Evaluating a Game-Based Workshop for Youth Urban Design Using Open-Source Tools

Abstract

Game-based learning (GBL) has shown promise for engaging citizens in participatory urban planning, with Minecraft-based initiatives such as UN-Habitat's Block by Block programme reaching over 55 countries. However, evaluations of these initiatives remain predominantly qualitative, and few studies combine pre/post statistical testing with behavioural process data. In Japan, participatory urban planning follows the tradition of machizukuri, a bottom-up approach to community-driven town-making that has progressively incorporated design games and digital tools, yet rarely targets younger populations with rigorous outcome measurement. This paper presents the design, implementation and mixed-methods evaluation of an open-source game-based workshop for youth participatory urban design. The workshop integrates Japan's national 3D city model data (MLIT's PLATEAU), the Blender modelling environment and the Godot game engine into a mission-based co-design activity in which students redesign a real public park. Unlike Minecraft-based approaches, it uses real geospatial data at actual scale, supports both design-time placement and first-person exploration, and relies exclusively on free software. Fifty sixth-grade students (aged 11–12) at a public elementary school in Yokohama, Japan, participated in groups of 3–5. Five data streams were collected: pre/post Likert surveys (n=36 paired), post-only items, conversation transcription and classification (21,612 words across 12 groups) and map annotations (n=17) with open-ended responses (n=34). In this single-group design, Wilcoxon signed-rank tests indicated increases in civic participation willingness (p < 0.001, paired d = 0.66), career interest in urban design (p < 0.001, d = 0.89) and interest in the target site (p = 0.017, d = 0.35), while perceived tool difficulty did not change; the first two results remain significant under Bonferroni correction, the third does not. Conversation analysis showed a shift from operational to design-oriented discourse (chi-square, p < 0.001, Cramér's V = 0.17), the design-to-operation ratio rising from 0.82 to 1.96. Post-only items showed 90.9% agreement that the game mode was exciting. Because the study included no control group and students worked in clustered groups, these results describe short-term associations rather than causal effects. The study nevertheless demonstrates the value of multi-modal evaluation for small-sample serious game interventions.