Recovery Home Networks as Social Capital

Abstract

AIMS:

Ensuring recovery home residents’ social integration into a home environment is important for preventing early dropout and facilitating sustained recovery. Social Capital Theory may provide an explanation for how recovery homes may protect residents and improve recovery rates. However, little is known about how social capital in recovery home environments is structured and accessed. Recovery homes may increase social capital by sharing bonds through friendships, lending money, and advice-seeking.

METHODS:

The current study describes social network cross sectional data obtained from a study of 42 Oxford House recovery homes, in three locations in the US (North Carolina, Texas and Oregon). The residents rated each member of their house on the dimensions of friendship, money loaning, and advice seeking to assess how each resident views one another on these dimensions. The research used baseline data from a larger longitudinal study, and although some data were presented for the full sample (APL, isolates, mean reciprocity and density), the results primarily focused on case studies for three of the participating Oxford Houses—with examples of low, median, and high “connected” houses respectively. Standard measures of network structures were calculated for each home.

RESULTS:

Although all Oxford Houses follow the same house rules, they were found to vary in network structure. Findings indicated a considerable range of interconnectedness among residents in these houses, with friendship being the most common relationship, willingness to lend money less common, and advice-seeking the least common.

CONCLUSIONS:

The findings on friendship, willingness to lend, and advice-seeking provide promising leads about what occurs among the social networks within these complex eco-systems, and may provide ways to better understand and facilitate resident social integration into these settings.

Recovery homes provide a substance-free living environment that may be beneficial for individuals with substance use disorders attempting to maintain sobriety following treatment (Jason, Davis & Ferrari, 2007). One such example are Oxford Houses (OH), the largest single network of over 2,000 recovery homes in the US, which are organized and run by residents and intended to encourage recovery-supportive relationships (Oxford House, 2019). There is evidence that these homes are effective for many with substance use disorders (Jason, Olsen, Ferrari, & Lo Sasso, 2006), particularly for those able to remain in a recovery home for at least six months (Jason, Davis, Ferrari, & Anderson, 2007).

Although there appear to be benefits of staying in an OH recovery home, nearly 50% of individuals leave a home before they have been there for 6 months (Jason et al., 2007). One reason for these high exit rates may be that some individuals have difficulties developing supportive social relationships with other residents. Models of successful entry into organizations suggest that newcomers need to feel welcomed in by their proximal group (Bauer, et al., 1998), under a model of social integration (Morrison, 1993). But, newcomers might have some difficulties gaining access to established networks (Reimer, et al., 2008). Turnover, then, could cyclically affect home structures since social capital losses from turnover can damage the well-being of a group (Shaw, Duffy, Johnson, & Lockhart, 2005).

But little is yet known about the actual relational structures that arise in these recovery environments. Given that the supportive function of these homes has been shown to facilitate sobriety of residents, Social Capital Theory (SCT) may account for the benefits provided by relationships within the home. Social capital is defined as “…the aggregate of the actual or potential resources which are linked to possession of a durable network of more or less institutionalized relationships of mutual acquaintance or recognition” (Bourdieu 1985, p. 248). In the case of recovery, it is important to understand how such resources are acquired and, ultimately, used in pursuit of sustained sobriety. Accordingly, there is a need to better understand the social networks within recovery homes in light of their intended function of providing recovery-related social capital to residents.

Applications of SCT to the recovery process (e.g., Best & Laudet, 2010) highlight access to specifically recovery-supportive resources provided by social relationships. These immersive sober living environments are specifically intended to augment non-member friend and family relationships by providing possibly hard-to-find companionship for those attempting the transition from newly initiated sobriety to long-term recovery. Shared goals are an essential ingredient of friendships in a wide variety of settings (e.g., du Plessis & Corney, 2011). Since individuals within a recovery house share the goal of long-term abstinence, joint commitment to this goal can positively affect group cohesion (e.g. Milner, Russell & Siemers, 2010; Portes, 1995). Such cohesive networks may be particularly beneficial for individual members since homogenous and insular networks of individuals can help to conserve existing resources and provide social support (Wellman & Potter, 1999).

