Being a social science professor, one becomes accustomed to seeing a variety of different ways of looking at the same problem, same data. These reflect the various methods in the social sciences of organizing data into a worldview. The problem with these data is that they are rarely, if ever, 100% in favor of one particular worldview. Two researchers with different worldviews can look at the same data and claim the data supports each of their views.
As more and more governmental policy comes to rely on social science research as a sort of apologia for the policy, the data gets pushed harder and harder as showing one of a number of potential explanations for the data as being correct. Since the data are rarely inconvtrovertible, using some data to guide or shape governmental policy is exceptionally dangerous, because the policy could entail exactly the wrong thing.
Social science is not hard science - and it's especially not physics. In hard science, from the point of view of a social scientist, there are true answers. The speed of light, for example, is a constant. Nothing can change that. Social science is the study of human action - there can be no universal laws here, because all human action shapes and is shaped by the environment in which it takes place.
What does all this have to do with agnosticism? Just this - in social sciences, we never know the true, or correct, answer. Often, there isn't one. This level of uncertainty should cause us to refrain from making public policy based on social science research, because we simply don't know for sure. I often hear, about some problem plaguing the world, that it is better to do something than nothing. But how do you know? How do you know?