Using daily diary methods to understand how college students in recovery use social support

Abstract

Collegiate Recovery Communities (CRCs) are important sources of support for college students building and maintaining recovery from substance use disorders. The current study used daily diary data from members of a CRC to examine with which sources of social support students engaged daily, and whether students connected with these sources more on days when they indicated higher-than-usual recovery difficulty, negative affect, and/or school stress. Results indicate that on days when students reported having greater difficulty with recovery maintenance than usual, they had higher odds of being in contact with family members and were expected to talk or spend time with family for longer than usual. Students also had higher odds of having recovery-focused conversations with both 12-step sponsors and CRC peers on days of greater-than-usual recovery maintenance difficulty. Recovery maintenance difficulty was uniquely associated with longer duration of family contact, above and beyond negative affect and school stress. Thus, the occurrence, amount, and nature of CRC members’ interactions with important social network members varied in relation to perceived recovery challenges that same day. Findings highlight the importance of providing college students with multiple sources of support that they can use to maintain their recoveries despite daily challenges.

1. Introduction

An estimated 600,000 college students are in recovery from a substance use disorder (SUD; Substance Abuse and Mental Health Services Administration, 2019). College campuses can be particularly difficult places to maintain recovery, in part because substance use is highly prevalent among college students (Johnston, O’Malley, Bachman, Schulenberg, & Miech, 2016) and social interaction with substance using peers is a risk factor for relapse (Godley & Godley, 2011). Thus, being in recovery can limit opportunities for social engagement and lead to feelings of isolation (Bell et al., 2009).

In response to these challenges, more than 130 colleges and universities have established Collegiate Recovery Communities (CRCs) to help students protect their recoveries (Association of Recovery in Higher Education, 2020). CRCs are institutionally sanctioned programs that provide the academic, social, and community support needed for recovering students to succeed in higher education while maintaining their recoveries (Brown, Ashford, Heller, Whitney, & Kimball, 2018; Cleveland, Harris, Baker, Herbert, & Dean, 2007; Laudet, Harris, Kimball, Winters, & Moberg, 2014). CRCs conventionally provide this support through a combination of peer-to-peer social support, academic support, drop-in centers, and sometimes sober living arrangements and clinical services (Bugbee, Caldeira, Soong, Vincent, & Arria, 2016). Program staff provide relapse prevention and life skills workshops, and organize sober recreational activities and service opportunities (Harris, Baker, & Cleveland, 2010; Laudet & Humphreys, 2013). CRCs also provide support to members for working their recovery programs, including providing on-campus mutual-help meetings and helping arrange 12-step sponsors, as well as providing guidance for managing relationships with family members and other important individuals. Of note, the term CRC is often used interchangeably with Collegiate Recovery Program (CRP); in some cases, CRP may be used to refer to the institutionally authorized program and its components (e.g., staff and physical space), whereas CRC may indicate the group of students that participate in the program (ARHE, 2020). Because this study focuses on the daily lives of students within the program and their interactions with other members of the community, we use the term CRC throughout.

1.1. Social recovery capital

Although CRCs provide support in multiple ways, one of their main objectives is to help students build and expand networks of social support. Viewed through the lens of recovery capital, the social network support that CRC involvement provides is one aspect of social capital (Cloud & Granfield, 2008; Vilsaint, Kelly, Bergman, Groshkova, Best, & White, 2017). Research has established that social support within CRCs is among the most important factors for preventing relapse (Smock, Baker, Harris, & D’Sauza, 2011), at least among women (see Smith, Franklin, Asikis, Knudsen, Woodruff, & Kimball, 2018). For example, having more social network connections within a community has been related to more months since relapse (Patterson, Russell, Nelon, Barry, & Lanning, 2020). In particular, recovering peers have been increasingly recognized as beneficial sources of support for individuals with SUD (Iarussi, 2018; Tracy & Wallace, 2016). Some of the earliest research on CRCs documented the high number of recovering peers they provide their members (Cleveland, Wiebe, & Wiersma, 2010). These peers create both a substance-free context for engaging in social interactions as well as opportunities for members to develop their recovery identities (Bell et al., 2009; Scott, Anderson, Harper, & Alfonso, 2016). These findings suggest that one of the primary values of CRCs is the access they provide to alternative peer groups comprising recovering peers.

In addition to having access to an alternative peer group, CRC members should receive recovery support from multiple sources (see Boeri, Gardner, Gerken, Ross, & Wheeler, 2015). Family may be one important source of continuous social support. Research has well documented that addiction affects the entire family system, and that family support for recovery is important (see Ventura & Bagley, 2017). For instance, research has shown that family involvement in SUD treatment increases program completion (McPherson, Boyne, & Willis, 2017) and, following treatment, research has shown that family cohesion, conflict, and social support are associated with substance use and substance-use related problems among adolescents (Godley, Kahn, Dennis, Godley, & Funk, 2005). Beyond the family, recovering students have noted the importance of ongoing communication with Alcoholics Anonymous (AA) sponsors (Terrion, 2012). Sponsor contact and a stronger sponsor alliance have been associated with both greater 12-step participation and abstinence among early adults in recovery (Kelly, Greene, & Bergman, 2016). Romantic partner support is important for many individuals as well: a recent meta-analysis found that SUD treatments involving significant others reduced substance use frequency above and beyond individually based treatments (Ariss & Fairbairn, 2020). Thus, students should ideally work both to develop new social connections and to maintain and improve important, supportive relationships that existed prior to college (such as family relationships) from which they can access diverse social capital to support recovery maintenance.

1.2. Accessing social recovery capital when it is needed

Implicit in the concept of social capital is that it provides a benefit for individuals who possess large amounts of it and draw upon it to support their recovery. That is, for social capital to be valuable, a person must not only have it, but also engage with it when they need it (Trulsson & Hedin, 2004). Students’ interactions with their social networks, an aspect of their social capital, are not static. Rather, successful recovery occurs through day-to-day efforts to build and maintain supportive social relationships and access these supportive relationships to manage daily challenges to well-being. As a result, although the size of individuals’ social networks may be relatively stable, contact with network members likely varies day-to-day based on the challenges to recovery that students encounter on different days. For example, on particularly challenging days students in recovery might seek support from friends in recovery to bolster their own recovery commitment that same day, and/or reduce the impact of a negative experience on their mood or craving that day. These dynamics may look different for different students, on different days. To understand these dynamics as they unfold across time, assessments need to capture students’ social interactions with other members of the community, as well as with sponsors, family members, and romantic partners, on a day-to-day basis.

