How Trump Wins

Printed from:

If you watch enough mainstream media, you know for sure that Donald Trump is down in the polls – probably hopelessly so – and that, above all, 2020 is not like 2016.

But actually, 2020 looks a lot like 2016. Now as then, Trump has to win Republican-leaning battleground states Florida, North Carolina, and Ohio. Now as then, he’s in decent shape to win all three.

Now as then, Trump needs one of Democratic-leaning battleground states Pennsylvania, Michigan, and Wisconsin. (This year you can throw in Minnesota, which also looks to be in play, though Biden leads there.)

Trump also needs Arizona in this scenario – though there’s another way to sneak in, if he can get one of Pennsylvania-Michigan-Wisconsin plus Nevada plus (if the one Rust Belt state is 10-electoral-votes Wisconsin) one electoral vote out of the second congressional district in central and northern Maine.

(This scenario assumes that Trump holds serve in Republican strongholds, that he keeps Republican-leaning states like Georgia and Texas, that he wins Iowa (where a late poll has him surging to a seven-point lead), and that he loses battleground New Hampshire.)

In 2016, the mainstream media could not conceive of a way for Trump to win any of Pennsylvania, Michigan, or Wisconsin. Neither of the three states had gone Republican in decades. Trump led in almost no public polls in those states. Yet he won all three.

That doesn’t seem possible this time around, if you go by the Real Clear Politics average of polls. It seems absurd if you go by mainstream media polls.

But what if most individual pollsters are off, and what if the average of polls is off, and what if one particular pollster is the one closest to the mark?

That’s what happened in 2016, when only Trafalgar Group, of Atlanta, Georgia, a Republican polling and political consultancy firm led by Robert Cahaly, correctly called the election, even down to the final tally in electoral votes. (Trafalgar got a couple of states wrong, but the final numbers for electoral votes came out exact.)

Cahaly says he and his firm have figured out better than other pollsters how to account for social desirability bias – which in this case means the high likelihood that Trump voters won’t tell a pollster that they’re planning to vote for Trump, for fear of what the consequences might be.

He also claims his samples are better than the samples of most other pollsters. For every state he surveys at least 1,000 voters, not 500 or 600 as some other pollsters do. He reaches younger voters by text message, on the theory that they won’t answer a land line telephone or even calls placed to their cell phones. When a particular survey is low on a particular demographic group – like, say, registered Republicans – he doesn’t simply extrapolate the numbers by weighting them according to their actual representation, but goes out and tries to get more respondents to the poll from that group.

Cahaly tries to keep his polls short (seven to nine questions) and anonymous. Where possible he uses digital platforms that make the respondent feel as though he isn’t being tracked. He also asks certain questions designed to ferret out the pollster-shy Trump voter. He won’t say exactly what those are, but he offered one example during an October 19 podcast interview with Rich Lowry of National Review:  whether someone has a hunting or fishing license.

“I’ve got to get past what you want to say in public and get to what you really feel. Because what’s in your heart is going to be what’s on that ballot,” Cahaly said during the October 19 National Review podcast.

Cahaly’s methods are laughed off by some mainstream pollsters, who claim his success comes from a broken-clock scenario – right on certain occasions in spite of itself. Yet it’s hard to argue with his success last time around.

So what does Cahaly’s Trafalgar Group see this time around?

Trafalgar’s late-race polls show Trump up 2.5 percentage points in Arizona and down 2.3 points in Nevada. While within the margin of error, the late October polls suggests a Trump win in Arizona (like last time) and a Trump loss in Nevada (like last time).

Trafalgar has Trump up 3 points in Florida. That suggests a Trump win in Florida. (Like last time.)

Trafalgar (as of Sunday, October 25) has Trump up 2.8 points in North Carolina. That suggests a Trump win in North Carolina. (Like last time.)

Trafalgar doesn’t seem to have released a poll of Ohio since early October, when it found Trump leading by 3.7 points. Most recent polls have the race close to even, though Ohio trends Republican in close races.

It’s also not clear when Trafalgar last polled Georgia, though Cahaly dismissed Biden’s chances there two weeks ago because of the state’s large Republican base. (Most polls released during the past week show the race within a point or two either way.)

The real shock for those watching mainstream polls is Michigan, where Trafalgar has Trump up by 2.5 percentage points. Cahaly believes Michigan is the most likely to vote for Trump out of the Rust Belt states. The black Republican U.S. Senate challenger there, John James, may actually boost Trump by drawing voters to the polls who otherwise might not vote and who certainly ordinarily wouldn’t vote Republican, Cahaly told Lowry on October 19.

In Wisconsin, Trafalgar has Biden up just four-tenths of a percentage point. That suggests a tossup.

As of Sunday, November 1, Trafalgar’s most recent public poll in Pennsylvania was released Sunday, October 25. It showed Trump leading by eight-tenths of a percentage point – though Cahaly told Lowry on October 19 that given Pennsylvania’s mail-in balloting uncertainty, the state could go for Biden even if Trump actually wins more votes there.

What’s the upshot?

Let’s assume Trafalgar is right for any polls showing a candidate up by more than 2 points, and that any states where the lead is less than 2 points (according to Trafalgar, that is) go to Biden. Let’s also assume that Biden wins all four electoral votes in Maine (including the contested second congressional district) and one electoral vote in Nebraska from the congressional district that includes Omaha.

In that scenario, Trump would carry battleground states Florida, North Carolina, Ohio, Arizona, Iowa, and Michigan – but lose battleground states Pennsylvania, Wisconsin, Minnesota, and New Hampshire.

The result?

Trump would get 280 electoral votes, Biden would get 258.

It takes 270 to win.

Four more years.