Disavowed ‘Power Pose’ Paper Illustrates Science’s Reproducibility Problem
What happens when junk science makes it into the mainstream?
In 2010, a team of researchers published a study that claimed to have discovered that so-called "power poses" (think Wonder Woman, head high and arms akimbo) can boost confidence and exude an aura of power.
It was bunk, but that wasn't clear at the time.
The study was chum in the water for the science press, and one of the researchers, psychologist Amy Cuddy, went on to forge a pop-science career out of the paper. In 2012, she delivered one of the most-viewed TED Talks ever, on power poses. Last year, she published a book on the power of body language called Presence. She currently lectures at Harvard Business School.
But late on Sunday, the lead author of the 2010 paper, Dana Carney at the University of California, Berkeley, posted a document on her personal faculty page completely disavowing her past research into power poses. "I do not believe that 'power pose' effects are real," Carney wrote." Moreover, she "[discourages] others from studying power poses.
"I have two words for you: Dr. Oz. Actually, two more: Donald Trump"
Carney doesn't want the original paper retracted, and she told retraction tracking site Retraction Watch that her study being more or less disproven in a 2015 study that was unable to reproduce her results was merely the normal functioning of good science. This is despite her own admission that the 2010 paper used extremely common methods to boost the attractiveness of results, including so-called "p-hacking" (essentially cherry-picking data) and excluding outliers.
"Reasonable and respected people may disagree with my opinion, but as new evidence came in, I merely updated my beliefs," Carney wrote me in an email.
"We can't use retraction as a neutron bomb to rid the literature of unverifiable results—there's too many of those sorts of articles to do it," said Adam Marcus of Retraction Watch in an interview. Indeed, last year the Open Science Collaboration project published a paper in Science that found less than half of psychology studies are reproducible.
"But this is a good example of how data, even produced honestly, can be so weak that it has no business being published in the first place—it never should have seen the light of day," Marcus continued.
When asked about why she believes the paper should not be retracted despite admitted methodological errors—again, very common ones—Carney wrote me in an email that, essentially, hindsight is 20/20.
"You realize over time the mistakes you make," Carney wrote. "They are clear now in hindsight. And what is p-hacking today was normative then and so most of us just went along with the flow—but we are now so clear about the error of our old ways."
Carney would not comment on Cuddy's continued promotion of the positive effects of power posing, only adding, "She may disagree with my position." Cuddy has not responded to Motherboard's request for comment.
Marcus had stronger words for Cuddy.
"Two words: Dr. Oz," he said. "Actually, two more: Donald Trump."
"People, unfortunately, believe what they want to believe, evidence notwithstanding," Marcus continued. "We're not a particularly science-literate culture, and we tend to endorse the worldviews that we hold and discard the facts that challenge them."
Get six of our favorite Motherboard stories every day by signing up for our newsletter.