Policies are supposed to be (and inevitably are) designed based on a balance between empirical analysis of their likely effects and a more holistic analysis of what people want. The holistic factors determine what kinds of empirical evidence are relevant and how much of it is required. If someone tells you they've designed a policy based purely on "science" or "data", they've baked in some implicit value judgments, and you've gotta understand what they are to properly evaluate the policy.