U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590
202-366-4000

Skip to content
Facebook iconYouTube iconTwitter iconFlickr iconLinkedInInstagram

Safety

FHWA Home / Safety / Local and Rural Road / Delta Region Transportation Development Program

Program Contact

Rosemarie Anderson
rosemarie.anderson@dot.gov
(202) 366-5007
 

Delta Region Transportation Development Program: Rural Safety Innovation Program Evaluation – Final Report

Chapter 2. Detailed Quantitative Evaluations

Detailed quantitative evaluations of the safety effectiveness of countermeasures or countermeasure combinations were performed for three RSIP projects. The specific countermeasures or countermeasure combinations that were evaluated included:

The results of each safety evaluation are provided below. For each evaluation, a description of the RSIP project is provided, followed by descriptive statistics, analysis approach, analysis results, and interpretation of results. A summary of the evaluations of the three projects is presented at the end of this chapter.

RSIP Project 27

Agency: Mississippi De partment of Transportation (MDOT)

Focus of Evaluation: Dual Application of Centerline and Shoulder Rumble Strips

Project Background

MDOT received funding through the RSIP to implement two types of safety improvements along rural state highways: the installation of centerline rumble strips and a clear zone restoration project. These improvements focused on reducing the number and severity of lane departure crashes.

The total project cost for both safety improvements was approximately $2,407,480. In 2009, MDOT spent $1,602,700 on the installation of centerline rumble strips. The remaining funds were spent on the clear zone restoration project which included removal of roadside objects, regrading of side slopes, and installation of cable barrier, covering about 5 mi of roadway.

Table 3 shows the highways, counties, and beginning and end points for the centerline rumble strips installed as part of the RSIP project. The total project covered approximately 468 mi of rural two-lane roads, but centerline rumble strips were not installed along the entire lengths of highways listed in Table 3. It was estimated that approximately 350 miles of centerline rumble strips were installed through the RSIP project. At many of the locations where centerline rumble strips were installed, shoulder rumble strips were already present and, in many cases, recently installed (i.e., within a year or two of installation of the centerline rumble strips).

After assessing the overall safety improvements implemented through the RSIP project, it was determined that the focus of this safety evaluation should be on the locations where centerline rumble strips were installed on the same routes where shoulder rumble strips were present. It was further decided that the improvements from the clear zone restoration project covered only a few miles of roadway so an evaluation of this project by itself would likely yield unreliable results. Also, the improvements were unique from the centerline rumble strip installations and other improvements implemented through other RSIP projects, so the safety effectiveness of the clear zone project was not investigated.

Table 3. RSIP Project 27: Approximate Locations of Centerline Rumble Strip Installations from the RSIP in Mississippi

Route Counties Begin Termini End Termini Mileage
MS 1 Washington and Bolivar Washington-Issaquena County Line Bolivar-Coahoma County Line 90
MS 7 Humphreys, Leflore, Carroll, Grenada, Yalobusha, Lafayette, Marshall Northern city limits of Belzoni Marshall-Benton County Line 125
MS 8 Grenada Leflore-Grenada County Line Grenada-Calhoun County Line 40
MS 18 Rankin Louis Wilson Road (Old MS 18) Shell Oil Road 5
MS 27 Warren, Hinds, Copiah Interstate 20 Northern city limits of Georgetown 60
MS 178 Marshall Desoto-Marshall County Line Western city limits of Holly Springs 15
MS 587 Lawrence, Marion Sand Road MS 586 25
US 49E Holmes, Leflore Yazoo-Holmes County Line Southern city limits of Greenwood 38
US 61 Warren, Issaquena, Sharkey, Washington Yazoo River Bridge Beginning of 4-lane section, south of Leland, MS 70

Thus, treatment sites considered in this safety evaluation included sites on rural two-lane roads where the centerline and shoulder rumble strips were both installed. The centerline rumble strips were installed as part of the RSIP project, and the shoulder rumble strips were installed separately as part of a previous safety improvement within a few years of the installation of the centerline rumble strips. Table 4 shows the beginning and ending locations of routes used as treatment sites in the analysis.

