Comparison of event-specific drinking motives reported at the first drink versus retrospectively the next morning

Abstract

Objective:

Drinking motives are proximal predictors of alcohol use and often conceptualized as trait-like constructs. Yet, research shows motives are dynamic, varying day-to-day. We compared associations between event-specific motives reported at first drink versus retrospectively the next morning and alcohol consumption and consequences.

Method:

Heavy-drinking college students (N=95) completed 28 days of ecological momentary assessments. At first drink of the day, participants reported whether they were drinking to reduce depression, reduce anxiety, have fun, get high/buzzed/drunk (“high”), and/or not be left out (“conformity”). These motives, total drinks consumed, and consequences were retrospectively reported the next morning.

Results:

Students were more likely to report “fun” and “high” motives retrospectively than at first drink. When assessed retrospectively, “fun” and “high” motives were associated with more drinks and “conformity” motives with fewer drinks. When reported at first drink, only the “high” motive was associated with consumption. Retrospective reports of “fun”, “high”, and “conformity” motives were associated with more positive consequences. When assessed at first drink, only “fun” and “high” motives were significant. Only “high” motives, reported at first drink, significantly increased likelihood of a negative consequence.

Conclusions:

Findings suggest event-level effects of motives on drinking outcomes depend on when motives are assessed. More associations between retrospective motives and outcomes suggest that drinking motives may change within an event and/or young adults reconstruct their drinking motives based on their experience. Findings have event-level assessment design implications and provide evidence that antecedent “high” motives present the greatest event-level risk for heavy drinking and negative outcomes.

Introduction

Alcohol is the most common substance used by young adults, with 65% of young adults ages 19 to 30 reporting alcohol use in the past 30 days and 27% reporting heavy episodic drinking (defined in the Monitoring the Future Survey as 5+ drinks/occasion) in the past 2 weeks (Patrick et al., 2023). Heavy alcohol use among young adults is associated with numerous adverse consequences including feeling depressed, anxious, or ashamed, harm to physical health, negatively impacting friend and family relationships, worse work/school performance, and driving under the influence of alcohol (Patrick et al., 2020). Given the high prevalence and adverse effects of heavy drinking, understanding proximal predictors of alcohol use and related consequences among young adults continues to be a public health priority.

One way to understand alcohol use behavior among young adults is to examine drinking motives or specific reasons for drinking alcohol. Drinking motives are robust predictors of alcohol use and alcohol-related consequences (Kuntsche et al., 2005; Votaw & Witkiewitz, 2021). Motivational models suggest that individuals drink based on what they expect to gain (positive reinforcement) or avoid (negative reinforcement). Moreover, the outcome individuals expect to achieve by drinking may be either external (i.e., social) or internal (i.e., emotional). Considerations of these dimensions has resulted in a four-factor model of alcohol motives: (a) to socialize or celebrate (social), (b) to enhance feelings of pleasure or positive mood/well-being (enhancement), (c) to conform to peer pressure and avoid social rejection (conformity), and (d) to cope with negative emotions/relieve stress (coping) (Cooper, 1994; Cox & Klinger, 1988; Kuntsche et al., 2005). More recently, a five-factor model has been explored with separate coping-anxiety and coping-depression factors to reflect the distinct relationships between coping-anxiety and coping-depression motives and alcohol consumption and related consequences (Grant et al., 2007).

The different motives for alcohol use are uniquely associated with drinking patterns and consequences (Cox, 1990; Cox & Klinger, 1988; Lyvers et al., 2010). Coping motives have been associated with greater drinking problems and alcohol-related consequences (Crutzen et al., 2013; Kuntsche et al., 2005) and coping-depression but not coping-anxiety motives have been associated with heavier alcohol consumption (Grant et al., 2007). Enhancement motives have the strongest associations with heavier drinking (Kairouz et al., 2002; Kuntsche & Cooper, 2010; Kuntsche et al., 2005). In contrast, social and conformity motives are associated with more moderate drinking (Kuntsche et al., 2005). Identifying one’s motives for drinking may be important for improving prevention and intervention efforts targeting risky drinking.

Much of the research has conceptualized motives as stable, trait-like constructs (Cooper et al., 2008; Crutzen et al., 2013; Littlefield et al., 2010). Such research that considers motives as individual-difference or between-person variables has explored differences in outcomes between people who drink for different reasons (e.g., examining differences in those who drink for social versus coping reasons). However, motivational theories posit that alcohol consumption is driven by proximal drinking motives, and emerging evidence suggests there are significant within-person or state-like variations in drinking motives. In other words, it appears that individuals may drink for different reasons in different situations (Arbeau et al., 2011; Armeli et al., 2016; O’Hara et al., 2015; O’Hara et al., 2014).

Experience sampling methods such as ecological momentary assessment (EMA) allows researchers to systematically assess the relationship between event- or day-level motives and alcohol use (Votaw & Witkiewitz, 2021). EMA allows for repeated assessments and reduces recall bias by capturing data about an individual’s current state or behavior in or close to real time in their daily life (Shiffman, 2009; Wray et al., 2014). As such, EMA affords the opportunity to disentangle within- from between- person effects; examining not only differences in people who drink for different motives, but also understanding how their drinking motives change across drinking occasions.

Research has found discrepancies between self-reports collected retrospectively (e.g., over weeks or months) and real-time/next day assessments of mood (Solhan et al., 2009), mental health symptomology and wellbeing (Christensen et al., 2024; Kiefer et al., 2024; de Vries et al., 2025), and behavior (Shiffman et al., 1997; Freeman et al., 2023). Studies across multiple fields suggest real-time assessments are more reliable and valid when compared to traditional retrospective reports that require longer recall and summarization of past experiences (Shiffman et al., 2008; Dulin et al., 2017; Leertouwer et al., 2022).

