More On That Q Poll Analysis

I was just saying yesterday, in response to this chart and analyis from Nate Silver, that "Democratic Members of Congress don't care about the party-wide approval numbers nationwide, but about their own approval numbers within their districts -- if anything, the overall approval numbers they probably care more about are those of Barack Obama." 

But that point is better made in this comment by political scientist Alan Abramowitz, as introduced by John Sides:

Finally, and most importantly, and please please shout this from the rooftops:

This brings up the most important point about evaluations of Congress. They have very little influence on how Americans vote in congressional elections. When it comes to choosing candidates for Congress, it is opinions of the president’s performance that matter.

I'll add: opinions of the president's performance and economic performance.  Those are the big, national, picture things that matter.

As long as I'm following up, I can be a bit more specific about the candidate influences in Congressional elections, too.  I said "approval rating."  What Gary Jacobson finds is that there are three variables about the candidates that move votes.  The first one is pure incumbency, which has a small effect (positive): all else equal, people are slightly more likely to vote for the incumbent.  The second is familiarity; all else equal, people are somewhat more likely to vote for a candidate they've heard of over one they haven't heard of.  The key variable has to do with things voters like and don't like about the incumbent, compared to things they like and don't like about the challenger.  Let me explain that last one...survey researchers have asked voters whether they know anything good or bad about the candidates.  At the extreme, if people can remember something they like and nothing they don't like about the incumbent, while they remember only something they don't like about the challenger, they are overwhelmingly more likely to vote for the incumbent -- in fact, the effect not only overpowers incumbency per se and familiarity, but it even overpowers party identification.  Of course, many voters don't know anything about their Member of the House, or even their Senators, but most of the information that's out there for most Members is positive, which helps explain very high reelection rates even when people generally express anti-incumbency attitudes. (The rest is explained by districts, since most Members of the House and most Senators are originally elected in districts (states) that favor their party, and party is the other big mover).

All told, as I said yesterday, I wouldn't put much weight in Nate's chart -- although I do think his political advice is exactly correct, the chart notwithstanding.