作者: Venky N. Shankar , Songrit Chayanan , Sittipan Sittikariya , Ming-Bang Shyu , Naveen K. Juvva
DOI: 10.3141/1897-20
关键词: Multivariate statistics 、 Transport engineering 、 Omitted-variable bias 、 Poison control 、 Crash 、 Statistics 、 Precipitation 、 Sample (statistics) 、 Multivariate analysis 、 Main effect 、 Environmental science
摘要: A multivariate model that incorporates the effects of design, traffic, weather, and related interactions with design variables on reported roadside crashes is presented. By providing for a framework accounts all measurable effects, minimizes impact omitted variable effects. Furthermore, presented partial observability stem from fluctuations in environmental conditions as well unobserved contribute to heterogeneity traffic safety network. sample 318 sections 1 mi long was used study. These represent state highway network Washington basis road classification factors therefore were collection detailed precipitation, snowfall, temperature data addition roadway parameters. The resulting suggests marginal weather both main interactive form, even after controlling observability, play statistically significant role crash occurrence. In particular, it found average monthly snowfall exceeding 4 in. between snow depths horizontal curves have effect frequency probabilities. these also significant; furthermore, contribution likelihood frequencies approximately 19%, contributed 33%, 6%. Weather 6% overall likelihood. Traffic 36%