Latent Motivational Profiles of Behavioral Regulations for Drinking Distinguish Groups by Alcohol Consumption and Consequences

Abstract

Background:

Self-Determination Theory characterizes drinking motives according to level of autonomy and locus of control and aligns with harm reduction approaches to alcohol use. This study used Latent Profile Analysis (LPA) to identify motivational profiles of self-determined behavioral regulations for drinking and to test their associations with sociodemographic variables and alcohol consumption and outcomes.

Methods:

Adults 18–57 (N=630, M_age=21.5, 55% female, 88% undergraduates) reported typical alcohol consumption (Daily Drinking Questionnaire), consequences (Brief Young Adult Alcohol Consequences Questionnaire, Importance of Consequences of Drinking), and five behavioral regulations for drinking including amotivation, external regulation, positive introjected regulation, identified regulation, and intrinsic regulation (Comprehensive Relative Autonomy Index for Drinking). 3-Step LPA tested models with one to five profiles and associations between the final model with sociodemographic variables and drinking behaviors and consequences.

Results:

A four-profile model fit best. The “Aimless Drinkers” profile (n=75, 11.9%) had low intrinsic regulation (e.g., enjoyment of drinking) and was disproportionately male (ORs = 2.35–2.65). The “Pleasure Drinker” profile (n=114, 18.1%) had average intrinsic and low other regulations and was significantly older (OR=1.07) than the “Externally Controlled Drinker” profile (n=177, 28.1%), which had high external regulation (e.g., drinking due to social pressure). Externally Controlled Drinkers reported greater drinking intensity (M_drinks/day=5.4) and more negative (M_neg=10.6) and positive (M_pos=7.2) consequences than all other profiles, including the “Flexible Drinker” profile (n=264, 41.9%), which was average across all regulations. Pleasure Drinkers experienced more positive consequences (M_pos=3.3 vs. 1.5) than Aimless Drinkers, despite similar alcohol consumption.

Conclusions:

Older individuals were more likely to drink due to intrinsic enjoyment, which may reflect a shift towards greater autonomy over alcohol consumption with increasing age. Drinking primarily for enjoyment also facilitated more positive drinking outcomes. Future research should investigate whether motivational profiles of behavioral regulations for drinking predict long-term trajectories of alcohol consumption and alcohol-related risks.

Introduction

Alcohol use, misuse, and negative consequences are public health problems across adulthood, particularly among young adults less than 30 years old. Over 50% of U.S. young adults used alcohol in the past 30 days, with 31% and 9% engaging in heavy episodic (4+/5+ drinks for women/men) and high intensity drinking (8+/10+ drinks for women/men), respectively in the past 30 days (Substance Abuse and Mental Health Services Administration Office of Applied Studies, 2021). Heavy episodic and high intensity drinking increase risk for serious mental and physical consequences including depression, anxiety, overdoses, injuries, and drunk driving (Terry-McElrath, O’Malley and Johnston, 2014; Hingson, Zha and White, 2017; CDC, 2020). The prevalence of heavy episodic and high intensity drinking is above 40% through age 25 (Patrick et al., 2017), after which many individuals transition towards lower risk behaviors that track throughout adulthood. Promoting transitions towards lower risk drinking requires adequate knowledge of factors influencing this transition, such as drinking motives (Patrick and Schulenberg, 2011).

Drinking motives strongly and proximally predict alcohol consumption and consequences (Cox and Klinger, 1988; Cooper, 1994; Courtney, Russell and Conroy, 2024). Motives are more modifiable than other factors associated with age-related transitions in alcohol consumption, such as changes in neurodevelopment, personality, or social roles (Patrick and Schulenberg, 2011; Sharma and Smith, 2011; Ryan and Deci, 2017; Richards, Pearson and Witkiewitz, 2021), making motives a potentially useful target for harm reduction interventions. Drinking motives change with age, with the proportion of people drinking to escape problems decreasing from 24% to 11% across ages 18–26 and the proportion drinking because it tastes good increasing from 40% to 62% across that same age range (Littlefield, Sher and Wood, 2010; Patrick et al., 2017). The effects of drinking motives on alcohol use also change with age (Littlefield, Sher and Wood, 2010; Patrick et al., 2017). For example, the effect of drinking due to anger on the odds of engaging high-intensity drinking decreased from 3.8 (odds ratio [OR]) among 18-year-olds to 1.9 (OR) among 25-year-olds (Patrick et al., 2017). Similarly, decreased coping motives with increasing age predict decreases in problem drinking across time, with some researchers positing that these changes may be a mechanism related to maturing out of problematic drinking (Littlefield, Sher and Wood, 2010). In sum, drinking motives predict alcohol consumption and consequences, motives change with age, and age-related changes in drinking motives parallel changes in drinking-related outcomes.

Drinking motives have previously been characterized according to the valence (positive or negative) and source (internal or external) of the outcomes people expect to achieve by drinking, corresponding with the four factors of social, coping, conforming, and enhancing motives measured by the Drinking Motives Questionnaire (DMQ) (Cox and Klinger, 1988; Cooper, 1994). A complementary approach for understanding drinking motives is Self-Determination Theory (SDT), which accounts for key psychological constructs, such as locus of control, the basic psychological needs for autonomy, competence and relatedness, and drinking identity, that predict alcohol consumption and consequences via behavioral regulations that capture a broad spectrum of motives for drinking alcohol (Levesque et al., 2006; Chawla et al., 2009; Ryan and Deci, 2017; Lassi et al., 2019; Courtney, Russell and Conroy, 2024, 2025). SDT suggests that human motivation stems from fulfilling the psychological needs for autonomy (i.e., having choice and control over one’s actions), competence (i.e., experiencing mastery of an activity), and relatedness (i.e., feeling connected to others). SDT further proposes that motives range on a continuum from low to high autonomy and impersonal to internal locus of control, with the behavioral regulations ordered along the relative autonomy continuum from low to high levels of autonomy and external to internal locus of control (Figure 1) (Ryan and Deci, 2017; Sheldon et al., 2017). As shown in Figure 1, amotivation is on the left end of the continuum because it reflects a complete lack of autonomy in which a person has no intentional motivation or has unclear motives for their behavior (Ryan and Deci, 2017; Sheldon et al., 2017). Next is the low autonomy behavioral regulation called external regulation in which people are motivated by the desired to avoid punishment or comply with others’ expectations, such as drinking alcohol because of pressure from friends (Ryan and Deci, 2017; Sheldon et al., 2017). As autonomy increases, people experience negative introjected regulation in which they are motivated by a desire to avoid loss of self-worth, such as drinking alcohol to avoid feeling uncomfortable (Ryan and Deci, 2017). In contrast, as autonomy increases and locus of control shifts from external to internal, people experience positive introjected regulation in which they are motivated by a desire to enhance their self-worth, such as drinking alcohol to boost one’s self-esteem (Ryan and Deci, 2017). Next is identified regulation, which is when a person is motivated to engage in a behavior because it results in valued outcomes, such as drinking wine to demonstrate one’s knowledge of viticulture (Ryan and Deci, 2017). Lastly, intrinsic regulation lies on the far right of the autonomy continuum. Intrinsic regulation is when a person is motivated by inherent enjoyment, such as drinking because one simply enjoys drinking alcohol (Ryan and Deci, 2017).

