Contrasting social climates of small peer-run versus a larger staff-run substance abuse recovery setting.

Abstract

This study compared the social climate of peer-run homes for recovering substance abusers called Oxford House (OH) to that of a staffed residential therapeutic community (TC). Residents of OHs (N=70) and the TC (N=62) completed the Community Oriented Programs Environment Scales (COPES). OHs structurally differ on two primary dimensions from TCs in that they tend to be smaller and are self-run rather than professionally run. Findings indicated significantly higher Involvement, Support, Practical Orientation, Spontaneity, Autonomy, Order and Organization, and Program Clarity scores among the OH compared to TC residents. Additional analyses found the OH condition was higher Support, Personal Problem Orientation, and Order and Organization scores among women compared to men residents. These results suggested that these smaller OH self-run environments created a more involving and supportive social milieu than a larger staff-run TC. These findings are interpreted within Moos' (2007) four theoretical ingredients (i.e., social control, social learning, behavioral economics, and stress and coping), which help account for effective substance abuse treatment environments.

Residential recovery settings provide individuals with substance abuse disorders a supportive place to live within a therapeutic milieu (De Leon, 2000). Sharing a living space with others in substance abuse recovery might encourage mutual self-help participation and increase social support, which are associated with longer periods of abstinence (Humphreys et al., 1999; Moos et al., 2006). Supportive social relationships within such settings might protect people in recovery from relapse and improve overall substance abuse recovery rates (Beattie et al., 1999; Groh et al., 2007; Moos et al., 2007; Vaillant, 1983).

In this study, we focus on two kinds of residential settings for substance abuse that employ mutual self-help principals. Therapeutic communities (TC) are among the most common types of professional, staff- run residential settings for substance abuse treatment (De Leon et al., 2008). Although TCs often do rely on mutual self-help principles, they also vary widely in overall capacity, staffing and training, program philosophies, and services for residents' personal or professional growth (De Leon et al., 2008; Moos et al., 1997).

Another kind of residential setting that employ mutual self-help principals are Oxford Houses (OHs), the largest network of self-help recovery settings in the U.S. serving over 10,000 people in recovery in over 1,400 OH homes (Oxford House Inc, 2009). In contrast to TCs, OHs all follow the same basic rules that emphasize mutual self-help principles, are entirely self-run without staff or supervision, and are limited in size with usually 7–10 people in each residence (Oxford House Inc, 2004). Living in an OH is not “treatment” in that there are no professionals or staff dedicated to providing therapeutic services (Jason et al., 2001; Jason et al., 2006). Although most mutual self-help settings share similarities (c.f. De Leon, 2004; Humphreys et al., 2004; Kirby, 2004), OHs structurally differ on two primary dimensions from TCs in that they tend to be smaller and are self-run rather than professionally or staff-run.

A number of studies have described the philosophies, conflicts, congruence, and approaches of professionals versus self-run mutual- and self-help groups in various rehabilitative settings (Bright et al., 1999; Hetherington, 1995; Humphreys et al., 1996a; Humphreys et al., 1996b; Maton et al., 1989; Moos et al., 1993; Moos et al., 1995; Polcin, 2009; Timko, 1995; Toro et al., 1985; Toro et al., 1988). One way to investigate differences between larger, professional run versus smaller, self-help oriented recovery settings to examine the social climate of these settings. According to Moos (1997), social climate emerges from the interaction between resident characteristics (i.e., health status, functioning, and preferences) and the objective characteristics (i.e., context, design, policies, and services) of the program. A common use for gathering social climate data is to evaluate a setting, compare settings, or to help determine what areas are effective and which are lacking in a setting to guide practitioners and program designers (Moos, 1997). Although the mechanisms of person-environment interactions are complex (Kelly, 2006), social climate is based on the concept that environments influence individuals and individuals influence their environment. Settings with social climates that are more involving, offer consistent support, and help people adapt to their unique life circumstances tend to promote better outcomes (Boydell et al., 1992; Holahan et al., 1982; Moos, 2003; Timko et al., 1998). A key question is what types of recovery communities might best promote these characteristics.

