Associations Between State Cannabis and Alcohol Policy Restrictiveness and College Students’ Cannabis Use, Binge Drinking and Co-use

Abstract

Introduction:

U.S. college students’ cannabis use and binge drinking occur in the context of state-level substance policies. The Cannabis Policy Scale (CPS), a continuous measure of state policy restrictiveness, may offer greater nuance than policy categories such as recreational cannabis legalization for examining associations with substance use.

Methods:

College students (n=49,211) aged 18–30 years participating in the 2018–2019 Healthy Minds Study reported past-month cannabis use, past 2-week binge drinking, and co-use. Associations of state-level CPS and Alcohol Policy Scale (APS) scores with each outcome were tested in logistic regressions, adjusting for individual- and institution-level covariates, and accounting for nesting of students in institutions.

Results:

Higher CPS scores (more restrictive state cannabis policies) were associated with lower odds of student-level cannabis use (OR=0.90; 95% CI=0.87, 0.93; p<.001) and co-use (OR=0.92; 95% CI=0.89, 0.95; p<.001). APS scores were positively associated with odds of cannabis use within NCL states. Students in recreational states used more cannabis, yet unexpectedly, within recreational states, CPS was associated with higher odds of cannabis use.

Conclusions:

College students’ substance use was sensitive to state cannabis policies after accounting for state alcohol policies. In states with less restrictive cannabis policies, cannabis use and co-use were more common, suggesting complementarity. Within non-legal states, more restrictive alcohol policies were linked to higher cannabis use odds, suggesting substitution. State CPS and APS scores may inform policymakers’ efforts to reduce college students’ substance use. Future studies should replicate findings in contemporary cohorts, explore social justice implications, and consider concurrent campus and community policies.

Introduction

Cannabis and alcohol are the two most commonly-used recreational substances reported among young adults and college students in the United States (Patrick et al., 2025). Most college students are young adults, and face risks for substance misuse that are specific to this developmental period (Sowell et al., 2001) and the college context (Borsari & Carey, 2001; Fromme et al., 2010; Napper et al., 2015; Welsh et al., 2019).

While alcohol use trends for young adult college students are decreasing, the prevalence of drinking remains high and comparable with that of noncollege peers. The prevalence of 2-week binge drinking, a pattern defined by the consumption of 5 or more standard drinks in a row on a single occasion for men, and 4 or more for women, was 22.8% among college students surveyed in 2024 (Patrick et al., 2025). In contrast, cannabis use has become more prevalent among college students over the past 10 years (Patrick et al., 2025). In 2024, 21.8% of college students used cannabis in the past month, and an average of 5.25% used cannabis daily (Patrick et al., 2025). Frequent cannabis use poses risks to young adults’ brain development and psychological and physical health (Fischer et al., 2020; Hall et al., 2020; Levine et al., 2017; Volkow et al., 2014).

Most young adults who use cannabis also use alcohol. For example, in one large study, approximately 75% of U.S. college students who used cannabis also used alcohol in the same month or year (White et al., 2019). Cannabis and alcohol co-use compared to use of either substance alone is associated with numerous negative consequences in college students (e.g., physical injury; physical injuries to others; drove after drinking; Reed et al., 2024) and in the general population (e.g., relationship, legal, and financial issues, mental health outcomes, and intoxicated driving; Subbaraman & Kerr, 2015). Furthermore, in other samples, two-thirds of adolescents and adults who used both substances had consumed them simultaneously (Lee et al., 2022; Pacula, 2022), increasing the risks already incurred by co-use (Subbaraman & Kerr, 2015). Therefore, identifying modifiable risk factors for college students’ cannabis use and co-use is a priority.

Effective prevention of substance misuse among college students often occurs at the person-level (e.g. protective behavioral strategies) (NIAAA, n.d.; Welsh et al., 2019). However, prevention efforts, such as drug policies, that are enacted at broader levels have the potential to impact entire populations. In particular, much of the regulation of alcohol and cannabis occurs at the state level, through policies that can impact individuals directly, such as through restrictions on price or physical access, and indirectly, such as by changing perceived norms, (Pacula et al., 2022). Indeed, there is considerable evidence that more restrictive state alcohol policies are associated with lower rates of drinking and related harms among young adults and college students specifically. (Kasimanickam et al., 2024; Kerr et al., 2025; Naimi et al., 2014; Nelson et al., 2005).