SCT identifies connections with goal-supportive others as a marker of cohesion in a social network (Almquist, 2011). For example, Best, Gow, Taylor, Knox, and White (2011) found that belonging to a social network that included others that were abstinent was one of the strongest predictors of a positive quality of life post-treatment. These ties to abstinent alters presumably provide social capital that can motivate and reinforce individuals trying to refrain from substance use. Social capital is thought of as being both created and distributed by such relationships, i.e. in facilitating trust, cooperation, and collective support among residents of recovery homes.

Friendship has been the primary focus of network science since its inception (Moreno, 1934). Some studies of friendship networks suggest that the association with a positively perceived individual represents a type of capital (Kilduff & Krackhardt, 1994). Individuals who are able to develop friendships in new environments have been shown to have higher levels of satisfaction, contentment, and significantly lower levels of homesickness and depression (Hendrickson & Rosen, 2009; Hendrickson, Rosen, & Aune, 2011). Thus, forming friendship ties would be expected to facilitate well-being, and also promote the goals of the home.

In addition to friendship, there are other types of important social linkages among those in recovery homes, such as a willingness to loan resources (money). Resource-sharing may be particularly important for new house residents who might not have the income or savings to pay for rent or food, a common issue in this population (Jason et al., 2014). Supportive relationships during this uncertain period could provide access, directly or indirectly, to practical solutions to quite likely short-term difficulties with money, as well as buffering against stresses of possible similar difficulties later on (Menjivar, 2000).

Another network resource within recovery homes is advice-seeking or mentorship, wherein residents help others with issues like decision-making, coping with negative feelings such as stress, anxiety, and loneliness, and so on. Possibly similar to the dynamics that occur within AA sponsorship (Stevens & Jason, 2015), residents seek advice from those that are trustworthy, make themselves available, provide information, and serve as empathetic friends. Those providing advice have often achieved a certain degree of their own recovery success and often appreciate the opportunity to help others, as the person providing help or assistance to another often also obtains benefits (Riessman, 1965). Having access to individuals who are further along in their recovery can provide valuable support, which could represent another benefit of social capital within house recovery networks for those less stable in their recovery.

Recent studies have examined friendship, willingness to share resources, and advice-seeking social network resources in recovery homes. Jason, Light, Stevens, and Beers (2014) and Light, Jason, Stevens, Callahan, and Stone (2016) collected baseline and three-month follow-up data on several OH recovery houses and found a distinction between mentors and willingness to lend money. Although willingness to lend money tended to be reciprocated between two individuals, mentorship relationships did not show this pattern, indicating that advice-giving and advice-seeking are not usually mutual in friendship dyads. In addition, willingness to lend predicted the formation of advice-seeking relationships. However, advice-seeking did not predict willingness to lend, suggesting that willingness to lend involves a more fundamental judgment about relationship-worthiness than advice-seeking. This may suggest that willingness to lend is part of a process of trust formation, and trust may provide the foundation for mentorship. This is suggestive of how different types of social relationships can overlap and affect each other, either positively or negatively. In a pilot study involving just one recovery home, Jason et al. (in press) found that friendships tend towards reciprocation (they are mutually selected) whereas advice-seeking relationships are more asymmetrical and tend to not be mutual (if A advises B, B is less likely to advise A). Willingness to loan relationships suggest that dyadic loaning relationships are more complicated and contextually-based.

There are many social network characteristics that have yet to be explored within these types of settings, and there are benefits of visually depicting such networks, and categorizing them in ways that illustrate different levels of interconnections. For example, McCabe (2016) found that friendship styles can be broken into three categories: where everyone’s connected, a group to fall back on, and a few small clusters of friends. Within recovery homes, some houses might be characterized as being highly connected, whereas others might be less connected, and the social networks of these varying environments might be distinct. In other words, being in a house where everyone is a friend of everyone else might be rather different from one where friendship is not widely shared. Viewing social networks of recovery homes of low, median, and high connection networks might be more descriptive, but this effort might provide insights that could lead to better theory in the field.

A social network that includes strong friendship, lending, and role model relationships with other house residents would appear likely to be recovery-supportive and capable of helping prevent relapse. But as noted earlier, residents lacking social support are prone to drop out (Jason, Olson, et al., 2007). Findings from social network data may ultimately be used to find ways to increase the number of residents who maintain residence for 6 or more months.

