The partisan divide over civil rights legislation widened abruptly in the 1990s and 2010s

Legislative support for civil rights in the United States Congress has grown over the past fifty years, but this support has increasingly divided along party lines. A new study published in the journal Nature Communications provides evidence that this polarization accelerated sharply during the 1990s and 2010s. This growing gap appears to be largely driven by a surge in bills sponsored by racial minority Democrats.

Civil rights legislation is a primary way the American democratic system formalizes social justice. Past studies suggest two competing ideas about how this policy area has evolved. One hypothesis proposes that civil rights legislation is becoming a more common area of focus due to shifting public attitudes and rising diversity in government. Another hypothesis suggests that support for civil rights is becoming increasingly polarized as political parties pull apart on social issues.

Researchers Joshua Conrad Jackson, Yuanze Liu, and Nour Kteily conducted this study to track the evolution of civil rights as a policy area. Jackson is affiliated with the University of Chicago and Harvard University. Liu is affiliated with the University of Chicago, and Kteily is at Northwestern University. They designed their research to see which social groups receive the most legislative attention and how this attention varies across different political and demographic lines.

The scientists analyzed 202,775 articles of legislation sponsored by the US House of Representatives between 1973 and 2022. They used a combination of human expert tags and a computerized language model to evaluate whether a bill genuinely supported civil rights. First, they looked at subject classifications assigned by the Congressional Research Service, a federal agency that categorizes bills.

Because some bills assigned to the civil rights category actually sought to roll back social justice programs, the researchers added a second layer of analysis. For instance, some historical bills aimed to stop school desegregation efforts but were still tagged as civil rights issues by the government. To fix this, the researchers used a language model to rate each bill’s civil rights prototypicality on a scale from 1 to 100. This score reflected how well a bill’s title aligned with the official US government definition of civil rights.

The researchers combined these two measures into a threshold system to accurately identify genuine civil rights legislation and filter out false matches. They then applied a structural change model, which is a statistical tool used to detect the exact timing of abrupt shifts in data trends. The authors found that civil rights bills did not simply rise in a slow, steady line. The historical increase in the percentage of civil rights bills relative to all legislation was marked by sudden surges, particularly an acceleration around the year 2014.

The scientists then looked at the data by political party and the racial identity of the lawmakers. They specifically focused on White Democrats, White Republicans, and racial minority Democrats. This focus was chosen because these three groups were large enough to yield stable statistical estimates across the entire historical timespan. The researchers used statistical models to compare the percentage of total bills each group dedicated to civil rights, controlling for dips in legislative activity that typically occur during election years.

The gap between Democratic and Republican sponsorship of civil rights bills widened in rapid bursts during 1994 and 2014. By the end of the study period, the share of civil rights legislation sponsored by racial minority Democrats had more than doubled. It jumped from 6.12 percent of all their sponsored legislation in the 2007 congressional cycle to 13.70 percent in the 2019 cycle. The historical trend for White Republicans was not statistically significant.

In a separate analysis, the scientists prompted the language model to identify which specific social groups each bill aimed to protect. A clustering algorithm organized these responses into 18 distinct categories. These categories included older adults, women, veterans, and racial and ethnic minorities.

Racial and ethnic minorities were the most frequently mentioned group in civil rights legislation. The researchers noted a growing gap between political parties in who they tend to support. Democrats were more likely than Republicans to sponsor bills supporting nearly every group.

The sharpest divergence appeared in legislation supporting racial minorities, the LGBTQ+ community, and women. The surge in civil rights bills during the 2010s was almost entirely explained by increased attention to these specific groups. At the same time, legislative attention toward older adults and people with disabilities declined across both political parties.

The observational nature of the data means these patterns cannot establish cause and effect. Changes in the volume of bills over time might correlate with major social events, but the data cannot confirm that specific historical events directly motivated lawmakers to introduce more legislation. The authors note that the jumps in 1994 and 2014 roughly align with high-profile cases of violence against Black Americans, but the statistical models only measure timing rather than the personal motivations of the politicians.

The study analyzes the percentage of total bills rather than the raw number of bills. Relying on percentages allows the researchers to fairly compare groups with different levels of representation in Congress. An increased percentage represents a relative shift in legislative priorities rather than a higher absolute count of bills. The authors verified that their main findings look similar when counting the raw sum of legislation.

The researchers lacked a large enough sample of specific intersectional identities to draw confident conclusions about very specific demographics. For example, there were too few racial minority Republicans in the dataset to include them in the primary analyses. Future research might apply similar language analysis tools to compare the text of political legislation with the communication habits of the general public.

The study, “The widening partisan gap in legislative support for civil rights in the United States,” was authored by Joshua Conrad Jackson, Yuanze Liu, and Nour Kteily.