We’re less than two weeks away and the polls would suggest that Joe Biden’s on his way to finishing above 300 in the Electoral College. Could Donald Trump still pull this off? Does he have a chance? Could some of the state polls be as off as they were in 2016? Or as accurate as they were in 2012?
Approval rating average of the president 11 days before his possible re-election v. how much of the popular vote they got (per 538)
45 Donald Trump – 42.5%
44 Barack Obama – 48.5% (W) 51.1%
43 George W. Bush – 48.4% (W) 50.7%
42 Bill Clinton – 57.5% (W) 49.2%
41 George Bush – 32.6% (L) 37.4%
40 Ronald Reagan – 54.5% (W) 58.8%
39 Jimmy Carter – 37.9% (L) 41%
38 Gerald Ford – 43.6% (L) 48%
37 Richard Nixon – 58.7% (W) 60.7%
Bill Clinton’s the only time he got less of the popular vote than his approval, but I chalk that up to Americans inflating his approval rating as a signal to Congress they didn’t think he should have been impeached, combined with 3rd-party man Ross Perot getting 8.4%.
Trump got 46.1% of the popular vote in 2016. I could see him getting that much again, maybe more. It’s a matter of where those votes will be. When I look at the correlation of approval rating and election turnout, the uptick of the more recent presidents hasn’t been as dramatic. About 2-3% for Barack Obama and George W. Bush. If Trump gets less than a 3% bump, the networks can call this one early.
The lesson of 2016 was that there was a lot of silent support for Donald Trump combined with the Democrats not realizing just how unpopular Hillary Clinton was. At the time I thought about 12 of the Republicans who ran in 2016 could have beaten Clinton outright, but they nominated Trump, who was thoroughly beaten in the popular vote but mostly in states like California and New York where it didn’t matter.
The lesson of 2012 was that there’s no such thing as “unskewed” polls. Mitt Romney’s team genuinely thought they were going to win, but they couldn’t move the margins in the Midwest.
Now when looking at this election, it’s important to frame it as an incumbent vs. an opponent. Is it a referendum on the incumbent or was he able to make it a referendum on his opponent? It’s the job of the incumbent’s team to highlight the good and portray the opponent as an unacceptable alternative. It’s the job of the opponent’s team to make it seem that we just can’t take another four years of this; our candidate will be better.
Anyway, back to the numbers. On Election Day 2016, 538 gave Trump a 29% chance of winning the popular vote, final estimate of 48.5%-44.9% for Clinton, and their model gave Clinton +77 electoral votes. Obviously they were within the margin-of-error on the popular vote and way off on the Electoral College. That led them to do a lot of model-adjustments when it came to state polling. National polls are neat, but the state polls determine the Electoral College, and if the past twenty years have made anything clear, it’s that the popular vote is ultimately irrelevant. The Electoral College is all that matters.
When I take the averages/forecasts of 538 and RealClearPolitics, it shows Biden getting 300+ EC votes. The RCP No-Toss-Up map shows Biden winning 357-181. I give less than 1% to Biden actually achieving such a landslide, barring some amazing October surprise like Trump on a taped conversation with David Duke cackling maniacally. (I don’t see him shooting someone on Fifth Avenue as a needle-mover.) An example of my lack of faith in Biden hitting 357: the No-Toss-Up map gives Iowa AND North Carolina to Biden and I can’t fathom that happening.
So in 2016, which states were the most off in predicting Clinton would win them?
Wisconsin – off by 6%
Michigan – off by 4.5%
Pennsylvania – off by 4%
Therefore, I’m going to assume even with their post-2016 model adjustments that the 538 model is still underestimating Trump by 2.5% in every state. And where does that put the final Electoral College count? Biden wins 319-219. Could Donald Trump be underestimated by 3.5%? Biden 279-259. The bellwether state will be Pennsylvania. Even if Trump wins Florida, where he’s down by 3%, he doesn’t have a path to 270 without Pennsylvania, where he’s down 5.7%.