The Moderating Role of Positive Affect in Understanding the Relationship Between Stress and Alcohol Use

Binge drinking among college students presents a significant public health concern given the frequency and the associated negative health and social consequences. There are a variety of reasons why college students decide to binge drink (and to drink more broadly), including drinking to cope with negative emotional experiences, including when stressed. One factor that may be playing a role in the relationships among stress, drinking to cope, and binge drinking is positive emotion. Positive emotion has been found to be linked to the ways in which individuals appraise stressful situations, with those higher in positive emotion being more likely to appraise stressful situations as moments in which they can gain some mastery rather than obstacles that will be difficult to overcome. Further, individuals who endorse drinking to cope have been found to be lower in positive emotions. These positive emotions may be serving as a buffer, limiting experiences of stress or changing the way stress is perceived, and therefore making people less likely to use alcohol to cope with stress. To explore these associations, college students from the University of Rhode Island completed an online survey assessing their self-rated levels of stress, intensity and frequency of experiencing positive emotions, and frequency of binge drinking. Approximately three-quarters of the sample reported having consumed alcohol during the past month and of those, nearly two-thirds reported at least one binge drinking episode during the past month. Stress and positive emotions did not appear to be significantly related to frequency of binge drinking, but were significantly correlated with coping motives for drinking. Regression analyses with all predictors, age, and gender entered into the model revealed the same results: stress and positive emotion were not significantly related to binge drinking, but were significantly related to coping motives. The interaction between stress and positive emotions predicting frequency of binge drinking or coping motives was nonsignificant. Findings of the present study indicate that, among college students, stress may not be a contributing factor to the decision to binge drink. It may be that college students are engaging in other behaviors to manage stress, or that measures included in the current study do not fully capture college students’ experiences of stress or positive emotional experiences. Findings of the present study suggest the potential utility of targeting positive emotions in efforts to decrease coping-related alcohol use among those college students who are receiving treatment due to their alcohol consumption.

Alcohol consumption, specifically binge drinking, among college students is of significant public health concern. Binge drinking is operationally defined by NIAAA as five or more standard drinks in a two-hour period for males and four or more standard drinks in a two-hour period for females, the amount that would be required to raise an individual's blood alcohol content (BAC) to .08 if they were of average height and weight (Health & Services, 2015). The college years represent the time in life with the highest level of alcohol consumption and the highest prevalence of alcohol use disorders (Dawson, Goldstein, & Grant, 2007), with 60% of college students reporting past-month alcohol use and two-thirds of those reporting past-month binge drinking (SAMHSA, 2015).
Binge drinking exposes college students to significantly increased risk of numerous alcohol-related consequences, compared to students who do not binge drink (Jennison, 2004). Binge drinking in college increases the risk of consequences such as hangovers, being late to class (Perkins, 2002), blacking out (Merrill et al., 2016), poor academic performance (Singleton, 2007;Singleton & Wolfson, 2009), legal ramifications (Wechsler, Lee, Nelson, & Kuo, 2002), and sexual assaults (McCauley, Calhoun, & Gidycz, 2010;Testa & Livingston, 2009), with at least half of sexual assaults occurring among college students involving the perpetrator of the assault, the victim of the assault, or both being intoxicated (Abbey, 2002). Those who binge drink 2 are also at increased risk for health-related consequences, such as cancers, cirrhosis of liver, and cardiovascular disease (Rehm et al., 2009). Each year, approximately 599,000 college students are unintentionally injured and 1,800 are killed as a result of alcohol-related injuries, including motor vehicle accidents (Hingson, Zha, & Weitzman, 2009). Additionally, college students who do not binge drink are placed at risk for disrupted studies, sexual assault, and injury from their binge drinking peers (Perkins, 2002).