Studies comparing alcohol use data derived from intensive longitudinal assessment (e.g., daily diary, EMA) versus traditional retrospective measures such as timeline follow back (TLFB) have been mixed. Some studies found lower alcohol consumption when reported retrospectively (Monk et al., 2015; Patterson et al., 2019), yet others have found higher estimates of drinking on a TLFB report (Merrill et al., 2020) or no difference (Chow et al., 2017). Additionally, limited work has examined differences in summary EMA assessments (e.g., next day morning report) and real-time assessments of drinking, finding less consumption reported retrospectively the next day versus in real-time (Monk et al., 2015). Discrepancies between assessments may be influenced by the context at the time of assessment (Shiffman et al., 2008) variability of one’s experiences (Perrine & Schroder, 2005; Gmel & Daeppen, 2007; Solhan et al., 2009), and consumption levels and/or consequences experienced during the drinking event (Freeman et al., 2023; Stevens et al., 2020). Studies have yet to examine differences in retrospective and real-time endorsement of specific drinking motives and how timing of assessment of motives relates to drinking outcomes.

While recent EMA studies have begun to examine motives in relation to event-level outcomes, most have only measured motives once per day (Votaw & Witkiewitz, 2021). A recent review of EMA studies examining substance use motives found most studies examine drinking motives after alcohol use, as opposed to before or during the drinking event (Votaw & Witkiewitz, 2021). Such retrospective reporting may be influenced by the totality of the substance use experience. It may also be that motives change across a single drinking event. Contextual factors (e.g., who the individual is with, time of day) and one’s experience during the event may lead to changes in drinking motives during the drinking occasion. For example, a young adult attending a party with new peers may initially drink to conform, but as the night goes on and they become more comfortable they may continue to drink to enhance their mood. Additionally, young adults may reconstruct their drinking motives to reflect the outcome of the drinking event and recall of motives may also be influenced by the time that has elapsed since use (e.g., directly after drinking versus the following day) (Piasecki et al., 2007; Shrier & Scherer, 2014). Therefore, it is likely that reports of drinking motives could differ depending on when drinking motives are assessed.

As an alternative to assessing drinking motives after the drinking event, some EMA studies have assessed drinking motives at a set time such as in the early evening (e.g., between 4pm and 6pm) which was assumed to be prior to a drinking event (Dvorak et al., 2014; Stevenson et al., 2019). The authors of these studies acknowledged that motives for drinking may change between the time they were reported, and the time people began drinking. If true, it may be more optimal to report motives right at the start of the drinking event. One EMA study assessed drinking motives if a participant reported currently drinking and hence during the drinking event, but motives were assessed just once per day (Joyce et al., 2018). Another recent EMA study reported assessing drinking motives via three assessments at random intervals in the morning, afternoon, and evening (O’Donnell et al., 2019). As such, motives were captured sometimes before and other times after the drinking event. The only outcome examined was alcohol consumption and this study did not examine whether timing of when drinking motives were reported influenced associations between motives and consumption.

While much of the event-level research has examined associations between motives and alcohol use (Dvorak et al., 2014; Hamilton et al., 2020; Stevenson et al., 2019) and (to a lesser extent) negative-related outcomes (Kuntsche & Labhart, 2013), it is also important to identify which event-specific motives are associated with positive consequences of drinking. Positive consequences stemming from use are likely key factors that contribute to the maintenance and escalation of subsequent alcohol use. One study examining event-level motives and drinking intensity found on days young adults reported greater enhancement and social motives (i.e., positive reinforcement motives) they were more likely to report heavier drinking and experience positive consequences (Patrick & Terry-McElrath, 2021). Individuals who drink for positive reinforcement and indeed achieve their desired effects may engage in more frequent and heavier drinking. Understanding which motives are associated with positive consequences that reinforce and escalate future use may inform intervention development to reduce risky drinking.

The gaps and limitations of the current research highlight the importance of considering the timing of when to assess event-level drinking motives. Thus, additional research is needed to (a) assess drinking motives at multiple time points within a day, (b) examine whether endorsement of drinking motives differ based on timing of assessment, and (c) determine if associations between event-level drinking motives and outcomes (i.e., drinking consumption, positive consequences, negative consequences) differ based on timing of when motives are assessed.

Purpose of Study

This study examined whether the timing of assessment of drinking motives (at the first drink of an event versus retrospectively the next morning) impacts (1) motive endorsement and (2) motive links to alcohol use, positive consequences, and negative consequences. Since retrospective assessments may capture additional motivations that arose during the drinking event, we hypothesized that each type of motive would be more likely to be reported retrospectively. Additionally, we hypothesized that compared to first-drink motives, retrospectively reported motives (relevant to any point during the prior day drinking event) would be more consistently associated with event-level drinking outcomes. Based on previous research (Kuntsche & Labhart, 2013; Patrick & Terry-McElrath, 2021), we also hypothesized regardless of timing that “high” and “fun” motives (i.e., enhancement/social motives) would be associated with more drinks and positive consequences, while “depression” and “anxiety” motives (i.e., coping motives) would be associated with negative consequences.

Understanding differences in endorsement of motives in real-time vs. retrospectively can inform future event-level study designs. If endorsement of motives differs by timing of assessment, this would call into question current practices of assessing these constructs once a day in event-level studies. Understanding when drinking motives are associated with outcomes of interest also may inform the best timing for event-level interventions.

Methods

Participants

One hundred participants completed a baseline assessment, 28 days of ecological momentary assessments (EMA), and a follow-up interview. Eligible participants were age 18–20, had access to a smartphone and data plan, were currently enrolled in a 4-year university, had no past two-week drug use other than cannabis, were not currently in treatment for a substance use disorder, and reported either (a) weekly heavy episodic drinking (based on sex-specific drinking thresholds of 4+/5+ drinks [females/males] in a single sitting) or (b) at least 1 (of 10 assessed) negative alcohol-related consequence in the past two weeks. Of the 100 participants, five did not report real-time drinking during the 28-day assessment period and were removed from analyses. Participant characteristics for the analytic sample of 95 are summarized in Table 1.

Table 1.

