Examining Socioeconomic and Physiological Factors Affecting Preferences for Cycling Infrastructure Using Virtual Reality Experimentation

Abstract

Understanding cyclist behaviors and preferences is critical to improve safety and increase mode share. However, real-world cycling data is scarce and on-road testing presents safety concerns. This study presents a novel way of studying cyclists’ perceptions of bicycle infrastructure design alternatives using immersive virtual environments modeled after a real-world corridor, and a previously validated bike simulator at the University of Virginia, U.S. Three infrastructure scenarios were tested: sharing lanes with motor vehicles marked with sharrows (with no bike lane), a separated bike lane, and a protected bike lane with flexible delineators. Surveys elicited data on participants’ preferences and perceptions of safety, as well as demographic and socioeconomic data. Smartwatches collected heart rate (HR) data. A multinomial logit model examined the relationship between sociodemographic and physiological variables and preferences for bicycle infrastructure. Results suggest gender, age, and abrupt changes in HR affect cyclists’ preferences for bike infrastructure design. Overall, gender emerges as the most practically significant predictor variable for bicycle infrastructure preference, with men more likely to prefer sharrows and women more likely to prefer protected bike lanes. Exploratory analysis also suggests that bicyclists who self-identified as “strong and fearless” were more likely to choose sharrows as the preferred design, while bicyclists who self-identified as “interested but concerned” more often chose the protected bike lane. These results highlight the importance of understanding preferences of not just current cyclists, but potential future cyclists. Virtual reality simulation offers a low-cost, safe, and efficient method to understand preferences of individuals interested but not yet choosing cycling as a mode.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 9 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.