The current study used resident-level data to characterize social networks found in the 42 housing participating in our study, and we aggregated resident-level data to characterize the social networks of residents in their 42 houses. This study describes patterns of friendship, willingness to lend, and advice-seeking among recovery home residents from case studies representing low, median, and high connectivity homes. The intent was to see if patterns of relationships previously found in smaller studies would be observed in a much larger study, and to identify the range of such relationships found in this larger and more geographically-varied sample.

Method

Settings and Participants

OHs are the largest network of recovery homes for individuals recovering from substance use disorders, housing over 20,000 individuals with substance use disorders in the US. OHs are rented, gender-specific single-family homes for 6 to 12 individuals. Each OH operates democratically with majority rule (i.e., > 80% approval rate) regarding membership and most other policies (Oxford House, 2019). Residents must follow three rules, namely, pay rent, abstain from using alcohol and other drugs, and contribute to the maintenance of the home. Violation of the above rules results in eviction from the House.

Data were collected from OHs located in North Carolina, Texas, and Oregon, states in which OH statewide organizations were strong and well-developed. Further, including residences from different geographical regions provided some ability to address the generalizability of findings to recovery homes in other locations. Communication with residences about possible participation helped field staff assemble lists of residences to approach, and recruitment attempts were made in approximately the order that residence contact information became available. Member-elected house presidents were asked to introduce the study by reading a description of it from a project-provided script; houses were accepted into the study if the house president and all other members (or all but, at most, one member) agreed to participate. The first thirteen consenting houses from each state were accepted, and three more houses were added for a total of 42. Once a house was recruited, all residents of those houses were invited to participate in the study by one of three recruiters who visited houses to explain the study to possible participants.

Participants were part of a longitudinal study that collected information every four months over a 2-year period, and the current study involves data from the baseline phase. Other manuscripts will be written on this longitudinal data set. Participants were recruited and interviewed by field research staff in face-to-face meetings. Participants completed measures of their demographics, stress, self-esteem, support and social networks. They were compensated $20 for completing their assessment. All participants provided written consent after being given an explanation of the study by one of the three recruiters. Permissions for this study were granted by the DePaul University Institutional Review Board.

Materials

Measures

The Social Network Instrument (SNI; Jason & Stevens, 2017) was utilized to capture the social dynamics within each OH. This instrument has been used in several investigations on the social networks of recovery home residents (Jason et al., 2014; Jason & Stevens, 2017; Jason et al., 2018; Light et al., 2016) The SNI measures several relationship characteristics, including friendship, loan, and advice-seeking. Data were also collected on help, frequency, and strength, but these measures are not included in the current study. The “closeness” of a relationship may also be used to indicate the strength of that relationship tie. Similarly, “frequency” of interaction has been used in the literature to gauge strength of relationships and their “closeness”. Because we did not use individual strength and frequency items, we still were able to differentiate from the concept of relationship strength inherent in the criteria used to define a friend (i.e. “a close friend or friend”) and an advice-seeking relationship (i.e. “if the respondent reported seeking advice from another resident very often or often, but not present otherwise…”).

The residents rated each member of the house on the relationships of friendship, money loaning, and advice seeking. Each social network relationship type was measured on a 5-point Likert scale. Friendship, intended to assess positive valence of the relationship, was measured by responses to “How friendly are you with this person?” Responses ranged from “close friend” to “adversary”. Participant’s friendship nominations were represented by an adjacency matrix with rows signifying an ego (senders of friendship nominations) and a column with alters (receivers of friendship nominations). If a nomination was present between an ego and an alter it would represent a degree. All values were dichotomized (0 = no degree; 1 = degree) and entered into the corresponding element of the matrix.

A friendship relationship was considered present if the respondent identified another resident as a close friend or friend, but not present otherwise (acquaintance, stranger, adversary). Money loaning (which represents a willingness to lend resources) was considered present if the respondent reported being willing to loan another resident great amounts of money ($500 or $100), but was not considered present if the respondent reported being willing to loan another resident smaller amounts of money (i.e., $0, $10, $50). Advice-seeking was considered present if the respondent reported seeking advice from another resident very often or often, but not present otherwise (i.e., regularly, rarely, never). We also explored each of these variables to determine the network threshold that differentiated the homes from each other as much as possible, since of course, definitions that provided little or no such variation would not be good predictors.