Figure 1.

Figure 1 shows Self-Determination Theory’s conceptualization of human motivation. This shows the six motivational behavioral regulations ranging in their level of autonomy from low (non-autonomous amotivation) to high (completely autonomous intrinsic regulation), their perceived locus of control from impersonal (amotivation) to completely external (external regulation) to completely internal (intrinsic regulation), and the relevant motivational processes for each individual behavioral regulation. Based on research testing the internal validity of the CRAI-Drinking survey (Courtney, Russell and Conroy, 2025), negative introjected regulation was not included in these analyses.

Some of SDT’s behavioral regulations share similarities with DMQ motives, such as intrinsic regulation and enhancement motives or external regulation and social motives (Courtney, Russell and Conroy, 2024, 2025). However, SDT’s behavioral regulations are also uniquely related to alcohol consumption and consequences beyond effects of DMQ motives, and amotivation may be a novel indicator of future alcohol-related risk (Courtney, Russell and Conroy, 2024, 2025). Not only can SDT be used to explain why people drink alcohol, but it also describes the process by which behavior change can occur, which constructs to target for interventions, and which individuals might be more likely to respond to an intervention (Sharma and Smith, 2011; Ryan and Deci, 2017; Richards, Pearson and Witkiewitz, 2021). Indeed, researchers have been encouraged to apply SDT to investigating alcohol use, because SDT is a framework that is well-suited for understanding harm reduction approaches to alcohol use, which may be particularly salient to high-risk college student drinkers (Sharma and Smith, 2011; Richards, Pearson and Witkiewitz, 2021).

Both SDT and harm reduction approaches to alcohol use are theoretically grounded in humanistic psychology (Ryan and Deci, 2017; Richards, Pearson and Witkiewitz, 2021). SDT argues that promoting high levels of individual behavioral autonomy supports the initiation and maintenance of health-promoting behaviors (Ryan et al., 2008; Ryan and Deci, 2017). The emphasis SDT places on maximizing individual autonomy aligns morally and practically with harm reduction approaches to alcohol use (Sharma and Smith, 2011; Richards, Pearson and Witkiewitz, 2021). Additionally, the psychological factors accounted for by SDT, including autonomy and locus of control, are established predictors of alcohol consumption and outcomes that can serve as intervention targets (Levesque et al., 2006; Chawla et al., 2009; Ryan and Deci, 2017; Bhowmick, Kulkarni and Gaikwad, 2019; Lassi et al., 2019). Despite robust evidence supporting the value of SDT in understanding human motivation across behavioral domains (including substances like tobacco), and the strong theoretical and moral links between SDT and harm reduction approaches to alcohol use, there are few studies using SDT to examine drinking motives and related outcomes (Levesque et al., 2006; Ryan and Deci, 2017; Richards, Pearson and Witkiewitz, 2021; Courtney, Russell and Conroy, 2024, 2025).

Existing studies assessing how SDT-based motives relate to drinking outcomes are typically limited to dichotomous comparisons of extrinsic versus intrinsic or controlled versus autonomous drinking motives (Knee and Neighbors, 2002; Neighbors et al., 2004; Chawla et al., 2009) (see Hagger et al., 2012 and Courtney et al., 2024 for exceptions). Consistent with SDT, these studies found that drinking due to controlled or extrinsic motives was associated with greater alcohol use (Knee and Neighbors, 2002; Neighbors et al., 2004; Chawla et al., 2009). Although these findings align with SDT, dichotomizing drinking motives fails to capitalize on two key features of SDT, the relative autonomy continuum (RAC) and organismic integration theory (Ryan et al., 2008; Ryan and Deci, 2017). The relative autonomy continuum specifies that behavioral regulations for drinking vary from low to moderate levels of autonomy, with four intermediate regulations underlying the broader concept of extrinsic motives studied in prior research, including external regulation, negative and positive introjected regulations, and identified regulation (Ryan et al., 2008; Ryan and Deci, 2017). These behavioral regulations for drinking have unique and sometimes opposite associations with alcohol consumption, with one study finding that greater positive introjected regulation was associated with higher drinking intensity, whereas greater identified regulation was associated with lower drinking intensity (Courtney, Russell and Conroy, 2024). These opposing effects can be explained by organismic integration theory, which specifies that the different behavioral regulations underlying extrinsic motivation are related to contextual factors that may promote or prevent a person from transitioning towards more autonomous self-determined motivation (Ryan et al., 2008; Ryan and Deci, 2017). For example, identified regulation promotes decreased drinking intensity through its effects on attitudes and subjective norms (Hagger et al., 2012), suggesting that targeting these constructs as a mechanism for behavior change could help people transition towards greater autonomy, a process referred to as internalization and integration (Ryan et al., 2008; Ryan and Deci, 2017; Richards, Pearson and Witkiewitz, 2021).

Simultaneously considering each of SDT’s behavioral regulations for drinking could benefit harm reduction interventions for several reasons. First, these regulations are uniquely associated with alcohol consumption and outcomes (Hagger et al., 2012; Courtney, Russell and Conroy, 2025). Second, these regulations do not exist in isolation. Rather, individuals can simultaneously report high levels of different behavioral regulations, which in turn combine and interact with each other to predict behavior (Howard, Gagné and Morin, 2020; Richards, Pearson and Field, 2020). Third, subgroups of individuals often share similar patterns of behavioral regulations that uniquely predict behavioral outcomes (Littlefield et al., 2013; Stapinski et al., 2016; Lee and Lee, 2018; Richards, Pearson and Field, 2020) and intervention responses (Tanaka and Nolan, 2018; Rijbroek et al., 2019; Beattie et al., 2022; Qu et al., 2022). This can inform prevention research by identifying which subgroups of individuals are at greatest risk and which behavior change mechanisms to use for those individuals, paving the way for targeted, scalable interventions. Therefore, to achieve the goals of understanding alcohol consumption across varying ages and of informing harm reduction behavior change interventions, researchers should simultaneously investigate each of SDT’s individual behavioral regulations. One challenge in modeling this range of behavioral regulations for alcohol consumption is that some level of collinearity is expected between adjacent behavioral regulations along the relative autonomy continuum (Sheldon et al., 2017). This collinearity creates a vulnerability for variable-centered analyses to provide overly reductionistic solutions to complex behavioral phenomena that might be driven by combinations of behavioral regulations instead of individual behavioral regulations (Bates, 2000; Howard and Hoffman, 2018).

