Person- and Context-level Moderators of State Alcohol Policy Environment Effects on Binge Drinking Among U.S. 4-year College Students, 2008-2019

Abstract

Introduction:

Binge drinking remains a serious issue among college students. While prior works showed more restrictive state alcohol policy environments are associated with less binge drinking in college students, the association may depend on individual, contextual, and institutional factors specific to this population.

Method:

Data were from repeated cross-sectional surveys among 902486 college students ages 18-24 years from 591 four-year institutions in 47 states during 2008-2019. Time-varying, state-level Alcohol Policy Scale (APS) scores and fourteen moderators were examined in relation to students’ past 2-week binge drinking (5+ drinks in a sitting).

Results:

Associations between higher APS scores (more restrictive state policies) and lower odds of binge drinking were moderated by several factors representing contextual features of college. For example, higher APS scores were associated with lower odds of binge drinking among students living on campus or with their parents and by those unaffiliated with the Greek system, but not among students in off-campus or Greek housing or members of the Greek system. Negative associations between APS and binge drinking were stronger for dating and cohabitating than single students. Additionally, APS scores were significantly more negatively associated with binge drinking for Asian than for non-Asian students. At the institution level, APS scores were negatively associated with students’ binge drinking at institutions in medium to large, but not small population centers.

Conclusions:

The identification of factors that moderate effects of alcohol policy environment on college binge drinking informs potential prevention approaches that are more culturally tailored and targeted to subgroups of interest.

Introduction

Binge drinking remains an important issue for young adults in college with a recent estimated prevalence of 28.9% among full-time college students ages 18 to 22 (U.S. Department of Health and Human Services, Substance Abuse and Mental Health Services Administration, Center for Behavioral Health Statistics and Quality, 2022), despite the steady decrease over the last two decades. As excessive and harmful alcohol use is a serious threat to public health given the immediate and long-term risks to the lives of drinkers and those around them (Alpert et al., 2022; Esser et al., 2024; Karriker-Jaffe et al., 2017), this population continues to be a critical target for prevention. Purchasing and consuming alcohol becomes legal during early adulthood, which also coincides with the peak prevalence of binge drinking (Patrick et al., 2024). Alcohol use patterns during college years predict future problem use including alcohol dependence and abuse (Jennison, 2004). Multiple layers of policies and programs are needed to reduce binge drinking among young adults in college, and studies support the effects of campus- (Ringwalt et al., 2011; Toomey et al., 2007), community- (Fagan et al., 2011; Holder, 2004), and state-level (Chaloupka & Wechsler, 1996; Nelson et al., 2005) efforts. However, prior evaluations of the protective potential of state-level alcohol policies for college students date back more than two decades.

In a recent analysis (Kerr, Naimi, et al., 2025), we updated and extended prior literature to show a significant association between state alcohol policy environments and binge drinking among college students in the U.S.. Specifically, we analyzed the Alcohol Policy Scale (APS) score, which quantifies the efficacy and implementation of 29 alcohol policies in every state-year and has been updated annually since 1999 (Blanchette et al., 2020; Naimi et al., 2014). In more than 900,000 students enrolled in colleges in 47 U.S. states, we found that more restrictive state alcohol policies were significantly associated with lower individual-level odds of binge drinking. This study supported that—like adults, older adolescents and underage young adults—college students are also responsive to the state alcohol environment (Naimi et al., 2014; Xuan, Blanchette, Nelson, Heeren, et al., 2015; Xuan, Blanchette, Nelson, Nguyen, et al., 2015). That is, it provided evidence that college campuses do not fully insulate students from the state policy environment, highlighting the importance of state policies as a key component of prevention efforts.

Given the significant association between more restrictive state alcohol policies and reduced college binge drinking we sought to determine whether some groups of students are more or less sensitive to the state alcohol policy environment with respect to binge drinking. This focus differs from research on the diverse individual and contextual risk and protective factors for binge drinking in college (Krieger et al., 2018), and instead asks who is or is not reached by state policies. Identifying subgroups at both the individual and contextual levels may inform prevention development as this could provide a basis for tailored and targeted prevention approaches in college programs, or additional state- or community-level responses. Previously, it was shown that the strengths of the relationship between the APS and binge drinking differed for some groups defined by race/ethnicity in adults and youth, whereas effects did not differ by gender (Xuan, Blanchette, Nelson, Heeren, et al., 2015; Xuan, Blanchette, Nelson, Nguyen, et al., 2015). However, analyses of whether the APS is associated differentially across groups of college students have not been conducted previously. Additionally, general population surveys lack information on some of the unique contextual features of campus life that may affect risk for binge drinking, such as being a first-year student, residing off-campus, or living in Greek system housing.