Sample Descriptives (n=95)

	N/% OR Mean (SD)
Age	18.66 (0.66)
Biological sex
Female	50 (52.6%)
Gender (check all that apply)
Men	44 (46.3%)
Women	50 (52.6%)
Trans Women	1 (1.1%)
Genderqueer/Gender non-conforming	2 (2.1%)
I prefer not to answer	1 (1.1%)
Race (check all that apply)
White	69 (72.6%)
African American	7 (7.4%)
Asian	21 (22.1%)
Native American/Alaskan	1 (1.1%)
Hawaiian/Pacific Islander	1 (1.1%)
Other	5 (5.3%)
Hispanic/Latino Ethnicity	14 (14.7%)
Year in School
First Year	77 (81.1%)
Sophomore	14 (14.7%)
Upperclass	4 (4.2%)
Greek Involvement	15 (15.8%)
Baseline Drinking Behavior
Drinks per typical week	10.42 (6.35)
Average drinks per drinking day	4.52 (2.20)
EMA Drinking Behavior
Average number of drinking days	4.43 (2.39)
Average drinks per drinking day	5.24 (2.80)
Average number of negative consequences per drinking day	0.78 (1.12)
Average number of positive consequences per drinking day	2.47 (1.88)
Number of Days Negative Consequences Endorsed
Any negative consequence	188 (45%)
Embarrassed yourself	40 (10%)
Became rude/obnoxious	20 (5%)
Hurt or injured yourself by accident	11 (3%)
Felt nauseated or vomited	49 (12%)
Behaved aggressively	11 (3%)
Neglected school-related obligations	73 (17%)
Had a hangover	76 (18%)
Forgot what you did	51 (12%)
Number of Days Positive Consequences Endorsed
Any positive consequence	352 (84%)
Made a new friend/acquaintance	176 (42%)
Talked to someone you probably wouldn’t have spoken to otherwise	163 (39%)
Made others laugh	216 (51%)
Something fun/exciting happened	143 (34%)
Something that normally would bother you failed to bother you	35 (8%)
Creative moment/experience	15 (4%)
Slept better	165 (39%)
Expressed your feelings more easily	126 (30%)

A priori power analysis (Rochon, 1997) was conducted for the primary aims of the study from which these secondary data are drawn. The primary aims of the study were to examine how real-time subjective evaluations of positive and negative alcohol-related consequences influence latency to and the amount of alcohol consumption at the next drinking event. This analysis indicated 100 participants were needed for a target power >.80 to detect moderate effects. Post-hoc power analysis for the present investigation suggested that a Level 1 sample size of 4.5 (4.5 drinking days) and Level 2 sample size of 95 (95 young adults who drink) for a target level of power > .80 would result in a minimum L1 detectable effect size of about .18 for small, medium, and large intraclass correlation coefficients (ICC) and a minimum L2 detectable effect size of .52 for large ICCs, .36 for medium ICCs, and .32 for large ICCs (Arend & Schäfer, 2019).

Recruitment and Orientation

Participants were recruited via flyers on and around local university campuses in the greater Providence, RI area and through social media advertising. Interested individuals were directed to complete an online screener and those who were eligible were then directed to an online informed consent. Those who consented completed an online baseline assessment and signed up for an in-person group orientation session. During group orientations, participants provided informed consent for the EMA portion of the study. Research staff reviewed study procedures and provided training on the definition of a standard drink as 12oz. of beer, 5 oz. of 12% table wine; 12oz. of a wine cooler; or 1.25 oz. of 80-proof liquor. Participants downloaded the EMA application on their phones and completed practice assessments.

EMA Procedures

The 28 days of EMA began the day after the last group orientation session.¹ During this EMA phase, real-time and retrospective next day data were collected through surveys sent to participants’ mobile phones. Surveys were programmed and delivered using the Metricwire Inc. application platform (www.metricwire.com). Participants were asked to report on previous day drinking events in a daily morning report. Morning reports were triggered at 7am each day and remained available all day, yet participants were instructed to complete the report as soon as possible each morning. Participants were also asked to self-initiate an EMA report when they began drinking and this first-drink report triggered hourly follow-up reports. Data for the current study analyses was solely collected via the morning report and first-drink report, the only reports where motives were assessed.

Measures

Baseline

Demographics.

Participants completed demographic information on age, sex, gender, race, ethnicity, year of school, and Greek involvement.

Alcohol Behaviors.

Participants reported their typical weekly alcohol use in the past 30 days using a 7-day grid similar to the daily drinking questionnaire (DDQ) (Collins et al., 1985). Data on typical drinks consumed each day of the week were summarized as total drinks and number of drinking days per typical week.

EMA Measures

Alcohol Use.

Each morning participants indicated whether they drank the previous day. If they reported drinking, they indicated the total number of standard drinks consumed during the drinking event.

Drinking Motives.

At their first alcoholic drink on a given drinking day, participants responded in real-time to “What are your reasons for drinking right now?” with five motive response options in which participants were asked to check all that apply including: to feel less depressed (“depression” motive), to feel less nervous/anxious (“anxiety” motive), to make the day/night more fun (“fun” motive)², to get high, buzzed, or drunk (“high” motive), and to not be left out (“conformity” motive). Additionally, the next day in the morning report participants were asked to respond to “What was your reason(s) for drinking yesterday?” with the same five motive response items as at first drink. Items were adapted from validated drinking motive measures (Grant et al. 2007; Cooper, 1994). Each motive was examined as a dichotomous variable: motive not endorsed (0) versus motive endorsed (1).

Alcohol Consequences.

When previous day drinking was endorsed, participants were asked to indicate if they experienced any of 18 positive and negative consequences.³ Consequences were derived from multiple sources including the Brief Young Adult Alcohol Consequence Questionnaire (Kahler et al., 2008), the Positive Consequences Questionnaire (Corbin et al., 2008), and a study by Lee et al. (2017) on event-level alcohol consequences. For this study, we examined 8 consequences typically viewed as negative experiences (embarrassed yourself, became rude/obnoxious, hurt or injured yourself by accident, felt nauseated or vomited, behaved aggressively, neglected school-related obligations, had a hangover, forgot what you did). Each consequence was asked about on a separate screen stating “During/after drinking yesterday, did you [consequence]?” with response options yes (1) or no (0). We summed the number of negative consequences experienced on a given day, but variability in number of consequences endorsed was very low; thus, negative consequences was represented as a dichotomous variable: no negative consequences experienced (0) versus at least one negative consequence experienced (1). For positive consequences, we summed across 8 consequences typically viewed as positive experiences to obtain a daily count of positive consequences (made a new friend/acquaintance, talked to someone you probably wouldn’t have spoken to otherwise, made others laugh, something fun/exciting happened, something that normally would bother you failed to bother you, creative moment/experience, slept better, expressed your feelings more easily).