One set of methods that are particularly well-suited for simultaneously investigating SDT’s individual behavioral regulations and identifying subgroups with similar patterns are person-centered approaches. Person-centered approaches focus on understanding the complex processes that characterize individuals by assuming that there are synergies or patterns of variables that contribute to behavioral outcomes and that there are subgroups of individuals with similar synergies or patterns within a given population (Bates, 2000; Howard and Hoffman, 2018). These assumptions are useful because they account for the fact that people can simultaneously experience multiple types (i.e., qualities) of behavioral regulations for drinking, alongside differing levels (e.g., quantities) of each behavioral regulation, all of which impact their alcohol consumption and outcomes (Littlefield et al., 2013; Stapinski et al., 2016; Lee and Lee, 2018; Richards, Pearson and Field, 2020). Despite the value of person-centered approaches, most prior research investigating drinking motives using SDT employed variable-centered approaches (Knee and Neighbors, 2002; Neighbors et al., 2004; Chawla et al., 2009; Hagger et al., 2012; Courtney, Russell and Conroy, 2024), which do not allow researchers to simultaneously account for multiple types and amounts of behavioral regulations for drinking (Bates, 2000; Howard and Hoffman, 2018).

One exception is the study by Richards et al. (2020) which used a person-centered approach to identify subgroups of college students who shared similar behavioral regulations for responsible drinking (e.g., drinking to take care of one’s own health) (Richards, Pearson and Field, 2020). They found three latent profiles of college student drinkers characterized by High Quality (i.e., high levels of introjected regulation and autonomous motivation), High Quantity (i.e., high levels of all behavioral regulations, which included amotivation, external regulation, introjected regulation, and autonomous motivation), or Low Quantity (i.e., relatively low levels of all behavioral regulations) behavioral regulations for responsible drinking (Richards, Morera and Field, 2020; Richards, Pearson and Field, 2020). The High Quality and High Quantity profiles engaged in less alcohol use and experienced fewer negative drinking consequences than the Low Quantity profile (Richards, Pearson and Field, 2020). Richards et al.’s study (2020) provides initial support for the existence of different motivational profiles of behavioral regulations for responsible drinking; however, we are unaware of studies investigating the presence of motivational profiles of behavioral regulations for general alcohol drinking behaviors. Motivations for general alcohol drinking behaviors are conceptually distinct from motives for responsible drinking (Richards, Pearson and Field, 2021; Courtney, Russell and Conroy, 2024). For example, a person may drink to enjoy drinking (i.e., intrinsic regulation) or its effects (i.e., identified regulation), but they may drink responsibly due to peer pressure (i.e., external regulation). In contrast, another person may drink due to pressure from others but may be intrinsically motivated to drink responsibly. These two people would require different interventions, highlighting the practical importance of the distinction between motives for general versus responsible drinking. Therefore, using a person-centered approach to identify subgroups of individuals with similar motivational profiles of behavioral regulations for general alcohol drinking behaviors may inform knowledge about how behavioral regulations for drinking cluster with one another and inform intervention development by identifying which subgroups of individuals may benefit most from harm reduction interventions.

Therefore, the purpose of this study was threefold. First, we used a person-centered approach (latent profile analysis) to determine whether we could identify distinct latent motivational profiles of behavioral regulations for drinking (Courtney, Russell and Conroy, 2025). These profiles reflect the main effects and interactions between behavioral regulations that drive alcohol use behavior and its consequences. Second, past studies suggest there may be sex (Cooper, 1994; Knee and Neighbors, 2002; Kuntsche et al., 2006; Stapinski et al., 2016) and age differences in drinking motives that could explain maturing out of problem drinking (Patrick and Schulenberg, 2011; Littlefield et al., 2013; Patrick et al., 2017); therefore, we examined whether those sociodemographic characteristics were associated with the likelihood of membership in a specific latent motivational profile. Third, we investigated whether latent profile membership was associated with alcohol consumption and consequences. Based on previous literature, we hypothesized there would be distinct latent profiles of behavioral regulations for drinking, with at least one profile being characterized by higher levels of no or low autonomy behavioral regulations (e.g., ‘Low Quality’ regulations) and at least one profile being characterized by higher levels of moderate or high autonomy behavioral regulations (e.g., ‘High Quality’ regulations) (Vansteenkiste et al., 2009; Emm-Collison et al., 2020; Richards, Pearson and Field, 2020; Moore et al., 2023). Similarly, we hypothesized that membership in profiles characterized by higher levels of no or low autonomy behavioral regulations would be associated with greater alcohol use and more negative drinking consequences, such that these subgroups would benefit most from harm reduction interventions promoting internalization and integration of motives (Ryan et al., 2008; Ryan and Deci, 2017; Richards, Pearson and Witkiewitz, 2021). We hypothesized membership in profiles characterized by higher levels of moderate or high autonomy behavioral regulations would be associated with less alcohol use and fewer negative drinking consequences (Vansteenkiste et al., 2009; Richards, Pearson and Field, 2020).

Materials and Methods

Participants, recruitment, and screening

Participants included two convenience samples of adults who typically consumed at least one drink per week. Sample 1 included adults 18 years or older (23.0 ± 6.6 years) and Sample 2 included adults 18–25 years (20.4 ± 1.5 years). Data from the two samples were combined. Of the 1,026 individuals who completed an online screening survey between November 2020 and December 2021, 197 (19.2%) did not qualify, resulting in 824 (80.3%) qualified participants, 630 (76.5%) of whom completed the online study survey and were compensated with extra credit or entered to win one of two $50 gift cards (Sample 1) or one of ten $30 gift cards (Sample 2). This study was approved by the university Institutional Review Board (Protocol #00016554). Participants provided informed consent to participate.

Measures

Self-Determined Motives for Drinking.

SDT-based drinking motives were assessed using the Comprehensive Relative Autonomy Index for Drinking (CRAI-Drinking) (Courtney, Russell and Conroy, 2025). A previous study validated and cross-validated the CRAI-Drinking measure via multi-dimensional scaling analyses and confirmatory factor analyses in two samples, finding that the 19-item, five-factor CRAI-Drinking has good structural validity (χ²(142) = 165.463 (p = .087), CFI = .99, TLI = .99, RMSEA = 0.025 (95% CI [0.000, 0.040]), SMSR = .06), with each of the 19-items having a significant loading on a single a priori assigned factor (p < .001, standardized loadings 0.56 – 0.81), and the measure showing configural, metric, and scalar invariance between males and females (Courtney, Russell and Conroy, 2025). The CRAI-Drinking includes 19 items assessing five behavioral regulations for drinking, including intrinsic (e.g., “Because drinking is fun”, 4 items, α = 0.82), identified (e.g., “Because I value the benefits of drinking”, 3 items, α = 0.72), positive introjected (e.g., “Because I want to feel confident in myself”, 4 items, α = 0.86), external (e.g., “Because I feel under pressure from others to drink”, 4 items, α = 0.75), and amotivated (e.g., “Honestly, I don’t know why I drink”, 4 items, α = 0.78) behavioral regulations (Courtney, Russell and Conroy, 2025). Participants were asked: “Why do I drink?”, and they rated each item on a 5-point Likert scale from (0) not true for me to (4) very true for me. In this manuscript, we used the same sample as the validation/cross-validation study (Courtney, Russell and Conroy, 2025) and calculated means and z-scores for each of the five factors for these analyses.

Alcohol Use.