In the present study, we quantified the relationship between a measure of state alcohol policy restrictiveness (i.e. APS) and binge drinking among college students at various levels of potential moderators. Given its large sample of 902,486 college students ages 18-24 surveyed from 2008-2019, this study is well positioned to identify individual-, institution-, and contextual-level moderators of the state alcohol policy environment on binge drinking. First, we hypothesized that race/ethnicity would moderate associations between state APS and students’ binge drinking; specifically, consistent with prior work (Xuan, Blanchette, Nelson, Heeren, et al., 2015; Xuan, Blanchette, Nelson, Nguyen, et al., 2015) we expected APS to be less strongly associated with binge drinking among Hispanic and Black students relative to their peers. Second, we also examined moderation of APS effects by gender. Although this was not supported by previous analysis in adults overall (Xuan, Blanchette, Nelson, Heeren, et al., 2015) and youth (Xuan, Blanchette, Nelson, Nguyen, et al., 2015), it requires further consideration and would be consistent with our prior work that college women were more responsive to cannabis legalization than men. Third, analyses of other moderators were exploratory, given the lack of literature to guide hypotheses. We selected moderators that a) could inform policy responses, b) pertain to identities or contexts known to be associated with risk for binge drinking (e.g., Greek housing and Greek involvement (S. E. McCabe et al., 2018), off-campus housing (Benz et al., 2017)), c) potentially offer protective effects against binge drinking such as romantic relationships (Blumenstock & Papp, 2021; Janota et al., 2024) and living with parents (Hamilton et al., 2021), and/or d) index shorter (e.g., first year students, international students) or longer term exposure to the state policy environment (public school attendees; students living with their family). We additionally conjectured that campus characteristics such as enrollment sizes and population of the city in which the campus is located may affect the extent to which students are insulated from and integrated with the broader drinking population, and thus may have interacting effects with policy environments on student binge drinking.

Materials and Methods

Study Participants

Individual-level data were drawn from repeated cross-sectional National College Health Assessment (NCHA) II surveys conducted from Fall 2008 to Spring 2019 (American College Health Association (ACHA), n.d.). With IRB approval, post-secondary institutions self-select to administer NCHA to a random sample of students in either Fall or Spring terms, and many institutions regularly participate biennially. For the present study, we selected undergraduate participants ages 18-24 years from 4-year institutions. To protect the privacy of institutions, ACHA staff released a de-identified dataset containing the policy data to our team; furthermore, they omitted all data from three states with low populations to protect institution identities. The final dataset included data from 902486 undergraduate respondents ages 18 to 24 years from 591 4-year institutions in 47 states (which may or may not include Washington D.C.).

Measures

Binge drinking outcome

Binge drinking was defined as consuming five or more drinks of alcohol in a sitting at least once (coded 1) or not (0) in the past two weeks. Binge drinking prevalence estimates from NCHA are strongly associated at the college-level with per capita on-campus arrests and disciplinary actions for liquor law violations (Kerr, Kasimanickam, et al., 2025).

Alcohol Policy Scale (APS)

