For years, scholars have argued that testosterone influences how people make decisions about money and competition. But new research co-authored by Wharton marketing professor Gideon Nave found no evidence that the hormone affects economic behavior.
The authors say their results suggest earlier studies linking the hormone to decision-making should be treated with caution. They add that future research needs to be more rigorous and transparent, with studies that include more people and explain their methods and results more clearly.
“Most behaviors are likely the result of many psychological and biological factors, each of them having a small effect on the outcome,” Nave said.
The paper, which was recently published in the Proceedings of the National Academy of Sciences, has produced robust findings. It tested the impact of testosterone on nine types of economic behavior, such as risk-taking, fairness, generosity, and competitiveness — and could not find an effect on any of them.
“One important strength of this study is that it’s very large,” Nave said. “The more people you have in the study, the more confident you can be in the results — just like when you use a large telescope, the picture becomes sharper and the faint signals you’re looking for begin to stand out from the noise.”
“Most behaviors are likely the result of many psychological and biological factors, each of them having a small effect on the outcome.”— Gideon Nave
Challenging Previous Testosterone Studies
Nave carried out the research with a large team from several academic institutions around the world, including Stockholm University in Sweden and the Nipissing University in Ontario, Canada. The authors wanted to settle a long-running debate about testosterone, namely that giving it to men impacts how they make decisions, such as whether to take risks, compete, or act fairly.
Many earlier papers have produced conclusions that appear to contradict each other. For example, some found the hormone can make men more generous, but others say it makes people less trusting. A few saw no clear effect at all.
To settle the debate, Nave and his colleagues ran one of the largest and most rigorous experiments on the topic to date.
They recruited 1,000 men — 10 to 20 times more than most earlier hormone studies — then randomly assigned half to receive a dose of testosterone, and the other half to receive a placebo. Neither the participants nor the researchers knew who got which in a “double-blind” study.
After the treatment, participants took part in a series of well-established economic tasks to measure behavioral tendencies such as risk tolerance, trust, fairness, altruism, and willingness to compete.
These included standard economic games — such as the “ultimatum” and “trust” games — in which participants decided how to split money, or whether to trust another person with it.
But the study found no measurable effect of testosterone on any of those outcomes, with economic behavior virtually the same between those men given the hormone and those taking the placebo.
“In the future, looking at people who take testosterone clinically might be somewhat helpful, but it won’t be a clean test.”— Gideon Nave
A Wider Pattern in Behavioral Science
The authors acknowledge the study’s limitations.
“We looked at one specific paradigm: giving men one dose of testosterone,” said Nave. “This is not the only channel through which the hormone could influence behavior. However, because long-term testosterone experiments aren’t safe or practical, the only causal evidence we had so far on this question came from the aforementioned studies that gave a single dose.”
He added: “In the future, looking at people who take testosterone clinically might be somewhat helpful, but it won’t be a clean test. These groups decide to take the treatment themselves, and without randomization or a real control group, we can’t tell how much of their behavior is due to testosterone versus other factors.”
Another limiting factor, the authors note, is that the study involved only men, so they cannot be certain the same results would apply to women. But they say there is scant evidence that such effects hold in women.
More broadly, they also say the findings chime with a wider pattern in behavioral science whereby many results that seem solid at the time of publication — especially those relying on small samples — fail to replicate in subsequent research.
The paper, then, adds to growing evidence that many bold claims about human behavior need stronger proof.
