Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2022 Jun 1.
Published in final edited form as: Soc Sci Med. 2021 May 10;279:114021. doi: 10.1016/j.socscimed.2021.114021

Substance use behaviors in the daily lives of U.S. college students reporting recent use: The varying roles of romantic relationships

Shari M Blumenstock 1,*, Lauren M Papp 1
PMCID: PMC8266259  NIHMSID: NIHMS1702997  PMID: 34004570

Abstract

Rationale:

While the detrimental consequences of problematic substance use behaviors in early adulthood are well-documented, the interpersonal predictors of substance use in the naturalistic context of daily life are less known.

Methods:

Using ecological momentary assessments to capture moments of binge drinking, marijuana use, nicotine use, and prescription misuse in everyday life, this study explored the romantic relationship contexts (status, quality, partner presence) of substance use among young adults (aged 18–21) attending a university in the Midwestern U.S. Between 2017 and 2019, 296 participants, who had endorsed recent substance use at screening for a larger study on prescription misuse, completed four reports per day for approximately 28 days, resulting in over 23,000 reports for this investigation.

Results:

Relationship status, partner support, and partner presence in the moment were all associated with at least one type of substance use behavior. Generalized multilevel models indicated that partnered participants engaged in less marijuana use, nicotine use, and prescription misuse in daily life compared to single participants. Higher partner support predicted less binge drinking, yet also predicted higher nicotine use and prescription misuse. When with a romantic partner in daily life, partnered participants were more likely to engage in binge drinking and marijuana use, and less likely to misuse prescription drugs. Nicotine use was more likely when with a partner, but only if partner support was high.

Conclusions:

Findings highlight the intricate links between interpersonal contexts and substance behaviors and indicate supportive relationships are not universally protective against substance use among this population.

Keywords: College students, Ecological momentary assessment, Health behaviors, Relationship quality, Romantic relationships, Substance use, United States, Young adults

Introduction

A major contributor to disease and mortality across the globe is the problematic use of substances such as alcohol, marijuana, nicotine, and prescription drugs, which collectively cause millions of deaths every year (United Nations Office on Drugs and Crime [UNODC], 2018; World Health Organization [WHO], 2019). In many industrialized nations the overall use of such substances is highest among late adolescents and young adults (ages ranging approximately 18–25) (UNODC, 2019), with particularly high rates among those in university settings (Schulenberg & Maggs, 2002), putting young adult college students at increased risk of numerous associated health problems such as substance use disorders and premature death (WHO, 2019). For instance, in 2017 young adults reported use at 85% for alcohol, 44% for marijuana, 27% for cigarettes, and 14% for prescription stimulant misuse (Schulenberg et al., 2019). Addiction and other substance use disorders are also highest among this age group (Substance Abuse and Mental Health Services Administration [SAMSHA], 2019), with over 10% of young adults (aged 18 to 25) in the U.S. estimated to have an alcohol use disorder and over 7% with an illicit drug use disorder, compared to 5% and 3%, respectively, among the overall population (SAMSHA, 2019). In light of elevated rates and dire consequences of problematic alcohol, marijuana, nicotine, and prescription drug use—combined with the substance-conducive environments of college and the critical developmental period of young adulthood that makes substance use particularly risky for later health and well-being—understanding the factors associated with college students’ substance use in daily life remains a pressing public health concern. Critically, young adulthood is also a period of increased interest and involvement in romantic relationships, which are known to play important roles in health behaviors such as substance use (Derrick et al., 2019; Johnson & Acabchuk, 2018; Rhule-Louie & McMahon, 2007). Drawing from a sample of college students in the United States at elevated risk of substance use, this study investigates the role of romantic relationship contexts in young adults’ various substance use behaviors throughout their daily lives.

Substance Use in Young Adults and College Students

Despite recently documented modest decreases in binge drinking and tobacco use among young adults, the rates remain high in many industrialized nations (UNODC, 2018). Marijuana is by far the most commonly used drug among young adults across the world, and rates of marijuana use have been increasing in most industrialized countries, including the U.S. (National Institute on Drug Abuse [NIDA], 2019; UNODC, 2019). In addition to the established risks of substance use disorders and premature death associated with the problematic use of alcohol, marijuana, and tobacco (UNODC, 2019; WHO, 2019), two other forms of problematic substance use have been recognized as major public health concerns. First, in the wake of targeted, successful global marketing programs, the use of electronic cigarettes (‘e-cigs’), which typically contain highly addictive nicotine but no tobacco, has risen dramatically, particularly among adolescents and young adults (Schulenberg et al., 2019; WHO, 2018, 2019). In 2018, use was estimated to be over 40 million adults worldwide, with annual prevalence rates up to 27% among young adults, in contrast to the steadily decreasing rates of tobacco use (Schulenberg et al., 2019; UNODC, 2019; WHO, 2019). Second, the medicalization and pharmaceuticalization of society and health (Abraham, 2010; Bell & Figert, 2012) have resulted in a steady rise of prescription medications dispensed to the public, thereby increasing access for misuse either by the intended patient or by someone else. Psychoactive prescription drugs, e.g., opioid pain relievers, stimulants, and anti-anxiety sedatives and tranquilizers, are particularly prone to abuse, with misuse rates estimated to be over 14% among young adults in the U.S. and up to 17% among youth in other countries (NIDA, 2019; EMCDDA, 2016; WHO, 2019). Notably, the rates of past-year prescription misuse in young adults are the highest among any other age group worldwide (NIDA, 2019; UNODC, 2018). Furthermore, prescription misuse serves as the leading cause of drug-related deaths among young adults in the U.S. (NIDA, 2018) and is highly correlated with other substance use (Martin, 2008). Thus, problematic substance use among young adults is highly prevalent with high potential for harm.

Elevated rates of overall substance use among young adults can be explained in part by sociocultural contexts. Following adolescence, the developmental period of young adulthood is characterized by increasing autonomy, identity exploration, and peer influence (Arnett, 2000), and, while not all 18–24-year-olds attend college, a substantial portion experience the transition from adolescence to young adulthood in a college or university setting—about 40% in the U.S. (Hussar et al., 2020), with the vast majority living away from parents (U.S. Department of Education, 2017). College attendance is typically accompanied by high expectations for academic achievement in addition to recreation and personal discovery. In response, college students may use a variety of substances to meet these expectations or to cope with numerous transitions and peer influence (Quintero et al., 2006; Rhule-Louie & McMahon, 2007; Schulenberg & Maggs, 2002). Combined with decreased supervision and increased unstructured time (Quintero et al.), college settings can become particularly encouraging environments for substance use. Thus, the developmental context of young adulthood, when coupled with the social context of college life, may make the use of substances especially appealing.