In contrast to research on the strengthening of state alcohol regulation over time and its protective impacts, studies of state cannabis policies have focused on liberalization and the increases in cannabis use following legalization. For example, after the passage of a recreational cannabis legalization initiative in Washington, D.C., positive attitudes, perceived norms, and willingness to use cannabis increased in cannabis-naïve college students (Clarke et al., 2018). Moreover, studies support that following state enactment of recreational cannabis legalization, prevalence of cannabis use and frequent use increased among young adults overall (Pawar et al., 2024) and among college students specifically (Bae & Kerr, 2020).

Cannabis policy environments have evolved rapidly in the past 20 years compared to alcohol policies, which have been more stable (Blanchette et al., 2020; Naimi et al., 2023). The prevailing approaches to studying state cannabis policy effects have limitations, however (Blanchette et al., 2022; Cerdá, 2020; Neeley & Richardson, 2022). Namely, the use of medical and recreational cannabis legalization policy categories may not capture the considerable variation in policy restrictiveness between states in the same category or changes in policies within states over time as their policies unfold or evolve. To address this limitation, researchers developed the Cannabis Policy Scale (CPS), a continuous measure of state cannabis policy environments that accounts for the differential efficacy of 17 policies and differences in implementation by state (Blanchette et al., 2022).

Consistent with its purpose, two recent studies support that higher state CPS scores (indexing more restrictive policies) are associated with lower rates of cannabis use in young adults (state-level prevalence; Pessar et al., 2025), and college students (person-level odds; Kerr et al., in press). Given the magnitude of change in states’ cannabis policy liberalization, studies of how CPS is associated with cannabis use in other college samples would provide further evidence on the utility of this novel tool.

Several studies also have tested the extent to which state cannabis policies may have unanticipated effects on alcohol use (Alley et al., 2020; Kerr et al., under review; Pacula, 2022). The interest in these so-called spillover effects is based on the logic that if cannabis and alcohol are used for similar reasons (Cooper et al., 2016) then changes such as increased cannabis access and decreased penalties could lead to use of cannabis use instead of alcohol (Hursh, 2005). If this substitution hypothesis is supported, it could be an argument in favor of less restrictive cannabis policies, given that the public health burden of alcohol misuse is greater than of cannabis misuse (Hall, 2017; Pacula et al., 2022). However, some research suggests cannabis and alcohol are complements; if there is evidence linking less restrictive cannabis policies to higher rates of not just cannabis use, but also alcohol use or co-use then this would be cause for public health concern.

Prior literature in both the general adult population and college students is inconclusive regarding the relationship that cannabis policy liberalization has with young adults’ alcohol use and co-use (Alley et al., 2020; De & Sun, 2025; Gonçalves et al., 2023; Kerr et al., 2023; Lee et al., 2022; Macha et al., 2022; O’Hara et al., 2023; Ross et al., 2023; Subbaraman, 2016; Weinberger et al., 2021; Williams et al., 2004; Zellers et al., 2023). Importantly, however, few studies have considered both cannabis and alcohol outcomes in the same study, or accounted for the concurrent state alcohol policy environment (Pacula et al., 2022). In a recent exception, Kerr et al (in press) found that more restrictive state cannabis policies were associated with lower odds of college students’ cannabis use and serious co-use but not with odds of alcohol use outcomes. Of note, the study adjusted for concurrent state alcohol policies that—consistent with complementarity--were protective against a range of alcohol, cannabis, and co-use outcomes. Still, such studies are few and require replication in other samples.

Thus, the current study was aimed at replicating and extending prior studies of the extent to which state cannabis policy environments are associated with college students’ cannabis use, binge drinking, and co-use. Specifically, this study uses the Cannabis Policy Scale (CPS) to test the hypothesis that more restrictive state policies are associated with lower student-level odds of cannabis use. Building on similar research with 18–24-year-old 4-year undergraduate students (Kerr et al., in press) this study uses a different sample that also includes 2-year/community and 4-year college students and graduate students, and extends the age range to 18-30 years. Given recent recommendations (Pacula, 2022), models adjust for the concurrent state alcohol policy environment when examining CPS effects on binge drinking and co-use. By doing so, these models also test the extent to which alcohol policies are associated with these same outcomes. A final contribution of this study was an explicit test of the utility of the CPS beyond the traditional categorical approach (e.g., non-legal, medical cannabis legalization, or recreational cannabis legalization) in predicting odds of students’ cannabis use. Within each state legalization category, higher CPS scores were expected to be associated with lower odds of cannabis use.

