Like-minded sources on Facebook are prevalent but not polarizing

Abstract

Many critics raise concerns about the prevalence of 'echo chambers' on social media and their potential role in increasing political polarization. However, the lack of available data and the challenges of conducting large-scale field experiments have made it difficult to assess the scope of the problem^1,2. Here we present data from 2020 for the entire population of active adult Facebook users in the USA showing that content from 'like-minded' sources constitutes the majority of what people see on the platform, although political information and news represent only a small fraction of these exposures. To evaluate a potential response to concerns about the effects of echo chambers, we conducted a multi-wave field experiment on Facebook among 23,377 users for whom we reduced exposure to content from like-minded sources during the 2020 US presidential election by about one-third. We found that the intervention increased their exposure to content from cross-cutting sources and decreased exposure to uncivil language, but had no measurable effects on eight preregistered attitudinal measures such as affective polarization, ideological extremity, candidate evaluations and belief in false claims. These precisely estimated results suggest that although exposure to content from like-minded sources on social media is common, reducing its prevalence during the 2020 US presidential election did not correspondingly reduce polarization in beliefs or attitudes.

Fig. 1. The distribution of exposure to content among Facebook users.

a, The distribution of the exposure of monthly active adult Facebook users in the USA to content from like-minded sources, cross-cutting sources, and those that fall into neither category in their Facebook Feed. Estimates are presented for all content, content classified as civic (that is, political) and news. b, Cumulative distribution functions of exposure levels by source type. Source and content classifications were created using internal Facebook classifiers (Supplementary Information, section 1.3). Source Data

Fig. 2. Day-level exposure to content from like-minded sources in the Facebook Feed by experimental group.

Mean day-level share of respondent views of content from like-minded sources by experimental group between 1 July and 23 December 2020. Sources are classified as like-minded on the basis of estimates from an internal Facebook classifier at the individual level for users and friends, and at the audience level for Pages and groups. W1–W5 indicate survey waves 1 to 5; shading indicates wave duration. Extended Data Fig. 3 provides a comparable graph of views of content from cross-cutting sources. Note: exposure levels increased briefly on 2 and 3 November owing to a technical problem; details are provided in Supplementary Information, section 4.11. Source Data

Fig. 3. Effects of reducing Facebook Feed exposure to like-minded sources.

Average treatment effects of reducing exposure to like-minded sources in the Facebook Feed from 24 September to 23 December 2020. a–c, Sample average treatment effects (SATE) on Feed exposure and engagement. b, Total engagement (for content, the total number of engagement actions). c, Engagement rate (the probability of engaging conditional on exposure). d, Outcomes of surveys on attitudes, with population average treatment effects (PATEs) estimated using survey weights. Supplementary Information 1.4 provides full descriptions of all outcome variables. Non-bolded outcomes that appear below a bolded header are part of that category. For example, in d, ‘issue positions’, ‘group evaluations’ and ‘vote choice and candidate evaluations’ appear below ‘ideologically consistent views’, indicating that all are measured such that higher values indicate greater ideological consistency. Survey outcome measures are standardized scales averaged across surveys conducted between 4 November and 18 November 2020 and/or 9 December and 23 December 2020. Point estimates are provided in Extended Data Table 3. Sample average treatment effect estimates on attitudes are provided in Extended Data Fig. 4. All effects estimated using ordinary least squares (OLS) with robust standard errors and follow the preregistered analysis plan. Points marked with asterisks indicate findings that are significant (P < 0.05 after adjustment); points marked with open circles indicate P > 0.05 (all tests are two-sided). P values are false-discovery rate (FDR)-adjusted (Supplementary Information, section 1.5.4). Source Data

Extended Data Fig. 1. Distribution of predicted ideology score by self-reported ideology, party identification, and approval of former president Trump.

Each histograms displays the distribution of respondents’ predicted ideology score according to Meta’s classifier for Facebook U.S. adult users (see Supplementary Iinformation, section 1.3) by subsets defined by their self-reported political characteristics. The histograms have bins of width equal to 0.10. Source Data

Extended Data Fig. 2. Pre-treatment exposure to Facebook Feed content by source type: Study participants and daily Facebook users.

Pre-treatment distribution of Facebook Feed exposure to content from like-minded sources (left column), cross-cutting sources (center column), and those that fall into neither category (right column). Estimates presented for all content (top row) and for content classified as civic (i.e., political; center row) and news (bottom row). Source and content classifications were created using internal Facebook classifiers (see Supplementary Information, section 1.3). The graph includes the distribution of exposure for both study participants and the Facebook population of users age 18+ who logged into Facebook each day in the month prior to August 17, 2020, when the study sampling frame was constructed. Source Data

Extended Data Fig. 3. Day-level exposure to content from cross-cutting sources in the Facebook Feed by experimental group.

Mean day-level share of respondent views of content from cross-cutting sources by experimental group July 1–December 23, 2020. Sources classified as cross-cutting based on estimates from an internal Facebook classifier at the individual level for users and friends and at the audience level for Pages and groups (see Supplementary Information, section 1.3). W1–W5 indicate survey Waves 1–5; shading indicates wave duration. (Note: Exposure levels briefly decreased on November 2–3 due to a technical problem; see Supplementary Information, section 4.11 for details). Source Data

Extended Data Fig. 4. Treatment effects on outcomes for primary hypotheses.

Average treatment effects of reducing exposure to like-minded sources in the Facebook Feed from September 24–December 23, 2020. The figure shows OLS estimates of sample average treatment effects (SATE) as well as population average treatment effect (PATE) using survey weights and HC2 robust standard errors. Exposure and engagement outcome measures were measured using Feed behavior by participants. Survey outcome measures are standardized scales averaged across surveys conducted November 4–18, 2020 and/or December 9–23, 2020. Sample size and P values for each estimate are reported in Supplementary Table 47. Source Data

Extended Data Fig. 5. Treatment effects on outcomes for research questions.

Average treatment effects of reducing exposure to like-minded sources in the Facebook Feed from September 24–December 23, 2020. The figure shows OLS estimates of sample average treatment effects (SATE) as well as population average treatment effect (PATE) using survey weights and HC2 robust standard errors. Engagement outcome measures were measured using Feed behavior by participants. Survey outcome measures are standardized scales averaged across surveys conducted November 4–18, 2020 and/or December 9–23, 2020, unless indicated otherwise. Sample size and P values for each estimate are reported in Supplementary Table 47. Source Data