The APS is a continuous scale (range: 0 – 100) that measures the aggregate state-level restrictiveness of the alcohol policy environment where higher APS scores index more restrictive environments. The APS was developed by a panel of 10 policy experts using a modified Delphi approach (Naimi et al., 2014) and calculated for each state-year from 1999-2021(Blanchette et al., 2020). After panelists chose 29 policies considered effective at reducing excessive drinking and related harm among adults and youth, panelists independently rated the efficacy and of each of 29 policies for reducing binge drinking among adults using a 5-point Likert scale (1 = low efficacy, 5 = high efficacy) and then rated its implementation in each state-year based on how broadly applicable, effective, or enforceable each policy was by statutory design. For example, for the Alcohol Beverage Control agency policy (mean efficacy=3.38), states with more agents/police dedicated to enforcement of alcohol laws per licensed outlet received a higher implementation rating than other states. Finally, APS scores were calculated by summing policies present in a particular state-year, after weighting each policy by its efficacy (invariant by state-year) and implementation (which can vary by state-year) ratings. APS was validated based on goodness-of-fit for adult binge drinking and youth drinking (Naimi et al., 2014; Xuan, Blanchette, Nelson, Nguyen, et al., 2015). More details about the construction of the scores and trends over a 20-year period (1999 – 2018) can be found in (Blanchette et al., 2020; Naimi et al., 2014).

Each NCHA respondent observation was matched with the APS score in the preceding year of the state in which the respondent resided. For example, NCHA respondents surveyed during the 2016-2017 academic year were matched with the respondents’ state-specific APS score from 2016 (reflecting policy updates as of June 2016). The change in the odds of binge drinking corresponding to a 10 point increase in APS (which approximately aligns with the interquartile range of state scores) is a meaningful metric used in prior studies (Naimi et al., 2014), and was used to interpret the odds ratios in the present study.

Covariates and moderators

Regression models adjusted for individual-, institution-, and contextual-level covariates.

Individual-level covariates included gender (female, male, or transgender), sexual orientation (heterosexual, bisexual, gay/lesbian, other identities, or unsure), and race/ethnicity indicators. Race/ethnicity comprised seven binary variables: black/African American; Hispanic or Latino/a; Asian or Pacific Islander; American Indian, Alaskan Native, or Native Hawaiian; biracial or multiracial; white; and other race. Students could select multiple responses so that these binary indicators are not mutually exclusive.

Contextual-level covariates the following variables. Relationship status had three categories: not in a relationship, in a relationship and living together, or in a relationship but not living together. Residence types had five categories: on-campus/university housing, off-campus housing, parent/guardian’s home, Greek (e.g., fraternity), and other. Other contextual-level covariates were binary variables: first year in school, legal age, international student, and membership in a fraternity/sorority.

Institution-level covariates included: enrollment size, categorized into <5,000, 5,000 – 20,000, and >20,000 students; city/town population, categorized into <10,000, 10,000 – 250,000, and >250,000]; and a binary indicator for private or public institution. Institutions elect to administer NCHA in Fall or Spring term; given that drinking patterns vary by term (whereas alcohol policies do not), we adjusted for time of year the survey was administered. Many institutions participate on different biannual schedules (odd or even year), which was also controlled.

Among these covariates, we selected gender, sexual orientation, and race/ethnicity indicators as potential individual-level moderators. Note that moderation of APS effects by legal age (minor vs. non-minor) was reported in (Kerr, Naimi, et al., 2025). Due to small counts of transgender students (0.7% in the present data), we only included males and females for gender in the moderation analysis; all genders and response options on the gender covariate were included in all other analyses, however. For similar reasons, we only examined the following race/ethnicity groups as moderators: black/African American, Hispanic or Latino/a, and Asian or Pacific Islander, and white. For contextual-level moderators, we chose relationship status, residence type, first year in school, international student, and membership in a fraternity/sorority. Finally, enrollment size, city/town population and private/public institution were selected as institution-level covariates.

Statistical Analysis

Prior to conducting the main moderation analysis, we ran a multivariate logistic regression model that included all of the covariates, but not the APS scores, to first identify and confirm risk and protective factors for binge drinking. Then, for each moderator, we ran a logistic regression which was consistent with a multi-level model in which students (level 1) are clustered in institutions (level 2); the models were of the following form:

$$logit(p)={\beta }_0+{\beta }_{1}AP{S}_{ijk}+{\beta }_{2}M+{\beta }_{3}M\ast AP{S}_{ijk}+{\beta }_{4}AY+{\beta }_{5}A{Y}^2+\sum _{l=6}^L{\beta }_l{X}_l+{\gamma }_j$$

In the model formulation, $p$ is the probability of the outcome variable $y_{ijk}$, where $y_{ijk}$ represents the binary outcome variable for the i^th individual nested in j^th institution in k^th state-year. $M$ represents the moderator of interest and $AY$ represents the academic year, which includes both the linear and quadratic terms. $X_l$ represents the observed values for covariates in the data (not including the moderator), and ${\gamma }_j$ represents the random intercept term for each institution to account for nesting of individuals within the same institutions. To adjust for multiple testing, we used a Bonferroni-corrected two-tailed significance level of 0.05/14 = 0.00357 for the tests of interaction terms. In models with significant interactions, we estimated and tested the simple effects of APS (i.e. effects of APS in each stratum of the moderator).