Table 4. RSIP Project 27: Location of Treatment Sites Used in Analysis of Dual Application of Centerline and Shoulder Rumble Strips in Mississippi

Site No. Route Begin Termini End Termini Length (mi)
Latitude Longitude Landmark Latitude Longitude Landmark
1 MS 1 33.86895 -91.01989 50 yds N of Rosedale Co Line 33.93449 -90.95056 Gunnison Corp. City Limit S 6.1
2 MS 1 33.94861 -90.93738 Gunnison Corp. City Limit N 33.98915 -90.90062 150 ft North of Bunge Rd 3.5
3 MS 7 33.48773 -90.32449 Ita Bena Corp City Limit/0.1 mi S of CR 514 33.39170 -90.31304 CR 511 7.1
5 MS 7 34.11651 -89.64865 0.8 mi N of CR 7 33.95933 -89.69366 0.6 mi N of CR 71(Mount Grove MB Church) 11.3
6 MS 7 34.72965 -89.46450 400 ft from MS 4 34.65681 -89.45878 550 ft N of Old MS 7 5.0
7 MS 8 33.78049 -89.76070 Grenada City Limit 33.76251 -89.50726 Grenada County Line 15.0
8 MS 27 31.97157 -90.25718 Beginning of NB passing lane 31.87526 -90.16793 500 ft N of MS 28 int. 8.6
9 US 61 33.09747 -90.88234 Sharkey County Line 32.97977 -90.82761 Anguilla Corp City Limit (Northern) 9.0
10 US 61 32.91724 -90.86683 Rolling Fork City Limit 32.96519 -90.82597 Anguilla Corp City Limit (Southern) 4.1
11 MS 587 31.34344 -89.97349 Ranch Rd 31.31425 -89.92480 Near Morgantown 4.0
12 MS 587 31.31175 -89.91358 Ballpark Ln 31.23504 -89.87185 W Division St 6.4

Nontreatment sites included in the analysis had similar characteristics to the treatment sites but had no rumble strips of any type present during the entire analysis period. All nontreatment sites were rural two-lane highways that had similar geometrics and traffic volumes as the treatment sites, but no rumble strips. Table 5 shows the beginning and ending locations of routes used as nontreatment sites in the analysis.

Table 5. RSIP Project 27: Location of Nontreatment Sites Used in Analysis of Dual Application of Centerline and Shoulder Rumble Strips in Mississippi

Site No. Route Begin Termini End Termini Length (mi)
Latitude Longitude Landmark Latitude Longitude Landmark
N1 MS 1 34.12048 -90.82658 Bolivar/Coahoma County Line 34.21038 -90.71105 MS 322 M9.1
N2 MS 3 32.84168 -90.43389 Rinalto Rd 32.70342 -90.52300 Edgeline rumble strips present 11.1
N3 MS 8 33.76251 -89.50726 Grenada County Line 33.81016 -89.3494 MS 9 9.8
N4 MS 8 33.8071 -90.88254 None 33.74943 -90.74514 Bishop Rd 8.9
N5 MS 14 32.99576 -90.58855 JCT 149 32.97409 -90.82049 Anquilla Corp City Limit (Eastern) 14.0
N6 US 49 33.00318 -90.32137 Yazoo Co Line 32.90033 -90.38420 Coker Rd 8.8
N7 MS 149 32.86546 -90.45191 Carter Rd 33.09863 -90.49826 MS 49 23.6
N8 MS 587 31.52922 -90.09777 Emanuel Peyton Ln 31.34724 -89.97308 Centerline rumble strips present 16.4

Objective

Considerable research has been conducted on the safety effects of both centerline and shoulder rumble strips installed by themselves on separate roadways. Current state of the practice recommends that the safety effectiveness of countermeasures, when implemented in combination, should be estimated by multiplying their effectiveness together. This approach assumes that the safety effects of the individual countermeasures are independent, which may not be accurate. The objective of this evaluation was to estimate the safety effectiveness of centerline and shoulder rumble strips installed in combination on rural two-lane roads based on available crash data. Only one recent study (Olson et al., 2013) was found that evaluated the safety effectiveness of centerline and shoulder rumble strips installed along the same roadway.

