Introduction
Despite a cell phone ban, news of a prominent political assassination rapidly spread through my high school, revealing stark political divisions across peer groups. These reactions towards a political party exemplify “affective polarization,” defined as “distrust, dislike, and contempt across party lines” (Garzia et al., 2025). While this may seem insignificant in a high-school setting, polarization at scale has been linked to political violence, economic uncertainty, and the erosion of democracy (Arbatli & Rosenberg, 2021; Baker et al., 2020; Piazza, 2023). Unfortunately, polarization in the United States is only increasing, making it important to understand the processes behind it (Boxell et al., 2020). One such processes is echo chambers, defined by Jamieson & Cappella (2008) as “a bounded, enclosed media space that has the potential to both magnify the messages delivered within it and insulate them from rebuttal.” While echo chamber research has historically centered on social media, researchers are increasingly recognizing that echo chambers can form in person as well (Balietti et al., 2021).
Even with growing polarization research, one group has been neglected – adolescents. Polarization among adolescents has risen sharply over the past few decades and is just as high as that in adults (Tyler & Iyengar, 2023). This matters because adolescence is a critical period of “political socialization,” which Neundorf & Smets (2017) defined as the “process by which citizens crystalize political identities, values and behaviors” that influence their politics later in life. These beliefs tend to remain relatively stable after crystallization in high school (Krosnick & Alwin, 1989). Among the factors shaping this belief, including parents, youth culture, personal identification, and high-school classes and activities (Pfaff, 2009; Polipciuc et al., 2023; Tedin, 1974; Wegemer, 2022), peer groups play a particularly significant role (Laursen & Veenstra, 2021; Tedin, 1980). I concur with the literature in defining peer groups as small, relatively intimate groups of friends who interact regularly, and similarly define peer influence as instances where a person is affected by one or multiple people of the same age within a peer group (Laursen & Veenstra, 2021). Note that in the context of the political polarization literature, peer and friend are used practically interchangeably.
During adolescence (including high schoolers aged 14-18), as a result of psychological developments, social interactions become increasingly important: adolescent spend more time with peers (Lam et al., 2014), grow more sensitive to peer approval than children or adults (Foulkes & Blakemore, 2016), and develop a stronger sense of belonging - defined as a “feeling of value and respect derived from a reciprocal relationship to an external referent that is built on a foundation of shared experiences, beliefs or personal characteristics.” (Mahar et al., 2013). These dynamics, combined with cognitive growth like executive control, solidifying beliefs while opening minds to other perspectives (Dumontheil et al., 2010), make late adolescence a particularly consequential period for political socialization and polarization. To bridge the gap between polarization research and adolescence, as well as to connect it with peer group processes, this study asks: How do American high school peer groups influence political polarization among high schoolers?
Methods
Study Design
Data collected through a Jotform survey and 30-45 minute interviews conducted over Zoom or in person, were triangulated with political science, anthropology, and psychology literature. Participants were entered into a raffle for a $50 gift card to encourage engagement. The survey was distributed primarily through school clubs – via in-person advertisement or communication platforms - to reflect the school accurately and reach a diverse range of peer groups based on different interests (Block & Grund, 2014). This study received IRB approval, followed informed consent procedures, and protected participant confidentiality through anonymization.
Interviews followed Bernard’s (2018) semi-structured approach, using an organized guide while allowing flexibility to pursue relevant leads. This balanced comparability with openness, especially important given the interviewer’s similar age to participants and the sensitivity of political topics.
Data Analysis
Respondents from both the survey and interview answered demographic questions as shown in Table 1. Although participant counts differ between methods, the relative percentages are compatible and reflect the school’s demographics.
Survey questions were designed based on peer influence factors as identified by Tilley & Hobolt (2025), including political salience, frequency of political interaction, and demographic homogeneity. Political salience in this study was defined as the importance of politics to the peer group. Political interaction was measured broadly and separated into social media and in-person communication. Homogeneity was assessed across political affiliation, race/ethnicity, religion, and gender. These variables were measured using five-point Likert scales (Likert, 1932), as shown in Tables 2 & 3, which are common in affective polarization research (McMurtrie et al., 2024). To streamline analysis, the five answer choices were grouped into three on the tables as represented by slashes.
Affective polarization was measured using 0-100 “feeling thermometer” scales toward Democrats and Republicans, the primary source of U.S. polarization (Iyengar et al., 2019). While there are many ways of analyzing polarization, “feeling thermometers” are the most common and well-validated measure in affective polarization research (Tyler & Iyengar, 2024). Affective polarization was calculated as the difference between in-party (most favorable party) and out-party (least favorable party) feeling thermometer scores, with higher differences indicating greater polarization (Iyengar et al., 2019). One-way ANOVA followed by the least-square means tests were used for analysis, with p < 0.05 considered statistically significant. To capture the range of sentiments that are defined as affective polarization, results from three feeling thermometer questions targeting feelings towards a party, feelings toward members of a party, and trust towards members of a party are shown in Tables 2 & 3.