This type of social network instrument has been found to be a reliable measure (Hlebec & Ferligoj, 2002). The SNI used with our sample had a Cronbach’s alpha of .85 and all items contributed positively. A multilevel confirmatory factor analysis of the SNI found an excellent fit and per-item contribution, and neither age nor sex significantly correlated with this instrument (Jason & Stevens, 2017).

Network Properties

This study examined patterns of bidirectional (or simply “directed”) relationships, with each resident (ego) rating every other person in the house (alters). Each resident is referred to as a node and each relationship between the nodes of a dyad is referred to as an edge. Each edge is a degree that can either be an out-degree (a rating that the ego node makes about each alter) or an in-degree (a rating that each alter makes about the ego node). Since each resident provides ratings of all other residents, rater’s may indicate different strengths of relationships, such that A’s rating of B is considered a different measure of the relationship than B’s rating of A. In the descriptions that follow, network measures are described intuitively; technical details are available from many sources, for example, Wasserman and Faust (1994).

The average path length (APL) is the average number of steps along the shortest paths for all possible pairs of network nodes, and it is a measure of the (typical) efficiency of information transfer within a network. A rating of one means everyone is directly connected, and higher values represent less direct linkages. Isolates count how many residents are not connected to any person in the network, and measures network fragmentation.

Two measures are used to characterize so-called “network closure”, i.e., the tendency for a relational network to add certain types of ties based on existing ties. Reciprocity measures the tendency for mutual (bidirectional) connection in a relationship. Reciprocity is the proportion of all of the edges for which a reciprocal edge is present. A related network measure of interconnections is density: the proportion of existing edges out of all possible such edges. Denser networks of positive sentiments, such as friendship, are considered to be more “cohesive” (Wasserman & Faust, 1994). Reciprocity is related to density, but reciprocity could be high even if density is not, indicating a strong tendency for mutual connection.

Analytic Approach

Social network analysis of 42 OHs were conducted using R (see https://www.r-project.org), which is an open-source software environment for statistical computing and graphics. We used the statistical package ‘sna’ (https://cran.r-project.org/web/packages/sna/sna.pdf), a range of processing tools for network data, to calculate the network metrics. The r package ‘igraph’ (https://igraph.org/redirect.html) was used to visually graph the social networks.

The networks of all 42 houses were visually represented as graphs for each network type—friendship, willingness to loan, and advice-seeking—which were used to identify broad structural patterns. The houses were then classified by judgment (a reasonable approach for an under-researched area with no clearly-better formal criteria available; Tyrer & Heyman, 2016) in terms of their density on each of the network types, as high, medium, or low. The resulting nine categories (low, medium, and high density for each of friendship, willingness to lend, and advice-seeking networks) formed the basis for the descriptive analyses that followed. Since showing all of these 126 individual networks graphically would be intractable (but they can be obtained by contacting the first author), we selected three example houses as case studies for visual representation (Figures 1–3). We determined the example houses by examining densities for friendship, loan, and advice-seeking, and then selected the median density house, the house with the highest density house, and the house with the lowest density.

Figure 1:

Sociograms for friend, willingness to loan, and advice-seeking relationships in a median small group network.

Figure 3:

Sociograms for friend, willingness to loan, and advice-seeking relationships in a high connected small group network.

Results

There were 6.14 participants per house, and we recruited an average of 5.45 of them (88.8% of potential participants). Of the 229 who consented to be part of this study, 85% (N = 195) of these consented participants provided social network data. Regarding gender, 55% of participants were males and 44.5% females with a mean age of 38.4 years (SD = 10.8). Their average age was 38.3 years (range 18–65; SD = 11.12). Participants mostly identified as European-American (82.1%), and other ethnicities included African American (9.2%), Latinx (6.5%), Alaskan Native (6.5%), American Indian (1.3%), and Pacific Islander (.4%). The average length of stay in an OH was 10.3 months (SD = 12.55, range from 7 days to 6.8 years).

