Unpacking the link between hormonal fluctuations and risk-taking: A systematic review and meta-analysis

Previous studies have suggested that hormonal fluctuations, specifically in testosterone, estradiol, and cortisol, may impact reward-related brain functioning and risk-taking behaviors. However, findings in this area have been inconsistent and sometimes contradictory. The current study aimed to conduct a meta-analysis to investigate the effects of both endogenous and exogenous testosterone, estradiol, and cortisol on risk-taking behaviors, as well as identify potential moderators of these effects. This meta-analysis systematically reviewed studies published up to February 20, 2025, encompassing both correlational and experimental designs. After screening 2544 records, 98 studies met inclusion criteria, yielding 162 effect sizes involving 8676 participants for testosterone, 55 effect sizes from 2510 participants for estradiol, and 66 effect sizes from 3933 participants for cortisol. Using the random-effects Bayesian meta-analytic models, our results showed that both testosterone and estradiol had a significant, albeit modest, effect on increasing risk-taking behaviors (testosterone: Hedge’s g = 0.22; 95 % CrI [0.14, 0.30]; estradiol: Hedge’s g = 0.20; 95 % CrI [0.03, 0.37]). However, cortisol was not associated with changes in risk-taking (Hedge’s g = −0.04; 95 % CrI [−0.17, 0.09]). Further analysis indicated that the effects of testosterone were moderated by the study design (experimental vs. correlational), the behavior type (sensation seeking vs. risk-taking vs. impulsivity), the measurement type of risky behavior (self-report vs. behavioral) and the measurement type of hormone (saliva vs. serum), but these moderators had no significant impact on the estradiol effect. Despite the potential for publication bias, no evidence of selective reporting (e.g. p-hacking) was found in the p-curve analysis. In summary, testosterone and estradiol may influence risk-taking behaviors, although further randomized controlled trials (RCTs) with larger sample sizes are necessary to confirm these findings.