Within larger, staff run recovery programs, well-trained professionals and paraprofessionals can have a positive impact on clients (De Leon et al., 2000). In addition, there is evidence that smaller, peer-led mutual-help groups can also have positive impacts on substance abuse outcomes (Groh et al., 2007). However, living with a larger number of peers in a staff administered program might differentially affect perceptions of involvement and structure compared to a smaller, peer run setting. Because smaller self-run settings might require residents to perform duties normally completed by staff in larger, staff-administered settings, the smaller, peer run settings might create social climates that are more supportive, involving, and oriented towards solving problems.

Consistent with Moos' (1997) model, personal characteristics of residents such as gender might affect the social climate. Several studies suggest that women enter substance abuse treatment with more complex personal problems than men, often involving health, trauma, child care, and child custody issues (Davis et al., 2006; Davis et al., 2005; Ferrari et al., 1999). In addition to receiving support to maintain a sober lifestyle, women seek to create supportive relationships in order to discuss and address personal problems (Davis et al., 2006; Davis, & Jason, 2005). These challenges might require women to create environments for their specific needs; thus, it is likely that women in comparison to men might report a social climate more focused on solving practical and personal problems.

Unfortunately, there have been no studies comparing the social climates of smaller, self-help recovery communities with larger, professionally run settings. The present study compared the social climates of 14 smaller, self-help run Oxford Houses to a larger, staff-operated TC. Clearly, there are two differences between these types of settings, one involving the self-help component and the other involving the size. As external validity was a primary consideration in the design of this study, we chose to select the types of self-help and professionally-run programs that exist in the US, with the self-help type programs being smaller and the professionally-run TCs being larger. We hypothesized that those residents in the smaller self-help OHs would rate the social climate dimensions (e.g., Involvement, Support, Practical Orientation, Spontaneity, Autonomy, Order and Organization, and Program Clarity, Personal Problems Orientation, Program Clarity, Anger and Aggression) as higher than those in a larger staff-run TC.

Method

Participants

We collected data from 70 residents living in 14 of approximately 60 OH in various cities in Illinois. The OH used in this study included 8 women's houses and 6 men's houses with a minimum of 4 to a maximum of six participants per house (M=5 residents) completing the survey. One of the researchers telephoned each house and explained the general purpose of the study to the house president, and requested to be present at weekly house business meeting to meet the residents and recruit participants. Not all residents were present at each house meeting. Of the 72 individuals approached to be in this study, 70 (97.2%) filled out the surveys.

For the comparison condition, we surveyed 62 male and female participants living in a residential TC program for substance abusers located in a major metropolitan area in Illinois. This TC provides services for up to 400 adults in several locations in a large metropolitan city and suburbs. The buildings maintained by the TC range from large single-family homes accommodating up to 15 adults to multi-unit apartment buildings containing up to 120 beds. This TC is organized such that large units are divided into smaller sub-units to enhance personal contact among its in-house staff. The staff in this TC consists of paraprofessionals: TC alumni paid via in-house residency credits or a salary for part-time work. Staff work as resident recovery coaches for each apartment unit, and mentors are assigned to each floor. The staff works on the day-to-day operations, perform client intakes and orientations, collect rent, perform maintenance duties for the buildings, and monitor residents' adherence to its program. Although TCs vary in size from 30 residents to up to several thousand in multiple facilities, 40 to 80 residents are sometimes recommended as optimal for creating a “stratification” of residents into less to more experienced residents (De Leon, 2000, p. 106). In terms of size, the TC we selected was typical of many programs for substance abuse.

After obtaining permission from TC management, we decided on a data-gathering procedure requiring the researcher to appear at the TC's weekly off-site meeting to explain the study and to recruit participants. Of approximately 100 residents in attendance at these meetings, 62 volunteers (62%) complete the surveys.

Procedures

For both conditions, the purposes of the study and confidentiality issues were described to residents. TC volunteers completed the survey in private in a quiet a room separate from the TC's group meeting; afterwards, participants would return to their gathering in-progress. OH volunteers completed the COPES at their individual houses. The total time for data collection during testing sessions did not exceed 45 minutes. We collected basic demographic information including age, gender, ethnicity, and length of time in the current setting.

Measure

This study used the Community Oriented Programs Environment Scale (COPES), form R(eal) (Moos, 1994, 2003). The COPES is a 100-item true/false pencil-and-paper survey organized into ten subscales distributed among three dimensions. The first dimension (Relationship) consists of Involvement, Support, and Spontaneity. Involvement measures participants' activity in the program. The Support subscale assesses the level of positive or constructive feedback clients receive from peers and staff in the setting. Spontaneity measures how freely people interact in their environment.

