The Impact of Managers

Scott Lemieux and David Brockington debate the question of whether managers “matter” in baseball. I don’t know much about baseball, but Brockington’s contention is very illogical:

These are superficial, anecdotal pieces of evidence; the sabermetric literature (that I am familiar with, I am now a couple years behind I’m afraid, although there is some interesting stuff here) has had a difficult time establishing that the field manager of a ball club has much measurable effect at all, and is negligible at best.

The link is to a research that indicates managers don’t have an impact on player performance. But insofar as some players perform better than others, and insofar as managers decide who plays and how much, I don’t see how sabermetrics could possible show that the field manager of a team has no impact on how many games the team wins. Say your team’s 8th-best offensive player is a slightly below-average defensive shortstop whereas your 12th-best offensive player is an above-average defensive shortstop. Who do you play? In what situation? Answering these kind of questions correctly seems incredibly important, and the importance of these issues is precisely why sabermetric research has been of so much interest. Or am I missing something?