drinking outcomes: the COMBINE Study. J Stud Alcohol Drugs, 71(6):837–46. doi: 10.15288/jsad.2010.71.837. [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Majer JM, Jason LA, & Bobak TJ (2021). An examination of abstinence social support among recovery home residents with psychiatric comorbidity. Drug and Alcohol Dependence. 10.1016/j.drugalcdep.2021.108971 [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Majer JM, Chapman HM, & Jason LA (2016). Comparative analysis of treatment conditions upon psychiatric severity levels at two years among justice involved persons. Advances in Dual Diagnosis, 9(1), 38–47. DOI: 10.1108/ADD-07-2015-0015 [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Majer JM, Jason LA, Aase DM, Droege JR, & Ferrari JR (2013). Categorical 12-step involvement and continuous abstinence at two-years. Journal of Substance Abuse Treatment, 44, 46–51. DOI: 10.1016/j.jsat.2012.03.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Majer JM, Jason LA, Ferrari JR, & North CS (2002a). Comorbidity among Oxford House residents: A preliminary outcome study. Addictive Behaviors, 27(5), 837–845. https://doi-org.ezproxy.depaul.edu/10.1016/S0306-4603(01)00214-3 [DOI] [PubMed] [Google Scholar]
  59. Majer JM, Jason LA, Ferrari JR, Venable LB & Olson BD (2002b). Social support and self-efficacy for abstinence: Is peer identification an issue? Journal of Substance Abuse Treatment, 23, 209–215. [DOI] [PubMed] [Google Scholar]
  60. McLellan A,T, Cacciola J,S, & Zanis D (1997). The Addiction Severity Index-Lite. Center for the Studies on Addiction. University of Pennsylvania/Philadelphia, VA Medical Center. [Google Scholar]
  61. McLellan AT, Luborsky L, Woody GE, O’Brien CP, & Druley KA.(1983). Predicting response to alcohol and drug abuse treatments. Archives of General Psychiatry, 40:620–625. [DOI] [PubMed] [Google Scholar]
  62. Meyers RJ, & Miller WR (2001). A community reinforcement approach to addiction treatment. Cambridge University Press. 10.1017/CBO9780511570117 [DOI] [PubMed] [Google Scholar]
  63. Mitchell A, Romano G, Groisman B et al. Adaptive prediction of environmental changes by microorganisms. Nature 460, 220–224 (2009). 10.1038/nature08112 [DOI] [PubMed] [Google Scholar]
  64. Moos RH, & Moos BS (2006). Rates and predictors of relapse after natural and treated remission from alcohol use disorders. Addiction, 101(2), 212–222. [DOI] [PMC free article] [PubMed] [Google Scholar]
  65. Morisano D, Babor TF, & Robaina KA (2014). Co-occurrence of substance use disorders with other psychiatric disorders: Implications for treatment services. Nordic Studies on Alcohol and Drugs, 31(1), 5–25. [Google Scholar]
  66. Ouimette PC, Finney JW, and Moos RH (1999) Two-year posttreatment functioning and coping of substance abuse patients with posttraumatic stress disorder. Psychology of Addictive Behaviors, 13(2):105–114. [Google Scholar]
  67. Oxford House, (2020). Annual Report, Fiscal Year 2019. Silver Spring, Md. [Google Scholar]
  68. Polcin DL. (2009). A model for sober housing during outpatient treatment. Journal of Psychoactive Drugs, 41(2):153–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
  69. R Core Team (2021). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/. [Google Scholar]
  70. Regier DA, Farmer ME, Rae DS, et al. (1990). Comorbidity of mental disorders with alcohol and other drug abuse: Results from the epidemiologic catchment area (ECA) Study. JAMA, 264(19):2511–2518. doi: 10.1001/jama.1990.03450190043026 [DOI] [PubMed] [Google Scholar]
  71. Ripley R,M, Snijders T,A,B, Boda Z, Voros A, & Preciado P (2020). Manual for SIENA version 4.0. Oxford: University of Oxford, Department of Statistics; Nuffield College. [Google Scholar]
  72. Ritsher JB, McKellar JD, Finney JW, Otilingam PG, & Moos RH (2002). Psychiatric comorbidity, continuing care and mutual help as predictors of five-year remission from substance use disorders. Journal of Studies on Alcohol, 63, 709–715. [DOI] [PubMed] [Google Scholar]