Young Adults’ Romantic Relationships and Substance Use

In addition to elevated substance use among young adults, a defining feature of the young adult developmental period is the increasing importance of romantic partners in behaviors and decision-making (Arnett, 2000). Indeed, scholars have long documented the powerful role that intimate relationships can play in potentially risky health behaviors like substance use (Johnson & Acabchuk, 2018; Meyler et al., 2007; Rhule-Louie & McMahon, 2007; Umberson, 1992). While previous findings have mostly been observed among married couples, some evidence suggests that any committed relationship reduces young adults’ likelihood of using certain substances. College students who reported being in a romantic relationship during data collection reported less frequent alcohol use in the previous 30 days compared to single participants (Braithwaite et al., 2010). Similarly, in a longitudinal study that retrospectively assessed young adults’ post-high school substance use, those who were in dating relationships reported decreased heavy alcohol and marijuana use compared to those who were single (Fleming et al., 2010). These two studies suggest that, among young adults, being in a committed romantic relationship is associated with less frequent heavy alcohol and marijuana use compared to being single. Conversely, tobacco (Brathwaite et al.) and cigarette (Fleming et al.) use were not associated with relationship status in these studies. Notably, these previous studies used retrospective reports, which could affect reporting, and they did not include e-cigarettes when assessing nicotine use, limiting applicability to the present day. In addition to e-cigarette use, prescription misuse has also not been widely explored within the context of romantic relationships. Given the established importance of romantic relationships in health behaviors and in young adults’ decision-making, relationship status may be an essential element of understanding, and potentially mitigating, several types of currently prevalent substance use behaviors in young adult college students. Furthermore, studies assessing the relationship context of substance use behaviors in daily life (versus retrospectively) remain rare.

The quality of such romantic relationships has also been linked to substance use behaviors (Rhule-Louie & McMahon, 2007). Low-quality relationships, such as those characterized by low levels of support, can be detrimental to well-being (Walen & Lachman, 2000), and the long-term connections between relationship quality and health may be mediated by unhealthy behaviors undertaken as an emotional response to relationship discord (Roberson et al., 2018). Such low-quality relationships can increase stress and subsequent coping via substances or can hinder substance use desistence (Johnson & Acabchuk, 2018). Among married couples, higher relationship quality has been associated with lower prescription misuse among wives (Homish et al., 2010) and a lower likelihood of alcohol problems (Leonard & Homish, 2008). In the longitudinal study of young adults, higher relationship quality was associated with decreased cigarette use (regardless of partner use) and decreased heavy drinking and marijuana use when the partner’s use was none to moderate (Fleming et al., 2010). For young adult men and women in romantic relationships, lower relationship quality predicted higher substance use problems (Simon & Barrett, 2010). Overall, low-quality relationships may be contexts in which substance use is relatively high. Extending these findings from retrospective reports, instances of substance use in daily life may be higher among individuals in less supportive relationships.

On a momentary level, individuals in romantic relationships could make substance use decisions based on the presence of their partner. Substance use is often a social behavior (Quintero et al., 2006; Schulenberg & Maggs, 2002), and given the heightened influence of romantic partners in young adults’ health behaviors, the presence of a partner may determine whether an individual chooses to use a substance in a given moment. Previous studies on substance use have investigated the role of a romantic partner’s use (Rhule-Louie & McMahon, 2007), yet do not assess the role of the romantic partner’s presence in moments when substances are being used. A handful of studies have investigated whether an individual’s overall substance use is congruent with a romantic partner’s use (Crane et al., 2016; Derrick et al., 2019; Homish & Leonard, 2005), and a diary study of prescription misuse among 46 mixed-sex couples in which at least one partner reported recent prescription misuse indicated that women who reported being with their partner more frequently overall were less likely to engage in prescription misuse (Papp & Blumenstock, 2016). These studies offer preliminary evidence that time spent with a partner may be related to substance use, yet little is known about whether being with a partner in a given moment makes substance use more or less likely for individuals who use substances.

Preliminary evidence further indicates a potentially complex interplay between partner presence and substance use. Dyadic studies on couple substance use indicate that relationship quality and individuals’ substance use are interrelated with the partners’ use, such that relationship quality is higher among those who use substances with a partner (versus at different frequencies) (Crane et al., 2016; Homish & Leonard, 2005). This suggests that any association between partner presence and substance use may depend on relationship quality. However, this has not been examined on a momentary level. If substance use is compensating for a low-quality relationship, the association between partner presence and substance use may be stronger for individuals in less supportive relationships. Alternatively, substance use could be a valued shared activity that promotes intimacy, and therefore the association could be stronger for those in more supportive relationships.

The Current Study

The purpose of the current study was to assess the roles of romantic relationship factors in college students’ substance use behaviors in daily life, using ecological momentary assessment (EMA) methods and drawing from a U.S. sample likely to report substance use behaviors (due to recent prescription misuse; Martin, 2008), optimizing the ability to capture substance use in daily life. The study expands previous research in three key ways. First, with EMA methods, reports are collected in real time and in participants’ real-world environments, thus minimizing recall biases and maximizing ecological validity (Shiffman et al., 2008). Time-based EMA methods collect data at randomized times within specific time periods, and are therefore designed to capture a reasonably representative sample of participants’ daily experiences. Combined with event-based assessments triggered by participants’ substance use behavior or behavioral intentions, EMA methods are well-suited to assess moments of substance use and contextual factors surrounding such use (Shiffman, 2009). The current study used time-based assessments for all substances, as well as event-based assessments for prescription misuse. Importantly, the collection of repeated measures allows for assessing between-person individual differences across the reporting period as well as within-person momentary differences that can compare individual’s moments of substance use versus moments of nonuse. This approach also offers unique insight into the moment-level roles of romantic relationships in health behaviors. Second, several types of substance use behaviors currently prevalent among young adult college students are assessed. In addition to binge drinking and marijuana use, we include e-cigarettes when assessing nicotine use; we also assess prescription misuse, which the study was specifically designed to capture. Third, the study design allows for documenting associations between relationships and substance use in three distinct ways: 1) by comparisons across relationship status, 2) as a function of partner support, and 3) by assessing within-person differences of real-time partner presence on a momentary basis and exploring moderating effects of partner support in this association. Accordingly, we tested the following research questions for four substance use behaviors: binge drinking, marijuana use, nicotine use, and prescription misuse.

RQ1: Do partnered participants report less substance use in daily life compared to single participants? We expect partnered participants will report lower likelihoods of substance use (Braithwaite et al., 2010; Fleming et al., 2010).

RQ2: For partnered participants, is partner support associated with substance use behaviors throughout daily life? We expect lower levels of support to be associated with higher substance use (Fleming et al., 2010; Simon & Barrett, 2010).

RQ3: a) Does real-time partner presence predict substance use behaviors in the moment? b) If so, does partner support moderate these within-person associations? Given indirect and mixed results in previous work we offer no definite hypotheses for these exploratory questions.

