Translating parenting styles to driver monitoring: the development and validation of the driver monitoring system-user experience (DMS-UX) scale

Driver monitoring systems (DMS) are increasingly used to sustain attention during partially automated driving, but existing measures do not assess how drivers experience the supervisory tone of DMS feedback. Drawing on parenting-style theory, we developed the Driver Monitoring System User Experience (DMS-UX) Scale and tested whether systems varying in demandingness and responsiveness produced distinct user experience profiles. Across two samples, 501 licensed U.S. drivers completed a between-subjects online vignette experiment and were randomly assigned to an authoritarian, authoritative, permissive, or uninvolved DMS. Scale development began with 80 items in Sample 1 ( n = 319). Exploratory factor analysis and item response theory yielded a 16-item, four-factor solution, which was tested in an independent Sample 2 ( n = 182) using confirmatory factor analysis. Inspection of factor loadings and residual correlations led to the removal of one poorly performing item and a refined 15-item model with comparative and residual fit emerged. The observed factors were Perceived Usage Benefits, Perceived System Judgment, Perceived Salience and Clarity, and Perceived Supportiveness. Internal consistency was good (McDonald's ω = 0.770–0.915; Cronbach's α = 0.745–0.914), and indices supported internal construct differentiation. Confirmatory model fit was mixed but improved after refinement (CFI = 0.916, TLI = 0.894, RMSEA = 0.097, SRMR = 0.073), providing qualified support for the four-factor structure. Monitoring style had a large multivariate effect on DMS-UX scores, Pillai's trace = 0.481, F (12, 1488) = 23.65, p < 0.001. Styles differed in perceived judgment, salience and clarity, and supportiveness, but not perceived usage benefits. The DMS-UX offers an initial descriptive instrument for comparing drivers' experiences of DMS supervisory feedback, though behavioral, longitudinal, and cross-cultural validation is needed before using it to infer safety outcomes or identify an optimal monitoring style.