The use of laboratory experiments to study issues in agricultural policy has grown in prominence within the fields of agricultural and environmental economics. Such experiments are often conducted with university students and framed in an abstract manner. This raises questions about whether the findings of these experiments provide reliable insights on the behaviour of actual agents in real settings. We contribute to this methodological debate by analysing the impacts of sample population and framing on behaviour in the experiment and on two policy effects: the direction and the magnitude of the policy impact. We also examine the channels through which differences in results may occur. For this, we test if behaviour is correlated with a set of covariates collected from our samples, including socio-demographics, social and risk preferences. Our main finding is that the type of subject significantly affects the magnitude of the policy impact. The two populations differ substantially in the representation of key characteristics and preferences, which in consequence affects behaviour in the experiment. We find no significant impact of framing on behaviour.