Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
161 A P P E N D I X F This appendix presents the generalized linear model (GLM) results to investigate the effect of treatment with shoulder rumble strips on the four crash types of interest: ⢠TOT crashes ⢠FI crashes ⢠SVROR crashes ⢠SVROR FI crashes Two groups of tables are presented in this appendix: ⢠Tables F-1 through F-4: GLM results based on all site types (i.e., all nontreatment and treatment sites in the study) ⢠Tables F-5 through F-8: GLM results based on all before-after sites and nontreatment cross-sectional sites (i.e., same sites as in Tables G-1 through G-4 but without treatment cross- sectional sites). These results should be most comparable to the SPF results since both approaches use the same types of sites. The statistics shown for each crash type, roadway type, and state (individually or combined) are: ⢠Intercept: estimate and standard error ⢠ADT (on natural log scale): estimate, standard error, and p-value (i.e., signiï¬cance level) ⢠Outside RHR: estimate, standard error, and p-value ⢠Rumble strip effect: estimate, standard error, and p-value. Rumble strip statistics are bolded whenever the effect is statistically signiï¬cant at the 0.10 level ⢠Overdispersion parameter: estimate and standard error GLM Analysis Results for Safety Effectiveness of Shoulder Rumble Strips Each regression model is represented by the following equation: where a (i.e., intercept), b, c, and d are the coefï¬cients whose estimates are shown in Tables F-1 through F-8. The coefï¬- cient d only applies to treatment sites since IRS is a 0,1 variable where 0 indicates absence and 1 presence of shoulder rumble strips. For states that treat both sides of a divided highway as separate sites (i.e., Missouri and Pennsylvania), the RHR vari- ables in the models represent the values for a single side of the divided highway. When both sides of a divided highway are treated as a single site (i.e., Minnesota sites), the RHR variables in the model represent average values for both directions of travel. Similarly, the RHR variable in the model for rural two-lane roads represents the average RHR for both sides of the roadway. Tables F-1 through F-8 are the companion tables to Tables 29 through 36. Number of sites and number of site-years for each model are provided in those tables. To obtain the percent change due to rumble strip treatment shown in Tables 29 through 36, use the rumble strip coefficient, d, shown in Tables F-1 through F-8 and calculate: Lower and upper 95-percent conï¬dence intervals for the percent change are calculated in a similar fashion based on lower and upper 95-percent conï¬dence intervals of the esti- mates in Tables F-1 through F-8 (conï¬dence limits calculated but not shown in these tables). Percent change d= ( )â[ ]100 1exp Expected crashes mi yr a blnADT cRHR dIOut= + + +exp RS( )