Descriptive Statistics

A total of 19 sites–11 treatment and 8 nontreatment sites covering approximately 80.1 mi and 101.7 mi of roadway, respectively–were identified for inclusion in the analysis. All sites, both treatment and nontreatment, were on rural two-lane roads.

For this analysis, the "before-period" years include only those years prior to the installation of either shoulder or centerline rumble strips, and the "after-period" years include the years after installation of the centerline rumble strips. Crash data were generally available from 2005 to 2012; crashes occurring during the treatment installation year or years were excluded from analysis. Typically, for treatment sites, the before period was from 2005 to 2008, and the after period was from 2010 to 2012. Years in which shoulder rumble strips were present prior to the installation of the centerline rumble strips were not included in the analysis, so that the dual application of rumble strips could be compared to a base condition of no rumble strips. All centerline rumble strips were installed in 2009, so depending upon the installation year of the shoulder rumble strips, the number of years in the before period differed slightly for treatment sites. As a result, the number of years of available data varied by site—from 1 to 4 years of data per site in the before period but always 3 years per site in the after period. For the nontreatment sites, the number of years of available data was 7 years per site for each site over the entire study period.

Crash types considered in this safety analysis are those expected to be impacted by the installation of both centerline and shoulder rumble strips, namely:

Three crash severity levels–total crashes (i.e., all severities), fatal and all injury crashes (FI), and fatal and serious injury crashes (FS)–were used and analyzed separately. Table 6 summarizes the treatment site data used in the analysis, separately for each period (before and after) and site. The table shows the number of segments within each site [corresponding to a change in average annual daily traffic (AADT) along the site]; the total site length; the number of years in the specific period; the average AADT for that period; and crash counts by severity level (total, FI, and FS)–these crash counts represent the combined SVROR, sideswipe-opposite direction, and head-on crash counts. The AADT for each site and year was calculated as the average AADT of the segments within each site-year. These average AADTs were then averaged over the before and after years, respectively, for each site.

Table 7 summarizes the corresponding data for nontreatment sites across the entire study period.

Table 6. RSIP Project 27: Summary Statistics for the Before and After Periods for Treatment Sites in Mississippi

Site No. Number of Segments per Site Total Site Length (mi) Before Period After Period
Number of Years Average AADT Total Crashes FI Crashes FS Crashes Number of Years Average AADT Total Crashes FI Crashes FS Crashes
1 3 6.1 4 1,742 3 3 1 3 1,333 2 2 2
2 1 3.5 4 2,000 0 0 0 3 1,300 0 0 0
3 3 7.1 4 1,567 11 10 1 3 1,589 5 4 1
4 4 11.3 4 3,338 16 10 4 3 2,792 5 1 1
5 2 5.0 3 4,633 10 7 4 3 4,333 3 1 0
6 4 15.0 1 2,650 13 8 1 3 3,050 14 7 0
7 2 8.6 4 2,338 9 6 0 3 2,267 3 0 0
8 2 9.0 1 2,700 3 1 1 3 1,850 5 4 4
9 1 4.1 1 3,500 0 0 0 3 3,000 0 0 0
10 1 4.0 4 238 7 5 0 3 190 6 4 0
11 2 6.4 4 1,413 24 12 1 3 1,417 20 12 2
Total N/A 80.1 N/A N/A 96 62 13 N/A N/A 63 35 10

Note: Crash types include: SVROR crashes (right or left), sideswipe-opposite direction crashes, and head-on crashes only.