Additionally, substance use is associated with adverse family environments during childhood, such as with caregivers who were abusive or neglectful, or who used substances themselves (for reviews, see Repetti et al., 2002 and UNODC, 2018, pp. 25–36). Such adverse early experiences are also linked to negative relationship outcomes in young adulthood (Connolly et al., 2014). The analyses, therefore, include a measure of childhood family adversity (Repetti et al., 2002), which captures several of these experiences, as a control variable.

Methods

Data is drawn from the first assessment of an ongoing longitudinal study of college students’ health and well-being (Papp et al., 2020). All procedures were approved by the associated Institutional Review Board. Participants attended two in-person lab visits with approximately 28-days of ecological momentary assessments (EMA) collected in between. The current study was preregistered before analyses (ED.: Include URL to registration).

Participants and Procedures

From Fall 2017 to Fall 2019, freshmen and sophomores aged 18–21 at a university in the Midwestern U.S. were continuously recruited via mass emails, flyers, and targeted ads on social media and student newsletters. A confidential phone screen procedure was used to determine recent (past 3 months) misuse of prescription pain medications, stimulants, tranquilizers, or sedatives (for details, see identical protocol from a different study, Papp and Blumenstock, 2016). During the first lab visit, participants completed informed consent procedures and questionnaires, received an iPod touch device, and completed training for the EMA portion of the study. Following the EMA reporting period, participants returned for a second lab visit, completed additional questionnaires, returned their device, and received payment up to USD$250 ($75 for visit 1, $55 for visit 2, and $120 for EMA completion, which includes a $36 bonus incentive for timely completion). All received a list of substance, relationship, and mental health resources. Data were then downloaded to secure servers.

Of the 300 participants endorsing recent misuse, two did not return their device, a glitch prevented one’s data from downloading, and another did not complete the family adversity measure, resulting in a final sample of 296. Of these, 205 (69.3%) were women and 96 (30.7%) were men. The majority identified as Caucasian/White (83.4%), with 7.1% identifying as multiple races, 5.8% identifying as Asian, and less than 3% identifying as American Indian or Alaskan Native, Black or African American, Native Hawaiian or Pacific Islander, or Other; 6.4% identified as Hispanic. The partnered sample consisted of 107 individuals in dating relationships (partner genders unknown). There were no statistically significant differences in proportions of gender (χ2=0.31, p = .602), race (based on all seven categories; χ2=8.05, p = .235), or Hispanic identity (χ2=4.16, p = .050) between partnered and single participants.

EMA procedures were designed to capture several substance use behaviors, with an emphasis on prescription misuse. Participants completed the EMA reports via the iPod touch device and could complete new reports at any time. All other functionality of the devices (e.g., internet connectivity) were disabled; live-monitoring was not possible. During the EMA reporting period, the iPod touch device sent a notification four times daily, once at random within four time periods (morning, afternoon, evening, night) (Shiffman et al., 2008), indicating it was time to complete an EMA report. The EMA reports took less than 2 minutes to complete on average, and included questions about participants’ location, who they were with, substance use (alcohol, marijuana, nicotine), mood, and recent events. Additionally, to assess prescription misuse, a portion of each EMA report asked whether participants were about to use any of four types of prescription medication (pain medications, stimulants, tranquilizers, or sedatives) either without a prescription or in a way a doctor had not prescribed. If they indicated intention to misuse any drug, a notification was sent 15 minutes after the EMA report was submitted, and the participants were given access to the follow-up report. The follow-up report asked whether or not any misuse actually occurred, along with other questions not relevant to the current study. Follow-up reports took less than a minute to complete on average.

To more fully capture prescription misuse in daily life, participants were also instructed to self-initiate an EMA report any time they intended to misuse a prescription medication (Shiffman, 2009). Thus, an EMA report could be completed at any time and there were no restrictions on the number of reports that could be submitted. The time-based and event-based reports were therefore identical (i.e., indistinguishable) in the data. This combined design resulted in 23,260 reports from which the current data are drawn. For more details about app development, completion rates, and sample demographic characteristics, see (Papp et al., 2020).

Measures

Relationship Variables

During the first lab visit, participants were asked whether they were currently in a romantic relationship (Yes/No). If yes, several aspects of that relationship were assessed. Relationship length was converted to months. Participants were also asked to choose which of six relationship types described theirs the best: married, engaged, dating exclusively, dating frequently but not exclusive, dating once in a while, or only having sex. The only option selected by any participant was dating relationship.

Relationship quality was assessed by the Partner Support and Partner Strain scales for intimate partners (Walen & Lachman, 2000). The four items for support, answered on a 4-point Likert-type scale (1 = a lot; 4 = not at all) assessed how much their partner understands their feelings and cares about them, and how much they can rely on their partner for help and open up to them about worries. Items were reverse-scored so higher values indicated higher support. Reliability as gauged by Cronbach’s α= .69. Partner Strain items exhibited less than adequate reliability (α = .55), indicating a weak measure for this sample, and were removed from analyses.

During the EMA reports, participants indicated whether they were alone or with others; if with others, they indicated who they were with (roommate or friend, romantic partner, parents, other family, and/or other). If romantic partner was endorsed, partner presence was coded as 1; if alone or not with a partner, this was coded as 0.

Substance Use

In the EMA reports, for each of the three non-prescription substances (alcohol, marijuana, nicotine) participants were asked about their use in the previous 15 minutes (Yes/No). If participants endorsed marijuana or nicotine use, their responses to these items were coded as 1; otherwise they were coded as 0. If participants endorsed nicotine use, they were asked what form (cigarette, chewing tobacco, and/or e-cigarette). If alcohol was endorsed, participants were asked the number of drinks they had consumed. If women reported 4 or more drinks and if men reported 5 or more drinks (occasion-based definition; National Institute on Alcohol Abuse and Alcoholism, 2019), binge drinking was coded as 1; otherwise, it was coded as 0. In the follow-up reports, if any misuse behavior was endorsed, the indicator for misuse was coded as 1; if misuse was not endorsed or a follow-up was not sent, misuse was coded as 0. Given our goal of assessing prescription misuse broadly, we did not differentiate between the types of medications.

Childhood Family Adversity

During the first visit, the Risky Families Scale (Taylor et al., 2004) assessed both nurturing and abusive aspects of the family environment during childhood. With a response scale ranging from 1 (rarely or none of the time) to 4 (most or all of the time), 11 items assessed how frequently participants experienced certain aspects, such as feeling loved and cared for; verbal abuse; and living with a substance abuser. Items were reverse-scored and summed so higher scores indicated greater childhood family adversity. Reliability in the current sample was Cronbach’s α = .83.

