Cannabis Ads & Use: What the 2026 Meta-Analysis Found

In February 2026, the journal Addiction published the first proper systematic review and meta-analysis of cannabis advertising exposure and cannabis use [McClure-Thomas et al., 2026]. It landed quietly. Most consumer-facing cannabis outlets did not cover it at all. The few that did reached for scare headlines or shrugged the findings off as “cross-sectional, so ignore it.”

Both of those reactions fail readers. The paper deserves a careful read. So does every paper of this shape: pooled effect sizes across 21 studies, a clear heterogeneity signal, honest limitations in the discussion section. We are going to walk through what it found, how it found it, and what it does and does not justify in the ongoing policy debate about how cannabis can legally be advertised.

What the study actually found

The headline number: across 21 studies that measured self-reported cannabis-advertising exposure and self-reported cannabis use, people who reported seeing cannabis ads had roughly 1.77 times the adjusted odds of reporting cannabis use compared with people who did not. The 95% confidence interval ran from 1.32 to 2.30. In plain language: the effect is not a rounding error, and the interval does not cross 1.0, which is the threshold where “no association” lives.

The authors — McClure-Thomas and colleagues at the University of Queensland’s National Centre for Youth Substance Use Research — then split the pool by exposure channel, and this is where the story gets more interesting than most coverage has acknowledged.

• General advertising (billboards, print, broadcast): aOR = 1.67 (95% CI [1.27, 2.21]) across 3 studies. Statistically significant.
• Internet and social-media advertising: aOR = 3.38 (95% CI [1.07, 10.66]) across 5 studies. Significant, but that wide interval is doing a lot of work.
• Storefront and sidewalk signage: aOR = 1.25 (95% CI [0.95, 1.66]) across 3 studies. Not statistically significant.

The storefront finding is the one almost nobody has written about. It complicates the tidy narrative that “all cannabis ads increase use.”

How the study was actually conducted

This was a systematic review following standard PRISMA methodology. The team searched PubMed, Scopus, and PsycINFO in January 2024, pulled 2,588 records through database searches, screened titles and abstracts, and full-text-reviewed 45 candidates. Twenty-one studies met inclusion criteria. Ten of those were eligible for the quantitative meta-analysis. The rest contributed to the narrative synthesis.

Eighteen of the 21 included studies were cross-sectional. Only three were longitudinal. That matters, and the authors flag it plainly. They used random-effects models — the appropriate choice when you expect genuine variation in true effect sizes across studies, not just sampling error. They reported heterogeneity as Q[11] = 22.73, p < 0.05, I² = 42.3%, which is moderate. Not so high that pooling is meaningless, but high enough that the subgroup analyses by channel are doing real explanatory work.

Methodological quality was rated good in 48% of included studies, satisfactory in 43%, and unsatisfactory in 10%. That is a typical distribution for an emerging literature. It is not the methodological quality of, say, the decades-old alcohol-advertising evidence base, but it is not garbage either.

The young-adult signal

The review also synthesized evidence on cannabis-use intentions — a separate outcome from current use. Here the pattern was consistent: in adolescents and young adults (roughly ages 13 through 25), exposure to cannabis advertising was associated with higher odds of intending to use cannabis in the future. One included study [Chaffee et al., 2024] found a 2.10 adjusted odds ratio for past-30-day ad exposure and intention to use within the next year, among never-users in California.

Why does 18–25 keep showing up as the most susceptible bracket? Developmental neuroscience has a reasonable answer. The prefrontal cortex — the brain’s executive-control region — is still wiring up through the mid-twenties. Risk-reward calibration, delay discounting, and social-conformity sensitivity are all more plastic in that window. Parallel findings exist in alcohol-advertising meta-analyses [Anderson et al., 2009] [Jernigan et al., 2017] and, more recently, in a 79-study e-cigarette systematic review [Paruk et al., 2025] published in BMC Health Services Research in December 2025. The cannabis signal is not anomalous. It looks a lot like the tobacco and alcohol signals, a generation earlier.

That parallel is worth sitting with rather than dismissing. The cannabis industry is not obligated to repeat tobacco’s mistakes. But the pattern of “digital ads in a loosely regulated environment reach young adults at elevated rates” is not new.