Intercept lnADT Outside RHR Rumble stripsa Overdispersion Roadway type State Estimate SE Estimate SE p-value Estimate SE p-value Estimate SE p-value Estimate SE Urban freeways PA â9.53 1.04 1.00 0.10 <.0001 0.15 0.10 0.111 â0.04 0.06 0.496 0.22 0.03 Combined â9.22 1.46 1.04 0.15 <.0001 0.01 0.07 0.890 0.23 0.03 MO â6.55 2.11 0.79 0.21 0.000 0.08 0.07 0.227 0.19 0.03 Rural freeways PA â4.27 1.85 0.50 0.19 0.008 0.08 0.11 0.480 0.21 0.05 Combined â5.90 1.10 0.64 0.12 <.0001 0.26 0.06 <.0001 0.18 0.08 0.021 0.24 0.03 MN â8.03 1.07 0.88 0.12 <.0001 0.15 0.06 0.010 0.15 0.08 0.053 0.11 0.02 MO â6.53 2.96 0.81 0.31 0.009 0.25 0.11 0.027 0.42 0.08 Rural multilane divided highways (nonfreeways) PA â0.97 4.23 0.20 0.46 0.653 â0.21 0.24 0.374 0.23 0.12 Combined â6.27 0.87 0.66 0.10 <.0001 0.26 0.04 <.0001 â0.15 0.11 0.178 0.54 0.04 MN â5.72 1.11 0.56 0.14 <.0001 0.31 0.05 <.0001 â0.04 0.07 0.594 0.27 0.04 MO â4.71 2.01 0.64 0.23 0.006 â0.18 0.88 0.836 0.99 0.11 Rural two-lane roads PA â4.55 0.80 0.57 0.10 <.0001 â0.28 0.19 0.145 0.36 0.06 a Statistics are bolded when p ⤠0.10. Intercept lnADT Outside RHR Rumble strips a Overdispersion Roadway type State Estimate SE Estimate SE p-value Estimate SE p-value Estimate SE p-value Estimate SE Urban freeways PA â11.19 1.27 1.08 0.12 <.0001 0.20 0.12 0.086 â0.10 0.08 0.222 0.28 0.05 Combined â10.55 1.26 1.08 0.13 <.0001 â0.08 0.07 0.284 0.10 0.03 MO â9.95 1.67 1.02 0.17 <.0001 â0.03 0.09 0.758 0.12 0.04Rural freeways PA â10.16 2.32 1.04 0.24 <.0001 â0.12 0.12 0.316 0.06 0.07 Combined â7.71 1.24 0.72 0.13 <.0001 0.23 0.07 0.001 0.01 0.11 0.956 0.22 0.04 MN â9.18 1.27 0.90 0.14 <.0001 0.11 0.07 0.131 0.07 0.09 0.474 0.09 0.04 MO â14.93 3.71 1.59 0.40 <.0001 0.05 0.20 0.799 0.38 0.12 Rural multilane divided highways (nonfreeways) PA 0.28 0.24 â0.54 0.27 0.045 0.40 0.22 Combined â7.16 0.77 0.65 0.09 <.0001 0.31 0.05 <.0001 â0.32 0.12 0.006 0.50 0.06 MN â6.45 1.10 0.53 0.14 0.0002 0.31 0.06 <.0001 â0.14 0.10 0.155 0.24 0.08 MO â8.37 1.95 0.61 0.25 0.013 0.71 0.19 0.0002 â0.51 0.75 0.501 0.51 0.11 Rural two-lane roads PA â5.44 0.86 0.61 0.10 <.0001 â0.18 0.22 0.415 0.37 0.08 a Statistics are bolded when p ⤠0.10. Table F-1. GLM estimates for TOT crashes based on all site types. Table F-2. GLM estimates for FI crashes based on all site types.
Table F-3. GLM estimates for SVROR crashes based on all site types. Table F-4. GLM estimates for SVROR FI crashes based on all site types. Intercept lnADT Outside RHR Rumble stripsa Overdispersion Roadway type State Estimate SE Estimate SE p-value Estimate SE p-value Estimate SE p-value Estimate SE Urban freeways PA â6.79 0.90 0.72 0.09 <.0001 â0.04 0.08 0.625 0.19 0.04 Combined â5.94 1.57 0.65 0.16 <.0001 â0.10 0.07 0.175 0.25 0.04 MO â5.41 2.31 0.61 0.23 0.009 â0.07 0.07 0.357 0.23 0.04 Rural freeways PA â3.42 2.40 0.29 0.25 0.256 0.26 0.14 0.057 â0.02 0.13 0.875 0.26 0.07 Combined â7.11 1.08 0.62 0.11 <.0001 0.39 0.08 <.0001 0.35 0.12 0.004 0.39 0.05 MN â8.74 1.21 0.83 0.13 <.0001 0.17 0.08 0.035 0.33 0.12 0.006 0.24 0.05 MO â12.64 2.98 1.26 0.32 <.0001 0.34 0.22 0.128 0.53 0.16 0.001 0.45 0.12 Rural multilane divided highways (nonfreeways) PA â0.59 6.16 0.12 0.65 0.855 â0.27 0.17 0.117 0.17 0.14 Combined â5.24 0.83 0.37 0.10 0.0002 0.42 0.05 <.0001 â0.35 0.17 0.037 0.72 0.07 MN â3.82 0.83 0.10 0.10 0.353 0.50 0.08 <.0001 0.18 0.13 0.180 0.51 0.13 MO â0.33 0.18 â0.09 0.79 0.906 1.23 0.18 Rural two-lane roads PA â2.64 0.89 0.28 0.11 0.008 â0.60 0.22 0.008 0.49 0.09 a Statistics are bolded when p ⤠0.10. Intercept lnADT Outside RHR Rumble stripsa Overdispersion Roadway type State Estimate SE Estimate SE p-value Estimate SE p-value Estimate SE