Resident and House Manager Perceptions of Social Environments in Sober Living Houses: Associations with Length of Stay

Abstract

Aims:

Studies have shown persons living in recovery homes for drug and alcohol problems make significant, sustained improvements. However, there is limited information about factors associated with outcome. This study examined how perceptions of social environment of one type of recovery home, sober living houses (SLHs), were associated with length of stay (LOS).

Methods:

SLH residents and their house managers (N=416) completed the Recovery Home Environment Scale (RHES) that assessed social model recovery characteristics and the Community-Oriented Program Evaluation Scale (CPES) that evaluated perceptions of the program environment.

Results:

Scales completed by residents predicted LOS, but those completed by house managers did not. Larger discrepancies between the two groups were associated with shorter LOS. The RHES was shown to be a stronger predicter of LOS than the CPES.

Conclusion:

Results highlight the importance of the social environment in SLHs, particularly those most closely aligned with social model recovery principles.

1. Introduction

The Substance Abuse and Mental Health Services Administration defines recovery houses (RHs) as safe, healthy, family-like substance-free living environments that support individuals in recovery from addiction (SAMHSA, 2019). Sober living houses (SLHs) are a type of RH that evolved from Alcoholics Anonymous members in California wanting to provide alcohol- and drug-free housing and social support to persons with substance use disorders (Wittman & Polcin, 2014;Mericle, Mahoney, Korcha, Delucchi, & Polcin, 2019). They are typically run by a house manager that is also in recovery, lives on-site, and is compensated in some manner, such as reduced rent. Although studies of SLHs are limited, several investigations showed that residents of these houses make significant improvements in a number of areas, including substance use, employment, arrests, and psychiatric symptoms (Bergman et al., 2015; Polcin, Korcha, Bond, & Galloway, 2010). Importantly, studies show improvements are largely maintained over 18 months even though most residents leave the houses by that time. Rather than relying on professional interventions (e.g., individual counseling, group counseling, and case management), SLHs use a social model approach to recovery (Borkman, Kaskutas, & Barrows, 1996) that emphasizes characteristics of the social environment, including peer support and peer involvement in house operations and decisions. However, limited attention has been devoted to understanding the extent to which social environments in SLHs reflect principles of social model recovery or how they are related to outcome.

1.1. Social Model Philosophy Scale

One of the earliest studies of social model recovery resulted in the development of the Social Model Philosophy Scale (SMPS) (Kaskutas, Greenfield, Borkman, & Room, 1998). The SMPS was designed to describe the physical characteristics, recovery philosophy, and operational structures of substance abuse programs. The scale measures the extent to which the design of programs adheres to a social model approach to recovery using six subscales: physical environment, staff role, authority base, view of dealing with substance abuse problems, governance, and community orientation. Although the scale provides an overall social model score, Mericle and colleagues (Mericle, Miles, Cacciola, & Howell, 2014) found wide variation of subscale scores among a sample of recovery homes in Philadelphia. For example, most directors or managers rated their homes high on having a 12-step-oriented recovery philosophy and home-like physical environment, but low on peer governance.

Using the SMPS, data are generated by interviewing program directors or managers who oversee delivery of services. Missing in these assessments are the perceptions and experiences of the persons receiving services. The SMPS was designed to reflect how programs are designed, not what actually occurs in terms of social model activities and behaviors among persons in the program.

1.2. Assessing Social Environments

Many issues need attention in studies of recovery homes, including how characteristics such as house capacity, gender of residents, and association with formal treatment programs are associated with outcome. In one of the few studies addressing these issues, Mericle et al. (2019) found houses that were part of a larger group of houses under one organization had increased odds of alcohol and drug abstinence. Houses that had an affiliation with a treatment program and were smaller had higher numbers of residents being employed.

Studies assessing social environment characteristics in recovery homes have been even more sparse, but a few studies have looked at social environments in related settings, including substance abuse and mental health treatment programs. One of the earliest studies to compare these client and staff views of a program’s social climate used the Community Oriented Program Evaluation Scale (CPES, formerly known as COPES) (Moos, 1997). Moos had originally developed this scale with the theoretical framework that the social climate evolves as a result from the patients’ personal factors (health status, functioning, preferences) interacting with program’s characteristics (context, design, policies, services) (Moos, 1972). Ten subscales measuring diverse aspects of the social environment were divided into three overall dimensions: Relationship, Personal Growth, and System Maintenance. Comparisons were made between staff and participant ratings across all three dimensions. On subscales related to the Relationship dimension (e.g., Involvement and Support), staff and client perceptions in the Moos (1997) study were similar. However, relative to clients, staff ratings were higher on scales related to Personal Growth opportunities (e.g., Practical Orientation and Personal Problem Orientation subscales) and relative to clients, they felt programs had clearer expectations and less staff control (Order and Staff Control subscales).