Alcohol Use and Stress
The college years tend to be a time of increased stress for young adults (Blanco et al., 2008), with college students reporting high levels of stress relating to academic performance in classes, pressure to be successful, and developing plans for after graduation (Beiter et al., 2015). Since the 1950s, the relationship between stress and alcohol consumption has been explained, in large part, by the tension-reduction hypothesis (Conger, 1951(Conger, , 1956. The tension-reduction hypothesis states that: a) under most circumstances, drinking alcohol will lead to a reduction in stress, and b) when stressed, people will be more motivated to drink alcohol. Much of the work that has explored this hypothesis has focused on the second prong, that in times of increased stress, people will be more likely to drink alcohol, possibly due to the popular belief that alcohol has stress-response-dampening (SRD) effects (Levenson, Sher, Grossman, Newman, & Newlin, 1980), though laboratory work focusing on both prongs has been notably mixed. For example, some work has found that alcohol increased objective measures of stress (for a review, see Greeley & Oei, 1999), while other work has found that, at sufficiently high doses (e.g., those which one might 3 consume when binge drinking), alcohol consumption has a direct dampening effect on physiological measures of stress (Donohue, Curtin, Patrick, & Lang, 2007), and that this effect is significantly more robust at higher levels of stress (Moberg, Weber, & Curtin, 2011). One study examining the effects of anticipatory stress on college student alcohol consumption found that those in a placebo condition (i.e., those anticipating completing a non-stressful activity) consumed significantly more alcohol than did those in the experimental condition (i.e., those anticipating giving a stressful speech). However, this effect was found to be moderated by past three month drinking, such that the effect was reversed for those reporting heavy drinking (Bernstein & Wood, 2017), indicating that those with higher baseline levels of alcohol consumption may be more motivated to drink alcohol to cope with stress. Another study examining the role of psychosocial stress on differences in alcohol consumption among undergraduate students found no difference in alcohol consumption between stress and no stress conditions (Larsen, Engels, Granic, & Huizink, 2013).
The idea that drinking motives may explain the relationship between stress and alcohol consumption is consistent with the tension-reduction hypothesis. Drinking motives refer to individuals' self-reported reasons for drinking alcohol. They have been found to be important predictors of quantity of alcohol consumption, maintenance of drinking behaviors and patterns, and likelihood of experiencing negative alcohol-related consequences among college students (Carey & Correia, 1997). Drinking motives have been found to fall into four categories: enhancement (i.e., drinking to induce, increase, or maintain positive affective states), social lubrication (i.e., drinking to enhance social situations and make them more enjoyable), 4 conformity (i.e., drinking because others are drinking, or to fit in), and coping (i.e., drinking to ameliorate negative affective states, including perceived stress) (Cooper, 1994;Read, Wood, Kahler, Maddock, & Palfai, 2003). Several studies have found that drinking to cope is a significant risk factor for alcohol-related problems and physiological dependence (Carey & Correia, 1997;Kassel, Jackson, & Unrod, 2000;MacLean & Lecci, 2000;Merrill, Wardell, & Read, 2014). Individuals reporting strong coping motives for drinking may lack other effective strategies for regulating negative affect and related states, such as stress (Cooper, Frone, Russell, & Mudar, 1995;Merrill & Thomas, 2013). However, findings regarding these relations among college students have been mixed. For instance, one prospective study found that, among college students in particular, endorsement of drinking coping motives was directly related to experiencing subsequent negative alcohol-related consequences (Merrill et al., 2014), and cross-sectional studies have found increased endorsement of coping motives to be related to increased alcohol consumption (Park & Levenson, 2002). On the other hand, some work has found this relationship to be less robust among college students, for whom heavy episodic drinking is more normative (Bradizza, Reifman, & Barnes, 1999;Perkins, 1999), thus suggesting that once drinking is established, there may be reasons other than coping for why individuals drink.

Positive Affect
There has been a relative dearth of research on the role of positive affective experiences in alcohol use. Much of the work in positive affect has been in the relapse/recovery literature (for a review, see Krentzman, 2013). Positive affect reflects 5 the degree to which a person feels enthusiastic, alert, and active (Watson, Clark, & Tellegen, 1988) and has been shown to be related to numerous positive outcomes including increased creativity (Isen, Daubman, & Nowicki, 1987) and intuition (Bolte, Goschke, & Kuhl, 2003), future positive self-rated health (Siahpush, Spittal, & Singh, 2008), and even increased lifespan (Danner, Snowdon, & Friesen, 2001). Frequent experiences of positive affect have also been shown to widen the scope of attention and expand our ability to identify potential behavioral options (Fredrickson & Branigan, 2005) and to potentially act as an antidote to stressful situations, such that individuals who are higher in positive affect bounce back from stress more quickly (Folkman, 1997(Folkman, , 2008. Further, it has been implicated in relation to the way individuals appraise potentially stressful situations, with those higher in positive affect being more likely to appraise stressful situations as challenges they are capable of handling (Folkman & Lazarus, 1985). Therefore, for these individuals, it is likely that they would not be as affected by stress and may see other alternative options to engaging in binge drinking to alleviate stress.