Analyses

Due to the nested nature of the data and to take full advantage of the nuanced momentary-level data (Hoffman & Stawski, 2009), 2-level generalized multilevel models (GMLM) were used. Specifically, due to the binary nature of the outcome variables, a Bernoulli distribution via logit link function was used to predict the likelihood of substance use occurring (Raudenbush & Bryk, 2002). The HLM 7 program, along with robust standard errors (Raudenbush et al., 2010) and full maximum likelihood estimation via adaptive quadrature with 10 points (due to low proportional odds of the outcome variables; Bauer & Sterba, 2011), were used. Four sets of models were run, each with one model for each substance type as an outcome: binge drinking, marijuana use, nicotine use, and prescription misuse. The moment-level substance use behavior was entered as a level-1 repeated measures outcome variable. Participant gender and childhood family adversity scores were entered as level-2 controls in each multilevel model to account for person-level variance in substance use explained by these variables. Random intercept models assessed whether current relationship status predicted substance use in daily life (RQ1), with relationship status entered as a level-2 intercept moderator. Random intercept models also assessed whether partner support predicted substance use in daily life (RQ2) among partnered participants, with partner support entered as an intercept moderator and relationship length and childhood family adversity entered as controls. Following standard, recommended practice, these continuous level-2 variables were grand-mean centered to reduce multicollinearity and provide meaningful intercepts (Raudenbush & Bryk, 2002). The intercept moderators are able to predict the average rate of substance use over the entire reporting period while still accounting for the within-person interdependence of the momentary data (Hoffman & Stawski, 2009).

Random slope models assessed whether real-time partner presence predicted the likelihood of substance use in the moment for partnered participants (RQ3a), with the variable for partner presence entered as a level-1 time-varying predictor. Lastly, to assess whether partner support moderated this association (RQ3b), this variable was added as a slope moderator for partner presence (cross-level interaction). To avoid conflating within- and between-person effects (Hoffman & Stawski, 2009), the partner presence variable was group-mean centered and the participants’ average score for partner presence (i.e., the ratio of reports with partner to all reports) was included in level 2. Additionally, because being with a partner and using a substance were both expected to be more likely in evening or later hours, these models also controlled for the time the report was submitted by using a dichotomous time-varying, non-random variable for Evening (0 = between 4am-4pm; 1 = between 4pm-4am). Partner support was also included as an intercept control. Continuous variables were again grand-mean centered. Although previous work considering similar substance use outcomes has not typically controlled for conducting multiple tests (Fleming et al., 2010), we established a more conservative approach by decreasing the standard significance level of .05 by half (alpha = .025).

Results

Report completion statistics and substance use instances are presented in Table 1; descriptive statistics and correlations in Table 2. There were no group differences in number of reports submitted by participants across relationship status (t(294)=−0.134, p=.894). During the EMA reporting period, the vast majority of the sample (over 95%) reported using at least one substance. Most participants (over 80%) reported at least one instance of alcohol use, with an average of about 4 instances being captured per participant using this method. About a third (34.3%) of the participants reported at least one instance of binge drinking. The majority of the sample also reported at least one instance of marijuana (58.8%) or nicotine use (60.1%). The most common form of nicotine consumption was e-cigarettes (94% of nicotine reports). About a third of the sample (35.5%) reported at least one instance of prescription misuse.

Table 1.

Completion Statistics and Instances of Substance Use Reported in Daily life, by Participant Relationship Status.

Single (n = 189)
 Partnered (n = 107)
M (SD) Range M (SD) Range
Total reports per participant 78.4 (25.1) 12–121 78.9 (27.8) 10–129
Number of reports per day 2.87 (1.0) 1–7 2.95(1.1) 1–8
N (%) N (%)
Total reports 14.823 (100.0) 8437 (100.0)
Reports of use
 Binge drinking 149 (1.0) 79 (0.9)
 Marijuana 1131 (7.5) 421 (4.9)
 Nicotine 2356 (15.8) 907 (10.8)
 Prescription misuse 245 (1.6) 81 (0.9)
Open in a new tab

Table 2.

Descriptive statistics and correlations for person-level variables.

Variable Descriptive Statistics
 Correlations
N M (SD) Range 1 2 3 4 5 6 7
1. Partner supporta 107 3.72 (0.4) 2–4
2. Relationship length (months)a,b 107 12.83 (13.9) 1–103b .27
3. Childhood family adversity 296 16.44 (4.8) 11–36 −.20 −.02
Total reported instances of:
4. Binge drinking 296 0.77 (1.7) 0–13 −.19 −.07 .00
5. Marijuana use 296 5.21 (9.8) 0–87 .02 −.10 .11 .18
6. Nicotine use 296 11.02 (16.7) 0–75 .18 .01 −.01 .17 .36
7. Prescription misuse 296 1.09 (2.4) 0–17 .10 −.05 .05 .06 .15 .21
8. Partner presencea 107 10.22 (10.3) 0–48 .21 .01 .10 .18 .16 .36 .12
Open in a new tab

Note. N = 296 participants. Bold font indicates significant correlation at p ˂ .05.

a

From partnered participants only.

b

Quartile segments: 4, 9, and 18 months.

RQ1: Results (Table 3) indicated that current relationship status did not predict the overall likelihood of substance use for binge drinking. Partnered participants reported 43% lower average likelihood of marijuana use, 38% lower average likelihood of nicotine use, and 39% lower average likelihood of prescription misuse in daily life compared to single participants.

Table 3.

Adjusted odds ratios from generalized multilevel models: Current relationship status predicting substance use in daily life.

Fixed Effects Binge drinking
 Marijuana use
 Nicotine use
 Prescription misuse
AOR 95% C.I. p AOR 95% C.I. p AOR 95% C.I. p AOR 95% C.I. p
Intercept 0.01 0.01–0.02 ˂.001 0.13 0.09–0.19 ˂.001 0.28 0.19–0.41 ˂.001 0.01 0.01–0.02 ˂.001
 Gendera 0.99 0.64–1.54 .964 0.53 0.35–0.82 .004 0.68 0.44–1.04 .076 1.67 1.16–2.41 .006
 Family adversity 1.01 0.97–1.04 .779 1.05 1.02–1.09 .002 1.01 0.97–1.05 .687 1.02 0.99–1.06 .223
 Current relationshipb 0.98 0.59–1.61 .931 0.57 0.39–0.83 .003 0.62 0.40–0.94 .026 0.61 0.42–0.88 .009
Open in a new tab

Note. AOR = Adjusted odds ratios. N = 296 participants.

a

Gender coded as 0 = men, 1 = women.

b

Current relationship status coded as 0 = single, 1 = partnered.

RQ2: Results (Table 4) indicated that greater partner support (one-unit increases) corresponded to a 69% decrease in the likelihood of binge drinking, along with an approximate 3-fold increase in the average likelihood of nicotine use and prescription misuse. Partner support was not related to marijuana use.

Table 4.

Adjusted odds ratios from generalized multilevel models: Partner support predicting substance use in daily life.

