Amy Cuddy’s 2012 TED talk on power posing has been viewed over 75 million times. Her own co-author publicly disavowed the original finding in 2016. A large replication failed to find the hormonal effects. Here is what the data actually shows, what survived, and what leaders should learn about a finding whose cultural life outlasted its scientific life by a decade.

In June 2012, Harvard Business School professor Amy Cuddy delivered a TED talk in Edinburgh called “Your Body Language May Shape Who You Are.” The premise was electric. Spend two minutes in a “high-power” pose --- feet on the desk, hands behind the head, or arms raised in a “V” --- and you will measurably change your hormones. Testosterone goes up. Cortisol goes down. You will be more willing to take risks. You will literally become more powerful.

The talk became one of the most-watched TED presentations of all time. It has now passed 75 million views. Cuddy followed up with a bestselling book, Presence (2015), which extended the framework into a general theory of self-presentation. The “power pose” became a culturally recognizable shorthand. Athletes adopted it before competitions. Job applicants used it in bathroom stalls before interviews. Business schools taught it in leadership modules. Wonder Woman became, briefly, a leadership consultant.

Four years after the TED talk, Dana Carney --- Cuddy’s first co-author on the original 2010 paper --- posted a public statement on her Berkeley faculty page that began with the sentence: “I do not believe that ‘power pose’ effects are real.”

This article walks through what the original 2010 paper actually claimed, what the major replication attempts found, what Cuddy’s own subsequent defense of the construct showed, what survived, and what any of this means for leaders, founders, or anyone considering whether to spend two minutes in a bathroom stall striking a Wonder Woman pose before an interview.

What the Original Paper Actually Said

The founding paper is Carney, Cuddy & Yap (2010), “Power posing: Brief nonverbal displays affect neuroendocrine levels and risk tolerance,” in Psychological Science. The sample was 42 college students --- 26 women and 16 men.

The study had three components. Half the participants were assigned to hold “high-power” poses (expansive, open postures like leaning back with hands behind the head, or standing with arms raised). Half held “low-power” poses (contractive, closed postures like sitting hunched with crossed arms). Each pose was held for two minutes --- one minute for each of two postures.

Then three dependent measures were collected. Subjective feelings of power on a self-report scale. Risk tolerance, measured by whether participants chose to gamble with a chance to win money. And saliva-based hormone measurements --- testosterone and cortisol --- taken before and after the pose manipulation.

The paper reported four main effects. High-power posers reported feeling more powerful. They were more willing to gamble. Their testosterone increased relative to baseline; low-power posers’ testosterone decreased. Their cortisol decreased; low-power posers’ cortisol increased.

This was a remarkable package of findings. A two-minute behavioral manipulation supposedly produced measurable hormonal changes --- the same hormones associated with dominance in primate research --- along with corresponding shifts in subjective experience and risky behavior. If true, it was a dramatic demonstration of mind-body coupling and a powerful, immediately actionable intervention. You could become more powerful just by standing differently for two minutes.

There were always reasons for skepticism. The sample size was small. The effect was novel and surprising. The hormonal changes implied unusually rapid neuroendocrine responses to a brief behavioral manipulation. But the paper appeared in a respected journal, the effect was vivid, and the implications for self-presentation, leadership, and gender dynamics were appealing enough that critical scrutiny took years to arrive.

The Cultural Explosion

What happened after the 2010 paper is one of the clearest examples of how a behavioral-science finding can become culturally dominant on the strength of presentation rather than data.

Amy Cuddy delivered her TED talk in June 2012. The talk was framed not as a tentative report of a preliminary finding but as a confident summary of “what the research shows.” Cuddy was a compelling speaker. The talk was visual, emotional, and concluded with a personal story about her recovery from a traumatic brain injury. The combination was extraordinarily effective. The talk went viral.

The “two minutes” framing was crucial to the talk’s cultural traction. A two-minute intervention is easy to remember, easy to try, and produces no observable cost if it doesn’t work. The intervention’s accessibility --- anyone, anywhere, can stand in a bathroom and try it --- meant that millions of people did. Many of them reported feeling more confident afterward. Their reports became personal validation that “power posing works,” even though feeling more confident after deliberately trying to feel more confident is not a particularly strong test of the underlying hormonal claim.

Cuddy’s 2015 book Presence extended the framework into a general theory of self-presentation under stress. Reviewers were enthusiastic. The book sold well. Cuddy became one of the most-quoted behavioral scientists in business media. Power posing became a fixture in leadership development programs, executive coaching curricula, and corporate confidence-building seminars.

The original 2010 paper had a sample of 42 students. By 2015, the cultural construct that had grown around it was a fully developed leadership philosophy with millions of practitioners.