Purpose of the Study
In sum, the relationship between stress and alcohol use among college students has been widely investigated with mixed findings. Literature examining the role of coping motives for drinking in college student alcohol use has also shown mixed results. Further, some work has suggested that those who endorse coping motives for drinking tend to experience lower positive affect (Armeli, O'Hara, Ehrenberg, Sullivan, & Tennen, 2014). Little work, however, has examined the relations among stress, positive affect, alcohol use, and coping motives. It may be that when 6 individuals who experience frequent positive affect are faced with a stressor, they are able to generate lots of different ways to cope with the situation (based upon Fredrickson's Broaden and Build Theory (Fredrickson & Branigan, 2005) and therefore less likely to drink. These individuals should also report decreased coping motives for drinking. Therefore, the purpose of this study is to examine the following hypotheses: 1. Frequency of binge drinking will be positively associated with levels of perceived stress and negatively associated with levels of positive affect.
2. Endorsement of coping motives for drinking will be positively associated with levels of perceived stress and negatively associated with levels of positive affect.
3. Positive affect will moderate the relationship between stress and binge drinking.
a. Specifically, perceived stress will be associated with decreased frequency of binge drinking for those individuals who report increased intensity and frequency of positive emotions. 4. Positive affect will moderate the relationship between stress and coping motives.
a. Specifically, perceived stress will be negatively associated with endorsement of coping motives for those individuals who report increased intensity and frequency of positive emotions.

Sample.
Participants for the current study were a sample of 351 undergraduate students at the University of Rhode Island collected in February 2018 through online advertisements in classes. Two participants were removed from the sample for missing more than 30% item-level data on any primary variable of interest (see Measures).

Measures.
Stress was measured using the stress subscale of the Rhode Island Stress and Coping Inventory (RISCI-S) (Fava, Ruggiero, & Grimley, 1998), which includes 7 items measuring the degree to which situations in an individual's life are appraised as stressful. Participants rate each item in terms of how often each statement was true for them in the preceding month with five possible response options (1 = never, 5 = frequently), with higher scores indicating increased levels of stress in the preceding month. Cronbach's alpha in the current sample was good, α = 0.87.
Positive affective experience was measured using the positive affect subscale of the Positive and Negative Affect Schedule (PANAS-P) (Watson et al., 1988)  (1 = very rarely or never, 5 = very often or always) for a possible total score between 6 and 30, and with higher scores indicating more frequency positive affective experiences over the preceding month. Cronbach's alpha for the PANAS-P and SPANE-P in the current sample were good, α = 0.90 and 0.91, respectively. To capture overall positive affect experiences, both intensity and frequency, a product term was created by multiplying scores on the PANAS-P and SPANE-P.
Alcohol use was measured using the NIAAA recommended 6 Question Set (Questions, 2003), which includes questions about past month frequency of alcohol use, quantity of drinks consumed, and frequency of binge drinking.
Coping motives were measured using the Drinking Motives Questionnaire -Revised (DMQ-R; (Cooper, 1994) Coping Subscale, which assesses the frequency with which adolescents and young adults choose to drink alcohol for reasons related to coping. Participants respond to each item with five possible response options (1 = almost never/never, 5 = almost always/always). A summed score is created with possible scores ranging from 0 to 20 and higher scores indicating more frequent endorsement of specific drinking motives. Cronbach's alpha for the DMQ-R in the current sample was good, α = .88.

Participant demographics
were assessed with questions regarding participants' age, gender identity, race/ethnicity, and college enrollment status.

Procedure.
Participants for the current study were undergraduate students recruited from the University of Rhode Island via e-mail and on-line course sites. Those interested in participating utilized a link within the advertisement to access a Qualtrics survey, at which point they were asked to read a consent form and confirm their willingness to participate by selecting "yes" at the bottom of the form. Participants who provided consent to participate were asked to complete the online Qualtrics survey, which took approximately 30 minutes to compete, and participants did not receive any compensation for completing the survey. No identifying information was collected during the course of this study and all study procedures were approved by the University of Rhode Island Institutional Review Board.