Transparency and Openness

We report how we determined our sample size, all data exclusions (if any), all manipulations, and all measures in the study. Data and study materials are available by emailing the corresponding author. Data analyses were run using the HLM 7.03 program (Raudenbush et al., 2013). This study’s design and analyses were not pre-registered.

Analysis Plan

We used multilevel modeling, analyzing only drinking days when both first drink and morning reports were completed (for the sake of direct comparison). Models were run using full information maximum likelihood estimation. Fully unconditional models (i.e., no predictors) were run to determine intraclass correlation coefficients (ICCs) and parse out information on the percentage of variation in motives at both the between-and within-person level.

To test hypotheses, we conducted two sets of analyses. First, to test the hypothesis that each of the five motives would be more likely to be reported retrospectively in the morning than at first drink, we used three-level models with report (Level 1) nested in day (Level 2) nested in person (Level 3). For these models, report type [first drink (real-time report, coded 0) versus morning (retrospective report, coded 1] at Level 1 was a predictor of endorsement of each motive. Day-level (Level 2) covariates included (a) weekday (Sunday-Thursday; coded 0) vs. weekend (Friday or Saturday; coded 1) given increases in alcohol use on weekends, (b) study day (1–28) to account for potential changes in drinking behavior over time, and (c) morning report submission time (person-centered). Morning report submission time was included as a covariate because, unlike the submit time of the first drink report, which was linked to the start of a drinking event, there was variability in when the morning report was submitted with respect to their prior day drinking event. The degree of retrospection varied across days as the morning report was made available to all participants from 7:00am-4:59pm. We also included a person-level (Level 3) covariate for sex (0=female, 1= male). Binomial distributions were used for the dichotomous motive outcomes.

For the next set of analyses, to test whether motives reported retrospectively the next morning (versus at first drink) would be more consistently associated with the event-level drinking outcomes, we examined two-level models with day (Level 1) nested in person (Level 2). For the first set of these analyses, all five motives at Level 1 (person-centered; assessed retrospectively) were entered as predictors of (a) total drinks, (b) number of positive consequences, and (c) odds of experiencing a negative consequence. A parallel set of three models using motives assessed at first drink were examined. The same day-level covariates were used as in the prior model. It should be noted that morning report submission time was only included in models that examined retrospectively reported motives. Additionally, when predicting consequences, number of drinks (person-centered) was controlled. Person-level covariates included sex and proportion of study days where each motive was endorsed (grand mean-centered). As noted, to isolate within person differences, Level 1 predictors of all five motives were centered around the person mean and Level 2 proportion of study days where each motive was endorsed were centered around the grand mean.

On drinking days, an average of 5.24 (SD=2.80) drinks were consumed each day with a range of 1–17 drinks. A visual inspection of the number of drinks consumed variable revealed a near normal distribution and there was not evidence of skewness (.779) or kurtosis (.665), so we specified a Gaussian distribution. Additionally, at the day-level, number of positive consequences resembled a count, average number of positive consequences endorsed on a drinking day was 2.47 (SD=1.88) with a range of 0–8 consequences experienced, therefore, a Poisson distribution adjusting for overdispersion was specified. Experiencing any negative alcohol-related consequence in this study was common, with any negative consequence reported on 45% of days when both first drink and morning reports were completed. Yet, variability was very low in terms of the number of consequences that were reported (M=0.78, SD=1.12; range of 0–6 negative consequences); thus, negative consequences were represented as a dichotomous variable. Binomial distributions were used for the dichotomous negative consequences models. For all models, intercepts were specified as random, to account for mean differences in outcomes across individuals, and slopes were tested for significance and fixed when non-significant.

Results

In the analytic sample (N=95), across 28 days of EMA, participants completed 2631 (98.9%) morning reports. On average, participants completed morning reports at 10:37 am, and 2103 (79.9%) were completed before noon each day. Participants reported prior day drinking on 484 days (18.4% of the morning reports completed). In comparison, participants completed 429 first drink reports (88.6% of the days that drinking was reported the next morning). Both first drink and retrospective morning reports were completed on 421 drinking days which represents 87.0% of the drinking days based on morning reports; this subset of days was used in analyses.⁴

Sample descriptives from both baseline measures and EMA reports are presented in Table 1. Table 2 reports the prevalence of drinking events where (a) motives reported retrospectively matched motives reported in real time (i.e., motives endorsed both at first drink and morning report, motives neither endorsed at first drink nor morning report) and (b) motives reported retrospectively mismatched motives reported in real time (i.e., motives endorsed only at first drink report, motives only endorsed at morning report). Table 3 provides the number of participants that endorsed each motive and the number of times each motive was endorsed for both retrospective morning and real-time first-drink reports. This table includes ICCs of each motive when reported retrospectively in the morning and in real-time at first drink. ICCs for motives reported retrospectively in the morning ranged from .27 to .41 and motives reported in real-time at first drink ranged from .28 to .37 suggesting much of the variance of event-level motives is attributable to within-person changes regardless of timing of assessment.

Table 2.

Prevalence of Motives Endorsed at First Drink and/or Morning Report

	N/% of days endorsed at:
Motives	Matched		Mismatched
	First-Drink and Morning Report	Neither First-Drink nor Morning Report	Only First-Drink Report	Only Morning Report
Depression	39 (9%)	369 (88%)	6 (1%)	7 (2%)
Anxiety	41 (10%)	358 (85%)	14 (3%)	8 (2%)
Fun	314 (75%)	55 (13%)	13 (3%)	39 (9%)
High	221 (53%)	140 (33%)	21 (5%)	39 (9%)
Conformity	46 (11%)	346 (82%)	17 (4%)	12 (3%)

Table 3.

