I investigate the relationship between attitudes towards ambiguity and the ability to reduce compound risks. The evidence from an experiment on adolescents shows that patterns identified in the previous literature are susceptible to experimental implementation and the characteristics of the subjects.
Cognitive skills and the way lotteries are presented affect reduction of compound risks differently to ambiguity neutrality. My results suggest that theoretical studies which model ambiguity preferences by relaxing the assumption of compound risk reduction should be viewed with caution, and I add to the evidence against the use of compound lotteries to represent ambiguous environments in experiments.