Methods

Participants

Analyses were based on survey data from the Healthy Minds Study. During the 2018–2019 academic year 76 institutions (k=14 community colleges and k=59 undergraduate colleges and universities), in 28 U.S. states elected to administer the survey to a random sample of students. Participants (n=49,211) were undergraduate [Associate (n=3,691) and Bachelor (n=35,186) degree enrollment)] and graduate students (n=8,858) aged 18–30 years.

The response rate was 16%; Healthy Minds Study staff calculated response propensity weights based on the gender, race/ethnicity, academic level, and grade point average of students in the sample relative to those of the full population of students at the institution (Healthy Minds Study, 2019). A weight value was calculated for each student who did complete the survey: students deemed less likely to complete the survey received larger weights (Healthy Minds Study, 2019). Institutions elected to administer the survey in either Fall 2018 (45.9% of respondents) or Winter 2019 (54.1%). As of Fall 2018, cannabis use remained illegal in 10 states, 11 states had legalized medical cannabis, and 7 had recreational cannabis legalization (i.e., a legal cannabis supply via dispensaries or home cultivation was available (see Kerr et al., 2023a). As this study used de-identified archival data, the Oregon State University IRB determined it did not require oversight.

Measures

30-day Cannabis Use

Participants answered: “Over the past 30 days, have you used any of the following drugs?” including “marijuana.” Responses were coded yes (1) or no (0).

2-week Binge Drinking

Respondents were asked, “Over the past 2 weeks, did you drink any alcohol?” Those who answered yes were asked: “Over the past 2 weeks, about how many times did you have 4 (to those identifying as female), 5 (male), or 4 or 5 (neither female or male) or more alcoholic drinks in a row? (1 drink is a can of beer, a glass of wine, a wine cooler, a shot of liquor, or a mixed drink.)” Binge drinking was coded yes (1) if a participant answered 1 or more times, or no (0) if they answered zero times or previously denied any 2-week alcohol use.

Co-use

Co-use was coded 1 if participants reported both past 30-day cannabis use and 2-week binge drinking, or 0 if they reported only one or neither behavior.

Cannabis Policy Scale (CPS)

Each state’s cannabis policy restrictiveness was quantified using the 2018 CPS. The CPS is based on 17 policies nominated by 10 cannabis policy experts who then rated each policy in terms of its efficacy at reducing excessive cannabis use, and then rated its implementation in each state in each year (1999-2021) based on characteristics that made each policy applicable, effective, or enforceable. The CPS score is a sum of the policies weighted by these efficacy and implementation ratings in each state-year, and then placed on a theoretical scale from minimum (0) to maximum (100) restrictiveness; see Blanchette et al. 2022 and Naimi et al., 2023 for further details on policy items, weights, and scale development procedures. For reference, Blanchette reported that CPS scores in 2019 ranged from 29.6 (Nevada) to 96.2 (Indiana). As in prior work, a 10-point unit was used to scale the regressions to assist with interpretability of the odds ratios.

Alcohol Policy Scale (APS)

The 2018 Alcohol Policy Scale (APS) score for each state was used as a continuous measure of state alcohol policy restrictiveness (Blanchette et al., 2020; Naimi et al., 2014). In parallel to the CPS, APS scores were the state- and year-specific sum of 29 alcohol policies weighted by experts’ efficacy and implementation ratings. APS scores also are scaled from 0-100, and are normally distributed (Naimi et al., 2014). A 10-point unit was created to increase interpretability of odds ratios.