Participants reported their typical weekly alcohol use via the Daily Drinking Questionnaire (Collins, Parks and Marlatt, 1985), with a standard drink defined in writing as 12 ounces (oz.) of 5% beer, 8–9 oz. of 7% beer, 4–5 oz. of wine, a 1.5 oz. shot of liquor (straight or in a mixed drink), or 12 oz. of hard seltzer (accompanied by a picture). Participants reported their typical number of drinks consumed and hours they drank on each day of the week, which was used to calculate typical weekly drinking volume (i.e., drinks/week) and frequency (i.e., days/week) and drinking intensity (i.e., drinks/day = drinking volume × drinking frequency).

Negative and Positive Drinking Consequences.

Negative drinking consequences were assessed using the 24-item Brief Young Adult Alcohol Consequences Questionnaire (Kahler, Strong and Read, 2005). Positive drinking consequences were assessed using 15-items from the Importance of Consequences of Drinking short form (Maggs, 1993; Patrick and Maggs, 2011). Participants responded ‘Yes’ or ‘No’ for having experienced each consequence, with ‘Yes’ responses summed to calculate the total number of negative or positive consequences.

Demographics.

Participants self-reported age, biological sex, ethnicity, race, education, work status, and student status.

Statistical Analyses

Latent profile analysis was used to identify groups (i.e., profiles) of participants with similar behavioral regulations for general alcohol drinking behaviors, and sociodemographic and drinking characteristics were investigated as predictors or outcomes of profile membership, respectively. The sample size for this study was based on heuristic sample size requirements for psychometric testing of surveys (Cohen, 1992; Courtney, Russell and Conroy, 2025). Vermunt’s three-step approach was used to investigate sociodemographic predictors (i.e., sex, age) of profile membership and the BCH three-step approach was used to investigate profile membership as a predictor of drinking-related outcomes (i.e., drinking frequency, drinking intensity, negative consequences, and positive consequences) (Bolck, Croon and Hagenaars, 2004; Vermunt, 2010). The three-step approach ensures that latent profile membership is not affected by covariates by fixing measurement parameters of the latent profile variable with covariates of the model at values from the unconditional latent profile model and, in BCH analyses, using a weighted multiple group analysis to avoid shifts in profile membership (Bolck, Croon and Hagenaars, 2004; Vermunt, 2010; Asparouhov and Muthén, 2014). The first step is estimating the measurement model without external variables (Research Question 1), with the model of best fit selected based on quantitative fit statistics (described below) and qualitative fit based on the alignment between the characteristics of each profile with theory and prior research (Nylund, Asparouhov and Muthén, 2007; Richards, Pearson and Field, 2020). The second step is assigning participants to their predicted latent profiles based on modal posterior probabilities. The third step is estimating the models of interest using the assigned profiles (Vermunt, 2010) (Research Questions 2 and 3).

In the first step we conducted a latent profile analysis (LPA) on z-scores for the five CRAI-Drinking subscales. Nested models with one to five profiles were compared based on several fit indices, because no single statistical indicator is sufficient for identifying which model fits best (Nylund, Asparouhov and Muthén, 2007). Fit indices included indicators for parsimony (e.g., Bayesian Information Criterion [BIC], Akaike Information Criterion [AIC], bootstrapped log-likelihood ratio test [BLRT]), clustering criteria (e.g., classification likelihood information criterion [CLC]) to estimate how accurately the model classifies participants into each profile, and parsimony plus clustering criteria (i.e., integrated classification likelihood with BIC approximation [ICL-BIC]). The BLRT was selected over the Lo-Mendell-Rubin LRT because it is more robust and can be performed on models with different model constraints (Nylund, Asparouhov and Muthén, 2007). We compared models with different assumptions for the profile covariance matrices to find the best fitting model that balances flexibility with parsimony, including models that were unrestricted (LPA-Free), imposed homoscedasticity (LPA-Same) (same covariance matrix for all profiles), and imposed proportionality (LPA-Prop) (equal correlations, proportional variances).

For research question 2, the structural model in the third step includes the covariate of interest (i.e., sex, age) as a predictor of latent profile membership within a multinomial logistic regression framework by using the assigned latent profile, which was fixed based on the results from the second step, as the sole indicator variable (Vermunt, 2010). Results of these models are odds of latent profile membership, which are reported as odds ratios (OR) and 95% Confidence Intervals (CI).

For research question 3, the structural model in the third step includes the assigned latent profile, which was fixed based on the results from the second step, as the sole indicator variable predicting drinking-related outcomes (Bolck, Croon and Hagenaars, 2004). The assigned profile is entered as a nominal indicator variable, thereby fixing membership to account for potential errors in latent profile assignment (a common problem with 1-Step approaches) (Asparouhov and Muthén, 2014). The model indicates whether one profile differs significantly from other profiles on each outcome. Results are reported as Chi-Square tests, including an omnibus test and pair-wise comparisons between the latent profiles for mean values of the outcome of interest, with Cohen’s d reported as an effect size estimate. Analyses were conducted in Mplus Version 8.9 (Muthen and Muthen, 2017) and statistical significance was set at p < .05.

Results

Participant Characteristics

The combined sample (N = 630) had a mean age of 21.5 ± 4.7 years (range: 18 – 57), was 55.1% female, 83.3% White, and 90.6% non-Hispanic. Participants drank alcohol on 3.2 ± 1.5 days per week, consumed 8.5 ± 4.9 drinks per day, and experienced 7.0 ± 5.8 negative and 8.5 ± 4.9 positive drinking consequences. Table 1 provides descriptive statistics for the sample.

Table 1.

Sample characteristics

Demographic Characteristics	Sample (N = 630)
Race (n (%)) a
White	525 (83.3)
Asian	48 (7.6)
Black	14 (2.2)
American Indian/Alaskan Native	2 (0.3)
Other	5 (0.8)
Mixed	27 (4.3)
Ethnicity (n (%)) b
Non-Hispanic	49 (7.8)
Hispanic	571 (90.6)
Education level (n (%)) c
High school degree or equivalent	227 (36.0)
Some college or Associate’s degree	321 (50.8)
Bachelor’s Degree	43 (6.8)
Post-Graduate Degree	38 (6.0)
Work Status (n (%)) d
Full-time	42 (6.7)
Part-time	2 (0.3)
Unemployed/Looking for Work	1 (0.2)
Student, not working	276 (43.8)
Student, working	306 (48.6)
Other work status	2 (0.3)
Student Type (n (%)) e
Undergraduate Student	555 (88.1)
Post-Graduate Student	24 (3.8)
Non-Student	45 (7.1)
Alcohol Use Characteristics	Mean ± SD
Alcohol Use – Typical Week
Total Drinks per Week f	15.2 ± 11.8
Days per Week	3.2 ± 1.5
Drinks per Day	8.5 ± 4.9
Negative Drinking Consequences	7.0 ± 5.8
Positive Drinking Consequences g	8.5 ± 4.9
CRAI-Drinking h
Intrinsic Regulation	2.8 ± 0.9
Identified Regulation	1.0 ± 1.0
Positive Introjected Regulation	1.1 ± 1.0
External Regulation	0.9 ± 0.8
Amotivation	0.8 ± 0.8

Notes: SD = standard deviation; Min = minimum; Max = maximum; 1QR = first quartile; 3QR = third quartile. CRAI-Drinking = Comprehensive Relative Autonomy Index for Drinking.

^aMissing data: n = 9.