There are a number of limitations to this study. First, Moos did not describe how perceptions of the social environment for staff or clients were associated with outcomes. Second, there was no assessment of potential effects of client and staff differences on outcomes. Finally, the samples studied focused on mental health and substance abuse treatment programs. Other settings, such as residential recovery homes that use a social model approach, may have different social climates.

A few recent studies used the CPES to collect data about social environment characteristics from recovery home residents. One study (Harvey & Jason, 2011) used the CPES to collect data from two types of RHs, Oxford Houses (OH) and therapeutic community (TCs) treatment programs. TCs are a common type of RH that are usually run by professional staff as residential addiction treatment programs. The capacity and philosophy may vary, such as their emphasis on 12-step models. Like SLHs, OHs employ mutual self-help principles and tend to have a smaller capacity. OHs have no paid staff and are completely run by the residents based off of the OH manual. Reliability was problematic for most of the CPES scales (Cronbach’s alpha <0.70). However, relative to the therapeutic community program, OH residents were found to have a higher level of Involvement. Another study of recovery home social environments used the CPES to examine resident perceptions of social environments in SLHs in California (Polcin, Mahoney, Witbrodt. & Mericle, 2020). Four of the CPES subscales had adequate levels of internal consistency and were found to be associated with recovery capital: Involvement, Support, Practical Orientation, and Order and Organization.

In addition, a scale designed to measure social model dynamics in the houses, the Recovery Home Environment Scale (RHES) was also associated with higher recovery capital. Polcin, Mahoney, & Mericle (2021) provides further information on the psychometric properties of this scale. Items on the RHES are based on social model theory (Borkman, Kaskutas, Room, Bryan, & Barrows, 1998; Wright, Mora, & Hughes, 1990). From a social model perspective, alcohol and drug problems are centered in reciprocal relationships between the individual and their surrounding environment. SLH service providers are therefore encouraged to focus on facilitating a culture of peer support for recovery, emotional support, empowerment of residents in decision making, attendance at 12-step recovery groups, and the practice of 12-step recovery principles during daily interactions (Wittman & Polcin, 2014).

Recovery capital assesses the physical, environmental, economic, social, and psychological assets a person possesses with the theory being that these resources can help individuals in recovery from substance issues. Rather than focusing on an intervention as a path to recovery, recovery capital looks at how the individual, their support system, and community can also impact their recovery. Recovery housing allows a stable living environment for residents to further develop their recovery capital (Jason, Guerrero, Salomon-Amend, Light & Stoolmiller, 2021). The social model speaks to how substance use service settings can be structured and organized to enhance recovery outcomes. The RHES measures the social model strategies that can develop recovery capital (Polcin, Mahoney, Witbrodt, & Mericle, 2020).

The RHES assesses resident perceptions about the extent to which these core principles of social model recovery are present, whereas the CPES measures the experiences of the residents in their social environment. The development of recovery capital could be viewed as a later outcome of social climate data in Moos’ model that he labeled “patients’ community adaptation (health status, social skills, functioning” (Moos, 1997). Despite the contributions of this work, several gaps in knowledge about RH environments remain, particularly with respect to differences between resident and manager perceptions, and how perceptions of recovery housing environments are related to resident outcomes.

1.3. Purpose

The overall goal of the current study was to investigate a broad view of social environment characteristics and their associations with length of stay (LOS) in SLHs. LOS was selected as the primary outcome because we surmised that characteristics of the social environment would impact the quality of residents’ experiences and their motivation to remain in the house. The importance of LOS has been supported by studies showing it is related to subsequent substance use (Polcin, Korcha, Gupta, Subbaraman, & Mericle, 2016).

