With people hyperventilating over minute shifts in the polls, let’s remember a little-known factoid: public opinion doesn’t shift. People are locked in. Remember the dramatic shift in public opinion over gay marriage? That was about a 12-point shift over 10 freaking years. That’s considered a dramatic upending of public opinion—about a one-point shift per year!
Today, there isn’t a single Donald Trump supporter who might be persuaded to vote for Hillary Clinton, and not a single Clinton supporter who might vote for Trump. So all that jumping up and down in the polls? Noise. Take a look at this, from 2012:
Ostensibly, the graph shows how much better Barack Obama’s polling was than Gallup, which isn’t too hard to prove given that Gallup had Mitt Romney winning in the end. But I’m actually interested in something else—how little that needle moved in Obama’s polling. From June to the end, Obama’s numbers shifted from 51 to 53 percent, and ended up back where he started, at around 52 percent. In other words, +/- one point. The convention? +2 boost. “47 percent” scandal? Another +1. After the first debate, it was back down two. That’s it.
Looking at this cycle’s aggregate polling, Clinton’s nadir since June has been 44 percent and her apex 48 percent—so again, four measly points. For Trump, the numbers are 40 and 42, respectively. Again, for all the sturm und drang, the numbers are simply not moving that much, and Trump has that severe demographic ceiling to contend with. Bernie Sanders busted through his demographic ceiling by winning young people of color, and that only got him from 30 to 40 percent. Trump, in the broader electorate, is stuck at 42. Further consolidation of the conservative vote and erosion in Gary Johnson’s support might get him closer to Mitt Romney’s 47 percent, but that still would leave him at Mitt Romney levels of support, aka losing.
This is a weird election, but we are still seeing the same stability we saw in 2012. Trump may gain among uneducated whites, but that demo was already heavily Republican. Clinton will gain among Latinos, but that demo was already heavily Democratic. The two will likely offset each other.
If anything, it is Trump who has bigger base worries, as white suburban women remain in play. They may come home (which appeared to be the case as the polls tightened), or they may not. Tribalism is so strong today that base Republicans can’t even reject, en masse, a literal white supremacist. And no, they’re not all racist fucks. But ultimately, they’re all rallying around their David Duke-style nominee.
The only real question is turnout, which is the difference between big Democratic years (presidential years), and Republican ones (off years). We’re even going to get a controlled study on the value of a ground game, given that one campaign has one and the other doesn’t.
But who supports who is already decided. If someone could correctly predict the exact demographics of November’s voter pool, I could use those numbers to calculate who will win and by exactly what margins. That’s how predictable and unmovable our voter preferences have become.
Let’s go down to the states, like Pennsylvania:
The entire difference in the margins over the past year is +/- 1.5 points or so.
North Carolina:
Same kind of fluctuation as Pennsylvania: looks like 1.6 points, tops.
Ohio:
That went from a +5 or so Clinton lead, to a half-point lead today. But really, much of that is consolidation of conservative support, as Clinton is pretty much exactly where she began, while Trump has crept up to near-parity. All that jumping around is just noise. In the end, Ohio is Ohio, and will be close as it always is because we can’t break free from our national partisan constraints. Note how immovable the margin between the two candidates is after May, when our nominees were decided.
And the results look the same almost everywhere—some movement on the margins, but things look eerily similar to 2012, and just as steady as then.
All that’s to say: Don’t sweat the minute shifts in the polling. People’s opinions really aren’t changing. It’s all just random statistical noise, even in the battleground states! If we win Romney states like North Carolina, Arizona, and even Georgia, it’ll be because we got more Latinos, African Americans, single women, and young voters to the polls. If Trump wins any Obama states, it’ll be because he squeezed out a few more white voters in historically tight white states, like Ohio and Iowa.
But there aren’t enough of those states to overcome Clinton’s big EV lead, and especially not when the electorate is as immovable as ours today.