Is the New Cannabis-Cortisol Scare BS Science or BS Journalism?
Bullshit Science, or Bullshit Journalism? Dissecting the Cannabis-Cortisol Scare
“Think Marijuana Helps Relax? New Study Links Cannabis With Higher Levels Of Stress Hormone.” That’s the headline that’s been making the rounds, courtesy of a health tabloid site that specializes in the kind of alarming framing designed to make you click before you think. It’s built to make stoners everywhere quietly reconsider their evening ritual. So let’s do what that headline actively discourages: read the actual study.
I went and found it. It’s a real paper, published in the peer-reviewed journal Cannabis, put out by the Research Society on Marijuana, written by three researchers at Oregon State University: Julia Donner, Alexia Obrochta, and Anita Cservenka. I want to be straight with you here, because being straight with you is the whole point of this column. This is not junk science. It’s a legitimate, modestly-scoped study that got run through a media telephone game until it came out the other end sounding like something it isn’t.
Here’s the design. The researchers recruited 39 frequent cannabis users and 43 non-using controls, 82 people total. Over two consecutive weekdays, participants collected saliva samples at three points: right when they woke up, thirty minutes later, and again in the evening. The researchers were measuring something called the cortisol awakening response, or CAR, which is the normal spike in cortisol that happens in the first half hour after you open your eyes. It’s a well-established marker researchers use to study how the body’s stress-response system, the hypothalamic-pituitary-adrenal axis, is functioning day to day.
The headline result, the one that’s blowing up group chats right now, is that cannabis users had higher cortisol levels at the moment of waking, before that morning spike even happened, compared to non-users. That’s real. It hit statistical significance at p = .047, just barely under the conventional 0.05 threshold researchers use to call a result unlikely to be chance.
Here’s what the headline conveniently leaves out. The actual size of the morning cortisol spike, the CAR itself, showed no significant difference between cannabis users and non-users at all (p = .23). The thing the study was actually designed to measure came back a wash. The one finding that did clear significance was a baseline difference in resting cortisol, not a difference in how the body responds to the stress of waking up.
Go back and read that headline again: “New Study Links Cannabis With Higher Levels Of Stress Hormone.” Linked. Not caused. Associated. The lead researcher herself, Anita Cservenka, said it about as plainly as a scientist can in the university’s own press release: “The research doesn’t prove frequent cannabis use causes people to wake up with higher cortisol levels, or vice-versa. We would need to study the same groups of people over a longer period of time to understand any cause-and-effect relationship.”
Read that last part again. The lead author is explicitly telling you the direction of causation is unknown, if there’s causation at all. Maybe chronic cannabis use nudges baseline cortisol upward. Or maybe people with naturally higher baseline cortisol, who are already running a bit more stressed day to day, are more likely to reach for cannabis in the first place and become frequent users. That’s not a hypothetical hedge researchers toss in to cover themselves. It’s a real, unresolved possibility that a single cross-sectional snapshot of 82 people cannot distinguish between, and Cservenka says so herself.
Eighty-two people is not nothing, but it’s not much either, especially once you split them into two groups of 39 and 43 and start running regressions on subjective stress ratings, three different collection timepoints, and group interactions. A result that clears p = .047 in a sample that size sits right at the edge of what statisticians call a fragile finding, the kind that can flip with a slightly different sample, a slightly different recruitment pool, or a slightly different analytic choice. Cservenka’s own team found exactly that kind of fragility when they tried to replicate an earlier study’s finding that cannabis blunts the CAR itself. They didn’t replicate it. Same broad research question, different sample, different result. That’s not a knock on either study. It’s exactly how small-sample psychophysiology research behaves, and it’s precisely why researchers describe single studies like this one as “a starting point” rather than an answer.
There’s also the demographic question the coverage never raises. This sample was recruited in and around Corvallis, Oregon, a college town built around a single university. We don’t know from the press coverage how representative that population is of cannabis users nationally in terms of age, use patterns, product type, or potency. A study of college-town cannabis users in a state with a mature legal market doesn’t necessarily generalize to a 45-year-old edible user in Ohio or a daily concentrate user in Colorado.