Data Analysis Plan.
As recommended by Tabachnik and Fidell (2001), all variables of interest were assessed for adherence to assumptions of generalized linear models. Then, Pearson product-moment correlations were calculated between relevant study variables to explore their bivariate associations. For all further analyses, only those who reported having a drink of alcohol in preceding 30 days were selected. To address the question of whether perceived stress and positive affect are uniquely related to frequency of binge drinking after controlling for the effects of age and gender, a poisson regression model was used with age, gender, perceived stress, and positive affect as predictors and frequency of binge drinking as the outcome. Poisson regression models were selected given the count nature of the data (Coxe, West, & Aiken, 2009). Then, to examine whether perceived stress, positive affect, and/or their interaction were associated with frequency of binge drinking, moderation analyses were conducted using the PROCESS macro in SPSS (Hayes, 2012). Second, to address the question of whether perceived stress and positive affect are uniquely related to coping motives after controlling for the effects of age and gender, a multiple linear regression model was used with age, gender, perceived stress, and positive affect as predictors and endorsement of coping motives as the outcome. Then, to examine whether perceived stress, positive affect, and/or their interaction predict coping motives, moderation analyses were conducted using the PROCESS macro in SPSS (Hayes, 2012). In both moderation analyses, predictor variables were mean centered prior to construction of the interaction term to aid in interpretation of parameter estimates and to lessen the correlation between the interaction term and its component variables. The PROCESS procedures use ordinary least squares regression and bootstrapping methodology, 12 CHAPTER 3. Further, nearly all participants reported some degree of coping motives for drinking (i.e., 18.3% reported "never" drinking to cope), and approximately half of participants scored higher on the DMQ-R coping subscale compared to mean scores presented in previous studies (e.g., (Neighbors, Larimer, Markman Geisner, & Knee, 2004).

Primary Analyses.
Hypothesis 1. Examination of Pearson product-moment correlations revealed that perceived stress did not appear to be significantly related to frequency (r(256) = .09, p=.17) or quantity of alcohol consumption (r(254) = .04, p=.57), or to frequency of binge drinking (r(344) = .01, p=.88). Positive affect also did not appear to be significantly related to frequency (r(254) = -.09, p=.17) or quantity of alcohol consumption (r(252) = .04, p=.50), or to frequency of binge drinking (r(339) = .03, p=.55). Next, a poisson regression model was estimated to evaluate the unique associations of perceived stress and positive affect to frequency of binge drinking after controlling for age and gender. Neither perceived stress (B = .01, Wald's χ 2 = 2.07, p = .15), nor positive affect (B = .02, Wald's χ 2 = 3.66, p = .06) appeared to be significantly related to frequency of binge drinking in this model.

Hypothesis 2.
Pearson product-moment correlations revealed that there was a significant positive correlation between perceived stress and coping motives (r(255) = .28, p < .001). Positive affect was significantly negatively associated with coping motives (r(253) = -.28, p < .001). Next, a multiple linear regression model was estimated to evaluate the unique associations of perceived stress and positive affect to coping motives after controlling for age and sex. Higher scores on perceived stress were significantly associated with greater endorsement of coping motives (β = .16, t = 2.86, p = .005), while lower ratings of positive affect were significantly associated with greater endorsement of coping motives (β = -.20, t = -2.87, p = .005). 15 CHAPTER 4.

