Home > News > How to think about the “racial realignment” in U.S. politics
409 views 19 min 0 Comment

How to think about the “racial realignment” in U.S. politics

Shifts in party coalitions don’t portend a shift in party positions.

- November 18, 2024

Right now, post-election surveys suggest that Donald Trump did better among Latino voters and, to a lesser extent, Black voters, compared to 2020. The longer-term changes from 2016 to 2024 are sometimes characterized as a “racial realignment” that is creating a “multiethnic” Republican party.

How true is that characterization? What is causing any changes in the Democratic and Republican party coalitions? And what does this all mean for the future? I think there’s a more complicated story than you’ll often hear. 

These are the two key points:

  • These changes in racial composition of the party coalitions are part of a larger “sorting” of the electorate based on identity and ideology. In a sense, Latino and Black voters are manifesting trends we have already seen among white voters.
  • These changes are actually not going to create what a “realignment” usually entails: long-term domination by one party and big changes in what the parties stand for.

In other words, the parties are simultaneously changing and staying the same.

Let’s dive in.

What exactly happened in 2020?

After the 2020 election, respected estimates from Catalist and the Pew Research Center suggested that a larger fraction of Black and Latino voters supported Donald Trump, compared to 2016. It’s always important to compare multiple sources because there is no one canonical estimate of how any particular group of Americans voted.

Among Black voters in 2020, Trump gained about 2 to 3 points of “two-way vote share” – that is, his share among those voting for Trump or Biden. Among Latino voters he gained about 8 to 10 points of vote share. He may have gained about 1 point among Asian voters.

What does this mean in absolute terms? Those two estimates, plus a third (from AP Votecast), suggest that in 2020 something like 6-8% of Black voters supported Trump. About 35-38% of Latinos voted for Trump.

[Sidenote: It is better to think about electoral changes in terms of vote share rather than the margin between the two candidates. Imagine an electorate that has 50 Democratic and 50 Republican voters. One Democratic voter switches sides. The percent Republican increases by one percentage point (51%) but the margin increases by 2 points (from 50-50 to 51-49, or +2 Republican). Using margin exaggerates the scale of the change.]

What about 2024?

We do not have a full set of estimates yet. Right now, the AP Votecast estimate is that Trump won 16% of the two-way vote share among Black voters and 43% among Latino voters. Both figures are about a 7-point shift from the AP Votecast estimates in 2020.

If those estimates are confirmed by other surveys, it would suggest that Trump made further gains among Black and Latino voters. This confirms the earlier predictions of some analysts that the GOP would continue becoming more racially diverse.

Are these changes driven by gender?

Some post-2024 commentary centers on gender and particularly Latino men or Black men. I would caution against this as the primary explanation. 

First, gender was not a big factor in 2020. According to the research by political scientists Bernard Fraga, Yamil Velez, and Emily West, Trump’s gains among Latinos in 2020 were not much different among men and women.

Second, the commentary about Latino men in 2024 leans on the traditional exit poll because it tells the more dramatic story. But in the AP Votecast data, Trump gained 9 points of vote share among Latino men and 6 points among Latino women. It’s not just a story about men. 

Among Black voters, Trump’s gains were greater with Black men (12 points) than Black women (3 points). That is a larger difference, one that is consistent with pre-election polls. This deserves more investigation. But, again, Trump’s gains were certainly not limited to Black men. Rather than start with gender, I would look for a broader explanation.

What about class, then?

It is also common to hear Black and Latino Trump voters described as “working class,” which usually means “having less than a college education.”

But again, the data complicate this story. In their research on the 2020 election, Velez, Fraga, and West did find that Trump won a larger share among Latino voters with less than a college education – but those voters did not contribute much to his total net votes since relatively few of them turned out.

In the AP Votecast estimates from 2024, Trump gained 7 points among non-white voters without a college degree. But he also gained 5 points among non-white voters with a college degree. 