Covariates

All models adjusted for state APS scores. Additionally substance use may vary with age, race/ethnicity, gender, sexual orientation, relationship status, fraternity or sorority participation, religiosity, type of degree enrolled, and residence (NIAAA, 2023; Bae & Kerr, 2020; Cranford et al., 2009; Fromme et al., 2010; Kerr et al., 2017; McCabe et al., 2005; Napper et al., 2015; Patrick et al., 2025; Rosenthal et al., 2023; Schulenberg et al., 2021; Subbaraman et al., 2020); therefore, these covariates were included to reduce unexplained variance in the outcomes and improve statistical power. For categorical covariates, the largest category was typically used as the referent. Degree enrolled was coded as Associate, Bachelor’s (referent), Graduate (Master’s, Ph.D., J.D., M.D., etc.), and Other. Age groups were categorized into 18–20 years old, 21–25 years old (referent), and 26–30 years old. Race/ethnicity categories (mutually exclusive) were African American/Black, Asian American/Asian, Hispanic/Latino/a, Native Hawaiian or Pacific Islander, American Indian or Alaskan native, Middle Eastern/Arab/Arab American, other, and non-Hispanic white (referent). Gender was coded cisgender female (referent), cisgender male, transgender and genderqueer, or other. Sexual orientation categories were homosexual (lesbian and gay), heterosexual (referent), bisexual, or queer/questioning/other. Relationship status was coded as single (referent), in a relationship, married, or other. Fraternity or sorority participation versus non-participation (referent) was a binary variable. Religious importance (“How important is religion in your life?”) was indicated as “Important or very important” or “not important.” Residence categories included on-campus (referent), with parents or relatives, off-campus, or other.

Finally, models were adjusted for survey administration period [Fall 2018 (referent), Winter 2019] and two institution-level covariates: public versus private (referent) designation, and enrollment size [under 1000 students, 1000–4999, 5000–9999, 10000–19999, or 20000 or more (referent)].

Data analysis

Analyses were pre-registered with Open Science Forum (OSF; osf.io/3m6fs) and initially did not include sampling weights, given the focus on regression models rather than prevalence estimates. Following concerns about representativeness raised in peer review, models were updated using weights. Models were tested using multilevel logistic regressions run in Mplus 8.11 (Muthén & Muthén, 2017), with students clustered within institutions, adjusting for covariates. In the first set of models each outcome was regressed on CPS and APS scores. Next, a model was run regressing students’ cannabis use on the broad legalization categories; specifically, the odds of cannabis use associated with attending an institution in a recreational or non-legal state versus a medical legalization state (referent). Finally, associations between CPS and cannabis use odds were tested within each of the three state legalization categories. Per the pre-registration, a two-tailed alpha level of p < 0.05 was used for Model 1, p<0.025 for Model 2 (two contrasts: non-legal vs medical states, and recreational vs medical states), p<0.017 in Model 3 (tests of three within-category associations) and p < 0.05 for Models 4 and 5.

Results

Descriptive statistics

Frequencies for covariates are reported in Table 1. Table 2 shows descriptive statistics for state CPS and APS scores overall and within states in the non-legal, medical, and recreational categories. Consistent with the purpose of the CPS, there was considerable variation in CPS scores within the medical and recreational legalization categories. Notably, the overall distribution of CPS scores was skewed left with a small standard deviation, indicating generally high policy restrictiveness with limited variability between states. In contrast, APS scores were normally distributed overall, and indicated generally moderate policy restrictiveness with wide variability between states.

Table 1.

Individual-level Demographic Descriptive Statistics.

	%(N)
Degree Enrolled
Bachelor’s	71.5% (35186)
Associate	7.5% (3691)
Graduate	18.0% (8858)
Other	3.0% (1476)
Age
18–20 years old	49.4% (24310)
21–25 years old	39.4% (19389)
26–30 years old	11.2% (5512)
Racial/Ethnic Identity
Black/African American	7.0% (3445)
Latino/a/Hispanic	10.1% (4970)
Non-Hispanic white	63.3% (31151)
Asian American/Asian	14.0% (6890)
American Indian/Alaskan Native, Native	2.0% (984)
Hawaiian and Pacific Islander
Middle Eastern/Arab/Arab American	2.1% (1033)
Other (write-in)	1.5% (738)
Gender
Cisgender female	67.2% (33070)
Cisgender male	30.7% (15108)
Transgender/Genderqueer/Other (write-in)	2.1% (1033)
Sexual Orientation
Heterosexual	78.4% (38581)
Gay or Lesbian	3.9% (1919)
Bisexual	10.1% (4970)
Queer/Questioning/Other (write-in)	7.6% (3740)
Relationship Status
Single	54.0% (26574)
In a relationship	38.9% (19143)
Married/Domestic Partnership/Engaged	6.1% (3002)
Other (Divorced/Widowed/Separated/Other (write-in))	1.0% (492)
Fraternity/Sorority Participation	10.9% (5364)
Religious Importance
Important	37.2% (18307)
Not important	62.8% (30905)
Residence
On-campus	41.6% (20472)
Off-campus	42.7% (21013)
With relatives	15.7% (7726)
“Traditional” College Students	68.6% (33759)
Survey Administration Period
Fall 2018	45.9% (22588)
Winter 2019	54.1% (26623)
Institution Sector
Public	62.5% (30757)
Private	37.5% (18454)
Enrollment Size
<1,000	3.5% (1722)
1,000–4,999	18.7% (9202)
5,000–9,999	24.0% (11811)
10,000–19,999	22.5% (11072)
20,000+ (REF)	7.5% (3691)
No Cannabis Legalization (NCL)	41.4% (20373)
Medical Cannabis Legalization (MCL)	27.8% (13681)
Recreational Cannabis Legalization (RCL)	30.8% (15157)

