In this paper we assess the importance of sample type in the estimation of risk preferences. We elicit and compare risk preferences from student subjects and subjects drawn from the general population, using the multiple price list method devised by Holt and Laury in their paper Risk Aversion and Incentive Effects (2002). We find evidence suggesting that under rank dependent utility, students exhibit approximately risk neutral preferences while subjects drawn from the general population exhibit risk loving preferences. However, when we assume an incorrect characterization of risk preferences, in particular we adopt the framework of expected utility theory, our estimation results lead to erroneous inferences. In this case, students are on average risk averse, while subjects drawn from the general population exhibit risk neutrality. Our results have implications for economic policy making under uncertainty.