Authors: Angelo M. DiBello (Center for Alcohol and Substance Use Studies & Department of Graduate and Applied Professional Psychology, Rutgers University, Piscataway, New Jersey, USA), Mary Beth Miller (Department of Psychiatry, University of Missouri‐Columbia, Columbia, Missouri, USA), Melissa R. Hatch (Department of Psychology, Rutgers University, Piscataway, New Jersey, USA), Nadine R. Mastroleo (Department of Psychology, Binghamton University, Binghamton, New York, USA), Kate B. Carey (Department of Behavioral and Social Sciences, Brown University School of Public Health, Providence, Rhode Island, USA)
Categories: Behavior, Treatment and Prevention, alcohol use, attitudes, college students, norms
Source: Alcohol, Clinical & Experimental Research
Doi: 10.1111/acer.70008
Authors: Angelo M. DiBello, Mary Beth Miller, Melissa R. Hatch, Nadine R. Mastroleo, Kate B. Carey
The study aimed to expand on existing research related to the theory of planned behavior (TPB) by exploring both between‐person and within‐person effects of descriptive norms and attitudes toward moderate and heavy drinking on drinking outcomes, including drinks per week, blackouts, and alcohol‐related consequences. While previous studies focus on between‐person effects, this study uniquely investigates both between‐person effects as well as within‐person changes over time, using longitudinal data collected at six points over 12 months.
Participants included 484 mandated college students (M
age = 18.66, SDage = 0.758; 55.6% male). Participants completed measures of descriptive drinking norms, attitude toward moderate and heavy drinking, drinking intentions, and drinking outcomes, including drinks per week, blackout, and alcohol‐related consequences. Multilevel models were run to examine the unique between‐and within‐person effects of descriptive drinking norms and attitude toward moderate and heavy drinking on drinking outcomes.
Overall, the results from the multilevel models showed that at the between‐person level, descriptive norms were associated with drinks per week, and a more favorable attitude toward heavy drinking was associated with higher weekly alcohol consumption and related consequences. At the within‐person level, within‐person fluctuations in descriptive norms and attitude toward heavy drinking were associated with higher weekly drinking, blackout, and alcohol‐related consequences, while favorable attitude toward moderate drinking were associated with lower odds of blackouts and fewer alcohol‐related consequences.
Attitudes toward drinking, particularly heavy drinking, at both between‐person and within‐person levels, are strong predictors of alcohol use and its consequences. Furthermore, attitude toward moderate drinking are protective. Interventions promoting a moderate drinking attitude and reducing heavy drinking attitude and descriptive drinking norms will likely be effective in reducing alcohol‐related harm.
Research has continually documented high rates of alcohol use and its associated consequences among college students. In the most recent findings from Monitoring the Future, 81% of students report drinking alcohol at least once in the past year; 63% have drunk in the last 30 days; 28% report at least one heavy‐drinking episode (five or more drinks) at least once in the previous 2 weeks; and 5% report high‐intensity drinking in the last 2 weeks (10+ drinks in a row; Patrick et al., 2023). Furthermore, hundreds of thousands of students report alcohol‐related consequences, including physical and sexual assault and academic difficulties (Grant et al., 2017; Hingson et al., 2017; Patrick et al., 2023). One alcohol‐related consequence of particular importance, alcohol‐induced blackout, also occurs at high rates among college students (Wetherill & Fromme, 2016). Approximately 50% of young adults who drink alcohol report blackouts, defined as partial or complete loss of memory as a result of alcohol use (Richards, Turrisi, et al., 2023; Wetherill & Fromme, 2016). Blackouts warrant special attention because they are uniquely associated with other negative outcomes, including injury and emergency department visits (Merrill et al., 2019; Mundt et al., 2012; Richards, Glenn, et al., 2023). Specifically, because individuals in an alcohol‐induced blackout are still conscious and able to engage with their environment, they are able to continue to engage in problematic behavior, putting them at increased risk for additional negative outcomes. This association with other negative outcomes distinguishes blackouts from many other consequences (e.g., embarrassing things, hangover, and tolerance), which do not necessarily perpetuate additional alcohol‐related harm.