Median Connected House

We identified a typical OH density sociogram for each theoretical relationship type: friend, loan, and advice-seeking. The first case study was the Median Connected house, identified as the most accurate representation of averages across each theoretical relationship. The median house size was six and the most highly connected relationship type was “friend” as seen in Figure 1. For friendship, among the 6 nodes or residents, there were 21 edges; the average resident was rated highly enough to be considered a friend by a little more than three other residents. However, relationships of friendship do not necessarily imply other pro-social relationships. For example, in this house, the average resident was friends with 3 other residents, would lend money to only one other resident and would seek advice from less than one other resident. There were 6 edges among the 6 nodes for lending money, and only 4 edges among the 6 nodes for advice-seeking. There is an isolate in both loan and advice-seeking for this house. Across other OH house data, houses showed lower density and more isolates for these advice-seeking and loan relationship types, again indicating that friendship need not necessarily lead to advice-seeking or loan relationships (see Table 1).

Table 1.

Median Connected house network statistics.

	Friend	Loan	Advice-seeking
Nodes	6.00	6.00	6.00
Edges	21.00	6.00	4.00
APL	1.30	1.67	1.20
Isolates	0.00	1.00	1.00
Density	0.70	0.20	0.13
Reciprocity	0.86	0.67	.50

The sociograms for the Median Connected house (Figure 1) show how isolation and reciprocity differ for each relationship type. Participant 5034 is the only isolate in both loaning and advice-seeking networks, yet this person has reciprocal relations in the friendship relationship type. Member 5032 was the central character in the loan willingness relationships, acting as a direct or indirect “broker” for all such relationships. Moreover, all 4 of these friendship out-choices were reciprocated; apparently, this individual was important in both friendship and loan willingness networks. In short, these results suggest that friendship networks in these recovery homes tend to be relatively dense and mutual, i.e. that individuals get along well with other residents. However, more specific support relationships involving advice and money are considerably more selective, often “glued” together by a single individual. This, again, suggests that the nature of these relationships involves trust capital, which may take longer to develop than friendliness.

Low Connected House

Next, we examined a case study of a house network that was had the lowest relationship densities, which we call the Low Connected house. The average resident of this home was considered a friend by only two other residents, and the average resident would either lend money to or seek advice from less than one of five (0.20) other residents. The Low Connected house had one individual that was not part of the loaning networks and two individuals who were not part of the advice-seeking networks (See Table 2). Friendship was distributed with little evidence of centrality.

Table 2.

Low Connected house network statistics

	Friend	Loan	Advice-seeking
Nodes	5.00	5.00	5.00
Edges	9.00	3.00	2.00
APL	1.18	1.00	1.00
Isolates	0.00	1.00	2.00
Density	0.45	0.15	0.10
Reciprocity	0.22	0.00	0.00

Apparently, willingness to loan to others is a relatively rare quality, but individuals who are willing to loan to one other person often are willing to loan to more than one, making them potentially a “central loan source.” This relationship may occur in part because these “central loan sources” hold dense and reciprocated friendship networks. Advice-seeking shows a similar pattern but with a very different interpretation. Such individuals are not providing resources, but rather trying to obtain them, and advice seekers tend to seek from more than one alter. This is perhaps evidence of someone who is “needy” or otherwise under-resourced and vulnerable. This finding serves as additional evidence to the SCT literature that advice seekers tend to seek advice within their friendship network, but may have difficulty determining who is the best source of such advice.

High Connected House

The next case study was with a High Connected house, which was the most strongly connected network, having the highest density for each relationship type (see Table 3). The average resident was rated a friend by almost 5 other residents, would lend to 3.5 other residents and sought advice from 1.7 other residents. Friendship and loaning relationship types were close to maximum density and the lowest APL possible. Compared to the less connected homes, these networks showed high reciprocity for friendship (0.97), moderately high reciprocity (0.67) even for advice-seeking, and moderate reciprocity for loan willingness (0.40).

Table 3.

High Connected house network statistics.

