Where does online hate happen?
By Daria Denti
Online hate and places
Despite being a digital phenomenon, online hate speech does not happen is a spatial vacuum. Digital hatemongers are grounded in a spatial environment which might influence their behaviors and beliefs. Nonetheless, the influence of places on cyberhate production is only rarely investigated.
Recent studies show that cyberhate does not happen with the same intensity across places, with some areas having more cyberhate production than others also when differences in resident population and Internet users are considered. The local level of crime, unemployment and social capital all relate to online hate (Bernatzky, Costello and Hawdon, 2021; Costello and Hawdon, 2018; Görzig, Milosevic and Staksrud, 2017; Kaakinen et al., 2018; Kowalski, Limber and McCord, 2019). A study in the US shows that US counties with higher levels of unemployment display higher shares of racist digital behavior (Anderson et al. 2020) and another research study identifies a causal link between the local spread of Covid-19 and cyberhate (Lu and Sheng, 2020).
Given this evidence, a better understanding of the influence of places on cyberhate matters also with respect to the policy debate. If some characteristics of places correlate to cyberhate production, then policy should also target these features to achieve an effective reduction in online hate.
So, it appears that we should investigate whether and how places can influence the observed pattern of cyberhate production.
Online hate and inequality
Among the spatial features worth analyzing with respect to their relationship with cyberhate, inequality is key. This is due to established research documenting that inequality is associated with many resentful behaviors. Places with high inequality display higher share of votes for anti-immigrant parties in US and Europe (Burgoon et al., 2019; Engler and Weisstanner, 2020), as acknowledged by the “revenge of places that don’t matter” literature (Rodríguez-Pose, 2018, 2020). Other works show that local inequality is strongly associated to several discriminatory behaviors: school violence (Denti, 2021), racism, sexism, opposition to social welfare, violent acts against minorities (Kunst et al., 2017), and scapegoating (Wilkinson and Pickett, 2017).
What is still missing is an investigation measuring the impact of inequality on cyberhate production. In a recent paper, co-authored with Alessandra Faggian, we contribute to fill this gap by measuring the effect of the geography of income inequality on cyberhate production in Italy.
Cyberhate and inequality in the Italian context
Our investigation “Where do angry birds tweet? Income inequality and online hate in Italy” aims at measuring whether income inequality has a causal effect on cyberhate production. We designed a novel database containing data on cyberhate production and income inequality at micro-regional level for Italy that we use to perform an econometric estimation of the effect of income inequality on cyberhate. To get a measure of causation rather than correlation, we use an instrumental variable approach and we also include in the estimation other potential features that could influence cyberhate production, such as unemployment, crime, social capital.
We consider the geography of cyberhate and income inequality across the 611 Italian Local Labor Market Areas, sub-national functional areas based on commuting containing the bulk of the labour force living and working there. They are particularly suitable to our study because they alleviate the unfeasibility in detecting whether the tweets are posted during working time, commuting time or leisure time.
Italy is a particularly interesting country to study in terms of the inequality/cyberhate nexus. It is one of the most unequal countries in terms of income distribution among the Western European countries (Eurostat, 2019). It also experiences huge volumes of online hate: nearly 80% of Italian internet users have witnessed some sort of online hate speech (SWG, 2017).
We measure cyberhate using the corpus of Twitter geo-referenced data extracted through a system of algorithms designed by Musto et al. (2016) and used to design the Italian Hate Map, in turn inspired by the Humboldt University Hate Map which covers cyberhate production across US. Our database contains more than 75,000 tweets generated in Italy in 2017 that we have weighted with the total geotagged tweets generated in the same places to get a measure of how many tweets have hate content.
Then, we measure income inequality through the well-established measure given by the Gini Index and we estimate how much of the spatial variation in hate tweet production across Italy is explained by the geography of income inequality.
Inequality drives cyberhate production in Italy
Our analysis supports income inequality as a cause for hate tweet production. Income inequality as a significant effect in determining the production of hate tweets. The higher the economic inequality in an area, the more hate speech online increases. We also check if this result holds when we control for the influence of other spatial features that relate to cyberhate production, such as the number of migrants in the area, the geography of refugee hosting centers, the prevalence of non-hate and hate crime, social capital, political preferences, disrupted families, marginalized areas, and population density.
Our investigation provides further support to the key role of inequality in determining the rising patterns of resentful behaviors. This calls for policy initiatives capable of targeting inequality to improve inclusiveness and sustainability of communities.
We get other interesting additional findings. First, economic insecurity—measured by the share of workers who perceive their job to be precarious—is another risk factor for cyberhate. This result aligns to existing evidence showing that the high level of job insecurity characterizing the Italian labour market fuels people’s discontent (Modena, Rondinelli and Sabatini, 2014).
Second, places where individualistic values are prevalent are associated to higher occurrence of online hate. Third, we find that education does not act as a protective factor against cyberhate in unequal places. This result might seem surprising, as it appears reasonable that educated people are more capable of overcoming stereotypes. At the same time, this result aligns with existing evidence showing that inequality may trigger intolerance, including among educated people, threatening the perceived stability of social positions (Decelles and Norton, 2016; Jetten et al., 2017). Fourth, neither immigrants nor crime appear to be relevant in influencing cyberhate. These results support existing evidence from behavioral experiments showing that inequality triggers fears which alter people’s perception of different ethnicities, to the point of believing that minority ethnic groups are far larger than their actual size (Krosch, Tyler and Amodio, 2017; Kunst et al., 2017).
Contrast to cyberhate should consider the economic outlook of places and communities
The evidence provided by the paper supports the policy approaches which go beyond banning content publication to address also contextual features. Further, this evidence outlines the importance of considering the local level and its economic characteristics in designing policy to counter cyberhate. Clearly there are limitations, starting from the fact that our analysis does not capture the entire cyberhate production, being limited to geotagged hate tweets. Also, our findings refer to Italy, hence research on other countries is needed.
Daria Denti is a Post-Doc Researcher at the Gran Sasso Science Institute and a Visiting Research Fellow at the London School of Economics, Department of Geography and the Environment. Daria’s research lies at the intersection between economics, economic geography and human rights violations. Her PhD thesis “Essays on the economic geography of oppressive violent deviant behaviours” won the 2020 Best dissertation award from the Regional Science Association International.