When the electoral college meets on Monday to ratify the election of Joe Biden, they will affirm the result that many polls had indicated would occur.
Yet during the 2020 U.S. presidential campaign, discussions in Taiwan and in the U.S. online ecosystems centered on how polls were unreliable because they had mistakenly expected Hilary Clinton to win in 2016. In the aftermath of the election, some have claimed that the 2020 polls were only “lucky” to have suggested a Joe Biden victory, and skepticism about the polls in Taiwan tracks with skepticism about the integrity of the entire U.S. election.
What’s often missing from these discussions about the polls is the social atmosphere in the United States and polling methodology.
Hilary Clinton’s defeat made absolute sense
The 2016 election had two central themes: one was hating on Clinton, another was the economy. Why did the criticism and negative coverage about Trump fail to make even a dent in his support? Clinton didn’t lose to Trump, but she lost to herself and Barack Obama’s eight-year-long political baggage.
As a seasoned politician, Clinton was beyond qualified. But her tough personal image and establishment background made it hard for her to score positive media coverage. “Emailgate,” the 2012 Benghazi attack, and the Clinton Foundation scandals — all of these controversies have tainted her political resume.
By contrast, Trump as a challenger locked in voters within the “blue wall” who were disappointed at Obama’s economic policies. Clinton, who was overconfident and neglected the blue states, chose to forego campaigning in Wisconsin and turned to Republican stronghold Arizona.
It’s hard to imagine any candidate skipping a rally in Wisconsin today. But four years ago, Clinton did exactly that.
“Why is Hilary Clinton so sure that we’re gonna vote for her?” was perhaps the question many voters in the Rust Belt regions were asking.
A less unlikeable Biden, a more hateable Trump
Although Biden was a vice president, he did not wield as much influence as Clinton. His policy preferences on foreign affairs were overridden by Clinton’s hawkishness on Libya and the Bin Laden raid. His image was at worst boring and milquetoast, which was an important factor affecting how majority voters saw him as a candidate.
In recent elections, whether inside or outside of the U.S., the benchmark has been “the level of disgust.” Whoever is less unpalatable to voters wins. According to RealClearPolitics, the average unfavorable rating for Clinton was 54.5% and 58.5% for Trump. They came head-to-head in that regard, but another central theme — the economy — allowed Trump to relentlessly preach his slogan: “Make America Great Again.”
Data from this year showed that Trump’s unfavorable ratings averaged at 53.7% and Biden was at 43.9%. Overshadowing Trump’s economic performance is the coronavirus pandemic, which erased Trump’s one claim to success. This time around, too, Trump was running against himself, not Biden. For this reason, we can understand why Trump wanted to hurry with lifting the lockdowns, because the economy was the only thing that could save his electoral campaign.
The changed atmosphere in the U.S. was one reason we shouldn’t apply the 2016 logic to the most recent election.
Poll accuracy has improved
Many have complained about the 2020 polls being grossly inaccurate. In reality, the polling predictions have vastly improved.
Polling serves two purposes, first to estimate the margins, second to reflect trends.
There were thousands of opinion polls prior to the U.S. election. Some were state-specific, others were nation-wide. One extreme example, however, was Wisconsin. The final vote counts in Wisconsin showed that Biden only had a 0.62% lead over Trump. But according to polls released by ABC News and the Washington Post on October 28, Biden had 57% support while Trump had 40%.
This number was untrustworthy, and not representative of what most polls published.
But casual observers would cite these polls to dismiss the credibility of polling in general. These media polls often used a sample size of fewer than 1,000, which left room for a wider margin of error. Data distribution would also skew towards urban areas with such a small sample size. Urban areas tend to lean Democratic, and if a poll was magnifying the representation in cities, the results were essentially meaningless.
Most pollsters work for profit, hence there’s an incentive for them to improve their probabilistic forecast. Since 2016, polling companies have made adjustments to move up the percentage of high school or less respondents for more accurate representation. They’ve also tried to balance their sample between urban, suburban, and rural areas.
What can we learn from “America’s most accurate pollster”?
Four years ago, Investor’s Business Daily was among the rare few who predicted Trump’s victory correctly.
IBD’s methodology consisted of 200 daily interviews, which kept its sample size well above 1,000. The IBD poll didn’t have individual state statistics, but it weighted on the demographic makeup provided by the Census Bureau.
The 2016 polls showed that Trump had at least a 10-point lead in the Midwest. This year, Biden was shown to either have a lead or tie with Trump in the same region. More importantly, IBD’s daily polls reflected a consistent trend in which Biden had a steady lead while Trump’s ratings fluctuated. The gaps between the two candidates barely closed in even in the month prior to the election, except for Pennsylvania.
Trump did fare better in the actual election than in the polls among swing states. But swing states were labeled swing states precisely because they’ve been unpredictable with a number of variables. Even then, the polling trends were largely accurate this year — that Biden was going to win.
READ NEXT: What’s Behind Taiwanese Support for Trump?
TNL Editor: Daphne K. Lee, Nicholas Haggerty (@thenewslensintl)
If you enjoyed this article and want to receive more story updates in your news feed, please be sure to follow our Facebook.
