You might remember the power pose.

The idea, popular in the early 2010s,  that if you hold the Superman position   your body reacts in ways that  make you feel more confident. The idea went viral, becoming one  of the most watched TED talks ever.

Then came the backlash, with  a slew of academics attacking   the science presented in that TED  Talk and claiming it was bogus. Years of back and forth ensued, with neither  side totally putting the debate to rest. And we can learn more from this science  beef than just the fact that it happened.

The power pose debate took  place at the same time as   an important shift – crisis, even – in psychology. It chronicles moving goalposts  and shifting expectations. So let’s see how the power pose  changes everything… or maybe nothing. [♪ INTRO] To start off, what was this whole power pose idea?

Huh, I think I do feel more confident… Or, do I?! The idea is basically that by changing  your body pose to postures of expansion,   you’ll feel more confident… and the pose might  affect your behavior and even your hormones! Now, the idea that by changing  our posture or facial expression   we can also alter our mental state isn’t new.

For example, yoga poses have been shown  to counteract the effects of stress. And no less than Darwin himself proposed  that by changing your facial expression,   you change the emotions you experience. But a particular study, and subsequent TED Talk,   really brought the idea into  the mainstream conversation.

In the late 2000s, three researchers decided to  dig into the idea of power posing and its effects. They brought 42 participants into the  lab and asked them to hold two positions,   either "high" or "low" power, for a minute each. Then they measured the effect of  holding those poses on three things.

One, they took saliva samples to measure  the levels of the hormones cortisol and   testosterone, supposedly corresponding  to feelings of stress and confidence. Two, they measured risk-taking behavior by seeing  if participants would undertake a gambling game. Finally, they measured participants’  self-reported feeling of power.

What they found was that holding high-power  poses significantly affected all three outcomes. They reported that holding high-power poses   lowered levels of cortisol  and increased testosterone. It also caused changes in behavior by making  participants more likely to take risks.

And participants reported feeling more powerful. In 2012, Amy Cuddy, one of  the authors of the paper,   stood on that famous stage and gave a TED Talk  telling audiences about these three findings. She finished the talk by asking the audience  to share her research with other people,   saying quote: "They need their  bodies, privacy and two minutes,   and it can significantly change  the outcomes of their life.” The TED Talk went viral, but the question is...   can those two minutes significantly  change the outcomes of our lives?

In 2015, three years after the viral TED Talk,   a study came out that changed the  conversation around the power pose. The study was initiated by a researcher  from the University of Zurich. She and her team recruited 200 participants,   more than 4 times the number of the  participants of the initial study.

But what they found was that while the  power pose does have an effect on the   feeling of power, the other two outcomes  of risk-tolerance behavior and hormones   didn’t pass muster – they didn’t clear  the bar of statistical significance. This is basically a way of deciding, with math,   if a result you see is unlikely to  be explained solely by random chance. And with that, we need to  zoom out and look at what was   happening in psychology in general in the 2010s.

This is the era of the replication crisis. Which we’ll talk about more  right after this quick break. Thanks for watching this SciShow video, and thanks  to JMP Statistical Discovery for supporting it!

JMP helps scientists turn messy,   complex data into something they  can visually explore and understand. And you don’t need to be an expert  statistician to get started. You can build these skills with JMP’s free,   on-demand workshop that lets you get hands-on  with data visualisation and exploratory analysis.

It’s designed for scientists,  engineers, and curious learners alike. There’s a sample data set of real world  data and webinars to give you all the   experience and guidance you need to make patterns,  relationships, and outliers stand out in seconds. Just download the free trial and  sample dataset to start your journey   into data visualisation and exploratory analysis.

It’s all in the link in the  description: jmp.com/scishow-dataviz Back then, some scholars were worried  that things like statistical luck and   bad methodology were behind some published papers. And to be fair, standards of psychology  methodology could definitely use some improvement! For example, small sample sizes, like the  42 participants in the power pose study,   were fairly typical and tended  to be statistically underpowered.

Or, instead of deciding on a fixed  number of participants, researchers   would keep recruiting new ones and running  tests until they reached positive results. And more systematically, there was, and still is,   a general positive publication bias where  significant findings are more likely to be   reported and published, and the only time  we saw experiments showing no significant   effect were people replicating or  arguing with earlier study results. With all these methodological problems and  findings that might not actually be positive,   a practice that can clear some  of the shadiness is replication.

In a replication study, researchers  perform similar experiments to see   if their new findings are consistent  with previously published results. If you can replicate something  somebody else has already found,   it’s a lot more likely to be a real effect. Considering that, in 2011, the Center  for Open Science led a project to   try to replicate the results of  100 popular psychology studies.

The results dropped in 2015, and they were  able to reproduce… drum roll please… ...39. Less than half of them. It was a real wakeup call, highlighting serious  flaws in the research methods used by psychology.

Meanwhile, in 2013, a group of researchers  co-founded Data Colada, a blog to stat-check   other academic research, evaluate shady  methodologies, and debunk bad research. One of these researchers went to school  with Cuddy and considered her a friend. But their relationship turned into  something more complicated when the   Data Colada crew addressed the power  pose study’s partial replication fail.

