Hi, it's Stupid: Election Edition

Haruspex examining the entrails of a sacrificed bull, after a drawing by the painter (Pierre?) Vauthier. Via.

Hi, it's Stupid to say the Democrats could win the Senate contest. Ask anybody!

But somebody's got to do it. I'm not saying they will win, or are likely to (I'm no good at handicapping anyway, but if I were to place a bet it would be for a 50-50 split), but in fact we know it's not impossible, Nate Silver is saying a 15% chance, and what I want to say here is that in the unlikely event that it does happen, I'm going to know why.

What Silver is saying now in prose, as opposed to the charts, is that if it does happen it will mean there's a "systematic error in the polling" he's working with:

By a systematic polling error, I mean one that occurs in a correlated way across every race, or in certain groups of races — not merely errors that happen on a one-off basis. Our models account for the possibility of several different types of systematic errors, but in this article, I’m going to focus on the simplest type of systematic error, which is a uniform swing that applies to every race.

But the error I think is most plausible is in the turnout model, something I believe the smartest professionals, of whom Silver is no doubt the best, consistently undervalue—in his house account of how his shop works, he relegates it to the "odds and ends" note at the bottom.

I think polling is almost always overfond of the concept of the "swing", of the idea of a large class of the judicious weighing their options and taking their time to decide finally moving in one direction or another relative to their position in the last election, in the last few days of the campaign, or grandly moving in two opposite directions like the waters of the Red Sea at the Lord's command. The data can more or less be interpreted this way more or less all the time, because you're not looking at individuals but at populations, other than exit polls where you can ask the individuals who they voted for previously, and in those you have no information about those who didn't vote.

Thus the conventional picture of the presidential elections of 2012 and 2016, where almost everybody thinks a whole class of people "swung" from voting Democratic when Obama was running to Republicans when Trump was, where I think Obama's edge was mostly in one set of people who don't normally vote and Trump's surprisingly close finish in a different set of people who don't normally vote while the Obama nonvoters went back to staying home (I went into this in some detail in a post last year).

And there are various signs around that there is some kind of really massive surge going on on the part of those who don't vote very often, in Florida, in Tennessee, in New Jersey (where excitement over some extremely attractive Democratic House candidates could pull through the endangered, personally yucky but politically often valuable Bob Menendez), in Arizona (where Republicans had the edge a week ago), nationwide. Cameron Joseph at TPM says of Texas:

“If the electorate looks like these polls look like no, Beto doesn’t have a chance. But the early vote does give you reason for optimism in terms of the electorate looking pretty dramatically different in Texas,” Democratic modeling expert Tom Bonier, who is doing some consulting for the Texas Democratic Party, told TPM on Wednesday. “My instinct on that is the polls are overly conservative in terms of the composition of the electorate and you’re looking at a margin of error race.”

According to Bonier’s model, at this point in the early vote in 2014 people 65 or older had cast 47 percent of the ballots. This time around, it’s down to 34 percent.  The percentage of the early vote cast by voters under 40 has climbed from just 11 percent in 2014 to 21 percent now.

Then there's the evidence from primaries in low-turnout states like New York, where it was off the charts in September. And perhaps ferment among the Native peoples of North Dakota and the black voters of Kansas and Georgia, hopefully enraged by Republicans' blatant attempts to suppress their voices with a kind of renewed Jim Crow technique. 

And nonvoters being characteristically younger, and/or minority members, that surge overall, in states with Senate contests and states without, certainly favors Democrats. Whether it favors us enough to make the difference nobody can say, but if there is an upset in the Senate predictions that's where it's going to be.