Here’s my first stab at what is likely to happen on Tuesday.
First off: my prediction for the Presidential race is Hillary Clinton winning by 7 points in the popular vote – 50.1 to 43.1 (rest to third party).
Electoral College at least 323 for Clinton though it could be higher.
Why am I so bullish on Hillary Clinton’s chances? Here is my explanation.
There are a few factors that inform the analysis and are worth discussing on their own.
1. Polls and polling aggregators – In the Moneyball Age of Big Data everybody wants to show off their mad forecast quantification skillz. Nate Silver is the most prominent, but he’s hardly the only one. The fact of the matter is: poll aggregators are only as good as the polls they aggregate and the models they use to aggregate them. Most of the aggregators are close to one another. The difference between, say, Sam Wang and Nate Silver has to do with whether or not national polls are considered, whether the polls are weighted for house effect, and in Silver’s case, whether there is a major Black Swan factor built into the model. As Nate has said throughout, he builds in about 12-14 points of a buffer at the high end in the event that “all the polls are wrong.” Personally, I think that’s silly. I can see maybe three or four percentage points insuring against systemic polling error, but no more. After all, systemic polling error can go in either direction.
Which leads to the next factor:
2. Turnout operations and models. Much ink has been spilt since 2012 about two potential paths forward for the GOP. One approach is to grow the party with non-whites, especially as the population becomes diverse. The other approach, taken by the Trump campaign, is to find the “hidden white vote” that never turns out and amp up support among non-college educated whites. Trump has had modest success attracting non-college whites at a better rate than Romney. But he hasn’t gotten the sizable “hidden white vote” to even register, which suggests that his gains will be fairly minimal in the end. He probably convinced a fair number Rust Belt ex-Democrats to cut the cord and register as Republicans. But registration numbers show he hasn’t brought large numbers of non-voting whites into the electoral process. Meanwhile, college educated whites (especially women) appear to have moved from their historic GOP home to the Democrats this year, largely offsetting any defections the other way.
Still, the bigger turnout question has to do with how many whites will turn out at all as a percentage of the population. In 2012, whites were 72 percent of the electorate. In 2008, whites were 74 percent. Pew did a large study and estimated that 69 percent of the electorate in 2016 will be white. This number is important. Poll models favoring Trump typically show likely voters at around 76 percent white or higher.
Considering the 90+% support for Clinton among blacks and the 79-18 percent support for her among Latinos (per Latino Decisions) these racial/ethnic turnout models are critical to understanding how the vote will look.
What does the ethnic makeup look like for this election and how can we determine it at this point? Here we can look to the third factor: Early Voters.
3 Early voting has become a routine part of elections now. 37 states have early voting now (without “excuse”). There’s a good chance nearly half of Americans will have already voted by Tuesday, with near 70 percent in Florida. What can we learn from Early Voting numbers? We can see how effective the GOTV programs are for each campaign. They set targets and try to meet them. The public only gets tidbits of what those targets are, but there are plenty of reports out there about campaigns meeting or not their in person and absentee early voting goals. Obviously, an early voting campaign can cannibalize Election Day votes, so it’s also good to see if these early voters are “low propensity voters.” From all the reporting in Florida and Nevada in particular, Clinton has achieved remarkable success in turning out “low propensity” Latino voters, many of them registered as No Party. Remember that these voters do not show up as “likely voters” in polls.
4. Finally, there is the matter of momentum and voting shifts in the late campaign. I am of the belief that opinions rarely move unless a truly earth shattering event happens, or if one of the candidates is thoroughly humiliated at, say, a debate or by a scandal. And even that is pretty rare. In this year, there were so many October Surprises that they canceled each other out and reinforced partisan views. Unfortunately, polls tend to pick up attitudinal shifts and measure them as voter preference shifts, when that doesn’t usually happen. On the other hand, it is very common for partisans to “come home” close to Election Day. Note that most of Trump’s polling gains happened the week before the Comey letter, but after the third debate chaos wore off. As I said at the time, the Comey letter would not move the dial unless there was an imminent indictment or some other bombshell. The same was true for any last minute revelations about Trump and Russia.
So where does this leave us? The major national poll aggregators give Clinton about 2.5 to 5 point lead. Because of the high turnout of “low propensity” minority voters and unusually high female early voting compared to 2012, I think Clinton is positioned to beat her final polling numbers by quite a lot. Clinton’s GOTV operation is working. How much will she beat final polls? Obama beat his final poll numbers by two whole points (3.9 percent v 1.9 percent). And Romney had a decent turnout operation. Trump has no turnout operation beyond what local Republicans muster up. Read all the state reports out of FL, NC, NV and OH and you’ll see more detail. Note that African American turnout looked slow at first but has now picked up pace a lot.
So I see Clinton beating her final poll aggregate by between 3 or 4 points. That gives her a national victory of about 7 points. So I’ll call it 50.1 (Clinton) – 43.1 (Trump) – 4 (Johnson) – 1 (Stein) – 1 (others).
As for the Electoral College, I think Clinton easily holds her Blue Wall, winning VA, PA, MI, WI, CO and NM and then taking NH and NV to get over the top. Strong turnout operations deliver FL comfortably and NC. The question marks will be OH, IA, GA and AZ. Those four have no effect on the margin, but I’ll have a better idea on Monday. Either way, Hillary Clinton clears 323 and can get as high as 374. I’ve said it all along, this will not be close. I’ll update again Monday morning with Senate predictions and any possible changes.