How a PAC sets the narrative

House Majority PAC, a Democratic political action committee, released 12 polls today that they say shows Republicans are in serious trouble. Sure enough, the sheep in the media printed exactly that.

This poll is actually very good news for Republicans.

Wait a sec, you're saying. All these Republicans have higher don't elect than elect. That's 12 seats Republicans will lose. There are three problems with that narrative.

1. This is a poll by a Democratic pollster for a Democratic PAC. Nate Silver has talked about partisan polls being off by as much as 6 points.

2. They polled registered voters. While they chose not to release the in-tabs, PPP has repeatedly had polls all year with the Democratic share of the electorate higher than it was in 2008, a great Democratic year, and often larger D/R spreads than in that year. PPP claims they don't weight by political party. It just happens that way. Yeah, right.

3. They ask the question, "would you like to re-elect Rep. XYZ or prefer someone else?" That question isn't on the ballot. People often express their unhappiness with their congressman when the alternative is the perfect congressman they've made up. Congressmen will always do better, sometimes a lot better, when faced with an actual opponent.

Since they didn't release in-tabs we don't know how the "someone else" broke down by party. These days a good share of the Republican electorate is unhappy with their congressman. They may say they wouldn't re-elect, as long as the alternative was another Republican. When faced with their Republican congressman against a Democrat, they'll vote for the Republican they're unhappy with. Since PPP conspicuously didn't release the in-tabs we don't know how it breaks down.

I'm going to add 7 points to the congressmen and drop the "don't elect" 7 points for a Democratic challenger. I think that's conservative considering the problems above, all of which could add 4-6 points to the Republican's total. If we add the 7 we end up with:

AR-02 R51-42
CA-07 R50-47
CA-10 R45-42
CA-26 R49-41
CA-36 R50-48
CA-52 R49-44
IL-10 R49-43
IL-11 R48-45
IL-13 D46-40
IL-17 R46-42
WI-07 R50-44

Ok, we did all that and it doesn't look so great for Republicans. In half of them the Republican leads by 4 points or less. In 9 of them it's 6 points or less. These Republicans could lose. Yes, they could, but look closely at the districts. There are four of them, CA-26, IL-10, IL-11, and IL-17, that the experts have already declared Democrats are sure to pick up. So not only aren't they sure Democratic pick-ups, but they might actually be leaning Republican. Three other districts, AR-2, CA-52, and WI-7, have big enough leads that they should be definite favorites.

It's not all good news. CA-10 and CA-36 are far closer than expected. There are probably 4 districts that Democrats are more likely to flip than these 12. There are probably 9 Democratic districts that are likely to flip and there will be more once all the redistricting is done.

The Democrats may need as many as 20 districts beyond these 12 to get enough seats to flip the House. If they can't count very many of these, they certainly won't get anywhere near 20 others.

So, yeah, it's a good group of polls.