![[Adobe Stock]](https://sp-ao.shortpixel.ai/client/to_webp,q_glossy,ret_img,w_750,h_375/https://www.psypost.org/wp-content/uploads/2025/04/doubtful-man-thinking-750x375.jpg)
Extreme liberals who relied on offline sources, such as newspapers, television, or friends and family, for scientific information were the most likely of any group to overestimate their scientific knowledge, according to an analysis of General Social Survey data. Extreme liberals who got their science information online were the least likely of any group to do so, a contrast that did not appear among extreme conservatives. The findings were published in Public Understanding of Science.
The internet has fundamentally changed how people access scientific information. It has made it easier than ever to encounter a wide range of opinions about scientific issues. Unlike traditional media, where editors and professional journalists usually decide which information reaches the public, online platforms allow almost anyone to publish their views, regardless of their scientific expertise.
As a result, people searching for scientific information encounter not only established scientific findings but also misleading claims and arguments challenging the scientific consensus. This is particularly important in the United States, where scientific issues such as climate change and the COVID-19 pandemic have become subjects of intense political disagreement and polarization.
Research suggests that political conservatives and liberals differ in their trust in mainstream scientific institutions and the sources they rely on for information. Conservatives are generally more skeptical of mainstream scientific authority, while liberals tend to place greater trust in established scientific institutions. Encountering information that challenges one’s views can affect not only what people think about science but also how confident they feel about their own scientific knowledge.
Study authors Marcus Mann of Purdue University and Cyrus Schleifer of the University of Oklahoma reasoned that the internet might therefore affect liberals and conservatives differently. Liberals who rely on offline sources may rarely encounter views that challenge mainstream science, while online they are more likely to run into such views. Conservatives skeptical of mainstream science, by contrast, may find more challenges to their views offline and more support online.
The researchers examined how political ideology and the sources people use to obtain scientific information relate to confidence in their own scientific knowledge, and how well that confidence matches what they actually know. In particular, they wanted to know whether the link between ideology and scientific confidence differs depending on whether people get scientific information mainly from the internet or from other sources.
To do so, they analyzed General Social Survey data collected between 2006 and 2018. The authors argue this period is well suited to their question because people still varied considerably in where they got scientific information: the share of respondents who named the internet as their main source rose from about 56% in 2006 to 72% in 2018.
The General Social Survey is a large national survey of U.S. adults that has been conducted since 1972 by NORC at the University of Chicago, with a new sample of respondents in each survey year. It asks participants about their attitudes, beliefs, behaviors, and social characteristics. The analysis included 7,177 respondents.
To measure perceived scientific knowledge, the researchers used a question asking how well participants felt they understood the term “scientific study” when they encountered it in news stories. Participants could say they had “a clear understanding of what it means,” “a general sense of what it means,” or “little understanding of what it means.”
To measure actual scientific knowledge, the researchers used the National Science Foundation’s science knowledge scale, which consists of 14 questions. Most are true-or-false statements, such as “The center of the Earth is very hot” and “Father’s genes decide sex of baby,” while others ask about probability and experimental design.
By comparing participants’ confidence with how they scored on the knowledge test, the researchers classified people as overconfident (which the authors call scientific arrogance), underconfident (scientific humility), or accurate in their self-assessment. Participants also reported where they would get information if they wanted to learn about scientific issues such as global warming or biotechnology.
The researchers compared those who named the internet with those who named any offline source, including newspapers, television, books, and friends or family, which the authors refer to as “old media.” The analysis also accounted for education, science coursework, religion, and demographic characteristics.
After accounting for these factors, about 37% of extremely liberal and 35% of extremely conservative participants were predicted to report having “a clear understanding” of what a scientific study is, compared with 29% of moderates. The difference for extreme conservatives, however, fell just short of statistical significance. Overall, people who got their science information online were no more or less confident than those who relied on other sources.
On the knowledge test, liberals tended to score somewhat higher than conservatives, although differences across the political spectrum amounted to no more than about 5 percentage points. Participants who primarily got their information from the internet scored slightly higher than those relying on other sources.
The source of information mattered most for extreme liberals. Among extreme liberals who relied on old media, 53% reported having “a clear understanding” of a scientific study, compared with 30% of extreme liberals who got their information online.
Extreme liberals who relied on old media were also the most overconfident of any group in the study: 39% were classified as overestimating their knowledge and 15% as underestimating it. In every other combination of political ideology and information source, about a third or fewer overestimated their knowledge.
Among extreme liberals who got their information online, only 20% were classified as overconfident, while 29% were underconfident. This contrast between the two groups of extreme liberals was the only difference between media types to reach statistical significance. Extreme conservatives who used the internet showed a small, non-significant shift toward humility, and moderates tended to overestimate rather than underestimate their knowledge regardless of where they got their information.
Additional descriptive analyses suggested that the humbling pattern among liberals was concentrated among those who went online to established outlets, such as online newspapers, magazines, and science websites, rather than social media, search engines, or other nontraditional sources. The authors speculate that such outlets expose people to information that challenges their self-perceptions without the reinforcing echo-chamber effects of social media.
The study contributes to the scientific understanding of the links between scientific knowledge and political identity. However, it should be noted that participants’ confidence in their scientific knowledge was assessed with a single question about understanding the term “scientific study,” while their actual knowledge was assessed using an unrelated set of factual questions. This makes direct comparisons between the two measures problematic, and studies comparing people’s confidence in specific scientific facts with their knowledge of those same facts might yield different results.
The data were also cross-sectional, so it is not known whether getting information online humbles extreme liberals or whether more humble liberals are drawn to online sources. Extreme liberals and extreme conservatives each made up less than 4% of the sample, the “old media” category combined very different sources, and the data end in 2018, before more recent shifts in the online information landscape.
The paper, “Does medium matter? Political orientation and variability in scientific self-perceptions,” was authored by Marcus Mann and Cyrus Schleifer.