By now, the details of the New York mayoral race are well-known to millions of people around the country. In late June, 34-year-old Zohran Mamdani, a member of the Democratic Socialists of America and a New York State Assembly member, secured the Democratic Party primary nomination decisively over the establishment candidate, former New York Gov. Andrew Cuomo. Since then, Mamdani charted out a course that culminated in his victory on Tuesday, Nov. 4, 2025.
We can trace the national attention around mayoral races in New York City to it being the largest city in the nation and one of the world’s chief financial and cultural hubs. But even relative to past elections, dialogues around the 2025 race have been especially contentious and consequential — so much so that the president reportedly weighed getting involved and ultimately endorsed Mamdani’s main opponent, Cuomo, who ran as an independent.
The reasons for the heightened stakes are that the mayoral race may reflect a swinging political pendulum away from the presidential administration in the U.S. But more than that, the race has reignited discussions that have dominated U.S. politics since the 2016 presidential election, when Donald Trump defeated Hillary Clinton in a political upset. These events highlighted questions about why people vote in the manner that they do. The questions are more than casual curiosities but can be formalized into a technical research agenda dedicated to investigating the sometimes-hidden forces that craft political identity.
Questions about voting patterns have long been the object of inquiry from thinkers across a breadth of fields, and especially in social science domains like political science, psychology, and economics. Studies in these realms are often reflective and retrospective, telling detailed stories of how, for example, the American electorate has changed over decades. These scholars ideally slow-cook these stories through meticulous analysis before they enter a peer review process that itself operates at a snail’s pace.
But because coverage of elections is such a time-sensitive endeavor, we often rely on the quick trigger of political journalists to tell us the stories at the timescale that we need to grasp the world around us. To have enough information to cast a vote on election day, we may need to know how the race for mayor has changed on a given day and what those changes mean. The problem here, well known throughout journalism, is that the rush to report can sensationalize or oversimplify voting dynamics.
In American politics, few demographic characterizations get more coverage than the manner in which ethnic groups are described as unified voting blocs.
In this sense, the study of political identity faces challenges similar to those in epidemiology. Reflections on the HIV pandemic from the 1990s, for example, helped to build a greater understanding of how epidemics happen. But when the next one arrives, public health officials will all sprint to do their best to tell us whether to wear a mask, get a vaccine, or lower our anxiety and live life as normal.
In the U.S., election discussions seem to be dominated by demographic stories about which groups are voting for which candidate. The strategy in campaigns often focuses on ways to secure the support from certain subgroups. For example, those aspiring for national office in the U.S. frequently aim to secure the votes of the working class, while not alienating the wealthy who run Hollywood, Silicon Valley, and Wall Street, and who are keys to large-scale fundraising.
But in American politics, few demographic characterizations get more coverage than the manner in which ethnic groups are described as unified voting blocs. For example, recent coverage of presumptive Democratic Party contender Pete Buttigieg has focused on his poor performance among Black voters, a historical bedrock of the party. Because of the historical ties between Black people and the Democratic Party, the narrative around the allegiance of certain groups to certain parties has persisted, based to some degree on those parties’ perceived stances on race and racism. A recent study by quantitative social psychologist Andrew Stier and myself sought to examine the relationship between implicit racial bias and voting patterns in U.S. cities. In it, we found that implicit racial biases often diverge from the narrative that Democratic‑leaning areas are less biased and Republican‑leaning ones more so.
In the study, we analyzed the results of racial bias testing using a mathematical model that predicts expected levels of implicit bias based on city size, diversity, and residential segregation. We identified many areas — technically called metropolitan statistical areas — whose actual bias levels don’t align with voting patterns. For example, some Democratic‑voting areas such as Minneapolis and Chicago exhibit higher than expected bias, and other Republication-voting metropolitan areas as Spokane and Knoxville show lower bias. We argue this misalignment highlights the limits of broad assumptions about political identity and points to the need for integrated, context-sensitive approaches that combine quantitative models, local cultural histories, and qualitative research. This approach could help researchers to more accurately understand how racial attitudes and the formation of political identity interact.
Similar surprises have accompanied analyses of the 2024 U.S. presidential race. Because Trump’s presidential campaigns featured inflammatory anti-immigration rhetoric and policy attacks on diversity and gender ideology, the regime is often associated with White nationalism and has been compared to the rise of the Nazi party in the 1930s. Given the flagrancy of these MAGA positions and policies, I was surprised to learn that the 2024 Trump electoral success can be fairly summarized as the product of a relatively multiracial Republican coalition. It is, of course, important not to overstate this point (the President has boasted about support from prominent Black Americans), but the data do speak of a meaningful shift.
For example, according to an analysis from the Pew Research Center, Trump achieved strong support (48 percent) from Hispanic voters in the 2024 election, which represents the best Republican showing among this demographic since 1960. And his improvements across different demographic groups included gains made among working class, Black, and Asian American voters. Furthermore, these gains came along with a high turnout: He was successful in recruiting voters who had skipped the 2020 election. This defies a longstanding assumption that higher turnout benefits the more liberal candidate.
To return to the 2025 New York City mayoral race, early data demonstrate how quickly tides can turn. It appears that Mamdani dominated many predominantly Black and Hispanic precincts, which may reflect a meaningful shift from voters who had backed Mayor Eric Adams in 2021. When Adams dropped out of the race in September, Cuomo expected such voters might shift their support to him, rather than the progressive Mamdani.
Given the large consequences of voting outcomes, we can’t treat the study of political identity as another quirk of the natural world.
More broadly, this highlights the idiosyncratic nature of political identity and the forces that craft it. Our old models are probably wrong. And we need new tools and vocabulary to rebuild them.
Given the large consequences of voting outcomes, we can’t treat the study of political identity as another quirk of the natural world. Like many other fields, the study of complex political identity contains many vectors and is encumbered with a large scientific and psychological burden: It requires creative models and methods for studying diverse populations. And it requires the disabuse of assumptions, narratives, and tropes.
I reiterate that these are not necessarily new questions, and political scientists have been examining them for many decades. If there is a difference in 2025, it is the rate at which our political identities change, which will require the application of diverse talents and skill sets to grasp.
If we are successful in our efforts, then the study of how and why people vote can offer more than electoral post-mortems that explain why the candidates we love or hate won or lost. Rather, the study of political identity can open a window into understanding identity more broadly, with its dynamism, and offer lessons for how society operates and where it is headed.