The moral for economists is that theory is a useful thing, but the real issue is whether the theory is in agreement with reality. If it isn't, then it's the theory that needs to be revised.I call bullshit. "Reality" is shaped by theory, as much as theory is shaped by reality. That is, there is no independently-described set of facts which theories can be compared to. Even if there is an objective (in the sense of mind-independent) reality, it does not follow that there is one discrete and finite set of descriptions of this reality. The way reality is described is a theoretical enterprise, thus the comparison of theory to reality is itself theory-driven. From this, it follows that, as I believe Quine once defended, when confronted with a conflict between a theory and an observation, one always has the option to discard the observation.
Now, of course, if I'm right that our understanding of facts -- our conceptual framework, if you like -- is shaped by theory, then what actually happens when we "discard the observation" is that we discard (or revise) one theory rather than another. That is, the theory which leads us to make the observation may have to go instead of the theory which the observation is in conflict with. Given that, it is also possible that the theory which the observation conflicts with may prove to be the weaker contender. The lesson, though, is that Coffin is wrong to think that the comparison of theory to reality straightforwardly favours reality: because reality is shot through with theory, what really happens is we favour the theory for which we have the strongest reasons, and discard (or revise) the theory for which our reasons are weaker.
It's funny how often these kinds of problems come down to reasons.