When does data settle arguments and when does it not?

It's often said that data wins arguments. Like many other things that are often said, it's mostly false. The history of ecology is littered with arguments that even copious data failed to settle. Using historical examples, I'll identify the rare, limited circumstances in which data is decisive, and the much more common circumstances when it's not. Particularly important is having a well-posed question or hypothesis. Empirical hypothesis testing is admirable—but too much empirical work in ecology tests pseudo-hypotheses with no theoretical basis, thereby starting arguments rather than settling them.