p-value Estimate SE Urban freeways PA â6.92 1.15 0.67 0.11 <.0001 0.02 0.10 0.862 0.20 0.07 Combined â8.53 1.48 0.76 0.15 <.0001 0.22 0.11 0.042 â0.15 0.08 0.078 0.19 0.06 MO â6.75 1.85 0.65 0.19 0.001 â0.13 0.10 0.173 0.20 0.07 Rural freeways PA â9.99 2.78 0.89 0.29 0.002 0.29 0.16 0.076 â0.14 0.16 0.383 0.17 0.10 Combined â7.19 1.48 0.58 0.16 0.0004 0.32 0.10 0.001 0.05 0.14 0.738 0.45 0.09 MN â9.50 1.49 0.88 0.15 <.0001 0.12 0.14 0.398 0.22 0.08 MO â15.15 3.96 1.56 0.42 0.000 0.17 0.21 0.422 0.59 0.20 Rural multilane divided highways (nonfreeways) PA â0.07 0.28 â0.40 0.28 0.156 0.58 0.32 Combined â5.46 0.77 0.32 0.10 0.001 0.41 0.06 <.0001 â0.47 0.16 0.004 0.85 0.13 MN â4.50 1.19 0.26 0.15 0.084 0.04 0.17 0.840 1.16 0.33 MO â8.46 2.18 0.40 0.24 0.094 0.95 0.15 <.0001 â0.90 1.38 0.515 0.19 0.14 Rural two-lane roads PA â2.96 0.94 0.25 0.12 0.035 â0.47 0.24 0.054 0.57 0.14 a Statistics are bolded when p ⤠0.10.
Table F-5. GLM estimates for TOT crashes based on before and after sites and nontreatment cross-sectional sites. Table F-6. GLM estimates for FI crashes based on before and after sites and nontreatment cross-sectional sites. Intercept lnADT Outside RHR Rumble stripsa Overdispersion Roadway type State Estimate SE Estimate SE p-value Estimate SE p-value Estimate SE p-value Estimate SE Urban freeways PA â9.30 1.23 1.03 0.12 <.0001 â0.05 0.06 0.372 0.22 0.03 Combined â8.83 1.75 1.01 0.18 <.0001 0.07 0.07 0.363 0.21 0.03 MO â5.38 2.63 0.68 0.26 0.010 0.11 0.07 0.113 0.19 0.03 Rural freeways PA â1.03 2.32 0.17 0.24 0.492 0.06 0.15 0.677 0.14 0.05 Combined â5.46 1.30 0.60 0.14 <.0001 0.24 0.08 0.003 0.25 0.08 0.004 0.29 0.03 MN â7.44 1.46 0.82 0.17 <.0001 0.14 0.08 0.086 0.15 0.12 0.185 0.12 0.03 MO â6.59 3.18 0.82 0.34 0.014 0.24 0.11 0.028 0.42 0.08 Rural multilane divided highways (nonfreeways) PA â1.11 4.54 0.22 0.49 0.656 â0.30 0.31 0.337 0.26 0.13 Combined â5.96 0.87 0.64 0.10 <.0001 0.23 0.04 <.0001 â0.06 0.14 0.680 0.59 0.04 MN â5.16 1.09 0.49 0.13 0.000 0.34 0.05 <.0001 0.16 0.09 0.058 0.31 0.06 MO â4.79 2.05 0.65 0.24 0.007 â0.17 0.93 0.858 0.99 0.12 Rural two-lane roads PA â4.55 0.80 0.57 0.10 <.0001 â0.28 0.19 0.145 0.36 0.06 a Statistics are bolded when p ⤠0.10. Intercept lnADT Outside RHR Rumble stripsa Overdispersion Roadway type State Estimate SE Estimate SE p-value Estimate SE p-value Estimate SE p-value Estimate SE Urban freeways PA â10.24 1.50 1.05 0.14 <.0001 â0.23 0.10 0.019 0.29 0.05 Combined â9.51 1.58 0.97 0.16 <.0001 â0.04 0.09 0.631 0.12 0.04 MO â9.70 2.26 0.99 0.23 <.0001 â0.01 0.10 0.888 0.13 0.04 Rural freeways PA â5.19 3.37 0.52 0.35 0.134 â0.09 0.15 0.555 0.08 0.08 Combined â7.49 1.52 0.70 0.17 <.0001 0.23 0.09 0.007 0.05 0.17 0.766 0.29 0.06 MN â8.70 1.48 0.87 0.15 <.0001 â0.20 0.18 0.278 0.08 0.05 MO â15.91 3.83 1.70 0.41 <.0001 0.02 0.21 0.929 0.38 0.12 Rural multilane divided highways (nonfreeways) PA â1.02 9.61 0.14 1.03 0.893 â0.58 0.29 0.045 0.41 0.22 Combined â6.86 0.75 0.63 0.09 <.0001 0.28 0.05 <.0001 â0.16 0.13 0.244 0.54 0.06 MN â5.99 1.06 0.47 0.13 0.000 0.31 0.06 <.0001 0.07 0.15 0.648 0.28 0.10 MO â8.46 1.91 0.63 0.24 0.010 0.70 0.19 0.000 â0.44 0.76 0.567 0.50 0.11 Rural two-lane roads PA â5.44 0.86 0.61 0.10 <.0001 â0.18 0.22 0.415 0.37 0.08 a Statistics are bolded when p ⤠0.10.
