Of course voting on the basis of economic growth in the two or three quarters leading up to an election plus military casualties is irrational. That's Kazin's point. The point that it can be predicted doesn't make it rational. Short-term economic growth and military casualties are related to good governance, but only very, very vaguely.Chait is correct that there's only a loose relationship between what politicians do and short-term results. But that doesn't make it irrational to reward good results and punish bad ones! Voters who do that provide extremely strong incentives for politicians to pursue good policy (that is, policy that tends to result in solid economic growth and peace, and avoids disasters such as Katrina).
Would Sides vote the way independents vote? Would he even want to be friends with anybody who did? I suspect the answer to both is no.
Strong partisans are also rational, even though they will support their party regardless of performance. How can that be? Because strong partisans (usually) want something that independents don't want: specific policies, and they're willing to basically swap their vote for party loyalty on those policies.
I'm not saying all voting is rational, but I certainly don't think that providing politicians with incentives to produce peace and prosperity would be high on my list of suspected irrational behaviors.
Update: Don't miss John's response.