^bMissing data: n = 10.

^cMissing data: n = 1. Some college or AA degree includes some college education, Associate’s degree, or trade school. Post-Graduate Degree includes Master’s degree, Doctoral degree, and professional degree (e.g., medical doctor).

^dMissing data: n = 1. Student, not working includes undergraduate and graduate students who are not working any part-time or full-time job. Student, working includes undergraduate and graduate students who are also working a part-time or full-time job.

^eMissing data: n = 6. Post-Graduate Student includes graduate and professional students.

^fSum of drinks per day for typical week, with outliers (n = 12) winsorized to 3 SDs above the mean.

^gNumber of positive consequences over the past month only collected in the second sample (n = 356).

^hCronbach’s alpha (α) and 95% Confidence Intervals (CI). Intrinsic regulation α = 0.82 (95% CI: 0.78, 0.84). Identified regulation α = 0.72 (95% CI: 0.67, 0.76) (excludes IDENT_1 item). Positive introjected regulation α = 0.86 (95% CI: 0.83, 0.87). External regulation α = 0.75 (0.71, 0.78). Amotivation α = 0.78 (0.74, 0.82).

Latent profiles for self-determined drinking motives

Models with 1- to 5-profile solutions and different assumptions for the covariance matrices were examined. Table 2 shows the fit statistics for the 11 models that converged and for which the best log-likelihood values were replicated. When comparing fit statistics between the models, the four-profile LPA-Prop model provided the best empirical fit across all fit statistics, with the exceptions of ICL-BCL and entropy. The three-profile LPA-Prop model had better ICL-BCL and entropy than the 4-profile model; however, the BLRT indicated that the 4-profile model showed statistically significantly better model fit than the 3-profile model. The 4-profile model also demonstrated strong theoretical fit, with each profile representing qualitatively different groups of individuals based on their behavioral regulations for drinking. After considering both empirical and theoretical fit, the 4-profile solution with proportional covariance matrices was determined most optimal, because it showed the best empirical fit based on the 7 of the 9 tests of model fit and made sense theoretically. Supplemental Table A reports the mean and standard deviation for each indicator variable within each latent profile and the intercorrelations. There were strong intercorrelations between the mean and corresponding standard deviation (r ≥ 0.98) for all behavioral regulations except intrinsic regulation (r = 0.32).

Table 2.

Summary of model fit for latent profile models ^{a, b}

			Model Fit Parameters
Assumptions	Profiles	Parameters	LL	AIC	BIC	CAIC	ssBIC	CLC	ICL-BIC	Entropy	BLRT (p)	n smallest group
Free c	1	10	−4458.641	8937.28	8981.74	8991.74	8949.99	8917.28	8981.74	-	-	-
	2	21	−3934.042	7910.08	8003.44	8024.44	7936.77	8002.58	8137.94	0.846	<.0001	239
Proportional Variances d	1	10	−4458.641	8937.28	8981.74	8991.74	8949.99	8917.28	8981.74	-	-	-
	2	17	−4038.784	8111.57	8187.15	8204.15	8133.17	8240.01	8349.59	0.814	<.0001	290
	3	24	−3917.473	7882.95	7989.64	8013.64	7913.45	8075.81	8230.50	0.826	<.0001	82
	4 e	31	−3841.912	7745.82	7883.64	7914.64	7785.22	8038.41	8238.23	0.797	<.0001	75
Same Variances	1	10	−4458.641	8937.28	8981.74	8991.74	8949.99	8917.28	8981.74	-	-	-
	2	16	−4142.209	8316.42	8387.55	8403.55	8336.75	8418.04	8521.17	0.847	<.0001	179
	3	22	−4068.210	8180.42	8278.23	8300.23	8208.38	8410.50	8552.31	0.802	<.0001	101
	4	28	−3996.724	8049.45	8173.93	8201.93	8085.03	8299.13	8479.61	0.825	<.0001	56
	5	34	−3944.272	7956.54	8107.70	8141.70	7999.75	8243.43	8462.58	0.825	<.0001	22

Notes: LL = Log-likelihood. AIC = Akaike Information Criterion. BIC = Bayes’ Information Criterion. CAIC = Consistent AIC. ssBIC = Sample size adjusted BIC. CLC = Classification likelihood criterion. ICL-BIC = Integrated Completed Likelihood Criterion with BIC approximation. BLRT = Bootstrapped log-likelihood ratio test.

^aN = 630.

^bBolded values are the best values across all models. Lower values are better for all model fit parameters except for entropy, for which a higher value is best.

^cModels with 3 or more classes failed to converge.

^dModel with 5 classes failed to converge.

^eSensitivity analyses were run, only including participants who were 18–25 years (N=582).

The 4-class model with proportional variances was still the best fitting model: LL=−3578.512, AIC=7219.024, BIC=7354.384, CAIC=7385.38, ssBIC=7255.971, CLC=7484.59, ICL-BIC=7681.95, entropy=0.836.

Profile 1 (n = 75, 11.9%) included individuals with below average scores across all five behavioral regulations for drinking, and particularly low intrinsic regulation (z-score = −1.58), making it unclear why they chose to consume alcohol; therefore, we labeled Profile 1 individuals as Aimless Drinkers. Profile 2 (n = 264, 41.9%) included individuals with average levels across all five behavioral regulations for drinking, suggesting that they were equally motivated to drink by a variety of factors; therefore, we labeled Profile 2 individuals as Flexible Drinkers. Profile 3 (n = 177, 28.1%) included individuals with above average motivation for external forms of behavioral regulations, including external regulation (z-score = 1.15) and positive introjected regulation (z-score = 1.20); therefore, we labeled Profile 3 individuals as Externally Controlled Drinkers. Profile 4 (n = 114, 18.1%) included individuals with below average levels of all behavioral regulations except for intrinsic regulation (z-score = 0.33), suggesting that they primarily consumed alcohol due to inherent enjoyment or pleasure associated with drinking; therefore, we labeled Profile 4 individuals as Pleasure Drinkers. Figure 2 shows z-scores on the five behavioral regulations for drinking for each profile.

Figure 2.

z-scores on the five behavioral regulations for drinking for the 4-class latent profile model.

Sociodemographic predictors of latent profile membership

Figure 3 shows the model-estimated proportion of participants in each profile by sex. As shown in Table 3, sex and age significantly predicted odds of latent profile membership. Specifically, compared to males, females had significantly lower odds of being Aimless Drinkers versus Flexible Drinkers (OR = 0.43, 95% CI: 0.22, 0.82), Externally Controlled Drinkers (OR = 0.41, 95% CI: 0.21, 0.80), or Pleasure Drinkers (OR = 0.41, 95% CI: 0.19, 0.88). There were no other significant differences in the odds of profile membership by sex.

Figure 3.

Figure 3 shows the proportion of participants estimated to be in each latent profile by biological sex.

Table 3.

