Beyond legalization: characterizing distinct recreational cannabis regulatory approaches across US states, 2013-2024

This study utilizes latent class analysis of US state data from 2013 to 2024 to identify four distinct recreational cannabis regulatory models, revealing that higher cannabis use prevalence and longer medical cannabis histories are associated with more access-oriented and more restrictive commercial approaches, respectively.

The Gist

Imagine the United States as a giant neighborhood where, over the last decade, every house (state) has been debating whether to open a "Cannabis Café."

Ten years ago, only two houses (Colorado and Washington) opened their doors. Now, many have. But here's the twist: just because they all opened a café doesn't mean they all serve coffee the same way. Some have drive-thrus, some have indoor seating, some only sell beans to take home, and some have strict rules about who can sit where.

This research paper is like a detective story that tries to sort these 50+ different café styles into four distinct "neighborhood vibes" to see what makes them tick.

The Four "Café Vibes" (Regulatory Models)

The researchers used a smart computer method (called Latent Class Analysis) to group the states based on their rules. They found four main types of "Cannabis Cafés":

1. The "Closed Shop" (No RCL):

   • The Vibe: These are the houses that haven't opened a café yet. The door is locked, and the sign says "No Cannabis Allowed."
   • The Reality: These are the states that haven't legalized recreational use at all.

2. The "Pop-Up Tent" (Pre-commercial):

   • The Vibe: Imagine a state that just got permission to sell cannabis but hasn't built a real store yet. It's like a farmer's market tent that's barely set up. They might let you grow a plant in your backyard, but there are no fancy shops, no delivery trucks, and no place to sit and smoke.
   • The Reality: These are states in the very early, messy stages of legalization, often before commercial shops are fully open.

3. The "All-You-Can-Eat Lounge" (Full Access):

   • The Vibe: This is the most relaxed café. You can buy your coffee, sit inside the shop to drink it, order it to be delivered to your house, and even hang out in a designated "smoking lounge" area. The rules are there, but they are friendly.
   • The Reality: These states (like Colorado or Nevada in recent years) allow you to buy cannabis and consume it right there in the store, or have it delivered. They restrict public smoking but don't ban it entirely.

4. The "Take-Out Only" Window (Dispensary Access):

   • The Vibe: This is a strict "Drive-Thru" or "Pick-up Window." You can buy the product, but you must leave the premises immediately to use it. No sitting in the parking lot, no delivery to your door, and absolutely no smoking in public. It's like a pharmacy where you grab your meds and walk away.
   • The Reality: These states (like Washington or Oregon in earlier years) are more cautious. They only allow off-site consumption and often have stricter rules on how much you can buy or how strong it can be.

Why Do Some Cafés Have Different Rules?

The researchers asked: "What makes a state choose the 'Lounge' style versus the 'Take-Out' style?" They looked at the "personality" of each state (its politics, how many people already use cannabis, and its history).

Here is what they found:

• The "High Demand" Theory: States where a lot of people already use cannabis (high "past-month use") tend to go for the "All-You-Can-Eat Lounge" style.
  • Analogy: If a neighborhood already loves coffee, the shop owners say, "Let's build a cozy place for everyone to hang out and enjoy it."
• The "Old School" Theory: States that have had medical cannabis laws for a very long time tend to stick with the "Take-Out Only" style.
  • Analogy: These states are like the cautious grandparents who say, "We've been doing this for 20 years, and we know how to keep it safe. Let's just sell it and keep the public use strict."
• The "Alcohol Connection": Interestingly, states that already have very high taxes on beer (strict alcohol rules) are less likely to open any cannabis café.
  • Analogy: If a town is very strict about alcohol, they are also likely to be strict about cannabis. They prefer to keep the "party" under tight control.

The Big Takeaway

For a long time, people thought legalization was just a "Yes or No" switch. This paper shows it's actually more like a menu.

States are choosing different combinations of rules. Some are building fun, social spaces (Full Access), while others are building strict, safety-first pharmacies (Dispensary Access).

Why does this matter?
If you are a public health official, you need to know which "menu" leads to fewer drunk drivers or fewer kids getting access to the product. You can't just say "Legalization is good" or "Legalization is bad." You have to ask, "Which kind of legalization?"

This study gives us the vocabulary to describe these different approaches so we can figure out which one works best for public health.

Technical Summary

1. Problem Statement

Since the legalization of recreational cannabis in Colorado and Washington (2012), 22 additional US states and the District of Columbia have followed suit. However, the regulatory frameworks governing these markets vary significantly. While existing research often relies on composite scoring systems to measure "policy restrictiveness," these additive scores fail to capture the specific configurations of policies that states adopt concurrently. Furthermore, there is a lack of understanding regarding how state-level characteristics (demographics, political climate, market conditions, and prior policy history) correlate with distinct regulatory models. The authors aim to move beyond a binary "legalized vs. non-legalized" distinction to identify and characterize the heterogeneity of Recreational Cannabis Laws (RCLs) and the factors driving their adoption.

2. Methodology

Data Sources and Scope:

• Timeframe: January 1, 2013, to January 1, 2024.
• Sample: 51 US jurisdictions (50 states + DC), resulting in 612 state-year observations.
• Policy Data: Primary data from the Alcohol Policy Information System (APIS), supplemented by the Network for Public Health Law, Marijuana Policy Project, and state websites.
• Covariates: State-level characteristics (2019–2023) including cannabis use prevalence, market conditions, medical cannabis history, political factors (e.g., Republican vote share), and socioeconomic demographics.