Table 7. RSIP Project 27: Summary Statistics for Entire Study Period for Nontreatment Sites in Mississippi

Site No. Number of Segments per Site Total Site Length (mi) Study Period
Number of Years Average AADT Total Crashes FI Crashes FS Crashes
N1 1 9.1 7 1,131 10 7 0
N2 2 11.1 7 1,629 14 5 1
N3 6 8.9 7 3,612 19 13 2
N4 3 9.8 7 1,819 18 7 0
N5 6 14.0 7 1,439 12 11 2
N6 2 8.8 7 2,579 11 6 0
N7 10 23.6 7 1,841 22 16 4
N8 3 16.4 7 301 21 14 2
Total N/A 101.7 N/A N/A 127 79 11

Note: Crash types include: SVROR crashes (right or left), sideswipe-opposite direction crashes, and head- on crashes only.

Analysis Approach

The safety effectiveness of this treatment was evaluated using the EB before/after method as outlined in the 14-step procedure of Appendix 9A in Chapter 9 of the HSM. The general procedure is as follows:

EB Estimation of the Expected Average Crash Frequency in the Before Period

Step 1 – Using the applicable SPF, calculate the predicted average crash frequency for site type x during each year of the before period. For roadway segments, the predicted average crash frequency will be expressed as crashes per site per year.

Step 2 – Calculate the expected average crash frequency for each site i, summed over the entire before period. For roadway segments, the expected average crash frequency will be expressed as crashes per site.

EB Estimation of the Expected Average Crash Frequency in the After Period in the Absence of the Treatment

Step 3 – Using the applicable SPF, calculate the predicted average crash frequency for each site i during each year y of the after period.

Step 4 – Calculate an adjustment factor to account for the differences between the before and after periods in duration and traffic volume at each site i.

Step 5 – Calculate the expected average crash frequency for each site i, over the entire after period in the absence of the treatment.

Estimation of Treatment Effectiveness

Step 6 – Calculate an estimate of the safety effectiveness of the treatment at each site i in the form of an odds ratio.

Step 7 – Calculate the safety effectiveness as a percentage crash change at site i.

Step 8 – Calculate the overall effectiveness of the treatment for all sites combined, in the form of an odds ratio.

Step 9 – The odds ratio calculated in Step 8 is potentially biased. Calculate an adjustment to obtain an unbiased estimate of the treatment effectiveness in terms of an adjusted odds ratio.

Step 10 – Calculate the overall unbiased safety effectiveness as a percentage change in crash frequency across all sites.

Estimation of the Precision of the Treatment Effectiveness

Step 11 – Calculate the variance of the unbiased estimated safety effectiveness, express as an odds ratio.

Step 12 – To obtain a measure of the precision of the odds ratio, calculate its standard error as the square root of its variance.

Step 13 – Calculate the standard error of the safety effectiveness measure from Step 10.

Step 14 – Assess the statistical significance of the estimated safety effectiveness.

Note that all predicted crashes for each site-year were estimated on a per site basis by multiplying the SPF by site length in the above calculations.

Prior to implementing the EB method, the following points were addressed:

1. Selecting an appropriate SPF: It was decided to use the SPFs for total and FI severity levels for rural two-lane roads from Safety Analyst. The Safety Analyst SPFs predict crashes for all collision types combined.

2. Obtaining the proportion of target crashes relevant to RSIP Project 27 evaluation (PR1): The proportions of both total and FI severity levels for the target crashes (SVROR, sideswipe-opposite direction, and head-on) were obtained from Table 10-4 in Chapter 10 of the HSM.

3. Obtaining the proportion of FS out of FI crashes for this project (PR2): The proportion of FS out of FI crashes was calculated from the FI and FS crashes that occurred on all the nontreatment sites and the before treatment sites combined in the project database (a total of 90 site-years).

4. Calibrating the SPFs to the local jurisdiction: Calibration was performed separately for total and FI crashes using all the nontreatment sites and the before treatment sites combined in the project database (a total of 90 site-years)

The Safety Analyst SPFs for rural two-lane roads for total and FI severity levels have the general form:

Equation 1. Predicted crashes per mile per year equals exponent open bracket a plus b times open parenthesis logarithm base ten of AADT closed parenthesis closed bracket.

