Does Legalizing Recreational Cannabis Increase the Use of Cannabis, Cigarettes, and E-Cigarettes?

This study examined whether legalizing recreational cannabis in U.S. states affected the use of cannabis, cigarettes, and e-cigarettes over five years. Researchers analyzed data from more than 55,000 adults across the United States using the Population Assessment of Tobacco and Health (PATH) study from 2013 to 2022. They compared states that legalized cannabis with those that did not. Cannabis use increased by 3.28 percentage points and e-cigarette (ENDS) use rose by 1.39 points after legalization, while cigarette smoking showed no significant change (–0.99 points). Increases in cannabis use were greatest after retail stores opened, especially in legalized states.

The Policy Scientist’s Perspective

This study addresses an increasingly important question as cannabis legalization continues to spread across the United States and globally: how legalization affects the substance use behaviors. Its relevance lies in evaluating the long-term public health implications of a rapidly changing policy environment rather than short-term behavioral shifts. The analysis benefits from the PATH dataset, a large, nationally representative longitudinal panel that enhances both reliability and generalizability to other high-income jurisdictions. The use of difference-in-differences methods represents a strong quasi-experimental approach for estimating causal effects. The paper builds on earlier studies that used shorter time horizons, offering a more comprehensive five-year view that captures both legalization and commercialization phases. It is among the most significant recent contributions to the cannabis policy literature because it isolates longer-term population-level effects using rigorous, transparent methods at a national scale.

Full Citation and Link to Article

Hyatt, A. S., Overhage, L., & Lê Cook, B. (2025). Use of tobacco and cannabis following state-level cannabis legalization. JAMA Network Open, 8(7), e2520093. https://doi.org/10.1001/jamanetworkopen.2025.20093

Extended Summary

Central Research Question

This study investigates whether the legalization of recreational cannabis at the state level in the United States influences the prevalence of cannabis, cigarette, and electronic nicotine delivery system (ENDS) use among adults over a five-year period. Specifically, it asks whether legalization and subsequent commercialization of cannabis—through the opening of retail outlets—are associated with measurable, sustained changes in these behaviors. The question is significant because it addresses both direct and indirect public health consequences of legalization. While early studies found modest, short-term increases in cannabis use after legalization, few examined the longer-term effects or the interaction between legalization and commercialization. The authors sought to fill this gap using a quasi-experimental, longitudinal design capable of capturing evolving behavioral patterns as cannabis markets mature.

Previous Literature

Prior research on the behavioral effects of cannabis legalization has produced mixed and often time-limited findings. Studies through roughly 2020, many using two- to three-year post-legalization windows, typically reported modest increases in adult cannabis use following legalization, particularly in states with established retail markets. For example, work by Cerdá and colleagues (2017, 2020) found increased cannabis use and cannabis use disorder prevalence following legalization, while Hasin et al. (2023) and Martins et al. (2021) documented demographic variations, especially across racial and ethnic groups. However, these studies generally lacked long follow-up periods and could not isolate the effect of commercialization—the phase when cannabis retail outlets open and market exposure intensifies.

Evidence regarding tobacco and ENDS use in the context of cannabis legalization has been inconsistent. Some studies suggested complementarity between cannabis and tobacco, with cannabis users showing higher rates of cigarette use and lower quit rates. Others proposed substitution effects, where cannabis use might displace cigarette consumption. A few studies, including Weinberger et al. (2020, 2022) and Dave et al. (2023), used national surveys or tax data to explore these relationships, finding limited or conflicting evidence about whether legalization influenced tobacco use trends. The emergence of ENDS products further complicated these dynamics. Vape devices can deliver both nicotine and tetrahydrocannabinol (THC), blurring distinctions between substances and raising questions about cross-use patterns.

In short, the literature prior to this study lacked long-term evaluations, had limited generalizability, and did not systematically distinguish legalization (the policy change) from commercialization (the market phase). Hyatt, Overhage, and Lê Cook aimed to address these shortcomings by examining longitudinal, individual-level data over a full five-year horizon using modern difference-in-differences (DiD) methods that account for staggered policy adoption across states.

Data

The analysis draws on the Population Assessment of Tobacco and Health (PATH) Study, a large, nationally representative longitudinal survey conducted jointly by the National Institute on Drug Abuse and the U.S. Food and Drug Administration’s Center for Tobacco Products. The dataset includes repeated observations of adults aged 18 and older across seven survey waves from 2013 through 2022, encompassing approximately 171,257 observations from 55,406 unique individuals after exclusions. The PATH data provide detailed measures of individual substance use, demographics, and socioeconomic characteristics, enabling robust modeling of behavioral change over time.