Results

Table 1 shows the descriptive statistics of the individual-, institution-, and contextual-level covariates for the participants included in the analysis. Table 2 reports prevalence of binge drinking and the mean APS scores by academic year. Students’ binge drinking prevalence shows a decreasing trend from 38.6% in 2008-2009 to 29.0% in 2018-2019. The results of the logistic regression model that included the entire set of covariates are shown in Supplementary Table 1.

Table 1.

Descriptive Statistics on Student- And Institution-Level Variables from The National College Health Assessment, 2008-2009 to 2018-2019 Academic Years.

Variable	Category	%
Sex/Gender	Female; not transgender	67.6
	Male; not transgender	31.7
	Transgender	0.7
Sexual Orientation	Heterosexual	88.1
	Gay/Lesbian	2.6
	Bisexual	5.5
	Other Identity	1.7
	Unsure/Questioning	2.1
White1		72.7
Black1		6.2
Hispanic1		10.8
Asian1		11.8
Native American1		1.8
Biracial1		4.3
Other race1		2.5
Legal Drinking Age, 21-24 (vs. 18-20) years		36.9
First Year Student		29.2
International Student		6.2
Fraternity Involvement		12.1
Relationship Status	Not in a relationship	55.3
	In a relationship, do not live together	37.8
	In a relationship, live together	6.9
Residence	On Campus	53.7
	Greek Housing	1.7
	Off-campus	29.7
	Parent Home	13.1
	Other	1.8
Private (vs. Public) Institution		37.3
Spring (vs. Fall) Survey Administration		75.4
College Enrollment Size	< 5000	21.8
	5000 - 20000	42.2
	> 20000	36.0
Population of City in which College is Located	<10000	8.5
	10000 - 249000	61.5
	> 250000	29.9
College Region	West	28.1
	Midwest	19.7
	Northeast	24.8
	South	27.5

¹Racial and ethnic identities are not mutually exclusive. Percentages indicate selecting vs. not selecting each identity (e.g., Black vs. not Black).

Table 2.

Descriptive statistics on student binge drinking from the National College Health Assessment and Alcohol Policy Scale (APS) scores by academic year.

Academic Year	2-week Binge Drinking	State Alcohol Policy Scale
	(%)	Mean*	STD
2008-2009	38.6%	43.3	8.9
2009-2010	37.0%	43.8	8.1
2010-2011	36.6%	43.4	8.8
2011-2012	36.6%	42.9	7.8
2012-2013	35.5%	44.4	8.7
2013-2014	36.2%	43.1	8.4
2014-2015	34.8%	43.5	8.8
2015-2016	34.4%	42.8	6.8
2016-2017	34.3%	43.6	8.4
2017-2018	29.9%	43.6	7.1
2018-2019	29.0%	43.4	8.3

Notes. AY = academic year; STD = standard deviation.

^*The mean APS was calculated based on the number of respondents in the NCHA

In the main interaction analysis, five moderators (race-Asian, residence type, relationship status, membership in fraternity/sorority, and population size) were significant at the Bonferroni-corrected significance level (p<0.00357), and two additional moderators (race-Hispanic and international student) were nominally significant (p<0.05). The results along with the estimated simple effects within each stratum for these seven moderators are summarized in Table 3. These simple effects also are illustrated in Figure 1.

Table 3.

