Okay, this NARAL/Langevin/Chaffee thing is getting out of hand. As Atrios said, NARAL's endorsement of Chaffee probably didn't drive Langevin out of the RI Senate race. And if it did, well, then Langevin's not much of a fighter anyways. But let's pretend NARAL's endorsement either way makes a shred of difference. Then the second point is this: they made an entirely rational choice from a strategic standpoint. Let's just run through the four scenarios NARAL's facing here:

1) Republicans keep the Senate in 2006 and Chaffee gets elected. Well, that's bad news. But notice, whenever the Republicans slap down some bit of legislation restricting abortion rights, Chaffee will be voting against it (remember, he votes pro-choice 100 percent of the time. 100 percent!).

2) Republicans keep the Senate in 2006 and Langevin gets elected. Worse news. Republicans are still in charge, but now whenever they slap down abortion restrictions, Langevin will likely vote for them, giving pro-life legislation one extra vote and making it more likely to pass. Clearly outcome #2 is worse for NARAL than #1. But then we have...

3) Democrats retake the Senate and Chaffee gets elected. Hooray! Now whenever Democrats want to push through some legislation expanding abortion rights, Chaffee votes for it, making it more likely to pass. Which is still better, from NARAL's perspective, than...

4) Democrats retake the Senate and Langevin gets elected. Now, that legislation expanding abortion rights suddenly becomes harder to pass—at the very least, you'll have to do Langevin some favor elsewhere to get him to vote for it. But odds are, he won't vote for it!

So NARAL's preferences here are, I think, ranked: 3, 4, 1, 2. Endorsing Chaffee, then, is a pretty optimal choice—it makes either 3 or 1 more likely, rather than 4 or 2. The wild card here, of course, is that control of the Senate may actually hinge on who wins, Chaffee or Langevin—in other words, the race itself may be between outcome #1 and #4. I don't really know what the probabilities are here, or how to model this, but presumably NARAL doesn't think this scenario is very likely (in other words, the probability is low that Senate control either way will depend on the RI race). I'm sure there's a rigorous way to calculate out what NARAL's optimal strategy is here, but I'm not smart enough to do that, so I'll just eyeball and say, yes, it makes sense to endorse the pro-choice Republican over the pro-life Democrat.