As heavy drinking and its associated negative consequences, including blackout, continue to be a significant concern among college students, it remains important that researchers continue to examine social‐cognitive factors that contribute to these risky behaviors. This may be particularly relevant to students mandated to receive an alcohol intervention after having violated a campus alcohol policy, as mandated students tend to report heavier alcohol consumption than those who are not mandated to intervention (LaBrie et al., 2006; Merrill et al., 2014).
The theory of planned behavior (TPB; Ajzen, 1985) is a social psychological theory that has been applied to the study of alcohol use and its associated consequences within the college environment. The TPB posits that an individual's behavioral intentions are influenced by three key attitudes, social norms, and perceived behavioral control. Within the model, attitudes refer to a person's overall evaluation of the behavior in question and can range from positive to negative. Attitudes are a key explanatory variable that has been shown to predict both intention to engage in a behavior and actual behavior (Bem, 1967; Glasman & Albarracín, 2006; Montano & Kasprzyk, 2008). Social norms describe the perception of what most people in a certain group approve of (injunctive norms) or do (descriptive norms; Cialdini et al., 1991). Perception of others' behaviors (descriptive norms; Cialdini et al., 1991) is one type of social norm that has demonstrated important predictive utility. Finally, perceived behavioral control is the extent to which a person feels they have the ability to engage (or disengage) in performing a certain behavior (Bandura, 1977, 1999).
A meta‐analysis documented that the social‐cognitive factors within the TPB strongly predict instances of alcohol consumption, intoxication, and heavy episodic drinking (Cooke et al., 2016). More recent research shows the TPB constructs have robust predictive ability in college drinking settings both cross‐sectionally and over time. For example, college students who hold a favorable attitude toward heavy drinking (drinking 4+/5+ drinks for females/males) engage in more problematic drinking and experience more consequences over time than those who hold less favorable attitudes toward drinking (DiBello et al., 2018, 2020). A positive attitude toward heavy drinking also serves as a distinct mediating mechanism in explaining alcohol‐related outcomes (DiBello et al., 2019). Furthermore, those who hold a favorable attitude toward moderate drinking (drinking three or less/four or less drinks for females/males) tend to report less drinking, less binge drinking, and fewer alcohol‐related consequences (DiBello et al., 2018). Similarly, descriptive norms are recognized as a reliable predictor of alcohol consumption, consequences, and blackout among young adults both cross‐sectionally and over time (DiBello et al., 2018, 2020; Neighbors et al., 2007). Specific to blackout, DiBello et al. (2020) found that favorable attitude toward blackout as well as descriptive norms were concurrent positive predictors of blackout frequency (DiBello et al., 2020). Subsequently, it was found that a more positive attitude toward heavy drinking was significantly associated with increased odds of blackout over time, beyond norms and intentions (DiBello et al., 2023). Finally, intervention research provides consistent support for changes in descriptive norms (i.e., reductions over time) as a key driver of personalized normative feedback efficacy specific to reducing alcohol use and related consequences (Reid & Carey, 2015). Thus, it is important to examine the TPB constructs as not just static predictors but rather dynamic/changing predictors that might evidence unique effects over time.
The research summarized thus far examined between‐person associations between TPB constructs and alcohol‐related outcomes, consistent with the original formulation of the TPB. That is, individuals who are higher in these constructs (e.g., favorable attitude toward heavier drinking) tend to engage in more drinking and experience more alcohol‐related problems compared to individuals with lower levels. However, attitudes and norms are not static. Presumably, individuals' attitudes toward drinking and perceptions of “normative” drinking can change across moments over time. Indeed, the common (albeit often facetious) comment that one will “never drink again” after a particularly heavy drinking episode indicates that attitudes likely vary over time. Little, if any, literature has examined these within‐person variations in TBP constructs and their influence on key drinking outcomes. Better understanding of the extent to which an individual's attitudes and norms fluctuate over time and the impact these fluctuations may have on drinking and associated consequences has the potential to inform efforts to intervene on drinking at opportune times.