The Personal Growth dimension consists of Autonomy, Practical Orientation, Personal Problems Orientation, and Anger and Aggression. Autonomy assesses the perceptions of the program's efforts to encourage and equip clients to make decisions and act on their own during and after treatment. Practical Orientation reflects the program's emphasis on treatment help in meeting daily needs for employment, education, training, and life skills. Personal Problems Orientation measures the program's emphasis on addressing residents' personal problems. Anger and Aggression assesses to what degree clients can express anger and how the setting manages conflicts.

The last dimension, System Maintenance, consists of Order and Organization, Program Clarity and Staff Control. Order and Organization measures the degree residents perceive their setting is well organized and orderly. Program Clarity represents the perceptions of how well residents know the rules, procedures, and responsibilities the program expects of them. Staff Control taps into perceptions of which parties have control in the setting, who can alter the procedures and rules, and how much control the staff or residents have over their own environment.

The participants in the TC condition were given the original 100-item COPES. For the participants in the OH condition, we eliminated or modified items relating to staff on the COPES surveys. If a resident of an OH could perform the action attributed to staff, we retained the item and changed the word “staff” to “resident”. These modifications resulted in 78 questions from the COPES survey, which we administered to the participants in the OH condition. For all data analyses, we eliminated the same 22 questions for the TC scores as we did with the OH condition. The Involvement (α=.70) and Personal Problems Orientation (α=.64) subscales were unchanged. The Spontaneity (α=.45) and Autonomy (α=.40) subscales retained 9 items, and 8 were retained for Practical Orientation (α=.53). The Anger and Aggression (α=.71) and Order and Organization (α=.59) subscales retained 7 items. Six items were retained for each of the following subscales: Support (α=.50), Program Clarity (α=.36) and Staff Control (α=.16).

Results

Sociodemographic characteristics

There were no significant differences between the OH versus the TC conditions on any of the sociodemographic items. In the OH sample, 57.1% (n = 40) were male and 42.9% (n = 30) were female, and in the TC environment, 61.3% (n = 38) were male and 38.7% (n = 24) were female [χ²(1, N = 132) = 0.23, p = .63]. No significant ethnic differences were found between the two settings [χ² (4, N = 132) = 7.27, p = .12]. In the OH environment, 64.3% were African-American (n = 45), 32.9 % were White (n = 23), 1.4% were Hispanic/Latino(a) (n = 1), and 1.4% did not indicate their ethnicity (n = 1). For the TC, 54.8% were African American (n=34), 33.9% were White (n = 21), 4.8% were Native American (n = 3), 1.6% were Hispanic/Latino(a) (n = 1), and 4.8% did not indicate their ethnicity (n = 3).

Participants resided in OHs for significantly more days (M = 298.9, SD = 352.9) than the participants in the TC (M = 145.8, SD = 209.7) [t(130) = 2.98, p < .01]. The length of residency in TCs are not permanent and vary from “short-term” of one to six months to stays of more than six months (Melnick et al., 1999); in contrast, OHs have no maximum length-of-stay prohibition (Oxford House Inc, 2004). Our goal was to maximize what residents of recovery homes are exposed to in the real world, and those in larger, staff-based settings often have shorter lengths of stay than in smaller self-help settings. However, we statistically controlled for length of stay in the settings in the analyses below to control this variable.

Social Climate Measures

We conducted a series of one-way analyses of covariance (ANCOVA) in which length of time in the program were used as a covariate for each of the COPES variables. Table 1 summarizes the ANCOVA results. Those in the OH setting had significantly higher scores for Involvement, Support, Practical Orientation, Spontaneity, Autonomy, and Order and Organization than those in the TC environment. Those in the OHs in contrast to those in TCs had directionally but not significantly higher Personal Problems and Program Clarity scores. There were no significant differences between the OH and the TC condition for Anger and Aggression.

Table 1.

