It’s partly because pollsters are changing whom they ask.
It’s partly because pollsters are changing whom they ask.
J im McLaughlin, the president of the McLaughlin & Associates poll, is fed up.
He told Newsmax on October 10 that major polling firms are intentionally “trying to drive down Republican support” in the November midterm elections. He claimed that a vast oversampling of Democrats in mainstream election surveys is intentional. “They’re trying to drive down Republican support both in the electorate, because you’re not going to show up if your candidates going to lose, and they’re trying to dry up money,” he stated. “Just like we get this biased mainstream media news coverage, their polls are biased.”
After studying polls for decades, I’ve come to believe that some polls have biases, especially in the way questions are worded. But in general they remain a useful instrument so long as you take into account the all-important “margin of error,” which can sometimes range above 4 percent.
But pollsters know they have real problems. Nate Cohn, the chief political analyst at the New York Times, admitted earlier this month that only 0.4 of attempted phone calls for the latest Times poll yielded a completed interview. “If you were employed as one of our interviewers at a call center, you would have to dial numbers for two hours to get a single completed interview,” he said.
It’s not only that the growth of cellphones make it harder to reach voters. It’s also that many people now are suspicious of pollsters and won’t tell strangers what they really think. In 2020, white Democrats were 20 percent likelier to respond to New York Times polls than were white Republicans.
But there is one persistent anomaly in the polls we all see that continues. Nate Silver of the analytics website FiveThirtyEight summed it up this way:
Polls of so-called likely voters are almost always more favorable to Republicans than those that survey the broader sample of all registered voters or all American adults. Likely voter polls also tend to provide more reliable predictions of election results, especially in midterm years. Whereas polls of all registered voters or all adults usually overstate the performance of Democratic candidates, polls of likely voters have had almost no long-term bias. . . . Because polls of likely voters are nearly unbiased, that implies polls of registered voters would usually have a Democratic bias.
If you examine the polling results included on the website RealClearPolitics, you’ll find an interesting and probably not accidental pattern that emerges on the most basic issue in the 2022 election. It’s called “the generic-ballot test” — i.e., will people prefer the Republican candidate for Congress or the Democratic one?
From the beginning of 2022 until late September, there have been 172 surveys that had generic-ballot test results reported on the site. Of those, 127 were from surveys of registered voters. The remaining 45 were from surveys of likely voters.
Surveys of likely voters are more relevant to campaigns and election outcomes. They are better predictors of election outcomes because they include people who routinely vote, follow the news, and have opinions. At this point in the process, no one should be paying any attention at all to surveys of registered voters. Likely voters are what matters now.
Among the 127 surveys of registered voters, the Democrats led the generic-ballot test in 61, the Republicans led in 50, and there were 16 ties. The average result of these surveys of registered voters was almost even between the parties. The 45 reported surveys of likely voters tell a very different story. In these surveys, the Republicans led the generic-ballot test in 42 of them. The Democrats led in just three. The average result for the ballot test in these surveys of likely voters was a clear Republican lead.
In the 21 polls released after October 1, that pattern has pretty much held. The five surveys of registered voters at RealClearPolitics show the two parties tied in the generic-ballot test. The 16 surveys of likely voters show the Republicans with an average advantage of three points.
Given the durable Republican advantage among likely voters on the generic-ballot test, it looks as if Republicans will win the national House vote and take control of that chamber. But if you were just looking at polls of registered voters, you would come to a different conclusion.
“It seems likely that surveys are systematically undercounting Republican votes” again, Mike McKenna, a columnist for the Washington Times and a former deputy director of President Trump’s Office of Legislative Affairs, told me. “In the wake of the president’s recent remarks, where he railed against Republicans as semi-fascists, it is very unlikely that each and every Republican feels comfortable self-identifying. There’s no telling the numerical effect of this undercount until we see the actual results in early November.”
The polls are now more reflective of reality — and becoming more Republican — between now and Election Day. Partly because inattentive voters are only now tuning in to the campaign, and most of them see basic issues such as inflation and crime as more important to their vote than, say, abortion and climate change.
Republicans hold a clear advantage on the key basic issues this year. For example, a new ABC News/Ipsos poll released on Sunday reports: “Inflation sees one of the larger gulfs — with 36% of Americans trusting the GOP and 21% trusting Democrats.” A similar gap exists around crime, with 35% Americans trusting Republicans and 22% trusting Democrats.”
So one reason for the GOP resurgence in polls is that the surveys have finally caught up by surveying the most likely electorate, and in most elections the likely-voter sample leans more Republican than do samples of registered voters or all adults.
No wonder so many Democrats want to push early or mail-in voting or are ducking debates this year. They want to lock in their base voters and not have a truly spirited debate on the issues that voters tell pollsters they care most about.