American issue polarization surged after 2008 as the left moved further left

The American public has grown increasingly divided on political issues over the past few decades, but this trend does not appear to be happening on a global scale. A recent paper published in Royal Society Open Science introduces a new way to measure these divisions using machine learning. The results reveal that polarization in the United States spiked heavily between 2008 and 2020, offering a fresh perspective on how political disagreements take shape globally.

Political polarization generally describes a society splitting into two distinct and opposing groups. Researchers often try to measure this by looking at the differences in opinion between people who identify with major political parties. Some scientists look at emotional divisions, which involves measuring how much opposing groups dislike each other.

Other researchers focus entirely on policy disagreements, which is known as issue polarization. The traditional approach to measuring issue polarization has some notable limits. In the United States, for instance, people might simply be getting better at matching their existing beliefs to the correct political party.

This alignment process is called “sorting.” If sorting increases, it can look like the country is becoming more divided, even if people’s actual opinions on the issues have not changed. Measuring polarization by comparing major political parties is also hard to do globally.

Many countries have more than two major parties, and some have only one. To navigate these hurdles, researchers at the University of Cambridge developed a new measurement tool. David Jack Young, a researcher in the Department of Psychology, led the work.

He was joined by Lee de-Wit, who leads the Political Psychology Lab at the university. They wanted to track polarization based on actual policy beliefs rather than the political labels people choose for themselves. The team turned to a machine learning technique called a k-means clustering algorithm.

This algorithm looks at a large group of people and automatically sorts them into clusters based purely on their shared opinions. Imagine plotting everyone’s political opinions on a giant graph, where similar views are close together and opposing views are far apart. The algorithm finds the center of these opinion clusters and assigns each person to the group they sit closest to.

It does not look at whether a person calls themselves a liberal or a conservative. After the algorithm forms two groups, the researchers look at three distinct features. The first is separation, which measures how far apart the average opinions of the two groups are.

The second feature is dispersion, which looks at how spread out or cohesive the opinions are within each group. The third feature is size equality, which checks whether the two groups contain a similar number of people. The researchers argue that a society is highly polarized when it splits into two groups that are far apart, internally cohesive, and roughly equal in size.

In their first study, the researchers applied this method to the United States. They analyzed over 35,000 survey responses collected between 1988 and 2024. This data came from the American National Election Studies, a long-running public opinion project.

The surveys asked Americans about their views on 14 specific social and economic topics. These items included the acceptability of abortion, the importance of traditional family ties, and whether the government should fund health insurance. The questions also covered economic inequality and government efforts to address racial discrimination.

The researchers found that the separation between the two opinion groups grew by 64 percent over this period. Notably, the distance between the clusters widened on every single issue measured.

Certain social issues showed especially dramatic changes. In 1988, people with generally conservative views did not necessarily hold restrictive views on abortion. By 2024, conservative opinions on various topics had become tightly packaged with restrictive views on abortion and a strong emphasis on traditional family ties.

Other topics saw the two groups move in entirely opposite directions. Opinions on government funded health insurance and the ongoing impacts of racial inequality divided the clusters sharply. On these issues, the groups actively drifted away from one another over the decades.

Most of this growth happened during a specific window of time. Between 1988 and 2008, the distance between the two clusters remained relatively flat. However, from 2008 to 2020, the two groups moved much further apart.

“Our study shows that 2008 was a major turning point for the divisions between left and right on many of the issues that define contemporary US politics,” said De-Wit. He noted that the public has generally moved left on many issues during this century. The data showed that the left-leaning cluster became substantially more progressive over the decades.

By contrast, the right-leaning cluster shifted only slightly more conservative. At the same time, the two groups remained internally cohesive and roughly equal in size. The researchers also noticed that Americans are aligning their self-identified political labels with their issue groups more closely than before.

Young explained that ideological consistency has become more common. “In the past, someone with left-wing views on one issue might have held right-wing views on another. That’s rarer now.”

This specific timeline challenges the idea that polarization is just an unavoidable feature of human psychology. If human nature were the sole cause, the divisions would likely increase at a steady pace. Instead, the rapid spike suggests that environmental factors like the financial crisis, changes in communication technology, or shifts in political leadership might play a role.

In their second study, the researchers expanded their scope to look at global patterns. They analyzed survey responses from more than 173,000 people across 57 countries. The data came from the World Values Survey and the European Values Survey, collected between 1999 and 2018.

This allowed the team to see if the trends observed in the United States were happening elsewhere. They found no clear evidence that issue polarization is increasing on a global scale. While separation between groups did rise slightly worldwide, the groups also became less internally cohesive.

When looking at what drives divisions globally, cultural issues proved to be the strongest factor. Topics such as the acceptability of abortion, divorce, and homosexuality created the largest splits between opinion clusters. Disagreements over economic policies or democratic systems did not separate the groups as sharply.

The researchers also discovered that a country’s level of economic and social development changes how its population divides. They measured this using the Human Development Index, a metric that considers life expectancy, education, and national income. In countries with lower development scores, the algorithm typically found a large culturally conservative majority and a small culturally liberal minority.

In countries with higher development scores, the two groups tended to be more equal in size and more liberal overall. The United States stood out globally because its two opinion groups have been roughly equal in size for a long time. Young noted that this even split might help explain why American political divisions feel so intense.

The team also looked for societal factors that predict how a country divides. They found that countries with higher ethnic diversity tended to have groups that were further apart on the issues. This diversity was measured by the probability of two random citizens belonging to different ethnic groups.

Additionally, countries with higher wealth inequality had opinion groups that were less cohesive internally. This suggests that economic disparities might create more disagreements within political factions. People who agree on cultural issues might still argue about the economy if wealth is distributed unevenly.

While the new measurement tool offers fresh insights, the researchers acknowledge some limits to their work. One limitation is that the clustering algorithm forces every person into one of two groups. In reality, many people might not fit neatly into either category.

Future research could explore alternative algorithms that allow individuals to remain unassigned if their views do not align with any major cluster. Scientists could also test algorithms that sort populations into three or more groups. This could capture different dynamics in countries with multiple political factions.

The study is also observational, meaning it can show patterns but cannot prove what causes them. It is impossible to say for certain if factors like wealth inequality directly cause changes in public opinion. The associations might be driven by outside factors that were not measured.

Finally, the findings are limited to the specific survey questions available in the datasets. Exploring how populations divide over a wider variety of contemporary issues could provide an even clearer picture of political polarization.

The study, “A new measure of issue polarization using k-means clustering: US trends 1988–2024 and predictors of polarization across the world,” was authored by David Jack Young, James Ackland, Andreas Kapounek, Jens Koed Madsen, Lara Jane Greening and Lee de-Wit.