Adjusted simple effects of Alcohol Policy Scale (APS) on individual-level odds of binge drinking in each stratum of significant moderator variables

Moderator	Categories	OR	95% LL	95% UL	p-value from test of simple effectsa	p-value from test of interaction
Race - Hispanics	Non-Hispanics	0.938	0.903	0.974	0.0009	0.0292**
	Hispanics	0.964	0.923	1.008	0.1079
Race - Asians	Non-Asians	0.940	0.905	0.976	0.0012	0.0006*
	Asians	0.899	0.860	0.940	<.0001
International Student	Domestic	0.939	0.904	0.975	0.001	0.0221**
	International	0.907	0.866	0.951	<.0001
Residence Type	On-campus Housing	0.944	0.909	0.981	0.0032	<0.0001*
	Greek Housing	0.976	0.920	1.036	0.423
	Off-campus Housing	0.967	0.930	1.005	0.0861
	Parent Home	0.888	0.850	0.928	<.0001
	Other	0.954	0.900	1.011	0.1101
Relationship Status	Not in relationship	0.957	0.922	0.995	0.0251	<0.0001*
	In relationship, Not live together	0.930	0.895	0.967	0.0003
	In relationship, Live Together	0.910	0.871	0.951	<.0001
Fraternity/sorority	Not in Fraternity/Sorority	0.936	0.902	0.973	0.0007	<0.0001*
	In Fraternity/Sorority	1.007	0.966	1.050	0.73
Population Size	Population: < 10k	1.051	0.984	1.122	0.1431	0.0026*
	Population: 10k - 250k	0.943	0.907	0.982	0.0041
	Population: > 250k	0.939	0.891	0.989	0.0185

Reported are the estimated simple effects of APS in each category of moderator variables derived from the interaction model, adjusted for study covariates.

^aSimple effects refer to the estimated effects of APS in each stratum of the moderator

^*The interaction term was statistically significant at the Bonferroni-corrected significance level (p<0.00357)

^**The interaction term was statistically significant at the nominal alpha level of 0.05.

Figure 1.

Illustration of adjusted simple effects of Alcohol Policy Scale (APS) in each stratum of significant moderator variables. The dots represent the estimated odds ratios and whiskers represent the corresponding 95% confidence interval. The vertical dashed line represents the null effect (i.e. odds ratio of 1).

We next probed only the individual-level moderators that yielded significant interactions. First, we observed differential effects of APS on binge drinking among Asian and non-Asian students (p=0.0006). The adjusted odds ratio (AOR) associated with 10 point increase in the APS score for Asian and non-Asian students were 0.899 (<0.0001) and 0.940 (p=0.0012), respectively, which exhibits that the effect of stricter alcohol policy environments on lower odds of binge drinking was significant in both groups, but was stronger among Asian students. Second, in a comparison of Hispanic and non-Hispanic students, the p-value from the test of interaction was nominally significant (p=0.029). It was observed that the negative relationship between the APS and binge drinking was significant among non-Hispanic students (AOR=0.938, p=0.0009), but not significant among Hispanic students (AOR=0.964, p=0.1079).

Next we examined the nature of the significant contextual-level moderators. First, in an analysis of five residence types among college students, the strength of the negative association between the APS scores and binge drinking was the greatest among students living in their parent/guardian’s home with AOR=0.888 (p<0.0001), followed by students living in on-campus housing with AOR=0.944 (p=0.0032). For the remaining housing types (off-campus housing, Greek housing¹, and other housing), higherAPS scores were associated with lower odds of binge drinking, but these results were not statistically significant. Second, on the basis of relationship status among college students, significant negative associations were shown in each stratum with different magnitudes. The AOR’s for binge drinking were 0.910 (p<0.0001), 0.930 (p=0.0003), and 0.957 (p=0.0251) for students in a relationship that live together with their partners, in a relationship that do not live together, and not in a relationship, respectively. Third, the relationship between higher APS scores and lower odds of binge drinking also differed significantly according to whether or not students were members of a fraternity/sorority¹. Specifically, the AOR was significant and lower among students who are not members of a fraternity/sorority (AOR=0.936, p=0.0007), while an association was not detected in members of a fraternity/sorority (AOR=1.007, p=0.73). Fourth, comparing the international and domestic students, the p-value from the test of interaction was nominally significant (p=0.029) although it did not reach the conservative Bonferroni-corrected significance level. The magnitude of the negative association between APS and binge drinking was stronger among international students (AOR=0.907, p<0.0001) than among domestic students (AOR=0.939, p=0.001).