Note. “Traditional” college student refers to students aged 18–24 years old who are enrolled in 4-year programs. Percentages are calculated at the individual level, not aggregated by state. Each percentage represents the proportion of individuals in the sample with the specified characteristic, rather than the proportion of states. Groups were separate for analysis, but any group under 1.0% was combined with others if possible to prevent student-level identification.

Table 2.

Cannabis Policy Scale (CPS) and Alcohol Policy Scale (APS) Descriptive Statistics.

	CPS mean (SD)	CPS range	APS mean (SD)	APS range
All states	76.88 (19.08)	29.55–96.20	43.01 (6.75)	33.75–61.00
No Cannabis Legalization (NCL)	93.09 (2.03)	89.55–96.20	41.69 (7.20)	33.83–61.00
Medical Cannabis Legalization (MCL)	77.18 (5.83)	71.18–93.30	44.35 (6.71)	33.92–57.22
Recreational Cannabis Legalization (RCL)	54.85 (17.51)	29.55–73.45	43.56 (5.76)	33.75–50.25

Note. SD = standard deviation.

Table 3 reports unadjusted outcome prevalence for the total sample and within the three legalization categories. 30-day cannabis use prevalence was lowest in non-legal states (21.6%), higher in medical states (25.7%) and highest in recreational states (28.4%). Co-use followed a similar pattern (14.7%, 16.6%, and 17.6% in non-legal, medical, and recreational states respectively. The opposite pattern was observed for 2-week binge drinking prevalence: 42.1%, 39.9%, 37.5% in non-legal, medical, and recreational states respectively.

Table 3.

Unadjusted Prevalence of Each Outcome and Within Legalization Subsamples.

	Total Sample % (N)	NCL subsample % (N)	MCL subsample % (N)	RCL subsample % (N)
	Cannabis use (past 30 days)	24.0% (48101)	21.6% (19878)	25.7% (28223)	28.4% (15160)
Binge drinking (past 2 weeks)	40.8% (48093)	42.1% (19720)	39.9% (28373)	37.5% (14883)
Co-use (cannabis use and binge drinking)	15.8% (46988)	14.7% (19232)	16.6% (27756)	17.6% (14625)
Abstinence (no cannabis or binge drinking)	51.6% (46988)	51.7% (19232)	51.4% (27756)	52.1% (14625)
Binge drinking only	24.9% (46988)	27.4% (19232)	23.2% (27756)	20.0% (14625)
Cannabis only	7.7% (46988)	6.2% (19232)	8.8% (27756)	10.3% (14625)

Note. NCL = no cannabis legalization, MCL = medical cannabis legalization, RCL = recreational cannabis legalization. Percentages are calculated at the individual level, not aggregated by state. Percentages are based on valid responses (N excludes missing values). Total values including missing values were 49206 for total sample, 20366 for NCL subsample, 28373 for MCL subsample, and 15160 for RCL subsample.

Relations between state policies and participants’ cannabis use, binge drinking, and co-use

As shown in Table 4, higher CPS scores (more restrictive state cannabis policies) were associated with significantly lower odds of 30-day cannabis use (OR=0.90; 95% CI=0.87, 0.93; p<.001) and lower odds of co-use (OR=0.92, CI=0.89, 0.95; p<.001). State APS scores were not significantly associated with any of the three outcomes.