Over the last several years, a few studies have examined within‐person associations between descriptive norms and alcohol use. Specifically, college students tend to engage in heavier drinking when they perceive their peer group is drinking more than usual (Angosta et al., 2023; Dumas et al., 2019), particularly at times when they strongly identify with their drinking group (Graupensperger et al., 2021). However, longitudinal studies incorporating other constructs of the TPB (e.g., attitudes toward heavy and moderate drinking) that allow for the evaluation of between‐ and within‐person effects remain scarce. Examining both types of attitudes and descriptive norms while disaggregating the between‐ and within‐person effects is crucial for improving interventions and advancing our understanding of how social influences shape drinking behavior, blackouts, and consequences.
The purpose of the current study was to extend the emergent work on key constructs related to the TPB by examining both between‐person and within‐person effects of descriptive norms, attitude toward moderate drinking, and attitude toward heavy drinking on drinking outcomes (drinks per week, blackout, and alcohol‐related consequences). Despite robust literature documenting the between‐person effects of these variables on drinking outcomes, few studies examine within‐person effects over time. Leveraging longitudinal drinking data collected at six time points over a span of 12 months, we are able to evaluate both between‐person and within‐person fluctuations of descriptive norms, attitude toward moderate drinking, and attitude toward heavy drinking. To test our hypotheses, we conducted a secondary data analysis of a longitudinal data set from a randomized controlled trial that did not find significant intervention effects (Carey, DiBello, Magill, & Mastroleo, 2024).
Based on existing literature examining between‐person effects of the TPB constructs (DiBello et al., 2018, 2020), we hypothesized the Individuals with higher descriptive norms, in general, will report more drinks per week, higher odds of blackout, and more alcohol‐related consequences, after accounting for the effects of attitude toward moderate drinking and attitude toward heavy drinking. Individuals with a more positive attitude toward moderate drinking, in general, will report fewer drinks per week, lower odds of blackout, and fewer alcohol‐related consequences after accounting for the effects of descriptive norms and attitude toward heavy drinking. Individuals with a higher attitude toward heavy drinking, in general, will report more drinks per week, higher odds of blackout, and more alcohol‐related consequences after accounting for the effects of descriptive norms and attitude toward moderate drinking.
Furthermore, specific to the within‐person effects of the TPB constructs, for which only norms have been examined to date (Angosta et al., 2023; Dumas et al., 2019), we hypothesized the At the time points where individuals report their descriptive norms are higher than usual (meaning higher than their own average), they will report more drinks per week, higher odds of blacking out, and more alcohol‐related consequences, after accounting for the effects of attitude toward moderate drinking and attitude toward heavy drinking. At the time points where individuals report that their attitude toward moderate drinking is higher than usual (higher than their own average), they will report fewer drinks per week, lower odds of blacking out, and fewer alcohol‐related consequences, after accounting for the effects of descriptive norms and attitude toward heavy drinking. At the time points where individuals report that their attitude toward heavy drinking is higher than usual (higher than their own average), they will report more drinks per week, higher odds of blacking out, and more alcohol‐related consequences, after accounting for the effects of descriptive norms and attitude toward moderate drinking.
Participants were college students from a Northeastern US University who had been cited for violating a campus alcohol use policy and had been referred to the Office of Alcohol, Tobacco, and Other Drugs (ATOD). All students mandated to participate in an alcohol intervention at the ATOD were screened to determine their eligibility to participate in the parent study (Carey, DiBello, Hatch, et al., 2025; Carey, DiBello, Magill, & Mastroleo, 2024) by a trained research assistant. The inclusion criteria for the larger study between the ages of 18 and 24; referred for a first campus alcohol‐related offense; full‐time undergraduate enrollment; and willingness to participate for a year in order to satisfy their sanction requirement.