Fixed Effects Binge drinking
 Marijuana use
 Nicotine use
 Prescription misuse
AOR 95% C.I. p AOR 95% C.I. p AOR 95% C.I. p AOR 95% C.I. p
Intercept 0.01 0.00–0.01 <.001 0.06 0.03–0.12 <.001 0.13 0.08–0.20 <.001 0.01 0.01–0.01 <.001
 Gendera 1.34 0.58–3.14 .490 0.65 0.31–1.36 .249 0.97 0.52–1.84 .935 2.01 1.22–3.30 .007
 Family adversity 1.02 0.95–1.09 .653 1.06 1.00–1.12 .064 1.02 0.97–1.08 .485 0.96 0.92–1.00 .046
 Relationship lengthb 1.00 0.97–1.02 .703 0.98 0.97–0.99 .004 0.99 0.97–1.01 .344 0.97 0.96–0.99 .002
 Partner support 0.32 0.19–0.54 <.001 1.32 0.61–2.85 .484 2.95 1.28–6.80 .011 2.93 1.58–5.46 <.001
Open in a new tab

Note. AOR = Adjusted odds ratios. C.I. = Confidence interval. N = 107 partnered participants.

a

Gender coded as 0 = men, 1 = women.

b

Months.

RQ3: Results (Table 5) indicated that participants were nearly 3 times more likely to engage in binge drinking and 1.4 times more likely to engage in marijuana use when with a partner versus not with a partner. When with a partner, prescription misuse was 55% less likely. Partner presence was not significantly associated with nicotine use.

Table 5.

Adjusted odds ratios from generalized multilevel models: Partner presence predicting moment-level substance use in daily life.

Fixed Effects Binge drinking
 Marijuana use
 Nicotine use
 Prescription misuse
AOR 95% C.I. p AOR 95% C.I. p AOR 95% C.I. P AOR 95% C.I. p
Intercept 0.00 0.002–0.004 <0.001 0.05 0.036–0.077 <0.001 0.14 0.108–0.177 <0.001 002 0.021–0.028 <0.001
 Gendera 1.32 0.92–1.90 0.128 0.91 0.623–1-32 0.610 0.79 0.55–1.15 0.224 1.36 1.11–1.66 0.003
 Family adversity 0.98 0.96–1.01 0.127 1.04 1.01–1.08 0.011 0.987 0.96–1.02 0.409 0.956 094–0.97 <0.001
 Relationship lengthb 0.999 0.991–1.007 0.812 0.992 0.985–0.999 0.020 0.987 0.975–0.998 0.024 0.986 0.981–0.992 <0.001
 Partner presence (avg)c 21.54 3.05–151.90 0.002 1.44 0.45–4.63 0.539 7.48 1.74–32.09 0.007 4.03 1.37–11.81 0.012
 Partner support 0.25 0.18–0.36 <0.001 1.09 0.70–1.69 0.712 1.33 0.83–2.15 0.231 1.56 1.12–2.17 0.008
Evening controld 4.95 4.22–5.80 <0.001 2.06 1.82–2.32 <0.001 1.60 1.45–1.76 <0.001 1.11 0.99–1.26 0.077
Partner Presence 2.84 2.33–3.45 <0.001 1.24 1.06–1.44 0.007 1.12 0.99–1.27 0.084 0.46 0.38–0.57 ˂0.001
Open in a new tab

Note. AOR = Adjusted odds ratios. C.I. = Confidence interval. N = 107 partnered participants.

a

Gender coded as 0 = men, 1 = women.

b

Months.

c

Between-person average.

d

Control variable for evening hours: evening = 4pm-4am.

Results from the moderation models indicated that partner support did not significantly moderate the association between real-time partner presence and the likelihood of binge drinking (p = .618), marijuana use (p = .470), or prescription misuse (p = .350). However, for nicotine use, the effect of partner presence was stronger when partner support was higher. Specifically, when partner support was average (i.e., zero, due to centering), being with a partner did not significantly predict the likelihood of nicotine use (AOR for partner presence slope = 1.03, p = .616). However, greater partner support (one-unit increases) corresponded to a higher likelihood of nicotine use when with a partner, whereas lower support corresponded to a lower likelihood of nicotine use when with a partner (AOR = 1.99, p < .001).

The models from RQ3 also indicated several person-level relationship factors that were included as intercept controls were significant predictors of substance use. Relationship length was associated with marijuana use, nicotine use, and prescription misuse in that for every one-month increase in relationship length above the mean, the average likelihood of use for each of these substances decreased by 1–2%; relationship length was not significantly associated with binge drinking. These results and the main results remained unchanged with the small number of extreme relationship length outliers (<2%) removed. The between-person average for partner presence was associated with greater average likelihoods of reporting binge drinking, nicotine use, and prescription misuse, but was not associated with marijuana use across the reporting period. As an intercept control, increases in partner support corresponded to a 74% decrease in the average likelihood of binge drinking and a 56% increase in the average likelihood of prescription misuse across the reporting period; the partner support intercept control was not associated with marijuana or nicotine use in these models.

Discussion

The goal of this study was to explore the role of romantic relationship contexts in substance use behaviors throughout the daily lives of young adult U.S. college students at increased risk of substance use. The results further highlight the importance of these young adults’ interpersonal contexts in substance use behaviors and their varied, intricate associations across different substances. Notably, the person-level associations between relationship factors and substance use remained while accounting for the potential confound of childhood family adversity.

As expected, the sample reported relatively higher levels of substance use than would be expected from population norms within the U.S. (Schulenberg et al., 2019). In line with previous literature indicating lower overall substance use and substance problems among partnered versus single individuals (Braithwaite et al., 2010; Fleming et al., 2010; Rhule-Louie & McMahon, 2007; Simon & Barrett, 2010), partnered participants reported less frequent marijuana use, nicotine use, and prescription misuse in their daily lives compared to single participants, and this was found in generalized multilevel models that accounted for gender, childhood family adversity, and the interdependent data. The contrast to previous studies from young adults suggesting tobacco use did not differ between partnered and single young adults (Braithwaite et al., 2010; Fleming et al., 2010) could be explained by our sample, which was more likely to use substances in general, or our inclusion of e-cigarettes, which may not limit users’ partner pool to only those who can tolerate the smell of cigarette smoke; rates of e-cigarette use may be more sensitive to other relationship contexts. While speculative, understanding how secondary social factors such as odors or public use restrictions may relate to use and romantic partnering of nicotine users could further elucidate these connections.

