Daily Life Assessment of Seven Alcohol-Use-Disorder Symptoms

Abstract

Understanding how alcohol use disorder (AUD) symptoms unfold in daily life is key to improving assessment and intervention. This study examined the retrospective and prospective validity of repeated daily assessments of AUD symptoms compared to retrospective self-reports. A community sample of young adults (N=496) completed daily reports over an 8-week period assessing a subset of AUD symptoms: hazardous use, social/occupational problems, failure to fulfill obligations, craving, tolerance, larger/longer consumption, and time spent obtaining/using alcohol. Retrospective self-reports were collected at baseline and 6-month follow-up. Several symptoms (e.g., hazardous use, social/occupational problems, time spent) showed strong convergence between daily and baseline reports, while others (e.g., craving, tolerance) showed weaker associations. Daily symptom totals predicted 6-month retrospective AUD severity, particularly for symptoms with greater convergence. Daily measures of total AUD symptoms were associated with both baseline and follow-up AUD severity. Findings support the value of daily assessment and underscore discrepancies in retrospective recall.

General Scientific Summary

This study examined how well individuals’ retrospective reports of alcohol use disorder (AUD) symptoms correspond to their daily experiences of those symptoms. Results showed that some symptoms, such as hazardous use and social or occupational problems, were consistently reported across both retrospective and daily assessments. However, other symptoms, such as craving and tolerance, showed weaker correspondence, suggesting that some aspects of AUD may be more difficult to recall accurately and may require different assessment strategies.

Introduction

Many theories of alcohol use disorder (AUD) describe its development as a dynamic process unfolding across days, weeks, and months. For example, tolerance is theorized to emerge through repeated episodes of heavy drinking (Siegel, 1983), which subsequently increases withdrawal symptoms, leading to negative affect and craving, and ultimately contributing to lapses from abstinence to consumption, reinforcing the cycle (Koob & Le Moal, 1997). Different AUD symptoms are also hypothesized to develop on distinct timescales. Some symptoms, such as consuming larger quantities of alcohol or drinking for longer durations than intended, are expected to emerge gradually as individuals adapt to alcohol’s effects and require more to achieve the same level of intoxication (Koob & Le Moal, 2008; Lipscomb et al., 1980; Siegel, 1983). In contrast, other symptoms, such as social/interpersonal problems or recurrent hazardous use, are thought to arise during or immediately after alcohol consumption. Further compounding the relationship between AUD symptoms and time, AUD symptoms are believed to be shaped by emotional and cognitive factors (e.g., negative affect, PTSD symptoms) that fluctuate at momentary or daily timescales (Emery et al., 2014; Lane et al., 2019; Simons et al., 2014; Waddell et al., 2021; Koob et al., 2023; Sinha, 2008).

Despite these rich theoretical frameworks that articulate AUD’s development over time, the temporal dynamics of AUD development have largely remained untested at the appropriate timescale due to the absence of validated measures designed to capture AUD symptoms as they unfold in daily life. Without such assessments, key assumptions about the micro-temporal progression of AUD and its symptoms remain speculative rather than empirically tested.

Typical measures of AUD rely on retrospective reports of symptoms, such as in the widely used Alcohol Use Disorders Identification Test (AUDIT; Saunders et al., 1993). Retrospective AUD assessments require participants to estimate symptom frequency and intensity (e.g., Alaybek et al., 2022; Menon et al., 1995; Robinson & Clore, 2002), infer the specific behavioral antecedent (e.g., Martin et al., 2014; Nisbett & Wilson, 1977), reflect on their intentions (e.g., Chung & Martin, 2005), or recall changes over time (e.g., Schacter et al., 2011; Shiffman et al., 1997). These assessment features introduce a number of potential limitations and biases (Alaybek et al., 2022).

First, memory recall plays an important role in estimating symptom frequency, and bias and cognitive distortion that may lead to incomplete or skewed representations of alcohol consumption and AUD symptoms (Robinson & Clore, 2002; Shiffman, Stone, & Hufford, 2008a). Second, attributing behaviors solely to alcohol use might oversimplify complex decision-making processes, potentially overlooking contributing factors influencing behaviors (Caetano & Babor, 2006; Nisbett & Wilson, 1977). For instance, certain consequences of alcohol use (i.e., getting into fights) have been critiqued because alcohol is often assumed to be the root cause of the behavior, whereas premorbid dispositional tendencies (e.g., impulsivity, aggression) may provoke that behavior regardless of extent of alcohol consumption (Martin et al., 2011, 2014; Martin & Sher, 2018). Third, retrospecting about one’s intentions to drink, such as how much and for how long, involves evaluating pre-alcohol mental states, which are subject to changes that are influenced by various factors, complicating assessment accuracy (Robinson & Clore, 2002). Fourth, retrospective assessments also include evaluating changes in AUD symptoms. For instance, reporting tolerance involves a subjective judgment that is susceptible to memory fluidity and personal interpretations, potentially resulting in inconsistencies in reported changes over time (Boness & Sher, 2020; Chung & Martin, 2005; Slade et al., 2013).

Thus, retrospective assessments cannot fully encapsulate the dynamic nature of AUD symptoms in daily life (Shiffman, 2014; Shiffman, Stone, & Hufford, 2008b). Employing shorter (e.g., daily, weekly, monthly) and even momentary (e.g., in-the-moment craving and intentions to drink) assessment intervals may offer valuable insights into the experiences of people with AUD, allowing researchers to directly model the dynamic nature of AUD symptoms at the time scales that they are experienced (Robinson & Clore, 2002). Repeated daily assessments capturing behaviors and experiences as they naturally occur in individuals’ everyday lives holds promise in capturing alcohol use behaviors at shorter timescales and likely aligns more closely with how people experience AUD symptoms in their daily lives, allowing researchers to better understand the immediate contexts in which AUD symptoms manifest.

Daily (or more frequent) assessments can capture the associations between the antecedents and consequences of symptoms to test theories of AUD development (Miranda et al., 2014; Ray et al., 2010) in daily life (Shiffman, 2014; Stone et al., 2007; Todd et al., 2005) while minimizing biases inherent to retrospective recall (Shiffman et al., 1997; Shiffman, Stone, Hufford, et al., 2008; Wray et al., 2014). In turn, because they separate the experience of the symptom with its time course (e.g. frequency, change, or recurrence), daily assessments may facilitate more precise tests of theories about the development of AUD in terms of the day-to-day or week-to-week fluctuations or changes in symptoms, and how the experience of those symptoms is related to cognitive, emotional, social, and environmental factors. These factors are integral to understanding the triggers and mechanisms that underlie not only core physiological AUD symptoms, such as tolerance and withdrawal, but also contextually influenced features of AUD, such as impaired control, hazardous use, and continued drinking despite negative consequences.

