I happened to be in the midst of reading Kuhn vs. Popper when I ran across this, by Nicholas Lemann, on the sometimes clumsy dance between narrative and analysis in journalism:
Forming a hypothesis. It’s healthier to admit to yourself that you have one than to go into a story with the idea that you have no presuppositions at all – that would be impossible. You should state a working hypothesis (to yourself, anyway), and then ask yourself what would prove the hypothesis false and what would be an alternate hypothesis to explain whatever it is you are investigating. As you report, you should try not just to prove but also to disprove your working hypothesis, and you should engage in a continuing process of revision of the hypothesis, if necessary. If you don’t design your reporting in such a way that if your hypothesis is flawed, you will find that out before you finish the story, then you are leaving yourself open to getting the story seriously wrong.