Factors Associated With Risky Drinking Decisions in a Virtual Reality Alcohol Prevention Simulation: Structural Equation Model.

Background: Risky alcohol consumption among adolescents is a significant public health concern in most Western countries. Various motives and factors (eg, sensation seeking, gender, reduced self-efficacy) known in the literature are associated with risky drinking decisions in real life. Efforts to tackle risky drinking decisions in real life through skills training to deal with social pressures have been successful. However, interventions of this nature require significant resources. Technological solutions, such as virtual reality (VR), offer advantages, as they enable immersive experiences that replicate real-life scenarios. However, a question persists pertaining to the fidelity of real-world behaviors within virtual environments. Objective: This study is exploratory and aims to ascertain if the established drinking motives and factors for risky drinking decisions are transferrable to the virtual environment in the simulation game VR FestLab and to uncover determinants linked to risky drinking decisions within the simulation. Methods: The study analyzed data from the intervention arm of a cluster-randomized study of 161 Danish students aged 14-18 years who tested the virtual alcohol prevention simulation VR FestLab. At baseline and before playing VR FestLab, independent variables such as age, gender, alcohol consumption, use of other drugs, sensation seeking, drinking refusal skills, knowledge of blood alcohol concentration, and refusal communication skills were recorded. The dependent variable, virtual risk decisions, was measured immediately after the gameplay. Confirmatory factor analysis and structural equation modeling were used to examine the latent variables in relation to virtual risk decisions. Moderation analyses for age and gender in relation to the latent characteristics and the primary outcome were also conducted. Results: The data indicate that 73.9% (119/161) of the participants engaged in binge drinking at least once in their lifetime. The confirmatory factor analysis demonstrated a good fit of the items for their respective constructs; therefore, they were adopted without modification in the structural equation model. The data suggest that individuals with prior alcohol experience are 4 times more likely to engage in virtual risk decisions within the simulated environment (odds ratio 4.31, 95% CI 1.70-10.84; P=.01). Knowledge and awareness of blood alcohol concentration were associated with a lower chance to engage in virtual risk decisions (odds ratio 0.32, 95% CI 0.11-0.93; P=.04). However, no significant associations were found between virtual risk decisions and other latent variables. Gender and age did not moderate the associations. Conclusions: The immersive and lifelike properties of VR partially reflected risk-related decisions. However, it remains unclear which factors favor the mapping of real-world behaviors in virtual simulations. Therefore, future research should address the mechanisms underlying behavioral dynamics in virtual simulations and explore the translation of virtual behaviors into real behaviors to gain a comprehensive understanding of the potential of virtual simulations for alcohol prevention.