After the 2015 study, the one that only  partially replicated their findings,   the original authors of the power pose  study published a paper in response. They reviewed 33 studies saying  that many studies in the past,   including the 2015 study, show an effect  of the power pose on feelings of power. They also pointed out a few studies  that replicated their hormonal findings,   and talked about why the 2015 study might not  have reproduced what they initially found.

For example, that study had 200  participants, much more than theirs. Participants held each pose for  three minutes, while theirs was one. Furthermore, the original power pose  experiment came with a cover story.

Participants were told the researchers  were collecting data about their hearts,   and nothing about the effect  of holding a power pose. On the other hand, the 2015 study told  participants what was being studied,   but experimenters were blinded to  the conditions of the experiment. To be honest, if you compare the two side by side,  the 2015 study definitely seems a lot stronger.

And the Data Colada people agreed. They analyzed the same 33 studies from the review,  looked at the significance of the findings, and   concluded that collectively, they didn't clear the  bar for any effect of power pose on participants. With the caveat that their analysis did not   include that feeling of power  that the original study showed.

The blog post got massive traction. The field went wild, attacks got personal  and mean, more blog posts popped up,   and Cuddy made accusations  of bullying within the field. In the following years, however, even  more failed replications popped up.

That included a 2016 study that showed  no effect of power pose on behavior,   and a negative effect of the  pose on feeling of power. Another study published the same year in the   journal Biological Psychology  showed no effect on hormones. And then another criticism was leveled at

Cuddy:   that she was the only one attacking critics of  the study, while her co-authors remained silent. Co-incidence or not, a day  later, the first author, Carney,   posted on her website a piece saying that she  does not believe in the effects of power pose. She cited some of the methodological  problems of the study, like how they   chose what results to report on, how  they picked analysis methods that got   them more favorable findings,  and how tiny their sample was. To be clear, she didn’t make any accusations   of made up data or academic fraud…  just questionable research practices.

Carney actively discouraged people  from studying the topic further. We usually say that the answer is  always more research, not less. But in this case, she’s saying their methods  were too questionable to lead anywhere.

Though spoiler alert, she did later throw her  weight behind some power pose replication studies,   with much stronger methodologies – like  having the studies be pre-registered. What this means is before you run the study,  you let others know what you plan to do,   so that you can’t pick and choose favorable  outcomes once you already have the data. Those studies showed a small effect on  feeling of power, and not much else.

At this point something kinda revelatory  happened for the power pose topic. A professor from Iowa State University  noticed something important had been   missing from this research all  along: a neutral pose, as a control. We do see a difference between the power  pose and the low-power contractive pose,   but is it the power pose’s positive effect  on feelings that’s driving the difference?

Is it the contractive pose’s negative effect? Or is it both?! See, when you look at studies that include a  contractive position – like, you know... this –   the effect we see is not so much because the power  pose is making us feel greater than the neutral   pose, but probably because that smooshed posture  makes us feel much worse than the normal pose!

A 2022 systematic review with an eye on  the neutral pose confirmed this idea. It showed a change in feelings  of power and behavior… but due   to the negative effect of contractive pose. The conversation continued into 2026.

In a LinkedIn post, Cuddy discussed some  of the newer evidence for the power pose. And yet, it was once again quickly  pointed out that the post had some issues. For example, Cuddy correctly mentions   that a 2022 meta-analysis shows the  behavioral impact of body posture.

But she doesn’t mention that the study  also reports that these findings are   more likely because of bias in  publishing only positive findings,   or because there were some big outlier  numbers that threw off the general findings. In another case, one of the references  she used doesn't seem to exist,   and kind of looks like an AI hallucination. Come on!

She even references a study while  failing to mention that it comes   to the opposite conclusion regarding power pose. Cuddy’s post was titled “The  "Power Posing Was Debunked" Myth”. And despite the possible problems of that post… It’s true that it's not really the place of  the scientific process to debunk a study,   only to say that a study cannot be replicated.

With that said… when the study’s first   author says it doesn’t amount to  anything, that has some weight. We can say with some confidence that there  is an effect of the body’s posture on how   we feel… and that effect mainly  comes from not being smooshed. And as for behavioral and hormonal changes,   those results don’t seem to  be out there… not yet anyway.

If you pay attention, you might  notice this is less of a story   about the power pose and more  about the process of science. When you go back in time, you see  fewer systems of checks and balances,   and probably some really bad ways  we communicated our disagreements. And even though we knew how to design studies  correctly, you still saw people not doing that.

But today, we have developed better strategies we  hope will help us avoid the mistakes of the past. With today’s standards, we could  preregister the power pose study. This way, researchers could commit  to the study design and analysis   method and avoid making changes  just to reach positive outcomes.

And we could run a power analysis,  a statistical practice that helps   us know how many participants we need  for our study to show an actual effect. That could potentially have  guided the power pose study   to commit to a higher number of participants. And while we still don’t see many negative  research findings published in journals, at least   we have some open repositories where researchers  can post their non-significant findings.

Maybe if we incentivized publishing  negative results more, researchers   wouldn’t feel like they need to tinker with  their numbers to get significant outcomes. And I really hope that we also get  better at disagreeing with each other. After their critical blog post, the  Data Colada folk wrote a piece about   civil criticism, and there has been a lot  of conversations around academic bullying,   criticisms, and figuring out  how we can disagree better.

In the end, is the scientific process perfect? It sure is not, but we’re trying… and  doing a bit better than yesterday. [♪ OUTRO]