Trump Takes Pennsylvania?

Poll Data Offers A Tempting Argument That He Will

As I observed yesterday, we need to be careful when assessing a single state’s electoral dynamics on the basis of a nationwide survey. American elections are by design state affairs, and there is considerable room for variation from national norms when assessing voter sentiment.

However, that does not mean that a state will vary significantly from a nationwide survey in all aspects. When trends are truly national, then we will see a great deal of confluence as we move from a national survey to a state survey.

We certainly see that with AtlasIntel’s polling in Pennsylvania, which mirrors their national poll in several regards.

In looking at the Pennsylvania numbers, we should again consider the composition of the sample population for the poll. In polling Pennsylvania, AtlasIntel increased the percentage of Democrats in the sample.

This is not itself controversial, as Pennsylvania has long been part of the Democratic Party’s “blue wall” of northern states. We should expect a larger proportion of Democrats in the overall state population.

However, this inference of a higher concentration of Democrats again leaves us with a possible risk of oversampling, and so once again we must acknowledge that the poll’s results are, in all probability, a “best case” for the Democrats and Kamala Harris.

Thus when the poll itself shows Donald Trump leading Kamala Harris by more than the poll’s margin of error (±2pp), we need to understand that Trump is leading Harris by at least that much.

Donald Trump could easily be leading by as much as 4.5pp. He is far less likely to be leading by less.

Pennsylvania also confirms another important aspect of AtlasIntel’s national poll: The presence of third party candidates is helping him by pulling support away from Kamala Harris.

Jill Stein is effectively increasing Donald Trump’s margin of victory by some 0.8pp.

We should also pause to note something remarkable about the diversity of Donald Trump’s support: Among voters who identify as a gender other than man or woman (or who refuse to identify)—i.e., voters who are in all probability non-binary—Donald Trump wins by several percentage points.

In Pennsylvania, Donald Trump leads in this “nonbinary” demographic by 9.6pp.

That is more than double the polling edge Trump has (+4.4pp) among this demographic nationwide.

When one considers the many forms of bigotry ascribed to Donald Trump, that voters who identify as anything other than male or female would gravitate to Donald Trump is truly astounding, and speaks volumes about how effectively Kamala Harris is getting her message across.

It is far less surprising to see that, in Pennsylvania, Harris is once again failing to reproduce Joe Biden’s 2020 voter coalition.

Harris is doing slighly better in Pennsylvania in this regard—she has only lost 8.3pp worth of Biden’s 2020 support in Pennsylvania as opposed to 9.8pp nationwide.

Still, these numbers broadly align with the national poll, which thus confirms that there is a significant cohort of Biden voters from 2020 which are disenchanted with Biden’s heir apparent.

This level of disenchantment becomes even more significant when we realize that Biden carried Pennsylvania by only 1.18pp.

The level of support Harris has lost from the 2020 Pennsylvania coalition is many times more than what is needed to move the state into the Trump column come Election Day.

Of particular note in Pennsylvania is that the shift is not merely about Donald Trump. The other notable race in Pennsylvania, the Senate race between Democrat Bob Casey and Republican David McCormick, is also shifted towards the GOP.

In a sign that Donald Trump may have some significant coattails, McCormick has gained 3pp in just the past month to pull ahead of incumbent Bob Casey, who has only lost some marginal support.

That the race is about more than just Donald Trump is underlined by the elevated level of disapproval towards Joe Biden in Pennsylvania.

As in the national poll, Biden is not well liked in Pennsylvania. In keeping with the theme, Biden’s disapproval numbers in Pennylvania roughly a percentage point higher than nationwide.

Given that Joe Biden is a native son of Pennsylvania, for him to be a worse draw than on a nationwide basis is the epitome of political irony.

Unsurprisingly, Biden gets a similarly bad review on how well he did the job of President.

Did I say Pennsylvania does not like Joe Biden? I meant Pennsylvania really does not like Joe Biden.

Moreover, the disenchantment is spread throughout the state. Urban, suburban, and rural voters all are unhappy with how Joe Biden has performed as President.

It is small wonder, then that the issues nearest and dearest to Pennsylvania voters are issues in which Donald Trump is the preferred candidate.

While it is premature to describe Donald Trump’s polling lead over Kamala Harris in Pennsylvania as “overwhelming”, the AtlasIntel poll data shows Trump is every bit as dominant over Kamala Harris in Pennsylvania as he is nationwide. If anything, Donald Trump is dominating in Pennsylvania more than he is dominating nationwide.

Based on the AtlasIntel poll, Donald Trump is winning in Pennsylvania for the same reasons the poll shows him winning nationwide: He is stronger on the issues most relevant to voters, and he has the broadest appeal among multiple voter demographics. In particular, Donald Trump is winning because he, even more than Kamala Harris, is “not Joe Biden.”

In Pennsylvania, as with the nationwide poll, the voters have formed the opinion that Kamala Harris is, in fact, Joe Biden 2.0. Her major support comes from former Joe Biden supporters, yet she has failed to hold Biden’s 2020 voter coalition together. Worse, she has shed enough support from that coalition to cost her the state on just that basis.

While we must always apply the constant caveat of “if the poll data is correct”, the reality of AtlasIntel’s Pennsylvania poll results is that Kamala Harris has failed to get her message across to voters. She has failed to persuade voters that she is able to “turn the page”, and whatever path she claims is the way “forward”, Pennsylvania voters have largely determined that is not the path they want for the country.

If the AltasIntel poll data is correct, Kamala Harris has failed to define herself and her candidacy, and she has allowed Donald Trump to do that for her—which is a political kiss of death in any race.

A poll is not a ballot box, and poll responses are not votes. We cannot, from this poll, automatically infer that Donald Trump is destined to take Pennsylvania. We always have to keep that thought in mind when evaluating polls—all polls can be wrong, and most polls are going to be wrong to some degree.

Yet when multiple polls, and multiple data points, converge on a common conclusion, it is difficult to dismiss the conclusion.

Multiple polls show Donald Trump winning in Pennsylvania.

Multlple betting sites show people willing to wager bigly that Donald Trump is going to win Pennyslvania.

Polls do not guarantee electoral outcomes, and the AtlasIntel poll does not guarantee Donald Trump will take Pennsylania come Election Day. Right now, however, one has to like the odds that Trump really will take Pennsylvania, and the election.