Effect sizes, without the scare

An adjusted odds ratio of 1.77 sounds dramatic. It is not trivial — but it also is not “seeing one ad doubles your odds of becoming a heavy user.” A few translations worth holding onto:

• Odds ratios are not risk ratios. When baseline rates are high, odds ratios overstate the absolute difference. Cannabis use in young adults is not rare, so the real-world shift in use prevalence implied by aOR = 1.77 is smaller than the number suggests.
• The 3.38 digital-ad figure has a confidence interval from 1.07 to 10.66. That is extremely wide. It tells you the true effect is probably positive, but almost nothing else about the magnitude. Five studies is a small base.
• Heterogeneity of 42.3% means the studies do not all point to the same number. The authors acknowledge this and respond appropriately with subgroup analyses.

The honest read: there is a real association, digital channels look stronger than physical ones, and the precision on the digital estimate is not yet good enough to make confident policy claims about exact effect size. It is enough to justify caution. It is not enough to justify panic.

What the study does NOT say

Every honest read of a meta-analysis should include a “does not say” section. This one is important.

It does not establish causation. Eighteen of 21 studies were cross-sectional. That means researchers measured ad exposure and cannabis use at the same time, in the same people. People who use cannabis may notice cannabis ads more, recall them more, and report more exposure — an effect the authors explicitly flag and that other researchers have called bidirectional [Hammond et al., 2026]. Longitudinal and experimental designs are the only way to separate the directions, and only three of the included studies were longitudinal.

It does not say cannabis advertising is uniquely harmful. The effect sizes here are broadly in the same neighborhood as alcohol-advertising and e-cigarette-advertising meta-analyses. If you believe this study implies sweeping cannabis-ad restrictions, the consistency of the evidence implies you should also favor similar restrictions on the other two — which is a coherent position, but not the one most commentators hold.

Self-reported exposure is noisy. People are poor historians of advertising. They over-report and under-report in different directions depending on whether they approve of the product. The authors mention this.

Publication bias is plausible. Studies that find significant associations with a hot-button product are more likely to reach publication. The authors did not find strong evidence of it in their funnel-plot analysis, but with only ten studies in the main pool, those tests have limited power.

Sidebar: The dispensary-proximity study

Published in the same window (Addiction, February 18, 2026), an ecologically different study looked at physical rather than media exposure [Harlow et al., 2026]. Alyssa Harlow and colleagues followed 2,277 California young adults from 2021 to 2023 and measured the number of cannabis dispensaries within a one-mile radius of each participant’s home. They built the count two different ways — from California’s government-maintained licensed-dispensary registry, and from a web-scraped list that also included unlicensed operators.

Every additional dispensary within one mile was associated with a 5–6% increased risk of past-six-month cannabis use using the registry data, and 3–4% using the web-scraped data. The smoked-cannabis IRR was 1.08 (95% CI [1.01, 1.15]); edibles ran 1.07 (95% CI [0.99, 1.15]); vaping and daily-or-near-daily use showed no consistent association.

This pairs interestingly with the meta-analysis. Storefront and sidewalk advertising did not show a significant association with use in the McClure-Thomas pool. But dispensary presence in Harlow’s cohort did. The difference may come down to what “exposure” actually means in each study: a glance at a storefront sign versus the repeated daily availability of a retail outlet within walking distance. Those are different kinds of exposure, and the data suggests they behave differently.

DOI for the dispensary paper: 10.1111/add.70356.

Why this shows up in the 2026 policy debate

As of spring 2026, U.S. advertising rules for cannabis are a patchwork. New York bans most outdoor cannabis advertising. California permits billboards under restrictions on placement and imagery. Michigan is actively debating tighter rules. A January 2026 Addiction analysis by Hammond and colleagues [Hammond et al., 2026] using International Cannabis Policy Study data found that stronger marketing restrictions were associated with meaningfully lower levels of self-reported marketing exposure — and that underage respondents (ages 16–20) still reported high exposure even in legal recreational markets with moderate restrictions.

Put those findings next to the meta-analysis and the picture is reasonably consistent: ads exist; people see them; stronger rules reduce how many they see; and greater exposure correlates with greater self-reported use, especially for digital channels, especially in younger age brackets. None of that settles the policy question — reasonable people will weigh adult autonomy, industry viability, enforcement cost, and youth-protection differently — but it sets the empirical floor the debate should start from.