Some prior research has shown that specific AUD symptoms can be measured using daily assessments. For example, measures of daily alcohol related consequences (Lee et al., 2017) align with AUD symptoms such as hazardous use (“I hurt or injured myself by accident”), social/interpersonal problems (“I got into a fight or argument” or “I embarrassed myself”), and failure to fulfill role obligations (“I was unable to do work/schoolwork”). Drinking in larger amounts or for longer periods of time than intended has been measured in daily life studies (Fairlie et al., 2019; Stevens et al., 2022), as have cravings for alcohol (Waddell et al., 2021). Only two prior samples have measured other symptoms of AUD—these studies used in-situ checklists of “acute dependence symptoms,” including drinking in larger amounts or for longer periods of time than intended, withdrawal, tolerance, and cutting down or quitting alcohol (Simons et al., 2010, 2014b, 2015). Across studies, evidence suggests that AUD symptoms measured via daily or more frequent assessments are associated with the quantity of drinks consumed (Dvorak et al., 2014; Simons et al., 2010, 2014b) and level of intoxication (Simons et al., 2010, 2014); however, none of these studies examined how these daily experiences correspond to retrospective reports of the same symptoms. Therefore, it remains unclear how accurately retrospective self-reports of AUD symptoms reflect individuals’ daily experiences with the same symptoms (or alcohol related consequences).

In the present study, we tested how well daily assessments of specific AUD symptoms reflect retrospective reports of AUD symptoms across two distinct timescales. First, participants completed a retrospective assessment of past-year AUD symptoms at baseline. We then conducted repeated daily assessments over a two-month period to examine the retrospective validity of these daily AUD symptom reports—that is, the extent to which daily symptom experiences align with individuals’ retrospective reports at baseline. Second, we conducted a six-month follow-up assessment that included another set of retrospective AUD symptom reports, allowing us to assess the prospective validity of the daily reports—specifically, whether individuals’ aggregated real-time symptom experiences predicted how they later recalled and reported those symptoms after six months. To our knowledge, this is the first study to directly compare participants’ retrospective self-reports of AUD symptoms with their daily symptom experiences and to examine how those daily reports relate to a subsequent retrospective assessment, thereby evaluating how well retrospective measures reflect both past and future real-world symptom expression. The four preregistered aims (https://osf.io/6ebg2) which we organize in this paper by the type of validity assessed are as follows:

Retrospective Validity Aims

Aim 1.

Whether baseline retrospective reports of past-year AUD symptoms predict the daily occurrence of those same symptoms during a two-month daily assessment period—for example, whether individuals who retrospectively report higher levels of alcohol craving also report higher craving levels during daily assessments.

Aim 3.

Whether baseline retrospective measures of overall AUD severity (i.e., AUDIT and total DSM-5 symptom count) predict overall daily AUD severity across symptoms.

Prospective Validity Aims

Aim 2.

Whether person-level averages of daily AUD symptoms over the daily assessment period predict symptom-specific retrospective reports at a six-month follow-up.

Aim 4.

Whether overall daily AUD symptom severity predicts retrospective AUD severity at 6-month follow-up.

Transparency and Openness

Preregistration.

All study methods and analytic aims were preregistered and are publicly available at https://osf.io/6ebg2. Analyses that were not included in the preregistration are clearly labeled as “exploratory analyses” in the manuscript.

Data, Materials, Code, and Online Resources:

The data, materials, and code for this study are not currently available in a public repository. Researchers interested in accessing de-identified data or study materials for replication or other scholarly purposes may contact the corresponding author, and reasonable requests will be reviewed in accordance with institutional data-sharing policies.

Reporting:

We report how we determined our sample size, all data exclusions, all manipulations, and all measures in the study.

Ethical Approval:

All study procedures were conducted online and approved by the local Institutional Review Board (IRB).

Methods

Sample.

Participants were recruited for a larger study on the development of alcohol and cannabis use problems during young adulthood (King et al., 2024). Participants were young adults at baseline (N = 496, age 18 – 22, Mage = 20.3, SD = 1.3). Participants were recruited from King, Pierce, and Snohomish Counties in Washington State from both college and non-college sources to ensure a representative sample of young adults in Washington State. We recruited using internet (Facebook, Instagram, TikTok, YouTube, Twitter, Craigslist, Reddit, and emails to university registrar lists and high school list-servs) and non-internet (newspaper advertisements and flyers) sources. Participants were required to be between the ages of 18 and 22 at screening, own a smartphone, be fluent in English, and report drinking alcohol or using cannabis “about once per week” or more over the past three months. Participants were excluded if they were not fluent in English or if they moved to the United States after age 12.

The sample consisted of 45% cisgender women, 42% cisgender men, 8.5% nonbinary, genderqueer, or gender nonconforming individuals, 4.0% transgender individuals, and 0.2% who identified as nongendered. At the time of recruitment, approximately 67% of participants were attending a four-year college. The racial and ethnic composition of the sample broadly reflected Washington census data from the counties where participants were recruited. Participants identified with a diverse range of racial and ethnic identities: 54% identified as non-Hispanic White, 28.5% as Asian, 6.6% as African American/Black, 8.4% as Hispanic/Latino, and 22.7% endorsed more than one racial/ethnic identity. Most participants identified as heterosexual (52%), with the remaining participants either identifying as LGBQ+ (47.6%) or declining to respond (n = 2).

Procedures.

Participants first completed an online screening survey that assessed basic demographic information, contact information, and past month substance use as well as other health behaviors to obscure the inclusion criteria. Eligible participants then completed a virtual training session on the daily assessment study procedures with research assistants (RAs), followed by a 20–30-minute online survey that included questions about their drinking behaviors, past-year AUD symptoms, and a battery of personality and mood measures. Over an 8-week period, during the social weekend (Thursday to Sunday), participants received text messages containing links to brief (5-minute) repeated assessment surveys, five times per day. On Monday mornings, an additional survey was sent out to capture behavior from Sunday nights. These repeated assessments were distributed randomly across 5 time blocks spanning from 9:00 AM to 11:00 PM (9–11 AM, 12–2 PM, 3–5 PM, 6–8 PM, and 9–11 PM). Although prompts were sent during these two-hour windows, participants had up to one hour to respond, effectively creating five three-hour assessment windows. Initially, participants received a single reminder at the 30-minute mark if they had not completed their survey. However, after the first 60 participants, this protocol was revised to send reminders at 20- and 40-minute intervals to enhance response rates. Compensation for participation included $50 for completing the baseline survey, along with $1 per completed daily assessment. Additionally, participants were eligible for a $5 bonus upon completing 80% (i.e., 17 out of 21) of the daily assessments for a given social weekend, totaling a possible compensation of $258.

All participants were contacted 6 months after the date of their completion of baseline study participation and invited to complete a brief longitudinal follow-up survey. Participants completed a 20–30-minute online survey, within which they answered questions about their drinking behaviors along with a battery of personality and mood assessments. Participants were compensated $40 for completion of the follow-up survey. Out of the 411 baseline participants included in the current analyses, 78% (n = 320) completed the 6-month follow-up survey.

Sample size justification.

