|Home | About | Journals | Submit | Contact Us | Français|
Social support may be considered from several different dimensions. While general social support promotes well-being, specific social support is tied to particular functions, such as alcohol use. Not only may the form of social support vary, but also the source (ie, friends vs. family). This study investigated the impact of general and specific support for alcohol use from family versus friends on alcohol use among 897 U.S. residents of abstinent communal-living settings (Oxford Houses). Results indicated that general support from friends and length of stay in Oxford House significantly predicted less alcohol use. Implications for alcohol recovery are discussed.
For individuals in alcohol recovery, contextual characteristics of the environment, such as social support, influence treatment acceptance and provide resources that affect posttreatment functioning.1 For this reason, it is important to explore the social support networks available to individuals in recovery. While social support is defined simply as the resources that other people provide,2 it is a concept that may be broken down into several different dimensions based on type (ie, general vs. specific) and source3–5 (ie, friends vs. family).
A major distinction in the measurement of social support is generality versus specificity. General social support is defined as support for one’s overall well-being, and measures of general support often combine structural aspects (ie, the number of people in a network) with functional aspects (ie., meaningfulness of the support).2 Studies indicate that general social support has an inverse relationship with alcohol use. People who receive more general support possess higher levels of subjective well-being, which is linked to improved post-substance abuse treatment outcomes.6 Social networks that are larger7 and include more supportive relationships8 may promote greater abstinence. People in alcohol recovery also report receiving more general social support when abstinent than when using.9 Similarly, less supportive friendships help perpetuate continued alcohol problems for people in recovery.10
Specific social support, in contrast, has either a positive or negative impact on substance abuse, depending on whether the relationships provide positive encouragement for abstinence (specific support for abstinence) or negative encouragement for drinking (specific support for alcohol use).11 Specific support for abstinence has been found to counteract the influence of substance-using networks.12 Zywiak, Longabaugh, and Wirtz7 found that relapse was more likely for those in alcohol recovery who remained in close contact with pre-treatment networks that provided specific support for alcohol use. However, abstinence was more likely for those whose networks contained more individuals who abstain or are also in recovery.
While in recovery, people tend to gain friendships with abstainers and decrease friendships with drinkers,13 presumably increasing their network support for abstinence. This emphasis on social network change is common to numerous recovery models. For example, cognitive-behavioral principles14 suggest that former substance abusers avoid people and situations that trigger the desire to use. Improving social networks also plays an integral role in AA, such that more of its twelve steps deal with changing relationships than abstinence.15
A few studies have examined the relationship between general and specific types of support and alcohol use. Regarding support from significant others, specific support for abstinence was found to be a stronger predictor of continued abstinence than general support.16 In addition, Beattie and Longabaugh17 found that while both general and specific support for abstinence support predicted short-term abstinence, only specific support for abstinence predicted long-term abstinence. On the contrary, one study demonstrated that positive general support counteracted the harmful effects of specific support for alcohol use in relation to well-being.18
The relationship between social support and alcohol use is further complicated by the fact that not only may the form of the support vary, but also the source of that support3,5 (ie, friends vs. family). Regarding social support from friends, peers’ use has been cited as the best predictor of lifetime substance use.19 Increased general support from friends was found to relate to fewer days of intoxication for women.20 Concerning specific support for alcohol use, Jones and Heaven21 found that 82% of adolescent alcohol users perceived that their friends endorsed their drinking. Furthermore, a Norwegian study found that most teenagers first consume alcohol at a friend’s house.22
Regarding family support, studies demonstrate that higher levels of general parental support are related to lower rates of substance abuse and drinking.23,24 While negative life events correlate with substance abuse, parental support may serve to reduce or buffer this effect.25 In adolescent studies, less supportive families were associated with earlier onset of alcohol use.22,26 Specific support provided by family members can have both positive and negative effects. In an African American sample, spouses, parents, and children were found to promote decreased drinking, whereas siblings and other family members promoted drinking.27
Findings regarding the differential impact of support from friends and family have proved inconsistent. Several studies found that while general support from friends influenced alcohol use outcomes, support from family members did not.28,29 Alternatively, multiple studies have produced opposing findings, suggesting that general support from family but not friends predicts drinking outcomes.21,30 Research has shown that, compared to friends, relatives are more likely to pressure drinkers to quit31 and convince them to seek treatment.32 Further, out of all relationship types, spouses31 and mothers33 have been deemed the greatest sources of pressure against drinking.