Figure 1. Equation 1 - General form of Safety Analyst SPF for rural two-lane roads for total and fatal and all injury severity levels.

where a and b are regression coefficients shown in Table 8 for each severity level (Total and FI).

Calibration factors (Cr) and proportions of target crashes are then used to adjust for local conditions as follows:

Equation 2. Predicted crashes per mile per year equals open brace exponent open bracket a plus b times open parethesis logarithm base ten of AADT closed parenthesis closed bracket closed brace times PR subscript 1 times PR subscript 2 times C subscript r.

Figure 2. Equation 2 – General form of Safety Analyst SPF for rural two-lane roads adjusted for crash type and local conditions.

where PR1, PR2, and Crare provided in Table 8 for each severity level.

Table 8. RSIP Project 27: SPF Coefficients, Target Crash Proportions, and Calibration Factors Used for Mississippi Data

Severity Level Number of Site-Yearsa Intercept (a)b log10AADT Coefficient (b)b Overdispersion Parameterb Proportion of Target Collision Type (PR1)c,d Proportion of FS/FI Crashes
(PR2)a		 Calibration Factor (Cr)a
Total 90 -3.56 0.55 0.45 0.563 1.00 0.25
FI 90 -4.89 0.53 0.45 0.606 1.00 0.64
FS 90 -4.89 0.53 0.45 0.606 0.17 0.64
  1. Calculated from RSIP Project 27 data.
  2. From Safety Analyst.
  3. From HSM Chapter 10.
  4. Crash types include: SVROR crashes (right or left), sideswipe-opposite direction crashes, and head-on crashes only.

Note that PR2 is simply set equal to 1 for Total and FI crashes since it does not apply to that severity level. The SPF for FS crashes is based on that for FI crashes with the additional PR2 multiplier (17 percent of FI crashes were FS crashes in the database used for analysis–see point No. 3 above).

Analysis Results

The EB method was applied to estimate the safety effectiveness of the dual application of centerline and shoulder rumble strips in reducing target collision types including SVROR, sideswipe-opposite direction, and head-on crashes. Analyses were performed separately for total, FI, and FS severity levels. The analyses were based upon before and after crash data from 11 treatment sites, crash data from the 8 nontreatment sites, and Safety Analyst SPFs for rural two- lane roads. The analysis results are shown in Table 9. The statistics shown for each crash severity are:

Table 9. RSIP Project 27: Safety Effectiveness of Dual Application of Centerline and Shoulder Rumble Strips on Target Crashes in Mississippi

Crash Severity Number of Treatment Sites Total Site Length (mi) Safety Effectieness (%) Standard Error of Treatment Effect (SE, %) Significance
Total 11 80.1 -35.0 10.5 Significant at 95% CL
FI 11 80.1 -39.6 12.3 Significant at 95% CL
FS 11 80.1 12.3 39.4 Not significant at 90% CL

Note: Crash types include: SVROR crashes (right or left), sideswipe-opposite direction crashes, and head-on crashes only.

The safety effectiveness estimates, which showed a reduction in total and FI target crashes, were statistically significant at the 95-percent confidence level, while the safety effectiveness estimate showed a resultant increase in FS target crashes that was not statistically significant at the 90- percent confidence level. However, there were only 13 FS target crashes in the before period and 10 FS target crashes in the after period. This small number of FS target crashes observed on the treatment sites contributed to the large standard error of the treatment effect, resulting in a non- statistically significant result for FS target crashes.