Our sample size was determined by the parent study (King et al., 2024) design rather than power calculations specific to our research questions. Given this fixed sample size, we conducted a sensitivity power analysis to determine the minimum detectable effect sizes for our analyses. Given an expected 80% compliance rate in the daily assessment surveys, we estimated that participants would complete an average of 102.4 daily reports over 25.6 days, resulting in approximately 51,200 observations across the sample. To evaluate the smallest detectable effect size, we conducted a sensitivity analysis, accounting for within-person variability in daily responses. Based on our pilot data, we assumed a maximum intraclass correlation coefficient (ICC) of 0.40 for our multilevel models. Under these conditions, our study was sufficiently powered to detect small effects in within-person analyses (e.g., standardized β = 0.013) and cross-level interactions (standardized β = 0.03).

Measures

Daily Assessment Measures.

We operationalized 7 daily AUD symptoms using a variety of measures related to drinking behaviors. Notably, we did not have measures that assessed AUD symptoms corresponding to drinking despite physical/psychological problems, giving up important activities due to drinking, desire to quit/cut down on drinking, or withdrawal.

Hazardous use.

At the morning assessment (or the 2^nd assessment of the day if they missed the morning assessment), participants reported which of the consequences they experienced as a result of their alcohol or cannabis use the previous day (“Did any of the following things happen to you as a result of your use yesterday?”; 1 = yes, 0 = no) (Lee et al., 2017), indicated they did not use, or that they did use but experienced no consequences of use. Because our pilot data indicated that participants frequently used both alcohol and cannabis, we did not ask participants to separately attribute consequences they experienced to a specific substance. We used 2 items (“I hurt or injured myself by accident” and “I couldn’t remember what I did”) from a measure of alcohol related consequences adapted from existing measures of AUD and extensively validated in young adults (Lee et al., 2017). If a participant endorsed either item, we coded it as 1, indicating a positive hazardous use symptom.

Social/interpersonal harm.

We used four items from Lee et al. (2017): “I did or said something I regret”, “I did or said something that embarrassed me”, “I was rude or obnoxious”, and “I got into a fight/argument” (1 = yes; 0 = no). If a participant endorsed any of these items, we coded it as 1, indicating a positive social/interpersonal consequence symptom.

Failure to fulfill role obligations.

We used one item from Lee et al. (2017): “I was unable to do work/schoolwork” (1 = yes; 0 = no).

Time spent (obtaining alcohol, drinking, or getting over the effects of alcohol).

This symptom was assessed through two methods. First, we used two items from Lee et al. (2017): “I had a hangover” and “I experienced nausea or vomiting,” (1 = yes; 0 = no) to encompass broader aspects of time allocation associated with alcohol consumption. We operationalized these items as time spent rather than withdrawal because we measured them only in the context of drinking episodes (e.g., the acute effects of alcohol wearing off the morning after drinking; Prat et al., 2009), whereas withdrawal is operationalized as symptoms that arise after people cut down or quit drinking. If a participant endorsed either item, we coded it as 1, indicating a positive symptom. Second, we used a single item where participants reported how many hours they drank alcohol the previous day, with response options ranging from 0 to 24. We rescaled the response range of the item from 0 to 1 to standardize the measurement.

Tolerance.

Tolerance was assessed using two methods. First, participants who reported consuming 20 or more drinks were classified as having a positive symptom (coded as 1), whereas those who consumed fewer were coded as 0 (negative symptom), consistent with prior research (e.g., Grant et al., 2015; Ruan et al., 2008; Hasin et al., 2015). Second, we calculated a ratio reflecting participants’ reported level of intoxication relative to the number of drinks consumed. During the morning report, participants indicated the number of drinks consumed the previous night using a visual analogue scale ranging from 0 to 30 or more drinks, alongside a standard alcoholic beverage definition. They also rated their level of intoxication on a visual analogue scale from 0 (‘Not at all/I didn’t consume alcohol’) to 100 (‘Very high’). To standardize this measure, we rescaled the intoxication-to-drinks-consumed ratio to range from 0 to 1, where values approaching 1 indicate higher tolerance and values nearing 0 suggest minimal tolerance.

Larger amounts/Longer periods than intended.

We measured drinking in larger amounts or for longer periods than intended with two items. During the afternoon assessment, participants indicated their intended drinking amount for later that day using a single item: “How many drinks do you intend to have later today?” Responses were provided via a slider bar ranging from 0 to “15 or more.” To identify this symptom, we calculated the discrepancy between participants’ intended and actual alcohol consumption reported the following day. Positive discrepancies—where participants drank more than planned—were coded as 1, indicating the presence of the symptom. Non-positive discrepancies (0 or negative values) were coded as 0, indicating the absence of the symptom.

Craving.

We measured alcohol craving using four items adapted from the Alcohol Urge Questionnaire (Bohn et al., 1995). Two items were randomly administered at each survey. For two items (“How much have you wanted to drink alcohol?”, “How much have you been thinking about drinking?”), participants responded on a 100-point slider bar with 0 = “Not at all” and 100 = “Very much”. For the other two items (“I felt like I could really use a drink.”, and “If I’d had some alcohol, I would probably have drunk it.”), participants responded on a 100-point slider bar with 0 = “Strongly disagree” and 100 = “Strongly agree”. To standardize the measurement of craving, we scaled the responses from 0 to 1. A score of 1 indicates the highest level of craving, while a score of 0 reflects no craving.

Baseline and Follow-Up Assessment (Retrospective Recall) Measures.

Alcohol Use Disorder Diagnostic Criteria.

We used 33 severity graded items as a baseline AUD assessment measure (Boness et al., 2019) designed to capture the 11 DSM-5 criteria (American Psychiatric Association, 2013) over the past 12 months at baseline. At the six month follow-up, we changed the retrospection period to the past six months. Each criterion was assessed with 3 items, allowing for the classification of symptom severity as low, moderate, or high. Response options ranged from 0 (indicating no experience) to 4 (indicating experiencing the behavior 4 or more times). For the present analyses, we recoded responses to create a binary variable: a score of 1 for items experienced 3 or more times, and a score of 0 for symptoms experienced fewer than 3 times. Next, for each of the 11 DSM-5 AUD criteria, we grouped the relevant three items and coded the criterion as 1 if any of the three items were endorsed (≥1), and 0 if none were endorsed. To create a total AUD criterion count score (range: 0–11), we summed the binary scores across all 11 criteria, producing an overall indicator of AUD severity based on the total number of DSM-5 criteria endorsed. The scores were then standardized to have a mean of 0 and a standard deviation of 1. This scale demonstrated excellent internal consistency in the present sample (Cronbach’s α = 0.93, 95% CI [0.92, 0.94])

Alcohol Use Disorders Identification Test (AUDIT).