The interview was designed to supplement the survey, with many parallel questions posed in open-ended form to capture stories, feelings, and context the survey could not. Responses were coded by question type and compared against survey findings. Quotes provided have been slightly modified for grammar while preserving the original meaning.
Participant was asked about their “closest friend” and “close friend group” as opposed to “best friend” and “best friend group,” respectively. This diction is consistent with literature, suggesting it is often easier for people to think of a closest friend rather than a best friend, which has a slightly more stringent connotation (Allen et al., 2022; Thurner et al., 2025). Furthermore, specifying a specific friend group ensured consistency, as many people have several groups with different characteristics.
Finally, a literature review was conducted to triangulate data from the survey and interview, connecting findings to the broader body of research. Although this single high school case study limits generalizability due to the geographical and demographic factors, combining survey, interview, and literature data - along with statistically significant results - allowed the study to propose a framework for understanding peer group influence on high school political polarization and can serve as groundwork for future research.
Results
Peer Groups Influence Affective Polarization
As shown in Table 2, respondents in peer groups with higher political salience tend to be significantly (p<0.05) more polarized than peer groups that find politics not important, suggesting a positive correlation between peer group political salience and affective polarization. This is best presented by a 16-year-old girl who said that, “sometimes around them [close friend group], I feel like I should just be the most liberal as I can possibly think,” explaining that political salience is high in her peer group as “[closest friend] is very politically aware and cares a lot” about politics. Tedin (1980) found that having a friend with a greater political salience on certain issues is linked with more political influence; my finding is more distinct by connecting political salience directly with polarization.
Table 2 shows that respondents in peer groups that interact about politics very often/often, are significantly (p<0.05) more polarized than peer groups that interact about politics rarely/never. Participants who sometimes interacted with peers over politics also tend to have higher polarization than those that rarely interact, suggesting a large gap between people who engage with politics at all and those who do not. Thurner et al. (2025) did a similar study for adults that found that more social interactions can lead to more polarization; my findings highlight how that trend applies to high schoolers as well.
Respondents in friend groups that interact more over social media about politics tend to be more polarized, though the relationship is weaker. Table 2 shows that friend groups interacting over social media very often/often are only slightly more polarized than those interacting rarely/never for two of the affective polarization questions, while those interacting sometimes showed significantly (p<0.05) higher polarization than rarely/never. This conflicts with the literature as Bail et al. (2018); Kubin & Von Sikorski (2021); and Thurner et al. (2025) all support a strong positive correlation with social media usage, whereas Lee et al. (2014) finds no correlation. A 16-year-old girl exemplifies the weak correlations, noting that although a friend would “bash Kamala Harris in the comments of social media,” she ultimately does not “let it affect my own opinion.” Critically, prior studies analyze social media use for consuming outside content and news broadly, whereas my study isolates peer-to-peer interaction specifically, providing clearer insights into social media as a vector for peer group influence.
High schoolers in peer groups that talk in person more about politics tend to be more polarized. Table 2 shows that respondents whose peer groups often discuss politics in person are significantly (p<0.05) more polarized than those whose peer groups rarely/never do. Mamakos & Finkel (2023) found that the person involved in the interaction matters more than the medium, which explain why both in-person discussion and social media interaction with the same peers yield positive correlations with affective polarization.
Closest Friend Influences Affective Polarization More Than Friend Group
The trends from the friend group analysis mirror those for the closest friend analysis, with all data showing a positive correlation between level of interaction or salience with the closest friend and affective polarization. The main difference is that the degree of affective polarization was consistently larger for the closest friend, suggesting they play a bigger role in influencing polarization than the broader peer group. This is consistent with anthropological literature finding that the closer a friend is, the more influence they have over a person (Morgan & Grube, 1991; Urberg, 1992). While those studies tested influence over non-political actions, my findings show how these trends can be extrapolated into politics.
Peer Group Demographics and Affective Polarization
Table 3 shows that participants with more politically homogenous friend groups tend to be significantly (p<0.05) more polarized than those with mixed or mostly/all different political affiliations, suggesting that greater exposure to people of different beliefs reduces polarization. This trend is consistent with literature on adults (Balietti et al., 2021; Overton & Cunningham, 2025).
By contrast, race/ethnicity, religion, and gender homogeneity showed no significant difference in polarization across groups, suggesting these are not important factors in determining affective polarization. This may explain the lack of literature on these dimensions. Rather, the literature suggests that it would be more conclusive to do research on each specific subgroup (e.g. whether the person is Muslim or Catholic) instead (Amalia et al., 2024; Hirschl et al., 2012).
