NYT (“Why the Polling in Michigan Was Off the Mark“):
Much of the pre-election polling in Michigan’s Democratic Senate primary showed Dr. Abdul El-Sayed, a progressive candidate, with a commanding lead over the more moderate Haley Stevens. Come election night, however, Dr. El-Sayed eked out a narrow victory, in a race so tight it was not called until Wednesday morning.
So, why did the polling miss by such a wide margin?
Why, indeed?
Conducting surveys in partisan primaries has long presented a challenge to pollsters, as it requires them to make assumptions about who will actually turn out to vote. Without mandatory party registration in Michigan, pollsters must guess who will show up on Election Day.
Many surveys assumed political independents would make up 30 to 40 percent of the electorate. These polls showed Dr. El-Sayed winning independents by a wide margin. While there is little authoritative detail available on the actual electorate, that assumption could have contributed to the challenge in polling the race, as voters who decline to identify with a political party tend to vote less frequently.
But it’s always the case that predicting who will actually turn up to vote is challenging. Pollsters apply “likely voter” screens, usually based on past behavior. But anomalous races—such as one with an unusually charismatic but wildly polarizing candidate—will defy the pattern.
This, though, strikes me as the issue:
Another challenge was a lack of high-quality polling. In The New York Times’s poll tracker, “select pollsters” are those who have met qualifications to be considered generally more accurate and trustworthy. In the final month before the primary, just one poll was conducted by such a group.
Much of the remaining polling was conducted by groups that were — either directly or indirectly — affiliated with Dr. El-Sayed’s campaign. Historically, campaigns and their allies have had an incentive to produce polls showing their candidate doing well to try to build a narrative about the race, even if those numbers do not always align with reality.
But here’s the thing: If there is only one “high quality” poll, news organizations should know that and report accordingly. Rather obviously, they should take candidate-financed polls with a grain of salt, if not dismiss them altogether.
Still, polling is expensive and there’s considerably less return on investment in statewide races as there is in national ones. In the latter, tons of media organizations commission reputable polls to generate news fodder and drive their own opinion pieces.
Primaries, let alone in a midterm cycle, are low-turnout affairs. For the 2024 presidential election, 2.8 million-plus Michiganders voted for President Trump and another 2.7 million-plus voted for Vice President Harris. Heck, 1.1 million voted in the Republican presidential primary, and 770,000 voted in the Democratic presidential primary, both of which were all but uncontested. By contrast, got 742,000 votes to Stevens’ 727,000 in Tuesday’s election.
Regardless, the results in Michigan were off considerably more than is usual, even given those challenges. Given the comparative enthusiasm El-Sayed generated, I suspect more voters than anticipated changed their mind at the last minute and voted for the “safe” choice in Haley.
By contrast, [El-Sayed] got 742,000 votes to Stevens’ 727,000 in Tuesday’s election.
And Mallory McMorrow got 61,000 votes (4%), despite having withdrawn from the race. The fact that her name was still on the primary ballot may have been enormously consequential.
@Eusebio Good point. The media reporting focused on the top two finishers, but there were indeed others on the ballot drawing significant votes.
steve222
Will be interesting to see the age composition of those who voted. My impression is that younger people get excited about candidates and talk a lot about voting for candidates then dont go to actually vote.
Steve
1
Charley in Cleveland
If El-Sayed and Mamdani had last names of Smith and Jones, are we having the same reaction from Wingnuttia?
And aren’t pollsters like meteorologists in that they can take shelter in the fact no one expects them to be 100% accurate?
Eusebio
The same reaction from Wingnuttia? Probably not. Although Mamdani does sound sort of Italian, so that one may be easier for some to take. I had a theory early in trump’s first term… that he kept Preet Bharara on as US attorney for Southern District of New York because he thought he was an Italian guy named Pete. That lasted only a few weeks.
1
Scott F.
But here’s the thing: If there is only one “high quality” poll, news organizations should know that and report accordingly. Rather obviously, they should take candidate-financed polls with a grain of salt, if not dismiss them altogether.
How then could news organizations do the breathless horse race coverage with which they are most comfortable and that is their bread and butter?
4
Kylopod
What I think people were surprised about was not that there was a polling error per se–that’s so common these days it’s practically taken for granted (and leads to all those hyperbolic claims about how polls are inherently garbage and should be ignored), and elections that are outside the regular schedule (including primaries as well as some special elections) are generally difficult to poll accurately. The real surprise was that it was the establishment candidate who outperformed the polls this time.
One of the factors that often distorts polls and causes certain candidates to be underestimated is a failure to pick up voter energy and enthusiasm; that’s why insurgents usually benefit the most in these situations. We’ve already seen several concerted efforts by establishment and pro-Israel groups to defeat these types of candidates, but that only seems to invigorate supporters of said candidates more. Maybe those groups have been getting more effective, and in that light it’s worth mentioning something that hasn’t gotten as much attention, the defeat of Cori Bush the same night–which was hardly unexpected, but she did not outperform the polls and may have even underperformed a little.
I live in the district where Avila Chevalia beat the incumbent Adriano Espiallat in June, and I was bombarded by ads from both campaigns in the weeks leading up to the election. I found the pro-Espiallat ads unfocused, and it did seem that he waited too long before taking the threat of her winning seriously. I don’t think that’s happening anymore.
As a side note, I think age is a factor that is getting less attention than it should. Espiallat is 71, Avila Chevalia 32. Janet Mills is 78 and would have become the oldest freshman (freshwoman?) Senator in history if she’d won. El-Sayed and Stevens are both in their early 40s. While this factor is related to the rise of left-wingers (as younger voters are likeliest to be attracted to such candidates), it probably had its own impact on some voters, separate from ideology.
MarkedMan
One thing I haven’t seen much discussion of is how many people wanted to vote for El-Sayed but went with Stevens because of electability concerns and so will be very happy to vote for him in the general, as compared to the number that had actual concerns about El-Sayed and so might stay home or vote R, and as compared to those who voted for Stevens because they felt she was wonderful but will drag themselves out of bed to vote for El-Sayed
2
Daryl
Vote Blue, no matter who.
MAGA held Fauci in contempt this morning which will lead directly to his indictment by the MAGADOJ.
Fuck these people.
2
Rick DeMent
My Wife and I were McMorrow supporter. When she dropped out of the race, we had a long discussion because my wife is a delegate for our precinct. She is also close to a lot of Jewish Dems who were staunchly anti-Abdul.
She also went to the nominating convention and said the place was filled with Abdul supporters (mostly younger people) who were getting borderline out of control and being aggressive. After that experience all of her older Jewish political friends were shocked over the strident attitude of the Abdul supporters. Most of them were also McMorrow supporters.
One of her Jewish friends was a bit more opened minded about Abdul but didn’t want to let on to her other Jewish friends (all of the people I’m talking about are in the local Democratic club. My wife reminded them that we have been witching about why more young people don’t get into politics that this is exactly what they said that wanted. But they wanted young people who saw the world the same way they do.
In the end the wife and I voted for Abdul. I have talked to him a number of times since 2016 (Stevens too). So both of the candidates were very familiar to us. I could have easily voted for any of them but in the end I went with AES because I felt he had a fresh perspective and I am bewildered by the timidness of the old guard.
Let’s hope the kids come out and vote. I think they will this time. I’m also glad McMorrow and Stevens got right behind AEL.
Kathy
I wonder how much is AI used now to design poll questions, or to draw up call lists, or other matters related to polling practices.
