Heterogeneous motivation and cognitive ability in the lab

I test whether economics experiments that estimate the relationships between cognitive ability and economic behavior are biased because they fail to account for differences in motivation among subjects when they measure cognitive ability without compensating them for performance. I find that monetary incentives do not significantly improve average performance on cognitive reflection test (CRT) questions, but subjects classified as low ability based on unpaid scores tend to improve their performance when they are paid for performance relative to high ability subjects. This heterogeneous response to monetary incentives appears to bias estimates of the relationships between cognitive ability and trust and risk aversion downward, but does not have the same systematic effect for strategic reasoning.