The Replication That Wasn’t Going to Stay Quiet

In 2015, Eva Ranehill and four colleagues published “Assessing the robustness of power posing: No effect on hormones and risk tolerance in a large sample of men and women,” in Psychological Science --- the same journal that had published the original Cuddy paper.

The Ranehill study was a direct replication of the original 2010 design, with one crucial change: a much larger sample. They tested 200 participants, nearly five times the original sample size.

The results were stark. The subjective feeling-of-power effect replicated --- participants who held high-power poses reported feeling more powerful. But the hormonal effects did not replicate. Testosterone did not change differentially between high-power and low-power posers. Cortisol did not change differentially. And the risk-tolerance effect --- the behavioral consequence that had been part of the original story --- also failed to replicate.

This was not a marginal failure. The Ranehill study had enough statistical power to detect substantially smaller effects than the original paper had claimed. The hormonal and behavioral effects that had defined the original finding simply weren’t there in a larger, more carefully run replication.

The paper’s title --- “No effect on hormones and risk tolerance in a large sample of men and women” --- was unusually direct for an academic publication. The replication failure was reported in major media. The power-posing construct entered a period of methodological scrutiny that would last for years.

Carney’s 2016 Disavowal

In September 2016, Dana Carney --- Cuddy’s first co-author on the original paper --- posted a statement on her UC Berkeley faculty webpage. The statement began with the sentence: “I do not believe that ‘power pose’ effects are real.”

Carney went on to describe specific problems with the original 2010 study. The sample size was very small. The data collection involved choices that, in retrospect, looked like researcher degrees of freedom --- including the decision to drop certain conditions and certain participants from analysis. She wrote, with unusual candor for an academic disavowal, that “the evidence against the existence of power poses is undeniable” and that she no longer studied them, taught them, or recommended them.

A first author publicly disavowing her own famous paper is rare. Carney’s statement was, in academic terms, devastating. It signaled that even the original team had lost confidence in the central claim.

Cuddy, by contrast, continued to defend the construct.

The 2018 P-Curve Defense and Its Problems

In 2018, Cuddy, Schultz, and Fosse published “P-curving a more comprehensive body of research on postural feedback reveals clear evidential value for power-posing effects,” in Psychological Science. Their argument was technical: across a wide set of studies on “postural feedback” --- a broader category than power posing specifically --- the p-curve analysis indicated evidential value for the construct.

P-curving is a statistical technique that analyzes the distribution of p-values across studies to detect publication bias and assess whether an underlying effect is real. The technique is valuable when used carefully.

Critics, including the methodology blog Data Colada (post 66 in their series), argued that Cuddy’s p-curve had a substantive logical problem. The studies she included compared “expansive” versus “contractive” postures, with the implicit framing that effects in either direction counted as evidence for “postural feedback.” But the original power-posing claim was specifically that expansive poses had positive effects. A study showing that contractive poses produced negative effects on confidence or behavior would, in Cuddy’s analysis, count as evidence for power posing --- even though it doesn’t bear on the original directional claim. This is a serious logical issue with the p-curve methodology as applied.

Other critics pointed out that the broader “postural feedback” literature includes studies of postures very different from the original power-pose paradigm, including some studies on body language interpretation and emotional expression that don’t really speak to the question of whether briefly holding a Wonder Woman pose produces measurable hormonal or behavioral changes.

In academic terms, the Cuddy 2018 defense is widely regarded as a substantially weakened version of the original claim, and its specific methodology has been substantively criticized. It does not reverse the verdict of the Ranehill replication on the hormonal and behavioral effects. It does provide some statistical evidence for a weaker construct --- something like “expansive postures produce small subjective feelings of power” --- but that is a much less impressive claim than the original “two-minute pose changes hormones” framing.

What Survived

The honest current verdict on power posing distinguishes between several specific claims, each of which has held up differently under scrutiny.

The subjective feeling-of-power effect survives, at small magnitude. Multiple replications, including Ranehill 2015, found that holding expansive postures briefly does produce small increases in self-reported feelings of power. This is the most robust component of the original finding. It is also the least surprising --- being told to assume a “powerful” posture and then asked how powerful you feel is the kind of priming-and-self-report combination that tends to produce small effects from demand characteristics alone.

The hormonal effects do not replicate. Testosterone and cortisol changes from brief postural manipulation have not been observed in any well-powered replication. The 2010 hormonal claim is, on current evidence, not supported. Gronau et al.’s 2017 Bayesian meta-analysis --- which was specifically scoped to the “felt power” subjective effect --- found support only for that narrower subjective claim, leaving the hormonal effects unsupported by the broader replication literature.

The behavioral effects (risk tolerance, persistence) do not replicate. The original claim that high-power posing made people more willing to take risks or persist longer on tasks has not survived rigorous replication.