	Friend	Loan	Advice-seeking
Nodes	6.00	6.00	6.00
Edges	29.00	21.00	10.00
APL	1.03	1.16	1.55
Isolates	0.00	0.00	0.00
Density	0.97	0.70	0.33
Reciprocity	0.97	0.67	0.40

Network Characteristics of 42 Houses

The findings presented so far described three case studies of homes in detail, but it is also important to examine network characteristics for the entire sample of 42 houses. In our next analysis, we focused on four indices of network health because of their informative strength of the network: APL, isolates, reciprocity, and density. Specifically, APL and isolates measure the connectedness of individuals with others in their network, if they are connected at all. Also, density and reciprocity mark the frequency of these relationships, and which relationships are reciprocated among residents. Thus, we focused on APL and density because they reflect the overall health of the network, and reciprocity and isolation because they are markers of resident-level health of the network.

It is ideal that residents are using all of the social capital present in the home by engaging in as many positive relationships as the environment provides. APL acts as a proxy to how social capital flows through a network, while isolates mark low levels of social capital. APL is a more suitable control for these outliers. When the APL of a house is low, it indicates that most residents have established relationships with many other residents in the home, and when APL is high, residents are friends with few others. APL should signify the movement of social capital within the network. Isolates, on the other hand, have few or no connections to others in the network and experience the least amount of social capital of those in the network. Isolation suggests that individuals are not properly assimilating into the house, and these individuals fail to attain helpful social capital. We measure both APL and isolates to illustrate the interconnectedness of the house, and how efficiently social capital spreads through the network.

Figure 4 shows the mean APL and isolates for the three network types examined in the 42 houses. Within the friendship network, there were no isolates, and the APL was quite low at 1.21. Within the loan willingness network, the APL was almost the same as the friendship network (1.20), signaling a dense network of loan willingness. However, there are far more isolates in the loan network than the friendship network; in other words, most loan willingness relationships were direct (one step), but a great many individuals were not part of this network at all. Within the advice-seeking network, the APL is also similar to the APL of the friendship and loan networks (1.22), but like loan willingness, it has far more isolates than the friendship network. Thus advice relationships were close, but many residents were not part of the advice-seeking network. In short, friendship networks were inclusive, whereas loan willingness and advice-seeking networks were far more exclusive.

Figure 4:

APL and isolates means across relationship type.

Density reflects the quantity of relationships in the house, signaling how many connections are present. However, density alone may be less informative than a multiplex model. For example, some networks may see more reciprocation in their relationships, like friendship, while other relationships, like advice-seeking, may not be reciprocated. Thus, reciprocity measurements enhance density measurements by shedding light on complex, directed, and multiplex networks.

Figure 5 represents the mean reciprocity and density across the three network types. Reciprocity and density are very similar to each other within the network types but vary across types. Within the friendship network, reciprocity and density are high suggesting that friendship (at least “friendly relations”) is the norm among residents. Compared to the friendship network, the loan network has far lower rates of reciprocity and density, suggesting that there are far fewer relationships of loaning, but those few are reciprocated. Finally, the advice-seeking network for reciprocity and density is similar to the loan network; both reciprocity and density are low but similar. Overall, loaning had the least reciprocity, with advice-seeking a bit more commonly reciprocated, and friendship the most typically reciprocated. Hence loan and advice-seeking relationships appear to be the hardest to obtain, while friendship is readily available.

Figure 5:

Reciprocity and density means across relationship type.

Discussion

The current investigation provided a description of three types of social capital-related networks—friendship, willingness to loan money, and advice-seeking—broken out by overall connectedness across all three types. Although all OHs follow the same house rules, they were found to vary in network structure. These settings comprise a significant portion of the resident’s day-to-day social experience—certainly as it relates to recovery, at the very least—and that whatever sobriety-reinforcing value-added this environment provides will be in some way dependent on social processes operating between and among the residents.

Because each house is a “complete network” of relationships, one may think of these social processes as a “multiplex” social network (involving several different types of relationships) that co-evolves over time, along with changing resident characteristics such as recovery-related attitudes and behaviors (Doogan, Light, Stevens, & Jason, 2019; Light et al., 2016). Typically social network analysis incorporates only a single network, thus providing more limited information about purpose-specific relationships. Our study utilized three different types of relationship networks; and in particular friendship, willingness to lend resources, and advice-seeking might play important roles in terms of how long recovery home residents stay in these settings, and whether they perceive these settings as supportive—i.e., as places where recovery-related social capital is available.

