In this paper we propose a new method in experimental economics, designed to evaluate the effectiveness of public policy incentives aimed at altering consumer behaviors. We apply this method to wide-ranging policies on food prices, which use subsidies to increase the consumption of healthy products and taxes to reduce that of unhealthy ones. Our protocol allows for observation of an individual’s daily food consumption before and after the policy. We examine two separate policies: the one subsidizes fruit and vegetables, while the other one combines taxes and subsidies. We measure their nutritional and economic impacts on the choices of low-income French consumers, compared to a reference group. Both policies have a positive effect on the nutritional quality of food choices of the two groups but initial gaps widen, especially with the subsidies. In the low-income group this can be explained by an initially unfavorable pattern and by weaker price elasticities. The redistributive effects are therefore doubly regressive. Moreover, the individual price elasticities, that the experimental approach enables us to measure, show widely diverse behaviors. They are counter-effective for close to 40% of our sample of poor women.