The AUDIT (Saunders et al., 1993) was also employed as it is one of the more widely used screening tools designed to assess past 12-month consumption and alcohol-related problems. For the six-month follow-up, we changed the retrospection period to the past six months. The AUDIT consists of 10 questions assessing four domains of AUD—Alcohol Consumption, Drinking Behavior, Adverse Psychological Reactions, and Alcohol-Related Problems—using a scale from 0 (“Never”) to 4 (“Daily or almost daily”). This measure was repeated at the follow up with a prompt asking participants to report about their experiences in the past 6 months. Total scores for this measure (range 0–40) were standardized to have a mean of 0 and a standard deviation of 1. Unlike the self-reported Alcohol Use Disorder Diagnostic Criteria (above), we chose not to categorize individual AUDIT items under the 11 DSM-5 AUD symptoms, as some items do not directly align with the DSM-5 criteria. Instead, we used the total AUDIT score to examine its overall association with AUD symptoms measured via daily assessments. The AUDIT demonstrated good internal consistency in the present sample (Cronbach’s α = 0.76, 95% CI [0.72, 0.80]).

Data Analytical Plans

To examine the retrospective validity of daily AUD symptom reports, we aimed to examine how well baseline self-reports of each past-year AUD symptom predict the daily occurrence of that specific symptom in the two months after the baseline assessment (Aim 1) and how well baseline retrospective self-reports of total past-year AUD symptoms predicted overall AUD severity from the two-month daily assessment (Aim 3). To assess prospective validity, we tested whether person-level averages of AUD symptoms from the two-month daily assessment predicted symptom-specific retrospective reports of past 6-month AUD symptoms, as assessed at the 6-month follow-up assessment (Aim 2), and whether overall AUD symptom severity from the two-month daily assessment predicted total retrospective AUD symptom scores at the 6-month follow-up (Aim 4) (see Table 1). Data processing and analysis were conducted in R (R Development Core Team, 2016), a flexible open source data analytic software, and models were fit with the lme4 package (Bates et al., 2015).

Table 1.

Analysis Aims

Aim 1. Baseline retrospective reports of past-year AUD symptoms predicting daily AUD symptoms over a two-month period
Predictors	Outcomes
Self-Report Hazardous Use	Daily Hazardous Use
Self-Report Social/Occupational Problems	Daily Social/Occupational Problems
Self-Report Failure to Fulfill Obligations	Daily Failure to Fulfill Obligations
Self-Report Craving	Daily Craving
Self-Report Time Spent	Daily Time Spent-1 (Lee et al., 2017)
Self-Report Time Spent	Daily Time Spent-2 (#hours)
Self-Report Tolerance	Daily Tolerance-1 (20+ drinks)
Self-Report Tolerance	Daily Tolerance-2 (15+ drinks)
Self-Report Tolerance	Daily Tolerance-3 (ratio)
Self-Report Larger Amounts	Daily Larger Amounts-1 (1+ drink)
Self-Report Larger Amounts	Daily Larger Amounts-2 (2+ drink)
Self-Report Larger Amounts	Daily Larger Amounts-3 (3+ drinks)
Aim 2. Daily AUD symptoms predicting AUD symptoms at 6-month follow up
Predictors	Outcomes
Person-Mean Daily Hazardous Use	Self-Report Hazardous Use
Person-Mean Daily Social/Occupational Problem	Self-Report Social/Occupational
Person-Mean Daily Failure to Fulfill Obligations	Self-Report Failure to Fulfill Obligations
Person-Mean Daily Craving	Self-Report Craving
Person-Mean Daily Time Spent-1	Self-Report Time Spent
Person-Mean Daily Time Spent-2	Self-Report Time Spent
Person-Mean Daily Tolerance-1	Self-Report Tolerance
Person-Mean Daily Tolerance-2	Self-Report Tolerance
Person-Mean Daily Tolerance-3 (ratio)	Self-Report Tolerance
Person-Mean Daily Larger Amounts-1	Self-Report Larger Amounts
Person-Mean Daily Larger Amounts-2	Self-Report Larger Amounts
Person-Mean Daily Larger Amounts-3	Self-Report Larger Amounts
Aim 3. Baseline AUD Severity Predicting Daily Measures of AUD Severity
Predictors	Outcomes
AUD total score	Means of all daily symptoms
AUDIT total score	Means of all daily symptoms
Aim 4. Daily AUD severity predicting AUD severity at 6-month follow-up
Predictors	Outcomes
Person means all daily symptoms	6-month AUD total scores
Person means all daily symptoms	AUDIT total scores

First, we used logistic generalized linear mixed models (GLMMs) and linear mixed models to examine the association between baseline retrospective reports of AUD symptoms and daily reports of the same symptoms, measured via daily assessment. For binary outcomes (i.e., daily hazardous use, social/occupational problems, failure to fulfill obligations; see Table 2), we applied logistic GLMMs, treating daily symptom reports as repeated observations nested within individuals. For continuous outcomes (i.e., daily craving, hours spent drinking, tolerance, daily number of drinks consumed), we used linear mixed models. In both cases, fixed effects included the baseline self-reports of each AUD symptom, while random intercepts accounted for between-person variability in overall daily assessment symptom levels. All models accounted for age, gender, college student status, and day of the week. Age was grand mean centered and gender and college/non-college status, binary covariates, were dummy coded (0/1), with male and college as the reference groups, respectively. At the daily level, we included time effects (i.e., week of study) as covariates.

Table 2.

Descriptive Statistics (final sample size included in the analyses: n=386)

Variables
	M (SD)	Range
Mean Age (SD)	20.36 (1.31)	[18–23]
Racial or ethnic identity	n	%
White	200	51.81
Asian	60	15.54
African American/Black	17	4.40
Middle Eastern	6	1.55
American Indian	17	4.40
Hispanic/Latino	16	4.15
Multi-racial	66	17.10
Other	4	1.05
Gender	N (%)
Male	164 (42.29)
Female	176 (45.60)
Other	46 (11.92)
College Student	N (%)
Yes	291 (75.29)
No	95 (24.61)
Baseline AUD Measures	M (SD)	Range
Mean #of AUD symptoms endorsed (SD), [range]	1.48 (1.75),	[0–11]
Mean AUDIT (SD), [range]	7.90 (5.53),	[0–37]
6-Month Follow Up AUD Measures	M (SD)	Range
Mean #of AUD symptoms endorsed (SD), [range]	1.50 (2.09),	[0–11]
Mean AUDIT (SD), [range]	6.41 (4.34),	[0–22]
Daily AUD Measures	n* (%)	ICC
Daily Hazardous Use (Lee et al., 2017)	152 (4.55)	0.10
Daily Social/Occupational Problems (Lee et al., 2017)	315 (9.43)	0.14
Daily Failure to Fulfill Obligations (Lee et al., 2017)	110 (3.29)	0.07
Daily Time Spent-1 (Lee et al., 2017)	549 (16.44)	0.14
Daily Tolerance-1 (20+ drinks)	18 (0.54)	0.11
Daily Tolerance-2 (15+ drinks)	51 (1.53)	0.18
Daily Larger Amounts-1 (1+ drinks)	3213 (96.23)	0.08
Daily Larger Amounts-2 (2+ drinks)	1774 (53.13)	0.23
Daily Larger Amounts-3 (3+ drinks)	1329 (39.80)	0.28
	M (SD), Range	ICC
Craving	0.15 (0.25), [0–1]	0.49
Daily Tolerance-3 (ratio)	0.01 (0.04), [0–1]	0.25
Daily Time Spent-2 (#hours)	0.16 (0.13), [0–l]	0.29

^*Number of cases. There were 3,339 drinking days in total.

