Two quick thoughts on this post by Mark Kleiman asking "Obama outperforms HRC in every match-up. His net favorable/unfavorable is +24; hers is +3. Any questions?" 1) Yes. Can we stop linking to individual horserace polls? They demonstrate extreme volatility. Better to go with the Real Clear Politics aggregation page of horserace polls, which lets you get a bit more stable view of what's going on. Here, for instance, we find that Obama is, on average, outperforming Hillary against Rudy Giuliani, posting up a 4.5% advantage to her 3.5% advantage. On the other hand, she's winning against McCain, with a 1.7% advantage to Obama's 1% advantage. And if you restrict to only like polls, Edwards is doing better than either (since Edwards gets polled less often, his averages include some older surveys than Obama and Clintons). But none of the candidates are routinely and wildly outperforming each other. Certain polls, for reasons I don't yet understand, show one or another candidate way up over their competitors, but those findings are not, as of yet, proving reproducible. We'll see, for instance, if Edwards posts up numbers like those he exhibited in the recent CNN poll, which showed him trashing the GOP. A Rasmussen poll from virtually the same few days didn't back up the CNN results. 2) As noted above, the differences exhibited in these horserace polls are not that great. Obama, with his 21 points favorability advantage over Clinton, is beating Rudy Giuliani by...1% more. He's beating John McCain by .7% less. Edwards, with his slight favorability disadvantage to Obama, is beating McCain by 4% more. Point being: It's not clear how well favorability ratings translate into votes, nor how stable they'll prove over the course of the campaign. So don't, at this point, take them too seriously.
