UCL Research Domains


Outcome Report: The Social and Physical Settings of Ethnic Hate Crime

2018-19 Social Science Plus Pilot Project (£10,800)

Outline of project 
The aim of this project was to examine situational factors that affect bystander intervention in ethnic hate crime. Specifically, we sought to assess (1) how the presence of other ethnic in- or outgroup members influences willingness to intervene and (2) whether intervention differs for offences in settings that are either ‘typical’ for the victim or perpetrator. Following a pilot study we constructed a multi-dimensional measure of ‘willingness for intervention’. We then conducted two online experiments in which participants reported willingness to intervene in a verbal attack by a majority-group perpetrator on a minority-group victim. The offence occurred either in a setting prototypical for the perpetrator, the victim, or a setting that was not specified (control group). Study 1 tested main effects of settings; Study 2 was a direct replication and considered possible explanatory processes. No main effects of settings were identified, which restricted our ability to develop an agent-based model (ABM). Indeed, a key insight has been that research and theories in the domain of bystander intervention, including our own work, are only partially apt to specify decision making processes - how and why action is taken - which is needed to define an agent-based model. In an effort to further elaborate on the behavioural model that is the basis for the ABM, we ran two large-scale online experiments as intended, examining the combined effect of settings as well as the presence of in- and out-group bystanders.

Main findings
Across four experiments we showed that settings of ethnic hate crimes - specifically, whether it is a setting that is prototypical for the minority-group member victim or the majority-group member perpetrator - has no direct effect on majority-group members’ willingness to intervene in the incident. This pattern was found for confrontational interventions, victim-oriented intervention, as well as passive intervention. Indeed, while previous research considered intervention as a one-dimensional construct, our studies show that different types of intervention must be distinguished. Importantly, the observed effects take into account bystanders’ beliefs and attitudes about the minority-group victim. While Study 1 and 2 only assessed the main effect of settings, Study 3 and 4 investigated the interaction between settings and the presence of fellow bystanders. Results replicated the well-established bystander effect - people are less willing to intervene when others are present - in a setting that is prototypical for the perpetrator but not in settings that are typical for the victim. An exploration of potential mediators showed that in particular victim-oriented intervention was less likely in the former context; intervention intentions were predicted by perceptions that such action would be risky and lead to repercussions from the perpetrator. These insights highlight that it is important to build strong norms in support of intervening in hate crime in order to facilitate bystander intervention. Findings also point to the application to the online context; this rationale is further explained below.

Key achievements and impacts (academic and non-academic), media coverage, etc.
We hosted a workshop on Agent-based Modelling for Social Scientists. Julia Eberlen (project partner), Tristan Caulfield (Co-I), and Amy Thornton (UCL, Department of Security and Crime Science) presented examples of the application of ABM as well as an introduction to the use of NetLogo software. Twenty-five participants attended, with diverse backgrounds, for example, in public health, terrorism studies, health psychology, and urban planning. Participants’ feedback demonstrated that there was/is a notable interest in this topic, especially its practical application and use of the software NetLogo. Following our call for participants, Tom Couch from the Research IT Services department in ISD attended to understand existing demand and needs in the domain of agent-based modelling. It is unclear whether UCL is now hosting its own workshops. All workshop materials were made available to participants and are now also hosted here as a public resource for beginners. Dr Schumann also presented the key findings of the three experiments at an internal seminar of the project partners at Université Libre de Bruxelles. The methodological approach of the original project was also discussed by Dr Schumann in the Introduction to Research module of the BSc in Crime Science. The four experiments will be presented in a paper that is in preparation for publication in the Journal of Interpersonal Violence. Lastly, our positive experience with this funding scheme encouraged our colleagues Toby Davies and Aiden Sidebottom to apply for a Social Science Plus grant; they were also successful in receiving support for their research

Detail your plans for external funding application(s) 
A key conclusion from the studies is the applicability of the insights to the online context, specifically, users’ intervention in incidents of ethnic hate speech on social media. In line with the presented findings, people should be more willing to intervene online than offline as doing so may be considered less costly and risky (i.e., due to anonymity). At the same time, while perceived risk is lowered, perceived responsibility may be reduced as well and intervention, thus, be reduced. Understanding these dynamics is crucial for developing policies for online platforms that tackle the prevalence rate of hate speech.  Following this rationale, we (PI Sanaz Zolghadriha, Co-I Sandy Schumann, Co-I Paul Gill) developed an application for the Digital Citizen Contribution Program of the Canadian government, Canada Heritage. We applied for funding for a two year project on 1.11.19, requesting CAD$ 1 090 911.60. We are awaiting the outcome.  Should this application not be successful, we will - after revising the proposal - apply for a research grant with the Economic and Social Research Council. This bid will likely be smaller, as it would include fewer public engagement activities.

Next steps
In the short-term future, Dr Caulfield and Dr Schumann aim to conduct a follow-up experiment with more nuanced assessment of participants’ decision making. Doing so, we aim to identify at which stage of the decision making process concerns of costs and riskiness affect subsequent choices to intervene in an ethnic hate crime. Results will allow more targeted proposals on how to encourage bystanders to intervene, or rather, when they need extra encouragement to overcome inhibitions to act. These insights will also offer the kind of evidence necessary to specify an agent-based model, which we then seek to develop.