Three-Step Results for Predictors: Multinomial logistic regression models regressing odds of profile membership on sociodemographic characteristics (Research Question 2) ^a

Profile 1 – Aimless Drinkers (n = 75) (reference)	b	SE	Odds Ratio	95% CI
Profile 2 – Flexible Drinkers (n = 264)
Intercept	1.79	0.31***	-	-
Female	−0.85	0.34*	0.42	0.22, 0.82
Age (in years)	0.02	0.04	1.02	0.95, 1.09
Profile 3 – Externally Controlled Drinkers (n = 177)
Intercept	1.60	0.31***	-	-
Female	−0.90	0.34**	0.41	0.21, 0.80
Age (in years)	−0.04	0.04	0.97	0.90, 1.05
Profile 4 – Pleasure Drinkers (n = 114)
Intercept	0.85	0.35*	-	-
Female	−0.89	0.39*	0.41	0.19, 0.88
Age (in years)	0.04	0.04	1.04	0.96, 1.12
Profile 2 – Flexible Drinkers (n = 264) (reference)	b	SE	Odds Ratio	95% CI
Profile 1 – Aimless Drinkers (n = 75)
Intercept	−1.79	0.31***	-	-
Female	0.85	0.34*	2.35	1.21, 4.55
Age (in years)	−0.02	0.04	0.98	0.92, 1.05
Profile 3 – Externally Controlled Drinkers (n = 177)
Intercept	−0.20	0.20	-	-
Female	−0.04	0.24	0.96	0.60, 1.53
Age (in years)	−0.05	0.03	0.95	0.89, 1.01
Profile 4 – Pleasure Drinkers (n = 114)
Intercept	−0.94	0.23***	-	-
Female	−0.04	0.29	0.96	0.55, 1.70
Age (in years)	0.02	0.03	1.02	0.97, 1.07
Profile 3 – Externally Controlled Drinkers (n = 177) (reference)	b	SE	Odds Ratio	95% CI
Profile 1 – Aimless Drinkers (n = 75)
Intercept	−1.59	0.31***
Female	0.90	0.34**	2.45	1.23, 4.79
Age (in years)	0.03	0.04	1.03	0.95, 1.12
Profile 2 – Flexible Drinkers (n = 264)
Intercept	0.20	0.20
Female	0.04	0.24	1.04	0.66, 1.66
Age (in years)	0.05	0.03	1.05	0.99, 1.12
Profile 4 – Pleasure Drinkers (n = 114)
Intercept	−0.74	0.23**
Female	0.01	0.28	1.01	0.58, 1.74
Age (in years)	0.07	0.03*	1.07	1.00, 1.14

Notes:

^*p < .05.

^**p < .01.

^***p < .001.

CI = Confidence Interval. b = Unstandardized beta coefficient. SE = standard error.

^aN = 630.

Compared to younger individuals, older individuals had significantly higher odds of being Pleasure Drinkers versus being Externally Controlled Drinkers, with each additional year in age above 18 increasing the odds of being a Pleasure Drinker by 7% (OR = 1.07, 95% CI: 1.00, 1.14). For example, compared to an 18-year-old, a 19-year-old had 7% higher odds of being a Pleasure versus an Externally Controlled Drinker, whereas a 28-year-old had 99% higher odds of being a Pleasure versus an Externally Controlled Drinker. There were no other significant differences in odds of profile membership by age.

Latent profile membership and drinking-related outcomes

Figures 4a – 4d show the model-estimated mean values for alcohol consumption and outcomes for each latent profile. As shown in the figures, the four profiles differed significantly regarding drinking frequency and intensity and number of negative and positive drinking consequences. Supplemental Table B shows the results of the Chi-square tests and pairwise comparisons. For drinking frequency, the omnibus Chi-square test was statistically significant (χ² = 61.606, p < .001). Pairwise comparisons indicated that individuals belonging to the Aimless Drinker profile drank alcohol on significantly fewer days/week (Mean [M] ± standard error = 2.19 ± 0.15) compared to individuals belonging to the Flexible Drinker (3.34 ± 0.11, p < .001, d = 0.73), Externally Controlled Drinker (3.60 ± 0.14, p <. 001, d = 0.91), and Pleasure Drinker (2.91 ± 0.15, p = .001, d = 0.51) profiles. Similarly, individuals belonging to the Pleasure Drinker profile drank alcohol on significantly fewer days/week compared to individuals belonging to the Flexible Drinker (p = .032, d = 0.25) and Externally Controlled Drinker (p < .001, d = 0.41) profiles. There were no significant differences in average drinking frequency between individuals belonging to the Flexible and Externally Controlled Drinker profiles (p = .16, d = 0.15).

Figure 4.

Figures 4a–4d show means and standard errors for each latent profile by drinking frequency (4a), drinking intensity (4b), negative consequences (4c), and positive consequences (4d). Statistically significant pairwise differences between classes are indicated by different superscript letters.

Similar between-profile differences were found with respect to drinking intensity (omnibus χ² = 75.543, p < .001), such that individuals belonging to the Aimless Drinker profile drank at a lower intensity, consuming significantly fewer drinks/day (2.6 ± 0.25) compared to individuals belonging to the Flexible Drinker (4.60 ± 0.18, p < .001, d = 0.76), Externally Controlled Drinker (5.44 ± 0.23, p < .001, d = 1.06), and Pleasure Drinker (4.64 ± 0.30, p < .001, d = 0.75) profiles. Individuals belonging to the Externally Controlled Drinker profile drank at a higher intensity, consuming significantly more drinks/day, compared to individuals belonging to the Flexible Drinker (p = .006, d = −0.29) and Pleasure Drinker (p = .034, d = 0.25) profiles. There were no significant differences in average drinking intensity between individuals belonging to the Flexible Drinker and Pleasure Drinker profiles (p = .841, d = 0.02).

There were significant differences between profiles in the number of negative drinking consequences (omnibus χ² = 106.49, p < .001). Pairwise comparisons indicated that individuals belonging to the Externally Controlled Drinker profile reported significantly more negative drinking consequences (10.63 ± 0.54) compared to individuals belonging to the Aimless Drinker (3.41 ± 0.54, p < .001, d = 1.20), Flexible Drinker (6.40 ± 0.37, p < .001, d = 0.64), and Pleasure Drinker (4.91 ± 0.51, p < .001, d = 0.90) profiles. Individuals belonging to the Flexible Drinker profile also experienced significantly more negative consequences than individuals belonging to the Aimless Drinker (p < .001, d = 0.56) and Pleasure Drinker (p = .028, d = 0.26) profiles. There were no significant differences in negative drinking consequences between individuals belonging to the Aimless Drinker and Pleasure Drinker profiles (p = .053, d = 0.29).

Similar between-profile differences were found with respect to positive drinking consequences (omnibus χ² = 86.331, p < .001), such that individuals belonging to the Externally Controlled Drinker profile reported significantly more positive drinking consequences (7.23 ± 0.51) compared to individuals belonging to the Aimless Drinker (1.52 ± 0.41, p < .001, d = 0.70), Flexible Drinker (4.66 ± 0.40, p < .001, d = 0.37), and Pleasure Drinker (3.27 ± 0.52, p < .001, d = 0.26) profiles. Individuals belonging to the Flexible Drinker profile also experienced significantly more positive drinking consequences than individuals belonging to the Aimless Drinker (p < .001, d = 0.89) and Pleasure Drinker (p = .048, d = 0.46) profiles. Lastly, individuals belonging to the Pleasure Drinker profile experienced significantly more positive drinking consequences than individuals belonging to the Aimless Drinker profile (p = .011, d = 0.38).

