This column by Don Boudreaux is primarily about the minimum wage. I find this excerpt to be highly relevant:
Empirical research in economics is notoriously difficult because wages, prices, unemployment rates, product qualities, and all other data of the social sciences are, as Friedrich Hayek said, "complex phenomena." Having so very much constantly going on in the real world, having no laboratory in which reliably to isolate more than a handful of these phenomena at any one time, and unable to read directly the minds of the many persons whose perceptions and choices combine to generate social outcomes, empirical researchers can easily overlook or misread important variables.
The data problems can be worked on and the theory can be grappled with as well. Ultimately, this problem points to the appropriateness of less ambitious objectives in the realm of social science.
This situation distinguishes the social sciences from the physical sciences in two notable ways. First, a higher proportion of empirical research in the social sciences is subject to legitimate -- oftentimes irresolvable -- dispute. Second, as a consequence, in the social sciences theoretical considerations inevitably play a larger role in navigating around these disputes and in forming judgments about desirable public policies.