“Defriending” in a polarized age: Political and racial homophily and tie dissolution

Transformation of social relationships in COVID-19 America: Remote communication may amplify political echo chambers

“…In addition, increasingly partisan elections frame important matters as “political matters,” and such a framing was likely amplified during the 2020 election because the response to the pandemic immediately became highly politicized and politically divisive ( 44 ). Earlier works show that individuals are more inclined to activate politically similar ties, reduce family time during holidays such as Thanksgiving, and even disengage from politically dissimilar friendships in politicized situations such as contested elections ( 11 , 13 , 45 ). In our future work, we will investigate how polarizing events and politicization of the pandemic responses shape racial and political homophily.…”

“…On the other hand, research shows that a sense of social isolation could arise from the politicization of topics considered "important matters" under heightened political polarization (10) through the process of political echo chambers within core relationships (11). During the 2016 US presidential election, Americans discussed important matters with a small number of confidants who share similar political views (12) and cut their close relationships with politically dissimilar others (13). This trend raises concerns, as traditional offline networks were believed to nurture political disagreement, essential for fostering a democratic society, even amidst the growth of online echo chambers (14).…”

“…Although concerns could be partially mitigated by using wider prediction intervals estimated from our benchmark data that include a diverse range of prepandemic datasets, it is important to note that our COVID-19 network data were collected through opt-in, nonprobability panels conducted exclusively online. Second, while we have applied statistical methods to estimate linear trends as benchmarks, existing literature indicates that network patterns, particularly political homophily, can fluctuate in different contexts, influenced by political events occurring at various times within the same year, such as elections (10,11,13,14). However, considering nonlinear trends might lead to overfitting due to the data limitation, and the benchmarks incorporating data collected at various moments would result in larger confidence intervals in case the assumption of a linear trend is not valid.…”