Relationship status was unrelated to binge drinking behaviors, which stands in contrast to previous findings indicating that heavy alcohol use and related substance problems are less frequent among partnered versus single young adults (Braithwaite et al., 2010; Fleming et al., 2010; Simon & Barrett, 2010). Because binge drinking more often occurs late at night and therefore after the EMA signals, instances of binge drinking were likely underreported using this method. Alternatively, partnered and single substance users may binge drink about the same amount (Salvatore et al., 2014). The low prevalence of binge drinking reports could have resulted in low statistical power for detecting effects, however, the reliable effects detected in other results involving binge drinking suggest this was not an issue. When collecting EMA data multiple times each day, binge drinking is a relatively uncommon behavior (Schulenberg et al., 2019); future researchers exploring multiple types of substance use behaviors in daily life (versus retrospectively) may consider employing different methods of data collection suited to each substance’s typical use. Indeed, such research on multiple or poly-substance use is needed given that the use of one substance often coincides with the use of other substances (e.g., Martin, 2008). While investigating the copious potential forms of polysubstance use across these four substances was well beyond the scope of the current study, this highly complex and often overlooked issue has been recognized as an important area for future research (UNODC, 2018).

For individuals in relationships, we tested whether partner support—an essential component of romantic relationship quality (Walen & Lachman, 2000)—was related to substance use behaviors in this sample. Higher support was related to lower binge drinking and marijuana use, as expected, yet related to higher nicotine use and higher prescription misuse. Findings echo recent studies suggesting complex associations between relationship quality and substance use behaviors. For instance, substances such as alcohol may decrease maladaptive problem-solving behaviors during conflict interactions for those in distressed relationships (Fairbairn & Testa, 2017), suggesting an explanation for why use may be higher among lower-quality relationships. Alcohol use with a romantic partner can also increase feelings of intimacy (Levitt & Cooper, 2010; Testa et al., 2019), indicating that the use of some substances may increase relationship quality via shared activities, for example, or could artificially increase perceptions of such quality. Taken together, findings suggest that high-quality relationships alone are not protective against substance use behaviors. While positive and supportive social relationships such as romantic partnerships are critical for supporting healthy behaviors (Johnson & Acabchuk, 2018), other individual, relational, and societal factors also play important roles. One relational variable likely relevant is the partner’s substance use, which has consistently been associated with concurrent and subsequent substance use among adolescents and young adults (Derrick et al., 2019; Rhule-Louie & McMahon, 2007). While assessing partner’s substance use was beyond the scope of the current individual-based study, we were able to capture another relational component that has not been directly tested before: partner presence during moments of substance use.

Binge drinking and marijuana use were more likely in moments when partnered participants were with their romantic partners, and this was found when accounting for evening-time hours (and other controls). Again, data was likely not representative of all binge drinking instances, but this was a robust finding. For the instances of binge drinking and marijuana use captured via the EMA, they were more likely to occur when a romantic partner was present. For nicotine use, partner presence became a significant predictor when partner support was entered as a moderator. These results reiterate the notion that substance use is often a social behavior (Quintero et al., 2006) and that couples often use substances together (e.g., Rhule-Louie et al., 2007). Conversely, partner presence did not significantly predict prescription misuse. Prescription drugs are often misused for the therapeutic purposes for which they were intended, such as for treating pain or sleep disorders or enhancing focus for academic study (e.g., McLarnon et al., 2012). Assessing motivations behind substance use was beyond the scope of the current study, but the negative association between partner presence and prescription misuse could be explained by the more practical (i.e., less recreational) nature of this type of substance use. When substances are used for more practical purposes, such as self-medicating or academic enhancement, the social context of a relationship may be less important in that moment. Some additional support for this explanation is offered by the evening-time control variable, which indicated that binge drinking, marijuana use, and nicotine use were more likely to occur in the evening or late night hours (suggesting a recreational purpose), yet the time of day did not predict the likelihood of prescription misuse. This corroborates other studies indicating that social contexts differ across substance type, with alcohol and marijuana being more socially oriented and prescription misuse more solitary (McCabe et al., 2014), and further indicates associations with relationship contexts are likely not equivalent across substances. Future research on the differences in social use of substances, and therefore how susceptible their use may be to romantic relationship contexts or to the drug-related attitudes of the partners, could prove useful for intervention efforts.

The association between partner presence and substance use was not significantly moderated by partner support for any substance use behaviors except for nicotine use. Specifically, the more supportive the relationship, the more likely nicotine use became when with a partner. For less supportive relationships, being with a partner reduced the likelihood of nicotine use. This finding from momentary experiences of substance use mirrors retrospective studies of couple substance use indicating that when substances are used together, relationship quality is higher (or vise-versa; Homish & Leonard, 2007; Testa et al., 2018). The findings also speak to the importance of how partners connect and what substances might be bringing to their relationships, such as feelings of intimacy or improved relationship functioning (Fairbairn & Testa, 2017; Levitt & Cooper, 2010; Testa et al., 2019). It is important to note that the instances of substance use among partnered participants were even lower than expected (especially binge drinking and prescription misuse; nicotine was most frequent) and could have resulted in these exploratory moderation analyses being underpowered; nevertheless, we present the findings here as a starting point for future studies.

Given the purpose of the larger project was to collect real-time predictors of prescription drug misuse, the recruitment strategies highlighted the health-based focus of the study and did not mention romantic relationships. This may have resulted in a wider array of romantic relationships regarding levels of support than are typically found in relationship-focused studies. This could also be a reason why the Partner Strain items exhibited poor reliability—psychometric properties previously documented for this scale were largely based on relationship-focused studies, which may not have included relationships exhibiting higher or more variable levels of strain. While potentially allowing for a more complete picture of how relationships are linked to substance use, it also points to the need for establishing scale reliability more broadly.

Limitations

The design and sample were optimized to capture the everyday contexts of prescription misuse behaviors in daily life. While ideal for the intended goal, this likely resulted in less representative samples of other substance use behaviors, limited generalizability to other young adult college students or to those within different geographical and cultural spaces, and unknown differences in results between prescription misuse and the other substance use behaviors. The design was correlational and findings should not be interpreted as causation; indeed, substance use may influence relationship outcomes, as use and relationships are bidirectionally interconnected (Derrick et al., 2019). Further, assessing partner characteristics, such as substance use, relationship quality, or gender, was beyond the scope of the current large, multifaceted study, yet represent important next steps as substance use within romantic relationships is highly interdependent and likely influenced by the dyadic nature of relationship quality (Derrick et al., 2019; Rhule-Louie & McMahon, 2007), and that substance use disorders are disproportionately high among sexual minority subgroups (Krueger et al., 2020). The study design also prevented identifying changes in partner status after the first lab visit, whether participants were dating each other, or whether reports were event- versus time-based, which may indicate important theoretical or statistical nuances, and are important factors to consider in future research.

