Democrats in midterm elections this fall are in for a drubbing that may reach generational, even biblical, proportions. Or so recent commentary has surmised, based on far-in-advance election polling and President Biden’s cratering approval ratings.
Perhaps there is a thumping ahead, an electoral bloodbath. But then, history suggests maybe not.
Poll-based previews of U.S. elections offered months in advance sometimes are far off-target. Not always, but it happens often enough to take note in 2022. Elections that were expected to be close have turned out to be landslides, or nearly so. Likewise, elections thought to be one-sided were far more competitive than months-in-advance assessments suggested.
A memorable example of a near-landslide that never materialized was Biden’s election to the presidency two years ago. Some of the most prominent national polls indicated Biden was headed for a double-digit popular vote victory over President Trump. CNN’s final pre-election poll, for example, estimated Biden’s lead at 12 percentage points, down slightly from 14 points that June. Quinnipiac University’s poll pegged Biden’s margin at 11 points, down from 15 points in July.
Biden won the popular vote by 4.5 percentage points. Senate and gubernatorial races in 2020 also diverged markedly from the expectations shaped by pundits and pre-election polls.
The overall polling performance was the worst in a presidential race in 40 years, according to a report by the American Association for Public Opinion Research.
To be sure, the dynamics of presidential elections differ from midterm elections. Voter turnout tends to be more robust in presidential elections. Campaigns for the presidency attract far more interest than midterm races. The party of the president seldom fares well in midterms.
But the salient point here is that expectations about national elections, confidently asserted months in advance, are prone to error — a historical reality tends to be overlooked as a dominant narrative takes hold about an unfolding national campaign.
Besides, midterms are not immune from off-target, poll-based expectations, either. Remember 2014? Republican candidates outperformed the polls overall and defeated Democratic senators in Colorado and North Carolina, and nearly ousted the Virginia incumbent whose summertime polling lead had reached 25 points.
It’s likewise instructive to recall the 2000 presidential campaign, when late-summer polls suggested George W. Bush’s run for president had turned hopeless, that his candidacy was “toast.” “Stick a fork in him,” one media analyst wrote. “He’s done.”
In the 1996 campaign, prominent pre-election polls signaled an overwhelming victory for President Bill Clinton in a three-way race against Republican Bob Dole and third-party candidate H. Ross Perot. The final CBS News poll before the election estimated Clinton’s lead at 18 points. Using the same data, the New York Times figured Clinton’s advantage was 16 points. The Pew Research Center’s survey placed Clinton ahead by 14 points. Clinton won by 8.5 points, a comfortable margin but not exactly the landslide that some polls had anticipated.
Another variant of expectations unfulfilled is the scenario of an anticipated close race that turns out to be a landslide. Polls in summer 1936 indicated a tight race between President Franklin D. Roosevelt and Alf Landon. In July 1936, the Gallup poll estimated that Landon was narrowly ahead in projected electoral votes, 272 to 259. At the time, the president’s wife, Eleanor, said she, too, expected a close race.
That was the year when the Literary Digest magazine — which had correctly predicted the winners of presidential elections since 1920 — announced that its massive mail-in poll pointed to a landslide for Landon, that the Republican would win 33 states and 370 electoral votes.
Landon carried two states and eight electoral votes in what was one of the most crushing elections in U.S. history. Roosevelt was returned to office with 523 electoral votes and a popular vote margin of 24 percentage points.
Errant or misleading polls are a major — but certainly not the only — reason that expectations can go unfulfilled in national elections. (Some pollsters argue, rather disingenuously, that election surveys are not predictions.) Campaigns tend to matter. President Gerald Ford, after all, nearly overcame a 33-point deficit in a midsummer Gallup poll in his race against Jimmy Carter in 1976. Whatever their cause, decisive shifts in voters’ sentiments are hardly uncommon in the months before voting.
The record retraced here encourages caution and humility about expectations developed six or seven months ahead of national elections. Such intervals can be simply too wide for accuracy.
W. Joseph Campbell is a professor of communication at American University and the author of seven books, including, most recently, “Lost in a Gallup: Polling Failure in U.S. Presidential Elections.” Follow him on Twitter @wjosephcampbell.