Preference clustering reveals divergent pathways for transfer-board design in a Living-Lab co-design study

Transfer boards demonstrate high abandonment rates despite mechanical simplicity, suggesting fundamental misalignment between current design approaches and user needs. Previous refinements have failed to improve adoption metrics over five decades.

To determine whether wheelchair users' transfer board preferences follow convergent or divergent distributions, and to identify distinct user segments with specific design priorities.

Mixed-methods Living Lab study combining qualitative focus groups and quantitative preference mapping. Twenty-eight wheelchair users (mean age 51 ± 12 years, 57% women) participated in semi-structured focus groups exploring usage patterns, limitations, and design preferences. Subsequently, 44 wheelchair users (mean age 48.2 ± 13.3 years, 52% women) completed a 31-item questionnaire measuring preferences across usage, materials, and enhancement dimensions. Hierarchical clustering with Ward's method identified user segments. Four-way ANOVA with Tukey post-hoc tests examined cluster differences.

Three distinct clusters emerged with divergent preferences. Cluster B1 (n = 8, 18%) demonstrated technology-embracing preferences (Technology subfactor: Δ= +1.70 ± 0.58 vs baseline, p < 0.001; Improvements: Δ= +0.66 ± 0.23, p < 0.001). Cluster B2 (n = 10, 23%) exhibited simplicity-focused preferences, rejecting technological features (Δ=-2.35 ± 0.68 vs B1, p < 0.001) and portability enhancements (Practicality: Δ=-1.36 ± 0.53 vs baseline, p = 0.002) while reporting fewer usage constraints (Δ=-1.50 ± 0.41, p < 0.001). Cluster A (n = 26, 59%) showed intermediate preferences without directional commitment. Significant cluster × factor interactions [F(4,1292)=11.36, p < 0.001] confirmed non-convergent preference structures.

Transfer board users distribute across three distinct segments with mutually exclusive design priorities rather than along a single sophistication continuum. These findings challenge the prevailing single-product optimisation paradigm and indicate that parallel development strategies targeting cluster-specific needs could reduce abandonment rates more effectively than iterative refinement towards universal solutions.