Policy Variables:
The study focused on 16 initial provisions across seven domains: governance, permitted products, access/retail controls, consumption restrictions, pricing/taxation, marketing, youth restrictions, and driving restrictions. Through an iterative selection process, 11 key provisions were retained for the final model to ensure meaningful class differentiation. These included:

• Role of local governments.
• Potency limits.
• Consumption restrictions (on-site vs. off-site).
• Access controls (home delivery, retail limits).
• Taxation and pricing controls.
• Marketing restrictions.
• Driving prohibitions.

Analytic Approach:

• Latent Class Analysis (LCA): Used to identify unobserved subgroups ("latent classes") of state-years based on the joint probability of policy provision endorsement. The study treated state-years as independent observations to maximize power, though a Multilevel LCA (MLCA) was used for sensitivity analysis.
• Model Selection: Fit indices (BIC, aBIC, CAIC, VLMR test, and cmPk) were used to determine the optimal number of classes.
• Regression Analysis: Multinomial logistic regression with clustered standard errors was employed (2020–2024 period) to assess associations between state characteristics and the identified latent classes.
• Sensitivity Analyses: Included MLCA, exclusion of DC, and temporal restrictions (2020–2024 only) to test robustness.

3. Key Contributions

• Typology Development: The study establishes a multidimensional typology of RCLs, moving beyond simple restrictiveness scores to identify distinct policy clusters.
• Differentiation of Commercial Models: It distinguishes between two primary commercial regulatory models that were previously conflated: one permitting on-site consumption and home delivery, and another restricting sales to off-premises only.
• Contextual Drivers: It identifies specific state-level correlates (e.g., cannabis use prevalence, duration of medical cannabis legalization, and alcohol tax rates) associated with specific regulatory pathways.
• Methodological Advancement: By using LCA, the study captures the complex interplay of policy provisions that additive scoring systems obscure, providing a more nuanced view of the regulatory landscape.

4. Key Results

Latent Class Identification:
The analysis identified four distinct regulatory approaches:

1. No RCL (79% of state-years): States without recreational legalization.
2. Pre-commercial (5%): Early-stage legalization with minimal commercial infrastructure. Characterized by limited local regulatory rights and prohibited public use, but lacking potency limits, advertising restrictions, or pricing controls.
3. Full Access (6%): A lenient commercial model.
   • Features: Allows on-site retail consumption and home delivery; restricts (but does not prohibit) public use; high probability of advertising and packaging restrictions.
   • Examples: Colorado (2020–2024), Nevada (2022–2024).
4. Dispensary Access (11%): A restrictive commercial model.
   • Features: Sales limited to off-premises consumption only; public use prohibited; high probability of potency limits and pricing controls; often prohibits home delivery.
   • Examples: Colorado (2014–2019), Oregon, Washington.

Associations with State Characteristics (2020–2024):

• Cannabis Use Prevalence: Higher past-month cannabis use was significantly associated with a greater likelihood of adopting the Full Access model relative to No RCL (Relative Risk Ratio [RRR] = 1.78; 95% CI: 1.21–2.62).
• Medical Cannabis History: A longer duration since the legalization of medical cannabis was associated with a higher likelihood of the Dispensary Access model (RRR = 1.47; 95% CI: 1.02–2.12).
• Alcohol Policy: Higher beer excise taxes were associated with a lower likelihood of adopting any RCL model compared to No RCL, suggesting that states with stricter alcohol controls tend to be more cautious with cannabis.
• Non-Significant Factors: The proportion of neighboring states with RCLs (policy diffusion), political affiliation (Republican vote share), and demographic factors (race, income, education) were not significantly associated with specific RCL classes in the multivariate model.

Temporal Trends:
Early-adopting states often transitioned from Pre-commercial to Dispensary or Full Access models over time. In contrast, late-adopting states tended to enter directly into Full Access or Dispensary Access models, bypassing the Pre-commercial stage.

5. Significance and Implications

• Public Health Policy: The study highlights that "legalization" is not a monolith. The Full Access model (permitting on-site consumption) may reduce secondhand smoke exposure in private homes but raises concerns regarding impaired driving and the normalization of use. Conversely, the Dispensary Access model imposes stricter controls, potentially mitigating some risks but limiting consumer access.
• Policy Evolution: The findings suggest two divergent pathways for RCL development:
  1. Market-Driven: States with higher existing cannabis use tend toward more access-oriented models (Full Access).
  2. Institution-Learning: States with a long history of medical cannabis regulation tend toward more restrictive, incremental models (Dispensary Access).
• Future Research: The authors note that as more states legalize, future studies should utilize Latent Transition Analysis (LTA) to track how states move between these classes over time. Additionally, incorporating county-level data is necessary to capture intra-state heterogeneity.
• Stakeholder Impact: The results provide a framework for evaluating how different stakeholder groups (industry lobbyists, public health officials, community advocates) influence the specific configuration of policy provisions.

In conclusion, this paper provides a critical, data-driven framework for understanding the complexity of US cannabis regulation, demonstrating that regulatory approaches are shaped by a combination of market demand and institutional history rather than a uniform national trend.