CONCLUSION
The goals of this study were twofold: 1) to examine the relations among perceived stress, positive emotions, frequency of binge drinking behavior and coping motives, and 2) to examine the potential moderating role of positive emotions on the relation between stress and frequency of binge drinking and endorsement of coping motives for drinking. We extended the tension-reduction hypothesis to explain not only actual drinking behavior, but also coping motives. We expected that greater positive affect would be associated with lower endorsement of coping motives based upon prior literature finding that coping motives for drinking is associated with lower daily positive affect (Armeli et al., 2014). Our results were largely consistent with the tension reduction hypothesis and suggested that increased perceived stress was associated with increased coping motives, while increased positive emotions was associated with decreased coping motives. This suggests that individuals who experience more positive emotions are less likely to endorse drinking as a way to alleviate distress or negative emotions, which aligns well with previous research that has found those higher in positive emotion to be better able to "bounce back" and effectively cope with difficult life situations (Tugade, Fredrickson, & Feldman Barrett, 2004).
Previous research has evaluated positive emotion in terms of recovery from substance use disorders (Krentzman, 2013), but positive emotion has largely been unexplored in the college student drinking literature. Our work attempted to extend the 16 role of positive emotion to binge drinking behavior and drinking coping motives.
Based on the broaden-and-build theory of positive emotions (Fredrickson, 2001;Fredrickson & Branigan, 2005), we expected that positive affect would be negatively associated with frequency of binge drinking and that it would serve as a protective factor in this way in the relationship between perceived stress and frequency of binge drinking. Contrary to our expectations, we found that neither perceived stress nor positive affect were significantly associated with frequency of binge drinking. It is notable, however, that the association of positive affect to frequency of binge drinking was trending towards significance (p = .053). As we expected, however, positive affect was revealed to be significantly related to endorsement of drinking to cope motives. Also contrary to expectations, moderation analyses did not reveal a significant perceived stress by positive affect interaction on either frequency of binge drinking or drinking to cope motives.
Findings of the present study are consistent with some research that has found that enhancement and social motives for alcohol consumption are most predictive of heavy drinking among undergraduate students, and that coping motives may be more predictive of experiencing negative-problems related to alcohol consumption than of actual consumption itself (Read et al., 2003). It may be that perceived stress is not a primary factor that is driving college students' decision to binge drink. This explanation would be consistent with what has been shown in prior lab studies, which have found participants in no-stress experimental conditions to have increased alcohol consumption compared to participants in induced-stress experimental conditions, (e.g., (Bernstein & Wood, 2017;Larsen et al., 2013), though some others have found that the opposite is true among those reporting recent patterns of heavy drinking (Bernstein & Wood, 2017). The present study adds to this mixed literature, finding that while students may not be binge drinking in particular in response to high levels of perceived stress but that despite this, some do report coping motives for drinking. One potential explanation for this discrepancy between the influence of stress on reported drinking motives and on actual observed drinking behavior may be that coping motives (as measured by the DMQ-R) assess drinking to cope regarding a variety of negative emotional experiences, only one of which is stress. It may be that other negative emotions are more related to reasons for drinking than stress (e.g., social anxiety, depressed mood). Further, it may be that students are engaging in behaviors other than binge drinking to manage stress, including marijuana use (McCormack, Laybold, Dickerman-Nelson, & Budd, 1993), and that rates of marijuana use may continue to rise as its legalization becomes more widespread. In recent years, rates of marijuana use have continued to rise, nearing but not yet reaching the level of heavy alcohol use (Lipari & Jean-Francois, 2013). Therefore, one avenue of future research may be to investigate these relationships with marijuana use as the outcome.
Alternatively, the relatively fewer males in our sample may also have presented a problem as some extant literature has found that male college students are significantly more likely to endorse heavy alcohol use as a viable means of coping with stressors as compared to females (McCormack et al., 1993;Park & Levenson, 2002). Therefore, future research should try to include a more representative sample of males and females.
Further, in the present sample, respondents appeared to be binge drinking nearly every time they drank, which may be related to the influence of permissive norms on widespread college student drinking (Borsari & Carey, 2003). This lack of variability in the frequency of binge drinking may have limited our ability to detect significant relationships among our variables of interest. This suggests that there may be strong drinking norms on college campuses that students are drinking for the purpose of drinking rather than for the purposes of stress or emotion management (Borsari & Carey, 2003). Future research should be done to examine motives for drinking related to specific drinking contexts or situations. For instance, University of Rhode Island undergraduates anecdotally report that students most commonly drink at specific bars on Tuesday and Thursday nights, and on Friday and Saturday nights at parties. Therefore, it seems likely that context is also an important component to consider in understanding both motives for drinking and actual drinking behaviors.
For example, where one drinks may be differentially related to different drinking motives and to drinking to cope with stress in particular. Drinking in certain establishments on certain nights might be more related to social enhancement motives and less because individuals are feeling stressed, whereas drinking that occurs in one's dormitory or off-campus apartment might be more likely to occur during times of stress (O'hara, Armeli, & Tennen, 2015).