To that end, daily diary approaches use technological tools to request feedback from participants on a daily basis (Stone & Shiffman, 1994). Such approaches permit researchers to identify the phenomena that matter at different times and contexts for a single individual as well as identify those phenomena that differentiate individuals from one another. However, few studies have leveraged such designs to examine daily patterns of social support usage during recovery. One exception is a study by Dennis and colleagues (2015) that used a combination of ecological momentary assessments (EMA) and ecological momentary interventions (EMI) to understand and support recovery among adolescents. The EMIs that adolescents could access at any time included social recovery support, relaxation, recovery motivation, and social networking (Dennis, Scott, Funk, & Nicholson, 2015). Adolescents reported recovery support, which included participating in discussion groups, reaching out to others, and contacting peer-support team members, as the most commonly used EMI.

Although the Dennis et al. (2015) study underscores the importance of understanding adolescents’ daily use of social recovery support, more work is needed to characterize the different relationships that make up “recovery support”, as well as to identify which supports are utilized in pursuit of sustained recovery, when, and to what extent. For instance, individuals may be more likely to benefit from supportive interactions from certain individuals or groups on particularly difficult days, when they need them most to maintain recovery. Further, when these interactions occur, they may differ both quantitatively and qualitatively on difficult recovery days. Quantitatively, individuals may want to spend longer amounts of time with supportive others on challenging days. Qualitatively, the content of their interactions may be more focused on addressing the challenges they are experiencing. Examining how the occurrence, amount, and nature of CRC members’ interactions with important members of their social networks on a given day varies in relation to perceived challenges to recovery would facilitate a better understanding of the daily lives of college students in recovery, and the development of better ways to help them.

1.3. The current study

Guided by the premise that individuals’ abilities to engage social relationships in support of their recovery goals is critical to recovery maintenance (Best, Lubman, Savic, & Wilson, 2014), the current study examined CRC members’ interactions with diverse sources of social support on days when they perceived recovery maintenance to be more difficult than usual. Our primary interest was in first understanding whether daily recovery maintenance difficulty was related to same-day contact with family, romantic partners, 12-step sponsors, and CRC peers. Then, follow-up analyses were conducted to examine whether associations revealed in prior analyses remained significant above and beyond negative affect and school stress (which themselves may be threats to recovery, but do not necessarily have to be). Specifically, the study addressed four research questions. First, was recovery maintenance difficulty on a given day associated with the occurrence of students’ contact with these different members of their social networks? Second, was recovery maintenance difficulty associated with the amount of contact that students had with the different members of their social networks? Third, was recovery maintenance difficulty associated with the occurrence of recovery-focused conversations with the different members of students’ social networks? Fourth, was recovery maintenance difficulty uniquely associated with the occurrence, amount, and nature of daily contact with social network members, above and beyond the potential effects of experiencing generally bad days (e.g., days marked by higher negative affect) and/or stressful school days?

2. Materials and methods

2.1. Participants

Data were collected from 55 members of a twelve-step CRC located in a southwestern U.S. public university. This university had one of the earliest and most well-established CRCs in the United States, making it ideal for studying how students utilize the resources that CRCs strive to provide. The university also offered its own supports in addition to those that the CRC provided, including counseling services, daily mutual help meetings, weekly recovery classes, and non-CRC sober activities such as sporting events. Participants were 71% male (n = 39) and 98% non-Hispanic white. To be included in the current analyses, participants were required to be between the ages of 18 and 29 and be in recovery primarily from a SUD. The research team made this decision because the age distribution was substantially skewed, with recent members of the CRC being younger, and some members of the CRC being substantially older. Older members had largely joined the community prior to it shifting to accommodate younger recovering individuals, and had different home lives and social networks than other members. Accordingly, four participants were excluded for being over the age of 30 (M_age = 38.75, Range = 32–54) and one was excluded for being in recovery primarily from an eating or food disorder, leaving n = 50 participants (M_age = 21.42, Range = 18 – 29) across freshman (n = 18), sophomore (n = 14), junior (n = 13), and senior (n = 5) years of college. All participants were full-time students (i.e., 12 credit hours or more). Alcohol was the first or second drug of choice for 52% of students in the sample (n=26), marijuana for 50% (n=25), stimulants for 40% (n=20), opiates for 26% (n=13), and club drugs or hallucinogens for 32% (n=16). Ninety-three percent had received professional alcohol/drug dependency treatment and 67% had received inpatient care, in most cases for 3 months or more. Eight participants (16%) had abstained from all substances from which they had made a commitment to remain abstinent for less than one year, 16 participants (32%) had been abstinent for 1–2 years, 11 participants (22%) for 2–3 years, 13 participants (26%) for 3–5 years, and 2 participants (4%) for five years or more.

2.2. Procedure

During recruitment, research staff explained that participation was voluntary, would require three weeks of end-of-day data entries, that data were confidential but not anonymous, and that compensation was $50. All participants provided written informed consent after study staff fully explained the study protocol to them. Interested participants were assigned to four different data collection flights, from fall 2004 to fall 2005, before filling out baseline questionnaires and receiving electronic data collection devices programmed with the Pendragon Mobile Forms Software for the daily diary portion of the study. Devices allowed participants to fill out surveys upon request. The data collection protocol requested that participants fill out three weeks of end-of-day surveys. Study staff called participants nightly and asked them to complete the survey that assessed mood, social interactions, and other daily experiences. After three weeks, some participants asked to keep taking the survey longer, which was allowed. Study staff made reminder phone calls each day until participants returned the devices. A university Institutional Review Board approved all study procedures.

2.3. Measures

2.3.1. Family contact.

Daily family contact was measured with two items. The first item was, “Did you spend time with or talk on the phone to anyone in your family for a few minutes or more today?” Responses were coded as 0 (did not contact family) or 1 (contacted family). On days when participants contacted family, length of time spent in contact with family was measured with a second item, “What was your total contact time (including phone calls) with your family today?” The response scale ranged from 1 (None to a few minutes) to 5 (4+ hours). Amount of family contact was coded as missing on days when students reported that no contact was made.

2.3.2. Romantic partner contact.

Among days when students reported having romantic partners, daily contact with romantic partners was measured with a single binary item, “Did you see your spouse/partner today?” Responses were coded as 0 (did not see spouse/partner) or 1 (saw spouse/partner). Students reporting not having a romantic partner were coded as missing for that day.