Table 4.

Results of Logistic Regressions Testing Associations between Substance Policy Measures and Substance Use Outcomes.

	Analytic n (total = 49211)	B (SE)	OR (95% CI)	p-value
Model 1: Cannabis use on state policies	48094
CPS		−0.109 (0.020)	0.897 (0.868, 0.926)	<.001***
APS		0.106 (0.062)	1.111 (1.004, 1.230)	.087
Model 2: Cannabis use on cannabis legalization categories	49199
NCL (vs MCL)		−0.161 (0.098)	0.852 (0.725, 1.000)	.101
RCL (vs MCL)		0.307 (0.106)	1.359 (1.141, 1.618)	.004**
APS		0.076 (0.059)	1.079 (0.980, 1.189)	.193
Model 3: Cannabis use on CPS utility within each legalization category
NCL	20416
CPS		0.162 (0.376)	1.176 (0.633, 2.182)	.667
APS		0.295 (0.102)	1.343 (1.136, 1.588)	.004**
MCL	14028
CPS		−0.070 (0.160)	0.933 (0.717, 1.213)	.662
APS		−0.068 (0.102)	0.934 (0.790, 1.105)	.505
RCL	13650
CPS		0.159 (0.049)	1.172 (1.081, 1.271)	.001**
APS		0.103 (0.168)	1.109 (0.841, 1.462)	.540
Model 4: Binge drinking on state policies	48086
CPS		0.002 (0.017)	1.002 (0.974, 1.030)	.929
APS		−0.007 (0.053)	0.993 (0.910, 1.083)	.890
Model 5: Co-use on state policies	46891
CPS		−0.087 (0.020)	0.916 (0.887, 0.947)	<.001***
APS		0.104 (0.070)	1.109 (0.988, 1.245)	.139

Note. CPS = Cannabis Policy Scale, APS = Alcohol Policy Scale, NCL = no cannabis legalization, MCL = medical cannabis legalization, RCL = recreational cannabis legalization. CPS and APS scores are divided by 10 to increase interpretability. All models adjusted for age, race/ethnicity, gender, sexual orientation, relationship status, fraternity or sorority participation, religiosity, type of degree enrolled, residence, survey administration period, public versus private institution designation, and institution enrollment size. Significance threshold was pre-registered to p<.017 for Model 3 to correct for multiple comparisons.

Table 4 also includes results of a regression testing associations between categorical state cannabis policy indicators and students’ 30-day cannabis use (see Model 2). This model adjusted for APS scores and other covariates, but not CPS scores. Students in recreational states had 1.36 times greater odds of 30-day cannabis use compared to students in medical states (OR=1.36; 95% CI=1.14, 1.62; p=.004; see Table 4 Model 2). Attending college in a state with no cannabis legalization (vs. medical legalization) was not significantly associated with odds of 30-day cannabis use (OR=0.85; 95% CI=0.73, 1.00; p=.101).

Associations between state-level policy scores and cannabis use within state legalization categories

Finally, Table 4 reports the effects of CPS and APS scores on students’ 30-day cannabis use within each legalization category. Within non-legal states, higher APS scores were associated with higher odds of cannabis use (OR=1.34; 95% CI=1.14, 1.59; p=.004). Within recreational states, higher CPS scores were unexpectedly associated with higher odds of using cannabis (OR=1.17; 95% CI=1.08, 1.27; p=.001).

Discussion

Using a relatively novel continuous measure of the state cannabis policy environment, the Cannabis Policy Scale (CPS), this study indicated that college students at institutions in states with less restrictive cannabis policies were more likely to use cannabis in the last month compared to peers in less restrictive states. Models using the coarser categorical measure of state legalization status showed the same pattern, consistent with prior studies of college students and young adults overall (Bae & Kerr, 2020; Pawar et al., 2024). The present associations between the CPS and cannabis use are consistent with those based on the general adult and youth population (Pessar et al., 2025), and are similar to those reported in a different college sample (Kerr et al., under review); however, the present sample included a wider demographic of older students, community college attendees, 4-year undergraduates, and graduate students. Taken together, these studies suggest that more relaxed state cannabis policies may be considered a population-level risk factor for college students’ cannabis use, and that the CPS may be a useful tool for indexing the state policy environment in which colleges are embedded. Of note, current study models accounted for the concurrent state alcohol policy environment, as recommended by Pacula (2022).