Second, we employed generalized linear models to examine the relationship between person-means of daily AUD symptoms (averaged across daily reports per participant) and 6-month follow-up reports of the same symptoms, derived from both the alcohol use disorder diagnostic criteria and AUDIT. These models treated person-means as fixed effects, assessing how well participants’ average daily AUD symptoms predicted their corresponding AUD symptom scores at the 6-month follow-up. For instance, in a model assessing hazardous use, the key coefficient of interest quantifies the association between the person-mean of daily hazardous use and its corresponding 6-month follow-up score. These models included fixed effects for age, gender, and college student status.

In addition to symptom-level analyses, we also ran models examining overall AUD severity by using total scores of retrospective measures and total person-means of daily symptoms. For daily assessment measures, we computed a composite daily AUD symptom score by summing the person-means of each symptom. For retrospective measures, we used two different total scores: (1) the total alcohol use disorder symptom count (0–11) and (2) the total AUDIT score (0–40). To examine the association between baseline retrospective AUD severity and daily reported symptoms, we estimated linear mixed-effects models predicting total daily symptoms from each of the three baseline retrospective AUD scores. These models included fixed effects for baseline total AUD scores, while random intercepts for participants accounted for between-person variability in overall daily AUD symptoms. All models accounted for age, gender, college student status, and day of the week. To assess the prospective relationship between total daily symptoms and retrospective AUD severity at the 6-month follow-up, we estimated generalized linear models with total AUD scores at follow-up as outcomes and mean daily symptoms as a predictor. These models included fixed effects for age, gender, and college student status.

Missing Data

We focused exclusively on AUD symptoms reported on days when participants consumed alcohol, as all symptoms—except craving—were only assessed on those days. Due to the study design, there were missing observations for drinking intentions. Specifically, morning reports of drinking were collected for Wednesday through Sunday evenings (e.g., on Thursday through Monday mornings), but pre-episode drinking intentions were only assessed Thursday through Sunday for those potential drinking days. As a result, we obtained complete AUD symptom data for 3,339 of 4,285 (78%) possible drinking days. Despite these missing data points, our analyses leveraged all drinking days by modeling symptoms at the item level, accounting for the nested structure of daily symptoms within participants.

Results

Descriptive Information

Table 2 provides descriptive statistics for all key variables. Across the sample (total n = 496), the overall compliance rate with the daily assessments was 61.89%. Further, 78% of participants (n = 386) reported at least one drinking day during the daily assessment period with a total of 3,339 alcohol use days. Among these drinking days, 2,689 days (80.54%) also involved cannabis use. See Table 2 for demographic information for this final analytic sample. The average number of drinks per day on drinking days was 4.03 (SD= 3.45), with an average intoxication score of 37.66 (SD=26.92). Approximately half of the sample met criteria for past-year AUD, with 37% meeting criteria for mild AUD (2–3 symptoms), 10% meeting criteria for moderate (4–5 symptoms), and 5% meeting criteria for severe (6+ symptoms) AUD. To evaluate the reliability of daily symptom reports, we calculated ICCs for each daily assessment variable. ICCs for daily assessment variables ranged from 0.07 (Daily Failure to Fulfill Obligations) to 0.49 (Craving), with an average of 0.21.

Retrospective Validity Analysis

Aim 1: Predicting Daily AUD Symptoms from Retrospective Baseline Measures.

Past-year retrospective self-reports of AUD symptoms were associated with their daily occurrence across most models (Table 3 presents a summary of the results, with full model details available in the supplemental materials #2). At baseline, reporting higher past-year levels of hazardous use (OR = 1.410, β = .344, p = 0.024), social/interpersonal harm (OR = 1.269, β = 0.238, p = 0.023), failure to fulfill role obligations (OR = 1.616, β = 0.480, p = 0.001), and time spent drinking or recovering (Lee et al., 2017: OR = 1.352, β = 0.302, p < 0.001; #hours: β = 0.009, p = 0.024) were all associated with the likelihood of reporting those same symptoms during the next two months of daily assessments. In contrast, craving (β = 0.003, p = 0.681), tolerance (20+ drinks: OR = 1.412, β = 0.345, p = 0.661; ratio: β < 0.001, p = 0.887), and drinking in larger amounts or for longer durations than intended (OR = 1.098, β = 0.094, p = 0.494) were not significantly associated with their daily indicators. See Table 3 for a summary of these results and Supplemental Material 1 for the complete set of findings.

Table 3.

Summary of Results

Aim 1: Baseline Self-Reports of Each Symptom Predicting Daily Reports
X	Y (Daily measures)	β	SE	p
Self-Report Hazardous Use	Hazardous Use (Lee et al., 2017)	0.344	0.153	0.024	*
Self-Report Social/Occupational Problems	Social/Occupational Problems (Lee et al., 2017)	0.238	0.105	0.023	*
Self-Report Failure to Fulfill Obligations	Failure to Fulfdl Obligations (Lee et al., 2017)	0.480	0.144	0.001	***
Self-Report Craving	Daily Craving	0.003	0.008	0.681
Self-Report Time Spent	Time Spent-1 (Lee et al., 2017)	0.302	0.075	<0.001	***
Self-Report Time Spent	Time Spent-2 (#hours)	0.009	0.004	0.024	*
Self-Report Tolerance	Tolerance-1 (20+ drinks)	0.345	0.787	0.661
Self-Report Tolerance	Tolerance-2 (15+ drinks)	0.429	0.379	0.257
Self-Report Tolerance	Tolerance-3 (ratio)	0.000	0.001	0.887
Self-Report Larger Amounts	Larger Amounts-1 (1+drinks)	0.094	0.137	0.494
Self-Report Larger Amounts	Larger Amounts-2 (2+ drinks)	0.270	0.073	<0.001	***
Self-Report Larger Amounts	Larger Amounts-3 (3+ drinks)	0.358	0.084	<0.001	***
Aim 2: Person Means of Daily Reports Predicting Self-Reports at 6-month Follow-up
X (Person-Means of Daily Measures)	Y	β	SE	p
Hazardous Use (Lee et al., 2017)	Self-Report Hazardous Use	0.108	0.063	0.086
Social/Occupational Problems (Lee et al., 2017)	Self-Report Social/Occupational	0.207	0.057	<0.001	***
Failure to Fulfdl Obligations (Lee et al., 2017)	Self-Report Failure Obligations	0.058	0.076	0.451
Daily Craving	Self-Report Craving	−0.006	0.062	0.929
Time Spent-1 (Lee et al., 2017)	Self-Report Time Spent	0.180	0.066	0.007	**
Time Spent-2 (#hours)	Self-Report Time Spent	0.157	0.071	0.028	*
Tolerance-1 (20+ drinks)	Self-Report Tolerance	−0.034	0.062	0.583
Tolerance-2 (15+ drinks)	Self-Report Tolerance	−0.014	0.077	0.851
Tolerance-3 (ratio)	Self-Report Tolerance	0.030	0.063	0.630
Larger Amounts-1 (1+ drinks)	Self-Report Larger Amounts	−0.010	0.062	0.867
Larger Amounts-2 (2+ drinks)	Self-Report Larger Amounts	0.194	0.063	0.002	**
Larger Amounts-3 (3+ drinks)	Self-Report Larger Amounts	0.156	0.064	0.016	*

Note. p < .05 (*), p < .01 (**), p < .001 (***). Asterisks indicate statistical significance thresholds.