2.3.3. Sponsor contact.

Among days when students reported having 12-step sponsors, daily contact with sponsors was measured with two items. The first item was, “Today, did you talk to your 12-step sponsor?” Responses were coded as “1” if students indicated talking to their sponsor in person or on the phone, and “0” if they had a sponsor but did not contact them. Days when students reported that they did not have a sponsor (n = 214), or that they had not yet contacted their sponsor but planned to later that day (n = 7), were coded as missing, due to our not knowing whether students actually contacted sponsors later. On days when students indicated that they contacted their sponsors, the nature of the conversation was measured with a follow-up item, “What did you talk about?” Since we were interested in recovery-focused conversations, we coded responses as “1” if students talked about abstinence-specific issues or “working” a specific step, and “0” if students reported talking about general life issues, just said hi, or none of the above.

2.3.4. CRC peer contact.

Multiple questions assessed daily contact (yes/no), length of contact, and recovery-focused contact with CRC peers. Daily contact was measured with two items, “Did you stop by [the CRC drop-in center] today?” and “Today, did you spend social time or talk to/with a [CRC] friend outside of the [drop-in] center?” As the drop-in center was only open on weekdays, we calculated a single measure capturing any CRC peer contact as having been in contact with CRC peers inside the center, outside the center, or both (coded as 1); or having neither been in contact with peers inside nor outside the center (coded as 0). On days when participants had contact with CRC peers, length of time spent with CRC peers was measured with two items, “How much time did you spend at [the CRC drop-in center] today?” and “What was your total contact time not at the [CRC drop-in center] with all CRC people today (including CRC roommates)?” The response scale for the first item ranged from 1 (stopped in/said hi) to 7 (2 hours or more), and the response scale for the second item ranged from 1 (none to a few minutes) to 5 (4+ hours). Because of the differing response scales, these items were analyzed separately. Recovery-focused conversations on days with at least some peer contact were assessed with two items: “Were issues [that you discussed at the CRC drop-in center] discussed in terms of recovery?” and “Were issues [discussed with CRC friends outside of the drop-in center] discussed in terms of recovery?” We calculated a single measure capturing any recovery-focused conversations with CRC peers as having discussed issues in terms of recovery inside the center, outside the center, or both (coded as 1); or having neither had recovery-focused conversations inside nor outside the center (coded as 0).

2.3.5. Recovery maintenance difficulty.

We measured students’ perceived difficulty with maintaining their recoveries with a single item, “Did you think about how hard it is to maintain your recovery today?” The response scale ranged from 1 (Not at all) to 5 (Extremely).

2.3.6. Negative affect.

We measured negative affect using 10 items from the PA-NA Scales (Watson, Clark, & Tellegen, 1988), a scale that research has previously examined among college student samples (e.g., Allan, Lonigan, & Phillips, 2015). The response scale ranged from 1 (Very slightly or not at all) to 5 (Very much). The items took the form, “Today, to what extent did you feel… ‘stressed’, ‘distressed’, ‘upset’, ‘scared’, ‘guilty’, ‘hostile’, ‘irritable’, ‘ashamed’, ‘nervous’, and ‘jittery’?” We calculated a composite negative affect score for each participant for each day as the average of the ten responses. We assessed reliability for this scale in two ways. First, we calculated Cronbach’s alpha across all assessments and all persons, yielding α = 0.87. Second, consistent with the generalizability theory approach for intensive repeated measures data (see Bolger & Laurenceau, 2013), we fit an ANOVA with random effects (i.e., intercept-only model) to understand the variability of negative affect across persons, time, and items. The measure provided reliable assessment of within-person change, R_c = 0.84.

2.3.7. School stress.

Daily school stress was measured with a single item, “Overall, was today a stressful school day?” The response scale ranged from 1 (Not at all) to 5 (Extremely).

2.4. Data analysis

2.4.1. Data preprocessing

First, we calculated the means for each of our respective predictors of interest, including recovery maintenance difficulty, negative affect, and school stress, separately for each person across all of their repeated measures to obtain usual levels of each variable for each person (e.g., Usual Recovery Difficulty). Then, we subtracted these person-level means from the observed value corresponding to each respective predictor reported by each person on each day of the study to derive a daily score for each predictor (e.g., Day’s Recovery Difficulty) that was standardized to the individual (i.e., person-mean centered). These scores represented the amount of within-person deviation in each of the predictors around a given person’s average level. Removing the person-means in this way allowed us to disentangle the within-person effect from the between-person effect for each predictor (Bolger & Laurenceau, 2013). However, we still wanted to adjust for potential between-person differences when estimating within-person parameters. For example, some students may have experienced more recovery difficulty than others on average. Thus, we accounted for this potential difference by including the person-level mean as a separate predictor in the model, and ensured that the relationship between recovery maintenance difficulty and the outcome of interest was purely a within-person, day-level relationship. Finally, person-mean variables for each of the three predictors were themselves centered around the sample-mean, such that between-person differences on each variable represented deviations from the average person in the sample. We coded time as days since study initiation so that intercepts described the first day of the study for each participant.

2.4.2. Multilevel models

A multilevel modeling (MLM) framework was used to accommodate the nested nature of the data (repeated measures nested within persons; Bolger & Laurenceau, 2013; Raudenbush & Bryk, 2002). We fit a series of four logistic multilevel models to test the first research question, which was whether recovery maintenance difficulty on a given day (the main predictor of interest) was associated with students’ odds of having contact with (1) family, (2) romantic partners, (3) 12-step sponsors, or (4) CRC peers that day. These models are represented by the general equation:

$$\begin{gathered} Level\text{1:}{\pi }_{it}=\mathrm{logistic}\left({\beta }_{0i}+{\beta }_{1i}Da{y}^{\prime }sRecoveryDifficult{y}_{it}\text{+}{\beta }_{\text{2i}}Da{y}_{it}\right) \\ Level\text{2:}{\beta }_{0i}={\gamma }_{00}+{\gamma }_{01}Se{x}_i+{\gamma }_{02}UsualRecoveryDifficult{y}_i+v_{0i} \\ {\beta }_{1i}={\gamma }_{10} \\ {\beta }_{2i}={\gamma }_{20} \end{gathered}$$

where the outcome (left side of the equation) has been “transformed” using the logistic link function, g(·)=log_e(p / 1−p), β_0i is the expected outcome on the first day of the study if all predictor variables are at their mean level for person i, and β_1i represents the predicted change in the outcome for each unit change in recovery maintenance difficulty for each individual.