Next, college students living in less restrictive cannabis policy environments were more likely to report co-use (both binge drinking and cannabis use), which poses multiple risks above using each substance alone (Reed et al., 2024; Thompson et al., 2021). This study adds to the prior literature which has been inconclusive regarding whether cannabis policy liberalization was related to increases in co-use (Pacula, 2022; Schlienz & Lee, 2018). Further, a prior study found that CPS scores were inversely related to high-risk co-use (binge drinking and cannabis use) in college students, (Kerr et al., in press). Although our co-use measure did not specify high-risk patterns, results are still consistent with these findings. Future studies should parse concurrent use (e.g., use of both alcohol and cannabis in the same month) from simultaneous use (i.e., at the same time so that effects overlap), an important distinction for public health implications (Subbaraman & Kerr, 2015).

A secondary focus of the present study was the extent to which state policy restrictiveness would be related to students’ cannabis use within states grouped into the traditional legalization categories (e.g., within recreational states). CPS scores were not significantly associated with students’ cannabis use within non-legal or medical states. However, within states where cannabis use is illegal, more restrictive state alcohol policy environments were associated with higher odds of 30-day cannabis use, which aligns with substitution theory (Hursh, 2005; Subbaraman, 2016).

A more unexpected finding was that within recreational states, more restrictive cannabis policy environments were associated with higher odds of cannabis use. The pattern was contrary to expectation and yet still consistent with the rationale that heterogeneity within broad policy categories is important. It is possible, for example, that for college students in more permissive states added restrictions are viewed as contradictory (e.g., few legal purchase options even though adult use is allowed), or might provoke resistance if lagging policies do not yet align with new social norms. There are, however, several important limitations to these analyses. Foremost, given that there were only 7 to 11 states in each legalization category and students were assessed in a single policy year, there may have been insufficient variability to robustly test CPS effects in within-category analyses. Further research is needed that compares the CPS with the categorical approach. Categorical thresholds may overshadow within-category nuances, particularly if data from only a subset of states are available. Additionally, cannabis use may not decrease linearly with increasing policy restrictiveness. Identifying specific policies that most strongly mitigate (or facilitate) cannabis use (e.g., physical availability restrictions, advertising differences) will also be important to understand what drives state differences in students’ use. These issues should be studied further in college students and other populations.

Strengths and Limitations

The current study has several strengths. First, it is one of the first studies to evaluate the effectiveness of the CPS, a novel and comprehensive measure (Blanchette, Pacula, Smart, Lira, Pessar, et al., 2022), and only the second to do so in a sample of college students (Kerr et al., under review). Second, by using the CPS to predict odds of students’ cannabis and alcohol use while adjusting for state alcohol policies, this study was also one of few that tested substitution and complementarity hypotheses in college students (Williams et al., 2004; Kerr et al., under review). Third, the Healthy Minds Study included community college students, graduate students, and older “non-traditional” college students, which means that these growing populations in higher education (Hittepole, 2019) were better represented here than in many prior studies.

Current findings should be understood in light of several limitations. Foremost, this study considered state cannabis and alcohol policies in only 28 states and during a single policy year. Therefore, the range of policy environments to which college students were exposed was constrained, and the cross-sectional design precludes inferences about directionality and causality.

A second set of limitations concerns measurement. The past-month cannabis use variable was binary, and did not capture frequency, mode (e.g., smoking, vaping, edibles), quantity, or potency. However, many studies that have examined effects of policy environment on “any use” have also found effects on frequent use (Bae & Kerr, 2020; Buckner et al., 2018; Everson et al., 2019), so the Healthy Minds Study monthly use variable may be a signal for more frequent use in many students. Additionally, for the measure of co-use, the time frames for 30-day cannabis use and 2-week binge drinking did not align and did not measure simultaneous use, a pattern associated with more negative consequences than co-use (Lee et al., 2022; Subbaraman & Kerr, 2015).