Social support has been found to be especially relevant to substance abuse treatment, and abstinent social support networks appear to be crucial for maintaining abstinence following treatment.7,17 Self-help/mutual-aid groups are therefore important community resources for providing peer-based support34 that serve to replace the substance-using networks that initially contributed to the problem. One such group for substance abuse recovery that has garnered recent interest is Oxford House. Founded in 1975, Oxford House provides a supportive, democratic, self-run, communal-living setting for individuals recovering from substance abuse. Unlike most treatment options, Oxford House does not involve professional therapists or treatment providers.35,36 Because residents maintain financial responsibility by paying for their own rent, food, and utilities and by sharing in house chores, Oxford House is no more expensive than any other place of residence. Oxford Houses are located in middle-class, low-crime, low-drug-traffic neighborhoods, and residents are free to stay in a house indefinitely, provided that they avoid substance use and disruptive behavior.37
Several studies have focused explicitly on general and specific social support within Oxford House. Regarding general support, residents rated “fellowship with similar peers” the most important aspect of living in an Oxford House.36 Regarding specific support, cross-sectional research suggests that the Oxford House experience provides residents with abstinent-specific social support networks consisting of other residents in recovery. For instance, among African Americans living in Oxford House, other house residents were found to contribute to abstinent support networks.27 Davis and Jason38 found that longer lengths of stay in Oxford House were related to less specific social support for alcohol and drug use, which was related to abstinence self-efficacy. Likewise, Majer, Jason, Ferrari, Venable, and Olson39 found that time spent in Oxford House combined with twelve-step participation was related to increased support for abstinence. While these previous studies have explored how general and specific types of support function in Oxford House, research has yet to compare the differential influence of general versus specific support in Oxford House. This present study attempts to fill this gap in the literature.
The first objective of the present study was to examine the impact of general and specific support for alcohol use with a prospective design. Baseline social support variables were used to predict alcohol use variables at the follow-up assessment four months later. It was predicted that baseline specific support for alcohol use would have a greater impact on follow-up drinking than baseline general support. The second objective aimed to investigate how family and friends differentially influence future alcohol use in Oxford House residents. It was predicted that support from friends at baseline would have a greater influence on follow-up drinking outcomes than support from family. To examine these effects, descriptive and regression statistics were used to determine the differential and combined impact of these social support variables on alcohol use.
Participants were recruited and surveyed using alternate strategies. The majority of participants (n = 797, 88.9%) were recruited through an announcement published in the monthly Oxford House newsletter that provided contact information for the study. Members of the research team then contacted Oxford Houses via letters to House Presidents, conducted follow-up phone calls to the houses, and where possible, arranged to visit houses. Of the 189 Oxford Houses that were approached, 169 (89.4%) had at least one individual who agreed to participate in the study. The remainder of the participants in the study (n = 100, 11.1%) filled out the baseline questionnaires at an annual Oxford House Convention. There were approximately 300 people at this convention, and the authors attempted to secure a sample of those in attendance (a table was set up in a room where individuals could complete the questionnaires with our research staff). Admittedly, this is a convenience sample of those who attended the conference and elected to participate, and self-selection factors were most likely in operation. However, analyses of data collected at the convention versus data collected using the first method did not reveal significant differences in outcome or sociodemographic variables.
In each case, the nature, purpose, and goals of the study were explained to the potential participants. As part of the consent process, research team members explained that participation was entirely voluntary and that withdrawal from the study was possible at any time. There was an initial and follow-up interview separated by a four-month interval, and $15 payments were made to participants following each survey. These data were gathered by research staff who primarily administered questionnaires in person to the participants. Some data were collected by telephone, which was often the case for those who had left Oxford House. No significant sociodemographic differences were found based on methods of data collection.
