WASHINGTON —
I live for the polls. I have one piece of advice for the coming weeks. “Please stop.”
It’s not that the polls are broken. Despite many criticisms, the survey continues to provide surprisingly accurate results and important insights into what different voter groups are thinking.
But most of the people who obsessively click on the latest latest survey aren’t looking for that. What they want to know — and if you’re reading this, you probably want to know — is the one answer that a close poll can’t provide: Who will win?
Democratic Party’s anxiety goes out of control
For the past week, it seems like Democrats have been making a fuss from country to country over Vice President Kamala Harris’ chances in the presidential race.
This shift in mood says more about public sentiment than opinion polls. On average, the polls have moved little, but in the final stages Democrats seem primed to be just as anxious as Republicans are tilted toward overconfidence.
Newsletter
You’re reading the LA Times Politics newsletter
Anita Chhabria and David Lauter bring insight into law, politics, and policy in California and beyond. Delivered to your inbox three times a week.
Please enter your email address
Please sign up
You may receive promotional content from the Los Angeles Times.
Check reality. Over the past 14 presidential elections, national polls have averaged a difference of 2.2 percentage points. (Harry Enten, then of the polling firm FiveThirtyEight.com, first calculated that average in 2016.)
That’s not bad. The best professional basketball players miss free throws 10% of the time. Olympic archers also miss the center ring at a similar rate. In a country as diverse as the United States, it’s pretty surprising to see margins of just a few percentage points in repeated elections, and there’s no evidence that polls are becoming less accurate. For example, the average number of electoral losses in 1996 and 2000 was higher than in any election since then.
The problem, of course, is that Harris and former President Trump are separated by narrow margins in key states. The average voting error is much larger than the superiority or inferiority of either candidate. And this situation is unlikely to change in the final two-and-a-half weeks, given that so few voters have yet to make a decision.
What the polls tell us: Decreasing racial polarization
In recent years, Democrats have gained support among white voters with college degrees, while Republicans have gained support among black and Latino voters, especially those without college degrees and those who are religiously conservative.
This means that American politics is somewhat less polarized along racial and ethnic lines, which is a good thing. But this is a difficult pill to swallow for many on the left, challenging the belief that voters of color naturally lean their way.
The key question is how big these trends have become, and how much they have benefited from each other. Do the shifts cancel each other out, or has one party gained a significant advantage?
Because Harris is of mixed black and Asian descent, some may be surprised that her strength among white voters is responsible for her clear leads in the northern battleground states of Pennsylvania, Wisconsin, and Michigan. There is.
Conversely, in the states where she is lagging, particularly Arizona, weakness among white voters is a major issue.
A New York Times/Siena College poll shows her trailing in Arizona and finds her lagging among white independents and moderate Republicans. Ta. As Kyle Kondik of the University of Virginia Center for Politics pointed out in an analysis this week, polls showing her lead in the state, as well as a recent Wall Street Journal study, show that she has a strong lead on just these groups. I found out that she was getting better grades.
By contrast, numerous surveys show that Mr. Trump enjoys historically high support among black voters, especially men.
For example, the latest Economist/YouGov poll shows that Harris still has a strong majority of black voters, 76% to 11%, with 8% uncertain. However, this is significantly lower than the 92% that President Biden received in 2020.
The question mark for Mr. Trump is whether the black supporters who show up in the polls will show up in the actual vote. Adam Carlson, a former Democratic pollster who compiled data from multiple surveys, said 2020 polls vastly overestimated Trump’s support among black voters. That might happen again this year.
Among Latino voters, Harris is currently doing about the same as Biden in 2020, 60% to 35% in the YouGov poll. That’s not as much as Democrats would like, but she eliminated a budget deficit that was much larger in the spring. However, many studies show that Latinos have a higher proportion of undecided voters than other groups.
Polling is less popular among Asian American voters, but a September study by the University of Chicago’s AAPI Data and NORC found Harris leading Trump 66% to 28%. It was a significant improvement for Democrats since Biden dropped out of the race.
and a big gender difference
In the AAPI survey, gender, rather than ethnicity, appears to drive much of Harris’ support, with Harris receiving greater support than Biden among Asian American women.
This is consistent with other studies that have found very large gender disparities this year. The gap could match or even exceed the gap in 2016, when Hillary Clinton’s campaign highlighted her status as the first woman to win a major party’s presidential nomination.
In the same year, there was a 26 point difference between men and women. According to a detailed analysis by Pew Research Center, Mr. Clinton won with women by 15 points (54% to 39%), while Mr. Trump won with men by 11 points (41% to 52%).