Second, we fit a series of three linear multilevel models to test the second research question of whether recovery maintenance difficulty on a given day was associated with the amount of time that students spent in contact with (1) family, (2) CRC peers inside the drop-in center, and (3) CRC peers outside the drop-in center that same day. These models are represented by the general equation:

$$\begin{gathered} Level\text{1:}{Y}_{it}={\beta }_{0i}+{\beta }_{1i}Da{y}^{\prime }sRecoveryDifficult{y}_{it}\text{+}{\beta }_{\text{2i}}Da{y}_{it}\text{+}{\epsilon }_{it} \\ Level\text{2:}{\beta }_{0i}={\gamma }_{00}+{\gamma }_{01}Se{x}_i+{\gamma }_{02}UsualRecoveryDifficult{y}_i+v_{0i} \\ {\beta }_{1i}={\gamma }_{10} \\ {\beta }_{2i}={\gamma }_{20} \end{gathered}$$

Third, we fit two logistic multilevel models to test the third research question of whether recovery maintenance difficulty on a given day was associated with students’ odds of having recovery-focused conversations with (1) sponsors and (2) CRC peers that same day. These models are represented by the same general equation as the previous logistic models, with the only difference being the outcome variable (left hand side of the equation).

Finally, variables representing day-level fluctuations (collected each day and person-mean centered) and person-level means (averaged across all daily reports) for both negative affect and school stress fit for research questions 1–3 to address the fourth research question of whether recovery maintenance difficulty on a given day was associated with the outcomes of interest above and beyond potential effects of negative affect and school stress. For example, again we modeled the continuous outcomes of interest in the second research question using linear multilevel models, but with the addition of negative affect and school stress variables, represented by the general equation:

$$\begin{gathered} Level\text{1:}{Y}_{it}={\beta }_{0i}+{\beta }_{1i}Da{y}^{\prime }sRecoveryDifficult{y}_{it}\text{+}{\beta }_{\text{2i}}Da{y}^{\prime }sNegativeAffec{t}_{it}\text{+} \\ {\beta }_{3i}Da{y}^{\prime }s{\mathit{SchoolStress}}_{it}\text{+}{\beta }_{\text{4i}}Da{y}_{it}\text{+}{\epsilon }_{it} \\ Level\text{2:}{\beta }_{0i}={\gamma }_{00}+{\gamma }_{01}Se{x}_i+{\gamma }_{02}UsualRecoveryDifficult{y}_i+{\gamma }_{03}UsualNegativeAffec{t}_i+ \\ {\gamma }_{04}{\mathit{UsualSchoolStress}}_i+v_{0i} \\ {\beta }_{1i}={\gamma }_{10} \\ {\beta }_{2i}={\gamma }_{20} \\ {\beta }_{3i}={\gamma }_{30} \\ {\beta }_{4i}={\gamma }_{40} \end{gathered}$$

We fit multilevel models with random intercepts, but constrained slopes to be equal across participants to ensure model convergence. We tested participant sex (0 = male, 1 = female) and day of study as covariates in all models, and we retained them when significant to account for differences in the outcome when examining the effect of the main predictors of interest. We used the lme4 package in R (Bates, Bolker, & Walker, 2015) to fit both linear and generalized linear mixed effects models. The final analytic sample consisted of 1,180 days of data from 50 students, with an average of 23.64 (SD = 4.26, Range = 9 – 33) days each. Roughly 88% of respondents (n = 44) provided between 19 and 29 days of data. Because the duration of data collection varied across participants, we calculated compliance rates based on percentage of days complete across the number of total days with the survey device. Compliance percentage rates varied from 47% to 100%, with the average rate being 87%. Because of these high compliance rates, analyses assumed missingness at random. To examine this assumption, we examined associations between both total data days provided and compliance rate for each participant and their average recovery maintenance difficulty, negative affect, and school stress. None of these six correlations were significant.

3. Results

3.1. Descriptive statistics

We present descriptive statistics for all variables of interest in Table 1 and correlations among all study variables in Table 2. Of note, students talked to some member of their family on 59.1% of days (n = 695). Mothers were the family member who students most contacted (M = 0.44, SD = 0.50), followed by fathers (M = 0.29, SD = 0.46). Collectively, students talked to either one of their parents on 55% of study days (M = 0.55, SD = 0.50). Students also reported talking to siblings on 17% of study days (M = 0.17, SD = 0.38). Among days when students were in contact with their family, the most frequently reported length of time spent in contact was half an hour (44.9% of days on which contact occurred), with the length of time being one hour or less on approximately 86% of days. Further, 46% of students (n = 23) had a romantic partner at some point during the study. Students with partners saw them in person on 72.2% of days (n = 319).

Table 1.

Descriptive statistics for all study variables.

Variable	Frequency (valid %) or mean (SD)
Daily contact (yes/no)
Family	695 (59.1%)
Partner	319 (72.2%)
Sponsor	384 (40.0%)
Peers	950 (80.5%)
Amount of daily contact
Family
None to a few minutes	176 (25.3%)
½ hour	312 (44.9%)
1 hour	112 (16.1%)
2–3 hours	44 (6.3%)
4+ hours	51 (7.3%)
Peers (inside center)
Stopped in/said hi	30 (7.0%)
About 5 minutes	17 (4.0%)
5–15 minutes	72 (16.8%)
15–30 minutes	80 (18.6%)
30–60 minutes	99 (23.1%)
1–2 hours	83 (19.3%)
2 hours or more	48 (11.2%)
Peers (outside center)
None to a few minutes	28 (3.2%)
½ hour	48 (5.5%)
1 hour	74 (8.5%)
2–3 hours	230 (26.6%)
4+ hours	486 (56.1%)
Recovery conversations (yes/no)
Sponsor	138 (35.9%)
Peers	440 (46.4%)
Predictors
Recovery difficulty	M = 1.63 (SD = 0.91)
	Min = 1, Max = 5
Negative affect	M = 1.64 (SD = 0.63)
	Min = 1, Max = 5
School stress	M = 0.94 (SD = 1.09)
	Min = 0, Max = 4

Notes. Min = minimum; Max = maximum. Frequency values for binary yes/no variables are the frequency of the “yes” response.

Table 2.

Correlations among all study variables.

	1	2	3	4	5	6	7	8	9	10	11
Daily contact (yes/no)
Family
Partner	0.21***
Sponsor	−0.03	−0.13*
Peers	0.08**	0.15**	0.09**
Amount of daily contact
Family	--	−0.25***	0.03	−0.16***
Peers (inside center)	−0.07	−0.06	0.11*	--	−0.09
Peers (outside center)	0.01	0.13*	−0.10**	--	−0.02	0.07
Recovery conversations
Sponsor	0.02	0.02	--	0.03	0.06	0.03	−0.01
Peers	0.12***	0.01	0.05	--	0.10*	0.07	0.13***	0.07
Predictors
Recovery difficulty	0.04	0.05	0.00	0.01	0.15***	−0.01	0.00	0.14**	0.11***
Negative affect	−0.03	−0.12*	0.03	−0.03	0.07	0.01	−0.05	0.03	0.07*	0.44***
School stress	−0.07*	−0.13**	−0.01	0.01	−0.06	−0.05	−0.08*	−0.01	0.01	0.16***	0.36***

Note.

