Log In

Reset Password

Have political polls outlived their usefulness?

After two embarrassing whiffs in consecutive presidential elections, polling has come under increasing criticism and skepticism.

A recent survey showed that public opinion polls in the 2020 presidential election had errors of “unusual magnitude.” That’s polite talk for saying that many of them were way off the mark.

The conclusion was reached by the American Association for Public Opinion Research, which studied more than 2,800 polls, including 529 national presidential polls and 1,572 state presidential polls.

Its study found that the polls overstated President Joe Biden’s margin of victory by 3.9% in national popular vote polls and 4.3% in the statewide polls. The study also found that the polls understated former President Donald Trump’s support in nearly every state by an average of 3.3.

The polls did a better job of estimating support for Biden, which showed that he polled 1% higher than his actual vote.

A much-publicized example of how far off the mark polls were for last November’s election was in Florida where polls predicted a 3%-4% Biden victory; instead, Trump won by 1.2% of the popular vote and, in the process, won Florida’s 29 electoral votes.

The only major local poll, conducted by Muhlenberg College/The Morning Call, showed a five-point lead (49%-44%) for Biden in Pennsylvania going into the week before the election.

The other well-known statewide poll - the Franklin & Marshall College Poll - had Biden winning by 6%.

Biden won Pennsylvania by 1.2% (50%-48.8%) to take Pennsylvania’s 20 Electoral College votes. The final tally had Biden besting Trump by 80,555 votes.

Among the five counties in the Times News region, three went for Biden and two for Trump.

Biden did best in Lehigh (53.2%-45.6%), followed by Monroe (52.6%-46.2%) and Northampton (49.8%-49.1%).

Trump overwhelmingly captured Schuylkill (69.2%-29.4%) and Carbon (6.4%-33.3%).

Other nearby northeastern Pennsylvania results showed a split with Biden winning Lackawanna County (53.7%-45.3%) and Trump taking Luzerne County (56.7%-42.3%).

An interesting note: Biden was the first Democrat to win the White House without carrying Luzerne County since Harry Truman won in 1948.

Biden won in 12 of Pennsylvania’s 67 counties, while Trump won the other 55.

The AAPOR said a similar study was undertaken following the 2016 election in which Trump upset Democrat Hillary Clinton. Although national polls accurately predicted that Clinton would win the popular vote (which she did by 2.87 million), most were inaccurate in determining state winners, and, as a result, concluded that Clinton would win the Electoral College count, as well. Trump had 304 Electoral College votes to Clinton’s 227.

So why were these polls off the mark twice in a row, and what does this mean for the future of polling? While the task force members ruled out failing to account for level of education, a problem in the 2016 polling, they were not able to reach a determination on what caused the discrepancies.

“Identifying conclusively why polls overstated the Democratic-Republican margin relative to the certified vote appears to be impossible with the available data,” the report states.

Of course, we have hunches. One theory is that some Trump supporters were not truthful to those polling, but Josh Clinton, chair of the study, said this theory was not supported by the data.

In 2016, there were a lot of late-deciding voters who broke for Trump, but four years later most voters had made up their minds well before Election Day, especially considering how popular mail-in voting had been.

Clinton believes that the back-to-back inaccuracies will not necessarily doom polling, although there is no question that it will be looked upon more skeptically.

If you’re looking around for an understatement when it comes to polling, the comment by Columbia University statistician Andrew Gelman is a good place to start: “Polling is an inexact science.”

So what’s to be done? Stop polling? Well you know that will never happen, not when there is money to be made and not when the public loves to hate polls.

I am at the point where I am wondering whether we just should ignore polls as untrustworthy.

By Bruce Frassinelli | tneditor@tnonline.com

The foregoing opinions do not necessarily reflect the views of the Editorial Board or Times News LLC.