We aimed to understand LOS in terms of its association with different aspects of the social environment in the houses from different perspectives and using different measures of it. These included views of the social environment among residents and house managers. We proposed two hypotheses. First, we expected more favorable perceptions about the social environment from both managers and residents would be associated with longer LOS. Based on a previous study showing associations between social environments in SLHs and recovery capital (Polcin, Mahoney, Witbrodt, & Mericle, 2020), we hypothesized CPES subscales measuring Support, Involvement, a Practical Orientation, and Order and Organization would be associated with longer lengths of stay. An additional hypothesis suggested that the RHES, which was designed to assess social model characteristics in the houses, would predict longer lengths of stay. We expected these associations to be the case for both managers and residents. A second hypothesis suggested managers would view social environments more favorably than residents and that higher levels of discrepancies between managers and residents would be associated with lower LOS. The rationale was that discordant perceptions between managers and residents might reflect higher levels of conflict and inconsistencies that could have detrimental effects on resident LOS.

2. Methods

2.1. Study Sites and Participants

Study participants consisted of persons residing in 45 SLHs in Los Angeles. Houses were recruited that reflected a diverse sample. SLHs in the study included 22 houses that were for men, 11 for women, and 12 for all genders. The U.S. Census Bureau’s American Community Surveys (ACS) were the primary source of data on neighborhood sociodemographic characteristics. These data are available for small areas such as Census tracts, which are the basic units we used to represent the “neighborhood” for each SLH. Houses were selected to include low (26.67%), medium (48.89%) and high (24.44%) socioeconomic statuses (SES) of the neighborhood in which the house is located. All houses were members of the Sober Living Network (SLN), an association of SLHs that provides certification to houses that comply with standards for health, safety, good neighbor relations, and good business practices. We did not approach houses that included children, houses that had less than 6 beds, houses that had more than 25 beds, and houses that had advertised fees that were over $4,500 per month.

To provide a broad depiction of all residents entering the houses and maximize generalization, we employed very few inclusion/exclusion criteria. Inclusion criteria included age 18 or older, able to provide informed consent, completed a baseline interview, and had been a resident in the house for less than one month to ensure that participants were new residents. If participants were still at the SLH 30 days after their entry, they were asked to complete the assessments of their perceptions of the social environment. The one-month time period allowed for residents to become familiar with the social environment and develop well-informed views All house managers of the homes participating in the study were contacted by phone and invited to participate. Interviews took place in-person at the houses or by phone.

2.2. Procedures

Houses were contacted using information obtained from the SLN. Residents entering the houses were recruited by experienced research interviewers primarily by phone within one month of entering the SLHs. Baseline assessments were conducted on average 16.47 days (sd=9.02) after entering the house. For the current study, baseline data included demographic information. One month after recruitment, residents participated in an interview assessing their perceptions of the social environment. Six months after the baseline interview, participants were contacted to complete a follow-up interview, whether or not they were still a resident at the SLH, as part of an intent-to-treat model. Because managers were responsible for oversight at the homes and nearly all had been in their roles for over one month (the range was from 27 days to 15 years) (Polcin, Mahoney, & Mericle, 2020), they were eligible to be interviewed at any point after their house entered the study. All study procedures were approved by the Public Health Institute IRB.

2.3. Measures

Demographics for residents and managers included gender and race.

House measures included SES of the local neighborhood, number of beds (house capacity), and gender of the house.

LOS (LOS) was calculated as the number of days in the house reported by the participant.

Perceptions of The Social Environment:

1. The Community Oriented Program Evaluation Scale (CPES, formerly COPES) (Moos, 1997) was used to assess manager and resident perceptions of the social environment in SLHs. The CPES consists of 100 items designed to assess social environments in programs for persons with substance use and mental health disorders. Ten areas of the program’s social environment are assessed: Involvement, Support, Spontaneity, Autonomy, Practical Orientation, Personal Problem Orientation, Anger and Aggression, Order and Organization, Program Clarity, and Staff Control. Items are scored true or false and subscale scores range from 0 −10. Psychometrics of the CPES have varied depending on the population and service setting. For example, Moos (1997) reported acceptable levels of internal consistency in his work assessing community-based substance abuse and mental health treatment programs. The average alphas for client ratings across subscales was 0.79 and for staff it was 0.78. However, in a study of Oxford Houses (Harvey & Jason, 2011), eight of the ten subscales had unacceptable levels of internal consistency (Cronbach’s alpha <0.70). For these analyses, we used a standard cutoff level of 0.70 which resulted in four subscales with acceptable levels of internal consistency: Involvement (0.81), Support (0.78), Practical Orientation (0.79), and Order and Organization (0.77). We dropped the other subscales, where alphas ranged from 0.38 to 0.67.