Aim 3: Predicting Overall AUD Severity from Baseline Measures.

Next, we examined how well retrospective reports of past 12-month AUDIT scores and AUD symptom count at baseline predicted aggregated daily AUD symptoms measured during the daily assessment period (see Supplemental Material #1). Results indicated that both the AUDIT (β = 0.038, p < 0.001), and AUD Symptom count (β = 0.018, p < 0.001) predicted the occurrence of daily AUD symptoms. More specifically, a 1 SD difference in AUDIT scores was associated with a 0.038 SD increase in the experience of daily AUD symptoms, while a 1 SD difference in AUD symptom count was associated with a 0.018 SD difference in the experience of daily AUD symptoms.

Prospective Validity Analysis

Aim 2: Predicting Retrospective AUD Symptoms at 6-Month Follow-Up from Daily Experiences.

Daily experiences of AUD symptoms across the two-month daily assessment period were associated with retrospective reports of the same symptoms at the 6-month follow-up across several key domains, after accounting for demographic covariates. People who reported more symptoms of social/interpersonal problems (β = 0.207, p < 0.001) and time spent obtaining or using alcohol (Lee et al., 2017: β = 0.180, p = 0.007; #hours: β = 0.157, p = 0.028) across the daily assessment period also reported more of those same symptoms at the 6 month follow up retrospective reports. In contrast, failure to fulfill role obligations (β = 0.058, p = 0.451), hazardous use (β = 0.108, p = 0.086), craving (β = −0.006, p = 0.929), daily reports of tolerance (β = −0.034, p = 0.583; β = 0.030, p = 0.630), and drinking larger amounts or for longer durations (β = −0.010, p = 0.867) did not predict these symptoms at the 6-month follow-up. See Table 3 for a summary of these results and Supplemental Material 2 for the complete set of findings.

Aim 4: Predicting Overall AUD Severity at 6-Month Follow-Up from Daily Measures.

Lastly, we tested how well aggregated person-means of daily AUD symptoms predicted total AUDIT and AUD symptom count at the 6-month follow-up. Results showed significant associations between daily symptom reports and subsequent 6-month AUDIT (β = 0.584, p < 0.001) and AUD symptom count (β = 0.322, p < 0.001) (see Supplemental Material #2). A 1 SD increase in daily AUD symptoms was associated with a 0.584 SD increase in AUDIT scores and a 0.322 SD increase in AUD symptom count at follow-up.

Exploratory Analyses (not pre-registered)

In the pre-registered analyses, tolerance was operationalized using the 20+ drinks criterion (Boness et al., 2019); however, the number of positive cases was low (n = 18). As an exploratory step, we adjusted the threshold for the tolerance symptom to include participants reporting 15 or more drinks on a single occasion. Results from this adjusted model still demonstrated an association between retrospective reports of tolerance and daily tolerance-related behaviors (OR = 1.535, β = 0.429, p = 0.257). The association between these daily reports and their self-reports at 6-month follow-up was not significant (β = −0.014, p = 0.851).

Additional exploratory analyses were conducted to examine the symptom of drinking in larger amounts than intended, using thresholds of 2+ and 3+ additional drinks earlier in the day. These analyses were motivated by the concern that a 1+ drinks threshold—used in our pre-registered analyses—may be relatively common in social or recreational contexts in young adult populations, where modest deviations from intended drinking may be perceived as more normative. Results indicated that retrospective reports of drinking in larger amounts were significantly associated with daily instances of this behavior at both the 2+ drinks threshold (OR = 1.309, β = 0.270, p < 0.001) and the 3+ drinks threshold (OR = 1.430, β = 0.358, p < 0.001). significantly predicted 6-month follow-up self-reports of this symptom: β = 0.194, p = 0.002 for 2+ drinks; and β = 0.156, p = 0.016 for 3+ drinks.

Discussion

This study sought to evaluate the predictive relationship between retrospective self-reports and daily experiences of AUD symptoms. Although retrospective measures have long been the standard for assessing AUD, they are unable to capture the temporal processes that are thought to underly the development of AUD symptoms in daily life. Prior research has demonstrated that certain AUD symptoms—such as hazardous use, cravings, and social consequences—are measurable in daily life through repeated daily or momentary surveys (Simons et al., 2010, 2014; Lee et al., 2017), and these and other studies showed that daily alcohol consumption patterns predicted the experience of certain AUD symptoms (e.g., cravings; Waddell et al., 2021) or clusters of symptoms (e.g. acute dependence symptoms; Simons et al., 2010; 2014, and Dvorak, 2014). No prior work had tested the degree to which, at the symptom level, people’s daily experiences of AUD symptoms correspond with the retrospective report of the experience of those same symptoms.

Our findings extend prior research by demonstrating that retrospective self-reports of AUD symptoms generally align with their daily occurrences, though the strength of these associations varies across symptoms and how they are characterized. Moreover, daily experiences of some AUD symptoms (e.g., social/occupational problems, time spent obtaining/consuming alcohol, drinking longer and in larger amounts) significantly predicted their retrospective reports at a 6-month follow-up, whereas others (e.g., tolerance, craving) exhibited weak or inconsistent associations. These results underscore the need to refine measurement approaches to better capture the dynamic nature of AUD symptoms in both daily life and long-term assessments. This is particularly important in young adult populations, where drinking patterns may be more variable, context-dependent, and developmentally distinct from those observed in clinical AUD samples, who are often older and may exhibit different symptom profiles (Chen & Jacobson, 2012; Schulenberg & Maggs, 2002; Windle & Windle, 2012).

Of note, the current study was not designed to develop a comprehensive or definitive daily AUD symptom measure, but rather to examine the validity of a subset of symptoms that could be reasonably approximated using the available data. This work serves as an initial step in evaluating the psychometric properties of repeated assessment-based AUD symptom indicators, with the goal of informing future efforts to improve the ecological operationalization of DSM-5 criteria. In this regard, our focus was on testing the extent to which these approximated daily symptom measures align with more established retrospective reports, rather than assuming one-to-one correspondence.

Baseline Retrospective Measures and Daily Assessments of AUD Symptoms

Our findings indicate that most baseline retrospective reports of AUD symptoms—such as hazardous use, social/interpersonal problems, failures to fulfill role/obligations, and time spent obtaining or using alcohol—are significantly associated with daily experiences of these symptoms. As such, daily assessments can reliably capture AUD symptoms that are traditionally measured using retrospective self-report measures. For instance, hazardous use (e.g., alcohol-related injuries, fights) and social/occupational problems (e.g., missing work or school) often involve distinct, memorable events, making them salient enough for both real-time reporting and retrospective recall.