Summary of Analyses of Covariance (ANCOVA) for COPES Subscale variables between Oxford House and Therapeutic Community (Covariate: Length of residency)

	OH		TC
COPES Subscale	M (SD)	95% CI	M (SD)	95% CI	F	η 2	p
Involvement	8.07 (1.90)	[7.45, 8.60]	6.63 (2.85)	[6.07, 7.30]	9.64**	.07	**
Support	4.94 (1.17)	[4.55, 5.24]	4.19 (1.71)	[3.88, 4.62]	6.12*	.05	.02
Practical Orientation	5.86 (1.65)	[5.39, 6.29]	4.97 (2.08)	[4.51, 5.47]	6.40*	.05	.01
Personal Problems Orientation	7.23 (2.12)	[6.69, 7.73]	6.47 (2.20)	[5.93, 7.04]	3.45	.03	.07
Spontaneity	6.46 (1.50)	[6.05, 6.79]	4.95 (1.58)	[4.60, 5.38]	26.73**	.17	**
Autonomy	6.99 (1.42)	[6.56, 7.28]	5.45 (1.61)	[5.15, 5.91]	26.85**	.17	**
Anger and Aggression	4.27 (2.00)	[3.75, 4.67]	4.60 (1.82)	[4.18, 5.15]	1.75	.01	.19
Order and Organization	5.87 (1.34)	[5.50, 6.28]	4.53 (1.88)	[4.10, 4.93]	21.97**	.15	**
Program Clarity	4.43 (1.11)	[4.12, 4.71]	3.98 (1.36)	[3.69, 4.32]	3.40	.03	.07
Staff Control	4.36 (0.85)	[4.11, 4.61]	4.31 (1.21)	[4.04, 4.57]	.11++	ns	.74

^*p < .05

^**p < .01

⁺⁺significant covariate interaction; ANCOVA not meaningful

Gender Analyses

We also investigated gender, and setting times gender interactions between the TC environment and the OHs (see Table 2; only significant main effects for gender or interaction gender by condition effects are reported). For Support, we found a main effect for gender [F(1,127) = 7.58, p =.007, partial η² = .06], a main effect for setting (OH vs. TC), F(1,127) = 5.34, p =.02, partial η² = .04], but no setting by gender interaction [F(1,127) = .02, p =.88, partial η² = .00]. An ANOVA analysis revealed that women in OHs report higher Support scores than men in OHs [F(1,68) = 5.23, p =.03, partial η² = .07]. No significant differences were found for Support among men and women in TCs [F(1,60) = 2.57, p =.11, partial η² = .04].

Table 2.

Means and Standard Deviations of All COPES Subscale by Setting and Gender

	Oxford House
	Men		Women
COPES Subscale	M (SD)	95% CI	M (SD)	95% CI
Involvement	7.90 (1.91)	[7.30, 8.50]	8.30 (1.90)	[7.61, 8.99]
Support	4.68 (1.29)	[4.32, 5.03]	5.30 (0.88)	[4.88, 5.71]
Spontaneity	6.30 (1.60)	[5.83, 6.77]	6.67 (1.35)	[6.12, 7.21]
Autonomy	7.10 (1.39)	[6.65, 7.55]	6.83 (1.46)	[6.32, 7.35]
Practical Orientation	5.82 (1.62)	[5.30, 6.35]	5.90 (1.73)	[5.29, 6.51]
Personal Problems Orientation	6.78 (2.09)	[6.12, 7.43]	7.83 (2.04)	[7.08, 8.59]
Anger and Aggression	4.10 (2.06)	[3.47, 4.73]	4.50 (1.93)	[3.77, 5.23]
Order and Organization	5.58 (1.39)	[5.16, 5.99]	6.27 (1.17)	[5.79, 6.74]
Program Clarity	4.42 (1.15)	[4.07, 4.78]	4.43 (1.07)	[4.03, 4.84]
Staff Control	4.40 (0.96)	[4.13, 4.67]	4.30 (0.70)	[3.99, 4.61]

	Therapeutic Community
	Men		Women
COPES Subscale	M (SD)	95% CI	M (SD)	95% CI
Involvement	6.32 (3.02)	[5.39, 7.24]	7.13 (2.54)	[5.96, 8.29]
Support	3.92 (1.78)	[3.37, 4.47]	4.63 (1.53)	[3.94, 5.31]
Spontaneity	4.87 (1.58)	[4.35, 5.39]	5.08 (1.61)	[4.43, 5.73]
Autonomy	5.34 (1.51)	[4.82, 5.87]	5.63 (1.77)	[4.97, 6.28]
Practical Orientation	4.58 (2.10)	[3.92, 5.24]	5.58 (1.93)	[4.75, 6.42]
Personal Problems Orientation	6.13 (2.18)	[5.43, 6.84]	7.00 (2.17)	[6.11, 7.89]
Anger and Aggression	4.84 (1.79)	[4.25, 5.43]	4.21 (1.84)	[3.47, 4.94]
Order and Organization	4.11 (1.94)	[3.52, 4.69]	5.21 (1.59)	[4.47, 5.95]
Program Clarity	3.84 (1.50)	[3.40, 4.28]	4.21 (1.10)	[3.65, 4.76]
Staff Control	4.29 (1.14)	[3.89, 4.69]	4.33 (1.34)	[3.84, 4.83]

