Investigating the underreporting of pedestrian and bicycle crashes in and around university campuses - a crowdsourcing approach.

In this paper, the non-motorized traffic safety concerns in and around three university campuses are evaluated by comparing police-reported crash data with traffic safety information sourced from the campus communities themselves. The crowdsourced traffic safety data comprise of both self-reported crashes as well as perceived hazardous locations. The results of the crash data analysis reveal that police-reported crashes underrepresent non-motorized safety concerns in and around the campus regions. The spatial distribution of police-reported crashes shows that police-reported crashes are predominantly unavailable inside the main campus areas, and the off-campus crashes over-represent automobile involvement. In comparison, the self-reported crash results report a wide variety of off-campus collisions not involving automobiles, while also highlighting the issue of high crash concentrations along campus boundaries. An assessment of the perceived hazardous locations (PHLs) reveals that high concentrations of such observations at/near a given location have statistically significant association with both survey-reported crashes as well as future police-reported crashes. Moreover, the results indicate the presence of a saturation point in the relationship between crashes and PHLs wherein beyond a certain limit, an increasing number of traffic safety concerns may not necessarily correlate with a proportional increase in the number of crashes. These findings suggests that augmenting our existing knowledge of traffic safety through crowdsourcing techniques can potentially help in better estimating both existing as well as emerging traffic safety concerns.