One of the three best pollsters in the country, Marquette Law School Poll (3.0),1 has a new poll out that might help explain why we are seeing Trump performing better than we would expect considering, well, everything – but particularly in light of how well Democrats have been performing in special and off-year elections everywhere since 2020.2
Today’s release has Biden beating Trump 52-48 nationally among likely voters. We have seen very few polls up to now that have reported results from likely voters. There is a good reason for this. It is early April and the election is in November, which means earlier polls were too far removed from the election to model a reliable likely voter electorate. It is arguable whether it is no longer too early to do this, but with a pollster like Marquette doing so we should take it seriously. However, Marquette has been surveying likely voters all along, so the fact that they are doing it now does not itself mean something’s changed in the electorate. But now that both parties have candidates with enough delegates to win their respective nominations, it is probably time to turn our attention to the general election.3 Perhaps regular ol’ voters are doing the same.
Here are some things to note from the survey:
This is not a pollster that has had Biden winning while other polls had him losing. Marquette’s previous surveys in February and November had Trump winning. In fact, Trump was up 52-48 in this poll in February.
Biden wins a much larger share of Democrats (93%) than Trump gets from Republicans (88%). Independents go for Biden with 53%. This is changed from the February poll. Then, Biden got 91% of Democrats and Trump won 93% of Republicans. While Biden’s change is not significant, Trump’s definitely is. Independents went for Trump with 54% in February.
Marquette also polled a five-way race with Kennedy, West, and Stein. In this scenario, Trump leads Biden 41-38 with Kennedy getting 14%, West 5%, and Stein 2%. This is almost identical to the February five-way results.
Kennedy takes 12% from Republicans and Democrats, so he is drawing equally from both major parties. This is the same for Democrats from the February poll, but down three points for Republicans. Stein gained some support among Democrats and independents, while West lost support. Overall, the results hardly changed.
I will skip discussion of the five-way race right now because I think the minor party candidates are just noise at the moment. It will only matter in states they get on the ballot, and history shows us that minor party candidates get much less support than they poll once the election comes. Of course, a small share of the vote could be determinative in a close election, but we can discuss that later. I want to focus on what’s going on in the two-candidate race.
It is interesting to see that Biden performs better among likely voters than others, but it is not surprising. Actual election results over the past four years suggest Biden and Democrats generally are doing better among actual voters than polls suggest. Marquette did not just survey likely voters, however. They also surveyed registered voters who were “not certain to vote”4 and adults who are not registered voters. Among the former category, Trump beats Biden 56-44. Among the latter, he wins 64-36!
Even if we exclude the garbage polling from outfits like Harris and Redfield & Wilton Strategies, which are flooding the airwaves with suspiciously good polling (that lack methodological transparency) for Republicans – as they did in 2020 and 2022 – most polling favoring Trump thus far has been of registered voters or without regard to voting status (usually called “all adults”). Marquette’s poll suggests that with likely voter screens, perhaps the result in these polls might flip to Biden. The results are still close. We’ve seen no consistent big leads for either candidate yet. This Marquette poll may be the beginning of increasingly better results for Biden in the national polling or it could be an outlier. As with all things in polling, time will tell.
538 has a new rating system. It is no longer using letter grades, which I think is a mistake since everyone understands what letter grades mean. It is using a number system in which 3.0 is the highest rating and 0.0 is the lowest. Currently, McLaughlin & Associates has the lowest rating at 0.5. Marquette Law is one of only three pollsters rated at 3.0. The others are The New York Times/Siena College and ABC News/The Washington Post.
Let’s keep in mind the obligatory caveat: any one poll is just a snapshot in time and it may be an outlier.
I think we can still glean some interesting trends for the fall by digging into the primary results, as you can see from my last post about the Wisconsin primary.
This category would appear to include some voters that other pollsters might consider likely voters. It depends on how they decide who is a likely voter. Some, probably most, would not use “certain to vote” exclusively for likely voters until very close to the election. Pollsters may ask respondents if they definitely will vote, probably will vote, probably will not vote, and definitely will not vote – or some variation of these. In many cases, pollsters will include the “definitely will” and “probably will” voters in their universe of likely voters.