For Personal Problems Orientation, we found a main effect for gender [F(1,127) = 6.61, p =.01, partial η² = .05], no main effect of setting (OH vs. TC), [F(1,127) = 3.12, p =.08, partial η² = .02], and no setting × gender interaction, [F(1,127) = .06, p =.80, partial η² = .00]. An ANOVA analysis revealed among residents of OH, women report higher Personal Problems Orientation scores than men [F(1,68) = 4.49, p =.04, partial η² = .06], but there were no significant differences in Personal Problems Orientation between the men and women living in the TC [F(1,60) = 2.34, p =.13, partial η² = .04],

Finally, we found a significant gender differences in Order and Organization between the two settings [F (1,127) = 10.23, p < .01, partial η² = .08]. Further ANOVAs indicated that women in the OHs scored significantly higher than men in OHs for Order and Organization, [F (1, 68) = 4.82, p = .03, partial η² = .07]. Likewise, women in the TC scored significantly higher than men in TCs for Order and Organization, [F (1, 60) =5.44, p= .02, partial η² = .08]. Men in OHs reported higher Order and Organization scores than men in the TC [F (1, 75) =14.30, p < .01, partial η² = .16], and women in OH reported higher Order and Organization scores than women in the TC [F (1, 51) =7.24, p = .01, partial η² = .12].

The COPES Ms and SDs among the 14 OHs are summarized in Table 3, and these data indicate that there is some variability among the social climate variables within OHs.

Table 3.

Oxford House Per-House N, Mean Length of Stay, and COPES Subscale Mean Scores

Oxford House	N	Length Of Stay	Involvement (0–10)	Support (0–6)	Practical Orientation (0–8)	Personal Problems Orientation (0–10)	Spontaneity (0–9)	Autonomy (0–9)	Anger and Aggression (0–7)	Order and Organization (0–7)	Program Clarity (0–6)	Staff Control (0–6)
1	6	479.3	9.17	5.67	7.17	7.17	7.00	6.00	5.17	6.50	4.16	4.33
2	4	218.8	8.25	5.50	8.50	6.50	6.25	8.50	5.00	6.25	4.25	4.00
3	6	443.8	9.50	5.67	6.33	8.50	7.50	6.67	5.00	6.50	4.00	4.16
4	4	63.8	8.50	5.25	6.00	7.25	6.25	7.00	1.25	6.75	5.25	4.75
5	4	142.5	8.00	5.25	6.25	8.50	6.25	7.25	5.75	5.50	4.50	4.50
6	4	606.8	8.25	4.75	6.25	8.00	7.25	7.50	4.75	5.50	5.25	5.25
7	6	151.3	9.50	5.17	5.66	7.50	7.00	6.50	3.67	6.00	4.50	4.00
8	6	586.5	6.16	3.83	4.00	5.33	5.16	6.67	5.00	4.83	4.50	4.17
9	5	166.8	7.20	4.60	5.80	5.20	5.40	6.00	4.20	5.00	2.80	4.60
10	6	482.5	8.33	5.00	7.17	7.50	7.33	7.67	2.33	6.33	4.50	4.67
11	4	98.8	8.25	5.00	6.25	7.00	5.25	7.00	2.00	6.50	5.00	4.25
12	5	202.4	5.80	4.60	4.80	7.00	5.80	6.60	5.20	5.20	5.20	4.40
13	5	141.6	8.20	5.00	4.40	8.20	6.80	6.80	4.60	6.40	4.00	4.20
14	5	197.6	7.60	3.60	4.80	5.60	6.20	7.60	5.00	4.80	4.60	4.00

Discussion

The main findings of this study were that residents living in smaller, self-run OHs in comparison to a larger, TC reported significantly higher scores on the following social climate scales: Involvement, Support, Practical Orientation, Spontaneity, Autonomy, and Order and Organization. As mentioned in the introduction, TCs and OHs represent two different types of residential programs available to individuals in recovery, with the TCs being both larger and run by staff, in contrast to the small, self-help run OHs. As both size and staff versus self-help were involved in this comparison, a limitation in the study is that it is unclear which factor might have been responsible for the differences in social climate. However, as mentioned in the introduction, the study was developed with a focus on external validity, and most residential programs tend to be both professionally run and large or self-help oriented and small. Another limitation of this study is that it used a cross-sectional sampling of fourteen OHs, and we did not take a random sample of houses within Illinois, and this limits the generalizability of the findings. Fortunately, most OHs operate with very similar rules and regulations, so the operational consistency across houses is high.

