Is Internet Use Really Polarizing Us? New Insights into Online Habits and Political Division
"A correction to a 2017 study sheds light on how internet usage impacts political polarization, revealing nuances in the data that challenge previous assumptions."
In an era dominated by screens and social media, the question of whether the internet is driving us further apart has become increasingly urgent. A 2017 study initially suggested a link between greater internet use and rising political polarization in the United States. However, a recently issued correction to that study offers a more nuanced perspective, urging us to reconsider the simple cause-and-effect relationship often assumed.
The original research, published in the Proceedings of the National Academy of Sciences (PNAS), pointed to a correlation between increased internet use and heightened political division. This seemed to confirm widespread anxieties about echo chambers and filter bubbles reinforcing existing biases. But what happens when the data itself is re-examined?
This article delves into the details of the correction, exploring how a seemingly minor error in data coding significantly altered the study's conclusions. We'll unpack the implications of these revised findings for how we understand the internet's role in shaping our political landscape, offering insights relevant to anyone concerned about online discourse and its impact on society.
Online News and Ideological Segregation
Research using internet traffic data has suggested that online news audiences are less ideologically segregated than people’s in-person interactions. The comparison specifically includes interactions with family, friends, coworkers, and political discussants. This finding complicates a blanket claim that online news audiences are more segregated than offline social life.
The Data Glitch: Unmasking the Error
The heart of the matter lies in a data coding error that affected the pre-2016 portion of the study. According to the correction, the researchers had inadvertently coded all females as males and all males as females in this segment of the data. While the 2016 data was correctly coded, this error had a ripple effect on the overall analysis.
- Tables S2 and S5 in the SI Appendix
- Figures S2, S4, and S12 in the SI Appendix
- Figure 4 in the main text
Limits of the Opposing-Views Experiment
The study reports that exposure to opposing political views on social media can increase political polarization, but its authors warn against interpreting the result as applying to every setting. They describe their work as among the largest field experiments on social media to date, yet say it should not be generalized to the entire U.S. population. Their stated reason is that a majority of Americans do not use Twitter.
Reducing Like-Minded Facebook Content
A Facebook study found that like-minded sources were prevalent, but reducing exposure to their content had no measurable effect on affective polarization, ideological extremity, or opinions on issues. The researchers’ exploratory equivalence-bound analyses let them rule out effects of ±0.12 standard deviations for the outcomes they examined. The result concerns this Facebook exposure intervention and does not establish that every platform or form of online activity has no effect.
Beyond the Correction: What Does This Mean for Our Online World?
The correction to this study serves as a crucial reminder of the importance of rigorous data analysis and the potential pitfalls of drawing hasty conclusions about complex social phenomena. While the initial findings reinforced existing anxieties about the internet's polarizing effects, the corrected data suggests a more nuanced picture.
Algorithms and Political Division
The Institute of Internet Economics describes algorithmic polarization as a measurable driver of social division. It points to the Oxford Internet Institute’s longitudinal “Computational Propaganda” project, which it says has documented recommendation engines and influencer networks amplifying partisanship across 65 countries. These are claims reported by the institute about the project and the wider role of algorithms.
Polarization Beyond Online Discourse
A 2024 perspective describes polarization as a critical threat to the stability of nations around the world. It identifies effects on climate change, populism, democracy, and global health. The perspective examines how polarization is understood, the challenges of measuring it, and potential interventions.
Anger and Social Unrest
A Stanford case study is designed to invite discussion of out-of-the-box thinking about major societal challenges. It focuses on increased polarization and the resulting anger, violence, and social unrest. The source presents these as the challenge explored by the case, rather than providing a quantified estimate of their effects.
It highlights the need to move beyond simplistic narratives and delve deeper into the multifaceted ways in which online interactions shape our political beliefs and behaviors. Are echo chambers as pervasive as we think? How do algorithms influence the information we consume? These are just some of the questions that warrant further exploration.
Ultimately, understanding the internet's impact on political polarization requires a commitment to careful research, critical thinking, and a willingness to challenge our own assumptions. The revised study encourages us to approach the online world with a more discerning eye, recognizing that the relationship between technology and society is far more intricate than we often assume.