2. The Recovery Home Environment Scale (RHES) (Polcin, Mahoney, & Mericle, 2021) is a new measure designed to assess perceptions of the SLH social environment that contribute to social model recovery. Eight scale items assess resident interactions relevant to social model recovery, including social support for recovery, integration of 12-step work into daily house interactions, general and recovery-oriented helping behaviors, perceptions of the effectiveness of house meetings, and empowerment of residents in decision making. Items are rated on a 5-point Likert scale ranging from “not at all” to “a lot.” A total mean score is calculated. Psychometric properties of the RHES include principal components analysis, which showed the scale is largely unidimensional. The one factor for the total scale contained an Eigen value greater than 1, which comprised 61% of the variance among the eight RHES items. Internal consistency of eight items was strong (alpha=.90). Construct validity was supported by correlations between the RHES and subscales scores on the CPES. The RHES was positively associated with the positive social environment characteristics on the CPES, including Involvement (r=0.66, p<0.001) and Support (r=0.63, p<0.001), but negatively associated with detrimental characteristics, such as the Anger and Aggression subscale (r=−0.191, p<0.01). Regression models to demonstrate predictive validity showed the RHES was positively associated with subsequent LOS (Beta=2.81, p=0.002) and negatively associated with subsequent number days of alcohol or drug use (Beta=−0.64, p=0.035) (Polcin, Mahoney & Mericle, 2021).

2.4. Analyses

Data were analyzed using SPSS. Analyses began with descriptive statistics depicting characteristics of houses, residents, house managers, and perceptions of the social environment. T-tests for independent means then compared CPES subscales and the RHES scores between each resident and their house manager. The magnitude of differences between house managers and residents in their perceptions of the social environment were also assessed by creating difference variables for each scale. Rather than calculating the difference between overall scores for the house manager and resident, the difference variable was the absolute sum of the differences for each question on that scale. These differences between the perceptions of house managers and residents were calculated to examine whether larger discordances were associated with shorter lengths of stay. Initial assessments of how social environment characteristics related to LOS were conducted using Pearson correlations. Correlations assessed how resident perceptions of the social environment, house manager perceptions of the social environment, and manager- resident differences in perceptions were associated with LOS.

These findings then informed the development of multivariable regression models testing various aspects of social environment influences on LOS controlling for demographic characteristics: 1) resident and manager perceptions of the social environment, 2) how differences in perceptions between the two groups predicted LOS, and 3) the relative influence of the RHES to predict LOS when CPES subscales were entered into the same models. Additional multivariate models examined the relative strength of the RHES to predicted LOS when CPES subscales were included in models.

Additional regression models were constructed to assess the relative strength of the RHES to predict LOS(LOS) controlling for demographics and subscales on the CPES. The RHES was designed to assesses social model characteristic in SLHs, which is the explicit approach to recovery used in SLHs. In contrast, the broader, more generic constructs assessed on the CPES were initially designed to assess a range of social environments in mental health and substance abuse treatment programs. We therefore wanted to explore the relative strengths of the different scales to predicts LOS. Separate models were created that included the RHES and each for the for subscale on the CPES.

3. Results

3.1. Demographics

Resident characteristics (n=371) included 37.2% who were women. The racial distribution was 52.6% white, 25.1% Hispanic/Latino, and 16.2% African American. The mean age was 40.23 (sd=12.49). Among house managers (n=45), 28.9% were women. The racial distribution was 51.1% white, 28.9% African American, and 17.8% Hispanic/Latino. Previous analyses (Polcin, Mahoney, & Mericle, 2020) showed the median amount of time managers lived in the house or served as the house managers was 2.9 years and ranged widely, from 27 days to 15 years. Similarly, there was variation in the total amount of time managers spent at the houses per week, which ranged from 20 to 168 hours (median=60).

