New research published in theory and society Research published in the social sciences suggests that it has consistently tilted to the political left for more than 60 years. The findings indicate that this leftward shift has strengthened over time, especially regarding social and cultural issues. This provides evidence that the political orientation of the academic publishing environment is becoming increasingly homogeneous.
Past research has consistently shown that U.S. university faculty tend to identify with left-wing political views. James Munsey, a researcher at the University of Oxford, wanted to know whether this political preference actually manifested itself in the published academic papers themselves. Prior to recent technological advances, analyzing the political content of hundreds of thousands of scientific documents was far too expensive and time-consuming for human readers.
Manzi decided to use artificial intelligence to read and code these vast amounts of text to see how academic output has changed over 65 years. Looking at published work gives us a stable way to see what areas we think are important and how we frame the problem. Rather than asking how individual scientists vote, this approach evaluates the actual ideas that make it into the scientific record.
Researchers collected exactly 599,194 English-language abstracts from social science articles published between 1960 and 2024. An abstract is a short summary of a research paper that appears at the beginning of an academic paper. This text is collected from 367 academic journals representing 11 different disciplines, including economics, sociology, and political science.
To process this huge dataset, Manzi used a large-scale language model. It is an artificial intelligence program that acts as a digital reader, recognizing patterns in large amounts of text. To ensure the program consistently evaluated all abstracts, researchers provided a rigorous rubric based on the U.S. political spectrum as it existed in 2025. The program rated each abstract on a scale of 0 to 10. 0 represents far right, 5 represents politically neutral, and 10 represents far left.
To help the show make sense of these ratings, scientists gave it a concrete, contemporary political anchor. The instructions linked each number on the scale to a contemporary policy position from a prominent politician or think tank. Using this rubric, the program answered specific questions about each summary, including whether the argument is relevant to contemporary political debates and where the idea fits on today’s economic and cultural spectrum.
Across the sample, the program identified 180,311 abstracts that directly discussed current political or social debates, and the scientists focused their primary analysis on this subset.
Before running the main analysis, the researchers validated the artificial intelligence by having it score summaries of articles published by known political groups. Computer programs have succeeded in putting these organizations on the right side of the political spectrum. Researchers also tested the tool on theoretical mathematics papers, but it didn’t exactly yield politically relevant results.
Key data showed that around 90% of politically relevant social science papers were left-leaning. The average political position of all social science fields remained left of center every year throughout the measurement period. The reason that the average is consistently left-leaning is largely due to the fact that there are very few right-leaning works in most fields.
Different disciplines have followed slightly different paths over time. Academic fields closely related to public policy, such as economics and political science, experienced a slightly slower rightward shift from the 1970s to the 1980s. After 1990, these policy-focused areas shifted back to the left.
In contrast, fields focused on identity and culture, such as sociology and gender studies, showed a sustained shift to the left over the entire period. Starting around 2010, a shift to the left began to accelerate in some of these areas. The researchers noted that this acceleration appears to be a continuation of a long-term trend, rather than a sudden break from the past.
The scientist also observed that fields with strong left-leaning tendencies tended to have less political diversity in published works. Over time, the social sciences have become more ideologically homogeneous. As the discipline drifted further to the left, the diversity of political perspectives represented in its compendium decreased.
To understand the causes of this ideological drift, scientists looked at whether individuals are changing their minds or whether fields are simply bringing in different people. The data shows that the leftward trajectory is driven primarily by new contributors who have entered the field with more progressive views. Older academics who changed their views over the course of their careers played a much smaller role in overall change.
In this analysis, we divided the evaluation into economic and sociocultural issues. Artificial Intelligence consistently rated abstracts as leaving more on social and cultural topics than on economic topics. This gap has widened significantly over the decades.
This suggests that even if scientists’ economic views remained slightly closer to the center, they became increasingly progressive on cultural issues. This reflects a broader societal trend in which highly educated professionals often hold moderate economic views along with progressive social ideals.
Readers should avoid misinterpreting these patterns as direct evidence of intentional bias or deliberate suppression of right-leaning views. Although this study measures the political tone of the final published work, it does not explain why that tone exists. It’s very likely that scientists are advancing research into topics like climate change and racial justice, and that artificial intelligence might naturally code as leftist.
The artificial intelligence also looked at only short summaries of every scientific paper, rather than the full text. Scientists often use abstracts as promotional summaries to attract the attention of journal editors. Researchers may rely on popular political buzzwords to increase their chances of publication. This means that the abstract can sound more ideological than the actual data and methods contained in the full paper.
The study also applied a strict U.S. political framework to all English summaries. Because science is a global enterprise, researchers in the United Kingdom or Australia may use concepts that are considered progressive in the United States but politically moderate in their home countries.
Finally, large-scale language models come with their own biases based on the internet data used to train them. This internal programming can influence how the software defines political neutrality or interprets sensitive academic debates.
This study, “Ideological Orientations in Academic Social Science Research 1960-2024,” was authored by James Manzi.
