Estimating risk preferences is tricky because controlling for confounding factors is difficult. Omitting or imperfectly controlling for these factors can attribute too much observable behaviour to risk aversion and bias estimated preferences. Agents often modify risky decisions in response to...