Is Donald Trump Already Doomed?
It is one of the canons of American political commentary: ignore general election polls taken during the primary. Pundits, prognosticators, and psephologists repeat it so often it’s become a mantra. John Sides, the political scientist who runs The Monkey Cage blog, penned a long post last July in which he argued that news stories about polls conducted so far in advance should include caveats about their lack of predictive value. Harry Enten of FiveThirtyEight was blunter when he declared in November that “History’s lesson is clear: Don’t pay attention to general election polls a year before the election.” As Enten noted, since WWII polls a year out have been off by an average of ten points, and sometimes much more. Just think of 1992, when George H. W. Bush led Bill Clinton by over twenty points.
Clinton’s comeback is a hopeful omen for Republican frontrunner Donald Trump. He currently lags the presumptive Democratic nominee, Hillary Clinton (wife of the aforementioned Bill), by 8.5 points in the RealClearPolitics average and 8.1 points in the HuffPost Pollster average. Conversely, her substantial lead offers no assurance to the former First Lady and Secretary of State. Early polls are not determinative, after all; the experts say so. That’s true, too – in July or November the year before an election. Now, however, it is April of election year. And if data from previous cycles is correct, then Hillary Clinton should be pleased. For it is quite possible that Donald Trump is already doomed.
Trump’s problem is that polls from this stage of the election cycle have a strong correlation with the eventual result. Political scientists Christopher Wlezien and Robert Erikson crunched data for the elections from 1956-2012. They found that 300 days before the election polls have little relationship to the outcome. But by the time you hit April their predictive value increases considerably. Wlezien and Erikson discovered, according to a summary of their research by Vox’s Andrew Prokop, that by “mid-April of the election year, polls explain about half the variance in the eventual vote split. And mid-April polls have correctly ‘called’ the winner in about two-thirds of the cases since 1952.” Polling changes registered in the spring tend to stick because they occur at the end of a contested primary. Voters have been exposed to the candidates for several months and have begun sizing up their general election potential. As late as mid-February Trump’s deficit was under five points. But since then his numbers have tanked. As Prokop makes clear, Trump’s polling collapse occurred at the worst possible time for him, as it happened just when such shifts begin to matter. If Trump’s drop keeps to the historical pattern, then the reality TV host and real estate mogul is in trouble.
To test this proposition, I looked at general election polls from April for the last three presidential races. First, 2008. There are sixteen April polls in RealClearPolitics’ table of general election polls from that year.
Of these sixteen polls, John McCain led in two, five were tied, and the remaining nine showed Barack Obama ahead anywhere from one to six points. That November, Obama won by seven.
Four years later the final outcome was also being foreshadowed by April.
The RCP table of general election polls for 2012 contains sixteen polls from April of that year. They too show Barack Obama as a strong favorite. He led eleven polls, two were tied, and Mitt Romney held leads in three. Obama beat Romney by four points.
This phenomenon is not limited to President Obama, either. Here is the comparable evidence from 2004.
RCP lists sixteen polls from April of 2004. George W. Bush was ahead in twelve while his opponent, Sen. John Kerry, led but four. Bush defeated Kerry by three points.
In the last three elections, the candidate who was ahead in April won the presidency. As previously noted, the April leader wins about two-thirds of the time. Hillary Clinton would seem to be in excellent position. In fact, history underrates her chances. A look at this year’s polling data reveals why.
There aren’t enough polls to focus on April alone, so I’ve also included those from March in the RCP table of Trump vs. Clinton polls.
Sixteen more polls, and Mrs. Clinton leads every one. In fact, Trump has led in only two polls from 2016 in RCP’s table, and none since mid-February. Clinton’s advantage appears even more dramatic when represented as a graph.
As this graph from HuffPost Pollster shows, Trump has never led Clinton. The margin was close in the fall but has gradually increased since January, and expanded rapidly in the last two months. That’s what happens when you trail in forty-seven consecutive polls, as Trump has.
Kerry, McCain, and Romney, though they were behind, led in a random poll here and there in April of their election years. Trump does not fare even that poorly. His numbers are deteriorating just when general election polls begin to mean something. The reason isn’t hard to find: voters loathe Donald Trump. His favorability rating has never risen above fifteen points underwater, and today the deficit hovers between thirty and forty points. Again, the graph tells the story best.
This graph is of polls of voters and adults. Trump does even worse with specific demographic groups, for example millennials, Latinos, and women. Such numbers are what make Trump what he is: the most unpopular major presidential candidate in modern history.
Numbers like this are why it matters not a whit that Trump’s convention manager, Paul Manafort, told a meeting of the Republican National Committee last week that Trump’s vulgar bombast and belligerence during the campaign was merely an act and that his behavior was “evolving.” Never mind that Trump’s new political director, Rick Wiley (late of the unlamented Scott Walker campaign), regaled the same meeting with tales of Trump’s ability to expand the map and compete in places like Wisconsin, Pennsylvania, even Connecticut, New Jersey and Illinois. And forget about Trump’s massive victories in the so-called Acela primary. Trump’s abysmal numbers render all of it moot. Reporters and Trump boosters can fantasize all they want about Trump’s path and the Rust Belt. Trump turns it all to dust. Trump has no path as long as he is on it. It is only April, yet we may already have reached the point where any discussion of what Trump will do in the fall is simply the punditry equivalent of shuffling deck chairs on the Titanic.
Trump hasn’t crashed into the iceberg yet. And not every observer regards April polls as bearing the metaphysical certitude of the writing on the wall. Nonetheless, every rubric available – favorability ratings, demographics, and head-to-head ballot tests – suggests that Trump is headed for a decisive, perhaps even crushing, defeat. The past is prologue. If history repeats itself – and, as I have argued, there is considerable reason to believe it shall – then Donald Trump’s only path is the one leading him to the biggest loss in a presidential election in nearly thirty years.
 I am only considering a Clinton vs. Trump contest. Bernie Sanders leads Trump (and all Republicans) by even larger margins than Clinton does. He is not the frontrunner, though, so his advantage is largely hypothetical. There is also reason to believe it is something of a mirage.
Image by Gage Skidmore