The interactive associations between positive and negative affect and alcohol in real-world social drinking environments

Abstract

Despite the widely known mood enhancing and anxiolytic properties of alcohol, its link with affective state is not universal and rather nuanced. We examined whether dispositional traits of positive affect (PA) and negative affect (NA) influence the link between consumed alcohol and affective states (i.e. state PA and state NA), and state arousal. Self-report measures of state affect, state arousal, and objective measures of blood alcohol concentration (BAC), were obtained in naturalistic social drinking settings (i.e. bar and drinking venues) as individuals engaged in real-world social drinking activities. The sample consisted of 60 participants (55.9% females) whose age ranged from 20 to 27 years old (M = 21.85, SD = 1.24). Trait affect and demographic measures were obtained 48 hours later via an online survey. Results revealed a between-subject interaction of trait affect by BAC on state PA and state: at higher BAC levels, individuals with high levels of trait PA, but not low trait PA, experienced significantly higher state PA. Further analyses of state PA subscales showed that this interaction was present in state calmness but not state well-being or vigor. In the context of arousal, those with high trait NA, but not low trait NA, scored higher in state high-arousal levels. Main effects of trait PA/NA on state affect were detected, but not for BAC on state affect or arousal. This highlights the impactful role that dispositional affect can have in permeating or enhancing emotional states experienced in real-world social drinking contexts.

Keywords:

• state and trait affect
• positive affect
• negative affect
• arousal
• alcohol
• social drinking
• real-world

It is widely understood that alcohol is an anxiolytic and rewarding substance (Galandra et al., 2018; Lac & Donaldson, 2018) that can facilitate social interactions, making it a historical and global staple of social activities. Past studies and theory have heavily focused on the stress dampening or mood enhancing properties of alcohol (Conger, 1956). Nevertheless, alcohol’s dynamics with affect are complex and influenced by other factors, like cognition, context, and psychosocial factors (Curtin & Lang, 2007; Sayette, 1999). Yet, it is still unknown how affect-related traits play a role in psychological states experienced during social drinking activities. Here we examine the association between consumed alcohol and state affect (i.e., in the moment short-lived emotions) as well as state arousal (i.e., non-valenced acute levels of intensity or activation). We assess how this link may be influenced by dispositional trait affect (i.e., affective tendencies lasting weeks, months, and/or throughout people’s lifespan). We extend from past studies’ focus on negative affect (NA) and pay particular attention to state and trait positive affect (PA) and state arousal levels, and we study this relation in real-world drinking environments.

Note that while NA includes a variety of unpleasant or distressful moods and feelings (e.g., sadness, anger, anxiety), PA is thought to be independent from NA and reflects feelings of pleasurable engagement with oneself and the environment, happiness, joy, excitement, or calmness (Clark et al., 1989). While NA and PA can be negatively correlated with one another, it has been suggested to study them as independent of each other (Diener & Emmons, 1984; Watson et al., 1988), illustrating the importance of considering both constructs separately when studying the link between alcohol and emotions. In fact, distinct brain systems have been implicated in the rewarding and anxiolytic effects of alcohol (Hendler et al., 2011) that manifest in elevated PA or reduced NA, correspondingly. Given the distinctions between affective and arousal states, examining the link between the positive and negative psychological states in social drinking settings and the role of trait PA and trait NA is important to better understand the nuances that lead to different patterns of alcohol use, particularly in social drinkers.

Early alcohol administration studies have shown that alcohol acutely reduces state negative affect, like anxiety, during conversations with strangers (Wilson et al., 1980). Similarly, during stressful social tasks (i.e., public speech task), when compared to a control group, those who consumed alcohol reported lower anxiety (Levenson et al., 1980) and displayed fewer facial expressions of negative affect (Sayette et al., 1992). More recently, a randomized controlled study showed alcohol administration decreased rumination, and this effect was explained by alcohol’s effect on state negative affect (Mollaahmetoglu et al., 2021). Interestingly, individuals assigned to intermediate and high levels of alcohol intoxication, but not those in no or low intoxication levels, displayed reduced startle reactions to unpleasant stimuli (Donohue et al., 2007). Reduction of stress responses as a function of increasingly higher alcohol doses have also been shown in dampened cardiovascular measures (Stewart et al., 1992) and lower self-reported state anxiety (Sher & Walitzer, 1986) when compared to low doses. Notably, when compared to a control injection of saline, alcohol intravenous infusion was associated with reduced brain responses to fearful stimuli and activation of reward brain pathways (Gilman et al., 2008), explaining some of the biological mechanisms at play. Overall, these experimental studies systematically demonstrate alcohol’s dampening effect in negative affect responses during stressful and social situations. However, as will be discussed below, a reduction in stress and/or negative affect is not the only outcome when drinking alcohol.

