Democratic victories in Michigan, Pennsylvania, and Wisconsin rest on the cities of Detroit, Philadelphia, and Milwaukee, respectively. This is also the order of Democratic strength in both the states and the cities. In all honesty, I hate being a crosstab snob, and I do think un-skewing a poll by simply assuming that a statewide electorate should be more weighted to one party or another gets you into delusion and wishful thinking. But there is also an obligation when you have extremely recent voting pattern data on the one hand, and polling on the other, you need to really ask what is more likely?
Taking a close look is particularly important here because being wrong in the cities will make your poll wrong statewide, because the cities are just that important. And unlike demographic cross tabs where you can’t know for certain how women voted, you do know for certain how a city voted.
Let’s begin comparing two of these cities and then we will turn to the third. In one city, let’s call it City A, Biden got 95% of the two-party vote in 2020. Democrats also cast 93.3% of the ballots cast in City A’s 2024 Presidential primary. In City B Biden got 78.83% of the vote in the 2020 election, and Democrats cast 77.8% of the votes in this year’s Presidential primary. In one of these cities Biden leads in the two-party ballot test in the NYT polling, 81% to 11%. In the other he leads 66%to 24%. If you guessed which polling went with which election results, you were likely wrong. City A is Detroit, where Biden leads in the NYT poll 66% to 24% despite winning 95-5 in 2020. Meanwhile, Biden has a larger lead than his 2020 result in City B, which is Milwaukee.
We compare these two cities because the lack of party registration means that anyone who wishes to switch parties and sides can begin to do so with no switching costs. The two cities also had somewhat similar turnouts in the recent primaries. Detroit cast about 25% of its 2020 Presidential turnout in the primary; Milwaukee reached 34%. In the actual voting we saw extremely similar results to the 2020 general with simply a smaller electorate. Milwaukee did see slightly more interesting down ballot race driving turnout, but nothing dramatic. But in the NYT poll we see similar cities, similar election contests, and yet radically different election results. Trump got no spike in votes in either city in the primary to suggest anything is changing, and yet the poll has him on track to receive almost 5 times the % of the vote in Detroit that he got last time and similarly five times the total vote we would project out by looking at the 25% of the voters who just voted in February. The NYT poll’s change in Detroit alone would have been almost enough to flip Michigan to Trump (and given similar communities statewide would have done so easily) But which result is more likely, the poll or November voting tracking prior patterns?
This brings us to our third and final city, Philadelphia. As does all of Pennsylvania, Philadelphia has party registration. Here again though we have recent election results, and 84.5% of Philadelphia ballots were cast by Democrats, 9% by Republicans and the rest independents. Even if you attribute the entire Dean Phillips vote to Trump, which seem doubtful, that would merely drop the Democratic share to 80.5% and raise the Republican number to 13.37%. If we looked only at those voting one of the two major parties, even with this generous accounting, we’d see a split of 85.6% to 14.4%. Biden took 81.21% to Trump’s 17.86% in 2020. The most recent primary saw 27% of the Presidential turnout. And yet the NYT poll result has Biden winning only 54% to Trump’s 30% In real terms that would take Biden’s 471,305 vote margin in Philadelphia down to 212,560 and turn a Biden win in the state by 82,166 into a loss by 176,579 , or basically exactly what this poll says of the Statewide margin. How likely is this?
When the polls and the election results clash, and the sample size from the primary electorate is considerably north of 20% of the full electorate and the primary results look basically exactly like the past results, and the same pattern appears across multiple states, it seems a much better bet to rely on observed reality rather the predictive polls. I am not certain whether Biden will win these three states. What I am extremely confident of is that Biden will get 85% in Detroit, 75% in Philadelphia and 75% in Milwaukee. In sports parlance, I would happily bet the over in all three cities.