Talent B.S.

The Power Poser

Some psychologists have made waves in leadership circles for saying that striking a specific pose can make you feel more powerful. The only problem? Not a lick of the research checks out.

By Marcus Crede

Lean back, kick your feet up onto your desk, and put your hands behind your head. Do you feel any different?

No? Okay, strike this pose: Stand up, feet shoulder-width apart, and put your hands on your hips, a la Wonder Woman. That should do the trick.

Still not feeling anything? Hey, maybe it just takes a while to kick in. Because according to research from Harvard University, adopting these kinds of expansive physical poses—a.k.a. “power poses”—can increase your testosterone, decrease your levels of the stress hormone cortisol, boost your willingness to take risks, and most importantly, make you feel more powerful and in charge.

It’s this last posing perk in particular that has made lead researcher Amy Cuddy a hit in leadership circles. She has shared her findings in speeches, a bestselling book, the pages of news outlets like the Wall Street Journal, and the second most-viewed TED Talk of all time. But after taking a closer look at the claims, it turns out that they can’t stand up to scrutiny.

The Problems with Power Posing

Let’s start with the original study, which Cuddy conducted with Dana Carney and Andy Yap and published in the journal Psychological Science in 2010. In the study, test subjects provided their saliva samples to establish baseline levels of cortisol and testosterone. 

They were then randomly assigned to hold either an expansive or contractive pose for 2 minutes. Immediately after, the study team asked the participants about their subjective feelings of being “powerful” and “in charge” in a simulated gambling situation, and then measured their testosterone and cortisol levels a second time.

Though this methodology sounds simple and straightforward, there are several holes we need to address:

1. The primary analysis described in the paper was based on only 39 participants—a sample size so small it not only becomes very difficult to detect any effects that may exist, but also implies that any “statistically significant” effects that are reported are likely to be vast overestimates of the real effects.

2. The study lacked a control group. This means it’s impossible to know if any differences between the expansive and contractive conditions reflected a positive effect of an expansive pose, a negative effect of a contractive pose, or some combination of the two.

3. Though the authors reported that the effect of the manipulation on the decision to take a risk was statistically significant, a re-analysis of the data shows that this in fact wasn’t correct.

4. The authors had incredible latitude in exactly how to conduct their statistical analysis, and our forthcoming paper in the journal Social Psychological and Personality Science suggests the inferences drawn by Carney, Cuddy, and Yap only find statistical support for one of the very many possible ways that the data could have been analyzed. That is, the reported effects are not robust to data analytic strategies.

5. It’s relatively well known that gambling—and even the opportunity to gamble—can raise testosterone and cortisol levels. The gambling tasks therefore represent a confound because it cannot be determined if any changes in these two hormones are due to the effect of the pose being adopted, or the fact that participants were given the opportunity to gamble.

These problems with the power pose paper had been widely discussed by various research methodologists and statisticians (such as the Columbia University statistician Andrew Gelman), but they only came to widespread public attention when Dana Carney, the first author of the study and an associate professor of management at Berkeley University, posted a document on her university web page outlining her position on power poses.

Carney actually said, “I do not believe that ‘power pose’ effects are real,” and discouraged others from even studying power posing. She also described numerous flaws with the design and execution of the original study, including her admission that the authors relied on “p-hacking”—a term used to describe the mining of data to find patterns that can be described as statistically significant. It’s well known that p-hacking typically results in researchers reporting findings that are spurious and cannot be replicated.

The second type of critique revolves around the inability to replicate the original results reported by Carney, Cuddy, and Yap, even when some of the problematic features of the original study—such as a very small sample size—are avoided. For example, a large-scale replication conducted by Eva Ranehill and her coauthors relied on a sample that was more than five times as large, but was still unable to find any effect of a power pose on hormones or risk taking.

Admittedly, some researchers report that they were able to replicate some of the effects, and Carney, Cuddy, and Yap had listed 33 such studies in a 2015 paper. However, Joseph Simmons and Uri Simonsohn, two noted methodological experts, used a technique known as p-curve analysis to examine these 33 studies. Their conclusion? The pattern of results in these studies is what one would expect to see if the average effect of power posing was zero.

In other words, the reporting of significant results in these studies is likely to be the result of p-hacking on the part of the authors or the tendency of journals to strongly prefer publishing studies that report significant effects, while studies that fail to find effects are never published.

The Power Trip

Recently, Cuddy and her crew have moved away from focusing on hormones and risk taking, and have instead emphasized the increase in subjective feelings of power. 

Problem is, asking participants to adopt an expansive pose or contractive pose and then asking them whether they felt powerful could create a “demand characteristic.” That is, participants may figure out the purpose of the experiment and respond in a way that satisfies the experimenter. 

In addition, saying that power posing increases subjective feelings of power particularly in women is problematic. In an article written for Harvard Business Review, Cuddy states that “standing in a bathroom stall like Wonder Woman before a stressful meeting … has the potential to substantially improve women’s ability to lean in … to take risks, face fears, and barriers.”

Facebook COO Sheryl Sandberg also reportedly invited Cuddy to develop a program for her Lean In organization, which is devoted to helping women unlock and achieve their ambitions.

Unfortunately, the data described in the original power pose article shows that power posing may have a statistically significant effect on subjective feelings of power, but only for men. Indeed, the authors of that article appear to have been fully aware of the fact that the effect does not exist for women because they explicitly wrote about it in an earlier draft of the paper that was supplied by Carney.

Further, a re-analysis of the data collected by Ranehill and her colleagues also shows that this effect doesn’t work for women in their much larger replication sample. It seems a little strange that power posing has been so explicitly promoted as some kind of magical life hack for women trying to improve their odds of overcoming barriers and obstacles.

Finally, it should also be noted that even the most optimistic reading of the data doesn’t support the implication that power posing is likely to benefit most people. In the original Psychological Science study, levels of the stress hormone cortisol actually increased for 62 percent of those participants who struck the power pose. Meanwhile, testosterone levels decreased for 48 percent of those participants who adopted it.

So should you strike a power pose? There appears to be no evidence that the strategy has any direct benefit—certainly not of the dramatic type—but it’s also unlikely to cause any harm. 

Still, there’s no substitute for talent, skill, and work ethic. Do it to compensate for underperformance in those areas, and you’re just being a poser.

Marcus Crede, Ph.D., is an assistant professor of psychology at Iowa State University. His areas of research include individual differences, predictors of performance in organizational and educational settings, research methodology, leadership, and decision-making.