Interpretation of the Results

A comparison of the analysis results from this research to the estimated safety effectiveness of centerline and shoulder rumble strips from previous research and to the current state of practice for estimating the safety effectiveness of countermeasure combinations provides further insight into the reliability of all the results and current state of practice for safety evaluations. Previous studies (Torbic et al., 2009; Persaud et al., 2003) estimated the safety effectiveness of centerline rumble strips on rural two-lane roads. Torbic et al. estimated that the installation of centerline rumble strips on rural two-lane roads can be expected to reduce all severities (total) of head-on and opposite-direction sideswipe crashes (i.e., target crashes) by 30 percent (Crash Modification Factor; CMF = 0.70) and FI target crashes by 44 percent (CMF = 0.56). The expected safety effectiveness estimates for centerline rumble strips provided by Torbic et al. were based on the combined results of their research and that of Persaud et al. Similarly, previous studies (Torbic et al., 2009; Patel et al., 2007) quantified the safety effectiveness of shoulder rumble strips on rural two-lane roads. Torbic et al. estimated that the installation of shoulder rumble strips on rural two-lane roads can be expected to reduce all severities of SVROR crashes by 15 percent (CMF = 0.85) and FI SVROR crashes by 29 percent (CMF = 0.71). The expected safety effectiveness estimates for shoulder rumble strips provided by Torbic et al. were based on the combined results of their research and that of Patel et al. Only one recent study (Olson et al., 2013) was identified that investigated the safety effectiveness of the dual application of centerline and shoulder rumble strips on the same roadway. Based on a simple comparison of crash rates before and after treatment, Olson et al. estimated that the dual application of centerline and shoulder rumble strips reduced all severities of lane departure crashes by 66 percent (CMF = 0.34) and fatal and serious injury (FS) crashes by 56 percent (CMF = 0.44).

Based on the results of this research, the safety effectiveness of the dual application of centerline and shoulder rumble strips on target crashes (combined SVROR, sideswipe-opposite direction, and head-on crashes) on rural two-lane roads is estimated as follows:

The analysis results for total target crashes are in close agreement with current state of practice; however, there is considerable discrepancy between the two in the results for FI target crashes. For example, as mentioned earlier, the current state of practice recommends that the safety effectiveness of countermeasure combinations should be estimated by multiplying their effectiveness together. This approach suggests that the dual application of centerline and shoulder rumble strips would be expected to reduce total target crashes (i.e., all severity levels) by 40 percent based on the combined CMFs from Torbic et al. (2009). The combined CMF for the dual application of centerline and shoulder rumble strips would be calculated by multiplying the CMF for centerline rumble strips (CMF = 0.7) by the CMF for shoulder rumble strips (CMF = 0.85) to obtain a combined CMF of 0.60, which translates to an estimated 40-percent reduction in target crashes. Similarly, the combined CMF for target FI crashes equals 0.40 (0.56 × 0.71) which translates to an estimated 60-percent reduction in target FI crashes. Thus, for total target crashes there is a relatively small difference of -5 percent (35 - 40 percent) between the two safety effectiveness estimates, while for FI target crashes there is a larger difference of -20 percent (40 - 60 percent) between the two safety effectiveness estimates. It is interesting to note that for both severity levels, the results of this analysis estimated a smaller reduction in target crashes as compared to the safety effectiveness estimates calculated for countermeasure combinations using current state of practice procedures. This suggests that the current state of practice approach for estimating the safety effectiveness of countermeasure combinations may overestimate the effectiveness of countermeasure combinations.

When comparing the analysis results to the results of the recent study by Olson et al (2013), it is only reasonable to compare results for FI crashes. As indicated above, so few FS target crashes occurred on the treatment sites that the safety effectiveness estimate for FS target crashes was not statistically significant, and Olson et al. did not perform an analysis for all severity levels combined (i.e., total). Therefore, when comparing results for FI target crashes, there is a relatively large difference of -26 percent (40 - 66 percent) between the two safety effectiveness estimates. Part of the difference between the results can likely be attributed to differences in analysis approaches. In this research, the EB before/after method was used to estimate the percentage change in crash frequency, while Olson et al. compared crash rates before and after installation of the treatment.

It is also interesting to note that the analysis results provided by Olson et al. (2013) for FI target crashes (i.e., 66-percent reduction in FI target crashes) are very similar to the safety effectiveness estimate calculated for countermeasure combinations using current state of practice procedures (i.e., 60-percent reduction in FI target crashes).

