We’re in an unfortunate place as a party when the GOP nominee’s down six points in the national four-way average and yet it’s Big News for his campaign manager to acknowledge that, yes, he’s trailing at the moment, as Conway does at the beginning of the clip below. It takes a lot of cocooning from some very influential people to make what she says “controversial,” and in this case the cocooning starts right at the top.
Drudge has also been a factor. His homepage as I write this is screaming about “SHOCK POLLS,” one which shows the race tied, one which has Clinton up a point, and one which has Trump up two points. Can you guess which polls those are? Right, the three Trumpiest polls of the cycle — IBD, the LA Times, and Rasmussen. A solid dozen other major national polls showing a sizable Clinton lead are being almost completely ignored in favor of the three that show a tight race. (Drudge does include a link to RCP’s “latest polls” page in a small font further down the page.) A low-information voter who gets his news mostly from the Drudge Report will be shocked on Election Day if, as seems likely, Clinton goes on to win by an Obama-esque margin. But I assume that’s the point — to flog the favorable polls, and only the favorable polls, in order to make Trump’s inevitable cries of “rigged!” seem more plausible to people who read Trump-friendly sources exclusively. If the LA Times and Rasmussen are the only polls you’re aware of, then obviously Clinton winning by five or six points is proof of cheating, right? Conway’s doing what she can here to reality-check Trump’s fans, but her spin-doctor duties require her not to dwell much on the state of the race. She’s going to have a rough couple of days two weeks from now, though, if it falls to her to go out there and tell them that the polls were right all along. And let’s face it, it probably will fall to her.
By the way, the “phony poll” Trump is referring to in his tweet is, I believe, the ABC/WaPo poll that dropped yesterday, which some Trump fans are scoffing at because it has many more Democrats in the sample than Republicans (36/27). Nine points is a big gap between the parties, no doubt. But let me show you something. Here’s HuffPo’s tracker of party ID dating all the way back to Election 2008. Notice anything?
Republican party ID is ever so slightly higher now than it was around this time in 2012 (a tenth of a point higher). It’s actually a bit lower than Republican ID circa February 2012. Democratic ID, however, is higher than it’s been since the summer of 2009. It’s nearly a full point higher than it was circa October 2012. How come?
Well, probably because … the polls really are accurate and we’re headed for a comfortable Democratic victory. It’s tempting to scrutinize the partisan gap in samples and conclude that it leads too far one way or another; I’ve done it myself, but I’m always mindful of the fact that partisan ID shifts during campaigns. It’s not D+4 unto eternity. It’ll turn bluer or redder in response to events. Saying that D+9 is outlandish implies that there’s zero chance that disgust with Trump and the “Access Hollywood” tape and the sexual-assault allegations have motivated some Democrats who weren’t planning to vote to turn out after all in the name of beating Trump. All of that might also have convinced some left-leaning independents to identify as Democrats, however temporarily, because they feel solidarity with Hillary as Trump’s foil and/or because they expect a Democratic victory and want to be on “the winning team.” That’s how partisan gaps widen. If anything, you would expect a bigger partisan gap in polling samples in the middle of an election that’s shaping up to be an easy win for one side or the other. (Note the 2008 gap in the graph above.) In fact, ABC explicitly noted in its story about the new poll that they’re picking up lower levels of enthusiasm among Republicans lately, a predictable result of the last few rough weeks Trump has had:
The previous ABC/Post poll found a sharp 12-point decline in enthusiasm for Trump among his supporters, almost exclusively among those who’d preferred a different GOP nominee. Intended participation now has followed: The share of registered Republicans who are likely to vote is down 7 points since mid-October.
They have a new story today following up on that. Fifty-six percent of Clinton voters now say they’re voting mainly for her rather than against Trump — a new high — versus 54 percent of Trump voters who say they’re voting mainly against Clinton rather than for Trump. “Affirmative support” shows that Democrats are increasingly energized. No wonder you’re seeing a bigger partisan gap in the sample.
All of this polling trutherism seems to be due to something in one of the leaked Podesta emails about “oversampling” certain groups. Supposedly, ABC “oversampled” Democrats in its poll yesterday (even though, as I just explained, there’s really no reason to think that). What Podesta probably meant by “oversampling” is best explained here by Pew. Typically, polls of the general population only include small numbers of blacks and Latinos because, after all, they represent relatively small portions of the overall population. That makes it hard to get a reliable sense of black and Latino opinion, though: If, say, a poll of the population includes only 100 Latinos, that will produce a gigantic margin of error in the numbers for that group. To reduce the MOE and get a more trustworthy read on what they really think, you need to oversample them — poll, say, 600 Latinos instead of 100. It’s not about rigging anything, it’s simply about getting more accurate measures of voter sentiment within small but key groups.