Standard measures of socioeconomic status (SES) were not collected, precluding comparisons across SES. The full-time undergraduate standing at a large state university that is generally attended by students from higher SES backgrounds does indicate a high sample-average SES. While this is typical for young adult U.S. college students (the population of interest), it does represent a limitation. Specifically, the well-documented systemic inequities that introduce barriers to higher education among historically disadvantaged groups (Tierney & Hagedorn, 2002), such as those with low SES, leaves these groups underrepresented, therefore deterring a full understanding of how substance use and romantic partnerships are linked. Given that such underrepresented groups report relatively lower academic success (Tierney & Hagedorn, 2002), and that substance use is linked to academic outcomes (UNODC, 2018), understanding these associations among students from underrepresented groups could provide critical information for those seeking to improve academic and health outcomes for these students.

Lastly, it is important to note that a clear line between substance use that is problematic versus harmless or even medicinal is often difficult to discern, and this study did not differentiate between problematic and more casual substance use behaviors. Nevertheless, documenting what factors are associated with more or less substance use in daily life is an important step in preventing problematic substance use in this at-risk population of college students.

Conclusions

Our findings suggest that, for young adult college students who use substances, substance use behaviors in daily life are strongly linked to romantic relationship factors. While being in a romantic relationship may make substance use less frequent overall, some (but not all) substance use behaviors are more likely when romantic partners are together. High-quality relationships are not universally protective against substance use; problematic substance use can occur even in highly supportive relationships. Results suggest that, depending on the partners’ social contexts and substance use characteristics, health behavior interventions designed to strengthen social support and intimate bonds among romantic partners may inadvertently provide avenues for increased substance use; given the established value of social support in health behaviors, more research is needed to confirm these potential nuances. The varying relationship contexts that may make some substance use behaviors more or less likely—both between and within individuals—as documented here represent fruitful paths for further in-depth research and effective prevention efforts.

Highlights.

  • Single participants used substances more frequently than dating participants.

  • Some substance use more likely when romantic partners are together at particular moments.

  • Momentary reports offer unique insights into relationship contexts of substance use.

  • High-quality relationships are not universally protective against substance use.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References