  73. Rosenberg M (1965). Society and the adolescent self-image. Princeton, NJ: Princeton University Press. [Google Scholar]
  74. SAMHSA (2011). SAMHSA announces a working definition of “recovery” from mental disorders and substance use disorders. Available at: https://www.samhsa.gov/newsroom/press-announcements/201112220800
  75. SAMHSA (2018). Recovery housing: Best practices and suggested guidelines. Available at: https://www.samhsa.gov/sites/default/files/housing-best-practices-100819.pdf
  76. Schaefer JA, Cronkite RC, & Hu KU (2011). Differential relationships between continuity of care practices, engagement in continuing care, and abstinence among subgroups of patients with substance use and psychiatric disorders. Journal of Studies on Alcohol and Drugs, 72, 611–621. [DOI] [PubMed] [Google Scholar]
  77. Scoones L (1999). New ecology and the social sciences: What prospects for a fruitful engagement? Annual Review of Anthropology, 28:1, 479–507 [Google Scholar]
  78. Snijders TAB, van de Bunt GG, & Steglich CEG (2010). Introduction to stochastic actor-based models for network dynamics. Social Networks, 32(1), 44–60. 10.1016/j.socnet.2009.02.004 [DOI] [Google Scholar]
  79. Snyder CR, Sympson SC, Ybasco FC, Borders TF, Babyak MA & Higgins RL (1996) Development and validation of the State Hope Scale. Journal of Personality and Social Psychology, 70 321–335. [DOI] [PubMed] [Google Scholar]
  80. Sklar SM, Annis HM, Turner NE. (1999). Group comparisons of coping self-efficacy between alcohol and cocaine abusers seeking treatment. Psychology of Addictive Behaviors, 13:123–133. [Google Scholar]
  81. Stevens EB, Jason LA, Ferrari JR, & Hunter B (2010). Self-efficacy and sense of community among adults recovering from substance abuse. North American Journal of Psychology, 12, 255–264. [PMC free article] [PubMed] [Google Scholar]
  82. Stoebenau K, Valente TW. (2003). Using network analysis to understand community-based programs: a case study from highland Madagascar. Int Fam Plan Perspect, 29(4):167–73. doi: 10.1363/ifpp.29.167.03. [DOI] [PubMed] [Google Scholar]
  83. Thelen E, Ulrich B, & Wolff P (1991). Hidden Skills: A Dynamic Systems Analysis of Treadmill Stepping during the First Year. Monographs of the Society for Research in Child Development, 56(1), I–103. doi: 10.2307/1166099 [DOI] [PubMed] [Google Scholar]
  84. The WHOQOL Group. (1996). The World Health Organization quality of life assessment (WHOQOL): Development and general psychometric properties. [DOI] [PubMed]
  85. Tierney N, Cook D, McBain M, & Fay C (2021). naniar: Data Structures, Summaries, and Visualisations for Missing Data. R package version 0.6.1. https://CRAN.R-project.org/package=naniar
  86. Torrens M, Rossi CP, Martinez-Riera R, Martinez-Sanvisens D, Bulbena A (2012). Psychiatric co-morbidity and substance use disorders: Treatment in parallel systems or in one integrated system? Substance Use and Misuse, 47 (8–9):1005–1014. [DOI] [PubMed] [Google Scholar]
  87. Walker M, Wasserman S, & Wellman B (1993). Statistical models for social support networks. Sociological Methods & Research, 22(1):71–98. doi: 10.1177/0049124193022001004 [DOI] [Google Scholar]
  88. Wickham H, & Miller E (2021). haven: Import and Export ‘SPSS’, ‘Stata’ and ‘SAS’ Files. R package version 2.4.1. https://CRAN.R-project.org/package=haven
  89. Wickham H (2021). dtplyr: Data Table Back-End for ‘dplyr’. R package version 1.1.0. https://CRAN.R-project.org/package=dtplyr
  90. Wickham et al. (2019). Welcome to the tidyverse. Journal of Open Source Software, 4(43), 1686, 10.21105/joss.01686 [DOI] [Google Scholar]
  91. Winder N 2007. Innovation and metastability: a systems model. Ecology and Society 12(2): 28. DOI: http://www.ecologyandsociety.org/vol12/iss2/art28/ [Google Scholar]

RESOURCES