Late Sunday night, Dana Carney, a co-author on the 2010 Psychological Science study that gave rise to the notion of “power posing,” the idea that adopting open, assertive postures for brief spans can have a variety of beneficial physiological and psychological effects, posted a rather shocking admission on her faculty website: She doesn’t think the effect is real. Carney wrote that she was swayed both by some recent failed replications of power-posing effects, but also by her realization that she and her co-authors engaged in what she now believes was p-hacking, a set of statistical tricks that can inflate the seeming importance of a research finding. It’s impossible to read Carney’s note and come away believing that she thinks the original power-posing study was well-conducted or, from a results perspective, particularly relevant.
Within the psychological-research community, this is a very big deal. While there had been a growing consensus that power-posing effects were wobbly at best and probably not real because of those failed replications, it’s another thing entirely for one of the original paper’s authors to come out so strongly against the original paper’s findings. Carney’s note also casts a harsh spotlight on her co-lead-author, Harvard Business School professor Amy Cuddy, who has built a lucrative research and speaking career out of promoting power posing and related findings — as is obligatory to point out in any journalistic mention of Cuddy’s success, her TED talk on power posing, with 36 million views, is the second-most-watched TED talk of all time. (The paper’s third author, Andy Yap, has, like Carney, been more or less silent about power posing ever since the original paper’s publication.)
There is a lot of dust that hasn’t yet settled, and it will be interesting to see how Carney’s post fits into psychology’s broader, rather urgent conversation about openness, replication, and scientific rigor and integrity in general. But one interesting and underappreciated element of the fall of power posing is the extent to which the idea was built on the foundation of other research findings that have, themselves, come under heavy failed-replication fire.
Most scientific-research papers are structured in roughly the same way. Early on, the authors explain the past findings that led them to be interested in whatever it is they are testing in the first place. In Carney, Cuddy, and Yap’s paper, there’s this paragraph:
In research on embodied cognition, some evidence suggests that bodily movements, such as facial displays, can affect emotional states. For example, unobtrusive contraction of the “smile muscle” (i.e., the zygomaticus major) increases enjoyment (Strack, Martin, Stepper, 1988), the head tilting upward induces pride (Stepper & Strack, 1993), and hunched postures (as opposed to upright postures) elicit more depressed feelings (Riskind & Gotay, 1982). Approach-oriented behaviors, such as touching, pulling, or nodding “yes,” increase preference for objects, people, and persuasive messages (e.g., Briñol & Petty, 2003; Chen & Bargh, 1999; Wegner, Lane, & Dimitri, 1994), and fist clenching increases men’s self-ratings on power-related traits (Schubert & Koole, 2009). However, no research has tested whether expansive power poses, in comparison with contractive power poses, cause mental, physiological, and behavioral change in a manner consistent with the effects of power. We hypothesized that high-power poses (compared with low-power poses) would cause individuals to experience elevated testosterone, decreased cortisol, increased feelings of power, and higher risk tolerance. Such findings would suggest that embodiment goes beyond cognition and emotion and could have immediate and actionable effects on physiology and behavior. [bolding mine]
The bolded paper, lead-authored by the German psychologist Fritz Strack, is one of the most famous in modern psychological history. The findings from that paper and its intellectual descendants are in textbooks everywhere: If you force people’s faces into a smile, they will feel happier, or find some amusing stimulus more amusing, relative to the members of a control group or a group of people whose faces are forced into a frown. There seems to be some deep connection between our bodies and our minds, and it doesn’t just go in the direction one would expect — rather, it’s a two-way street. This is solid, accepted, ensconced-in-the-pantheon psychological science.
Except: Maybe not. As Daniel Engber wrote in Slate last month, the most recent, rigorous attempt to replicate Strack’s finding delivered a decisively meh result. This, combined with issues Engber discovered in the original paper, suggests the smile effect, a foundational finding in the broader field known as embodied cognition, may not be real. If the smile effect is as shaky as it appears to be, there’s solid reason to think many other embodied cognition effects could be, too — like the head-tilting and posture studies mentioned by Carney and her colleagues.
From the point of view of understanding that original 2010 power-posing study, this is important. Researchers don’t p-hack in a vacuum — they (usually) don’t sit around smoking cigars contriving ways to gin up fake results and hoodwink the public. Rather, when they cut corners it’s partly because they are too confident in their theory. They don’t need to bother with due diligence, because the thing they are testing fits so neatly into past findings, and makes too much sense. Again, it’s very unlikely these sentiments are uttered aloud, but confirmation bias can worm its way into human systems in subtle ways. (Other incentives and influences also chip away at the integrity of science, of course — in her note Carney explicitly mentioned the team’s limited research budget as one limitation that led to certain types of corner-cutting.)
In the case of the Carney, Cuddy, and Yap study, we now know they p-hacked. And one possible reason for that could be their confidence in those past findings, in the idea that they were building upon a sound foundation. Power posing made sense because it was standing on the shoulders of exciting, established, sturdy(-seeming!) research.
This isn’t to excuse naughty research practices, of course. It’s to point out how vitally important it is for there to be a relatively high threshold before researchers accept a new set of ideas as “true,” given how complex and fuzzy a term that can be in science. When researchers accept a new idea on the basis of a smattering of early, not-particularly-rigorous studies, it can have pretty damaging ripple effects — it leads the discipline down false corridors, wastes research dollars, and in some cases squanders years of the careers of otherwise promising scholars.
That’s partly just how science works, of course, since science is by definition the process of sifting exciting hypotheses from failed ones, and that’s going to entail some missteps. But still: Premature excitement today can lead to a lot of wasted effort tomorrow. That may be part of the story with power posing.