^***= p < 0.001.

^**= p < 0.01.

^*= p < 0.05.

Additionally, 92% of students (n = 46) had a 12-step sponsor at some point during the study. These students contacted their sponsors on 40% of days (n = 384) and had recovery-focused conversations on 35.9% of the days on which they had contact with their sponsors (n = 138). Adjusting for the campus drop-in center being closed on weekend days, participants stopped by the center on approximately 51% of possible days. Among days when they stopped by, the most frequently reported length of time spent at the center was between 30 and 60 minutes (23.1% of days), with the length of time being 30 minutes or more on 53.6% of the days they stopped by. Students spent time with CRC friends outside the center on 74.1% of days (n = 874) and, among days when they had at least some contact, the most frequently reported length of time spent was 4+ hours (56.1% of days). In total, students were in contact with CRC peers, whether inside or outside the center, on 80.5% of days (n = 950). Moreover, on days when they had contact with CRC peers, students had conversations with peers where the content discussed was recovery on 46.4% of days (n = 440).

3.2. Research Question 1: Is recovery maintenance difficulty on a given day associated with students’ odds of having contact with social network members?

Tables 3 and 4 provide results from the logistic multilevel models that assessed whether recovery maintenance difficulty was associated with students’ odds of having contact with different members of their social network that same day. The significant within-person, day-level association between recovery maintenance difficulty and family member contact indicated that, on average, students’ odds of having contact with family were significantly higher on days when they perceived recovery maintenance to be more difficult than usual. In contrast, students’ odds of having contact with sponsors, romantic partners, or CRC peers were not significantly higher on days with higher-than-usual recovery maintenance difficulty.

Table 3.

Associations between recovery maintenance difficulty and the odds of daily contact with family and romantic partners.

Model 1: Recovery Maintenance Difficulty
	Family Contact					Romantic Partner Contact
Fixed effects	Est	SE	p	CI	OR	Est	SE	p	CI	OR
Intercept	0.13	0.23	0.56	−0.33, 0.60	1.14	2.24	1.21	0.06	−0.18, 5.16	9.41
Day’s recovery difficulty	0.24	0.10	0.02	0.05, 0.43	1.27	−0.49	0.32	0.13	−1.15, 0.14	0.61
Usual recovery difficulty	−0.28	0.38	0.46	−1.04, 0.48	0.76	−0.01	2.26	0.99	−5.19, 4.85	0.99
Sex	1.42	0.49	0.004	0.46, 2.42	4.15	3.20a	2.49	0.20	−2.07, 9.04	24.64
Day of study	−0.01a	0.01	0.55	−0.02, 0.01	0.99	0.06a	0.04	0.07	−0.004, 0.14	1.07
Random effects	Est			CI		Est			CI
Intercept variance	1.70			1.03, 2.91		26.36			10.44, 82.75
Effect size estimate	Est					Est
R2	0.06					0.003
Model 2: Adding Negative Affect and School Stress
	Family Contact					Romantic Partner Contact
Fixed effects	Est	SE	p	CI	OR	Est	SE	p	CI	OR
Intercept	0.08	0.22	0.70	−0.35, 0.52	1.09	2.50	1.24	0.04	0.06, 5.54	12.21
Day’s recovery difficulty	0.20	0.11	0.07	−0.01, 0.41	1.22	−0.56	0.35	0.11	−1.27, 0.11	0.57
Usual recovery difficulty	0.17	0.40	0.67	−0.62, 0.97	1.19	1.59	2.52	0.53	−3.75, 7.53	4.91
Day’s negative affect	0.12	0.16	0.46	−0.20, 0.44	1.13	−0.08	0.57	0.89	−1.19, 1.08	0.93
Usual negative affect	−1.32	0.71	0.06	−2.78, 0.08	0.27	−5.93	4.94	0.23	−18.54, 3.38	0.003
Day’s school stress	−0.05	0.08	0.54	−0.20, 0.11	0.95	0.35	0.28	0.20	−0.17, 0.92	1.42
Usual school stress	−0.18	0.45	0.69	−1.09, 0.73	0.83	0.42	2.42	0.86	−4.73, 6.05	1.52
Sex	1.61	0.47	0.00	0.67, 2.59	5.02	3.09a	2.49	0.21	−2.20, 8.89	21.92
Day of study	−0.00a	0.01	0.64	−0.02, 0.01	1.00	0.06a	0.04	0.10	−0.01, 0.13	1.06
Random effects	Est			CI		Est			CI
Intercept variance	1.46			0.87, 2.51		25.71			10.06, 79.38
Effect size estimate	Est					Est
R2	0.10					0.10

Note. Est = Estimate. SE = Standard error. p = p-value. CI = 95% confidence interval. OR = Odds ratio. R² = Marginal coefficient of determination.

^a= Non-significant covariate, removed from the final model.

Table 4.

Associations between recovery maintenance difficulty and the odds of daily contact with sponsors and CRC peers.

Model 1: Recovery Maintenance Difficulty
	Sponsor Contact					CRC Peer Contact
Fixed effects	Est	SE	p	CI	OR	Est	SE	p	CI	OR
Intercept	−0.18	0.30	0.54	−0.79, 0.41	0.83	1.79	0.36	0.00	1.11, 2.55	5.98
Day’s recovery difficulty	0.19	0.11	0.07	−0.02, 0.40	1.21	−0.16	0.13	0.22	−0.42, 0.10	0.85
Usual recovery difficulty	−0.25	0.47	0.60	−1.19, 0.71	0.78	−0.17	0.59	0.77	−1.38, 1.03	0.84
Sex	0.83a	0.63	0.19	−0.43, 2.10	2.28	2.44	0.81	0.003	0.86, 4.13	11.51
Day of study	−0.03	0.01	0.00	−0.05, −0.01	0.97	−0.02a	0.01	0.16	−0.04, 0.01	0.98
Random effects	Est			CI		Est			CI
Intercept variance	2.65			1.55, 4.73		3.74			2.13, 6.95
Effect size estimate	Est					Est
R2	0.02					0.10
Model 2: Adding Negative Affect and School Stress
	Sponsor Contact					CRC Peer Contact
Fixed effects	Est	SE	p	CI	OR	Est	SE	p	CI	OR
Intercept	−0.24	0.30	0.43	−0.85, 0.36	0.79	1.74	0.36	0.00	1.06, 2.51	5.68
Day’s recovery difficulty	0.14	0.12	0.23	−0.09, 0.37	1.15	−0.10	0.15	0.50	−0.39, 0.19	0.90
Usual recovery difficulty	−0.42	0.56	0.46	−1.55, 0.71	0.66	0.02	0.66	0.97	−1.32, 1.34	1.03
Day’s negative affect	0.14	0.18	0.45	−0.22, 0.49	1.15	−0.31	0.22	0.15	−0.74, 0.11	0.73
Usual negative affect	0.62	1.01	0.54	−1.40, 2.67	1.86	−1.12	1.19	0.35	−3.50, 1.34	0.33
Day’s school stress	−0.03	0.08	0.76	−0.19, 0.14	0.97	0.09	0.11	0.39	−0.12, 0.31	1.10
Usual school stress	−0.02	0.61	0.98	−1.25, 1.23	0.98	0.51	0.73	0.49	−0.99, 1.99	1.66
Sex	0.78a	0.66	0.23	−0.54, 2.14	2.19	2.87	0.86	0.00	1.21, 4.67	17.58
Day of study	−0.03	0.01	0.00	−0.05, −0.01	0.97	−0.02a	0.01	0.11	−0.04, 0.004	0.98
Random effects	Est			CI		Est			CI
Intercept variance	2.63			1.54, 4.71		3.66			2.05, 6.92
Effect size estimate	Est					Est
R2	0.02					0.13