While earlier studies have focused on the anxiolytic effects of alcohol, more recent studies have also highlighted alcohol’s impact on state positive affect and related behaviors in social drinkers (Capito et al., 2017). For example, acute alcohol administration results in an acute increase of high-arousal positive affect measures, like elatedness, excitement, energetic, talkativeness, vigorous, among others, and this effect was more robust in heavy drinkers (Holdstock et al., 2000). Furthermore, a laboratory-based study of over 700 social drinkers detected within-subject changes in objective measures of affect as a function of alcohol consumption: a moderate dose of alcohol during social interactions with strangers increased positive affect and decreased negative affect, measured via facial expressions. Notably, the authors of this study found that social bonding was significantly higher for individuals who consumed alcohol compared to a control group (Sayette et al., 2012). These studies demonstrate the importance of alcohol research in social drinkers beyond the context of negative emotions. Nevertheless, it is still not fully understood what factors influence the experienced state PA when drinking in social settings, which could inform our understanding of people’s distinct motivations to drink.

The studies above support the idea that alcohol serves as a rewarding and anxiolytic substance, yet its influence on mood and behaviors are more nuanced (Bartholow et al., 2018; Lac & Donaldson, 2018). For instance, in a within-subject examination of behaviors before and after alcohol administration, it was found that alcohol significantly increased risk-taking behaviors (Lane et al., 2004). And aggressiveness was also found to be elevated in individuals who drank alcohol versus those assigned to a placebo beverage; this effect was true for males but not for females, demonstrating the role of psychosocial differences in the effect of alcohol on affect and behavior (Giancola, 2002b). A possibility is that dispositional traits are differentially linked to behaviors and emotions experienced under the influence of alcohol, and that the acute effect of alcohol on emotions and behaviors are of instrumental value in individuals depending on their traits. For instance, it has been shown that highly extraverted individuals benefit more from mood enhancing effects of alcohol compared to those low in extraversion; the authors of this study also showed that only for highly extraverted participants, positive affect behaviors, like smiling, were associated with higher mood enhancement and social bonding (Fairbairn et al., 2015). This study highlights the role that dispositional traits can play to obtain the beneficial effects of alcohol in the context of social drinking. Yet it is unclear how affective traits, such as trait-like PA and NA, differentially moderate the link between alcohol consumption and affective states in social settings.

Higher trait PA could explain individuals’ motivation to engage in social drinking activities and the benefits obtained from it. For instance, it has been postulated that for individuals high in traits like positive emotionality, alcohol use in social settings is characterized by a motivation and expectancy to enhance positive emotions as well as the reward from alcohol and enhanced social experience (Creswell, 2021; Creswell et al., 2015). Notably, trait PA is positively associated with obtained positive social benefits (Lyubomirsky et al., 2005; Schimmack et al., 2004) and with an openness to experience that is beneficial in social contexts (Fredrickson, 2001). Based on these PA-related frameworks and on Creswell’s social drinking framework, those with high trait PA may be able to better capitalize on the mood-altering benefits of alcohol via social interactions, as they are more likely to increase their state PA when surrounded by positive and social activities (Lyubomirsky & Layous, 2013), amplifying the rewarding properties of alcohol. Furthermore, high trait PA is associated with an enhanced ability to benefit from social rewards (Aderman, 1972), enhanced willingness to start conversations (Batson et al., 1979), and ability to achieve higher quality social interactions (Berry & Hansen, 1996). Of relevance, alcohol increases smiling and talkativeness among strangers (Sayette et al., 2012), so it can serve as a facilitator for positive social engagement (Sayette et al., 2019). These findings in conjunction with studies showing that rewarding responses and higher state PA increases as a function of more consumed alcohol (Aan Het Rot et al., 2008; Kastl, 1969; Tizabi et al., 2007) suggest that increases of consumed alcohol in conjunction with high trait PA may robustly enhance experienced positive emotions in social drinking settings.

Conversely, trait NA may interact differently with alcohol in social settings because elevated NA dispositions are thought to influence perceptions of social experiences and their emotional reaction to them, possibly influencing their drinking expectancies or motivations. In fact, there is evidence on the association between higher NA tendencies and drinking to cope motives and more alcohol consumption (Bravo et al., 2017; Hussong, 2003). In the context of social activities, people with elevated NA, like depressive symptoms, are more likely to experience increased daily negative social experiences (i.e., social conflict or negative interactions) and reduced well-being and sense of belonging (Steger & Kashdan, 2009). They also react more strongly to social exclusion and interpersonal rejection (Nezlek et al., 1997), suggesting that people with higher trait NA may be more emotionally vulnerable to negative social experiences (Machell et al., 2015). This is important in the context of social drinking, as research has shown alcohol can influence mood when concerned about social evaluation (Sayette & Wilson, 1991). This aligns with the early studies overviewed above that highlight the anxiolytic effects of alcohol during perceived social threats.