A small “felt power” effect may be partly demand characteristic. Because Cuddy’s TED talk has been viewed 75 million times, many participants in subsequent power-posing studies have prior knowledge of what the study is “supposed” to show. This makes interpretation of the surviving subjective effect complicated --- some portion of the effect is plausibly explained by participants doing what they think the experiment is testing for, rather than by any genuine bodily-feedback mechanism.

The honest current picture: a small subjective effect probably exists. The dramatic hormonal and behavioral effects that made the original finding famous do not.

Why the Original Looked Real

The power-posing story is a clean case study in how a single small study can become culturally dominant.

Small sample plus novel finding plus prestigious journal. Forty-two participants is a small sample. A novel finding with a small sample published in a respected journal is the demographic profile of “claims that may not survive replication.” This was true in 2010 and is true now. The publication did not validate the claim; it merely made it visible to a broader audience.

A charismatic communicator and a viral platform. Amy Cuddy was an exceptional speaker. The TED format is designed to maximize emotional impact and minimize methodological caveats. The combination of a compelling speaker, an emotionally affecting personal story, and a visually memorable intervention was extraordinarily effective at cultural propagation. The talk would have spread regardless of whether the underlying claim was true.

An intervention with zero observable cost. Power posing is free to try. It takes two minutes. There is no observable downside if it doesn’t work. This made it spread further than interventions with friction or expense would have. It also made personal validation easy --- millions of people tried it, many felt more confident afterward (because they were trying to feel more confident), and reported personal validation back into the cultural conversation.

Cultural alignment with empowerment narratives. The framing --- “you can change your body chemistry and feel more powerful in two minutes” --- aligned with a broader cultural movement around self-presentation, confidence, and women’s leadership. Cuddy’s talk was particularly resonant for women in professional environments. The cultural alignment provided amplification beyond what the underlying evidence would have generated.

The replication crisis was still nascent in 2012. Power posing went viral in the same year that some of the foundational replication-crisis literature was being published. Most readers, journalists, and even academic colleagues did not yet have the calibration to read a small N=42 finding with appropriate skepticism. By the time the field had developed that calibration, power posing had become a cultural construct that was much harder to displace than a journal claim would have been.

The Honest Verdict Today

A small subjective “felt power” effect probably exists. The strong claims that defined the original finding --- that two minutes of high-power posing produces measurable hormonal changes, that it increases risk tolerance, that it changes behavior in ways linked to those hormonal changes --- are not supported by current evidence. The first author of the original paper has publicly disavowed those claims.

If you are deciding whether to spend two minutes in a bathroom stall striking a Wonder Woman pose before a job interview, the honest answer is: it probably won’t change your hormones, but it might make you feel slightly more confident, partly because you are deliberately trying to feel more confident. Whether that is worth two minutes of your time is up to you. The science does not support the dramatic original claim, but the placebo-like subjective effect appears to be small but real.

If you are a leadership development professional, executive coach, or someone selling training products built around power-posing claims, the honest verdict is more uncomfortable: the construct as originally framed is not well-supported, and continuing to teach it as if the original 2010 hormonal findings were established science is not defensible on the current evidence.

If you are evaluating behavioral-science claims more generally, the power-posing story is a useful template for what to look for: small original sample, charismatic communicator, viral platform, accessible intervention, cultural amplification, and then several years of accumulating replication failures that the cultural construct largely ignores.

What This Means If You’re a Strategist

Three takeaways for leaders, founders, consultants, or anyone whose work involves either using or selling behavioral-science claims.

1. Viral cultural reach is not evidence. Power posing has been viewed by tens of millions of people through Cuddy’s TED talk. That cultural reach is enormous. It is also not evidence of any kind about whether the underlying claim is true. There is essentially no relationship between how many people have heard of a behavioral-science finding and how strong the evidence for that finding is. Cultural reach is downstream of communication quality, narrative fit, and amplification platforms --- not downstream of evidence quality.

When you encounter a behavioral-science claim with enormous cultural reach --- power posing, the marshmallow test, ego depletion, the various “decision-making” framings popular in business media --- apply the same evidentiary scrutiny you would apply to a claim with no cultural reach. The cultural reach is informative about communication quality, not about the underlying claim. Treating “everyone knows about this” as a credential is a systematic mistake.

2. Original-author disavowals are signal-rich. When the first author of a famous paper publicly disavows the central claim, that is unusual and informative. Carney’s 2016 disavowal was about as direct as academic disavowals get: “I do not believe that ‘power pose’ effects are real.” When you see this kind of statement from an original author, it should substantially update your confidence in the construct, regardless of how many other researchers continue to defend it.

