Republicans believe that the polls favor them. Sometimes, this belief has led them to misread large political races. For example, in 2012, many Republicans believed that Mitt Romney, the GOP’s presidential nominee, was going to defeat Barack Obama. A lot was said about “unskewing the polls.” Obama won easily over Romney.
In the years that Donald Trump has been in national politics, there have been many instances of polling bias or even mistakes. Many times, pollsters were unable to gauge Trump’s popularity, particularly in state-level polls. The 2016 Trump vs. Hillary Clinton race saw Clinton win in three critical states: Wisconsin, Michigan and Pennsylvania. The RealClearPolitics average polls for Election Day in Wisconsin showed Clinton winning with 6.5 points. Trump actually won by 0.7 percentage points. The polls in Michigan showed Clinton ahead by 3.6 percentage points. Trump won by just 0.3 points. In Pennsylvania, Clinton was ahead by 2.1 percentage points. Trump won by 0.7 point.
The Trump-Joe Biden 2020 race was a close one. Even though the polls had the winner correctly, sometimes they were off by quite a bit. For example, Joe Biden won Wisconsin by 6.7 points according to the RealClearPolitics average polls. Biden actually won by 0.7 points. The poll by Washington Post-ABC News, which was published in October, showed Biden leading Trump 17 points. It was a scandalous result that pollsters must not forget.
What does this all mean for us today? One month is left for the midterm elections, and polls are all around. Is it the same old problem?
We don’t know. However, there are some reasons to suspect that problems persist. RealClearPolitics is now comparing the midterm polls today to the known errors from polls in 2016, 2018, and 2020. The election results could change if the same mistakes occur again.
RealClearPolitics looked at the current polling average for each contested state in midterm elections. It then averaged the polling errors from 2016, 2018, 2020 — two presidential election and one midterm — all with Trump. It then took the polling error and applied it to today’s poll to see how things looked.
Let’s start with the Pennsylvania Senate race. The average polls show John Fetterman, the Democratic candidate, ahead of Republican Mehmet Oz by 3.7 points. Next, take a look at the Pennsylvania polls in the past. The 2016 polls had Trump ahead by 8.6 points. The 2018 midterm polls had Sen. Bob Casey (D–PA) ahead by 16 points. And the 2020 polls had Biden ahead by 6.5 points. Comparing the final election results of those years to this point, the polls underestimated the Republican performance by 5.9 percentage.
Is Oz really 3.7 points behind Fetterman, or is that just me? Maybe. Oz could be up to 2.2 points ahead if the polls are wrongly reading the electorate today, as they have in past elections. This is just at the moment, and not on Election Day. The polls may be wrong about the horse race.
Others races may be closer than you think. New Hampshire’s Democratic Senator Maggie Hassan is ahead of Republican challenger Don Bolduc, by 5.3 points according to the RealClearPolitics Average. Factoring in poll errors from previous years, Hassan’s lead rises to 0.9 points. This is a much more competitive race. The RealClearPolitics average in North Carolina shows Ted Budd, a Republican, leading Cheri Beasley, Democrat by 1.5 points in Senate race. Budd leads by 6 points if you take into account the old mistakes.
These are huge differences. There is however one race that is closely monitored where polls seem to be on the money, at least according to this measure. Georgia’s Democratic Senator Raphael Warnock leads Republican Herschelwalker by 3.8 points. The historical polls for 2016, 2018, and 2020 show no significant differences from the final results. This means that Warnock’s lead of 3.8 points over Walker is real. This is only a snapshot of the race at this point, 28 days before Election Day. But Warnock’s lead remains solid.
In one race, the 2016-2018 and 2020 polls underestimated Democratic performance. Today, Nevada’s Republican Adam Laxalt leads Democratic Senator Catherine Cortez–Masto by 2.1 percentage points. However, past polls show that this might be an exaggeration and that Laxalt actually leads by 1.1 points. It’s still a close race, but he is still ahead.
We do know that state polls were wrong in the past. Serios are trying to figure out how this happened and what it means for today’s midterm elections. Right now, the picture is better for Republicans than what the polls indicate.