The propensity to experience negative emotions in high trait NA individuals in social settings is important to understand alcohol’s link to state affect. This is because according to the social attributional approach model, alcohol disables social threat appraisals to access social rewards more readily (Fairbairn & Sayette, 2014). Given people’s alcohol expectancies of relieving negative emotions (Jones et al., 2001), people with higher trait NA who tend to experience more negative and/or unstable outcomes in social situations (Fairbairn & Sayette, 2014) could be motivated to drink for the anxiolytic properties of alcohol to relieve their negative emotional reactions to social experiences. This is consistent with past studies and Creswell’s models, which illustrates that in contrast with people with high trait PA, extraversion, and social ease, people with higher negative emotionality, introversion, and social discomfort, drink to cope with negative emotions and are likely to hold beliefs that alcohol decreases NA and tension (Creswell, 2021; Creswell et al., 2015). Aligned with this idea, an early systematic review of over 82 studies showed similar trends in 10- to 25-year-olds: extraversion and sensation seeking traits positively predicted mood enhancing drinking motives, and neuroticism and anxiety was strongly linked with coping drinking motives (Kuntsche et al., 2006). Lastly, given alcohol’s sedative or anxiolytic alcohol effects increase as a function of dose (Donohue et al., 2007; Imperato & DiChiara, 1986), it can be expected that higher BAC can be linked with reduction of state NA in people with high trait NA.

The intensity or physical activation felt during psychological experiences (i.e., arousal) (Barrett, 1998; Russell, 1980) is a non-valenced dimension of affective states that is less studied in social drinking contexts. Independent of valence, arousal can be measured via physiological measures, like heart rate or skin conductance (Sforza et al., 2000) or it can be self-reported under affect-related survey items, like feeling active versus passive (Yik et al., 2011). Incorporating arousal measures to move beyond studying emotions as discrete to approach them as multidimensional processes can help detect the nuances of the link between alcohol and affect (Stritzke et al., 1996). Thus far, what is known about the association between arousal and alcohol is inconsistent and nuanced, with classic studies showing that alcohol diminishes arousal (Stritzke et al., 1995), and other researchers, like Burish et al. (1982), finding that alcohol does not influence stress-induced arousal and only affects state PA and NA in non-stressful situations. This highlights the importance of alcohol-related research in not just isolated scenarios, but also in real social settings that are contextually rich. Examining individuals’ PA and NA traits can help clarify the association between alcohol and arousal in social drinkers. This is because their affective dispositions to engage in these activities may differ and distinctly influence how aroused or activate, they feel during their drinking experiences.

The relation between consumed alcohol and emotions as a function of affect traits can also depend on the context in which it is examined, but past findings are mixed. For instance, more robust alcohol effects on state PA and NA have been detected in controlled laboratory procedures when compared with measures from real-world investigations (Fairbairn et al., 2018). In contrast, earlier experimental studies have shown that contextual factors related to social activities, like the friendliness of the social activity, predicted enhanced mood in people assigned to drink either alcohol or placebo (Fromme & Dunn, 1992). This study highlighted the importance of context and social factors when studying experienced emotions during social drinking activities. To study the effect of alcohol on social drinkers, it is beneficial to examine it in ecologically valid settings, like real-world drinking establishments and bar venues, as people engage in their real social drinking activities with their own peers, friends, and/or drinking companions. The ambiance of naturalistic bar settings (e.g., music, lighting, and crowds) could also influence people’s social interactions and experienced emotions. Furthermore, although drinking often occurs in social settings for many people, particularly young people, much of the alcohol-mood research uses either self-reported retrospective assessments, or it is performed in laboratory settings, potentially threatening the validity and generalizability of past findings. Thus, more studies assessing state affect in real-time as drinking occurs can help better detect sources of variance mentioned above, like contextual factors of social and ecological valid settings.

Here we address several gaps in the alcohol-affect literature. Generally, studies tend to examine the impact of alcohol on affect in laboratory settings (Creswell et al., 2015; Curtin et al., 2001; Hefner & Curtin, 2012), focus on negative emotions (Foo et al., 2018; Fox et al., 2007; Gilman et al., 2008), or are limited to subjective measures, such as self-reports of alcohol use (Monk et al., 2020). Given the known complexity of emotions during social interactions, we aim to understand how PA and NA dispositional traits distinctly interact with objective measures of consumed alcohol (i.e., blood alcohol concentration, BAC) and predict emotions and arousal captured in real-time and real-world social settings. Based on evidence presented above (Creswell, 2021; Fairbairn & Sayette, 2014; Fredrickson, 2001), we expect that for individuals with high trait PA, higher levels of BAC would predict higher state PA. In contrast, for individuals with high trait NA, we expect state NA to be lower when BAC is higher. These linear relations between state affect and BAC are expected based on past evidence showing the effects of alcohol are dose dependent, with affect, physiological, and behavioral responses occurring as a function of alcohol increases (Aan Het Rot et al., 2008; Curtin & Lang, 2007; Donohue et al., 2007; Kastl, 1969; Sher & Walitzer, 1986; Stewart et al., 1992; Tizabi et al., 2007), which also align with animal studies showing neuronal activation involved in reward and anxiolytic processes occurs as ethanol levels increase (Imperato & DiChiara, 1986; McBride, 2002). Lastly, we perform exploratory analyses on the link between BAC and state high-arousal and low-arousal and the moderating role of trait PA and NA in this association.