Discussion: Affective Polarization through Belonging-Induced Echo Chambers
Peer groups’ formative roles in building a person’s identity extend to the political sphere through the psychological desire for group belonging. The combination of adolescent-specific cognitive developments, increased peer interaction time, and growing autonomy all contribute to increased peer influence among high schoolers. The main form of peer influence is the maintenance of belonging. The desire to fit in leads to the repetition and reinforcement of similar views, at the expense of the entry of opposing views that could challenge the group’s identity and decrease a person’s belonging within a group. Echo chambers reward expressions that reinforce majority opinion while leading people to overvalue their political sides and discredit opposing viewpoints (Nyhan & Reifler, 2010; Schachter, 1951), resulting in strong in-group favorability and out-group disfavorability that increase affective polarization. Based on the convergence of survey data, interview evidence, similar trends identified in literature, and the prevalent need for belonging in most high schoolers, I propose the framework of belonging-induced echo chambers to offer a useful lens for understanding how adolescent peer groups affect political polarization in high school. Although broad generalization is limited by the case study’s unique geographical and demographical traits, this finding contributes to ongoing theory development on adolescent polarization and opens avenues for future research.
Several factors drive belonging-induced echo chambers. A peer group must care about politics (political salience) and interact about it—both on social media and in person—to reinforce views. Repetition of similar opinions, strengthened by political homogeneity, also matters. Individuals must seek belonging within the group. Thus, echo chambers form when these factors positively correlate with affective polarization. This study analyzed respondents’ “closest friend group”, where belonging is strongest, and found positive correlations between affective polarization and political salience, political interaction, social media interaction, in-person interaction, and political homogeneity. Greater salience centers group identity on politics; because high schoolers seek belonging, they tend to reinforce rather than challenge shared beliefs. More political interaction increases opportunities for reinforcement, with in-person conversations especially powerful in solidifying belonging and discouraging dissent. Social media interactions also showed a positive correlation, though weaker, suggesting it can still facilitate echo chambers but less effectively than in-person interaction. The interview data corroborates this with examples of people interacting indirectly by commenting on other posts rather than direct online discussions. It is worth noting that these social media echo chambers differ from those in previous studies (Cinelli et al., 2021; Justwan et al., 2018), which examine algorithmic content absorption among other factors, whereas my study isolates social media strictly as a peer communication platform. Finally, greater political homogeneity increases uniformity of beliefs and the likelihood of similar views being recirculated among the friend group. Together, these factors - salience, interaction, homogeneity, and closeness – drive high schooler into belonging-induced echo chambers and ultimately increasing affective polarization.
Echo chambers are meant to be closed systems where no differing views can enter and similar views are magnified. However, high schoolers are exposed to more influences such as their parents and other peer groups that could introduce opposing views and potentially break the echo chamber. Because adolescence is a time in a person’s life when they are especially influenced by peer groups, similar views from the peer group become magnified compared to other influences and opposing views outside of it diminish in importance, allowing echo chambers to still thrive.
Conclusion
This study demonstrated that greater political salience, more political interaction, more social media political interaction, and more in-person political discussion with the peer group all correlate with greater affective polarization - supported by interview examples including one girl who felt pressured to become more liberal due to her politically salient friend group. The weakest positive correlation was that of social media interaction. Closest friend trends mirrored those of the peer group, with the key difference that extreme values of each variable with a closest friend produced greater polarization than with the broader group. Greater political homogeneity also increased affective polarization, whereas racial/ethnic, religious, and gender homogeneity played no significant role.
Past research into affective polarization has analyzed the formation of echo chambers through algorithms on social media, focusing primarily on adults. The belonging-induced echo chambers framework offers a novel approach into understanding affective polarization by analyzing the formation of echo chambers through the psychological process of peer group influence, focusing primarily on adolescence. While this framework emerged from a single-school case study, its consistency with established research on peer influence, echo chambers, and adolescent development suggests potential applicability to other high schools, though further testing is warranted. Future researchers could apply similar survey and interview methods across schools with different demographics to broaden generalizability. Longitudinal studies can also be conducted to analyze change in polarization over time to potentially identify a more causal relationship between peer group and polarization. Future work isolating belonging as a variable and connect that to affective polarization either directly or through multidimensional studies could deepen understanding of the framework.
Research on adult polarization has yet to produce interventions that actually reduce democratic resentment (Voelkel et al., 2022), suggesting that prevention during adolescent political socialization - when beliefs are still forming - may be more effective. Understanding how polarization emerges in the first place is a necessary step toward developing solutions that decrease political polarization to protect our democracy, people, and livelihoods.
Acknowledgements
I would like to thank Dr. Austin Bryan for providing me with guidance throughout my research journey, Mrs. Michelle Jedlicka for helping with IRB approval, and Dr. Yang-Yi Fan for helping with statistical analysis and always supporting me through the process.
Conflict of Interest
The author declares no conflicts of interest.