In addition, The OHs were all communal living spaces in ordinary single-family homes, whereas the TC participants lived in a number of settings in communal spaces in modified multi-unit apartment blocks, one of which contains rooms to house 120 residents. The TC program attempted to offset residents' possible isolation by offering shared living spaces and a “floor manager” who brings together a smaller number of residents for floor and unit meetings. Clearly, there are variations across TCs so the fact that only one was used in this study is a limitation. As TCs vary in size from 30 residents to up to several thousand in multiple facilities (De Leon, 2000, p. 106), the TC we selected was typical of many programs with this orientation. Hopefully, future studies will include a larger number of staff-run recovery communities to compare to smaller, self-run settings. Differential recruitment rates are another limitation of this study. A final limitation is that residents of TCs overall have shorter aggregate time living in the TC, and this may reflect shorter sober time overall, and possibly more severe substance abuse problems.

One way to interpret these findings is to view social climate as an indicator of the “activition” of Moos' (2007) proposed four theoretical ingredients - social control, social learning, behavioral economics, and stress and coping. Social control processes emphasize equality and peer identification, influence over one's surroundings, and a clear structure of how to succeed in the setting. Social control processes might be more engaging and active in smaller, self-help recovery settings that require participants to be directly involved in decision-making and performing administrative tasks that affect the entire house. For OHs, these decisions include creating and enforcing house rules and admission of new residents. In contrast, it is staff in larger settings who mandate and enforce the rules and content of the program and perform resident intakes. Therefore, a setting's social climate for involvement and autonomy might be higher in smaller, self-run recovery communities than larger professionally administered programs. It is possible that democratic decision-making, leadership, and group responsibilities, which occur in OHs, will account for some of the differences among the treatment conditions processes. For example, tasks associated with leadership positions in an OH are not token responsibilities; individual OH communities can and do cease to exist when residents do not or cannot fulfill the OH program's obligations (Mueller et al., 2009).

Moos' (2007) second ingredient of social climates is social learning, which involves being able to discern clear norms about positive, non-using behaviors and obtaining feedback from others on appropriate behaviors. Social learning typically occurs when staff and other residents perform behavior monitoring and have opportunities to provide supportive feedback. In a smaller self-help setting, residents might have strong incentives to monitor, support, and learn from each other because any resident who relapses will disrupt the house immediately both financially and operationally. In contrast, residents in a larger, professional administered setting might have fewer opportunities to monitor each other because this task is the responsibility of the staff. A resident who relapses while living in settings with a larger number of residents might be less disruptive to the overall program than to a smaller, self-help oriented community.

Moos' (2007) third ingredient is stress and coping processes, which can enhance life skills, promote self-efficacy to cope with a non-using lifestyle, promote autonomy, and reduce stress involved in practicing these skills. Typical of larger, staff-run recovery settings, it is staff, not residents, who monitor residents and visitors, collect rental payments, and perform bookkeeping. In such settings, residents are shielded from these tasks, thus they might not be offered the opportunity to learn these skills and practice mutual reliance with others tasked with similar and interdependent duties. In contrast, in smaller, self-help run settings, residents are required to adopt a self-governance model to handle their houses' activity in order to meet the operational needs of their setting. To facilitate the house duties of collecting rent and paying bills, residents in smaller, self-operated OH settings elect their own administrative officers for the house (president, treasurer, comptroller, secretary) every six months. Therefore, it is possible that these smaller, self-help settings in contrast to larger, staff oriented facilities, will have social climates characterized by higher autonomy and practical orientation.