At the baseline assessment, this study utilized a modified version of the Important People and Activities Inventory40 (IPA), omitting the Activities portion (IP) from the inventory. The IP is the only known measure to assess both general and specific support for alcohol use in various relationships. In the first section of the IP, the Important People section, the participant may identify up to 12 social contacts over the age of 12 years old. For each network member listed, this section examined the type and duration of the relationship, the frequency of contact, and the network member’s drinking habits. In the next segment, the Most Important People section, the participant listed the four people who were the most important over the past three months. Finally, the participant rated the importance of the network member, how much they liked the person, and how the person reacted to the participant’s drinking. The IP contains 11 indices, which are used to calculate two composite scores: investment in the identified network and support for drinking, along with an overall summary measure, network support for drinking. Strong construct validity was demonstrated across IP subscale indices (0.80) based on respondents’ verbal self-reports when compared to information collected from significant others.12,41 The IP has been found to have 2–3 day test-retest reliability of 0.95.42
There are several concerns with using the IP as intended by the authors. Schmitt18 questioned the validity of combining scores of four and five-point scales along with the assumption that opposition toward drinking is equivalent to support for recovery. Internal consistency and reliability for the IP has also been found to be suspect. While Beattie et al.6 found acceptable overall item and subscale alphas for the IP (Cronbach’s alpha = 0.66–0.67), Davis and Jason38 found the composite scores to have fairly low Cronbach’s alphas (investment in the identified network: 0.31; support for drinking: 0.68). In the present study, the composite scores yielded a Cronbach’s alpha of 0.62. Furthermore, the authors of the scale note that some indices correlate weakly or even negatively with the summary score,7 which is problematic because these indices are summed to create the summary score. In an effort to address these concerns, several studies have scored the IP in different ways, such as combining indices7,18 or utilizing the separate indices as measures of social support.13,17,27
Due to reliability/validity concerns and because the network support for drinking summary score was not designed to be divided between family and friends, the present study chose not to use the summary score to measure specific support for alcohol use. Instead, a baseline question from the Most Important People section was used: How has or how would this person react to your drinking? On a five-point scale, responses range from 1 (left or made you leave) to 5 (encouraged your drinking). This question can be split between family and friends and is both conceptually and empirically sound. Research supports using individual IP indices as measures of specific support for alcohol use,13,17,27 and some of these indices are computed based solely on a single IP question. Moreover, the use of a single question helps avoid the internal reliability problems associated with the summary measure, which combines data that do not all relate to support for drinking.
It was necessary to examine the construct validity of the new measure of specific support for alcohol use (How has or how would this person react to your drinking?). Thus, correlations were run between our measure, the network support for drinking summary score, and relevant alcohol use variables. Both measures of specific support for alcohol use produced similar correlations with alcohol use variables (length of abstinence: r = −.04 and −.06; days consuming alcohol in the past three months: r = .07 and .19; length of stay in Oxford House: r = −.05 and −.07). Additionally, the two measures of specific support positively correlated with each other (r = .63), suggesting that they assess similar constructs.
Because the IP collects information on general support but does not produce a general support summary score, a method to measure this construct was needed. Following the recommendations set forth in the literature, the study assessed general support through both structural and functional aspects.2,4 Two baseline IP questions were multiplied: one measuring structure (Is this person generally supportive of you?) and another measuring function (During the past three months, how frequently have you been in contact with this network member?). This method of assessing general support is conceptually sound and validated by previous research that supports the combining of indices (which are often based off of single questions) to create new outcome measures.7,18
The outcome variables in this study were measured by questions taken from Miller and Del Boca’s43 Form 90 Timeline Followback given at the four-month follow-up. This instrument assesses activities that occurred during the past 90 days, such as general health care utilization, residential history, and substance use. The specific outcome question used in the present study was 90 Days Consumed Any Amount of Alcohol. Although many Oxford House residents are poly-substance abusers, the present study did not examine drug use outcomes because the Important People Inventory was designed to measure specific support for alcohol but not drug use.
In order to examine how general and alcohol-specific social support provided by family and friends influenced alcohol use, descriptive analyses were run exploring demographic, substance use, and social support characteristics of the sample. Subsequently, a regression model was tested using social support variables to predict alcohol use. Finally, how interaction terms factor in this model were determined.
The baseline sample consisted of 897 Oxford House residents, including 293 females (32.7%) and 604 males (67.3%). The sample was ethnically diverse, with 58.4% European American, 34.0% African American, 3.5% Hispanic/Latino, and 4% others. At baseline, the average age of the sample was 38.4 and the average education level was 12.6 years. Regarding marital status, 49% were single/never married; 46.2% were divorced, widowed, or separated; and only 4.8% were married. With respect to employment, 69.3% reported being employed full-time, 13.9% part-time, 11.6% unemployed, and 3.8% retired/disabled.
Of the original baseline sample of 897 participants, 687 (76.6%) completed the second wave of data collection. Chi-square analyses indicated that gender, race, religion, and marital status were similar for those who dropped and those who were retained. Those from the northeast were most likely to complete the follow-up assessment, while those from the midwest and Texas were least likely (X2 [3, n = 897] = 59.75, p < .001). Furthermore, those who dropped out had experienced more serious depression (X2 [1, n = 872] = 3.97, p < .05) and thoughts of suicide (X2 [1, n = 868] = 4.73, p < .05), and were prescribed more medication for psychiatric/emotional problems (X2 [1, n = 863] = 3.10, p = .05). Independent samples t tests indicated that dropouts and those who completed both waves were similar on the variables of education, income, and employment. However, those who completed Wave 2 were older (t = −2.29, p < .05) and had longer lengths of stay in OH (t = −5.90, p < .05). Regarding alcohol use, those who dropped out had shorter lengths of alcohol sobriety (t= −3.19, p < .005) and consumed alcohol on more of the past 90 days (t = 3.50, p < .005).