Motive Endorsement at Retrospective Morning and Real-time First-Drink Reports

	Retrospective Morning Reports			Real-time First-Drink Reports
Motives	ICC	N (% of 95 Reported)	# days (% of 421 drinking days)	ICC	N (% of 95 Reported)	# days (% of 421 drinking days)
Depression Motive	0.38	21 (22.1%)	46 (10.9%)	0.29	25 (26.3%)	45 (10.7%)
Anxiety Motive	0.41	21 (22.1%)	49 (11.6%)	0.37	24 (25.3%)	55 (13.1%)
Fun Motive	0.27	90 (94.7%)	353 (83.8%)	0.34	89 (93.7%)	327 (77.7%)
High Motive	0.28	80 (84.2%)	260 (61.8%)	0.28	77 (81.1%)	242 (57.5%)
Conformity Motive	0.36	27 (28.4%)	58 (13.8%)	0.35	29 (30.5%)	63 (15.0%)

ICC= Intraclass correlation coefficients

Depression Motive = to feel less depressed, Anxiety Motive = to feel less nervous/anxious, Fun motive = to make the day/night more fun, High Motive = to get high/buzzed/drunk, Conformity Motive = to not be left out

Note: Based on the 421 days when both first drink and corresponding morning reports were completed

Discrepancies in Motive Reporting

In multilevel models, as hypothesized, we found that “fun” (OR=1.72, 95% CI=1.28, 2.31) and “high” (OR=1.32, 95% CI=1.05, 1.65) motives were more likely to be reported retrospectively in the morning than at first drink. Endorsement of “depression” (OR=1.04, 95% CI=0.75, 1.44), “anxiety” (OR=0.80, 95% CI=0.47, 1.37), and “conformity” (OR=0.87, 95% CI=0.67, 1.12) drinking motives did not differ by assessment timing (Table 4).

Table 4.

Discrepancies in Motives Reported in Retrospective Morning versus First-Drink Report

	Depression Motive		Anxiety Motive		Fun Motive		High Motive		Conformity Motive
	OR	95% CI	OR	95% CI	OR	95% CI	OR	95% CI	OR	95% CI
Intercept	0.04	(0.01, 0.12)	0.06	(0.02, 0.18)	7.00	(3.19, 15.33)	0.40	(0.20, 0.82)	0.06	(0.02, 0.16)
Level 1 Effects
Report Type	1.04	(0.75, 1.44)	0.80	(0.47, 1.37)	1.72	(1.28, 2.31)	1.32	(1.05, 1.65)	0.87	(0.67, 1.12)
Level 2 Effects
Study Day	1.04	(0.99, 1.10)	1.04	(0.99, 1.08)	0.98	(0.94, 1.01)	1.01	(0.98, 1.05)	1.02	(0.98, 1.06)
Weekend	0.70	(0.34, 1.45)	0.38	(0.18, 0.80)	1.12	(0.64, 1.94)	2.42	(1.40, 4.19)	1.19	(0.43, 3.26)
Morning Report Time	0.91	(0.79, 1.05)	0.99	(0.86, 1.16)	0.99	(0.90, 1.10)	1.07	(0.96, 1.20)	0.89	(0.76, 1.04)
Level 3 Effects
Sex	1.39	(0.53, 3.67)	2.42	(0.84, 6.96)	0.62	(0.28, 1.39)	1.93	(0.92, 4.07)	0.87	(0.67, 1.12)
Random Effects	SD	Variance	SD	Variance	SD	Variance	SD	Variance	SD	Variance
Intercept	1.32	1.74	1.29	1.65	0.94	0.89	1.09	1.19	0.99	0.99

Significant effects of interest are bolded

OR=Odds Ratio

Depression Motive = to feel less depressed, Anxiety Motive = to feel less nervous/anxious, Fun motive = to make the day/night more fun, High Motive = to get high/buzzed/drunk, Conformity Motive = to not be left out, Morning Report Time = morning report submission time

Report Type = Real-time (First Drink) Report =0, Retrospective (Morning) Report =1; Weekend = Sunday-Thursday =0, Friday-Saturday=1; Sex = Female =0, Male=1

No random slopes

Motives Predicting Number of Drinks

When examining retrospective morning reports, “fun” (B =1.48, p<0.001) and “high” (B =1.45, p<0.001) motives were associated with having more drinks, and “conformity” motives (B = −0.90, p=0.043) were associated with drinking fewer drinks. However, when examining motives from the first-drink report, only “high” motives (B =0.91, p=0.013) significantly predicted an increased number of drinks consumed during the drinking event (Table 5).

Table 5.

Motives Predicting Number of Drinks

	Retrospective Morning Reports				Real-time First-Drink Reports
	B	SE	t	p	B	SE	t	p
Intercept	3.70	0.35	10.47	<0.001	3.74	0.33	11.17	<0.001
Level 1 Effects
Depression Motive	0.63	0.59	1.07	0.288	0.65	0.46	1.41	0.169
Anxiety Motive	0.64	0.38	1.67	0.095	0.52	0.42	1.24	0.223
Fun Motive	1.48	0.34	4.39	<0.001	0.33	0.31	1.07	0.292
High Motive	1.45	0.29	5.02	<0.001	0.91	0.35	2.54	0.013
Conformity Motive	−0.90	0.44	−2.04	0.043	−0.68	0.41	−1.64	0.104
Study Day	−0.01	0.01	−0.68	0.497	−0.01	0.01	−0.56	0.583
Weekend	0.92	0.22	4.23	<0.001	1.09	0.25	4.31	<0.001
Morning Report Time	0.06	0.05	1.16	0.245	-	-	-	-
Level 2 Effects
Depression Motive	−0.27	0.95	−0.28	0.778	−0.08	0.97	−0.08	0.936
Anxiety Motive	−1.93	0.93	−2.06	0.042	−1.39	0.74	−1.88	0.064
Fun Motive	1.18	0.68	1.73	0.088	1.16	0.58	1.99	0.049
High Motive	1.34	0.53	2.54	0.013	1.52	0.58	2.62	0.010
Conformity Motive	1.59	0.70	2.26	0.027	1.07	0.72	1.50	0.138
Sex	1.49	0.39	3.78	<0.001	1.10	0.35	3.18	0.002
Random Effects	SD		Variance		SD		Variance
Intercept	1.59		2.53		1.43		2.05
Weekend Slope	-		-		0.60		0.36
High Motive Slope	-		-		1.93		3.73
Conformity Motive Slope	-		-		1.38		1.89