Limitations and Future Directions.
While findings of the present study add to the mixed literature regarding the relations among stress, positive affect, binge drinking, and drinking to cope motives, it should be considered within the context of its limitations. Firstly, the Rhode Island Stress and Coping Inventory (Fava et al., 1998) asks about the perception of stress over the past month rather than using a more immediate measure of stress, which may be one potential explanation for the discrepancy between findings of the present study and prior work that has found college students consume more alcohol on days that they perceived as being more stressful (Park, Armeli, & Tennen, 2004). Park and colleagues (2004) examined the relationship between stress and alcohol consumption at a daily level, asking respondents to rate the degree to which they had perceived the most challenging thing that had happened that day as stressful. A scale reflecting specific stressful experiences (e.g., financial concerns, studying for exams, difficulties making friends, romantic relationships) that students may be more likely to face could potentially better reflect students' actual discrete experiences of stress (e.g., the Revised University Student Hassles Scale (Pett & Johnson, 2005), including those situation-specific stressors that serve as triggers for high levels of alcohol consumption. Further, while many self-report measures are the gold standard for collecting information on relevant constructs, the nature of self-report data has its limitations. For instance, respondents may not be accurately describing actual drinking behaviors, either because of misremembering or fear about ramifications of accurately reporting drinking. Regarding stress, respondents may not be accurately describing their overall levels of stress over the past month, and may have instead responded according to the level of stress they were experiencing in the moment rather than over the specified time period that questionnaire asks about (Gorin & Stone, 2001;Schwarz, 1999). Future research could instruct respondents to hone in on a particularly stressful week and examine positive affect and drinking behaviors during that time 20 frame only. Alternatively, real-time data collection procedures (i.e., ecological momentary assessment, experience sampling methods, daily diary methods) may provide a means by which to increase accuracy of reporting, as well as providing an opportunity to examine temporal associations between moments of stress and alcohol consumption (Stone & Shiffman, 1994), which are not possible to examine with data of a cross-sectional, correlational nature.
Further, the use of the Positive and Negative Affect Schedule (PANAS) (Watson et al., 1988) and Scale of Positive and Negative Experiences (Diener et al., 2010) may have influenced outcomes. Specifically, they do not measure positive affective experiences over the same time scalethe PANAS asks the respondent to consider the last seven days, while the SPANE asks the respondent to consider the last month. Future research could do several different things. One may be to adjust the timing so the two scales ask about the same time frame. Alternatively, perhaps a more intense measure may be to use daily data collection methodologies to improve in the validity of measurement of affect and to examine the temporal relationships between positive affect and health behaviors, including drinking. However, while the PANAS in particular has been criticized for measuring only those positive affective experiences with high activation (Barrett & Russell, 1999), it remains the gold standard for self-report assessment of affect (Crispim, Archer, & Cruz, 2014).
Together, the results of the present study help to increase our understanding of reasons for binge drinking use among college students. Our findings suggest that the relations among stress, positive affect, drinking behaviors and drinking reasons are complex. In the present sample, drinking to cope motives, but not frequency of binge drinking, appeared to be particularly related to perceived stress and positive affect.
Given prior literature suggesting that endorsement of coping motives for drinking is associated with increased negative alcohol-related consequences (e.g., car accidents, problems with friends or family), assessment of motives for alcohol use may be particularly important in screening for points of intervention, and it may be useful to offer skills building around coping with stress for students referred for alcohol treatment. Further, given the relationship found between positive affect and endorsement of coping motives, it may be beneficial to incorporate elements of positive psychotherapy (Seligman & Csikszentmihalyi, 2000), which seeks to increase positive affective states, into treatment for students seeking to address their alcohol consumption, specifically among those who endorse drinking to cope motives.
Positive psychotherapy interventions have been used as an effective treatment to target negative emotionality and increase subjective well-being (Bolier et al., 2013;Sin & Lyubomirsky, 2009), and may be well-suited to provide college students who are drinking to cope with alternative methods of coping with distress.