RSIP Project 36

Agency: Illinois Department of Transportation (IDOT)

Focus of Evaluation: Delineation and Signing at Horizontal Curves

Project Background

IDOT developed a systematic, low-cost initiative to reduce crashes at high-risk curves on local roads in four Illinois counties in the Delta Region. Of the 16 Delta Region counties in Illinois, Franklin, Jackson, Randolph, and Williamson counties had the highest total fatal and serious injury curve-related crashes during a five-year analysis period prior to the RSIP project. Improvements planned for implementation through the RSIP project for these counties aimed at reducing roadway departure crashes on curves to reduce serious injuries and fatalities.

IDOT partnered with the four counties to identify high-risk curves based on crash data and local knowledge. Countermeasures were chosen by local agencies with the assistance of IDOT. The total project cost for the safety improvements was $430,000. Table 10 shows the types of countermeasures implemented in the four counties, the number of curves improved, and the cost of the improvements by county. At some locations, new signs were installed, while at other locations older signs were upgraded. The safety improvements were completed during calendar years 2009 and 2010.

Table 10. RSIP Project 36: RSIP Safety Improvements in Illinois by County

County Countermeasures Number of Curves Improved Cost
Franklin Advanced curve warning signs, speed plates, and chevrons 17a $74,000
Jackson Advanced curve warning signs, speed plates, and chevrons 9 $44,000
Randolph Advanced curve warning signs, speed plates, chevrons, and raised pavement markings (RPMs) 21 $63,000
Williamson Advanced curve warning signs and chevrons 37b $249,000
Total $430,000

aAt two of the curves, some tree trimming and tree removal took place. At another curve, some existing guardrail was extended.

bAt three curves, paved shoulders were to be installed.

As described in the Research Approach in Chapter 1, the research team conducted field visits to confirm the types of safety improvements completed at each treatment site. During the data collection trip to Illinois, the research team could not confirm that the safety improvements were completed at curves in Williamson County. Therefore, only the safety improvements implemented at curves in Franklin, Jackson, and Randolph Counties are included in this evaluation of delineation and signing at horizontal curves. Table 11 shows the location of the treated curves and the types of safety improvements implemented at each curve included in the safety evaluation. It should be noted that the research team was unable to confirm the characteristics of the treatment sites prior to installation of the safety improvements. As a result, assumptions about existing conditions were made based on IDOT's treatment description for each site and on the research team's post-installation field inspection.

Table 11. RSIP Project 36: Location of Treatment Sites in Illinois and Types of Safety Improvements

