Polling, punditry and plenty of poor predictions

Hey there, time traveller!
This article was published 11/04/2023 (1257 days ago), so information in it may no longer be current.

The votes were last counted in the United States following the mid-term elections on Nov. 8, 2022, in which all the seats were contested in the House of Representatives and about a third of the 100 seats in the Senate.

Yet that was merely the end of the most recent two-year cycle. The very next day marked the beginning of a new one, because the election campaigning, incessant (almost pathological) polling and non-stop media speculation about potential winners and losers the next time Americans cast their votes never truly stops.

Early in November 2024, all the seats in the House will again be up for grabs, as will another third of the Senate seats, and the election of a president.

Incumbents in the House of Representatives have tremendous advantages over opponents in terms of financing and media access. They also benefit from decades of manipulations of electoral districts, referred to as “gerrymandering,” by state governments (mainly Republican ones, recently) that have created districts where the election of a Republican or Democratic candidate, depending on the state in question, is pretty much guaranteed.

The race to become president naturally receives the most attention. It starts close to two years before the vote, which is why Donald Trump — trying against all odds to become the first former president since Grover Cleveland in 1892 to serve a second term four years after he lost an election — and former South Carolina governor Nikki Haley, along with a few others, have declared their candidacies to become the Republican nominee for the next election.

With each new declared candidate, the pollsters and journalists can’t help themselves. How significant is it that in early April — 10 months before the first primary vote — Trump is pulling ahead of Florida Gov. Ron DeSantis in a few polls? The answer: likely not at all.

Considering Trump’s various legal problems and the fact he has been indicted in New York, with more indictments from other jurisdictions for other alleged crimes to come, his standing right now — given that DeSantis, former vice-president Mike Pence and other potential candidates have yet to declare their presidential intentions — as the Republican to beat means little.

And yet, the results of such polls are reported and commented on in the media as if they were representative of how voters might cast their ballots. (True, Trump’s polling numbers among Republicans have risen since the New York indictment —and DeSantis (who seems out of his league on a national stage), Haley and Pence have attracted only mild support among Republicans — but, overall, this voting demographic keeps shrinking.)

A case in point: the polls were wrong about the 2022 midterm elections that predicted a massive Republican victory, as they were about predictions for elections in 2018 and 2020. (It’s wise to keep in mind former prime minister John Diefenbaker remark: “I never trusted a poll. Only dogs know best what to do with poles.”) It is difficult to say how much, if any, influence polls have on the behaviour of voters.

Pundits equally contribute to the confusion with incorrect, and even absurd, speculations based upon a reflection of events at a particular moment in time and frequently without regard to any realistic political perspective.

In mid-October 2015, Jonathan Bernstein, a respected political opinion columnist for Bloomberg, declared that there was no way Trump would become the Republican nominee and held that position into the spring of 2016. He was convinced, wrongly as it turned out, that the other Republicans running at the time would figure out a way to stop him. They didn’t.

Similarly, in mid-January 2020, the usually clear-eyed New York Times columnist Bret Stephens tried convince himself that based on primary polls in Iowa and New Hampshire he was wrong to think that left-wing Vermont Senator Bernie Sanders did not have a real chance to become the Democratic Party nominee — a mistaken premise if there ever was one. The Democratic Party was never going to choose Sanders over a moderate like Biden, who could take on Trump and defeat him — which is exactly what happened.

Pollsters and pundits are quick to tout the potential success of any politician who, for a variety of reasons, becomes a front-runner even for a short time. Victory for these politicians seem a sure thing — until it is not.

There was Republican Herman Cain in 2012; Hilary Clinton in 2008 (a year before the November 2008 election, a USA Today/Gallup poll gave Clinton a 29-point lead over relative newcomer Barack Obama); in 2008, Rudy Giuliani was going to defeat John McCain as the Republican nominee; in 2004, journalists declared that former Vermont governor Howard Dean was the Democratic candidate to beat (that was before his famous battle-cry scream at the end of a speech); in 1992, former California governor Jerry Brown was polling well ahead of Arkansas governor Bill Clinton, who was given little chance of winning.

And, so, it has gone. Hyped predictions about a particular candidate fade into the sunset and lessons about predicting results of elections months before a vote are never learned. Expect more of the same in the coming months.

Now & Then is a column in which historian Allan Levine puts the events of today in a historical context.