Why polls may be underestimating Republicans

CNN/Stylemagazine.com Newswire | 2/22/2022, 10:32 a.m.
Most polls you're looking at right now are likely underestimating Republicans' position heading into the midterm election cycle. It's not …
Voting booths on November 3, 2020 in the Brooklyn borough of New York City Mandatory Credit: David Dee Delgado/Getty Images

Originally Published: 19 FEB 22 12:17 ET

By Harry Enten, CNN

(CNN) -- Most polls you're looking at right now are likely underestimating Republicans' position heading into the midterm election cycle. It's not that the polls are "wrong." Rather, it's that most polls at this point are asking all registered voters who they're going to vote for in November, when it's likely only a distinct subset of voters who will cast a ballot.

The voters who will actually turn out for the fall election are likely going to be disproportionately Republican based on current polling data and history.

Take a gander at our CNN/SSRS poll from last week. It featured a rather close race when all registered voters were asked who they would vote for if the election for Congress was held now. The generic Republican candidate garnered 44% to the generic Democratic candidate's 43% -- a result well within the margin of error. This poll is basically in line with the average of all polls, which has generally featured a low single digit Republican advantage on the generic ballot.

If that held through November, Republicans would be small favorites to take back the House. Democrats winning the House popular vote and holding on to control given how redistricting has gone (relatively well for Democrats) would be quite plausible.

The election is still well over half a year away and we obviously don't know who will turn out in November -- hence, why we are not whittling down the electorate just yet.

Dig into the poll a little more, and you can see where Democrats' problem lies, though. Our CNN poll asked respondents how enthusiastic they were about voting in this year's election "extremely, very, somewhat, not too or not enthusiastic at all."

Among those who said they were extremely enthusiastic (24%), Republicans held a 59% to 39% lead over Democrats on the generic congressional ballot.

If we expand our universe to include those who were very enthusiastic as well (43% of all voters) of all voters, Republicans were ahead 55% to 42%.

No matter how you measure enthusiasm, the voters who are most enthusiastic about voting this year lean Republican. A similar signal was seen in an NBC News poll last month in which Republicans registered far more interest in the upcoming midterms than Democrats, so this poll is no outlier.

Of course, a vote cast by an enthusiastic voter is worth the same as a vote cast by an unenthusiastic voter.

The fact is, though, that enthusiastic voters have been more likely to say they'll cast a ballot than unenthusiastic voters, at least in CNN polling historically. The relationship isn't perfect, but it's clear one exists.

In the final poll before the last midterm in 2018, voters were asked the enthusiasm question as well as how likely they were to vote on a scale from definitely not to definitely will/already have. Among those who said they were extremely enthusiastic to vote, 96% said they were definitely going to vote or had already voted.

A relatively low 73% of those who were not extremely likely to vote said they were definitely voting or had already voted. This dropped to 64% among those who were not extremely or very enthusiastic.

Of course, the exact levels of enthusiasm between the parties could easily differ between now and Election Day.

It's also true that the party that has been more enthusiastic about voting at this point in the campaign has been more enthusiastic to vote in the final CNN poll asking about enthusiasm before the midterm. Republicans were more enthusiastic in 2010 and 2014, while Democrats were more enthusiastic in 2018. The party that was more enthusiastic at this point went on to major gains in the November elections.

Indeed, history would need to be upended for there not to be a significant difference between the voting patterns between registered voters and midterm voters who turn out to vote.

In a previous story, I examined the difference in party identification between all registered voters and those who actually voted in midterm elections since 1978 in post-election surveys from the ANES and CES. The years in which there was a Republican president (like in 2018), there was a minimal difference in the party identification between midterm voters and voters overall.

In years in which there was a Democratic president (like right now), Republicans on average made up 5 points more of the midterm voter pool than registered voter pool. In none of the years with a Democratic president did Republican turnout advantage shrink below 3 points.

Averaging across different national pre-election generic ballot polls in both 2010 and 2014, we see how this impacted the result. The Republican lead widened by about 5.6 points and 4.8 points on average in 2010 and 2014 respectively. In other words, it was very close to what we'd expect given what the ANES and CES signaled.

Could 2022 be different? Past isn't always prologue, but history does seem more likely than not to repeat itself. Beyond the polling we have nationally right now, the polling ahead of the Virginia gubernatorial election a few months ago painted a similar picture.

Fox and the Washington Post, two national pollsters, who presented registered and likely voter results in that race found Republican Glenn Youngkin doing 5 points better on average with likely than registered voters. He would go on to win the election.

Anyway you look at it, Democrats are probably in considerably worse shape at this point than polls of registered voters on the generic ballot might lead you to believe.