Descriptive analyses were completed for baseline substance use histories. Regarding lifetime data, 99.5% reported any alcohol use, and the average participant had consumed alcohol for 18.3 years (SD = 103). In addition, 10.1% of the sample consumed alcohol in the 90 days prior to baseline (only 5% in the 90 days prior to follow-up). Participants had undergone alcohol treatment an average of 2.8 times (SD = 4.2) and had abstained from alcohol for 1.7 years (SD = 2.2). Drug use outcomes were not assessed in this study because the Important People Inventory does not measure specific support for drug use; however, many participants in this study were poly-substance abusers. For example, 15.7% of participants had used either alcohol or drugs in the 90 days prior to baseline, and 13.3% had only used drugs. Regarding lifetime data, the drugs most commonly used in this sample were cocaine (79.1%) and cannabis (69.2%). Participants also used amphetamines (45.6%), hallucinogens (44.4%), heroin (29.6%), sedatives/hypnotics/tranquilizers (28%), other opiates/analgesics (24%), barbiturates (13.5%), and methadone (10%). Additionally, 73.9% of participants reported using more than one substance in a day (including alcohol). On average, participants underwent drug treatment 2.9 times (SD = 3.5) and had been abstinent from drugs for 1.9 years (SD = 3.1).
In order to examine the participants’ social networks, descriptive analyses were performed on baseline Most Important People (MIP) networks, the data from which were used to calculate the social support variables. Out of a possible four people, participants listed an average of 3.2 (SD = .9) members in their MIP networks. On average, participants had contact with MIP network members slightly more than once or twice a week. Almost all network members (97.9%) were considered to be generally supportive. These networks on average contained 48.7% family and 51.3% friends. Overall, 77.6% of MIP network members were current nondrinkers and 22.4% were current drinkers. Additionally, Oxford House members comprised 28.2% of MIP networks.
One aim of the present investigation was to examine general and specific social support from family versus friends. Out of a scale ranging from 0–7 (0 = low general support, 7 = high general support), participants reported significantly more general support from friends (M = 5.96, SD =1.23) than family (M = 5.21, SD = 1.33; t = −9.77, p < .001). Out of a scale ranging from 1–5 (1 = low specific support for drinking, 5 = high high specific support for drinking), participants received significantly more specific support for drinking from friends (M = 2.08, SD = .83) than family (M = 2.07, SD = .78; t = − 1.52, p < .001).
Because studies indicate that longer lengths of stay in Oxford House are related to decreased social support for alcohol use38 and increased social support for abstinence,37 correlations were run at baseline between length of stay in Oxford House and characteristics of social support networks. Length of stay in Oxford House positively correlated with the percentage of friends (r = .11, p < .01) and Oxford House residents in MIP networks (r = .10, p < .01). However, no significant correlation was found between length of stay and percentage of network drinkers. Because baseline length of stay in Oxford House was related to key social support variables, it was included into the regression model along with general and alcohol-specific support. It is likely that longer stays in Oxford House are related to increased social support, and thus the regression model was tested with the addition of interactions between social support variables and length of stay in Oxford House.
A regression analysis was run to determine if baseline social support variables predict alcohol use variables during the 90 days prior to the follow-up assessment. The variables entered in the model were general support from family members, general support from friends, specific support for alcohol use from family, specific support for alcohol use from friends, and length of stay in Oxford House. In addition, interactions were tested between length of stay in Oxford House and the different types of social support. However, because the interaction between length of stay and general friend support was the only interaction that significantly contributed to the model, the other interactions were removed. This empirically supported interaction is also in line conceptually with the Oxford House model. The baseline variables of general support from friends (β = −.32, t = −4.57, p < .001), length of stay in Oxford House (β = −.78, t = −2.93, p < .01), and the interaction between general support from friends and length of stay in Oxford House (β = .70, t = 2.60, p < .05) were significant predictors of less alcohol use during the 90 days prior to the follow-up assessment. The model significantly accounted for 8.0% of the variance in alcohol use (F[6, 312] = 4.50, p < .001). However, general support provided by family and specific support for alcohol use provided by friends or family did not significantly influence alcohol use.