Pew found that the gender gap narrowed in 2020. Now I’m back. A recent NPR/Marist poll showed Harris winning by 18 points among women and Trump winning by 16 points among men, a 34-point margin. This has been on the high side in recent polls, but polls this fall have averaged 22 points. Carlson found the difference.
One notable part of the gender gap has to do with young voters.
A study by the New York Times and Siena College drew a lot of attention when it found that while men under 30 are leaning toward Trump, young women are leaning sharply toward Harris.
But as Vanderbilt University political scientist John Sides recently pointed out, other high-quality surveys show much smaller gender differences among young people, with both young men and young women supporting Harris. shown.
Who is right? Exit polls conducted by major television networks and the Associated Press may provide some preliminary clues, but we won’t know for sure until researchers can fully examine the data, several months after vote counting. I don’t know.
underestimate trump
“If you’re reading this right now, you’re probably thinking, ‘But don’t polls always underestimate the Republican vote?’
no.
In both 2000 and 2012, polls underestimated Democrats.
Polls underestimated President Trump’s vote in 2016. The pollster then investigated what went wrong and set about fixing it.
they were not successful. The investigation missed many of his supporters in 2020 as well.
So pollsters tried new methods to get the right results. Maybe I’ll be successful this time. Perhaps they will underestimate Trump supporters again.
Or maybe you’ve over-corrected and your results are skewed in a different direction.
Errors are expected. I don’t know which way it will go.
I’m sorry, but even an “internal” poll doesn’t tell us.
Don’t think you’ll get better results just by clicking on a headline about a leaked internal investigation.
Yes, campaigns have more data than pollsters.
But there is no guarantee that they know very well who will win. For example, in 2012, Mitt Romney’s campaign polls showed him in the lead. He was unprepared when President Obama hit him. Romney is not alone in this. Campaign research is subject to the same limitations that public opinion polls face.
In fact, getting accurate results from horse races is not a top priority for election pollsters. Campaigns primarily use polls to test messages, trying to see which words most effectively sway groups of voters. It requires consistency from one study to the next, not absolute top-line accuracy.
And campaigns are selectively leaked. If you take 20 polls in a close race, you’re almost guaranteed to see some candidates trailing, and some candidates trailing. This is how probability works. Experience has shown that anything that supports the narrative they want to cultivate will leak.
That’s why, as Nate Silver recently detailed, analysts who aggregate polling averages typically discount internal polling by several points.
This year, both camps seem intent on pushing the idea that Harris is the underdog. The vice president has consistently embraced this label in his speeches, perhaps hoping to galvanize Democratic voters. Of course, Trump can’t stand being anything but top dog.
So it’s no surprise that both parties leaked polls showing Trump with a slight lead. It’s a shame that they don’t even distribute tranquilizers to anxious Democrats.
Who will show up?
One of the major sources of error in campaign polls, and the main reason why surveys differ from each other, is that polling organizations are trying to measure a population that doesn’t yet exist: people who will vote in an election that hasn’t yet taken place. That’s what it means.
The nonpartisan polling team that conducts the NBC News poll recently conducted an experiment to show the impact of pollster assumptions on turnout.
Their latest poll shows Harris and Trump tied nationally, 48% to 48%, using a preset turnout model.
But with a few small adjustments to the turnout model, assuming women make up 53% of voters instead of 52%, or white voters making up 70% instead of 72%, Harris He led with 49% and 46%. Similarly, small adjustments in the other direction would give Trump an advantage.
Making good estimates is especially complicated when turnout is high. In low-turnout elections, nearly everyone who votes is a regular voter with a consistent track record. Elections with high turnout attract many new voters.
In 2020, when voter turnout hit an all-time high, about 1 in 4 voters hadn’t voted four years ago, according to Pew data. As a measure of voter enthusiasm this year, turnout is expected to be even higher going forward, despite the uncertainty surrounding the survey.
So, are you worried about the election? Go for a long walk. Look at the baseball playoffs. Or, if you feel committed, volunteer for your favorite candidate. Knock on the door or make a phone call.
But if you want to stay sane, stop scrolling catastrophically through polls. They won’t tell you what you want to know.
What else should I read?
This week’s poll: Why do Asian immigrants come to the U.S. and how do they view life here?
Saturday’s article read, “Do they really believe that? America’s political turmoil has psychological causes.”
LA Times Special: Donald Trump threatens revenge on California. Should we believe him?
—
Was this newsletter forwarded to you? Sign up here to get it delivered to your inbox.
