New Publication: What Drives Public Attention to Victims of Police Violence on Social Media?

By /

I’m happy to share that our new article, “Waves of Attention to Racial Injustice on Social Media: Extrajudicial Police Killings in the United States as Focusing Events”, is now published open access in Social Science Computer Review.

The study addresses a crucial yet underexplored question: Why do some victims of police violence become powerful symbols of protest—while others remain publicly invisible?

We examined over 1.5 million tweets related to 795 extrajudicial police killings in the U.S. (2015–2016), using the Washington Post Fatal Force database as our source of events. Our goal was to identify the factors shaping whether a killing receives attention on Twitter/X—and how much attention it gets.

Data on 1952 fatal encounters from the Fatal Force Database curated by the Washington Post. Distribution of events in terms of Twitter coverage: events without tweets in 10% Twitter Gardenhose sample, events that received at least one tweet, and events that received above-average coverage in 2015 and 2016.

Two processes: Thresholding and Focusing

We conceptualize public attention as the result of two distinct but connected processes:

  • A thresholding process, which determines which victims are tweeted about at all. Here, we find that victims who are Black, female, or very young are significantly more likely to receive attention. Victims who were unarmed or did not clearly threaten the police also received more visibility.
Logistic regression estimates (log odds) for the likelihood that a victim’s name is tweeted, which we conceptualize in the paper as a thresholding process.
  • A focusing process, which determines how intense the wave of attention is. This is shaped by several factors, including the timing of the first tweets (waves are bigger if first tweets occur on Fridays), the population size of the city where the killing occurred (with attention peaking in medium-to-large cities), and the early involvement of elite accounts, including media and activists. We also found that moderate use of hashtags and visuals boosts attention, while overuse can diminish it—a pattern suggesting a platform-specific “Goldilocks effect.”
Parametric coefficients of the GAM models for the categorical variables, which we conceptualize in the paper as a focusing process.

For the analysis, we employed—besides common Logistic Regression—Generalized Additive Models (GAMs), which is a flexible, nonlinear modeling strategy that is still rarely used in this domain. This approach allowed us to detect nonlinear and interaction effects that would be difficult or impossible to capture with conventional linear models.

Partial effects of the nonlinear terms in the GAM models.

Broader Implications

Our findings show that digital activism is shaped by sociocultural and platform-driven dynamics. Who gets seen—and how much—depends on the interplay of social meanings, media practices, and the structural logic of platforms like Twitter.

This publication is the result of an international collaboration among excellent scholars from five institutions across Europe and the U.S., and is freely available here: https://journals.sagepub.com/doi/10.1177/08944393251364290

Citation: Waldherr, A., Righetti, N., Gallagher, R. J., Klinger, K., Stoltenberg, D., Kumar, S., Ridley, D., & Foucault Welles, B. (2025). Waves of Attention to Racial Injustice on Social Media: Extrajudicial Police Killings in the United States as Focusing Events. Social Science Computer Review, 0(0). https://doi.org/10.1177/08944393251364290

Scroll to Top