Back to Search
Start Over
Speed limit reduction in urban areas: A before–after study using Bayesian generalized mixed linear models
- Source :
- Accident Analysis & Prevention. 73:252-261
- Publication Year :
- 2014
- Publisher :
- Elsevier BV, 2014.
-
Abstract
- In fall 2009, a new speed limit of 40km/h was introduced on local streets in Montreal (previous speed limit: 50km/h). This paper proposes a methodology to efficiently estimate the effect of such reduction on speeding behaviors. We employ a full Bayes before-after approach, which overcomes the limitations of the empirical Bayes method. The proposed methodology allows for the analysis of speed data using hourly observations. Therefore, the entire daily profile of speed is considered. Furthermore, it accounts for the entire distribution of speed in contrast to the traditional approach of considering only a point estimate such as 85th percentile speed. Different reference speeds were used to examine variations in the treatment effectiveness in terms of speeding rate and frequency. In addition to comparing rates of vehicles exceeding reference speeds of 40km/h and 50km/h (speeding), we verified how the implemented treatment affected "excessive speeding" behaviors (exceeding 80km/h). To model operating speeds, two Bayesian generalized mixed linear models were utilized. These models have the advantage of addressing the heterogeneity problem in observations and efficiently capturing potential intra-site correlations. A variety of site characteristics, temporal variables, and environmental factors were considered. The analyses indicated that variables such as lane width and night hour had an increasing effect on speeding. Conversely, roadside parking had a decreasing effect on speeding. One-way and lane width had an increasing effect on excessive speeding, whereas evening hour had a decreasing effect. This study concluded that although the treatment was effective with respect to speed references of 40km/h and 50km/h, its effectiveness was not significant with respect to excessive speeding-which carries a great risk to pedestrians and cyclists in urban areas. Therefore, caution must be taken in drawing conclusions about the effectiveness of speed limit reduction. This study also points out the importance of using a comparison group to capture underlying trends caused by unknown factors. Language: en
- Subjects :
- Automobile Driving
Engineering
Percentile
business.industry
Speed limit
Accidents, Traffic
Public Health, Environmental and Occupational Health
Linear model
Contrast (statistics)
Poison control
Bayes Theorem
Human Factors and Ergonomics
Models, Theoretical
Bayes' theorem
Controlled Before-After Studies
Statistics
Linear Models
Humans
Seasons
Point estimation
Cities
Safety, Risk, Reliability and Quality
business
Simulation
Empirical Bayes method
Subjects
Details
- ISSN :
- 00014575
- Volume :
- 73
- Database :
- OpenAIRE
- Journal :
- Accident Analysis & Prevention
- Accession number :
- edsair.doi.dedup.....70a237a14fb74e7e65ce3ef9bca52694