I referred to the Larry Bartels general election model yesterday, but I think it's worth revisiting it, because probably my biggest question about 2012 is whether Bartels is right or not.
At least, that's my reaction after seeing two things today. One is a Philip Klein post about the latest GDP numbers, in which he makes the case that Barack Obama is in very weak shape compared with other postwar presidents seeking re-election. It's okay...the biggest problem here is that Obama can do worse than Bill Clinton in 1996, and quite a bit worse than Ronald Reagan in 1984, and still win.
The other is an excellent Monkey Cage piece by John Sides with new survey numbers about whether Obama or George W. Bush is blamed for the condition of the economy. The key result, after you get past the partisans (who behave as one would expect), is that independents are far more likely to blame Bush. Even now, three plus years in.
The obvious question is whether people will vote to throw the bums out based on a still-weak economy, or if they'll transfer their blame from Bush to Romney and keep Obama in office. That's where the Bartels model comes in. Larry Bartels finds that GDP growth in the first year of a president's term has a strong negative association with re-election vote, and even second-year GDP growth has a mild negative relationship with re-election. If this is true -- and, again, small n and all that -- then that's consistent with the finding of voters blaming Bush for the current economy, and therefore rewarding Obama for even mild improvement.
So is it true? I have no idea. There's really no way to know whether the minor effects observed in past elections were real or just flukes, nor is there any way of knowing whether the mild effect observed in the past will produce a large effect when a president takes office during a massive recession. I guess that's why they run elections...to give political scientists more data to study. Right?