1. Method

1.1. Sample

Participants who completed all parts of the study were included in the main analysis for this investigation.¹ Our final sample size (N = 60) aligned with a power analysis indicating a required sample of at least 54 participants to achieve large-size effects F = .35, Power = .95, and alpha error-probability = .05 (Cohen, 1988; Faul et al., 2007) in a moderation regression analysis with three predictors in each model, including the predictor (X), outcome (Y), and interaction term (XY). A sample size of 60 approximated the sample size required (i.e., 63 to 70 participants) if two to four covariates were added to the models. Likewise, when comparing samples of similar studies, a sample size of around 48 to 72 participants was common (e.g., Hefner & Curtin, 2012; Ostafin & Brooks, 2011; Pihl & Zacchia, 1986; Simons et al., 2005). The age for the full sample ranged from 20 to 27 years old (M = 21.85, SD = 1.24). Weight ranged from 110 to 215 lb (M = 152.14, SD = 29.45). Gender for those who reported it included 33 females (55.9%) and 26 males (44.1%). Ethnicity/race of participants was Caucasian 53 (88.3%), Black/African American 3 (5%), Asian 2 (3.3%), and Hispanic 2(3.3%). This study was approved by the University of Kansas Institutional Review Board (IRB).

1.2. Measures

An Alco-Sensor breathalyzer from Intoximeters, Inc. was used to obtain objective measures of consumed alcohol. To prevent cross-contamination, a new mouthpiece was used for each participant. State PA and NA (i.e., feelings present “right now”) and Trait PA and NA (i.e., feelings “over the past weeks”) were measured by a shortened version of the Profile of Mood States (POMS-SV; Usala & Hertzog, 1989). “The POMS is a validated and reliable mood scale that has been used in numerous past investigations, including alcohol administration studies in which investigators measured participants’ state affect occurring under the influence of alcohol (Corbin et al., 2008; Knowles & Duka, 2004; Lusher et al., 2004; Reeves & Nagoshi, 1993). The scale ranged from 0 to 4 for each item: Not at all accurate (0), A little accurate (1), Moderately accurate (2), Quite a bit accurate (3), and Extremely accurate (4). State and trait PA were calculated from averaging scores in PA subscales of vigor, well-being, and calmness. State and trait NA were calculated from the overall mean of fatigue, depression, and anxiety. High-arousal (i.e., mean of self-report scores on intense and active scores) and low-arousal (i.e., mean calculated from quiet and passive scores) levels were calculated to measure non-valenced dimensions of affect from the POMS-SV. State affect or experienced emotions during social drinking engagement was collected in real-time in part 1 of the study (i.e., bar settings). Trait affect and demographics assessments were sent out to participants (48 hours later) via an internet survey to be completed for part II of the study.

1.3. Procedure

Local bar venues and alcohol-serving establishments were contacted and obtained approval for the recruitment of participants at their respective bar locations. Recruitment posters and a table with study instruments and information about the study were set up at the bar venues. Interested individuals were invited to participate after explaining the study to them. If they agreed to participate, an informed signed consent was obtained. Lastly, to protect participants’ identity, they were not required to provide names or identification. All procedures were in adherence with approved IRB. In part I of the study (i.e., bar settings), participants completed the POMS-SV. We asked the participants to not drink, eat, chew gum, or put their hands near their mouth while completing the survey to increase the accuracy of the breathalyzer. After the survey was completed, the participants were administered the breathalyzer test to measure their BAC. Then, they were given an informational card with an explanation of what BAC is, their personal BAC reading, the Kansas legal BAC limit for driving, and contact information for the University of Kansas’ “Safe Ride Program.” At this time, participants were compensated $2 for their brief in-bar participation. Forty-eight hours later, participants completed self-reports on trait affect and other psychosocial factors. Upon full completion of this two-part study, participants were compensated $8.

1.4. Analytic approach

Pearson’s correlation analyses were conducted to assess any age or weight covariances. Analysis of variance (ANOVA) and t-tests were also conducted to assess overall differences driven by gender or race/ethnicity groups. To assess our main hypotheses on the role of trait affect in the link between BAC and state affect and arousal, we conducted a series of moderation regression analyses using IBM SPSS statistics (Version 29, 2022). For each analysis, BAC was set as the main predictor (X), state affect and arousal levels as the outcome variable (Y), and trait affect the moderator (M). Weight and sex were added as covariates in all analyses. For all linear regressions involving interactions, the variables were mean-centered and multiplied together to create an interaction term. Simple slope probing was conducted to assess the nature of the associations in the moderation analysis. Confidence intervals are included and effect size for the full models were calculated using Cohen’s f² (Selya et al., 2012).

