Boys take more risks than girls. This is one of the more robust findings in developmental psychology, and it is repeated so frequently that it has acquired the status of a fixed fact.
It is also considerably more contingent than the summary suggests, and the contingencies are where anything useful lives.
What is measured
Risk research uses several methods, and they do not all agree.
Behavioural tasks in laboratories — inflating a virtual balloon for increasing reward until it bursts, choosing between certain and gambled outcomes — typically find modest differences, with boys taking somewhat more risk. Effect sizes are usually small to moderate.
Self-report questionnaires find larger differences. Observed real-world behaviour, particularly in physical domains, finds larger differences still.
The pattern of that divergence is informative. The gap is smallest where the environment is most controlled and the behaviour is least socially visible, and largest where social interpretation is doing the most work. That is not what you would expect from a purely dispositional difference.
The supervision asymmetry
Observational studies of playgrounds have repeatedly found that adults intervene at different thresholds depending on the child's gender. Girls attempting the same climbing structure are more likely to be cautioned, assisted or stopped; boys are more likely to be allowed to continue, and more likely to receive instruction on how to do it rather than advice not to.
This finding has been reported in several countries and across several decades of research. Effect sizes vary and some studies find no difference, but the direction is consistent where a difference appears.
Consider what that produces over ten years. Two children with identical dispositions, one of whom has been stopped hundreds of times and one of whom has been coached hundreds of times, will not have identical expectations about attempting the next physical challenge. The disposition need not differ at all for the behaviour to.
Punishment and interpretation
The second asymmetry is what happens after a risk goes wrong.
Studies of how adults describe children's misbehaviour find that the same act tends to be attributed differently — high spirits and boundary-testing in one case, poor judgement and inappropriateness in the other. The words matter because they determine the response, and the response determines what the child learns about the cost of trying.
There is a related finding in the workplace literature that is worth putting next to this: when a decision fails, the penalty applied to women in leadership positions has been found in several studies to be larger than that applied to men for comparable failures. If the downside of a failed bet is greater, then taking fewer bets is not timidity. It is arithmetic.
This reframing has some empirical support. Studies that equalise the consequences of failure within an experiment tend to find risk-taking differences narrow considerably.
Domain matters more than the summary suggests
"Risk" covers behaviours with almost nothing in common. Physical risk, financial risk, social risk, and health risk are separate constructs that correlate only weakly with each other.
When they are separated, the picture changes. The male-higher pattern is strongest in physical and, in some studies, financial domains. In social risk — speaking against a group, ending a relationship, raising an uncomfortable subject — several studies find no gap or a reversal.
A person willing to have a difficult conversation but unwilling to jump off a wall is not less willing to take risks. She is willing to take different ones. Aggregating those into a single trait score obscures more than it reveals.
The consequence that matters
Risk-taking is not intrinsically good. Most of what is romanticised as boldness is simply poor estimation, and the domains where boys take more risk are also the domains where boys are more often injured or bankrupt.
The reason the pattern matters is narrower: several routes to opportunity are gated by willingness to make a claim that might fail. Applying for something you are not obviously qualified for. Pitching an idea in a meeting. Putting a project in front of someone who might dismiss it. Starting something without permission.
Each of these is a small social risk with an asymmetric payoff, and each is more available to someone whose experience says that a failed attempt is survivable.
What actually shifts it
Experience of survivable failure, again. This is the same lever as in the self-efficacy literature, which is reassuring rather than repetitive — it suggests the underlying mechanism is the same.
The practical version is unglamorous: environments where attempting and not succeeding is routine and low-cost. Competitive sport at a level where losing is frequent. Making things that might not work. Any structure with a high attempt rate and a low penalty per attempt.
The adults' half of it is quieter and harder. It involves noticing the threshold at which you intervene, and asking whether it is the same one you would apply to a different child doing the same thing. Most people find, on honest inspection, that it is not.