Significant effects of interest are bolded

Depression Motive = to feel less depressed, Anxiety Motive = to feel less nervous/anxious, Fun motive = to make the day/night more fun, High Motive = to get high/buzzed/drunk, Conformity Motive = to not be left out, Morning Report Time = morning report submission time; Weekend = Sunday -Thursday =0, Friday -Saturday =1; Sex = Female =0, Male=1

Random slopes for real-time first-drink high and conformity motives and weekend

As studies often combine “depression” and “anxiety” motives, and since “fun” motives may be functioning as either a social motive and/or enhancement motive, in addition to models described above we also conducted models where both coping motives were combined into one dichotomous coping variable and where both “fun” and “high” motives were combined into one dichotomous social/enhancement motive variable. The majority of results predicting number of drinks remained consistent (see Supplementary Table 1), with the exception that combined coping motives reported retrospectively were associated with increased number of drinks consumed (B =0.92, p=0.023).

Motives Predicting Positive and Negative Consequences

When assessed retrospectively in the morning, reports of “fun” (ERR=1.70, 95% CI=1.33, 2.18), “high” (ERR=1.32, 95% CI=1.09, 1.59), and “conformity” (ERR=1.26, 95% CI=1.05, 1.52), motives were associated with increased likelihood of experiencing positive consequences the day/night before, controlling for number of drinks consumed. When assessed in real-time at the first-drink report, only “fun” (ERR=1.36, 95% CI=1.11, 1.68) and “high” (ERR=1.19, 95% CI=1.01, 1.40), motives significantly predicted positive consequences. “High” drinking motives reported at first drink (OR=2.14, 95% CI=1.22, 3.74) significantly predicted odds of experiencing a negative consequence when controlling for drinks consumed (Table 6).

Table 6.

Motives Predicting Negative and Positive Consequences

	Likelihood of any Negative Consequences				Number of Positive Consequences
	Retrospective Morning Reports		Real-time First-Drink Reports		Retrospective Morning Reports		Real-time First-Drink Reports
	OR	95% CI	OR	95% CI	ERR	95% CI	ERR	95% CI
Intercept	0.54	(0.23, 1.23)	0.61	(0.27, 1.42)	2.39	(1.87, 3.06)	2.28	(1.77, 2.92)
Level 1 Effects
Depression Motive	1.41	(0.39, 5.14)	1.54	(0.53, 4.43)	0.91	(0.74, 1.12)	1.00	(0.75, 1.34)
Anxiety Motive	0.96	(0.27, 3.42)	0.77	(0.28, 2.15)	0.88	(0.69, 1.13)	0.96	(0.76, 1.21)
Fun Motive	1.79	(0.74, 4.32)	1.29	(0.65, 2.58)	1.70	(1.33, 2.18)	1.36	(1.11, 1.68)
High Motive	1.69	(0.96, 2.96)	2.14	(1.22, 3.74)	1.32	(1.09, 1.59)	1.19	(1.01, 1.40)
Conformity Motive	0.99	(0.35, 2.79)	0.76	(0.30, 1.90)	1.26	(1.05, 1.52)	1.03	(0.84, 1.26)
Study Day	1.02	(0.99, 1.05)	1.02	(0.99, 1.04)	0.98	(0.97, 0.99)	0.98	(0.97, 0.99)
Weekend	1.01	(0.56, 1.83)	0.96	(0.53, 1.76)	1.13	(0.93, 1.37)	1.19	(0.98, 1.46)
Morning Report Time	0.91	(0.80, 1.04)	-	-	1.03	(0.99, 1.07)	-	-
Total Drinks	1.58	(1.33, 1.90)	1.59	(1.33, 1.89)	1.07	(1.04, 1.10)	1.09	(1.06, 1.12)
Level 2 Effects
Depression Motive	0.67	(0.11, 4.13)	0.24	(0.34, 1.74)	1.25	(0.68, 2.33)	0.91	(0.46, 1.80)
Anxiety Motive	4.26	(0.80, 22.76)	8.24	(1.64, 41.28)	1.36	(0.79, 2.34)	1.59	(0.98, 2.57)
Fun Motive	2.32	(0.42, 12.86)	1.88	(0.48, 7.33)	2.37	(1.36, 4.13)	1.62	(1.05,2.49)
High Motive	1.51	(0.49, 4.70)	1.86	(0.61, 5.62)	1.11	(0.78, 1.57)	1.11	(0.80, 1.54)
Conformity Motive	2.29	(0.41, 12.69)	2.38	(0.51, 11.03)	1.12	(0.69, 1.83)	1.36	(0.89, 2.09)
Sex	0.95	(0.45, 2.00)	0.95	(0.46, 1.94)	1.03	(0.82, 1.30)	1.05	(0.84, 1.31)
Random Effects	SD	Variance	SD	Variance	SD	Variance	SD	Variance
Intercept	1.29	1.66	1.25	1.57	0.44	0.19	0.44	0.19

Significant effects of interest are bolded

OR=Odds Ratio; ERR=Event Rate Ratio

Depression Motive = to feel less depressed, Anxiety Motive = to feel less nervous/anxious, Fun motive = to make the day/night more fun, High Motive = to get high/buzzed/drunk, Conformity Motive = to not be left out, Morning Report Time = morning report submission time; ERR=event rate ratio; Weekend = Sunday-Thursday =0, Friday-Saturday =1; Sex = Female =0, Male=1

No random slope

In a model with combined coping motives and combined social/enhancement motives, most results predicting consequences remained consistent (see Supplementary Table 2), with the exception that enhancement/social motives reported retrospectively were associated with odds of experiencing a negative consequence (OR=4.11, 95% CI=1.19, 14.20).

Discussion

Identifying optimal timing to assess drinking motives, arguably one of the most proximal predictors of drinking, will help us understand risky drinking behavior and inform intervention development focused on targeting motivational processes of drinking. Given the momentary and dynamic nature of drinking motives, we conducted a novel examination of whether the timing of assessment of drinking motives influences motive endorsement. This study was also the first to examine effects of both real-time and retrospectively reported motives on event-level drinking outcomes, including not only alcohol use, but also positive and negative alcohol-related consequences.