  1. Abraham J (2010). Pharmaceuticalization of society in context: Theoretical, empirical and health dimensions. Sociology, 44, 603–622. 10.1177/0038038510369368 [DOI] [Google Scholar]
  2. Arnett JJ (2000). Emerging adulthood: A theory of development from the late teens through the twenties. The American Psychologist, 55, 469–480. 10.1037/0003-066X.55.5.469 [DOI] [PubMed] [Google Scholar]
  3. Bauer D, & Sterba S. (2011). Fitting multilevel models with ordinal outcomes: Performance of alternative specifications and methods of estimation. Psychological Methods, 16(4), 373–390. 10.1037/a0025813 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Bell SE, & Figert AE (2012). Medicalization and pharmaceuticalization at the intersections: Looking backward, sideways and forward. Social Science and Medicine, 75, 775–783. 10.1016/j.socscimed.2012.04.002 [DOI] [PubMed] [Google Scholar]
  5. Braithwaite SR, Delevi R, & Fincham FD (2010). Romantic relationships and the physical and mental health of college students. Personal Relationships, 17, 1–12. 10.1111/j.1475-6811.2010.01248.x [DOI] [Google Scholar]
  6. Connolly J, McIsaac C, Shulman S, Wincentak K, Joly L, Heifetz M, & Bravo V (2014). Development of romantic relationships in adolescence and emerging adulthood: Implications for community mental health. Canadian Journal of Community Mental Health, 33(1), 7–19. 10.7870/cjcmh-2014-002 [DOI] [Google Scholar]
  7. Crane CA, Testa M, Schlauch RC, & Leonard KE (2016). The couple that smokes together: Dyadic marijuana use and relationship functioning during conflict. Psychology of Addictive Behaviors, 30(6), 686–693. 10.1037/adb0000198 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Derrick JL, Wittkower LD, & Pierce JD (2019). Committed relationships and substance use: Recent findings and future directions. Current Opinion in Psychology, 30, 74–79. 10.1016/j.copsyc.2019.03.002 [DOI] [PubMed] [Google Scholar]
  9. Fairbairn CE, & Testa M (2017). Relationship quality and alcohol-related social reinforcement during couples interaction. Clinical Psychological Science, 5(1), 74–84. 10.1177/2167702616649365 [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Fleming CB, White HR, & Catalano RF (2010). Romantic relationships and substance use in early adulthood: An examination of the influences of relationship type, partner substance use, and relationship quality. Journal of Health and Social Behavior, 51, 153–167. 10.1177/0022146510368930 [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Hoffman L, & Stawski RS (2009). Persons as contexts: Evaluating between-person and within-person effects in longitudinal analysis. Research in Human Development, 6(2–3), 97–120. 10.1080/15427600902911189 [DOI] [Google Scholar]
  12. Homish GG, & Leonard KE (2005). Marital quality and congruent drinking. Journal of Studies on Alcohol, 66(4), 488–496. 10.15288/jsa.2005.66.488 [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Homish GG, & Leonard KE (2007). The drinking partnership and marital satisfaction: The longitudinal influence of discrepant drinking. Journal of Consulting and Clinical Psychology, 75, 43–51. 10.1037/0022-006X.75.1.43 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Homish GG, Leonard KE, & Cornelius JR (2010). Individual, partner and relationship factors associated with non-medical use of prescription drugs. Addiction, 105(8), 1457–1465. 10.1111/j.1360-0443.2010.02986.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Hussar B, Zhang J, Hein S, Wang K, Roberts A, Cui J, Smith M, Bullock Mann F, Barmer A, & Dilig R (2020). The Condition of Education 2020 (NCES 2020–144).
  16. Johnson BT, & Acabchuk RL (2018). What are the keys to a longer, happier life? Answers from five decades of health psychology research. Social Science and Medicine, 196, 218–226. 10.1016/j.socscimed.2017.11.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Krueger EA, Fish JN, & Upchurch DM (2020). Sexual orientation disparities in substance use: Investigating social stress mechanisms in a national sample. American Journal of Preventive Medicine, 58(1), 59–68. 10.1016/j.amepre.2019.08.034 [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Leonard KE, & Homish GG (2008). Predictors of heavy drinking and drinking problems over the first 4 years of marriage. Psychology of Addictive Behaviors, 22, 25–35. 10.1037/0893-164X.22.1.25. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Levitt A, & Cooper ML (2010). Daily alcohol use and romantic relationship functioning: Evidence of bidirectional, gender-, and context-specific effects. Personality and Social Psychology Bulletin, 36, 1706–1722. 10.1177/0146167210388420 [DOI] [PubMed] [Google Scholar]
  20. Martin CS (2008). Timing of alcohol and other drug use. Alcohol Research and Health, 31(2), 96–99. [PMC free article] [PubMed] [Google Scholar]
  21. McCabe SE, West BT, Veliz P, Frank KA, & Boyd CJ (2014). Social contexts of substance use among U.S. high school seniors: A multicohort national study. Journal of Adolescent Health, 55(6), 842–844. 10.1016/j.jadohealth.2014.06.017 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. McLarnon ME, Barrett SP, Monaghan TL, & Stewart SH (2012). Prescription drug misuse across the lifespan: A developmental perspective. In Verster JC, Brady K, Galanter M, & Conrod P (Eds.), Drug Abuse and Addiction in Medical Illness: Causes, Consequences and Treatment (pp. 213–230). 10.1007/978-1-4614-3375-0 [DOI]
  23. Meyler D, Stimpson JP, & Peek MK (2007). Health concordance within couples: A systematic review. Social Science and Medicine, 64, 2297–2310. 10.1016/j.socscimed.2007.02.007 [DOI] [PubMed] [Google Scholar]
  24. National Institute on Alcohol Abuse and Alcoholism. (2019). Drinking Levels Defined. https://www.niaaa.nih.gov/alcohol-health/overview-alcohol-consumption/moderate-binge-drinking
  25. National Institute on Drug Abuse. (2019). Drug and Alcohol Use in College-Age Adults in 2018. https://www.drugabuse.gov/related-topics/trends-statistics/infographics/drug-alcohol-use-in-college-age-adults-in-2018
  26. Papp LM, & Blumenstock SM (2016). Momentary affect and risky behavior correlates of prescription drug misuse among young adult dating couples: An experience sampling study. Addictive Behaviors, 53, 161–167. 10.1016/j.addbeh.2015.10.016 [DOI] [PubMed] [Google Scholar]
  27. Quintero G, Peterson J, & Young B (2006). An exploratory study of socio-cultural factors contributing to prescription drug misuse among college students. Journal of Drug Issues, 36, 903–932. [Google Scholar]
  28. Raudenbush SW, & Bryk AS (2002). Hierarchical linear models: Applications and data analysis methods (2nd ed.). Sage Publications, Inc. https://doi.org/0750 [Google Scholar]
  29. Raudenbush SW, Bryk AS, & Congdon RT (2010). Hierarchical linear modeling with the HLM7 programs. Scientific Software International. [Google Scholar]
  30. Repetti RL, Taylor SE, & Seeman TE (2002). Risky families: Family social environments and the mental and physical health of offspring. Psychological Bulletin, 128(2), 330–366. 10.1037/0033-2909.128.2.330 [DOI] [PubMed] [Google Scholar]
  31. Rhule-Louie DM, & McMahon RJ (2007). Problem behavior and romantic relationships: Assortative mating, behavior contagion, and desistance. Clinical Child and Family Psychology Review, 10(1), 53–100. 10.1007/s10567-006-0016-y [DOI] [PubMed] [Google Scholar]
  32. Roberson PNE, Shorter RL, Woods S, & Priest J (2018). How health behaviors link romantic relationship dysfunction and physical health across 20 years for middle-aged and older adults. Social Science and Medicine, 201, 18–26. 10.1016/j.socscimed.2018.01.037 [DOI] [PubMed] [Google Scholar]
  33. Salvatore JE, Kendler KS, & Dick DM (2014). Romantic relationship status and alcohol use and problems across the first year of college. Journal of Studies on Alcohol and Drugs, 75(4), 580–589. 10.15288/jsad.2014.75.580 [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Schulenberg JE, Johnston LD, O’Malley PM, Bachman JG, Miech RA, & Patrick ME (2019). Monitoring the Future national survey results on drug use, 1975–2018: Volume II, College students and adults ages 19–60. Institute for Social Research, The University of Michigan. [Google Scholar]
  35. Schulenberg John E., & Maggs JL (2002). A developmental perspective on alcohol use and heavy drinking during adolescence and the transition to young adulthood. Journal of Studies on Alcohol, 63(SUPPL. 14), 54–70. 10.15288/jsas.2002.s14.54 [DOI] [PubMed] [Google Scholar]
  36. Shiffman S (2009). Ecological Momentary Assessment (EMA) in studies of substance use. Psychological Assessment, 21(4), 486–497. 10.1037/a0017074 [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Shiffman S, Stone AA, & Hufford MR (2008). Ecological momentary assessment. Annual Review of Clinical Psychology, 4, 1–32. 10.1146/annurev.clinpsy.3.022806.091415 [DOI] [PubMed] [Google Scholar]
  38. Simon RW, & Barrett AE (2010). Nonmarital romantic relationships and mental health in early adulthood: Does the association differ for women and men? Journal of Health and Social Behavior, 51(2), 168–182. 10.1177/0022146510372343 [DOI] [PubMed] [Google Scholar]
  39. Substance Abuse and Mental Health Services Administration. (2019). Key Substance Use and Mental Health Indicators in the US: Results from the 2018 National Survey on Drug Use and Health. Center for Behavioral Health Statistics and Quality, Substance Abuse and Mental Health Services Administration. [Google Scholar]
  40. Taylor SE, Lerner JS, Sage RM, Lehman BJ, & Seeman TE (2004). Early environment, emotions, responses to stress, and health. Journal of Personality, 72(6), 1365–1393. 10.1111/j.1467-6494.2004.00300.x [DOI] [PubMed] [Google Scholar]
  41. Testa M, Wang W, Derrick JL, & Leonard KE (2018). Marijuana use by intimate partners: Does discrepant use impair relationship functioning? Psychology of Addictive Behaviors, 32, 475–484. 10.1037/adb0000357 [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Testa M, Wang W, Derrick JL, & Leonard KE (2019). Does drinking together promote relationship intimacy? Temporal effects of daily drinking events. Journal of Studies on Alcohol and Drugs, 80(5), 537–545. 10.15288/jsad.2019.80.537 [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Tierney WG, & Hagedorn LS (2002). Increasing Access to College: Extending Possibilities for All Students. State University of New York Press. [Google Scholar]
  44. U.S. Department of Education. (2017). 2015–16 National Postsecondary Student Aid Study (NPSAS:16).
  45. Umberson D (1992). Gender, marital status and the social control of health behavior. Social Science and Medicine, 34(8), 907–917. 10.1016/0277-9536(92)90259-S [DOI] [PubMed] [Google Scholar]
  46. United Nations Office on Drugs and Crime [UNODC]. (2018). World Drug Report 2018. United Nations. 10.1080/00909887909365203 [DOI] [Google Scholar]
  47. UNODC. (2019). World Drug Report 2019. United Nations. 10.18356/bdc264f4-en [DOI] [Google Scholar]
  48. Walen HR, & Lachman ME (2000). Social support and strain from partner, family, and friends: Costs and benefits for men and women in adulthood. Journal of Social and Personal Relationships, 17, 5–30. 10.1177/0265407500171001 [DOI] [Google Scholar]
  49. World Health Organization. (2019). Tobacco. https://www.who.int/news-room/fact-sheets/detail/tobacco

RESOURCES