Divergent patterns of engagement with partisan and low-quality news across seven social media platforms

Summary

Mosleh, Allen, and Rand offer a corrective to a literature that has long treated Twitter/X (and to a lesser extent Facebook) as a stand-in for “social media” writ large. By comparing user engagement with news content across seven platforms, they show that the relationships between engagement, partisan slant, and source quality are not uniform — they diverge meaningfully from platform to platform. The argument is both empirical and methodological: claims about misinformation and partisan news circulation built on single-platform data risk generalizing patterns that are in fact platform-specific.

Key Contributions

• One of the first systematic, like-for-like comparisons of news sharing across seven social media platforms rather than one or two.
• Empirical evidence that engagement patterns with partisan and low-quality news are platform-dependent rather than a universal feature of “social media.”
• A methodological warning against extrapolating from Twitter/X or Facebook studies to the broader, increasingly fragmented platform ecosystem.
• Groundwork for theorizing how platform affordances shape the visibility and reach of partisan and low-quality news.

Methods

The authors conduct a comparative analysis of engagement data drawn from seven social media platforms. For each platform, they link engagement metrics on shared news content to two domain-level attributes: political slant and a quality/reliability rating. This allows direct cross-platform comparison of how partisan and low-quality domains fare in user engagement.

Findings

• Engagement with partisan news varies substantially across platforms rather than following a single pattern.
• Engagement with low-quality news likewise diverges across platforms.
• The contrasts imply that conclusions drawn from any single platform — particularly Twitter/X — do not straightforwardly transfer to others.
• Platform-level differences are large enough to matter for both scholarly inference and policy debate about misinformation.

Connections

This work speaks directly to cross-platform efforts to map misinformation and partisan content, complementing Bollenbacher2026-vz and DeVerna2025-dl on multi-platform information dynamics, and extending the Facebook-centric findings of Gonzalez-Bailon2024-rq and Bakshy2015-rn as well as the broader misinformation-exposure accounting in Budak2024-ef. It also intersects with platform-comparative work on alternative and fringe ecosystems such as Efstratiou2025-gs and with research on partisan news sharing behavior like Rossini2026-jn.