3.2. Bivariate Analyses

T-tests for independent means were conducted to compare how perceptions of the social environment differed between managers and residents. Table 1 shows that significant mean differences were noted on three of the four CPES scales and the RHES. As hypothesized, managers rated CPES subscales, including Support, Involvement, and Practical Orientation, and the RHES higher than residents. All of these social environment measures showed significant differences at p<0.010. Although managers also rated the Order and Organization subscale higher than residents, the difference did not reach statistical significance (p=0.139).

Table 1.

T-Tests Comparisons of Perceptions of Social Environment Scores by Sober Living House Role

Social Environment Measure	Resident Mean (sd) n=371	House manager Mean (sd) n=45	p-value of t-test
CPES Scale: Support	49.80 (11.8)	57.51 (8.2)	0.001
CPES Scale: Involvement	50.75 (11.5)	57.05 (10.1)	0.001
CPES Scale: Practical Orientation	46.95 (13.5)	54.12 (11.3)	0.001
CPES Scale: Order and Organization	52.35 (12.3)	55.19 (7.9)	0.139
RHES Social	26.66 (8.0)	30.40 (5.2)	0.003

The 371 residents who participated in the study stayed in the homes an average of 128.19 days (sd=56.00). However, number of days varied widely, from 2 to 180. Since we were looking at 6-month outcomes, the maximum number of LOS days was set to 180. Table 2 shows Pearson correlations between social environment scales and LOS. Correlations are shown separately for residents and managers. Although coefficients were relatively small, ranging from .16 to .22, resident ratings for all for all four of the CPES subscales and the RHES had significant relationships with LOS. None of the scales for the house managers had significant relationships with LOS. We also calculated differences between manager and resident ratings on the social environment scales and tested whether larger differences between the two groups were associated with shorter lengths of stay. As hypothesized, larger differences between managers and residents on three scales showed negative associations with LOS, Involvement (r=−0.103, p<.05), Order and Organization (r=−0.172, p<0.001), and the RHES (r=−0.22, p<0.001).

Table 2.

Correlations for Length of Stay and Perceptions of Sober Living House Social Environment

Social Environment Measure	Coef.	p-val
CPES Resident Subscales
CPES Support	0.156	0.003
CPES Involvement	0.211	0.001
CPES Practical Orientation	0.159	0.002
CPES Order and Organization	0.184	0.001
RHES Social Resident	0.222	0.001
CPES House Manager
CPES Support	0.020	0.697
CPES Involvement	0.072	0.166
CPES Practical Orientation	0.010	0.853
CPES Order and Organization	0.061	0.241
RHES Social House Manager	0.080	0.129
Differences between Resident and House Manager on CPES
CPES Support	−0.052	0.327
CPES Involvement	−0.103	0.049
CPES Practical Orientation	−0.004	0.943
CPES Order and Organization	−0.172	0.001
Differences for RHES Social between Resident and House Manager	−0.220	0.001

3.3. Multivariate Analyses

As Table 3 indicates, resident perceptions of the social environment were consistent predictors of LOS controlling for demographic characteristics and perceptions of managers. The table shows that all five social environment scales for residents predicted LOS. Betas ranged from 0.73 (SE=0.22, p<.01) on Practical Orientation to 1.69 (SE=0.39, p<.001) on the RHES. None of the social environment scales from the perspective of house managers were significant.

Table 3.

Association with Length-of-Stay for Sober Living House Resident and House Manager Perceptions on RHES and CPES Subscales

	Beta (SE)
CPES Support
Resident	*** 0.89 (0.25)
Manager	−0.10 (0.37)
CPES Practical Orientation
Resident	** 0.73 (0.22)
Manager	−0.06 (0.26)
CPES Order and Orientation
Resident	*** 0.94 (0.25)
Manager	−0.07 (0.43)
CPES Involvement
Resident	*** 1.07 (0.26)
Manager	0.10 (0.30)
RHES Social Total Score
Resident	*** 1.69 (0.39)
Manager	0.05 (0.56)

^***p<.001

^**p<.01

^*p≤.05

Note. All models shown above controlled for age, sex, ethnicity, income, house capacity, and house gender.

Table 4 shows shorter lengths of stay were predicted by larger differences between house manager and resident ratings on three social environment scales: Order and Organization (beta=−4.95, SE=1.65, p<.01), Involvement (beta=−3.51, SE=1.90, p<.10), and the RHES (beta=−2.47, SE=0.60, p<.001). Differences on Practical Orientation and Support were not associated with LOS.

Table 4.