Discussion

This study aimed to enhance the existing literature by examining drinking motives from the perspective of SDT and by using a person-centered approach to identify subgroups of adult drinkers who reported similar quantities of the five qualitatively distinct behavioral regulations for drinking. To our knowledge, this is the first study examining latent motivational profiles of behavioral regulations for general alcohol drinking behaviors, which is advantageous over variable-centered approaches due to the ability to capture the complex, higher-order interactions between behavioral regulations. Previous research examined latent profiles of behavioral regulations for responsible drinking (Richards, Pearson and Field, 2020) or examined self-determined drinking motives or behavioral regulations using variable-centered approaches (Knee and Neighbors, 2002; Neighbors et al., 2004; Chawla et al., 2009; Courtney, Russell and Conroy, 2024). We found four profiles of behavioral regulations for drinking in this sample that were uniquely associated with sociodemographic characteristics and alcohol consumption and outcomes. Overall, our findings were consistent with SDT in that the profile characterized by lower autonomy behavioral regulations was associated with worse alcohol consumption and outcomes (Vansteenkiste et al., 2009; Richards, Pearson and Field, 2020; Courtney, Russell and Conroy, 2024). Highly autonomous, intrinsic regulation was a key feature distinguishing between latent profiles and, consistent with SDT, the profile characterized by drinking primarily for intrinsic reasons was associated with more positive drinking experiences (Knee and Neighbors, 2002; Neighbors et al., 2004; Chawla et al., 2009; Courtney, Russell and Conroy, 2024).

There were four distinct motivational profiles of behavioral regulations for drinking in this sample, which is consistent with research in other behavioral domains, like physical activity, academics, and work (Vansteenkiste et al., 2009; Gillet, Morin and Reeve, 2017; Emm-Collison et al., 2020; Moore et al., 2023). The largest profile, Flexible Drinkers, included 41.9% of individuals who demonstrated average levels of all behavioral regulations (Gillet, Morin and Reeve, 2017; Emm-Collison et al., 2020). Another 28.1% of individuals belonged to the Externally Controlled Drinker profile. This profile included high levels of amotivation, external regulation, and positive introjected regulation, all of which reflect minimal autonomy and are considered low quality behavioral regulations (Vansteenkiste et al., 2009). The presence of this specific profile aligned with our hypotheses and with other studies informed by SDT finding ‘Low Quality’ profiles for behaviors such as physical activity and responsible drinking (Vansteenkiste et al., 2009; Gillet, Morin and Reeve, 2017; Emm-Collison et al., 2020; Richards, Pearson and Field, 2020; Moore et al., 2023). Smaller proportions of individuals belonged to either the Pleasure Drinker (18.1%) or Aimless Drinker (11.9%) profiles, both of which were characterized by below average levels of four behavioral regulations but were distinguished from one another by intrinsic regulation. Specifically, Aimless Drinkers showed lower than average intrinsic regulation, whereas the Pleasure Drinkers showed average intrinsic regulation. We interpreted this to indicate that Pleasure Drinkers were motivated primarily by their inherent enjoyment of drinking alcohol, whereas Aimless Drinkers had little enjoyment of drinking and no clear motivation for drinking alcohol. These two profiles differed from profiles in the previous literature, although the Pleasure Drinker profile was somewhat consistent with ‘High Quality’ motivational profiles characterized by higher levels of autonomous regulations for behavior, such as intrinsic (e.g., enjoying a behavior) and identified (e.g., valuing the outcomes of a behavior) regulations (Vansteenkiste et al., 2009; Richards, Pearson and Field, 2020). Overall, each of the profiles found in this sample were consistent with SDT in that the profiles varied by both the type and the quantity of behavioral regulations for drinking.

In general, the four profiles found in this sample differed somewhat from the previous study examining latent profiles of behavioral regulations for responsible drinking (Richards, Pearson and Field, 2020). In their study, Richards et al. (2020) found three latent profiles for responsible drinking, the first being a ‘High Quality’ profile with high levels of autonomous motivation and introjected regulation (e.g., drinking responsibly out of feelings of guilt or shame). The second was a ‘High Quantity’ profile with high levels of all behavioral regulations for responsible drinking (Richards, Pearson and Field, 2020). The third was a ‘Low Quantity’ profile with low levels on all behavioral regulations for responsible drinking (Richards, Pearson and Field, 2020) that overlapped somewhat with our Aimless Drinker profile. Differences between the present study and Richards et al.’s (2020) are likely attributable to measurement differences. Specifically, Richards et al. (2020) assessed behavioral regulations for responsible drinking, as the Treatment Self-Regulation Questionnaire (TSRQ) focuses on engaging in healthy behaviors, like drinking responsibly (Richards, Morera and Field, 2020; Richards, Pearson and Field, 2020). In contrast, the Comprehensive Relative Autonomy Index for Drinking assesses behavioral regulations for engaging in general alcohol drinking behaviors (Courtney, Russell and Conroy, 2025), making it conceptually distinct from behavioral regulations for responsible drinking (Richards, Pearson and Field, 2021). Indeed, behavioral regulations for responsible drinking have been previously shown to be distinct from traditional DMQ drinking motives (Richards, Pearson and Field, 2021), and preliminary evidence suggests that behavioral regulations motives for general alcohol drinking behaviors predict alcohol consumption and consequences (Courtney, Russell and Conroy, 2024). Other measurement differences include the TSRQ assessing four behavioral regulations whereas the Comprehensive Relative Autonomy Index for Drinking assesses five behavioral regulations (Richards, Morera and Field, 2020; Richards, Pearson and Field, 2020; Courtney, Russell and Conroy, 2025). Notably, the TSRQ assesses autonomous motivation by including items for both intrinsic and identified regulation (Richards, Morera and Field, 2020; Richards, Pearson and Field, 2020), whereas the Comprehensive Relative Autonomy Index for Drinking includes separate factors for intrinsic and identified regulations (Courtney, Russell and Conroy, 2025). The latent profiles in this sample for Pleasure and Aimless Drinkers were distinguishable from one another based solely on their levels of intrinsic regulation, and it is possible that combining intrinsic and identified regulation into a single factor would have resulted in one latent profile like Richard et al.’s (2020) Low Quality profile. However, the value intrinsic regulation brought in distinguishing between subgroups supports examining the unique contribution of each individual behavioral regulation for drinking and aligns with SDT by accounting for the underlying relative autonomy continuum and organismic integration theory (Ryan et al., 2008; Ryan and Deci, 2017). This distinctive and important role of intrinsic regulation is also supported by findings related to differences in latent profile membership by sociodemographic characteristics and drinking-related outcomes.

