For pollsters and campaign professionals looking at the available data, here are eight big questions we are asking:
Are the public polls correctly modeling the 2024 electorate?
Pollsters already use standard demographic data to frame their surveys by gender, age, ethnicity, and region. Some pollsters are now adding sample frame quotas or weighting to a combination of gender/education, recalled 2020 vote, or partisan affiliation. All of these are defensible, but if they’re even slightly off in their modeling of the 2024 electorate, we could see a polling miss.
I suspect that many public pollsters are under-polling a small-but-significant cluster of non-college educated white voters and shy President Trump voters. In 2016 and 2020 the public polling underestimated Trump’s support. It wouldn’t be a surprise if this happens one more time.
Can the incumbent secure victory in a heavily wrong track environment?
The mood question, sometimes called the “right direction – wrong track”, is a basic tool for estimating the electoral environment. The recent NBC survey is showing 66% wrong track and only 27% right direction. Negative voter moods are not conducive to incumbent reelections. A 66% wrong track environment is extremely difficult for an incumbent and it favors the challenger.
Exacerbating this challenge is the fact that President Biden’s approval is at an average of 40.7% and his disapproval is at an average of 56.6%. Can the VP Harris team defy political gravity? Can an incumbent win a change election? This is a very large and open question, but the data suggests that it’s not easy.
Can Harris generate an Electoral College victory with anything less than a 2-3 point popular vote win?
Most pollsters think that Harris needs to be +2 to +3 in the national popular vote in order to secure a win in the Electoral College. The logic is that Democrats have wasted votes in New York and California. These votes add to a Democrat’s popular vote margin, but they don’t help secure an Electoral College win in the swing states. It’s possible that Trump could improve his performance nationally in (1) unwinnable blue states and (2) deep red states and still come up short in the Blue Wall (PA, WI, MI), but that seems unlikely. A marginal increase in Trump support nationally would almost certainly be felt in the swing states, including the Blue Wall.
What are we to make of Trump outperforming his polling in 2016 and 2020?
The aggregated average for national polling is showing a tied race. On this day in 2020 the aggregated average had Biden 6.9 points ahead. On this day in 2016 the aggregated average had Clinton 3.2 points ahead. Trump is polling better now than he did against his previous two opponents. What does this tell us? The most basic interpretation of this data is that Trump is poised to perform better than he did in 2016 and 2020. He is also polling better among Hispanics and Black men (especially younger Black men). An alternative analysis is that public pollsters have fixed their sampling, the Trump vote isn’t being under-polled, and the race is simply too close to call.
Are the public polls under-polling Trump voters again?
This leads naturally to the next question. Are public pollsters as a group under-polling Trump supporters again? They have in the past, and explanations are that (1) Trump voters may have lower levels of social trust and avoid surveys, (2) that non-college educated whites may be less likely to complete surveys, or that (3) some Trump voters simply don’t want to share their vote preference – similar to the “shy Tory” effect in Great Britain in some elections. Time will tell.
The Dobbs Effect: Could we be missing a wave of female voters that shift the composition of the electorate in Harris’ favor?
Are women, especially younger women, turning out in larger numbers relative to men? Will a surge of female voters concerned about abortion rights shift the electorate 1-2 points more female and deliver Harris a victory? There is certainly a large gender gap in the polling, possibly the largest I have seen. Early voting appears to also be marginally more female, although it is also true that rural, red county, and registered Republicans are over performing on early voting.
If female voter intensity alters the composition of the 2024 electorate in a material way, this would almost certainly help Harris. Single women are the Democratic party’s strongest gender/marital status group. Single women are traditionally 23% of the electorate. They historically support the Democratic party by 68% to 31%. Strong overperformance here could deliver Harris a victory, assuming that Harris limits her losses among male voters.
How will Election Day undecided voters break?
Our ability to analyze undecided voters is limited by their relatively small percentage of the sample. An example is the recent TIPP national survey. With only 2% undecided and a sample size of 1411 likely voters, this survey has a raw total of only 28 undecided voters. A subsample of only 28 undecided voters is far too small to tell us anything reliable. Late undecideds are usually lower information, younger, independent voters. They traditionally break against the incumbent, and we would expect them to break against the incumbent in a wrong track environment. But the challenger was a sitting President, is well-known, and well-defined. Again, time will tell.
What are we missing?
Every election has its wrinkle, its surprise pattern. Is the late-breaking Selzer survey a harbinger of increased Democratic support from women and a challenge for Trump in the Midwest? Or is it an outlier? Is the early voting energy among Republicans and the lackluster early voting among Democrats a sign of a turnout gap favoring Republicans? Or is this simply a reversion to the new normal after a COVID election in 2020? Are relatively weak polling results for Harris in New Hampshire and Virginia telling us something? We will know soon.
Some Clues
The publicly available polling suggests that we’re looking at a close popular vote nationally. That would present a significant, but not insurmountable, challenge to Harris in the swing states and the Electoral College. Likewise, the Harris campaign is fighting the gravity that any incumbent would face in a heavily wrong track political environment.
Again, this isn’t insurmountable, but it’s a historic challenge. Further, Trump appears to be outperforming 2016 and 2020 in head-to-head matchups nationally. And in the swing states, Trump is polling better now than he was polling in 2016 and 2020. In the swing states Republicans have increased their share of registered voters, most notably in Pennsylvania, North Carolina, and Nevada. And early voting suggests stronger than anticipated turnout in rural areas and Republican counties.
All of these load in Trump’s favor and their collective weight would suggest a marginal advantage at the end of the campaign. On the other side of the equation, the Harris campaign has had a significant cash and advertising advantage. Moreover, we can assume that Harris campaign GOTV operations will be strong and focused on sharpening its historic advantage with female voters. It’s hard to imagine a more dramatic finish.
Robert Moran is a management consult, public opinion expert, and Partner at the Brunswick Group