For context on where the broader legal landscape stands, our state-by-state cannabis laws guide and the U.S. legalization overview cover the policy terrain. The hemp-vs-cannabis regulatory landscape piece is also relevant — hemp-derived THC products are currently the least-regulated advertising category of all, and they are not what this meta-analysis measured.

What this means for readers who track their own use

TIWIH’s editorial position is that consumers are better off when they understand how cannabis research actually works, including the parts that are uncomfortable. The meta-analysis does not say you should feel guilty for seeing an ad. It does say that ad exposure is one of many environmental inputs that correlate with use patterns, especially digital ad exposure, especially in younger adults.

For parents, the relevant takeaway is simpler: if you are worried about your teenager and cannabis, the media-environment piece is worth paying attention to alongside the developmental-risk piece. Our teens-and-cannabis primer for parents covers the adolescent-brain research in more depth. For the broader question of what cannabis actually does in the body, the endocannabinoid system guide is a good place to start.

For everyone else: this is one more data point in a complicated picture. Stronger than nothing, weaker than proof. The researchers treated it that way. So should we.

We do not run outbound cannabis advertising on TIWIH. Our model is editorial plus disclosed affiliate relationships on gear. That is not a moral flex — it is a disclosure. You deserve to know where your information is coming from, and what the incentives around it look like. Separately, if you want to understand why crowd-sourced strain reviews and lab-testing standards complicate the “just read the label” advice, those two pieces go deeper on the information-environment problem.

Key takeaways

• The McClure-Thomas et al. (2026) meta-analysis in Addiction found a pooled adjusted odds ratio of 1.77 (95% CI [1.32, 2.30]) linking self-reported cannabis-ad exposure to cannabis use across 21 studies.
• The signal is strongest for digital and social-media advertising (aOR = 3.38), weaker but still significant for general advertising (aOR = 1.67), and not statistically significant for storefront or sidewalk signage (aOR = 1.25).
• Eighteen of 21 studies were cross-sectional, which means the evidence is associational, not causal. The authors treat it that way, and so should every honest summary.
• The young-adult signal (ages 18–25) is consistent with developmental neuroscience and with prior meta-analytic findings for alcohol and e-cigarette advertising. It is not anomalous.
• A companion Addiction paper (Harlow et al., 2026) found that each additional dispensary within one mile of home was associated with a 5–6% higher risk of past-six-month cannabis use in young adults, using the government-maintained registry.
• Stronger marketing restrictions (Hammond et al., 2026) appear to reduce self-reported exposure, but underage respondents still report high exposure even under moderate restriction regimes.
• None of this settles the policy question. It narrows the empirical floor the debate should start from.

Sources

• McClure-Thomas C, Yimer T, Strong C, Sun T, Hall WD, Chan GCK, Connor JP, Leung J. (2026). A systematic review and meta-analysis of self-reported exposure to cannabis advertising and its association with cannabis use and intentions. Addiction. doi: 10.1111/add.70310. PubMed
• Harlow AF, Williams MP, Pacula RL, Leventhal AM, Pedersen ER, Cockburn MG, Thompson LK, Barrington-Trimis JL, Haley DF. (2026). Cannabis dispensary exposure and smoked, vaped and edible cannabis use among young adults: Comparison of web-scraped and government-maintained registries. Addiction. doi: 10.1111/add.70356
• Hammond D, et al. (2026). Cannabis Marketing Restrictions and Exposure to Cannabis Marketing in Legal Recreational Markets: Findings from the International Cannabis Policy Study. Addiction. PMC open access
• Berg CJ, et al. (2025). A Narrative Review of Research on Cannabis Advertising in the United States. Current Addiction Reports, 12(92). doi: 10.1007/s40429-025-00703-1
• University of Queensland News. (2026, March 24). Does advertising increase cannabis use? Interview with Caitlin McClure-Thomas. UQ News
• Paruk S, et al. (2025). Impact of e-cigarette advertising, promotion, and sponsorship on cognition and behavior: a systematic review of public responses. BMC Health Services Research. doi: 10.1186/s12913-025-13929-6

This article summarizes observational and meta-analytic research. Correlation is not causation, and policy decisions involve value tradeoffs beyond the evidence base. Nothing here constitutes medical or legal advice.