Abstract / Summary
Conventional fire-safety awareness campaigns built on brochures, leaflets, and instructional videos have shown limited success in developing transferable hazard-recognition skills, and simply moving this content into a virtual reality (VR) headset does not resolve the problem. This Project Report synthesizes fifteen cross-cutting lessons learned from a three-phase FireGuard case study conducted in Dubai and Sharjah, spanning a 149-resident needs-assessment survey, an expert-validated True-Context (TContext) game design, and an 80-participant controlled experiment in which the contextual group achieved higher learning gains and outperformed a non-contextual VR baseline on usability, engagement, and motivation. Rather than presenting new experimental data, this report integrates findings from three reported phases into a single cross-phase account and an accompanying six-layer evaluation architecture. The fifteen lessons are organized into three clusters. The Phase 1 survey yields three diagnostic lessons that show a clear distinction. Awareness is not preparedness. Passive delivery persists for cost reasons, not effectiveness. Training access remains structurally unequal and employer-dependent. The TContext framework and Phase 2 expert review yield a further set of design lessons: Passive content does not become contextual simply by appearing inside a headset, Contextual design - more than immersion alone, was associated with the larger learning gains observed. These design lessons also point to the need for explicit temporal and situational escalation. Distributed situational awareness and progressive interactivity should be built into scenario design. Gamification should function as a teaching tool rather than a descriptive layer. Structured, multi-domain expert review before user trials is likewise associated with reduced redesign costs. Phase 3 and the integrative discussion yield five lessons on evaluation and deployment. Balance interaction realism against accessibility for non-technical users. Sustain motivation through real-time, scenario-anchored feedback. Isolate the independent variable through rigorous experimental design. Adopt a six-layer evaluation architecture, spanning reach to durable behaviour change. Data quality is shaped through field logistics and human factors. Together, these fifteen lessons offer practical, theory-grounded guidance. They are intended for researchers, safety educators, and policymakers designing similar VR interventions in other high-density, multilingual urban settings. However, their transferability beyond the present study has not been directly tested.