Comparing Interactive and Panoramic Virtual Reality for Low-Carbon Old-Housing Retrofit Communication: Effects on Residents’ Adoption Intention and Willingness to Pay

Old-housing retrofit uptake may be constrained by limited understanding of concealed construction processes and household financial concerns. This study compared how two VR modes, an interactive experience and a panoramic video, were associated with residents’ retrofit knowledge, subjective perceptions, adoption intention, community support intention, and willingness to pay (WTP). The study’s originality lies in directly comparing task-based interactive VR with panoramic-video VR while keeping the principal retrofit content and process sequence broadly comparable and in integrating knowledge, psychological, WTP, and interview evidence to distinguish between-mode differences from associations among post-test constructs. The final analytical sample comprised 60 residents allocated to the two conditions through a non-random, investigator-administered, pretest-informed procedure. Pretest and post-test surveys, exploratory partial least squares structural equation modelling with 5000 bootstrap resamples, ordinal WTP analyses, and 37 semi-structured interviews were used. No consistent between-mode advantage was observed in retrofit knowledge, perceived operational transparency, adoption intention, or community support intention. In the exploratory structural models, the assigned VR intervention mode did not significantly predict perceived operational transparency, perceived value, task-related cognitive load, or adoption intention, and none of the tested indirect effects involving intervention mode was statistically significant. The interactive group reported higher post-test WTP in the unadjusted comparison, but neither within-group change nor the between-group difference in change was statistically significant. The adjusted group coefficients were also positive, indicating higher odds of reporting a higher WTP category in the interactive group, but they were not statistically significant in Model 1 (OR = 2.633, 95% CI [0.911, 7.608], p = 0.074) or Model 2 (OR = 2.905, 95% CI [0.998, 8.451], p = 0.050). The interaction between household income and post-test perceived operational transparency was not significant. Interview participants reported that the VR experiences made some concealed material layers and retrofit procedures easier to understand, while concerns about construction quality, household expenditure, disruption, and neighbourhood coordination remained. The findings distinguish associations among post-test psychological constructs from differences between the two VR intervention modes. Because the study did not include a non-VR control group, the observed pretest–post-test changes cannot be attributed specifically to VR. The mediation findings should be interpreted as exploratory given the modest sample size.