Importantly, the associations between daily and retrospective reports were weaker or inconsistent for certain symptoms (i.e., tolerance, drinking in larger amounts or for longer than intended, craving). First, tolerance—defined as the need to consume increasing amounts of alcohol to achieve the desired effect or experiencing a diminished effect with continued use (American Psychiatric Association, 2013)—is particularly challenging to operationalize in daily reports. In this study, we used two indices: (1) the quantity of alcohol (20 or more drinks in a single day) and (2) a ratio of alcohol quantity to subjective intoxication level. The first index (20+ drinks) had very few positive cases (n = 18 out of 3,339 drinking days), and even when we lowered the threshold to 15+ drinks (n = 51), it was not significantly associated with retrospective measures of tolerance. Although the 20+ drinks threshold aligns with prior work using a similar item to indicate extreme levels of tolerance (Grant et al., 2015; Hasin et al., 2015; Ruan et al., 2008), our item did not explicitly assess subjective intoxication level in relation to the quantity consumed. Although drinking at such extreme levels presumes the presence of at least some tolerance, the lack of subjective anchoring may partly explain why this index showed poor convergence with retrospective reports of tolerance.

The second index (ratio of alcohol quantity to subjective intoxication level) was also not significantly associated with retrospective measures of tolerance. Whereas tolerance is generally conceptualized as reduced sensitivity to alcohol’s effects at the person level (e.g., individuals with greater tolerance report feeling less intoxicated for a given number of drinks), our data suggest that this definition does not align well with daily experiences of AUD symptoms. Of note, higher scores on the ratio-based measure of tolerance were associated with fewer other AUD symptoms (see Supplemental Material 3 for the full correlation matrix). This finding may indicate that the daily ratio measure of tolerance does not fully align with how individuals perceive or retrospectively report their tolerance, further highlighting the challenges of capturing this symptom in real-time assessments. For example, tolerance may be better characterized by episodes of over-consumption, where people are drinking larger amounts than usual to get some intended effect of alcohol, rather than a broader measure of intoxication for a given level of drinking. Given these complexities, further research is needed to refine how tolerance is conceptualized, operationalized, and measured in the context of daily assessments. Nevertheless, our challenges assessing tolerance have been noted by researchers who use retrospective assessments, as well (Boness et al., 2021; Chung et al., 2001; Chung & Martin, 2005; S. P. Lane et al., 2016; Slade et al., 2013), which motivated the DSM-5 Substance Use Disorders workgroup to consider removing it as a diagnostic criterion (Hasin et al., 2013).

Second, our pre-registered operationalization of drinking larger or longer than intended was not significantly associated with retrospective self-reports of that same criterion. However, exploratory analyses using a cutoff of 3 or more additional drinks than intended revealed a significant association with retrospective measures. In our data, consuming 1 or 2 extra drinks was relatively common among young adults, and it is important to note that diagnostic criteria often emphasize consuming significantly large amounts than intended (e.g. First et al., 2016), meaning our pre-registered cutoff was misaligned with that AUD symptom. Retrospective reports may be more sensitive to higher-severity episodes, while more moderate or ambiguous instances—such as drinking slightly more than intended—may be underreported. Further, it is also possible that individuals do not conceptualize drinking intentions in terms of specific, rigid limits but rather set more flexible or heuristic goals, such as whether they intend to get drunk rather than a specific drink count (Stevens, Boyle, Miller, et al., 2022). Future research could benefit from cognitive interviewing methods (Boness & Sher, 2020) and other qualitative approaches to better understand how people conceptualize their drinking intentions. In fact, among people with severe AUD, it is possible that people do not set limits on their drinking, potentially compromising the validity of this AUD criterion (Boness & Sher, 2020; Chung & Martin, 2005).

Third, craving presented unique challenges in linking daily and retrospective reports due to its inherently temporal and fluctuating nature. Unlike other AUD symptoms that may reflect more stable behaviors or consequences over time, craving is typically momentary, capturing acute desires to drink that can vary widely throughout a single day depending on context, mood, and environmental cues (Kavanagh et al., 2013; Kühn & Gallinat, 2011; Ooteman et al., 2006). In this study, craving was assessed multiple times per day by asking participants to rate their past-hour craving; however, to align with other daily AUD symptom measures, we averaged these ratings across the day to create a single daily craving score, which may have obscured meaningful within-day variability. Future research should consider additional metrics—such as peak craving or context-contingent craving—to better capture the dynamic nature of this symptom (Serre et al., 2012; Wray et al., 2014).

Further, the way craving was assessed in daily versus retrospective reports likely contributed to the weaker or inconsistent associations observed. The daily assessment questions asked participants about their craving in the past hour (e.g., How much have you wanted to drink alcohol in the past hour?; If I’d had some alcohol, I would probably have drunk it.), reflecting immediate, momentary experiences that are heavily influenced by contextual factors. In contrast, retrospective craving measures asked participants to reflect on their typical craving frequency, severity, and difficulty resisting alcohol over an extended period (e.g., How often have you thought about drinking?; At its most severe point, how strong was your craving for alcohol?). These questions may prompt participants to recall only the most intense episodes of craving, rather than a true average across time. Thus, rather than simply averaging momentary craving reports, retrospective responses may be shaped by recall biases, peak experiences, or heuristics that emphasize the most memorable or extreme episodes of craving (Kavanagh et al., 2013). Further supporting this interpretation, internal consistency for the Alcohol Urge Questionnaire (Bohn et al., 1995) used in our daily assessments was relatively low (McDonald’s ωt = .51), possibly reflecting the challenge of applying a multi-item trait-based questionnaire in a state-based context. Together, these findings suggest that craving may be better understood and measured as a state-like, context-dependent construct, and future research should consider designs that capture momentary craving intensity, duration, and fluctuations rather than relying on daily or retrospective summaries to represent its occurrence.

Daily Reports Predicting 6-Month Follow-Up

Our study is, to our knowledge, the first to examine the relationship between daily reports of alcohol use consequences and retrospective self-reports of these outcomes at follow-up. Daily reports of social/occupational problems, time spent drinking, and using alcohol more (3+ drinks) than intended significantly predicted the corresponding self-reports at 6-month follow-up, demonstrating convergence between daily and retrospective assessments of AUD symptoms, meaning that daily measures of some AUD symptoms can predict retrospective self-reports. In contrast, we did not observe significant associations between daily and 6-month reports of tolerance and craving. Of note, these symptoms were also weakly associated with baseline self-reports, which may also explain why they did not predict 6-month outcomes.

Tolerance, in particular, presents a challenge for daily measurement, as it reflects gradual physiological adaptations to alcohol use that unfold over weeks or months rather than daily fluctuations. In contrast, craving is highly dynamic, influenced by both internal states and external cues, making it difficult to compare momentary craving reports with retrospective assessments that require participants to summarize their experiences over an extended period. While symptoms like tolerance may be better captured through longer-term or retrospective measures, craving may require momentary assessment approaches that focus on within-day variations rather than broad retrospective summaries.