We will need to break down this “non-white” category and examine racial and ethnic groups separately. And we’re still waiting for more data. But at the moment, Trump’s gains among non-white voters in 2024 were not centered among the “working class” – if that term means voters with less than a college degree. 

And regardless, we cannot interpret educational differences as purely about “class.” Education levels also correlate with different political identities and values.

Is it just inflation?

You could interpret the 2024 changes among Latino and Black voters as part of a much broader shift brought about by inflation and the ensuing unpopularity of the incumbent, Joe Biden. I wrote about that the day after the election. And I think broader national conditions could be a factor. 

But that doesn’t help us understand the changes between 2016 and 2020, when Trump gained support from Black and Latino voters despite being an unpopular incumbent himself. So we should consider other factors.

What else might be at work?

There are two crucial factors in particular: identity and ideology. They’re not always separable, but let’s start with identity. 

For a long time, many scholars have pointed out that Latino voters do not all strongly identify as Latino. And Latinos who do not identify as Latino – or who identify more as American than Latino – are a more Republican group.

Indeed, this was true even before 2020. In our book on the 2016 election, Michael Tesler, Lynn Vavreck, and I discussed why more Latinos didn’t support Hillary Clinton, despite the notion that they were the “sleeping giant” who would awaken and help defeat Trump. 

Part of the challenge was rallying Latino voters who don’t strongly identify as Latino. Here is a graph (eventually cut from the book) that shows the relationship between Latino identity and voting behavior in 2012 and 2016.

Ethnic identity also matters for how Latinos feel about other racial groups. For example, Latinos who identify primarily as American also express more anti-Black prejudice, according to research by Efrén Pérez, Crystal Robertson, and Bianca Vicuña. 

This matters for 2024 because, as Michael Tesler and I showed previously, the Latinos shifting away from the Democratic Party in 2020 and afterward were less sympathetic to Black people. Indeed, between 2019 and 2022, the Latinos who shifted their party identification away from the Democratic Party were concentrated among those who said that America had “gone too far” in “giving Black people equal rights.” 

Interviews with Latino Republicans also find evidence of less sympathetic views of Black Americans.

In other words, even as we are becoming less racially polarized, we are becoming more polarized by racial attitudes.

What is happening among Latinos has already happened among white voters. White voters who moved into the Republican Party in the 1990s and 2000s also had more racially conservative attitudes, as shown in the research of Michael Tesler and Andrew Englehardt

So, to explain why some Latinos may be moving to the Republican Party, we have to study how they feel not only about Latinos but also about other racial groups.

What about ideology?

Research on the 2020 election by political scientists Bernard Fraga, Yamil Velez, and Emily West shows that Republican gains among Latino voters came almost exclusively among those with more conservative attitudes. This includes attitudes about crime and immigration, as well as self-identification as conservative or very conservative. Here’s a graph from their article:

Of course, issues like crime and immigration are not purely “ideological.” They are linked to identity as well. Latinos who do not strongly identify as Latino tend to be more conservative on immigration.

For example, in the 2016 American National Election Study, only 49% of Hispanics who said that being Hispanic was not important or only a little important to them agreed that immigrants are good for the economy. Among Hispanics who said being Hispanic was extremely important to them, 75% said that immigrants were good for the economy.

Nevertheless, what appears to be happening is that the broader ideological “sorting” of the American electorate – whereby liberalism goes with Democratic party identification and conservatism with Republican party identification – appears to be affecting Black and Latino voters as well. As Efrén Pérez noted after the election:

Indeed, in 2020, you could see this ideological sorting among all major racial and ethnic groups. This graph, from our book about the 2020 election, shows that Trump gained among conservatives in every racial and ethnic group relative to 2016, while Biden gained among liberals.

So, do both parties now have “multiethnic coalitions”?

After the election, the conservative writer Reihan Salaam said this:

As we approach the midpoint of this decade, what’s emerging instead is something entirely different – a clash between two multiethnic coalitions. The multiethnic coalition on the left, united by a particular vision for racial progress, is arrayed against a multiethnic coalition on the right, united by its opposition to that vision.

