User-Articulated Associations in VR: A Co-Occurrence Network Map of Presence, Realism, Agency, and Time

Research on VR often isolates constructs on separate scales, obscuring how users themselves link presence, agency, realism, fictionality, temporal structure, and embodiment during play. We address this gap with a qualitative, participant-level mapping of user-articulated associations across nine constructs defined a priori from theory and standardized for analysis: presence (PI; feeling "there"), visual realism, physical realism, fictionality, agency, temporal structure, narrative comprehension, attentional guidance, and embodiment. Fifty-seven design-trained participants played two commercial VR games (A Fisherman's Tale; Moss on Quest 2/3) and answered openended prompts. Using codebook thematic analysis, we binarized construct mentions per participant and computed normalized co-occurrences (Jaccard) with bootstrap 95% CIs and permutation baselines; metrics are descriptive, valenceagnostic, and non-causal. Findings: Presence co-occurred most with layered attentional guidance and embodiment, exceeding its association with visual fidelity. A"design-side" cluster linked visual and physical realism with temporal pacing and guidance; visual realism clustered more with fictionality and narrative discourse than with presence. These patterns refine theory toward coherence, guidance, and embodiment-centered pathways to sustaining place illusion and engagement, consistent with claims that internal consistency and sensorimotor contingencies outweigh photorealism. Methodologically, the participant-level, normalized map makes qualitative associations inspectable and comparable via effect sizes with uncertainty, complementing solo-construct scales and supporting reuse on other construct sets. Practically, the results recommend prioritizing consistent world rules and physically plausible interactions, scaffolding agency with multimodal diegetic cues, and providing robust, synchronous self-representation.