European Parents' Risk Perception of Online Hazards: A Data Science Approach.

In this study, we explore the determinants of risk perceptions of European parents about online hazards concerning their children. The study is based on a multi country survey with a stratified sample of EU parents from eight countries (n=6400). We analyse the main domains of risks that emerge from data identifying three clusters of risks: persuasion risks, personal safety and financial ones. After, we model their determinants in terms of parents’ past experience online, several different type of their digital skills and their parenting style regarding access of online resources. Last, we explore countries differences using a novel statistical method from data science, model-based recursive partitioning, which represents a synthesis of a theory-based approach and a data-driven set of constraints to the model validation and further development. The latter method allows us to obtain a more nuanced picture of the differences across countries and of their sub populations in terms of risk perception and their determinants.