How the electoral math flipped against Democrats
Demographics aren’t destiny. But Democrats are losing support among the groups that once gave them their biggest advantage.
I’ll be conducting a Substack Live with Mike Pesca at 12:30 PM Eastern tomorrow (Wednesday) — 9:30 AM out here in Vegas — where we’ll cover some of the themes discussed in today’s newsletter along with other topics. You can watch us live in the Substack app, or I’ll send the video via the newsletter in case you miss us.
I want to acknowledge that there’s a lot of news this week. Israel is at war with Iran, and the United States may get involved, too. A U.S. Senator was handcuffed at a press conference, and a mayoral candidate in New York was arrested at an immigration hearing. A member of the Minnesota State Legislature was assassinated. And then there were massive political protests on Saturday. It’s one of those times when you can really feel the acceleration into an uncertain future.
But can we get a little “meta” here? One of the tough lessons I learned at FiveThirtyEight is that you don’t need to cover every story. Instead, focus on the places where you add distinctive value. (I’m not much of a foreign policy expert, for instance.) So at times like these, I’m inclined to stick to some bread-and-butter topics like today’s post.
Two weeks ago, I covered a new Catalist report with detailed data — data widely considered best-in-class by political practitioners — on how the electorate has evolved over the past several presidential elections. My topline conclusion was that Kamala Harris can’t primarily blame her defeat last year on turnout. But I largely skipped over the demographic trends described by the report, so let’s catch up on those today.
I would like to note that I can find this type of analysis reductive. At the risk of sounding a little woke, voters contain multitudes of identities. A Gen Y Latina woman raised in a Catholic household by wealthy Republican parents has lots of traits that push and pull her in different directions. So do I, for that matter.1 So do all of us. But the Catalist report contains gold-standard data, and it reveals clear enough trends that I think it’s worth a deeper dive.
How did Democrats go from a winning coalition to a losing one?
Democrats won the popular vote four times in a row from 2008 through 2020. Extend the window back to 1992, and they went 7-for-8. It had almost gotten to the point where Democrats took it for granted that the popular vote was theirs to lose, and the only real question was the Electoral College. The thesis of the “Emerging Democratic Majority”, in which Democrats would be buoyed to perpetual success by an increasingly nonwhite and college-educated electorate, appeared to be irrefutable.
But then Harris narrowly lost the popular vote to Trump last year, contributing to perceptions of a conservative vibe shift. Ironically, the Electoral College/popular vote gap narrowed significantly (more than models like mine had predicted). But since she lost the popular vote, it didn’t matter. Harris lost every key battleground.
How did this happen? Let me introduce a concept that doesn’t have a sexy name, but that I’ll call the “net contribution to popular vote margin” or NCPVM. (This acronym probably won’t catch on.) For instance, say that the Democrat wins 90-10 among Black voters, i.e. by an 80-point margin. And say that the Black vote is 12 percent of the electorate. Democrats’ NCPVM is +80 points x 12 percent or +9.6 points. That means that if the Black vote had been split 50-50 instead, Democrats would perform a net of 9.6 points worse in their overall popular vote margin against Republicans.
I like NCPVM for two reasons. First, as I discussed previously, it recognizes that line between “turnout” and “persuasion” is blurry. You can win an election either by getting more of “your” voters to participate — or by flipping swing voters. But often, there isn’t a clear distinction between these. A voter who grows disillusioned with the Democratic Party might shift to Trump — or she might sit out, or she might vote for a third-party candidate. In general, people with strong political preferences (i.e. those who nearly always vote for one party or the other) are also more likely to vote. But it’s the squishy middle in which elections are won and lost.
Second, it recognizes that the relative size of these various voting groups matters. Frustration with Democrats among Palestinian-American and Arab-American voters undoubtedly cost Democrats in places like Dearborn, Michigan last year, for instance. But Arab-Americans are something like 1 percent of the electorate nationwide. A big shift among a smaller group is usually outweighed in electoral politics by a small shift among a bigger group.
To wit, here is the NCPVM by race for the past four presidential elections, according to Catalist’s detailed data:
White voters are now down to “just” 72 percent of the electorate, as compared with 76 percent in 2008. These demographic shifts, which are fairly predictable based on the age of voters and immigration patterns, were once widely assumed to favor Democrats.
But what happened instead? Well, Kamala Harris lost white voters by 15 points last year, according to Catalist, about the same as Barack Obama’s 14-point defeat among whites in his successful re-election attempt in 2012. Because white voters are now a smaller fraction of the electorate, Trump actually gained less from white voters overall in 2024 than Mitt Romney did in 2012.2
However, Black voters contributed a NCPVM of +11.0 points to Democrats in 2012 but just +7.4 in 2024. Turnout among Black voters was only slightly lower for Harris as a proportion of the electorate than for Obama in 2012 (11 percent versus 12 percent) — and because overall turnout was high last year, similar to if not higher than 2012 on an absolute basis. But Harris won the Black vote by “just” 70 points in 2024, as compared to — are you ready for this? — 91 points (95-4) for Obama in 2012.
Black voters remain overwhelmingly supportive of the Democratic Party, so it would be misguided to blame them in any sense for Harris’s loss. But Democrats had relied on not merely strong but near-unanimous support among Black voters to make up for losing the white vote to make the math balance.
Latinos are a growing share of the electorate, meanwhile — 10 percent in 2024 versus 7 percent in 2012. But Harris won them by only 8 points last year, versus 35 points for Obama in 2012 and an even larger 38 points for Hillary Clinton in 2016. Thus, Hispanics3 added much less NCPVM for Democrats in 2024 as compared with prior elections. The same holds to a lesser degree for Asian American voters.
Let’s explore these trends among other groups in the Catalist data. I’ve been negligent about running paid articles lately, so I’m going to put the second half of this article behind the paywall for people who generously signed up for the paid tier.