Third, there were limitations imposed by the Healthy Minds Study design. Social desirability bias may influence self-report of stigmatized or illegal behavior, and may have differentially affected responses from underage participants or those living in states where cannabis use was illegal. Additionally, the low response rate (16%) for the 2018–2019 Healthy Minds Study suggests potential for sampling bias, despite safeguards including random sampling of students from schools, and use of sampling weights. Indeed, one study found that cannabis use prevalence was higher for college men participating in HMS from 2018–2019 compared to those in the National College Health Assessment, National Survey on Drug Use and Health, and Monitoring the Future studies. On balance, prevalence of cannabis use among college women, and binge drinking among men and women did not differ across data sources (Kerr & Bae, 2025). Finally, given that institutions self-select to participate in the Healthy Minds Study, the sample is not designed to be representative of all U.S. institutions or college students.

Future Directions

The current study may inspire many future research directions. Given evidence linking high-frequency, high-potency, and long-term cannabis use to adverse health outcomes (Asbridge et al., 2014; Petrilli et al., 2022; Steeger et al., 2021), future research should utilize stronger measures, including more nuanced measures of use and co-use or indicators of cannabis use disorder symptoms. Additionally, longitudinal and causal inference designs may examine person-level data to explore the possibility that differences of cannabis use between policy environments may be a result of different prevalences of use in these areas prior to legalization.

A natural next step from the current study is to investigate potential mediators of the relationship between cannabis policy and college students’ substance use, such as proximity to cannabis retailers (Bostean et al., 2023; Hust et al., 2020; David C. R. Kerr et al., 2023), advertising presence (Firth et al., 2022; Hust et al., 2020), cognitions about cannabis use such as risk perceptions (Clarke et al., 2018; Malain et al., 2023), and perceived peer and parental norms (Koval et al., 2019; Napper et al., 2015). Along this line, the role of policy in substance use experiences unique to the college context such as social events (21st birthday parties and sporting events; Fromme et al., 2010) and Greek life involvement (McCabe et al., 2005) could be examined. With this knowledge, mechanisms through which policies affect substance use can be identified, and modifiable factors can be targeted to inform prevention efforts.

Furthermore, campus and community policies may exert additional influence on students’ substance use. State policies do not fully capture the policy environment that students experience, as institutional and local policies may further shape cannabis use—for example, whether the effects on cannabis use that have been observed from campus-level interventions (Buckner et al., 2018) and local restrictions on availability (Matthay et al., 2022; Young-Wolff et al., 2024) build on the effects of strong state-level policies or offset weak ones. One recent study tested this combined impact of state- and campus-level policies on students’ binge drinking (Trangenstein et al., 2025), but similar studies of students’ cannabis use and co-use outcomes are needed. Knowing that students in less restrictive cannabis policy environments are more likely to use cannabis, it is possible for institution and community stakeholders to collaborate on efforts to reduce substance use. For instance, campuses might assess substance use prevalence in their own student body, decide whether reducing it is a priority, examine their own state’s relative rank on APS and CPS, and determine how local or campus policy or prevention could enhance or compensate for their state environment.

A final set of future directions concern the social justice implications of state substance use policies, which the CPS and APS were not designed to assess. While these measures are scaled in relation to substance use outcomes, they do not illuminate how policy environments create, exacerbate, or alleviate other disparities. For example, Matthay et al. (2022) found that some local policies that reduce cannabis outlet density may increase the placement of outlets in marginalized neighborhoods (2022). There also are significant racial disparities in the enforcement of cannabis policies (reported incidents and arrests) (Butcher et al., 2022) that may improve with liberalization. Future research should consider how policies and their enforcement interact with these structural inequities, and should ensure that enacted policies do more good than harm.

In conclusion, college students in states with less restrictive cannabis policies were more apt to use cannabis and to report co-use of cannabis use and binge drinking than peers in other states. In states with no cannabis legalization, students exposed to more restrictive alcohol policy environments were more likely to use cannabis, which may support substitution. Overall, the current study and other recent studies support the use of continuous measures of state cannabis and alcohol policies to characterize the broad policy environments within which other efforts to reduce substance use and promote well-being in college students can occur.

Data availability statement:

Study and methods were first pre-registered in the Open Science Framework (OSF) on October 3, 2024 (Hummel & Kerr, 2024). Data are available from the Healthy Minds Study team. Authors interested in the APS and CPS can contact the scale developers. Teams wishing to use these data, or to replicate or extend our analyses, can request data from the stewards of those data sources.