Who’s to fear? Implicit sexual threat pre and post the “refugee crisis”

The perceived sexual threat of ethnic outgroups has been argued to contribute to anti-immigrant attitudes within societies. The current study investigates to what extent news media might contribute to such negative outgroup perceptions by analyzing implicit sexual threat associations in Dutch news. The study draws on a sample of more than two million news articles published between 2008 and 2018 in five major Dutch newspapers. To identify implicit bias in this corpus, we use word embeddings—an advanced language modeling technique in Natural Language Processing (NLP). The results show that ethnic outgroups (i.e., nationalities that were most prominently represented among asylum applications in the Netherlands during the refugee crisis) are associated more strongly with sexual threat than ethnic ingroups. These findings proved robust across a diverse set of outgroups and ingroup nationalities and names, and point to the existence of considerable implicit bias in the coverage of ethnic minorities. Moreover, the sexual threat associated with Arabic names has grown stronger since the “refugee crisis”. Such implicit biases reflect and potentially reinforces individuals’ implicit associations between ethnic minorities and sexual threat, potentially explaining growing anti-immigration attitudes, especially since the refugee crisis.