Regressions for Length of Stay and the Sums of Absolute Differences between Resident and House Manager Perceptions

	beta	(SE)	p-val
CPES Support	−1.678	1.826	0.359
CPES Order and Organization	−4.948	1.654	0.003
CPES Involvement	−3.508	1.899	0.066
CPES Practical Orientation	1.169	1.787	0.513
RHES Social Total	−2.469	0.595	0.000

Note. All models shown above controlled for age, sex, ethnicity, income, house capacity, and house gender.

Table 5 shows that in each of the four models the RHES was a strong predictor of LOS. Beta ranged from 1.14 (SE=0.47, p<.015) when Involvement was included in the model to 1.46 (SE=0.46, p<.002) when Practical Orientation was in the model. When included in models with the RHES, none of the CPES subscales were significant predictors, although Involvement was a statistical trend (beta=0.54, SE=0.32, p<.10).

Table 5.

Length of Stay and Resident Social Environment Measures Jointly Entered into Four Multivariate Models

	beta	(SE)	p
Model 1
RHES Social Total score	1.139	0.468	0.015
CPES Involvement (I) Subscale	0.540	0.323	0.096
Model 2
RHES Social Total score	1.464	0.459	0.002
CPES Practical Orientation (PO) Subscale	0.166	0.275	0.545
Model 3
RHES Social Total score	1.252	0.432	0.004
CPES Order and organization (OO) Subscale	0.446	0.277	0.108
Model 4
RHES Social Total score	1.418	0.471	0.003
CPES Support (S) Subscale	0.219	0.304	0.472

Note. All four models shown above controlled for age, sex, ethnicity, income, house capacity & house gender

4. Discussion

SLHs and other types of recovery homes use a social model approach to recovery that emphasizes peer support, 12-step recovery principles, and resident involvement in decisions that affect house operations. This study assessed whether measures of the social environment at sober living houses was associated with longer retention. Previous studies have shown longer lengths of stay in SLHs are important indicators of subsequent substance use outcome (Polcin et al., 2016). We also assessed whether broader measures of the social environment that were designed to assess a variety of mental health and substance abuse programs were associated with LOS. These included subscales on the CPES: Support, Involvement, Practical Orientation, and Order and Organization. The social environment was assessed from the perspectives of residents and SLH managers.

With the exception of Order and Organization, there were significant differences between house managers and residents on all measures of the social environment. House managers viewed the social environment more favorably (e.g., higher peer support and greater resident involvment) than residents. The differences noted between perceptions of house managers and residents were important because on two scales (i.e., Order and Oganization and the RHES), larger differences between the two groups were associated with shorter lengths of stay. Differences on the Involvement subscale showed a trend (p=0.066) toward significance. Though these were small to medium effect sizes, the difference in views can inform future research on how to conduct research in RHs. Moreover, when manager and resident perceptions were both entered into the same regresssion models, resident perceptions predicted LOS, while manager perceptions did not. Overall, study findings affirm the importance of social environment factors in recovery homes from the perspective of residents but not house managers.

Because the house manager role focuses on oversight of SLHs, it is understandable that they might rate the quality of the social environment more favorably. They may want their homes to be viewed in a positive light. This finding is consisistent with Moos (1997), who found staff perceptions of social environments in mental health and substance abuse programs to be more favorable than perceptions of clients receiving services. However, the differences might also refect other issues that are detrimental, such as disconnection of house managers from resident expereinces or conflict between house managers and residents.

Although bivariate analyses and regression models showed that all five of the social environment scales had significant associations with LOS, factors most closely aligned with social model recovery principles were shown to be the most important. This was evident when we constructed regression models assessing the RHES and each of the CPES subscales in the same models. In each model, the RHES was a significant predictor of LOS, while each of the CPES subscales was not. However, it should be noted Involvement showed a trend (p<0.10).

4.1. Implications for Service Providers and Researchers

Study findings suggest house managers and residents should strive to facilitate social environment factors in SLHs that are associated with LOS, which is an important factor related to subsequent substance use outcome (Polcin et al., 2016). The finding that resident perceptions of the social environment were much stronger predictors than the perceptions of managers suggest that understanding the strengths and weaknesses of the social environment in the homes and changes that need to be made should draw from the experiences and perceptions of residents.

