If there’s one message we’re getting from the legacy media, it’s that Romney has blown this election. It’s all over, Obama has won, and we really all ought to stay home and accept that government will be running our lives from now on.
In the past, I’ve resisted the temptation to deny the polls are wrong. Basically, that strikes me as loser talk, and I don’t want to indulge in it. In a rout like 1996 or 2008, I’d rather just accept reality. But this election is different. If you look at most of the polls, they simply don’t reflect reality.
The thing is, pollsters don’t simply ask people who they will vote for and report the results. After polling, they reweight the results to some desired balance of Democrats, Republicans, and independents in the sample. That weight is not determined by polling; they simply make it up. And that weight is the single greatest factor in the result of the poll. So despite all the talk of scientific polls, they aren’t.
If they guess the turnout correctly, weighting makes the polls more accurate, but the weights that most polls are using are simply insane. In 2008, Democrats outnumbered Republicans in the electorate by 7%. Anyone who thinks that the electorate is more Democratic now than in 2008 is just crazy, but most polls are overweighting Democrats by more than 7%, often even in double digits. Basically, they’re lying.
In fact, according to Rasmussen’s poll of party identification, Republicans actually outnumber Democrats by 4.3% now (37.6% Republican, 33.3% Democrats, 29.2% independent). That’s roughly a ten-point difference from the targets that most polls are using, which corresponds to roughly a ten-point swing in the polls. The website Unskewed Polls takes the major polls and reweights them with Rasmussen’s party numbers and finds that every one of them puts Romney ahead, by an average of 7.8 points.
Now I would caution against taking too much comfort in those numbers. I doubt that things are going that well. Rasmussen was the most accurate poll in 2008, and they have the race dead even. (Today they have Obama up by one.) We’re not where we want to be, but I think our chances are better than even.
The other reason I discount the polls that show Obama taking a big lead is they are completely at odds with the tracking polls. The great thing about tracking polls is they use the same methodology all the time, so even if you don’t believe their bottom-line result, they are effective at tracking movement. Here’s the Rasmussen tracking poll:
Except for a fleeting convention bump for Obama (nobody tells tall tales quite like Bill Clinton), the state of the race hasn’t moved much. Obama may have improved his standing by about 2 points.
And here’s the Gallup tracking poll for the last month, which has supposedly been a disaster for Romney:
All of Romney’s imaginary gaffes of the last month have amounted to essentially nothing.
So why are the media lying about the state of the race? Of course they want Obama to win, that’s a given, but how do they expect skewed polls to help make that happen? One theory is that they want to depress Romney voters into staying home. I’m skeptical of that theory. Here’s the problem: The polls today are gimmes; they can say whatever they want because there’s no benchmark to compare them against. The important poll for their credibility is the final poll before election day. Between now and then they have to de-skew their polls, and when they do, it’s going to look like a big shift toward Romney. That will help his turnout, not hurt it.
I think this is all about fundraising. No one wants to contribute to a losing cause, but everyone loves a winner. I think they are trying to depress Republican donors and — especially — to encourage Democratic ones. (Romney and his allies have much more money on hand than Obama.)
And sure enough, last night I accidentally watched about ten seconds of 60 minutes and what was Steve Kroft asking Romney about? How Romney can convince his donors that he hasn’t already lost the race.
(Via PJ Tatler.)
UPDATE: I’ve been reading elsewhere that, contrary to some accounts, most polls do not weight for targets for party identification. But then I don’t understand how they come up with such outlandish numbers. Even if we stipulated that Democrats somehow will increase their turnout from 2008 — despite Democrats’ lessened enthusiasm and despite the observed shift in party identification from Democratic to Republican — the number ought not be all over the map. If the skewed populations are not deliberate, then these polls have some fundamental methodological problem.