Trump’s looming defeat
In the coming US presidential election, I have my own bias, and it’s occupational. I don’t care whether the Democrats or the Republicans win; I’m rooting for the pollsters as a group. The 2020 election is their chance to redeem themselves from the big surprise of 2016.
Every election, in every country, is an opportunity for survey doers to publicly exhibit the quality of their work, and for anti-survey pundits to hope for a mistake. As a fellow pollster at a conference remarked, “We’re like tightrope walkers in a circus: the audience expects us to cross safely to the other size, but they know, and we know, that there’s always danger of a slip.” Every few decades, a circus accident happens, and makes the headlines.
In 2016, the polls did slip—see “The polls missed Trump. We asked pollsters why” (11/9/16), by FiveThirtyEight.com, which I heartily recommend as a reader-friendly aggregator of US polls. Nate Silver, head of 538, isn’t a pollster himself, but a smart bookmaker who knows the odds on the outcome of sports events.
538 is the number of votes in the US electoral college; the presidency is won by a majority of 270 electoral college votes, not by the majority of votes cast by Americans nationwide. Luckily for Trump, he outperformed his swing state polls. He won in the electoral college, unexpectedly but within the margin for error; but he lost the national vote as expected.
Throughout his term, Trump’s popularity has been sub-majority (“The unpopular Donald Trump,” Opinion, 2/23/20). The Democrats won control of the House of Representatives in the 2018 election. Trump was impeached in the House, but acquitted because the Republicans controlled the Senate. His latest disapproval rate is 53.8 percent.
Right now, 538 says that Biden’s chance of winning the election is 88 percent, and Trump’s chance is 12 percent (“Where the presidential race stands before the last debate,” 10/22/20). It says Biden has an 86 percent chance of winning Pennsylvania, the probable tipping-point state, and has a 72 percent chance of winning Florida. From what I see in 538, Biden is up in the following battleground states (see my 2/22/20 column): Colorado, Florida, Michigan, Minnesota, Nevada, New Hampshire, North Carolina, Pennsylvania, Virginia, and Wisconsin; Trump is up only in Ohio; and there is a tie in Iowa.
In national polls, Biden’s average lead is 9.9 points. Biden is +5 in a poll by Fox News, +6 from Suffolk University, +8 from Quinnipiac University, and +10 from CNN/SSRS. The bigger the national lead, the greater the likelihood of winning the electoral college.
What I particularly enjoy about 538 is its meticulous description of each poll it reports and its rating of the pollster that did it. For each poll, it states fieldwork dates, sample size, and whether the respondents are likely voters (LV) or registered voters (RV). Virtually all statewide samples have 625+ respondents, the cut-off for plus/minus 4 percent sampling error. An error of plus/minus 3 points needs a sample of 1,000; anything larger is probably not a probability sample, and so sampling error cannot be calculated.
538 rates the pollsters, from A+ to F, and states their modes of interview (live is best). My personal favorite is ABC News/Washington Post (A+, live). Marist University is also rated A+. Suffolk University, CBS/New York Times, and SurveyUSA are all rated A. NBC News/Wall Street Journal is A-. Fox News is A/B. Quinnipiac University and CNN are both B+. YouGov is B. I ignore SurveyMonkey (D-, online).
So how does Trump plan to game the election of 2020? For one thing, he has packed the US Supreme Court, where the seats are for life, and hopes for as many state races as possible to be litigated all the way there. He has refused to say that he will concede peacefully if he loses.
The US election story is all about Trump. Not about how Biden is bound to win, but about how Trump is poised to lose.
——————
Contact [email protected].
