Posting advisory speed signs at sharp horizontal curve sites is a practice well established in the United States. The purpose of these signs is to provide the driving public with a safe speed to negotiate such curves; however, the link between these signs and safety performance has not yet been clearly established. A number of research projects have investigated the operational effectiveness of these signs by focusing on the direct response of drivers to the signs in terms of speed selection and reduction. This paper presents a recent Oregon effort to model the safety performance as it relates to these curve advisory speed signs. The authors developed a Generalized Linear Model that parameterizes the crash frequency at 2-lane rural highways in the state of Oregon in terms of curve advisory speed signs and other factors. Though ultimately the paper presents an analysis based on the Poisson model as this model provided the most appropriate fit to the data, the authors also tested an alternative Negative Binomial (NB) model. Because the data for this analysis was not over-dispersed, the model interpretation is valid for both the Poisson and NB model specifications. This research found that a bi-linear interpolant polynomial, contained in the selected statistical model, convincingly establishes a link between the presence of advisory speed signs and the expected numbers of crashes at these sites. Such a link also proved meaningful from the engineering and human factors perspectives. The authors anticipate that the alluded bivariate function should find applications in safety assessment and future speed posting practices. By using the developed sub-model, the authors estimated the safety effectiveness of advisory speeds. This research estimates that, for the state of Oregon, these signs are linked to an approximate reduction of 27% in the expected number of crashes. In general, this research found that advisory speed signs tend to enhance safety. However, the authors also determined that, under certain conditions, advisory speed signs may not be displaying the value that offer the greatest potential for safety enhancement. Furthermore, some advisory speeds can actually be negatively associated with safety performance. Based on the findings of this research, this negative relation can occur at sites with either excessively prohibitive or excessively permissive advisory speeds.
[1]
Mashrur Chowdhury,et al.
ARE THE CRITERIA FOR SETTING ADVISORY SPEEDS ON CURVES STILL RELEVANT
,
1998
.
[2]
Klaus F. Zimmermann,et al.
A new approach for modeling economic count data
,
1991
.
[3]
法律.
Manual on Uniform Traffic Control Devices
,
2010
.
[4]
Jeffrey D Miles,et al.
Measuring the Non-circular Portions of Horizontal Curves: An Automated Data Collection Method Using GPS
,
2011
.
[5]
R. Scheaffer,et al.
Mathematical Statistics with Applications.
,
1992
.
[6]
Hiroe Tsubaki,et al.
Bivariate Negative Binomial Generalized Linear Models for Environmental Count Data
,
2006
.
[7]
Richard W Lyles.
ADVISORY AND REGULATORY SPEED SIGNS FOR CURVES: EFFECTIVE OR OVERUSED?
,
1982
.
[8]
Raul Avelar,et al.
Safety evaluation of curve warning speed signs.
,
2011
.
[9]
Jeffrey David Miles,et al.
Evaluation of Alternative Procedures for Setting Curve Advisory Speed
,
2009
.
[10]
Thomas Brambor,et al.
Understanding Interaction Models: Improving Empirical Analyses
,
2006,
Political Analysis.