It’s worth sitting with what the researchers actually measured before assuming it translates to “cannabis makes you stressed” in any everyday sense. Cortisol at the moment of waking, before the morning spike, is a single snapshot of one hormone at one instant across two mornings. It’s not a diagnosis of anxiety. It’s not a measure of how a person feels day to day, and in fact the study found that subjective stress ratings correlated with cortisol regardless of cannabis use status, meaning the users who felt more stressed had higher cortisol for the same reason non-users who felt more stressed did. The interesting wrinkle Cservenka herself highlighted is that the elevated baseline correlated with self-reported negative consequences of cannabis use, not with cannabis use itself. In other words, the strongest relationship in the data wasn’t “cannabis user versus non-user.” It was that among cannabis users, the ones reporting more problems from their use also tended to have higher morning cortisol. That’s a meaningfully narrower finding than “marijuana raises your stress hormone,” and it points toward a much more interesting, much less clickable question: whether cortisol dysregulation might eventually serve as a biological marker for problematic use specifically, not recreational use generally. That’s exactly the framing Cservenka used, and it’s exactly the framing that vanished by the time the story reached a health tabloid headline writer.
None of this is really Oregon State’s fault. Their press release, to their credit, includes Cservenka’s causality disclaimer right in the body text, along with an honest acknowledgment that they failed to replicate prior findings. The distortion happens downstream. A university PR office writes a careful release. A science aggregator picks it up and tightens the caveats to make room for a punchier headline. A health tabloid site picks up the aggregator’s version, strips out what’s left of the nuance, and slaps on a headline engineered for maximum alarm and maximum clicks, because “study finds weak, unreplicated, correlational link in small sample” doesn’t generate ad revenue the way “cannabis linked to stress hormone” does.
This is the actual scandal, and it’s not really about cannabis. It’s about an entire media ecosystem built on an incentive structure where nuance is a liability. Every field of science gets this treatment, from nutrition to psychology to climate research. Cannabis just gets it more often and more aggressively, because sixty years of prohibition-era messaging built an audience primed to believe the worst about it on minimal evidence, and outlets know exactly which headlines that audience will click.
Next time a cannabis headline crosses your feed, run it through three questions before you let it change your Tuesday night routine. First, is the language “linked to” or “associated with,” or does it say “causes”? Those are not the same claim, and reputable coverage should never blur them. Second, what’s the sample size, and does the effect size look proportional to the confidence of the headline? A p-value of .047 in 82 people is not the same statistical weight as a finding replicated across ten thousand people in five countries. Third, what do the researchers themselves say about their own findings, buried in the original press release or the abstract, before three layers of aggregation stripped it out?
The Oregon State study is legitimate, careful, appropriately hedged science asking a real question about how frequent cannabis use might interact with the body’s stress machinery. It deserves to be taken seriously and it deserves follow-up research, exactly as its own authors say. What it doesn’t deserve is a headline telling you cannabis is quietly wrecking your stress hormones when the study’s own lead author won’t even claim that much. The science isn’t bullshit here. The journalism wrapped around it is.
Sources
Donner, Obrochta, and Cservenka, “Subjective Stress and Diurnal Salivary Cortisol in Adults with Frequent Cannabis Use,” Cannabis, June 9, 2026
Oregon State University Newsroom, “Frequent cannabis users wake up with more of the ‘stress hormone’ in their system,” July 21, 2026
Health and Me, “Think Marijuana Helps Relax? New Study Links Cannabis With Higher Levels Of Stress Hormone,” July 27, 2026
HOW CANNABIS HELPS STRESS, READ ON…
HOW CANNABIS CAN HELP WITH STRESS AND ANXIETY?
Slang Drug Combinations – Candy Flips, Speedballs, and the Secret Language of Mixing What Does Cannabis Rescheduling Mean for Hemp Beverages?
Comments are currently closed.