2. Results

Prior to our main analysis, we examined BAC and state-trait affect in the context of theoretically meaningful covariates, including gender, age, weight, and race (Kleiner et al., 2004; Vogel-Sprott & Barrett, 1984; Wilsnack et al., 2000). No significant correlations between these covariates and state-trait affect were detected, except for gender differences in trait NA, with males displaying lower trait NA (M = .64, SD =.48) than females (M = .94, SD = .70), t(58) = −1.86, p < .05). In the context of BAC, no correlations were found for either age r(60) = .05, 95% CI (−0.21, 0.3) p = .70 or weight r(59) = .22, 95% CI (−0.036, 0.45), p = .09. While gender was significantly related to BAC, with males having higher BAC (M = .08, SD = .05) than females (M = .05, SD = .04), t(57) = 2.44, p < .05). No BAC differences were detected based on race F(3, 56) = .58, p= .63, possibly due to the lack of ethnic diversity in our sample. Given the role of weight and sex in alcohol metabolism, both were adjusted for in our main analyses. Lastly, the number of peers, friends, and/or drinking companions (M = 3.72, SD = 2.57) was examined in relation to affect, given that high trait PA and low trait NA may be related to social network characteristics (Van Kleef, 2009); however, no significant associations were found between this factor and BAC, or state-trait affect.

2.1. Correlations between state-trait affect and arousal

As it would be expected, trait PA was significantly correlated with state PA, r(54) = .40, 95% CI (.16, .60), p = .002, and negatively correlated with state NA r(54) = −.27, 95% CI (−.49, −.01), p = .04. Moreover, trait NA was positively associated with state NA r(54) = .32, 95% CI (.07, .54), p = .02, but interestingly, it was not associated with state PA r(54) = −.17, 95% CI (−.41, .09), p = .21, suggesting our findings on state PA are independent of NA. In the context of activation, trait PA was not correlated with high-arousal r(54) = .25, 95% CI (−.01, .48), p = .07 or low-arousal r(54) = −.02, 95% CI (−.28, .24), p = .90. Similarly, trait NA was not significantly associated with high-arousal r(54) = .00, 95% CI (−.26, .26), p = .10 or low-arousal r(54) = .15, 95% CI (−.12, .39), p = .28. Refer to Table 1 for descriptive statistics of BAC, state and trait affect and arousal.

Table

Measures	Mean	SD	Min	Max	Skewness
BAC	.06	.05	.00	.19	.81
State PA	2.74	.67	.83	3.67	−.79
State NA	.66	.51	.00	2.00	.83
Trait PA	2.43	.77	.33	3.81	−.56
Trait NA	.81	.62	.00	3.07	1.48
State High Arousal	1.84	.95	.00	4.00	.26
State Low Arousal	.88	.81	.00	3.00	.82

Notes: Descriptive statistics of Blood Alcohol Concentration (BAC), state and trait positive affect (PA), state and trait and negative affect (NA), state arousal, including means, standard deviations, range, and skewness.

2.2. Moderation analyses of the association between trait affect, BAC, state affect, and state arousal

First, when examining interactions of trait affect and BAC on state PA, we found a significant moderation regression model R² = .28, F(3, 56) = 6.71, p = .009, f² = .62 (Figure 1), which reflected a significant interaction of trait PA and BAC on state PA (ß =.31, t(53) = 2.30, 95% CI [−3.59,4.22], p = .03. In other words, higher BAC levels were associated with elevated state PA, but this was only for people with higher trait PA scores. As such, no significant main effect of BAC on state PA was detected: ß = .19, t(53) = 1.49, 95% CI [−4.46, 4.84], p = .14. There was a main effect of trait PA ß = .36, t(53) = 2.99, 95% CI [0.16, 0.57], p = .004, indicating that for each one unit increase in trait PA, there is a .36 unit increase in state PA. When probing simple slopes, conditional effects of the interaction were only significant at values above 1SD over trait PA’s mean (i.e., $\ge$3.24), ß = 5.60, t(53) = 2.01, p $\le$ .05, but not significant for either average levels of trait PA ß = 1.78, t(53) = .95, p $>$ .05 or lower than average trait PA ß = −1.84, t(53) = −.82, p $>$ .05. This suggests that state PA was linked to higher BAC only if scoring above average in trait PA, so alcohol itself does not predict state PA.

Figure 1. Interaction of trait PA with BAC on the effects of alcohol on state PA.

Notes: Depiction of the significant interaction between trait PA and blood alcohol content (BAC) on state PA. State PA is represented on the y-axis, and BAC is represented on the x-axis. The aquamarine simple slope (S.E. = 2.72) represents state PA as a function of BAC for those with high trait PA (.76) or above the mean. The red line represents the simple slope (S.E. = 1.88) at mean trait PA (.00), and the blue line (S.E. = 2.25) represents the simple slope at low trait PA values below the average (-.76). This moderation graph shows that higher BAC is positively associated with higher state PA, but only for people with higher trait PA (>1SD) above trait PA’s mean. The opposite pattern can be seen for people with low trait PA but was not statistically significant. Dichotomous splits were used for figure purposes only.

