The fourth task of Work Package 6 in the RAYUELA project aims to clarify and refine the results obtained for the cybercrimes under consideration (affecting minors). Given the sensitivity of the subject matter, it is crucial to review and interpret the data-driven results and conclusions from the project through a multidisciplinary lens of expertise, including law enforcement agencies, psychologists, sociologists, educators, and more. This holistic approach is essential to clarify and refine the knowledge acquired. 

In this document, we combine the quantitative results obtained applying causality and Bayesian statistics to analyse the data gathered through the videogame and the results of previous work packages addressing human and technological factors. All these results are critically discussed with the other members of RAYUELA. This discussion occurred during an interactive workshop held in Zagreb (Croatia) on September 28 2023. 

The main objective of this document is to clarify the impact of the analysed human and technological factors in cybercrime affecting minors. Subsequently, the document also aim to provide recommendations and guidelines that may be useful for law enforcement agencies and policymakers. Our results suggest that considering the available data and technical limitations of the methodologies employed: 

• The RAYUELA serious game represents a helpful social science research tool to study the cybercrimes under consideration. 
• The data obtained through the serious game are relevant to explaining and predicting the risk of suffering/committing cybercrimes. 
• Demographic variables or those obtained through psychological tests do not have significant relevance when considered independently in any of the cybercrimes studied except in the case of cyberbullying offences, especially the variable indicating previous victimisation, which strongly influences the experiments. 
• Future research might also include risk taking behaviour to contribute to the overall findings. 
• Some of the methodologies used in the experiments to “interrogate” the model once it is trained are non-standard since we have had to adapt them to the specific needs of our problem. This implies that some results and conclusions could be distorted. 

These findings confirm the suitability of our chosen methodology, not just for predictive purposes but also for its prescriptive capability, which has been validated with expert input. However, it is imperative to reiterate the methodological limitations, emphasising the need to interpret these results with caution. We hope the insights gained through this analysis will be valuable in advancing our understanding of young cybercriminality and developing public policy safeguard minors from engaging in inappropriate online activities and improve their online experience