Those who consented to participate in the parent study completed a baseline assessment, were randomized to either a control or self‐affirmation writing activity, completed the eCHECKUP TO GO intervention (Moyer et al., 2004), and did a postintervention assessment within one sitting. Those in the control writing activity populated a list of the foods they had eaten in the last 24 h and then wrote a description of the first listed food. Those in the self‐affirmation writing activity listed their top five most important values in order of importance and then wrote about their most important value and how they have displayed that value in their life. They completed follow‐up surveys remotely at five additional time 1, 3, 6, 9, and 12 months. All study procedures were approved by the university's Institutional Review Board.
Participants provided information on their biological sex, race/ethnicity, and age at baseline.
Drinks per week were measured using the Daily Drinking Questionnaire (Collins et al., 1985). Participants were asked to estimate the number of drinks they typically consume on each day of the week in the past 30 days, and the number of drinks was then summed across days to determine average drinks per week. A standard drink was defined as 12 oz. of beer, 5 oz. of 12% table wine, 12 oz. of wine cooler, or 1.25 oz. of 80‐proof liquor.
An item from the Brief Young Adult Alcohol Consequences Questionnaire (BYAACQ; Kahler et al., 2005) was used to assess blackout experience in the past month. Participants responded yes/no to the item, “I have not been able to remember large stretches of time while drinking heavily.”
Alcohol‐related consequences was assessed using a modified version of the Brief Young Adult Alcohol Consequences Questionnaire (BYAACQ; Kahler et al., 2005). The modified version of the BYAACQ included all 24‐items, with the exception of the blackout item noted above. Responses are dichotomous (yes/no) and refer to the past month. The BYAACQ demonstrates strong psychometric properties and is free of gender bias (Kahler et al., 2005). Example items include, “I have felt very sick to my stomach or thrown up after drinking” and “I have often found it difficult to limit how much I drink.” The 23‐item BYAACQ variation used in this study evidenced strong internal consistency (α
Baseline = 0.87; α
1‐month = 0.90; α
3‐month = 0.89; α
6‐month = 0.92; α
9‐month = 0.91; α
12‐month = 0.93).
Attitude toward moderate drinking was assessed using an adapted version of the attitudes measure developed by Hagger et al. (2012). For moderate drinking, the stem read, “Keeping my alcohol drinking within what is considered moderate drinking for adults (i.e., at 4 or fewer drinks for men or at 3 or fewer drinks for women) on each individual occasion over the next month would be…” Participants responded to five semantic differential scales ranged from 1 to unenjoyable‐enjoyable, bad‐good, harmful‐beneficial, foolish‐wise, and unpleasant‐pleasant. The five items were averaged to create a scale representing attitude toward moderate drinking, (α
Baseline = 0.87; α
1‐month = 0.86; α
3‐month = 0.90; α
6‐month = 0.92; α
9‐month = 0.91; α
12‐month = 0.92).
Attitude toward heavy drinking was assessed using a 5‐item measure with demonstrated reliability and validity (DiBello et al., 2018). The heavy drinking stem read “Having five or more drinks (for males)/four or more drinks (for females) in a sitting over the next month would be…” Participants responded to five semantic differential scales ranged from 1 to unenjoyable‐enjoyable, bad‐good, harmful‐beneficial, foolish‐wise, and unpleasant‐pleasant. The five items were averaged to create a scale representing attitude toward heavy drinking, (α
Baseline = 0.91; α
1‐month = 0.90; α
3‐month = 0.89; α
6‐month = 0.90; α
9‐month = 0.92; α
12‐month = 0.91).
Descriptive norms for same‐gender students at the university were measured using a 7‐day grid adapted from the Daily Drinking Questionnaire (Collins et al., 1985) by Krieger et al. (2016). Participants were asked to report how much they believed peers consumed on each day of a typical week (Krieger et al., 2016). Responses were summed to generate a “normative” perception of peer drinking quantity in a typical week.
A modified version of the Daily Drinking Questionnaire (Collins et al., 1985) was used to assess participant drinking intentions at baseline. Specifically, participants were asked to estimate the number of drinks they intended to drink on each day of the week in the next month. The intended number of drinks was summed across days to determine average intended drinks per week.