Additionally, we also examined state PA subscales (i.e., state calm, state wellbeing, and state vigor) in a series of regression models. In contrast with the other state PA subscales, we only found a significant model for the state calm subscale: R² = .27, F(3, 56) = 5.92, p = .001, f²= .61, reflecting a significant interaction of trait PA by BAC on state calm (M = 2.54, SD = .98), (ß = .38, t(53) = 2.79, 95% CI [−5.69, 6.45], p = .01). Results revealed a significant main effect of trait PA, ß = .27, t(53) = 2.24, 95% CI [−0.04, .59], p = .03; this indicated that with each unit increase in trait PA, there was a .27 increase in state calm, but no main effects were found for BAC on state calm, ß = −.04, t(53) = −.28, 95% CI [−5.47, 5.39], p =.78. Further analysis of simple slopes yielded a significant interaction at levels below 1SD of trait PA average ($\le$1.62), ß = −6.85, t(53) = −2.01, p $\le$ .05, but this was not the case for average scores in trait PA, ß = −.81, t(53) = −.29, p $\ge$ .05 or at 1SD above trait PA average, ß = 4.72, t(53) = 1.19, p $\ge$ .05. These between-subject conditional effects show that the lower the trait PA levels, the less state calm was experienced if scoring high in BAC, mirroring the BAC by trait PA interaction on state PA mentioned above.

No significant interactions were detected for other state PA subscales, including state well-being (M = 3.14, SD = .73), ß = .26, t(53) = 1.95, 95% CI [−4.26, 4.79], p = .06, which did yield a main effect of trait PA (ß = .29, t(53) = 2.41, 95% CI [.06, .53], p = .02 but not of BAC, ß = .25, t(53) = 1.92, 95% CI [−3.80, 4.29], p = .06. Similarly, a main effect by trait PA was found for state vigor ß = .27, t(53) = 2.01, 95% CI [−.03, .57], p = .04 but not of BAC ß = .26, t(53) = 1.92, 95% CI [−4.78, 5.30], p = .06, or BAC by trait PA interactions influencing state vigor levels (M = 2.34, SD = .87), ß = .06, t(53) = .43, 95% CI [−5.58, 5.70], p = .67. Similarly, when examining if trait NA interacted with BAC to reduce state NA, we found a not statistically significant moderation regression model R² = .14, F(3, 56) = 2.69, p = .06, f² = .16, reflecting that there were no significant interactions of trait NA and BAC state NA (ß = .22, t(53) = 1.49, 95% CI [−5.58, 6.01], p = .14 or BAC main effect (ß = .08, t(53) = .55, 95% CI [−2.84, 2.99], p = .59; nevertheless, we did find a main effect of trait NA on state NA (ß = .41, t(53) = 2.82, 95% CI [−5.33, 6.21], p = .007, which indicated that trait NA enhanced negative emotions experienced in social drinking settings independent of BAC.

In the context of arousal, a moderation model R² = .22, F(3, 56) = 2.85, p < .05, f² = .54 (Figure 2) showed a significant interaction of trait NA by BAC on state high-arousal (ß = .37, t(53) = 2.71, 95% CI [−10.51,11.25] p = .009). In this model, there were no significant main effects on high arousal levels by either BAC, ß = .21, t (53) = 1.62, 95% CI [−5.26, 5.68], p = .11 or by trait NA, ß = .17, t (53) = 1.22, 95% CI [−0.27, 0.61], p = .23. Further analysis of simple slopes revealed a significant conditional effect on state NA at 1SD above trait NA’s mean ($\ge$ 1.42), ß = 11.12, t(53) = 2.23, p $\le$ .05 but not at trait NA’s mean ß = 2.72, t(53) = .96, p $\ge$ .05 or 1SD below its average ($\le$.16) ß = −5.68, t(53) = −1.52, p $\ge$ .05. These findings suggest high trait NA individuals are significantly more likely to experience elevated state high-arousal at higher between-subject BAC scores, but alcohol itself did not significantly predict state arousal (Figure 2). In contrast, no significant interactions were found between trait PA by BAC on high-arousal (ß = −.08, t(53) = −.56, 95% CI [−6.52, 6.35], p = .58), nor did we find a trait PA main effect (ß = .23, t(53) = 1.74, 95% CI [−.11, .57], p = .09) or a BAC main effect on high-arousal (ß = .14, t(53) = .99, [−5.62, 5.89], p = .33).

Figure 2. Interaction of trait NA with BAC on the effects of alcohol on state high-arousal.

Notes: Depiction of a significant interaction between trait NA and blood alcohol content (BAC) on arousal. State High-Arousal is represented on the y-axis and BAC on the x-axis. The aquamarine simple slope (S.E. = 4.98) represents state High-Arousal as a function of BAC for those with high trait NA (.63) or above the mean. The red line represents the simple slope (S.E. = 2.82) at mean trait NA (.00), and the blue line (S.E. = 3.74) represents the simple slope at low trait NA or below the average (-.63). Here, we see that for individuals with higher trait NA, higher arousal levels were higher for people showing higher BAC levels. This correlation between BAC and high arousal levels was not detected in individuals with lower trait NA. Arousal dichotomous splits were used for figure purposes only.