This is rare enough that you can use it as a heuristic. When evaluating a contested behavioral-science finding, check whether any of the original authors have publicly walked back the claim. If yes, the construct is in unusually weak shape. If no, you have less direct signal --- but the absence of disavowal doesn’t mean the construct is strong.

3. Personal validation is not evidence either. The most resilient cultural defense of power posing is personal: “I tried it before a job interview and I got the job.” Personal validation of any behavioral intervention is essentially worthless as evidence. Confirmation bias, placebo effects, demand characteristics, and selection bias (you remember the times it worked, not the times it didn’t) all produce systematic personal validation of interventions that don’t actually work. This is true of power posing, lucky socks, pre-meeting rituals, breathing exercises, and most other small behavioral interventions.

For leaders and consultants, the implication is that you should be very skeptical of personal anecdotes --- both other people’s and your own --- when evaluating whether to recommend, teach, or build strategy around a behavioral-science claim. The plural of anecdote is not data, and the plural of personal-validation anecdote is especially not data. Look for properly designed studies with adequate sample sizes and preregistered analyses. If those don’t exist, hold the construct provisionally regardless of how many people swear it changed their lives.

Sources

This article is part of an ongoing series on famous behavioral science studies that did not survive replication. Other entries cover the Stanford Prison Experiment, the marshmallow test, ego depletion, the bystander effect, and the Mozart Effect. The full hub lives at /replication-crisis/.

If you’ve built leadership development, training, or coaching programs on power-posing or similar behavioral-science claims and want a careful evidence review, book a consultation.

FAQ

Does power posing work at all? A small subjective effect on self-reported feelings of power probably exists. The dramatic hormonal and behavioral effects that defined the original 2010 finding do not survive rigorous replication. If you mean “will it make me feel slightly more confident for a few minutes,” probably yes. If you mean “will it change my hormones and make me more powerful,” probably not.

Why is Amy Cuddy still teaching it? Cuddy has continued to defend a modified version of the construct (the “felt power” component) and disputes the strength of the replication failures. Reasonable people can disagree about how much weight to give her defense versus the Ranehill replication and Carney disavowal. The current academic consensus is that the strong original claims are not supported, but Cuddy and her current collaborators have not accepted this verdict.

What does Dana Carney say now? Carney published a public statement in 2016 disavowing the original finding. She has not retracted that statement. She is no longer a power-posing researcher and reports that she does not teach or recommend the construct. Her position has been consistent for nine years.

Is the TED talk still up? Yes. TED has not removed Cuddy’s talk despite the replication failures and Carney’s disavowal. This is consistent with TED’s general policy of not retracting popular talks even when their underlying claims have been substantially challenged. The talk is still recommended in many leadership development contexts.

Should I still try it before an important meeting? If you find that briefly standing in an expansive posture before a stressful situation makes you feel slightly more confident, the cost is essentially zero and the small subjective benefit appears to be real. Just don’t expect hormonal changes or dramatically different behavior. Treat it as a self-confidence ritual, not a neuroendocrine intervention.

Are there other “body language” claims that didn’t replicate? Many. The broader “embodied cognition” literature has had a tough time with replication. Several specific findings --- that holding a warm coffee makes you judge people as warmer, that physically washing your hands reduces moral judgment of past actions, that nodding affects belief --- have either failed direct replication or had effect sizes substantially revised downward. The “physical posture changes psychological state” framing remains a contested area of social psychology.

replication-crisis behavioral-science leadership evidence-evaluation social-psychology

Free Tool

Built for Experimentation Teams

GrowthLayer is the experimentation platform I built for CRO teams --- test management, AI-powered insights, and pattern recognition across your entire program.

Explore GrowthLayer → (opens in new tab)

· Start Free →

Share this article

LinkedIn (opens in new tab) X / Twitter (opens in new tab)

Copy link

Go deeper

Methodology The PRISM Method Case Studies $30M+ in Results Work Together Services & Mentoring

Experimentation and growth leader. Builds AI-powered tools, runs conversion programs, and writes about economics, behavioral science, and shipping faster.

About LinkedIn Newsletter

← Previous

The Stanford Prison Experiment: How a Famous “Truth” About Human Nature Came Apart

Next →

The Mozart Effect: How a 36-Person Study Became a State Policy --- And Why It Was Never There

replication-crisis behavioral-science leadership evidence-evaluation social-psychology
Share this article
LinkedIn (opens in new tab) X / Twitter (opens in new tab)
Atticus Li

Experimentation and growth leader. CXL-certified CRO practitioner, Mindworx-certified behavioral economist (1 of ~1,000 worldwide). 200+ A/B tests across energy, SaaS, fintech, e-commerce, and marketplace verticals.

About LinkedIn Newsletter