Prior to analyses, descriptive statistics and bi‐variate correlations among variables were examined. Outcomes of interest included drinks per week, experiencing a blackout, and alcohol‐related consequences, which were assessed up to 6 times (baseline, 1‐month, 3‐month, 6‐month, 9‐month, and 12‐month follow‐up). Analyses were conducted using generalized linear mixed models (GLMM) and an extension of general linear models allowing for multilevel data with non‐normal distributions (Atkins et al., 2013; Hedeker & Gibbons, 2006) using the GLMMIX procedure in SAS version 9.4. Missing data were assumed to be missing at random. Preliminary analyses indicated that negative binomial models were most appropriate for count outcomes, such as drinks per week and alcohol‐related consequences; blackout was modeled as a binary outcome (0/1; no/yes) using a logistic model. Results from negative binomial models are log based and are therefore exponentiated to aid in interpretability; the resulting exponentiated coefficients are interpretable as incident rate ratios (IRRs), which are the expected proportional change in the outcome for each unit change in the predictor.
Primary analyses were conducted using multilevel modeling with maximum likelihood estimation to account for the nested two‐level data structure; specifically, assessments (Level 1) within person (Level 2). Models tested three variables (descriptive norms, attitude toward moderate drinking, and attitude toward heavy drinking) at the between‐ and within‐person levels as predictors of three alcohol outcomes (drinks per week, experiencing a blackout, and alcohol‐related consequences). Level 1 predictors were centered within person, and Level 2 predictors were centered between persons.
The substantive questions of interest were examined in the following ways. First, an unconditional GLMM (i.e., random intercepts for person only) with time and no other predictors was estimated for each of the three outcomes to confirm within‐person variation over time. Next, we estimated three separate conditional models (one for each outcome) where within‐person norms, attitude toward moderate drinking, and attitude toward heavy drinking deviations at Level 1 and between‐person norms, attitude toward moderate drinking, and attitude toward heavy drinking at Level 2 were included as predictors. Biological sex (0 = female, 1 = male), condition (0 = control writing and 1 = self‐affirmation writing), and drinking intentions were included in all models as Level 2 covariates. For the models evaluating blackout and alcohol‐related consequences, drinks per week at Level 2 were also included as a covariate. Time was coded as 0, 1, 3, 6, 9, and 12 to reflect the interval between assessment points and included in all models. All analyses included random intercepts.
Participants included 484 individuals (M
age = 18.66, SDage = 0.758; 55.6% male). They identified as the 11.8% Asian, 3.7% Black/African American, 0.2% American Indian/Alaska Native, 0.6% Native Hawaiian/Pacific Islander, 6.0% Other, and 77.6% White. Additionally, 11.6% identified as Hispanic/Latino. On average, in the past month, participants reported consuming 8.90 (SD = 7.74) drinks per week, drinking 4.83 (SD = 3.91) drinks per drinking occasion, and experiencing an average of 4.64 (SD = 4.39) total consequences over the past month at baseline. Differential attrition, defined by missing any of the 1‐, 3‐, 6‐, 9‐, or 12‐month follow‐ups (n = 216), was examined using logistic regression. Attrition status was not significantly associated with baseline drinking, blackout, alcohol‐related problems, race/ethnicity, or sex.
The results from the model examining drinks per week over time (no other predictors) revealed that drinks per week decreased by 2% at each time point (IRR = 0.982, 95% CI [0.973 0.990], p < 0.001). Similarly, the odds of blackout decreased by ~10% (OR = 0.907, 95% CI [0.878 0.937], p < 0.001) and alcohol‐related consequences decreased by 6% at each time point (IRR = 0.943, 95% CI [0.931 0.955], p < 0.001). Adjusting for covariates did not alter these trajectories over drinks per week (IRR = 0.983, 95% CI [0.975 0.992], p < 0.001); odds of blackout (OR = 0.907, 95% CI [0.878 0.937], p < 0.001); and alcohol‐related consequences (IRR = 0.945, 95% CI [0.934 0.957], p < 0.001).