Our findings suggest that young adults report different reasons for drinking in real-time at the start of a drinking event versus retrospectively the next morning. We also found timing of assessment has implications not only for whether some motives are endorsed, but also for whether they are related to drinking outcomes. More associations between specific motives and outcomes (drinks and positive, though not negative, consequences) were observed for motives assessed next morning relative to when drinking began.

Participants were more likely to endorse “fun” and “high” (i.e., social/enhancement) motives in their retrospective assessments. Frequent endorsement for social/enhancement motives found in this study correspond to findings from other event-level studies examining drinking motives of young adults (Arbeau et al., 2011; Armeli et al., 2014). A recent review of event-level studies of substance use motives found college students reported enhancement motives during more than 80% of drinking episodes and social motives during more than 50% of drinking episodes (Votaw & Witkiewitz, 2021). In our study, “fun” and “high” motives were associated with greater consumption and positive consequences (as hypothesized) and “high” motive was associated also with odds of experiencing a negative consequence. This study provides evidence that “high” motive may present the greatest event-level risk for heavy drinking and negative outcomes among college students.

We did not see differences in endorsement of “depression” and “anxiety” (i.e., coping motives) and “conformity” motives retrospectively versus in real-time. Also, “depression” and “anxiety” motives each did not increase event-level risk. These results add to the mixed findings in the event-level literature which have found no associations between event-level coping motives and outcomes (Cook et al., 2019, Stevenson et al., 2019) or associations with higher alcohol consumption (Dvorak et al., 2014, O’Hara et al., 2014) or more negative consequences (Patrick & Terry-McElrath, 2021). For example, O’Hara and colleagues (2014) found among a young adult African American sample that day-level endorsement of drinking motives was associated with heavier drinking that day. The authors suggest stress from racial discrimination and stigma may be related to drinking to cope and its impact on drinking levels in this population. Interestingly in supplemental analyses where we combined “anxiety” and “depression” motives into one dichotomous coping motive variable, we did find coping motives reported retrospectively were associated with increased number of drinks consumed. Thus, mixed findings between coping motives and negative outcomes may be due to differences in assessment across studies. More research is needed to examine (a) how differences in daily assessment of coping motives influences associations with drinking outcomes and (b) how time varying constructs such as context, affect, and stress may influence the relationship between day-level coping motives and drinking outcomes.

Participants in our study reported lower endorsement of “depression”, “anxiety”, and “conformity” motives compared to other motives which aligns with findings from prior event-level (non-treatment seeking) young adult studies (Arbeau et al., 2011; O’Hara et al., 2014). This low endorsement may explain why we did not see differences in endorsement of these motives retrospectively versus in real-time and saw few associations with outcomes. Another possible explanation is the effects were smaller than those that our sensitivity power analysis suggested we were powered to detect.

A plausible explanation for why retrospective assessments yielded more frequent endorsement of “high” and “fun” motives, and more associations with outcomes is that young adults who drink may reconstruct their motives based on their experience. When asked to recall one’s motives for use retrospectively the next day, young adults may be influenced by the experience of the drinking event such as how they felt when drinking or the consequences they experienced and report motives that correspond to their drinking outcomes. For example, on occasions when young adults experience a fun, positive drinking experience, they may be more likely to recall drinking to have fun. To empirically examine whether young adults are reconstructing their motives based on their drinking experience, future work could compare assessment of motives taken at specific intervals during the drinking event until participants report they have stopped drinking and compare those to retrospectively reported motives the next morning.

Additionally, more motive endorsement and associations with drinking outcomes found for retrospective assessments may be due to morning reports reflecting a cumulation of motives that occurred across the entire drinking event. It may be that motives change across the drinking event and the accumulated set of motives is reflected in young adults’ next day morning assessments. For example, dynamic contextual factors and/or experiences during a drinking event may lead to changes in motives across that event, effects of which cannot be captured at first drink. Social learning theories highlight the influence of both mood (internal context) and the social environment (external context) on cognition, suggesting that such contextual factors may play a role in influencing drinking motives across drinking occasions leading to changes in drinking motives when these contextual factors change during the event (Maisto et al., 1999). Prior research has found that young adults drink for different reasons depending on the situation (Kairouz et al., 2002; Kuntsche et al., 2006), but future research is needed to understand how and what contextual factors may influence drinking motives and related outcomes during a drinking event.

It is also important to note that not only did the morning and first drink report differ in their time of assessment, but also in the timeframe being assessed. The first drink report motive assessment asked about a narrow timeframe (“right now”) while the morning report assessment asked about a broad timeframe (“yesterday”). It is possible that endorsement of motives at first drink would look different if young adults are asked about a broader timeframe such as “What are your reasons for drinking today?” or had motives been assessed later in the drinking event. Young adults may endorse more motives as they anticipate or experience potential changes in context (i.e., mood, location, who they are with) during their drinking event. More research examining whether and how motives change over different time intervals is needed.

If motives change across a drinking event and retrospective reporting reflects the accumulation of motives across that event, this may prompt researchers to consider retrospective reporting to be the most accurate and post-drinking to be the more valid time to assess drinking motives. Indeed, the majority of event-level studies to date have chosen to assess motives after alcohol use (for a review, see Votaw & Witkiewitz, 2021). Yet, if young adults are reconstructing their motives based on their experience, researchers may want to take caution when using motives assessed retrospectively as proximal predictors, as such assessment may reflect drinking outcomes and not necessarily motives for drinking during the drinking event. Additionally, retrospective reporting even as soon after as the next morning may be susceptible to recall bias. For example, young adults who drank heavily the night before may recall their reasons for drinking inaccurately or simply not remember all the reasons they drank the prior day. Studies have shown that level of intoxication impacts reporting of drinking behavior (Stevens et al., 2020) and thus, may also impact recall of one’s motives for drinking. This suggests retrospective reporting of drinking motives may be more susceptible to recall bias following occasions of heavier drinking. Alternatively, young adults may just vary in how they report drinking motives at different assessments. Prior research has found individual reporting may be influenced by the context an individual is in when they complete an assessment (Shiffman et al., 2008). Drinking cognitions including motives are dynamic constructs and this may be reflected in fluctuating responses to assessments across a day or event within a drinking occasion. Each of these potential influences of event-level motive assessment should be considered when choosing when to assess motives as there may not be one correct time to assess drinking motives.

