Methodology to Predict the Safety Performance of Urban and Suburban Arterials (2008) / Chapter Skim
Currently Skimming:
Pages 88-121
The Chapter Skim interface presents what we've algorithmically identified as the most significant single chunk of text within every page in the chapter.
Select key terms on the right to highlight them within pages of the chapter.
From page 88... ...
85 CHAPTER 5. BASE MODELS AND ADJUSTMENT FACTORS This chapter presents the base models and adjustment factors developed for the HSM safety prediction methodology for urban and suburban arterials.
|
From page 89... ...
86 • shoulder width • on-street parking In most cases, ADT and the other independent variables are significant at the 5-percent significance level or better. The model coefficients are presented in Tables 32 and 33 for roadway segments in Minnesota and Michigan, respectively.
|
From page 90... ...
87 TABLE 33. Initial models for multiple-vehicle nondriveway collisions on Michigan roadway segments Regression coefficient (standard error)
|
From page 91... ...
88 data points represent a two-lane undivided arterial with zero shoulder width (e.g., a curb-andgutter cross section) and no on-street parking.
|
From page 92... ...
89 TABLE 34. Initial models for multiple-vehicle nondriveway collisions on roadway segments for Minnesota and Michigan combined Regression coefficient (standard error)
|
From page 93... ...
90 TABLE 35. Base models for multiple-vehicle nondriveway collisions on roadway segments Regression coefficients Road type Intercept (a)
|
From page 94... ...
91 Figure 12. Plots for base models for fatal-and-injury multiple-vehicle nondriveway collisions on roadway segments by roadway type.
|
From page 95... ...
92 The values of for the base models, which represent the improvement of the model over an intercept-only model, range from 0.10 to 0.37. The only models with relatively low values of are those for fatal-and-injury accidents on roadway types with relatively few accidents.
|
From page 96... ...
93 The independent variables included in these models, when statistically significant, in addition to ADT, which was always included, are: • shoulder width • on-street parking • roadside hazard rating The model coefficients are presented in Tables 37 and 38 for roadway segments in Minnesota and Michigan, respectively. As in the case of multiple-vehicle nondriveway collisions, no statistically significant effect was found for lane width and only counterintuitive effects were found for speed category.
|
From page 97... ...
94 TABLE 37. Initial models for single-vehicle collisions on Minnesota roadway segments Regression coefficient (standard error)
|
From page 98... ...
95 TABLE 38. Initial models for single-vehicle collisions on Michigan roadway segments Regression coefficient (standard error)
|
From page 99... ...
96 TABLE 39. Initial models for single-vehicle collisions on roadway segments in Minnesota and Michigan combined Regression coefficient (standard error)
|
From page 100... ...
97 TABLE 40. Base models for single-vehicle collisions on roadway segments Regression coefficients Road type Intercept (a)
|
From page 101... ...
98 Figure 15. Plots for base models for fatal-and-injury single-vehicle collisions on roadway segments by roadway type.
|
From page 102... ...
99 TABLE 41. Single-vehicle collision factors for roadway segments Single-vehicle collision factor (frsv)
|
From page 103... ...
100 Driveway-Related Collisions The models presented above for multiple-vehicle collisions addressed only collisions that were not related to driveways. Driveway-related accidents are also generally multiple-vehicle collisions, but are addressed separately because the frequency of driveway-related accidents on an arterial roadway segment depends on the number and type of driveways.
|
From page 104... ...
101 Table 43 presents values for model coefficient a, which represents the intercept; coefficient b, which represents the exponent of the traffic volume factor; and coefficients c through i, that are proportional to the average annual accident frequencies for driveways of specific types. The factors in Table 43 were derived with multinomial regression from Minnesota data.
|
From page 105... ...
102 TABLE 43. Initial models for driveway-related collisions based on Minnesota data Regression coefficients for intercept, traffic volume, and specific driveway types Road type Total number of driveways Intercept (a)
|
From page 106... ...
103 Table 44 each represent the average number of accidents per driveway per year for specific driveway types, an arterial of the specified type for an ADT of 15,000 veh/day. The normalization or centering to an ADT of 15,000 veh/day was done to facilitate comparisons between values of the driveway coefficients across roadway types.
|
From page 107... ...
104 TABLE 45. Pedestrian safety adjustment factors for roadway segments in Minnesota and Michigan Pedestrian safety adjustment factor (fpedr)
|
From page 108... ...
105 TABLE 47. Bicycle safety adjustment factors for roadway segments Bicycle safety adjustment factor (fbiker)
|
From page 109... ...
106 Multiple-Vehicle Collisions Negative binomial regression models for multiple-vehicle collisions were developed in the form shown below: Nbimv = exp (a + b lnADTmaj + c lnADTmin)
|
From page 110... ...
107 TABLE 50. Initial models for multiple-vehicle collisions at North Carolina intersections Regression coefficient (standard error)
|
From page 111... ...
108 The values of for the models shown in Table 51 are relatively high, with the proportion in the variation in accident frequency explained by the models ranging from 0.30 to 0.61. The severity distribution for multiple-vehicle intersection collisions is shown in Table 52 based on combined data for Minnesota and North Carolina.
|
From page 112... ...
109 Figure 18. Plots for base models for fatal-and-injury multiple-vehicle collisions at intersections by intersection type.
|
From page 113... ...
110 Single-Vehicle Collisions Negative binomial regression models for single-vehicle collisions were initially developed in the form shown below: Nbisv = exp (a + b lnADTmaj + c lnADTmin)
|
From page 114... ...
111 TABLE 53. Initial models for single-vehicle collisions at Minnesota intersections Regression coefficient (standard error)
|
From page 115... ...
112 TABLE 54. Initial models for single-vehicle collisions at North Carolina intersections Regression coefficient (standard error)
|
From page 116... ...
113 TABLE 55. Base models for single-vehicle collisions at intersections in Minnesota and North Carolina combined Regression coefficient (standard error)
|
From page 117... ...
114 Figure 20. Plots for base models for total single-vehicle collisions at intersections by intersection type.
|
From page 118... ...
115 Figure 22. Plots for base models for property-damage-only single-vehicle collisions at intersections by intersection type.
|
From page 119... ...
116 The severity distribution for single-vehicle collisions at intersections is presented in Table 57.
|
From page 120... ...
117 Vehicle-Bicycle Collisions Table 59 presents adjustment factors for the average frequency of vehicle-bicycle collisions at arterial intersections. The table includes values for Minnesota, North Carolina, and both data sets combined.
|
Key Terms
This material may be derived from roughly machine-read images, and so is provided only to facilitate research.
More
information on Chapter Skim is available.