Results from the full multilevel model can be found in Table 1; only the substantive predictors of interest will be highlighted here in the text. At the between‐person level (Level 2), there were significant effects of descriptive norms and attitude toward heavy drinking on drinks per week. Specifically, those who believed their peers drank more alcohol reported consuming more alcohol than those who believed their peers drank less (every additional perceived peer drink was associated with 20% more drinks per week). Similarly, those who reported more favorable attitude toward heavy drinking consumed more alcohol than those with less favorable attitude toward heavy drinking (every one‐unit increase in attitude was associated with 49% more drinks per week). There was no significant effect of attitude toward moderate drinking on drinks per week at the between‐person level.
Within persons (at Level 1), participants reported drinking more at times when their norms and attitude toward heavy drinking were higher than their own average. For every one‐unit increase in descriptive norms above their own mean, there was a 23% increase in drinks per week consumed; and for every one‐unit increase in attitude toward heavy drinking above their own mean, there was a 24% increase in drinks per week consumed. Attitude toward moderate drinking was not associated with drinks per week at the within‐person level.
Results from the full multilevel model can be found in Table 2. At the between‐person level, we observed no significant effects of attitude toward heavy drinking, descriptive norms, or attitude toward moderate drinking on the odds of blackout.
When examining the within‐person effects, the results revealed that for every one‐unit increase in descriptive norms above a person's mean, the odds of experiencing a blackout increased by nearly 32%, and for every one‐unit increase in attitude toward heavy drinking above a person's mean, the odds of experiencing a blackout also increased by 29%. Furthermore, for every one‐unit increase in attitude toward moderate drinking above a person's mean, the odds of experiencing a blackout decreased by 20%.
Results from the full multilevel model can be found in Table 3. At the between‐person level (Level 2), there was only a significant effect of attitude toward heavy drinking, suggesting that a one‐unit increase in a more favorable attitude toward heavy drinking was associated with experiencing 30% more alcohol‐related consequences. There was no significant effect of descriptive norms or attitude toward moderate drinking on alcohol‐related consequences at the between‐person level.
When examining the within‐person deviation effects (Level 1), for every one‐unit increase in descriptive norms above a person's mean, there was a 16% increase in alcohol‐related consequences and for every one‐unit increase in attitude toward heavy drinking above a person's mean, there was a 10% increase in alcohol‐related consequences. Finally, for every one‐unit increase in attitude toward moderate drinking above a person's mean, there was a 9% decrease in alcohol‐related consequences.
The current study tested how between‐ and within‐person variation in two different attitudes (toward moderate drinking and toward heavy drinking) and descriptive norms are linked to college students' drinking outcomes. Importantly, this is among the first studies to document that attitudes toward drinking do vary within persons over time. Moreover, attitude toward heavy drinking at the between‐person level was positively associated with drinks per week and consequences, and at the within‐person level, it was positively associated with drinks per week, odds of blackout, and alcohol‐related consequences. This is consistent with previous research on the predictive validity of TPB with regard to alcohol outcomes (Cooke et al., 2016) as well as work documenting these associations at the between‐person level across a variety of alcohol outcomes, including drinking, general alcohol‐related consequences, and blackout (DiBello et al., 2018, 2020, 2022). It advances this work by demonstrating not only that people with a more favorable attitude toward heavy drinking drink more and experience more consequences—but that individuals drink more and experience more consequences at times when their attitude toward heavy drinking is especially favorable. This highlights the importance of within‐person effects of attitudes and norms on alcohol outcomes.
Descriptive norms were also associated with more drinks per week at the between‐ and within‐person levels and were associated with blackout and other alcohol‐related consequences at the within‐person level. This is generally consistent with existing work examining norms and alcohol‐related outcomes (DiBello et al., 2022, 2023). Consistent with the limited previous work (Dumas et al., 2019; Graupensperger et al., 2021) on within‐person effects of descriptive norms on drinking, higher‐than‐usual descriptive norms were associated with worse alcohol outcomes in each model examined. Thus, we have demonstrated that over the course of a year, periods during which perceived descriptive drinking norms of one's peer group are elevated, relative to their own mean, appear to be associated with an increase in how much a college student will drink and similarly increase their risk for consequences. This finding points to the relevance of situationally specific perceptions of peer norms on negative outcomes. It also suggests that the socialization effects of normative influence may have somewhat transitory effects based on exposure to heavier drinking peers (c.f., Reifman et al., 2006).