Finally, Moos' (2007) last domain involves behavioral economics, which focuses on choosing to participate in activities that are protective and in which pro-social, non-using behaviors are rewarded. Although pro-social behaviors are emphasized in both smaller, self-help environments and larger staff-run settings, a smaller, non-hierarchical setting might offer rewards that are more engaging. For example, in order to gain the privilege of living in an OH house, residents are required to participate in house activities and to seek harmony in close daily living with peers. Common goals and rewards in the OH setting might encourage more meaningful relations than interactions with residents in larger staff run settings. In addition, in larger staff administered settings, where staff responsibilities include tasks distributed among numerous residents, staff might have fewer and less meaningful opportunities to interact with residents.

Undoubtedly, these underlying theoretical ingredients and processes also occur in larger, staff-run TC environments. Many residents living in TCs and insurance companies expect and pay for services provided by a competent staff. However, staff involvement might mediate and possibly diminish the impact of these underlying processes that occur in a typical TC (Moos et al., 1997). For example, in terms of social control, residents in a TC typically have little say who staff admits into the program, with whom they share a room, or whether or not a peer is active in a recovery program. As most TCs enforce a limited length of stay, social learning from experienced peers is inherently limited, and peer-staff are likely to have the longest-term sobriety. However, even if staff members are also in recovery (as in the TC used in this study), there is always a power and responsibility differential that separates staff and residents (McDonald et al., 2005). Although many TCs employ live-in staff members who are also in substance abuse recovery to offer support and gain credibility among residents, the effects of this characteristic on residents is unclear (Moos et al., 1997). We maintain that differences in social climate are possibly attributable to a system whereby persons living in OHs both run the “program” and receive the “treatment.”

A community is likely to be most influential when it operates with clear sanctions, rules, norms, and expectations of its members, yet are flexible enough to serve the needs of its members (Forys et al., 2007). The finding that OH had higher scores on these social climate subscales than the TC condition might be due to the fact that every OH operates under relatively few core principles; however, each OH may create and enforce its own particular set of house rules, expectations of its residents, and the contractual obligations each resident must abide.

Overall, Personal Problems Orientation was not significantly different between the two conditions. Perhaps this was because discussing personal problems are not required for an OH (or a TC) to function unless personal problems result in behaviors disruptive to other residents or requiring intervention. However, women living in OHs reported higher Personal Problems Orientation scores than men in OHs, and women in the OH sample reported significantly higher levels of Support than men in OH. This suggests that women in contrast to men residing in OHs established more supportive relationships and felt freer to discuss and address personal problems. In addition, higher scores for Support, and Order and Organization among women in OHs in contrast to men in OHs might represent intentional creation of structure in their environments to manage complex situations, such as employment and child visitations. The finding that men in OH had higher Order and Organization scores than men in the TC setting might be due to the impact of self-governance and smaller size.

These social climate data might inform policymakers whether to favor OH settings that are smaller and self-help in orientation over larger, more expensive settings for populations that are seeking social reintegration. Policymakers often make difficult decisions that directly or indirectly affect whether and what kinds of substance abuse settings receive funding, how (or if) such settings are coordinated with other treatment programs, and whether the cost-benefit of the settings are justified and sustainable (Humphreys et al., 2004; Jason et al., 2005). Further complicating these political decisions, identifying clear goals for substance abuse treatment is often complex and idiosyncratic to the individual context and community in which treatment occurs (De Leon, 2004; Vaillant, 2005), and person-environment fit is crucial (Moos, 2003). Oxford Houses represent just one point on a continuum of care; both TCs and Oxford Houses create communities and prepare former substance users to return to non-sequestered living. However, OHs may offer greater opportunities for social reintegration into community, family, and employment systems because of OH's policies of unlimited length of stay while requiring individual residents and houses to be self-supporting (Jason et al., 2008). If a policymaker's goal is to encourage social reintegration of former substance abusers, OHs seem like a worthwhile investment in public funds.

In summary, this study explored the social climates within smaller self-run vs. a larger staff-run setting, and found significant differences in social climate, which might be indicators of “activation” of the four theoretical ingredients of social control, social learning, behavioral economics, and stress and coping processes (Moos, 2007). Up to now, very few studies have been published that use social climate as a predictive measure of therapeutic outcomes (Holahan, & Moos, 1982). Although items such as support and involvement might have an impact on substance abuse recovery, this study was limited to exploring social climate rather than substance abuse outcomes. Given the thousands of people current do reside in these types of settings, more research is needed to know whether the differences in social climate also influence both short term and longer term substance abuse outcomes.