Given the differences between homes, we were able to discover similarities within social networks that speak to recovery home social networks at large. Friendship did appear to be the most common relationship, with a willingness to lend money less common, and advice-seeking the least common. These results are consistent with other, smaller studies of recovery home relationships (Jason et al., 2014, 2019). On average, the friendship relationship had no isolates in any house. This confirms earlier studies that friendships in OH recovery homes are normative, along with the social capital associated with being accepted and worthy of friendly treatment. However, other important elements of recovery-relevant social capital—small loans and advice—are considerably harder to come by, and apparently, require something more than just being a fellow resident in order to access them.

Findings for APL were very similar for all three network types, except that advice-seeking and (especially) willingness to loan networks left out many residents entirely. Further, the APL values of these two types of networks were close to 1, implying that any non-isolate usually had a direct connection to every other non-isolate. This pattern is also seen in the density of network types, which is quite high (about 77%) for friendship but drops to about 37% and 27%, respectively, for advice-seeking and loan willingness. Thus a clear agenda for future research is to investigate predictors of acceptance into the loan and advice networks. Possibilities include characteristics of the individual, e.g., will they repay a loan? Are they worthy of advice, perhaps because their prospects for ongoing recovery seem better, or instead because those prospects seem worse, indicating a greater need? Another set of possibilities involves the nature of house norms involving advice and money. Although a formal set of organizational rules establish a basic principle of mutual help for OH residences, these rules are general enough to admit a wide range of implementation; thus, the “culture” of helping other residents may vary from one home to another. In some homes, lending money may be seen as helpful, while in others, a way of enabling poor choices. These are all interesting topics for future work.

Our study provides a methodology that could be used to further explore how residents fit into house ecologies or fail to do so. These types of analyses might help us better understand whether there are more systematic ways prospective residents could prepare for a successful stay. In addition, there is a need to investigate how residents’ relationships within the house, as well as within their own (non-house) personal networks, interact to fulfill recovery requirements.

There have been few social network investigations of recovery homes, and the completeness of the current data and the sheer number of networks are strengths of this investigation. In addition, another positive feature of this study was that social network data were collected from 85% of the participants, with complete participation from all members of many of the homes in the sample. The high level of cooperation provides some confidence that the results for this set of homes are meaningful, even if it is less certain that they are truly representative in a statistical sense. Such work would also shed light on whether individual features (like tenure in the house or age) could affect networks within the house. There are several limitations in the current study. First, there are a limited number of recovery homes which were not selected using probability sampling methods, and they might not represent OH settings and are even less likely to represent non-OH settings, which generally have staff versus the self-help management of OHs. Second, the data presented here are only cross-sectional, so causative explanations are not possible with these analyses, and this design is not able to capture the fact that relationships are dynamic and ever-changing house composition will affect house-level network characterizations. In addition, single items were used to measure friendship, willingness to lend, and advice-seeking, but these concepts are likely to be multidimensional. This problem occurs in much of the social network literature, as investigators need to avoid burdening participants, whose task can grow quickly with the size of the whole network being rated. In addition, some of the social network measures are mathematically related and therefore correlated. These types of relationships often co-occur—i.e. friends may provide advice and/and or lend money to one another.

There is a need to examine house-situated longitudinal interrelationships among individual related recovery attitudes and behaviors, dyadic interpersonal friendship, lending resources and advice-seeking relationship formation processes using a dynamic social network approach. Such research would allow us to treat both relationships and behaviors/attitudes as endogenous, mutually-interacting entities that co-evolve over time. Recovery house social environments may change quite quickly in response to composition change or individual change because they are relatively small (5–10 individuals, generally) and, by design, highly interdependent both socially and instrumentally. Additionally, such research could indicate how new members into OHs are able to successfully embed themselves in their new network. Thus an empirical and related modeling framework is required (Kornbluh & Neal, 2016) that does not take either the social environment or the behaviors and attitudes of the individuals comprising them as fixed.

Figure 2:

Sociograms for friend, willingness to loan, and advice-seeking relationships in a low connected small group network.