In contrast to attitude toward heavy drinking and descriptive norms, attitude toward moderate consumption had a protective effect, but only at the within‐person level. That is, when participants were feeling more positively about moderate drinking, they experienced reduced odds of blackout and fewer alcohol‐related consequences; but people who felt more favorably toward moderate drinking did not differ in the experience of consequences from those who felt less favorably toward moderate drinking. This is generally consistent with previous work that did not differentiate between‐ from within‐person effects (DiBello et al., 2018). The finding that this effect may occur at the within (not between) persons level is somewhat counterintuitive. However, it may also indicate promise for moderate drinking attitudes as a target of prevention and intervention efforts—as promoting safer drinking behavior (moderate drinking) may be more acceptable to college students than condemning risky behavior (heavy drinking). This perspective is consistent with harm reduction approaches to substance use; specifically, making alcohol use safer and reducing its adverse impacts, rather than focusing exclusively on avoiding all use (Collins & Clifasefi, 2023; SAMHSA, 2024). These approaches emphasize individual autonomy through the acknowledgment that not all individuals want to abstain from substance use and many individuals are able to live productive lives while engaging in moderate substance use. More generally, these findings underscore the importance of specifying the attitude object in studies of alcohol‐related attitudes, as attitude toward heavy drinking and attitude toward moderate drinking demonstrated associations in opposite directions.
This work has implications for clinical practice and future research. Most interventions for alcohol use focus on normative correction rather than changing attitudes. Although our results support that reducing perceptions of peer descriptive norms could impact a significant risk factor for problematic use, they also suggest that such a narrow focus may be missing other theoretically important intervention targets. Alcohol prevention interventions could benefit from focusing on reducing one's favorable attitudes toward heavy drinking and/or increasing one's favorable attitude toward moderate drinking. As one example, motivational interviewing (MI; Miller & Rollnick, 2012) involves eliciting internal motivations for change and could be used to target attitudes toward moderate drinking. We encourage future studies in this area, as attitudes are an understudied mechanism of change in existing work (Reid & Carey, 2015). Additionally, the existing robust literature on attitude change within social psychology can be used to develop new strategies for altering attitudes toward drinking. Research on reducing substance use, such as cigarette smoking, has effectively used attitude change processes to alter behavior (Simmons et al., 2013, Simmons et al., 2004; Simmons & Brandon, 2007). Indeed, research applying these techniques to reduce alcohol use and related harms has shown promise (Carey, DiBello, Magill, & Mastroleo, 2024; DiBello et al., 2018).
The study is not without its limitations. First, participants consisted of predominantly White, mandated college students, who are not representative of all college students. Second, the absence of perceived behavioral control did not allow for the full test of all TPB constructs at the between‐ and within‐person levels. Third, these longitudinal relationships were tested in the context of an RCT (albeit one that did not find significant intervention effects) and during the COVID pandemic. Thus, future research should replicate these findings in more diverse and non‐intervention samples to enhance generalizability.
Attitudes toward drinking are a key independent predictor of alcohol use and its related consequences. Even when considering other factors like drinking intentions and norms, attitude toward heavy drinking consistently emerge as significant predictors of alcohol‐related outcomes. This underscores the potential value of targeting attitudes in interventions aimed at reducing alcohol use and its negative consequences. Importantly, the attitude toward heavy drinking seems to increase the risk for harm, while the attitude toward moderate drinking appears to be protective. As such, prevention interventions that encourage a more favorable attitude toward moderate drinking may hold promise for reducing alcohol‐related harm when interventions aimed at decreasing heavy drinking fail.
The authors have no conflicts of interest to report.