Attention should be paid to general characteristics of the social environment, such as support, resident involvement, and practical skill building. However, the relatively stronger influence of the RHES suggests that managers and residents may want to focus their efforts most intensively on ways to influence issues related to social model recovery. Social environment characteristics such as support, involvement, and skill building might be most helpful when they are related to residents’ recovery.

A number of items on the RHES are framed to capture interactions and activities specifically related to social model recovery. Examples include items assessing the extent to which residents practice 12-step principles in their daily activities, the effectiveness of house meetings in terms of decision making and resolving conflicts, the extent to which residents use supportive confrontation to remind their peers about the potential harm of not maintaining recovery, and the frequency of residents attending 12-step meetings together. Strategies for maximizing some of these principles have been addressed by Polcin et al. (2014) and include strategic interventions to enhance social model behaviors in various situations, including house meetings, daily resident interactions, assisting residents in crisis, planning attendance at 12-step meetings and social activities, and conducting house operations, such as admission of new residents and evictions of residents for rule violations. The paper includes important considerations for mobilizing residents to implement social model behaviors with each other, particularly senior residents who have long periods of recovery and longer tenure in the house.

The National Alliance of Recovery Residences (NARR) has taken a leading role in dissemination of research to residences throughout the U.S. However, there are still many residences that are not affiliated with NARR or other recovery home organizations that promote health, safety, and quality standards. In California, some SLHs are opting out of remaining in voluntary associations such as the SLN. In 2013, there were 490 houses that were affiliated with the network (Wittman & Polcin, 2014). The most recent listing on the SLH website indicates 224 homes (Sober Living Network, 2021). The number of members is not indicative of the number of SLHs, but an overall number is difficult to measure due to the independence of the self-run houses and lack of requirements for registration. As more SLHs choose to operate without memberships in associations, further research is needed into how homes with no affiliation can keep abreast of advances in recovery services and whether these associations are needed. Polcin, Mahoney, and Mericle (2020) surveyed SLH managers in Los Angeles County and found two-thirds received no training related to their manager role over the past year. The limited number of trainings that were attended were typically brief, under one hour. Addressing ways to ensure oversight of houses with standards and mechanisms for dissemination of research represent important challenges for RHs.

4.2. Limitations

A number of limitations bear noting. First, the study was limited to one type of RH, SLHs, and other types of RHs (e.g., Oxford Houses and residential treatment programs) may have different social environment characteristics associated with outcomes. Second, there is need for studies addressing how social environment characteristics are associated with other outcomes, such as substance use, psychiatric symptoms, and severity of drug, alcohol, and other related problems. Third, the study recruited houses in Los Angeles County and houses in other geographical areas could have other factors influencing retention. Fourth, although most of the CPES scales had acceptable levels of internal consistency (>0.70), the Order and Organization subscale for house managers had an alpha of 0.53. Finally, many other individual and environmental factors could impact LOS beyond the social climate, so future research is needed on other predictors of LOS.

Identifying the most effective ways to facilitate the social model activities measured on the RHES will require additional research. For example, some houses might require extensive efforts from house managers of facilitate a healthy social environment. However, in other houses, that might be counterproductive, particularly in houses with residents who have long periods of recovery and previous experience in SLHs. Similarly, the best way to facilitate house meetings or increase the practice of 12-step principles may be dependent on characteristics of residents and dynamics within the home. Another limitation of this research is that the social environment was only measured at one timepoint. As the residents interact with their environment, the social climate and their perceptions of it may change. This study measured their perceptions early on in their stay, but future research could look at how these change throughout their stay and its impact on outcomes.

5. Conclusions

Despite limitations of this current study of perceptions of the social environment at SLHs, this work highlights how these findings can advance recovery housing research and practice. This study aimed to understand LOS in terms of its association with different aspects of the social environment in the houses from different perspectives and using difference measures of the environment. These analyses found that a higher discordance of the ratings between the house manager and resident is associated with shorter lengths of stay and that the residents’ perceptions of the social environment can predict their LOS. Further research is needed into other factors that may impact LOS and the evaluation of methods to help foster these social model principles in SLHs and to increase the house manager awareness of resident perceptions. Since the perceptions of the house manager were not associated with LOS, these results suggest that future investigations should measure how well the residents perceive these principles are followed at the SLH using the RHES and how improving these principles influences other resident outcomes beyond lengths of stay.