An alternative approach would be to assess drinking motives frequently throughout the drinking event. Although not measured in the current study, an EMA study could be designed to assess motives at the start of drinking and at subsequent assessments while individuals are still drinking. For example, participants could be asked to self-initiate a drinking report when they start drinking and report their drinking motives and then receive triggered drinking follow-up reports every half-hour at which time they are asked to report their motives for continuing to drink and these assessments could continue to be triggered until participants report they are no longer drinking. Such real-time assessments would be advantageous in that they would allow for an exploration of multiple constructs together including drinking context, drinking behavior (e.g., level of intoxication, current consequences experienced), and motives for drinking/continuing to drink. Such real-time assessments would reduce recall bias and may also reduce reconstruction of motives based on one’s reflection of their experience since the drinking event is still occurring. It would also be important to examine whether making motives more salient across drinking events (through frequent assessment) would alter consumption patterns and consider the burden placed on participants when asking them to complete repeated assessments. While a recent meta-analysis did not find EMA assessment burden to influence compliance rates, the authors of this meta-analysis suggest study protocols may have been informed by piloting procedures with a focus on reducing participant burden (Jones et al., 2019). Future work should consider assessing motives repeatedly across a drinking event, to capture within-event variations in motives and how they predict acute changes in drinking behavior across an event. Yet, it will be important to test and consider what is the fewest assessments needed to examine changes in drinking motives to minimize participant burden.

While the present investigation has several strengths, including fine-grained assessments with high compliance and a novel research question, it is not without limitations. The sample was a non-clinical college student sample which may limit the generalizability of these findings to other populations. Additionally, the sample predominately identified as White, and thus findings may not be representative of a more diverse sample. All the data was collected during a single month in the spring semester. Given that drinking motives have been found to change across the academic year (Boyle et al., 2022), future work should extend this research to look at large college samples who are recruited and enrolled in the study throughout the academic year.

As mentioned above, some motives (“anxiety”, “depression”, “conformity”) had low endorsement rates. Large studies, longer assessment windows, and examination of specific populations that may be more likely to endorse these motives would provide additional insight. For example, clinical samples may be more likely to endorse event-level coping motives (“anxiety”, “depression”) and higher endorsement of these motives may be associated with outcomes. Also, specific motive domains were represented by single items. In an attempt to create shorter less burdensome questions/responses based on pilot participant feedback the wording of the item “to make day/night more fun” ended up being ambiguous as to whether this item was referring to a social motive or enhancement motive resulting in no clear social motive item. This highlights the challenge of capturing motivations often assessed with multi-item scales in an EMA context. There is always a balance between the number of assessments to ask in EMA reports and the burden being placed on participants. Future studies should ensure examination of all four factors of drinking motives: social, enhancement, conformity, and coping. Additionally, since the dynamic nature of drinking motives has been established, future research should prioritize a more comprehensive assessment of drinking motives (i.e., multiple assessment items per domain) at the event-level. Lastly, while we believe there is value in distinguishing between drinking events where no negative consequences occurred versus where any negative consequences occurred, future studies should examine samples where individual consequences are more highly endorsed. This would allow further examination of motives in relation to the number of both positive and negative consequences reported, and in relation to specific individual consequences. Also, one’s evaluation of how positive or negative specific consequences are can vary from day to day. As shown in a prior study using these data (Merrill et al., 2023), the specific negative consequences assessed in this study were evaluated on average as negative and the specific positive consequences were evaluated on average as positive. However, future work might examine event-level motives as a predictor of only those consequences evaluated as negative or positive on each drinking event.

Study findings highlight the importance of considering timing of motive assessments when designing event-level studies. Future work should consider investigating motives using frequent assessment, close in time to when drinking is occurring. Research examining event-level motives would also benefit from examining contextual factors that may influence and relate to reasons of use. For example, we would expect social reasons for drinking to be associated with being in a social context or anxiety-coping motives for drinking would occur when people report increased levels of anxiousness. Additionally, beginning a drinking event with elevated “high” motives was predictive of drinking more and experiencing negative consequences. Further work understanding drinking events characterized by antecedent “high” motives may be important for reducing heavy drinking and negative outcomes. A more complete assessment of drinking motives across drinking events will help us to better understand motives related to risky drinking and inform prevention and intervention development.

The momentary and dynamic nature of drinking motives may complicate the development and dissemination of drinking interventions that target drinking motivations. Often, substance use interventions are delivered distal in time to actual substance use events, missing the opportunity to target cognitions or behavioral skills when they are most salient and opportune for modification. Just-in-time adaptive interventions could be used to target motives of substance use in real-time. For example, we found “high” motive at the start of a drinking event was associated with increased drinking and experiencing both positive and negative consequences. Our data suggest that this could be considered a proximal motivational determinant of risk. Therefore, if participants report at the start of a drinking event they are drinking for enhancement motives, then a just-in-time adaptive intervention aimed at reducing the amount of use or preventing consequences could be provided during the drinking event. Since additional motives were found to be reported retrospectively the next morning, future interventions could include having young adults monitor and report their alcohol use and consequences each day and reflect on what their motives were for drinking. Additionally, the current study found positive consequences were associated with “fun”, “high”, and “conformity” drinking motives. If individuals endorse such motives and report risky alcohol behavior, interventions could highlight the concurrent risk of negative consequence, and provide or enhance the salience of alternative reinforcers. Feedback trends of one’s alcohol use and negative consequences in relation to motivations to drink presented by clinicians or via computer/smartphone delivery may lead to behavior change.

Public Health Significance:

This study highlights that associations between event-specific drinking motives and outcomes (use, consequences) depend on precisely when those motives are assessed. More motives were endorsed, and more effects were observed the next morning, relative to when drinking began, calling into question current practices of assessing motives once per day. Future studies should consider using ecological momentary assessment to measure drinking motives throughout the course of a drinking event to capture motives in real-time and such assessments may inform timing for event-level interventions focused on targeting motivational processes of drinking.