Note. Est = Estimate. SE = Standard error. p = p-value. CI = 95% confidence interval. OR = Odds ratio. R² = Marginal coefficient of determination.

^a= Non-significant covariate, removed from the final model.

3.3. Research Question 2: Is recovery maintenance difficulty on a given day associated with students’ amount of same-day contact with social network members?

We present results in Table 5 from the linear multilevel models that assessed whether recovery maintenance difficulty was associated with the amount of time students spent in contact with various members of their social network that same day. First, a significant within-person, day-level association between recovery maintenance difficulty and amount of daily contact with family revealed that on days when students were in contact with family, the length of time they spent in contact was significantly higher on days when they perceived recovery maintenance to be more difficult than usual. In contrast, recovery maintenance difficulty was not associated with the amount of time that students spent in contact with CRC peers inside nor outside the drop-in center.

Table 5.

Associations between recovery maintenance difficulty and the amount of daily contact with social network members.

Model 1: Recovery Maintenance Difficulty
	Family				CRC Peers (inside center)				CRC Peers (outside center)
Fixed effects	Est	SE	p	CI	Est	SE	p	CI	Est	SE	p	CI
Intercept	2.20	0.09	0.00	2.02, 2.38	4.38	0.16	0.00	4.07, 4.70	2.98	0.11	0.00	2.77, 3.19
Day’s recovery difficulty	0.21	0.05	0.00	0.12, 0.31	0.03	0.10	0.75	−0.17, 0.23	−0.06	0.04	0.15	−0.15, 0.02
Usual recovery difficulty	0.14	0.17	0.40	−0.18, 0.47	−0.04	0.30	0.90	−0.62, 0.54	0.09	0.18	0.64	−0.27, 0.44
Sex	0.18a	0.21	0.40	−0.23, 0.59	−0.54a	0.38	0.16	−1.28, 0.19	0.32a	0.22	0.16	−0.11, 0.74
Day of study	0.00a	0.00	0.44	−0.01, 0.01	0.01a	0.01	0.31	−0.01, 0.03	0.01	0.004	0.00	0.01, 0.02
Random effects	Est		CI		Est		CI		Est		CI
Intercept variance	0.32		0.18, 0.50		0.97		0.55, 1.55		0.37		0.21, 0.58
Effect size estimate	Est				Est				Est
R2	0.03				0.0003				0.01
Model 2: Adding Negative Affect and School Stress
	Family				CRC Peers (inside center)				CRC Peers (outside center)
Fixed effects	Est	SE	p	CI	Est	SE	p	CI	Est	SE	p	CI
Intercept	2.21	0.09	0.00	2.04, 2.38	4.42	0.16	0.00	4.10, 4.73	3.02	0.10	0.00	2.82, 3.22
Day’s recovery difficulty	0.27	0.06	0.00	0.16, 0.37	0.02	0.11	0.88	−0.20, 0.24	−0.04	0.05	0.39	−0.13, 0.05
Usual recovery difficulty	0.10	0.20	0.63	−0.28, 0.47	−0.02	0.35	0.94	−0.69, 0.64	0.19	0.20	0.36	−0.19, 0.56
Day’s negative affect	−0.13	0.09	0.13	−0.31, 0.04	0.07	0.16	0.64	−0.23, 0.38	−0.06	0.08	0.43	−0.21, 0.09
Usual negative affect	0.21	0.34	0.54	−0.44, 0.86	0.10	0.58	0.86	−1.01, 1.22	−0.06	0.32	0.85	−0.67, 0.54
Day’s school stress	−0.06	0.04	0.12	−0.15, 0.02	−0.04	0.08	0.63	−0.18, 0.11	−0.03	0.03	0.37	−0.10, 0.04
Usual school stress	−0.14	0.22	0.53	−0.57, 0.29	−0.43	0.39	0.27	−1.18, 0.31	−0.38	0.22	0.10	−0.80, 0.05
Sex	0.14a	0.22	0.53	−0.28, 0.55	−0.64a	0.39	0.11	−1.37, 0.09	0.30a	0.21	0.17	−0.10, 0.70
Day of study	0.00a	0.00	0.43	−0.01, 0.01	0.01a	0.01	0.31	−0.01, 0.03	0.01	0.004	0.00	0.01, 0.02
Random effects	Est		CI		Est		CI		Est		CI
Intercept variance	0.31		0.17, 0.48		0.97		0.51, 1.48		0.32		0.17, 0.48
Effect size estimate	Est				Est				Est
R2	0.04				0.01				0.04

Note. Est = Estimate. SE = Standard error. p = p-value. CI = 95% confidence interval. R² = Marginal coefficient of determination.

^a= Non-significant covariate, removed from the final model.

3.4. Research Question 3: Is recovery maintenance difficulty on a given day associated with students’ odds of having recovery-focused conversations with social network members?

Table 6 provides results from the logistic multilevel models that assessed associations between recovery maintenance difficulty and students’ same-day recovery-focused conversations with social network members. The significant within-person, day-level association between recovery maintenance difficulty and recovery-focused conversations with 12-step sponsors indicated that on days when students were in contact with their sponsors and perceived recovery maintenance to be more difficult than usual, the odds of having conversations about recovery were higher relative to the odds of not having recovery-focused conversations. Similarly, the significant day-level association between recovery maintenance difficulty and recovery-focused conversations with CRC peers demonstrated that on days when students were in contact with CRC peers and perceived recovery maintenance to be more difficult than usual, the odds of having recovery-focused conversations with peers were higher relative to the odds of not having recovery-focused conversations.

