A meta-analysis on age differences in risky decision making: adolescents versus children and adults

Abstract

Despite evident heightened adolescent risk-taking in real-life situations, not all experimental studies demonstrate that adolescents take more risks than children and adults on risky decision-making tasks. In the current 4 independent meta-analyses, neurodevelopmental imbalance models and fuzzy trace theory were used as conceptual frameworks to examine whether adolescents engage in more risk-taking than children and adults and whether early adolescents take more risks than children and mid-late adolescents on behavioral risk-taking tasks. Studies with at least 1 of the aforementioned age comparisons met the inclusion criteria. Consistent with imbalance models and fuzzy trace theory, results from a random-effects model showed that adolescents take more risks (g = .37) than adults, and early adolescents take more risks (g = .15) than mid-late adolescents. However, inconsistent with both perspectives, adolescents and children take equal levels of risk (g = -.00), and early adolescents and children also take equal levels of risk (g = .04). Meta-regression analyses revealed that, consistent with imbalance models, (a) adolescents take more risks than adults on hot tasks with immediate outcome feedback on rewards and losses; however, contrary to imbalance models but consistent with fuzzy trace theory, (b) adolescents take fewer risks than children on tasks with a sure/safe option. Shortcomings related to studies using behavioral risk-taking tasks are discussed. We suggest a hybrid developmental neuroecological model of risk-taking that includes a risk opportunity component to explain why adolescents take more risks than children in the real world but equal levels of risks as children in the laboratory.