Table F-7. GLM estimates for SVROR crashes based on before and after sites and nontreatment cross-sectional sites. Table F-8. GLM estimates for SVROR FI crashes based on before and after sites and nontreatment cross-sectional sites. Intercept lnADT Outside RHR Rumble strips a Overdispersion Roadway type State Estimate SE Estimate SE p-value Estimate SE p-value Estimate SE p-value Estimate SE Urban freeways PA â7.26 1.11 0.77 0.11 <.0001 â0.10 0.11 0.339 0.19 0.05 Combined â5.16 2.02 0.58 0.21 0.005 â0.10 0.08 0.214 0.26 0.04 MO â5.15 3.12 0.58 0.31 0.063 â0.06 0.08 0.453 0.26 0.05Rural freeways PA 0.18 0.08 â0.14 0.18 0.450 0.16 0.08 Combined â7.12 1.25 0.64 0.13 <.0001 0.33 0.09 0.0003 0.51 0.14 0.0002 0.42 0.06 MN â8.58 1.41 0.86 0.14 <.0001 0.30 0.10 0.002 0.20 0.06 MO â13.09 3.08 1.31 0.34 <.0001 0.33 0.22 0.137 0.51 0.16 0.002 0.46 0.12 Rural multilane divided highways (nonfreeways) PA â0.46 6.31 0.11 0.67 0.874 â0.26 0.17 0.141 0.18 0.14 Combined â5.23 0.83 0.38 0.10 0.0002 0.41 0.05 <.0001 â0.30 0.22 0.177 0.73 0.08 MN â3.60 0.85 0.06 0.11 0.578 0.52 0.08 <.0001 0.11 0.18 0.523 0.46 0.16 MO â0.32 0.18 â0.06 0.79 0.940 1.21 0.17 Rural two-lane roads PA â2.64 0.89 0.28 0.11 0.008 â0.60 0.22 0.008 0.49 0.09 a Statistics are bolded when p ⤠0.10. Intercept lnADT Outside RHR Rumble stripsa Overdispersion Roadway type State Estimate SE Estimate SE p-value Estimate SE p-value Estimate SE p-value Estimate SE Urban freeways PA â6.95 1.39 0.67 0.13 <.0001 â0.10 0.13 0.43 0.18 0.08 Combined â6.61 1.89 0.63 0.19 0.001 â0.19 0.10 0.05 0.21 0.07 MO â8.20 2.37 0.79 0.24 0.001 â0.15 0.11 0.18 0.22 0.08 Rural freeways PA â1.86 3.25 0.14 0.34 0.680 â0.25 0.20 0.21 0.17 0.13 Combined â7.07 1.79 0.57 0.20 0.004 0.30 0.12 0.014 0.14 0.20 0.50 0.52 0.11 MN â8.15 1.71 0.74 0.18 <.0001 â0.16 0.19 0.40 0.16 0.09 MO â16.48 3.99 1.70 0.43 <.0001 0.11 0.21 0.60 0.60 0.21 Rural multilane divided highways (nonfreeways) PA â0.04 0.29 â0.24 0.20 0.23 0.59 0.32 Combined â5.39 0.80 0.31 0.10 0.003 0.41 0.06 <.0001 â0.50 0.20 0.01 0.86 0.14 MN â3.83 1.22 0.17 0.16 0.286 â0.26 0.25 0.30 1.26 0.47 MO â8.49 2.16 0.41 0.24 0.087 0.94 0.15 <.0001 â0.84 1.43 0.56 0.19 0.14 Rural two-lane roads PA â2.96 0.94 0.25 0.12 0.035 â0.47 0.24 0.05 0.57 0.14 a Statistics are bolded when p ⤠0.10.