Table 6.

Associations between recovery maintenance difficulty and the nature of daily contact with social network members.

Model 1: Recovery Maintenance Difficulty
	Sponsor					CRC Peers
Fixed effects	Est	SE	p	CI	OR	Est	SE	p	CI	OR
Intercept	−0.72	0.16	0.00	−1.07, −0.41	0.49	−0.58	0.22	0.01	−1.02, −0.15	0.56
Day’s recovery difficulty	0.28	0.14	0.05	0.001, 0.56	1.32	0.24	0.10	0.02	0.05, 0.44	1.27
Usual recovery difficulty	0.41	0.27	0.13	−0.15, 0.97	1.51	0.25	0.30	0.41	−0.35, 0.86	1.28
Sex	−0.20a	0.35	0.56	−0.91, 0.51	0.81	0.75	0.37	0.04	0.02, 1.49	2.12
Day of study	0.00a	0.01	0.74	−0.02, 0.03	1.00	0.01	0.01	0.15	−0.00, 0.03	1.01
Random effects	Est			CI		Est			CI
Intercept variance	0.35			0.08, 0.95		0.90			0.50, 1.63
Effect size estimate	Est					Est
R2	0.02					0.04
Model 2: Adding Negative Affect and School Stress
	Sponsor					CRC Peers
Fixed effects	Est	SE	p	CI	OR	Est	SE	p	CI	OR
Intercept	−0.70	0.16	0.00	−1.05, −0.39	0.50	−0.60	0.22	0.01	−1.04, −0.17	0.55
Day’s recovery difficulty	0.29	0.17	0.09	−0.04, 0.63	1.34	0.15	0.11	0.16	−0.06, 0.37	1.17
Usual recovery difficulty	0.58	0.31	0.06	−0.05, 1.23	1.79	0.42	0.33	0.21	−0.25, 1.10	1.52
Day’s negative affect	−0.01	0.28	0.98	−0.57, 0.55	0.99	0.27	0.18	0.13	−0.08, 0.61	1.31
Usual negative affect	−0.71	0.58	0.22	−1.92, 0.42	0.49	−0.43	0.57	0.46	−1.59, 0.73	0.65
Day’s school stress	0.01	0.12	0.94	−0.23, 0.25	1.01	0.05	0.08	0.54	−0.11, 0.21	1.05
Usual school stress	0.08	0.36	0.82	−0.63, 0.82	1.09	−0.14	0.39	0.72	−0.91, 0.64	0.87
Sex	−0.14a	0.37	0.71	−0.86, 0.59	0.87	0.81	0.37	0.03	0.06, 1.56	2.24
Day of study	0.00a	0.01	0.83	−0.02, 0.03	1.00	0.01	0.01	0.13	−0.004, 0.03	1.01
Random effects	Est			CI		Est			CI
Intercept variance	0.34			0.07, 0.93		0.87			0.47, 1.58
Effect size estimate	Est					Est
R2	0.03					0.05

Note. Est = Estimate. SE = Standard error. p = p-value. CI = 95% confidence interval. OR = Odds ratio. R² = Marginal coefficient of determination.

^a= Non-significant covariate, removed from the final model.

3.5. Research Question 4: Is recovery maintenance difficulty uniquely associated with the occurrence, amount, and nature of daily contact, above and beyond negative affect or school stress?

Tables 3–6 provide results from models assessing whether recovery maintenance difficulty was uniquely associated with the occurrence, amount, and nature of students’ daily contact with social network members, above and beyond daily negative affect and school stress. Results indicated that amount of family contact remained significantly higher on days when students perceived recovery maintenance to be more difficult than usual, above and beyond effects of negative affect and school stress. The odds of students having contact with family and having recovery-focused conversations with CRC peers remained marginally higher on days when they perceived recovery maintenance to be more difficult than usual (ps < .10), though these associations did not reach statistical significance at the p < .05 level after adding the nonsignificant effects of negative affect and school stress to the model. Recovery maintenance difficulty was not associated with students’ odds of having contact with romantic partners, sponsors, or CRC peers, nor with the amount of daily contact with CRC peers, after accounting for negative affect and school stress.

We also provide pseudo-R² values for each model in Tables 3–6 to gauge the substantive importance of the fixed effects portions of each of our models (i.e., the marginal coefficient of determination). Effects were relatively small (ranging from 0.0003 to 0.13), indicating that although these variables are important, future research may also want to consider additional variables to further explain contact with social network members.

4. Discussion

An important yet understudied question regarding the social support that CRCs provide is how students access their sources of support on days when they need it most for the purposes of maintaining recovery. The current study used daily diary data to understand with which social contacts students engaged daily (including family, romantic partners, sponsors, and CRC peers) and when students connected with them (e.g., on days when students experienced higher-than-usual recovery maintenance difficulty, negative affect, and/or school stress). On days when students reported having greater difficulty with recovery maintenance than usual, they had higher odds of being in contact with family members and were expected to talk to or spend time with family for longer than usual. In addition, on days with higher-than-usual recovery maintenance difficulty, students had higher odds of having recovery-focused conversations with both CRC peers and 12-step sponsors. Adding negative affect and school stress to the models revealed that recovery maintenance difficulty was uniquely associated with longer duration of family contact.

These findings should be interpreted keeping in mind that students in a large CRC, such as the one in this study, likely have large amounts of recovery capital. As just one example, 92% of the students in this study had a 12-step sponsor. Prior work on this sample has shown members of this recovery community have more than double the number of abstainers than drinkers in their social networks (Cleveland, Wiebe, & Wiersma, 2010). Possibly due to large amounts of capital, levels of daily recovery maintenance difficulty, negative affect, and school stress were all relatively low, suggesting that this may have been a fairly low-risk sample. Still, recent research has found that within-CRC differences in capital, such as number of social network connections, were related to months since relapse (Patterson et al., 2020).

Against this backdrop, one interesting aspect of our findings was that higher-than-usual recovery maintenance difficulty was not associated with the daily occurrence or amount of CRC peer contact. These findings may indicate that students’ amount of time at the CRC was routinized into their daily schedules. For example, students may have had time in their course schedule to spend at the drop-in center on certain days of the week and not others. Moreover, descriptive findings indicated that students were in contact with CRC peers, whether inside or outside the center, on roughly 80.5% of days and had recovery-focused conversations on 46.4% of days. In other contexts—where CRC peer contact is not a regular part of the members’ lives to the same degree—there might be more meaningful variability in contact with CRC peers. But at a large CRC where members often lived with each other in apartments or recovery residences, the occurrence or amount of daily contact with peers did not vary based on perceived need, whereas the content of the conversations was more focused on recovery when needed. Likewise, contact with romantic partners was also very frequent (72.2% of days among students with partners), and thus the occurrence of contact was unlikely to vary based on perceived need.