Further analyses explored the interaction between general friend support and length of stay in Oxford House (see Figure 1). Median splits divided the sample into high and low levels of general friend support (greater or less than 6.33) and length of stay (greater or less than six months) to create four mutually exclusive groups. In addition to being the median length of stay, six months is often considered a critical turning point in the Oxford House recovery process.44 For participants with shorter residencies in Oxford House, levels of general friend support had a greater impact on alcohol use than for participants with longer residencies in Oxford House. In the shorter residency group, individuals receiving low general friend support were almost twice as likely to consume alcohol as those receiving high general friend support (10.5% vs. 5.6%). In contrast, participants in the longer residency group were almost equally as likely to drink regardless of their level of general friend support. In this subset, rates of alcohol consumption were 1.9% for those with high general friend support and 2.4% for those with low general support.
The present study investigated the relationship between general and alcohol-specific social support from family versus friends on alcohol use in Oxford House residents. Overall, results demonstrated that out of the different social support variables, general social support provided by friends had the greatest impact on drinking behaviors in this sample. In addition, length of stay in Oxford House was a significant predictor of less drinking. It is suggested that relationships, especially those consisting of friends, should not be overlooked during alcohol recovery. Friends who provide general support may offer a great asset to recovery, especially in the Oxford House setting.
General support from friends was the strongest predictor of fewer days consuming alcohol at four months follow-up. The impact of friends in this sample is consistent with the fact that Oxford House residents live with friends instead of family members. This finding is similar to research showing that Oxford House residents considered “fellowship with similar peers,” the single most important aspect of the Oxford House living experience.34 In essence, not only do participants value companionship with Oxford House friends, but this companionship helps them remain abstinent. In the social support literature, this present finding is consistent with one study29 but inconsistent with others.21,30 Results may have varied because much of the existing literature focused on adolescent drinking, whereas this investigation focused on adult drinking. Social support from family members may be more important in childhood, while support from friends may become more important as people age.45
It was not at all unexpected that length of stay in Oxford House predicted less alcohol use, supporting the effectiveness of this recovery model. However, the significant interaction between length of stay and general friend support was less anticipated. It was discovered that general friend support had a greater impact on less alcohol use for those with less time in Oxford House than for those with more time in Oxford House. In other words, levels of general friend support have less of an effect on drinking behaviors after a considerable stay in Oxford House. It is suggested that the Oxford House experience is so effective that it counteracts the negative effects of low general support. It is also possible that after staying six months in an Oxford House, most residents have developed highly supportive networks.
Contrary to predictions, specific support for alcohol use, regardless of whether provided by family or friends, did not have a significant impact on alcohol use. Thus, general support had a greater impact on alcohol use than alcohol-specific support. Possibly Oxford House is such a positive and supportive setting that general support from peers undergoing a similar experience is more relevant to recovery than any type of specific support. This finding is consistent with results by Schmitt18 but contradictory to two studies finding the opposite effects.16,17 Participants may have received lower levels of general support in these non-Oxford House recovery samples. It is also suggested that because this present sample received little specific support for alcohol use, it would be more valuable to measure specific support for abstinence. This type of support is almost certainly more readily available in Oxford House and may play a major role in Oxford House recovery. Unfortunately, this study was unable to measure specific support for abstinence because the Important People Inventory was only designed to measure specific support for alcohol use.
There were several limitations to this study. Attrition occurred between the baseline and four-month follow-up assessment; however, based on baseline data, there were only small differences between those who left versus those who were retained. Some selection bias may have occurred during recruitment, and the low rates of alcohol use in this sample may indicate that only the more successful or motivated Oxford House residents chose to participate. In addition, these analyses have limited generalizability because this study only measured alcohol use outcomes even though many participants were poly-substance abusers. Future research with this sample could therefore focus on specific support for any substance use (both alcohol and drugs). While significant, the regression model did not account for a considerable proportion of the variance in alcohol use. It may be easier to predict alcohol use among individuals who consume more alcohol, and therefore, it is suggested that future studies examine social support in new Oxford House residents. Possible limitations also arise from our utilization of the Important People Inventory to measure general and specific support. Although several studies have similarly modified the scoring of the IP, it is recommended that future studies aim to determine the most valid and reliable methods of scoring this measure.
The authors appreciate the financial support from grant number DA13231 from the National Institute on Drug Abuse, Bethesda, Md. (Dr. Jason).
Portions of this paper were taken from the unpublished Master’s Thesis of the first author.