Finally, the population size of the city in which the college is located was a significant moderator of the relationship between the higher APS scores and lower odds of binge drinking. Significant and similar AOR’s were observed in medium (AOR=0.943, p=0.0041) and large (AOR=0.939, p=0.0185) population sizes, although the positive association was not signfiicant in small population sizes (AOR=1.051, p=0.1431).

Discussion

The present study examined whether the strength of the association between more restrictive state alcohol policy environments and less binge drinking among college students depends on an array of individual-, institutional-, and contextual factors. Prior studies of APS and binge drinking have examined moderation on the basis of age, gender, and race/ethnicity among adults (Xuan, Blanchette, Nelson, Heeren, et al., 2015) and on the basis of grade level, gender, and race/ethnicity among youth (Xuan, Blanchette, Nelson, Nguyen, et al., 2015). However, these questions have not been examined in a sample of young adults ages 18-24 attending college, who represent an important target for prevention. The current study provides evidence that the protective effects of more restrictive state alcohol policy environments in relation to students’ binge drinking is moderated by several distinct factors.

The strongest evidence for moderation of APS effects came from contextual features: residence type, relationship status, and affiliation with Greek life. We found that the relationship between more restrictive policy environmentsand lower odds of binge drinking differs depending on housing type. Specifically, we did not find evidence that the state alcohol policies exert significant influences on binge drinking among college students who live in off-campus housing, Greek housing, or other housing facilities. Conversely, among students who live in their parent or guardian’s home, the strength of the association between more restrictive alcohol policies and reduced binge drinking was the strongest, followed by students in on-campus housing. It is possible that the state policy environment influences students’ binge drinking through the greater supervision and control over alcohol use that is possible in homes and residence halls or through effects on their parents who in turn affect the availability of alcohol or approval of its use in the home. Prior studies similarly indicate that residence type in college is a risk factor for increased substance use as well as other risky health behaviors (DiBello et al., 2018). Perhaps lower levels of campus engagement among students in off-campus housing facilities (DiBello et al., 2018) contributes to their decreased sensitivity to the state policy environment.

We also failed to detect a significant association between state alcohol policies and binge drinking among Greek-affiliated students, apart from where they lived. Thus, participation in Greek life, rather than Greek housing per se, may insulate students from the effects of state alcohol policy restrictiveness. In the Greek system, alcohol may be more readily available from older peers and purchasing may be more communal, thus dulling the effects of policies (e.g., pricing; retail hours of operation) designed to curb drinking by individual consumers. Our findings may relate to some of the reasons why students who are members in Greek organizations exhibit the highest levels of substance use including binge drinking (Scott-Sheldon et al., 2008). The present analyses cannot rule out selection effects; for example, students who are unresponsive to policy controls may be less apt to live with family and more inclined to affiliate with the Greek system. Still, taken together, these moderators present unique features in college life, and suggest targeted approaches to reduce binge drinking in these sub-groups are warranted. For instance, our analysis shows that fraternity/sorority involvement is associated with a more than two-fold increase in the odds of binge drinking (see Supplementary Table 1);if those students are unresponsive to existing state policy environments, then more specialized state, community, or campus-level prevention efforts may be necessary.

Relationship status provided a novel lens through which the association between the APS and binge drinking could be examined among college students. Higher APS was significantly associated with lower odds of binge drinking in all three relationship groups, but with a clear gradient of effect sizes from partnered students who live together (AOR=0.910), partnered students who do not live together (AOR=0.930), and students who are not in a relationship (AOR=0.957). Interestingly, partnering and cohabitation are strong protective factors for binge drinking in the NCHA data with a direct opposite trend (see Supplementary Table 1). Partnering and cohabitation are associated with less frequent heavy alcohol use (Fleming et al., 2010) and more rapid declines in alcohol use across early adulthood (Bachman et al., 2014; Kerr et al., 2011), perhaps in part due to decreases in contact with alcohol-using peers, a primary influence (Kim et al., 2013). As partnering and cohabitation also involve movement toward conventional norms and social controls, college students in this life stage may become more sensitive to state laws and regulations, and more similar to the general state population of adults (who are known to be responsive to APS). Nevertheless, the mechanisms that yield the varying effects of the state alcohol policy environment on binge drinking by relationship status are not immediately clear. Still, the results highlight the importance of these young adults’ interpersonal contexts in relation to the alcohol policy environment and binge drinking.