This establishes a sort of equivalence between the parties: both are “multiethnic coalitions.” While that is true in a literal sense – both parties include voters of different racial and ethnic backgrounds – it also papers over one key fact: the Democratic Party remains much more multiethnic than the Republican Party.

In 2020, the Catalist estimates suggested that about 15% of Trump voters were not white. By contrast, 39% of Biden voters were. Pew Research Center data generated nearly identical estimates

In 2024, we don’t yet have the Catalist or Pew estimates. But my back-of-the-envelope calculation from AP Votecast surveys is that in 2024 about 16% of Trump voters were not white, compared to 13% in 2020. In 2024, 34% of Harris voters were not white, compared to 39% of Biden voters in 2020.

Again, we need more data and more estimates. But if the 2020 to 2024 changes in the AP Votecast numbers are largely correct, it shows that the changes in party coalitions occur slowly, especially when those changes involve groups of voters, such as Black and Latino voters, who are a small fraction of the overall electorate.

Thus, the changes in 2020 and 2024 have only begun to make the Republican Party as “multiethnic” as the Democratic Party. Right now, the GOP is still largely a party of white people.

Does all of this constitute a “realignment”?

It is difficult to answer that question because people mean different things by “realignment.” At times, people seem to use the term simply to signify that Something Dramatic Has Happened. In the academic literature, scholars continue to debate the concept of a realignment. 

One definition of realignment is what I wrote after the 2008 election: “For a realignment to occur, there has to be a dramatic and permanent shift in the party coalitions. That shift then ushers in an extended period of party control, which in turn brings with it a notable shift in policy.”

This leads to an important irony: The “racial realignment” may indeed shift the demographic composition of the party coalitions, but it likely will create neither long-term control for either party nor much of a shift in policy. Here’s why.

Go back to the fact that the racial changes to the party coalitions are part of a broader process of sorting based on identity and ideology. On the one hand, this fact means that Republican gains among people of color are likely to be durable, as Yamil Velez argued when I interviewed him after the election.

But ideological sorting has two sides. Republicans and Democrats both gain ideologically aligned voters. The result is not a single dominant party, but two parties operating essentially at parity, with close elections producing frequent swings in which party has power in the White House and in Congress. John Judis and Ruy Teixeira recently made this point as well.

And instead of shifts in policy, the changes in the party coalitions are actually reinforcing the policies the parties already support. For the Republican Party, the influx of conservative-leaning voters of color means that the party will remain conservative-leaning – including on issues related to race and immigration. Becoming more “multiethnic” may not remake the party.

In short, if demographic changes in the parties don’t produce clear ideological changes, any “realignment” is less of a big deal.

What does any “racial realignment” mean for the future elections?

Maybe not much. As always, commentators love to assume that the results of a single election have lasting implications. Here’s one recent pronouncement after the 2024 election: “So conservative power is consolidated in a way that makes the Biden administration look like the fluke, the last gasp of a dying order.”

I don’t know what that prediction is based on. But you certainly cannot base it on the demographic changes in the two parties that underlie this “realignment.” I give you the cold, hard facts from a new paper by economists Richard Calvo, Vincent Pons, and Jesse Shapiro:

Many observers have forecast large partisan shifts in the US electorate based on demographic trends. Such forecasts are appealing because demographic trends are often predictable even over long horizons. We backtest demographic forecasts using data on US elections since 1952. We envision a forecaster who fits a model using data from a given election and uses that model, in tandem with a projection of demographic trends, to predict future elections. Even a forecaster with perfect knowledge of future demographic trends would have performed poorly over this period – worse even than one who simply guesses that each election will have a 50-50 partisan split.

In other words, we can’t predict 2028 from the demographic shifts in 2024. Maybe that dying order isn’t dead just yet.

Changes in the party coalitions help us understand what happened in the 2024 election. They just don’t tell us much about who will win four years from now.