Using virtual reality to promote outdoor walking on a college campus

Abstract

Walking carries many individual and community-level benefits, including physical and mental health effects, climate action, community building, and equitable transportation options. When designing for walkability, transportation planners are underserved by the traditional design process and have voiced a need for more rigorous research. Common research approaches into built environmental factors of walking struggle to address local context and causality. As a research method, virtual reality (VR) has the ability to provide this much-needed site-specific experimental data. The purpose of this dissertation is to use VR to pretest potential changes to the built environment on a college campus in order to promote safe walking. In two studies, this dissertation (1) assesses the feasibility and acceptability of a novel VR simulation and study protocol examining the effects of street modifications to a real-world university campus and (2) pretests two potential interventions under consideration by decisionmakers. Study 1 details the development of a VR simulation depicting a 104,000 m2 section of Colorado State University’s (CSU) campus. Collaboration with CSU Transportation and the City of Fort Collins aided in identifying appropriate features to render in VR and in designing a study protocol which addresses safety concerns in the area. This protocol and the simulation are piloted in a feasibility and acceptability study (n = 21) following the guidelines of the CONSORT standards’ extension to feasibility studies. The pilot study’s objectives are to (1) detect issues with the VR simulation and protocol and (2) monitor induction of cybersickness and presence. The pilot data met prespecified criteria for missingness, cybersickness, and presence but failed to pass the study duration criterion, indicating that efficacy study (Study 2) sessions should be scheduled for a longer duration. Pilot data revealed limited variability in behavioral measures and a strong possibility of learning and differential carryover effects confounding the investigated factors. Solutions to these unexpected issues are discussed in Study 1 and implemented in Study 2. Study 1 centers reproducibility by outlining a VR development workflow and study design recommendations for academic-governmental collaboration. Thus, Study 1 will support additional universities and municipalities interested in conducting similar research to efficiently alter their built environments to promote walking for transport. Study 2 used this VR simulation to quantify the effects of two types of separated bike lane on pedestrians, a notable gap within active transportation literature. Specifically, this study investigates how perceptions of Alfonzo’s four environment-related walking needs (accessibility, safety, comfort, & pleasurability), walking routes, and community opinion are affected by the City of Fort Collins and CSU Transportation’s intervention options: (1) two new marked crosswalks and a separated bike lane created with flexible delineators and (2) two new marked crosswalks and a separated bike lane created with concrete barriers. To do so, 75 university students and employees who frequently visit the intervention area participated in a within-subjects protocol. Over 12 trials, they were asked to repeatedly cross a virtual campus street depicting the two interventions and the as-built conditions. At the end of each trial, participants self-reported their perceptions of walking needs. Multilevel structural equation modeling indicated that both interventions boosted perceived safety and comfort compared to the as-built condition when accounting for control variables; however, the flexible delineator condition also enhanced perceptions of accessibility and pleasurability. Furthermore, the four assessed walking needs created a within-level latent variable, representing exposure-based audiovisual walkability perceptions, which provides additional support for Alfonzo’s walking needs model. Both interventions resulted in higher values of the exposure-based audiovisual walkability perceptions variable. Based on multilevel logistic regression models, the contiguous concrete barrier condition resulted in the highest odds of using a marked crosswalk although the flexible delineator environment also showed improvement compared to the as-built condition. In a post-VR survey, participants showed an overwhelming preference for the flexible delineator environment, noting in free response items that it promoted safety while preserving a pleasant, open environment and allowing free movement of road users. Discrepancies between data collected in-VR and post-VR are discussed with reference to salience and recall bias. Thus, this study highlights the value of simulation-based research enabling real-time data collection in a controlled setting. This study progresses walkability research through reinforcing the influence of street-scale interventions and addressing a gap in the literature regarding a common type of safety countermeasure. Lastly, this dissertation provides an actionable recommendation to local decisionmakers based on site-specific causal evidence for the design of CSU’s campus.