With respect to race/ethnicity, a stronger effect of the policy environment on binge drinking was observed for Asian students (AOR=0.899, p<0.0001), while a significant association persisted in non-Asian students (AOR=0.940, p=0.0012). Conversely, there was a lack of statistical evidence linking the APS to binge drinking among the Hispanic students. The results illustrate that Asian students are more sensitive to the policy environment, whereas the Hispanic students may be less sensitive to the same policy environment. One explanation could be that, given ethnic differences in drinking cultures (Cook et al., 2012) and drinking norms (B. E. McCabe et al., 2019), awareness of and adherence to state alcohol policies may also differ between these subgroups. Asian and Hispanic students represent significant proportions of campus communities (11.8% and 10.8%, respectively of the present sample) and were at differential risk for binge drinking (lower for Asian and higher for Hispanic students relative to white non-Hispanic peers). These groups’ different patterns of responsiveness to state policies may suggest that culturally tailored campus prevention approaches could further reduce binge drinking.

Contrary to expectation, the binge drinking of international students, who arguably have much less exposure to U.S. drinking norms much less state-level alcohol policies, was more associated (nominally significant) with APS than that of domestic peers. It is possible that relative to peers, international students are more sensitive to the perceived legal (including immigration-related) consequences of alcohol misuse or to policies impacting alcohol pricing, or that they are more often exposed to domestic peers who live on campus (who are responsive to APS) as models of state drinking patterns.

Finally, APS was more strongly associated with lower binge drinking among students attending institutions in larger population centers, whereas APS was unrelated to drinking for those in cities with fewer than 10,000 people. We speculate that students in larger cities may have more exposure to adult drinking environments (e.g., bars, clubs) that are in turn shaped by state norms and policies, whereas students in small towns may be more sensitive to the campus culture rather than the broader state culture.

Limitations

A primary weakness of the study is that it is not based on a random sample of institutions, and student-level response rates are low (mean = 24%). Therefore findings should be replicated in more representative samples. Second, findings may not generalize to community colleges students, those in vocational training programs, or young adults who are not in college; such young adults may be less mobile (across state lines) and one might expect state policy effects to be stronger for young adults who are not exposed to the heavy drinking culture associated with college campuses. Third, we did not measure community- or campus-level alcohol policies. Fourth, we could not estimate the effects of powerful policies (e.g., minimum legal drinking age) that were uniform across states during this time period. Finally, we lacked a measure of how long students had been exposed to the state policy environment. For example, an out-of-state transfer student in their first quarter could not be discerned from a student who spent their entire life in the state. Duration of exposure to the state policy environment may be confounded with moderators of interest; specifically, students at private institutions, those in their 1^st year, and international students have less state policy exposure on average than their public, later-year, and domestic counterparts.

Conclusion

In sum, the public health impacts of more restrictive state alcohol policies on less binge drinking among college students can be understood and enhanced if the protective effects are quantified for various subgroups. Given the varying effects at the individual-, institution-, and contextual-levels, granular analyses in the future should query specific policies and policy combinations, which will further inform recommendations to states, communities, and colleges. For example, Oregon supplemented alcohol policies aimed at the general population (ranked the 10^th most restrictive state according to 2018 APS score (Blanchette et al., 2020)) with additional policies restricting alcohol manufacturers from sponsoring or promoting events on college campuses (Oregon Liquor and Cannabis Commission: Activities on College Campuses, n.d.) or with off-campus student living groups. Furthermore, the College Alcohol Intervention Matrix (College AIM) summarizes the evidence on individual- and environment-focused policies and programming that campuses and communities can implement to discourage alcohol misuse among college students (Cronce et al., 2018). One study pointed out that the majority of college students are aware of the alcohol policies at their campus, but less than half accept such policies (Marshall et al., 2011); whether this also applies to state policies is unknown. Policymakers at the level of states, communities, and colleges should focus on the promulgation of effective policies, and also emphasize awareness and acceptance of policies among students, including through greater policy transparency, more effective sanctions, and consistent enforcement (Jernigan et al., 2023).