Differences in profile membership existed for sex and age. Compared to males, females had lower odds of being in the Aimless Drinker profile versus the other three profiles, corresponding with prior literature finding sex differences in drinking motives (Cooper, 1994; Knee and Neighbors, 2002; Kuntsche et al., 2006; Stapinski et al., 2016). These differences may relate to sex-specific drinking norms and expectations, such as the normative belief that men drink more than women (Knee and Neighbors, 2002). Although the Aimless Drinkers in this sample engaged in the lowest frequency and intensity of alcohol use, they also showed significantly lower than average intrinsic motivation, suggesting that males may be more likely than females to consume at least some alcohol in response to peer norms and expectations despite not enjoying consuming alcohol. Additional research should investigate how sex, peer norms/expectations, and drinking motives interact in predicting alcohol consumption and outcomes.

Intrinsic regulation also factored into age differences in profile membership. It is notable that age differences were detected given that the range of ages was restricted with most participants being young adults aged 18–29 years. Specifically, older participants were more likely than younger participants to be in the Pleasure Drinker versus the Externally Controlled Drinker profile, indicating that older individuals had higher odds of drinking for autonomous versus controlled reasons. This aligns with prior research finding age-related differences in drinking motives (Patrick and Schulenberg, 2011; Littlefield et al., 2013; Patrick et al., 2017). It also aligns with the general progression of human development because autonomy increases with age, particularly during emerging adulthood (Wray-Lake, Crouter and McHale, 2010), and increased autonomy is a feature of transitioning towards adulthood (Arnett, 2001). Central to SDT is the concept that motives characterized by greater autonomy correspond with better health and positive behavioral outcomes (Ryan and Deci, 2017). As such, SDT promotes transitioning towards more autonomous self-determined behavioral regulations (i.e., internalization) as a mechanism to promote positive behavior change (Ryan et al., 2008; Ryan and Deci, 2017). Our findings suggest that this process of internalization may occur naturally with increased age in some individuals, as would be expected with general human development (Arnett, 2001; Wray-Lake, Crouter and McHale, 2010), and may help partly explain the typical transition away from high-risk alcohol consumption across the late 20s (Littlefield et al., 2013; Patrick et al., 2017). Promoting internalization of behavioral regulations for drinking at an earlier age, particularly by emphasizing intrinsic regulation, may be a useful strategy for harm reduction behavior change interventions to help hasten an earlier transition towards lower risk alcohol consumption in emerging adulthood.

The associations between profile membership and alcohol consumption and outcomes in this sample aligned with SDT and our hypotheses, such that membership in the Externally Controlled Drinker profile was associated with greater drinking frequency and intensity and with experiencing more negative drinking consequences. This is consistent with prior research finding that high amotivation for drinking is associated with more negative consequences after accounting for the amount of alcohol consumed (Courtney, Russell and Conroy, 2024), and drinking due to controlled motives or an external locus of control corresponds with greater alcohol use, worse drinking-related outcomes, a higher likelihood of hazardous drinking, and higher levels of alcohol dependence (Knee and Neighbors, 2002; Neighbors, Walker and Larimer, 2003; Neighbors et al., 2004; Chawla et al., 2009; Bhowmick, Kulkarni and Gaikwad, 2019; Lassi et al., 2019; Courtney, Russell and Conroy, 2024). Also consistent with prior research, membership in the Pleasure Drinker profile was associated with less alcohol use and better drinking-related outcomes (Knee and Neighbors, 2002; Neighbors et al., 2004; Chawla et al., 2009; Courtney, Russell and Conroy, 2024). Overall, membership in the Pleasure Drinker profile appeared to correspond with more enjoyable drinking experiences. Participants in the Pleasure Drinker profile consumed less alcohol and experienced fewer negative consequences than the Externally Controlled and Flexible Drinker profiles. Notably, despite consuming more alcohol than Aimless Drinkers, Pleasure Drinkers experienced the same number of negative consequences and more positive consequences than the Aimless Drinker profile. Overall, this finding suggests that drinking primarily for intrinsic reasons may facilitate enjoyable drinking experiences over the short-term (Courtney, Russell and Conroy, 2024) and may support transitions towards better alcohol consumption and outcomes. Future research should investigate how motivational profiles of behavioral regulations for drinking predict transitions towards lower risk alcohol consumption and/or towards incident alcohol use disorder (Bates, 2000; Patrick and Schulenberg, 2011; Howard and Hoffman, 2018). Such information could inform harm reduction interventions targeting individuals with high-risk motivational profiles for drinking.

Limitations of the present study should be acknowledged. First, the study was limited to two convenience samples who were predominantly young adults 18–29 years (93.7%), White, non-Hispanic undergraduate students. The sensitivity analyses indicated that the 4-class model fit well in a subsample of young adults 18–25 years (Table 2); however, more diverse samples should be tested to determine if the four latent profiles found in this sample are replicable and to provide additional information about sociodemographic and cultural differences between profiles, such as in countries with lower drinking ages or different drinking norms (Rahav et al., 2006; Nordlund and Østhus, 2013; Room et al., 2019), as research suggests that drinking motives mediate cultural differences in alcohol use among adolescents (Kuntsche et al., 2015). Comparisons of behavioral regulations for general alcohol drinking behaviors versus responsible drinking are also warranted (Richards, Morera and Field, 2020; Richards, Pearson and Field, 2020, 2021). Third, the cross-sectional nature of the study limits causal inferences and does not allow for examining the long-term effects of motivational profiles on alcohol consumption and outcomes. Prospective, longitudinal studies are needed to examine the predictive value of these profiles, how they relate to changes in alcohol consumption and outcomes across time and across developmental transitions (Bates, 2000; Patrick and Schulenberg, 2011; Howard and Hoffman, 2018), and how they relate to maturing out of problem drinking (Patrick and Schulenberg, 2011; Littlefield et al., 2013; Patrick et al., 2017). Finally, participants self-reported their alcohol consumption, introducing the risk of recall and social desirability biases (Davis, Thake and Vilhena, 2010). Collecting complementary data via device-based measures of alcohol use would provide valuable information about how the dynamics of participants’ alcohol exposure, such as peak intoxication and total alcohol exposure (Russell, Turrisi and Smyth, 2022), relate to motives.

Conclusion

The present study used a person-centered approach to provide novel information about self-determined behavioral regulations for drinking, contributing to the calls for research examining drinking motives across ages and for using SDT as a framework for understanding alcohol use. Our findings supported the presence of four motivational profiles for drinking that differed in terms of sociodemographic factors and that were associated with alcohol consumption and consequences in a manner consistent with SDT and prior research. Older individuals were more likely to drink due to intrinsic regulation, which may reflect a shift towards greater autonomy over alcohol consumption with age. Drinking primarily for intrinsic reasons also facilitated more enjoyable drinking experiences. Future research should investigate whether motivational profiles of behavioral regulations for drinking predict long-term trajectories of alcohol consumption and alcohol-related risks to inform harm reduction interventions.

80-Word Overview of Key Findings and Significance.

This study identified four motivational profiles of self-determined behavioral regulations for drinking alcohol in young-to-middle-aged adults (N=630, ages 18–57, 55% female). Drinking for enjoyment facilitated more positive outcomes and was more common for older participants, which may reflect a shift towards greater control over drinking with age. In contrast, drinking due to social pressure corresponded with more alcohol use and negative outcomes. Future research should investigate whether motivational profiles for drinking predict long-term trajectories of alcohol consumption and risks.