Lastly, our exploratory analyses did not yield any significant moderation regression models for state low-arousal. This is reflected in the non-significant interactions of trait NA by BAC on state low-arousal ß = −.22, t(53) = −1.46, 95% CI [−9.80, 9.37] p = .15, or any main effects by either trait NA, ß = .05, t(53) = .34, 95% CI [−.34, .44], p = .72, or by BAC, ß = −.18, t(53) = −1.28, 95% CI [−4.50, 4.64], p = .21. This was also the case for trait PA, with no significant interactions of trait PA by BAC on state low-arousal, ß = −.11, t(53) = −.73, 95% CI [−5.71, 5.48], p = .47), or main effects by trait PA, ß = −.04, 95% CI [−.33, .26] t(53) = −.28, p = .78) or BAC, ß = −.19, t(53) = −1.30, 95% CI [−5.20, 4.82] p = .20) on state low-arousal.

3. Discussion

In this study we examined if trait-like affect moderates the associations between alcohol consumption and affective states in social drinkers. We assessed this relation using objective measures of consumed alcohol (i.e., BAC) and real-time experienced emotions or state affect and state arousal in ecologically valid environments (i.e., real social drinking activities in real-world bar venues and drinking establishments). First, although no main effects of BAC were detected on state affect or arousal, our primary findings showed that affective traits significantly moderated the association between BAC and state PA in social drinking settings. In other words, there was a positive association between state affect and consumed alcohol as a function of dispositional traits: higher state PA was positively associated with higher BAC, but only for individuals with high trait PA, while no significant associations were detected between BAC and state PA at other levels of trait PA. When further assessing state PA subscales, this interaction was only significant for feelings of calmness, but not for state well-being or state vigor. In contrast with numerous past studies highlighting the mood-altering effects of alcohol, here we found that experienced state PA during social drinking activities does not seem independently linked with BAC, emphasizing the role of affective traits over alcohol in social drinking environments.

The rather unexpected lack of associations between consumed alcohol and emotions and arousals did, however, align with an earlier study in which contextual factors during social activities, like friendliness, significantly enhanced mood in people regardless of whether they drank alcohol or placebo (Fromme & Dunn, 1992). Suggesting alcohol is not always the only factor responsible for changes in affect in social drinking settings. This is important because many past alcohol studies have been conducted in controlled laboratory settings or during solitary drinking sessions. Therefore, this study highlights the importance of studying behaviors in real-world settings, which shows that the link between alcohol and affect is more nuanced than it has been postulated to be. It is also possible that we could not detect any associations due to the timing in which consumed alcohol was measured. In other words, we did not measure if the BAC levels were at the ascending or descending limb of the blood alcohol curve, which have been shown to influence affect differently (Earleywine & Martin, 1993; Rueger & King, 2013). Thus, we suggest that future studies taking place in naturalistic drinking environment still measure or control the timing in which alcohol is consumed to compare emotions experienced at various timepoints of the blood alcohol curve.

Why did trait PA interact with BAC to predict state PA, but no interactions between trait NA and BAC to reduce NA were found here? First, this study steers away from the conventional focus on solitary drinking and the anxiolytic effects of alcohol in clinical populations. Instead, we examined alcohol and both positive and negative emotions as well as arousal levels experienced in naturalistic settings in young social drinkers. Although we did not assess for specific alcohol use related diagnoses in our sample, our findings point out that state PA is likely to be enhanced in social drinkers with positive dispositional traits engaging in these social related activities. The observed low range values of trait and state NA relative to observed high scores in state PA and trait PA in our sample could imply that during social drinking activities, experienced emotions are more positive due to people’s already inherently positive affective dispositions and higher likelihood of participating in these activities. Consistent with Cresswell’s model (Creswell, 2021), our findings could shed some light in the contrast between young people’s solitary drinking patterns and its link with negative affect and problematic alcohol use (Skrzynski & Creswell, 2020), while social drinkers (likely captured in this study) may reserve alcohol consumption as a “social lubricant” and thus less likely to drink to reduce negative emotions.

Our study highlights the importance of affect dispositions that can impact individuals’ overall social experiences, explaining why people with higher trait PA can benefit from engaging in social activities compared to people with high trait NA. Although alcohol consumption did not predict state PA, the data from this study supports the emotional enhancement model Cresswell’s social drinking model, as those with high positive emotionality tendencies were more likely to experience higher state PA, albeit only at higher BAC at the between levels. Trait PA may be more involved in alcohol’s influences on mood in social drinking contexts because of the reward mechanisms at play during social interactions (Charlet et al., 2013). Trait PA might be especially helpful because it is known to broaden attentional scope (Fredrickson et al., 2001), perhaps counteracting alcohol’s narrowing of perception of negative cues in the environment (Fromme & Dunn, 1992). Thus, individuals’ motives and expectations of the possibility of positive experiences when drinking can allow high trait PA individuals to focus on the broader pleasant aspects of real-world drinking settings and capitalize more from positive social interactions (Mansell et al., 2002; Reis & Gable, 2003).