The key exposure variables were based on the timing of state-level recreational cannabis legalization (RCL) and the subsequent opening of retail cannabis outlets. These dates were extracted primarily from the National Institute on Alcohol Abuse and Alcoholism’s Alcohol Policy Information System and supplemented by the Marijuana Policy Project. Respondents were coded as residing in a legalized state beginning in the year the law took effect and continuing thereafter, with observations during the transition year excluded to avoid ambiguity about exposure.

Outcome variables were binary indicators capturing past 30-day use of cannabis, cigarettes, and ENDS. The analysis also incorporated relevant time-varying state-level covariates—such as cigarette and ENDS tax rates, indoor smoking restrictions, and inclusion of vaping in clean air laws—to reduce policy confounding. Individual-level covariates included age, gender, race and ethnicity, and whether respondents received public financial assistance. The sample’s demographic profile was diverse and closely matched national distributions, enhancing external validity and generalizability across U.S. states and, potentially, to other high-income nations with similar regulatory contexts.

Methods

The researchers employed modern difference-in-differences (DiD) estimation techniques following the Callaway and Sant’Anna (2021) approach, which accommodates variation in treatment timing across states and corrects for biases common in traditional two-way fixed-effects DiD models. This quasi-experimental design approximates a causal inference framework by comparing pre- and post-policy changes in states that legalized cannabis to changes in states that did not.

The analysis proceeded in several stages. First, the authors verified the key DiD assumptions: (1) the parallel trends assumption—that, absent legalization, treatment and control states would have exhibited similar outcome trajectories—and (2) the no-anticipation assumption—that states did not experience behavioral changes prior to legalization. Parallel trends were assessed using pre-treatment event study estimates and Wald tests; states or cohorts violating these assumptions were excluded.

After validating these assumptions, the authors estimated percentage point changes in 30-day use for each substance. They also conducted a two-stage DiD to separately estimate the effects of legalization itself and subsequent commercialization (after retail outlets opened). Robust standard errors were clustered at the individual level to adjust for repeated measures. Sensitivity analyses tested the robustness of results to alternative specifications, including restricting the sample to states with five full years of follow-up, excluding decriminalized states from the control group, limiting analysis to adults aged 21 and older, and re-estimating results using traditional two-way fixed-effects models for comparison.

The methodological approach is notably strong within the context of observational policy evaluation. While it does not reach the rigor of a randomized controlled trial, the longitudinal DiD framework, coupled with extensive robustness checks, provides credible quasi-causal estimates of policy impact.

Findings/Size Effects

Over the five years following legalization, the study found statistically significant increases in cannabis and ENDS use, but no significant change in cigarette smoking. Specifically, recreational cannabis legalization was associated with a 3.28 percentage point increase (95% CI, 2.29–4.27) in past 30-day cannabis use relative to control states. The effect grew larger after retail outlets opened—rising to 3.74 percentage points post-commercialization compared with 1.17 percentage points during the legalization-only phase.

ENDS use also rose modestly by 1.39 percentage points (95% CI, 0.44–2.35), with similar estimates before and after commercialization (1.41 vs. 1.21 percentage points). This pattern suggests that cannabis legalization may have indirectly encouraged greater vaping overall, possibly reflecting the shared delivery technology between nicotine and cannabis products.

In contrast, cigarette use declined slightly (–0.99 percentage points) but not significantly (95% CI, –2.25 to 0.27), suggesting that legalization neither slowed nor reversed the long-term downward trend in cigarette smoking. The results therefore contradict concerns that cannabis legalization might lead to renewed tobacco use through behavioral complementarity.

Event study plots revealed a steady rise in cannabis use through year four post-legalization, leveling off or slightly declining thereafter—a pattern possibly reflecting market stabilization or data interruptions due to the COVID-19 pandemic, which paused data collection in 2020. Sensitivity tests confirmed the robustness of cannabis findings, while ENDS results were somewhat more sensitive to the inclusion or exclusion of certain large states (notably California and Michigan).

The magnitude of these effects, while modest in absolute terms, represents a meaningful population-level shift given the scale of legalization across 24 U.S. states by 2024. The findings suggest that legalization primarily influences the prevalence of occasional or moderate cannabis use, rather than escalating heavy or dependent use, as there were no significant changes in weekly or greater cannabis consumption.

Conclusion

This study provides one of the most comprehensive long-term evaluations to date of how recreational cannabis legalization and commercialization affect substance use behaviors in the United States. Using a large, nationally representative dataset and advanced quasi-experimental methods, the authors demonstrate that legalization is associated with moderate, sustained increases in cannabis and vaping use but has no measurable effect on cigarette smoking.

The findings have several implications. First, they confirm that legalization effects are not transient; rather, they persist and even grow as legal retail markets expand. Second, the evidence refutes the notion that cannabis policy liberalization undermines progress in tobacco control. Finally, the observed rise in ENDS use underscores a potentially new area of public health concern, as vaping technologies bridge nicotine and cannabis consumption.