Digging deep on the presidential electorate

After every federal election, Pew Research Center conducts an incredibly useful bit of polling: asking Americans what they did on Election Day and validating that those who reported voting were actually registered voters. It takes a few months for the analysis to be completed, but the result is one of the most useful and detailed assessments of the electorate — and the non-voter pool — that is publicly available. Think exit polls, but more reliable.

I’ve used this analysis (the 2024 iteration of which was published in June) in the past to evaluate how Donald Trump’s base has evolved since the 2016 election. But after seeing a social-media post this week (which I’ll share in a moment), I realized that there’s a lot more information that can be presented to help explore how the 2024 election unfolded.

I came up with a little game. Below is a blank scatterplot, a chart that contrasts the margin of support in the presidential race (from more Democratic at left to more Republican at right) with the percentage of voters a particular demographic group constituted last year.

I know that’ complicated, so let me just explain what to do. Figure out what percentage of the electorate were White men last year and what margin they preferred the Democratic (Kamala Harris) or Republican (Trump) candidate in the election. If you think that they were 100 percent of voters and split evenly between Harris and Trump, you’d be wrong. But you’d click/tap right at the top center of the chart — at the place where 100% of the electorate meets zero on the left-right spectrum. If you think they were 50 percent of voters and backed Trump by 50 points, you’d tap/click at the intersection of the second horizontal line down and the third vertical line from the left.

That’s confusing, too. So just try it.

Once you do, it tells you how close you were. Was the actual value where you thought it would be?

Now try to guess where the values for Black voters (all of them, not just men) landed last year.

Contrast that with the values for White evangelical Protestant voters.

Why that comparison? Because it was a post from my former colleague Perry Bacon that inspired this experiment.

The number of white evangelicals backing Harris in 2024 is around the same as Black voters backing Trump. (15 percent of the group).

— Perry Bacon (@perrybaconjr.bsky.social) 2025-09-28T13:55:00.671Z

You’ll notice that his point is about shares of voters for Harris and Trump, not voters overall. We can test that, too. (Note the change in the question below.)

We can run the same experiment for years past, given that Pew’s been doing this since 2016.

The Pew data encapsulates an enormous range of demographic groups, particularly in more recent years. (There are some caveats I mostly elide here; feel free to go read the methodology in detail.)

At the bottom of this article is a version of the scatterplot that lets you compare demographics and years directly. One thing you might want to play around with is how voting shifted by age. For example, try your hand at how young voters voted in 2024…

…versus 2016.

And then compare that with older voters.

One demographic group I found interesting to consider, given the amount of attention that has been paid to it over the past few years, is Jewish Americans. Where do you estimate their support lands on our chart?

Anyway, here’s the whole thing. Lots of categories and cross-tabulations to explore. There are some gaps in the data for 2020 and (more often) 2016, so if a dot doesn’t show up on the chart for some combination, that’s likely why.

Find something particularly surprising? Let me know on social media or by email!

Photo: Balloon drop at the 1972 Republican convention. (National Archives)