In our exploratory analyses on arousal, we did not find significant interactions of trait PA by BAC on either high or low arousal levels. Instead, moderation analyses showed that for individuals with high trait NA, between-subject levels of BAC were positively correlated with experiencing elevated state high-arousal. In other words, people with high trait NA or negative emotion dispositions were more likely to feel more activated or aroused during social drinking activities at higher levels of consumed alcohol. It is plausible that wanting to feel more aroused or activated in social settings is more likely to occur in some groups that possess higher negative dispositional traits. For instance, a correlational study showed that higher scores in trait-like depression levels were correlated with more risky behaviors under the influence of alcohol (Hubicka et al., 2010). And in a survey-based study of over six hundred college students, self-reported risky behavior tendencies, like drunk driving, were positively correlated with higher extraversion and negatively correlated with constraint personality trait (Vollrath & Torgersen, 2008); Similarly, drunk driving was linked to higher sensation seeking traits and negatively associated with altruism (Ge et al., 2014). Thus, arousal seeking tendencies should be considered when studying risk factors,

Moreover, our findings suggest that it is not the case that reduced stress or state NA is the only outcome when individuals drink alcohol, and it suggests that dispositional factors may play a role in how young individuals feel and behave under the influence of alcohol. This is important to consider, particularly in young adults who are more prone to experience negative affect (Charles et al., 2001) and lower emotion regulation tendencies (Gross et al., 1997) relative to older groups. In fact, an EMA study of young adults measured daily and momentary alcohol consumption and affect, it was found that momentary elevated arousal, but not affect, predicted subsequent increased likelihood of drinking (Peacock et al., 2015). These findings highlight the relevance of arousal over affect in some age groups. To better detangle the effect of alcohol on arousal in social drinking settings, future studies could benefit from using objective measures of arousal, such as physiological markers like heart rate and skin conductance and study across different ages.

Despite these relevant implications, the current study had limitations. First, although our ecologically valid approach was a strength of the study, we did not measure state affect before and after alcohol consumption, limiting our ability to detect changes in emotions as a function of alcohol over time. This was partially due to our main goal of unobtrusively assessing the interactions between dispositional affect and alcohol on experienced emotions and arousal. By performing this unobtrusive one-time assessment on psychosocial factors and state affect (rather than several time-point affect assessments), we intended to lower participants’ exposure to the true nature of the study, as to not influence their expectations/behaviors or induce participant biases. Due to the geographic location of where this study took place, another limitation of this study is the low diversity in the sample in terms of race and ethnicity and cultural background, which could have helped generalize our findings to other groups. It is highly encouraged for future studies to assess if cultural and/or ethnic background are plausible influential factors associated with emotions experienced social drinking settings. Furthermore, it is possible that for our sample (primarily young adults engaging in social drinking activities in a college town area), the main drive to consume alcohol could have been a social one, explaining the elevated state PA found here. Nevertheless, these findings cannot be generalized to other populations, like middle age and older adults.

Lastly, the interactions seen here could have unfolded differently in individuals whose motivation to drink is not only to engage in social activities. While high trait PA people seem to benefit from social drinking activities, it has also been suggested that those that most robustly benefit from this alcohol’s mood enhancing feature can be at a higher risk of developing alcohol use disorders (Creswell, 2021). In fact, elevated state PA has been linked to heavy drinking: a daily diary study showed that PA at the between and within levels predicted heavy drinking (Jones et al., 2021). This is consistent with more recent findings from a meta-analysis of over 12,000 people, which showed that in daily diary and ecological momentary assessment studies, state PA (but not NA) was associated with elevated drinking behaviors (Dora et al., 2023). Here we provided insights into which dispositional traits play a role in emotions experienced in social drinkers, which could be implemented in prevention frameworks of alcohol misuse, like binge-drinking in young people. Future research should combine within-subject assessments with both ecologically valid methods and controlled laboratory experiments to tease apart contributing factors present in the real world and elucidate the nature of the detected associations. Overall, despite widely held expectations for the mood-enhancing properties of alcohol, here we show alcohol’s link with emotions is not always present or universal and may be more dependent on individuals’ affective profiles and the context in which it is studied.

Acknowledgements

The authors would like to thank Chris Maxwell and our laboratory research assistants for their work on data collection. We also thank Kasey Creswell for her comments on an earlier version of this manuscript.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

At this moment we cannot make the data available to the public due to current IRB permissions obtained for this study and individuals’ participation agreement.

Additional information

Notes on contributors

Veronica Ramirez

Veronica Ramirez is a PhD student in Psychological Science at the University of California, Irvine. Her work focuses on the biological and physiological factors underlying negative and positive emotions, as well as addictive behaviours.

Claire M. Gorey

Dr. Claire Gorey obtained her PhD in clinical psychology from the University of South Florida. Her research background is largely in the substance use area and the processes underlying addiction.

Sarah D. Pressman

Dr. Sarah Pressman is a Professor of Psychological Science and the Associate Dean of Undergraduate Education at the University of California, Irvine. She has published extensively on the beneficial effects of emotional and social well-being for physical health and the mechanisms by which these experiences “get under the skin” to influence biology.

Notes

1. Two subjects completed both parts but were removed from analyses due to missing BAC data and for reporting affect values outside the scale options.