Two AUD symptoms from Lee et al. (2017), daily hazardous alcohol use behaviors and daily failures to fulfill obligations, did not significantly predict 6-month follow-up self-reports in our sample. These discrepancies may be attributed to the unique characteristics of our sample, which primarily consisted of college students (75% of sample) with a mean age of 20. Hazardous use symptoms, such as getting hurt or blacking out from drinking, may occur infrequently, especially in young adult social drinking contexts where occasional high-risk episodes are relatively common but not necessarily indicative of persistent problems.

However, our data suggest that even when these events occur, they may not always be reported retrospectively. One possibility is that memory biases and recall difficulties play a role—high-risk drinking episodes often occur in social or intoxicated states, where encoding details of the event may be impaired (Bruce et al., 1999). For example, blackouts involve fragmented or lost memory, making retrospective recall inherently unreliable for these experiences (White, 2003). Lastly, young adults’ responsibilities, particularly those in college settings, are more fluid and often change rapidly. For example, college students’ commitments—such as academic responsibilities or social engagements—may shift significantly over short periods. By the time of a 6-month follow-up, these obligations might be entirely different from those reported at baseline, potentially making it more difficult to link failures to fulfill obligations at one point in time to another.

Beyond individual AUD symptoms, we also examined how well aggregated daily reports of AUD symptoms aligned with global retrospective measures of AUD severity, including the total AUDIT score and total DSM-5 AUD symptom count. Baseline AUDIT and total AUD symptom count scores were associated with greater daily AUD symptom reports, suggesting that individuals who retrospectively reported greater alcohol-related problems also tended to experience more frequent symptoms in daily life. Further, aggregated daily AUD symptom reports significantly predicted AUD severity at the 6-month follow-up, reinforcing the validity of daily assessments in capturing meaningful, long-term patterns of alcohol-related impairment. Importantly, because the daily assessments overlapped with one third (2 months) of the follow-up assessment period, significant associations between these measures suggest a degree of convergence between daily and retrospective reports.

Limitations and Future Directions

The current study did not capture the full spectrum of AUD symptoms that individuals may experience in daily life, and not all symptoms were assessed using multiple items. Our symptom indicators were drawn post-hoc from available daily survey items, therefore the mapping between item content and DSM-5 diagnostic criteria is imperfect. For example, our operationalization of the “time spent” symptom included hangover and nausea/vomiting items, which reflect time spent recovering from alcohol’s effects (e.g., First et al., 2016; Grant et al., 2015) but may not fully capture the broader construct of time spent obtaining, using, or recovering from alcohol use. Similarly, our operationalization of the “more than intended” symptom used a 3+ drinks threshold, which may have missed lower-level deviations that could still be clinically meaningful depending on context. Future studies would benefit from prospective repeated assessment designs that incorporate multiple, construct-aligned items per symptom domain to improve both coverage and validity of daily AUD symptom assessment.

Relatedly, an important consideration is that the majority of drinking days in our sample (2,689 out of 3,339 total) also involved cannabis use. While the study focused on alcohol-related consequences, we did not assess whether daily symptoms could be attributed specifically to alcohol or co-use with cannabis. As such, co-use presents an interpretive challenge, particularly in understanding the substance-specific origins of reported symptoms. We acknowledge this as a limitation and suggest future work to assess alcohol- and cannabis-related symptoms separately and/or to design studies specifically focused on co-use contexts to better isolate substance-specific effects.

Neither of the retrospective measures used gold-standard diagnostic approaches, such as semi-structured clinical interviews. Future research should aim to incorporate a more comprehensive assessment of AUD symptoms that could be experienced in daily life including desire to cut down or quit, giving up important activities due to drinking, continued use despite awareness of physical or psychological harm, and withdrawal symptoms. Importantly, expanding the assessment of AUD symptoms will require measuring symptoms beyond drinking episodes, as several symptoms—such as craving, desire to quit, and withdrawal—can occur (or in the case of withdrawal, must occur) on non-drinking days.

In the current study, daily assessments were limited to social weekends (e.g. Thursday to Monday morning, capturing alcohol use from Wednesday to Sunday nights) to reduce participant burden and align with patterns of peak drinking in young adults (Del Boca et al., 2004); however, this restriction may have further constrained the ability to observe a full range of symptom expression, particularly those that emerge on Mondays or Tuesdays or among individuals with more persistent or severe AUD presentations. It is also important to note that the timeframe of daily assessments (8 social weekends) did not fully align with the broader retrospective reporting windows (past-year at baseline and past 6 months at follow-up), limiting the direct comparability of the two assessment methods. Given these limitations, a broader and more temporally inclusive approach could yield different conclusions regarding the psychometric properties of daily AUD symptom assessments. Future studies should strive to incorporate both weekday and weekend assessments and capture a wider range of experience that represent the spectrum of daily AUD symptoms, similar to how prior research has expanded retrospective assessments of AUD (Boness et al., 2019).

Approximately half of our sample met DSM-5 criteria for past-year AUD. Among those individuals, 71% met criteria for mild AUD, 19% for moderate AUD, and 10% for severe AUD. While this distribution is broadly consistent with national epidemiological data (Fairbairn & Kang, 2025; S. P. Lane & Sher, 2015), expanding representation of individuals with more severe AUD could enhance our understanding of how symptoms emerge and fluctuate in daily life across the full severity continuum. Relatedly, while our sample of young adults highlights the value of tailoring assessment tools to reflect the unique drinking behaviors and symptom expressions characteristic of this age group, future research should expand the population of interest to include clinical samples of people with AUD—who are often older and may exhibit different symptom profiles—to ensure broader generalizability and validity across diverse clinical contexts.

Conclusion

Understanding how AUD symptoms manifest in daily life and how they align with both baseline and longitudinal retrospective self-reports is essential for improving alcohol use assessment and intervention. This study examined both the retrospective and prospective validity of daily measures of AUD symptoms by evaluating their correspondence with retrospective reports at baseline and again at a 6-month follow-up. Although certain symptoms (e.g., hazardous use, social/interpersonal problems) showed strong associations across both timeframes, others (e.g., tolerance, exceeding intended drinking limits) exhibited weaker or inconsistent relationships. These findings highlight the need to refine how certain AUD symptoms are measured in real-time and over extended periods, suggesting that retrospective reports may better capture broader patterns and cumulative experiences of alcohol use. Future research should focus on enhancing the validity of daily and longitudinal assessments, ensuring they accurately reflect the dynamic and evolving nature of AUD symptoms.

Supplementary Material

Supplemental Material: Yes. We include supplemental materials 1, 2, and 3 to be posted on the journal’s website.

Figure 1a.

Associations between Baseline Retrospective Reports of AUD Symptoms and Daily AUD Symptoms

Figure 1b.

Associations between Baseline Daily AUD Symptoms and Retrospective Reports of AUD Symptoms at 6-Month Follow-Up