Online abuse in sport is often most visible when it is directed at individual athletes. But those cases are only part of the picture. Every day, fans react to performances, results and decisions across online communities, shaping a much wider climate of conversation around sport. By analysing discussions on Flashback and Reddit, researchers at Stockholm University are examining that wider environment to better understand the patterns of harmful and threatening communication that emerge around sport.
Inside the Research is UAOA’s Erasmus+ Researcher Interview Series, spotlighting the researchers contributing to the European Union co-funded programme Developing a Pan-European Framework to Combat Online Abuse in Sport. Each edition looks inside one national study to explore the core questions being asked, the emerging evidence and what the findings could mean for the future of sport.
For Lisa Kaati, Associate Professor and Amendra Shrestha, PhD, understanding online abuse means examining the wider conversations in which it appears.
Lisa Kaati is a Senior Lecturer and Associate Professor at Stockholm University, specialising in harmful and threatening online communication, digital hate, threat assessment and violent extremism. Amendra Shrestha, PhD, is a computer scientist whose research applies machine learning and computational linguistics to hate speech, threatening communication and violence assessment.
Together, they apply computational methods to analyse discourse in online sport communities. Their study, Online Abuse Targeting Athletes: A Computational Analysis of Hateful Language and Threats on Flashback and Reddit, separates targeted hate speech from general profanity and ordinary criticism. By doing so, the research provides a clearer picture of the extent and nature of athlete-targeted abuse and, through its comparative phase, will examine how these patterns may vary across sports and platforms.
Below, we discuss the preliminary findings from the first part of the research, covering the analysis completed of the Motorsport forum on the Swedish platform Flashback. A broader comparative analysis of motorsport and fifteen other online sport communities on Reddit is now underway.
Looking beyond individual incidents
Individual cases reveal the occurrence of online abuse and the potential severity of its consequences. However, they are less suited to measuring the overall frequency of harmful communication within the wider volume of everyday sport discussion.
Online discussion forums and communities provide a different perspective.
For the Flashback study, the researchers analysed the complete Motorsport forum, comprising 1,020 threads and 60,188 posts spanning more than two decades.
Confirmed targeted abuse was identified in 2.6% of all posts, equivalent to approximately 1 in every 38 posts. Within this specific dataset, abusive content represented a minority of the overall conversation, while remaining sufficiently present for distinct patterns to emerge.
Distinguishing criticism from abuse
A central challenge is determining what should, and should not, be classified as abuse. Sport naturally generates criticism: a negative assessment of a driver’s performance or disagreement with a sporting decision is not automatically abusive.
The research therefore distinguishes ordinary criticism and untargeted profanity from communication directed at an identifiable person or group that becomes demeaning, threatening or identity-based. It also includes a category particularly relevant to sport: competence- or performance-based abuse.
Within the Flashback dataset, competence- or performance-based abuse was the most prevalent category, appearing in 68.9% of confirmed abusive content. Racist or ethnic abuse appeared in 13.6%, body-shaming in 12.3%, violent threats in 8.2%, dehumanising language in 7.2%, and sexist or misogynistic abuse in 4.5%. Other discriminatory content, sexual harassment and homophobic abuse were also identified.
Because different forms of abuse can occur together, these figures should not be read as mutually exclusive categories. Notably, dehumanising language tended to occur alongside racist or violent content rather than in isolation, suggesting that some of the more severe forms of abuse can cluster within the same communication.
The researchers interpret these findings within the culture of motorsport itself: a highly competitive, historically male-dominated environment in which drivers and their performances are continuously evaluated. This context may help explain the prominence of competence-based abuse.
The analysis also identified gendered insults targeting male drivers perceived as insufficiently competitive. This pattern aligns with broader research demonstrating how insults are weaponized to enforce traditional expectations of masculinity in male-dominated online spaces.
The findings highlight why online abuse cannot be understood through vocabulary alone: context, target and usage all matter, and the frequency of a category does not necessarily reflect the severity or nature of the harm it represents.
At this stage, the Flashback analysis does not identify which individual athletes or groups were targeted most frequently. The completed work provides evidence on the forms and prevalence of abusive content, while systematic analysis of specific targets remains a separate area for further investigation.
From depth to comparison
The second phase substantially broadens the research. Kaati and Shrestha have collected 213,625 Reddit posts and nearly 16 million comments from sixteen online sport communities between 1 January and 31 July 2026, covering sports including soccer, Formula 1, tennis, rugby, gymnastics, cycling, Formula E and WRC.
The two datasets offer complementary perspectives. Flashback provides depth within one motorsport community and a specific linguistic and national context, while Reddit provides breadth, allowing comparisons across sports, fan cultures and community structures.
To analyse this much larger dataset, the researchers use a staged process combining automated toxicity screening, control samples and comparable samples of around 10,000 comments per subreddit. This helps account for content the initial screening may miss and reduces the risk that differences in community size distort comparisons.
The aim is not simply to identify offensive language at scale, but to examine what forms of targeted abuse appear, how frequently they occur and how patterns differ across online sporting communities. The presence of motorsport-related communities on Reddit may also help explore how similar sporting contexts differ across platforms.
The Reddit analysis is still underway, so findings from Flashback should not yet be assumed to apply across other sports or platforms. However, the comparative analysis will help identify similarities and differences in the patterns and extent of online abuse in motorsport compared with other sport disciplines.
What the findings mean for sport
The completed Flashback analysis has a clear practical implication: different forms of harmful communication may require different responses.
Within this dataset, competence- and performance-based abuse accounted for the majority of confirmed cases. Racist abuse, violent threats and sexualised or gendered harassment occurred less frequently, but represent distinct categories of harmful behaviour that may warrant different levels of safeguarding, moderation and support.
Grouping all flagged content under a single measure of “toxicity” risks obscuring these distinctions.
For sport organisations, governing bodies, moderators and athlete-support teams, more detailed classification could help identify where attention and resources are most appropriate and urgent. General hostile commentary may raise different considerations from racist abuse or a direct threat of violence.
The broader Reddit analysis will examine whether the prevalence and characteristics of these categories vary across different online sport communities. If such differences are identified, they could provide a stronger evidence base for more context-specific approaches to prevention, moderation, reporting and athlete support.
The Flashback study already points to one important conclusion: Measuring online abuse is not only a question of how much harmful content is present. Meaningful analysis also requires distinguishing what form it takes and the context in which it occurs.
Key takeaway
Kaati and Shrestha’s analysis shows that targeted abuse formed a small but persistent part of the Flashback Motorsport conversation, appearing in 2.6% of more than 60,000 posts.
Almost seven in ten confirmed cases involved competence- or performance-based insults. Alongside this dominant category was a smaller layer of identity-based and violent content, including racist and ethnic abuse, body-shaming and threats of violence.
The findings demonstrate why volume alone cannot determine how online abuse should be understood or addressed. Effective responses depend on distinguishing between different forms and levels of harm.
