In 2010 three researchers published a paper about posture, hormones, and risk. Forty-two people held a pose for a couple of minutes and their testosterone and cortisol moved in the directions the paper predicted. Two years later one of the authors gave a TED talk that CNN and Oprah picked up, and by 2017 it had been watched about forty-seven million times, making it the second most popular talk in the history of the conference. Political parties adopted the stance for photo calls. And then, in 2016, the lead author of the original paper posted a statement on a university website saying she no longer believed any of it. What makes the story worth an evening is not that a famous study failed to replicate. It is what the reanalysis found sitting underneath it.
🧠 Forty-Two People and a Bathroom Stall
The question was simple and the design was small. Dana Carney, Amy Cuddy, and Andy Yap, writing in Psychological Science, had participants hold either an expansive posture — the open, limbs-spread shape that got nicknamed the Wonder Woman pose — or a contractive one, arms folded, shoulders curled. Hormone levels were measured before and after. The expansive group showed a rise in testosterone and a drop in cortisol; the contractive group went the other way. The paper’s own summary of what this meant is the sentence that did the travelling: these findings suggest that, in some situations requiring power, people have the ability to fake it ‘til they make it.
The applications the authors sketched were concrete, and that is part of why it spread. Stand like this for two minutes before a job interview. Stand like this before a negotiation. One popular image of the technique in practice, as the encyclopaedia entry puts it, is candidates locking themselves in bathroom stalls before interviews to make victory V’s with their arms.
Cuddy became the public face of it. Her interest in the subject came out of her own life: she had recovered from head trauma after a car accident, and the research spoke to people who feel chronically powerless because of money, rank, or the group they were born into. She put it in a 2015 book and argued for it from stages. In 2015 a set of British politicians began standing with their feet planted unusually wide at high-profile events, and the press gave it a name: the Tory power pose. Sajid Javid, George Osborne, David Cameron and Theresa May were all photographed in it, and Tony Blair had been doing it for years before any of the science existed.
🤔 Then the Lead Author Published a Statement
The failure arrived in stages, and the first stage is the interesting one. In 2015 Ranehill and colleagues ran a conceptual replication with a larger sample. They reproduced the part about how people felt. They could not detect a single physiological or behavioural effect. The feeling survived; the mechanism did not.
In 2016 Simonsohn and Simmons took the p-curve to the body of evidence and concluded that once selective reporting is accounted for, the research does not suggest an effect exists at all. Two more 2016 attempts went the same way: in one, an expansive pose actually reduced participants’ feelings of power, and in a stress-test study the posture manipulation moved nothing — not hormones, not behaviour, not subjective ratings.
Then Carney, the first author of the original paper, published a statement of her own: she did not believe power pose effects were real, and she called the evidence against them undeniable. The person with the most to lose from saying so said it anyway. A special issue of Comprehensive Results in Social Psychology followed in 2017 with eleven new preregistered studies and a meta-analysis. The effect on power behaviour did not replicate. The effect on feeling powerful came in at a medium size. In the same year Cuddy and colleagues p-curved fifty-five studies and argued there was solid evidential value in postural feedback, strongest for mood, attitudes, and feelings about the self.
Cuddy kept working. She also reported harassment, including death threats, after the replication failures. In the spring of 2017 she left Harvard.
So where does that leave the pose? A comprehensive meta-analysis of 128 studies, published and unpublished, is the closest thing to a verdict. Its finding splits cleanly. On thoughts and feelings — positive mood, self-esteem, a sense of dominance — the effect is reliable. On physiology, which was the entire point of the 2010 paper: nothing. Hormones, blood pressure, skin conductance, all flat. On behaviour there may be an effect, but the authors can’t say whether it is real or an artefact of the significant results getting written up.
And then the cut. Almost none of those studies included a neutral posture as a control group. They compared a dominant pose to a slumped one, and the meta-analysis points out that this leaves the question open: we do not know whether the effect comes from the expansive pose at all, or from the contractive one doing damage. The problem has a name, borrowed from medicine. Compare a drug to a poison and find that the drug group lives longer, and you have not shown that the drug works. You may have shown that the poison kills. Maybe the Wonder Woman pose does nothing and the slouch is the whole villain of the literature.
🔗 The Layer That Actually Pays
Sit with the shape of that for a second, because it is a product lesson wearing a psychology costume. A small physical act, a couple of minutes long, produces a change in how someone feels, which then shows up in how they behave. That is a real and useful chain. It is also the only part of the chain that survived contact with replication, and it was the part everyone treated as a footnote while the hormones got the headlines.
This is roughly the situation any prediction or guidance product lives in. The mechanism story is what sells — the reading, the model, the instrument, the thing that claims to explain why. The part that actually gets delivered is much narrower and much less glamorous: how the user feels walking out, and what they do in the next hour. Getting clear about which layer you are actually selling is uncomfortable, because it means your most impressive-sounding feature may be decoration, and your most modest-sounding one may be the product. The reanalysis here is the cautionary version. The forty-seven-million-view version of the finding has been shot down for a decade, and the modest version that survives is the one nobody put in a TED talk.
There is a second lesson about honesty and it is the middle of the story rather than the end. The science did not get corrected because a hero auditor arrived. It got corrected because the authors of the original work went looking for the replication and then said so in public, one of them against her own interest, and because the field had built machinery — preregistration, multi-lab designs, p-curve analysis — that could survive the answer. A system that can absorb “we were wrong” is not a weak system. It is the only kind that produces anything durable.
And there is a smaller one for anyone building characters. A ritual is a better literary device than a mechanism. A person hiding in a bathroom stall making victory V’s with her arms before the worst interview of her life is a scene, regardless of what her cortisol did. The hormone explanation is a footnote that got promoted by accident.
🎲 The Free Half of the Trick
Two things to take away, and the second one is a free lunch.
The trivia first: the missing control group means the cheapest version of the intervention might be the one that got no press. If the low-power condition is what suppresses, then the finding, read honestly, is not “adopt a power pose” but “stop collapsing.” Posture feedback may be real and still cost you nothing except the slouch. That is a much duller promise, which is exactly why it never made it onto a stage.
The exercise is short. Before the next thing you are nervous about, notice what your body is doing right now, before you change anything. Most people reach for a confidence trick while sitting folded around their own phone. Run the comparison on yourself: no pose, and then upright. Those are the two conditions the literature never properly separated, and you can run both for free in about a minute.
One line of the original paper is still worth keeping, minus the hormones. In some situations requiring power, people have the ability to fake it until they make it. The evidence took away the endocrine explanation. It did not take away the thing you are doing in the bathroom stall.