A similar story emerged for 12-step sponsors. Although contact with sponsors was less frequent than with peers (students who had sponsors contacted them on 40% of days), it was the content, rather than the occurrence, of interactions that varied according to daily levels of perceived recovery maintenance difficulty. That is, interactions were not any more or less likely to occur with sponsors on more difficult recovery days, but when they did occur, the odds of their content being recovery focused were higher on days when recovery maintenance difficulty was higher than usual. Thus, findings suggest that students may have capitalized upon existing interactions with both peers and sponsors by shifting the focus to recovery-specific issues on days when they were experiencing more recovery maintenance difficulty than usual.

In contrast, students appeared to utilize support from family members more sparingly and specifically on days characterized by higher-than-usual recovery maintenance difficulty. Contact with family members was less routine than CRC peer or partner contact: students talked with family on fewer days (59.1%) and for less time when they did talk. This amount of contact time likely reflected the context of the large CRC at a large university, and might be higher at smaller CRCs based at institutions that serve local communities. However, both the occurrence and amount of family contact were uniquely higher on more difficult recovery days. Thus, students may have been strategic and timely with their use of support from family—they had contact with them on fewer days and for a shorter amount of time than peers or romantic partners, yet did so more often and for longer on days when recovery maintenance was more difficult than usual. Family members may be well-positioned to be important sources of social recovery support, specifically on days when students are struggling more with their recoveries. Families should be in the best position possible to provide effective support, given that members can simultaneously be impacted by another member’s SUD and a source of recovery support (Ventura & Bagley, 2017).

Taken together, results are consistent with previous research that suggests that students need access to multiple sources of social support that they can draw upon to maintain recovery (e.g., Boeri et al., 2015). Students may access the resources they need (such as family support) more so on especially difficult days, while also capitalizing on existing interactions that are already built into their everyday lives (such as with peers and sponsors) by discussing important issues relevant to their recoveries on days when recovery maintenance is more difficult than usual.

4.2. Implications

These findings have implications for future research, as well as program and intervention development within and outside the CRC context. First, results align with the notion that an ability to access recovery support when it is needed (i.e., in response to a perceived need) is itself an important aspect of recovery capital (Best et al., 2014). Even if capital is available, as in the case of individuals who enjoy substantial social support for recovery by virtue of belonging to a CRC, the individuals’ ability to access this capital on days when they need it most may have important implications for sustaining recovery. Future research is needed to understand whether individual differences in the ability to access support on a daily basis have implications for the maintenance of recovery in the longer-term. Moreover, a better understanding of students’ daily use of social support might assist treatment interventions. For example, interventions could help students to recognize the social support that they have at their disposal, identify different relapse risks as they unfold in daily life, and learn important strategies for accessing the sources of support that are most appropriate for them during critical moments.

4.3. Limitations

These findings should be viewed in light of several important limitations. First, the students in this sample came from one of the largest and most well-established CRCs in the United States, and thus likely represented a “best case scenario” in terms of the amount of recovery capital they could access. A large amount of variability exists across CRCs regarding the amount of capital that they can provide. In addition, the generalizability of the findings may be further limited by the fact that the sample was almost exclusively non-Hispanic white, the data were collected several years ago (2004–2005), and older participants were dropped from the analytic sample because some of the aspects of the data analyzed herein (in particular the items about family social contact) were designed with early adults in mind. Future studies should examine the extent to which findings extend across time, to smaller CRCs or other recovery community contexts, and to samples that are more diverse in terms of race/ethnicity and age. For example, studies could examine similar processes among those living in recovery residences, whose developmental, educational, and economic experiences differ from participants in this study.

Second, the once-a-day daily diary data collection protocol substantially limited our ability to critically examine the temporality of the associations that we examined. For example, we do not necessarily know whether perceiving more recovery challenges on a given day preceded students’ contact with family members, or if it was the other way around. However, we suggest that several aspects of the findings lend support to the interpretations that we provided. First, higher recovery maintenance difficulty was not only positively associated with students’ daily contact with family, but also with having conversations with both CRC peers and 12-step sponsors that were recovery focused (the study did not collect data on the content of interactions with family members or romantic partners, a limitation that makes comparisons more difficult). Second, adding negative affect to the model revealed that it was not associated with the occurrence, amount, or nature of daily contact with social network members. We would hypothesize that if the temporal order were actually reversed, such that more daily family contact preceded higher recovery maintenance difficulty, then family contact would not only be linked with recovery difficulty, but would also be associated with higher negative affect. The lack of association with negative affect lends support to our interpretation. Third, responses to the baseline questionnaire indicated that students in this sample experienced very high levels of family support on average (Median = 17 on a 20-point scale), suggesting that family was more likely to support rather than challenge recovery within this sample. Still, collecting data from participants multiple times per day, such as via ecological momentary assessment (EMA), would permit further testing of same-day temporality assumptions.

Fourth, the findings regarding the potential role of romantic partners were limited in that only about half of the sample had a romantic partner at any point during the study. In contrast, all or nearly all participants reported at least some contact with family members, 12-step sponsors, and CRC peers. Thus, findings should not be taken to suggest that romantic partners do not play an important role in recovery maintenance; these relationships should be the subject of more thorough investigation in future research. Finally, these analyses did not address the mechanisms by which utilizing social relationships may be helpful to recovery maintenance. Future studies could explore whether daily contact with social network members serves to reduce perceived recovery challenges by, for example, increasing recovery commitment and/or mitigating the impact of negative mood or craving.

5. Conclusions

Findings from daily diary data suggested that the occurrence of students’ interactions with sponsors and peers did not vary significantly according to day-to-day fluctuations in perceived recovery difficulty, but that when interactions did occur, the odds that they were recovery focused were higher on more difficult recovery days. In contrast, students connected with family members more sparingly and specifically on days when they perceived recovery maintenance to be more difficult. Collectively, these findings underscore the importance of providing multiple sources of social support to college students that they can use on a daily basis to maintain their recoveries despite the challenges that they face.

Highlights.

• Students contacted family more often and for longer on challenging recovery days

• Conversations with sponsors and peers were more recovery-focused on difficult days

• Findings highlight need to provide students multiple sources of daily support

• Understanding students’ daily use of support might make it possible to intervene