Role-Contingent Visual Decision Support for Age-Friendly Furniture in Conversational Recommendation Interfaces: Appraisal Pathways to Trust and Decision Confidence

Visual information in conversational recommendation interfaces can function as decision evidence, but its value may depend on decision role. We tested this in a 2 × 3 between-subjects experiment with 210 participants, crossing self versus proxy decisions with product-only, home-context, and home context with an older-user cue. Trust showed a role-contingent reversal: home context produced the highest mean for self decisions, whereas the older-user cue produced the highest mean for proxy decisions. The Role × Visual interaction remained significant after Holm correction. Decision confidence showed a visual-presentation main effect but no role interaction. Appraisal analyses indicated two selective pathways. Contextual presentation was associated with recommendation verifiability, with supported indirect associations for trust and confidence. The older-user cue showed a role-contingent association through target-user suitability, driven by a negative self-decision pathway. These findings support a role–representation fit account and suggest aligning visual evidence with the decision task and represented user.