I mentioned in my previous post on this topic that Richard Miller was one of my foes on the issue of politics and objectivity. He discusses the issues in this book, among other places. Three dogmas of the social sciences which Miller sets out to slay are the ideal of value freedom (most relevant for our purposes), methodological individualism, and the Hempelian covering-law model of explanation.
To take them in reverse order, Miller rejects the Hempelian covering-law model of explanation in favor of a causal model. Hempel's model is, uncontroversially, quite oversimplified and does not work in many cases. Many of us still think it's a useful model, and that one of the best ways of proceeding is to describe this model, describe its problems, and try to understand specific cases in the philosophy of science by applying the model and keeping alert to the known ways it can break down (many introductory philosophy of science books endorse essentially this approach, either openly or implicitly in their practice). Miller's criticisms are actually a lot weaker than the standard set.
He complains that the covering-law model does not account for the fact that an explanation may be judged satisfactory even if there's no very general law that would account for it. In one of his examples, he notes that a historian may successfully explain a counter-revolutionary uprising in one region of 18th century France by appealing to a clerical monopoly on access to outside sources of information, even if such a clerical monopoly on outside information did not produce counter-revolutionary uprisings in 9th century Japan. Conversely, he notes that an explanation may be unsatisfactory even if there is a law that would account for it. Though being appointed by doddering old coots of heads of state is reliably connected with people becoming heads of government, that Hitler was appointed chancellor by Hindenburg would, according to Miller, be an inadequate explanation of his seizure of power if the Nazi's support among the German economic elite would have resulted in their eventually seizing power regardless of Hindenburg's actions.
To summarize this argument of Miller's, sometimes an explanation is acceptable even if it's not very good because it's the best we can do, and sometimes a good explanation isn't good enough because it's possible to do better. How these obvious points are supposed to show that we need Miller's causal account of explanation (which I won't bother to try to summarize, as it's never stated very clearly) rather than Hempel's model is, to put it mildly, unclear.