County Site No. Route Latitude Longitude Countermeasures
Franklin 1 FAS 2863 (Bessie Rd) 38.00388 -88.81216 Add advanced warning, speed plates and chevrons
2 FAS 873 (Akin Blacktop) 37.98761 -88.71155 Upgrade advanced warning, add speed plates, add chevrons
3 FAS 873 (Akin Blacktop) 37.98538 -88.85292 Upgrade advanced warning, add speed plates, add chevrons
4 FAS 868 (Ewing Rd) 38.08746 -88.81339 Add advanced warning signs
5 FAS 1878 (Deering Rd) 37.91333 -88.89322 Upgrade advanced warning, speed plates and chevrons
6 FAS 1878 (Deering Rd) 37.90676 -88.89964 Upgrade advanced warning signs
7 FAS 1886 (Number 9 Blacktop) 37.88872 -88.86073 Add advanced warning signs and chevrons
8 FAS 1886 (Number 9 Blacktop) 37.88059 -88.80361 Upgrade advanced warning and add chevrons
9 FAS 1877 (Yellowbanks Rd) 37.95015 -88.99392 Upgrade advanced warning and add chevrons
10 FAS 1972 (Peach Orchard Rd) 38.05281 -89.01185 Upgrade advanced warning signs
11 FAS 1873A (Orient Rd) 37.91231 -88.86084 Upgrade warning signs and add chevrons
12 FAU 9496 (Country Club Rd) 37.87682 -88.95136 Add advanced warning signs and chevrons
13 FAS 876 (Freeman Spur Rd) 37.86339 -89.00713 Add advanced warning signs and chevrons
14 FAS 869 (Elkville Blacktop) 37.87685 -89.13827 Upgrade advanced warning signs
15 FAS 869 (Elkville Blacktop) 37.87867 -89.1423 Upgrade advanced warning, add speed plates, add chevrons
Jackson 1 FAS 1916 (Elkville Road) 37.90647 -89.24431 Upgrade advanced warning and add chevrons
2 FAS 1919 (Boskeydell Rd) 37.67128 -89.21367 Upgrade advanced warning and speed plates and add chevrons
3 FAS 919 (Giant City Road) 37.65337) -89.16955) Upgrade advanced warning signs, speed plates and chevrons
4 FAS 917 (Town Creek Rd) 37.74844 -89.35211 Upgrade advanced warning signs, speed plates and chevrons
5 FAS 2912 (Marina Road) 37.77795 -89.40972 Upgrade advanced warning signs, speed plates and chevrons
6 CH 10 (Big Lake Road) 37.75174 -89.52092 Upgrade advanced warning signs
7 CH 10 (Big Lake Road) 37.7691 -89.52713 Add advanced warning and speeds plates, and update chevrons
8 CH 10 (Big Lake Road) 37.77853 -89.53449 Add advanced warning and speeds plates, and update chevrons
9 CH 9 (Neunert Road) 37.71667 -89.48951 Upgrade advanced warning signs, speed plates and chevrons
Randolph 1 CH 1 (FAS 849) 38.00805 -89.82202 Upgrade advanced warning and speed plates, add chevrons and RPMs
2 CH 1 (FAS 849) 38.00846 -89.82399 Upgrade advanced warning and speed plates, add chevrons and RPMs
3 CH 1 (FAS 849) 38.01405 -89.82511 Upgrade advanced warning and speed plates, add chevrons and RPMs
4 CH 1 (FAS 849) 38.01714 -89.83398 Upgrade advanced warning and speed plates, add chevrons and RPMs
5 CH 1 (FAS 849) 38.06551 -89.83324 Upgrade advanced warning and speed plates, add chevrons and RPMs
6 CH 1 (FAS 849) 38.06721 -89.84531 Upgrade advanced warning and speed plates, add chevrons and RPMs
7 CH 1 (FAS 849) 38.07847 -89.84699 Upgrade advanced warning and speed plates, add chevrons and RPMs
8 CH 1 (FAS 849) 38.08107 -89.84923 Upgrade advanced warning and speed plates, add chevrons and RPMs
9 CH 1 (FAS 849) 38.19741 -89.84601 Upgrade advanced warning, add chevrons and RPMs
10 CH 1 (FAS 849) 38.21327 -89.84835 Upgrade advanced warning and speed plates, add chevrons and RPMs
11 CH 1 (FAS 849) 38.21711 -89.85127 Upgrade advanced warning sign and add chevrons
12 CH 4 (FAS 862) 38.08864 -89.82124 Upgrade advanced warning, add chevrons and RPMs
13 CH 3 (FAS 859) 38.00854 -89.78905 Upgrade advanced warning, add chevrons and RPMs
14 CH 3 (FAS 859) 38.00441 -89.79291 Upgrade advanced warning, add chevrons and RPMs
15 CH 2 (FAS 853 & 1870) 37.99927 -89.74601 Upgrade advanced warning and speed plates, add chevrons and RPMs
16 CH 2 (FAS 853 & 1870) 37.96921 -89.74239 Upgrade advanced warning, add chevrons and RPMs
17 CH 2 (FAS 853 & 1870) 37.95559 -89.74092 Upgrade advanced warning and speed plates, add chevrons and RPMs

Nontreatment sites included in the analysis had similar characteristics to the treatment sites but did not have recently installed or improved delineation and/or signing present. Typically, nontreatment sites considered in before/after safety evaluations have similar characteristics to the treatment sites, but the countermeasure being evaluated is not present at the nontreatment sites. However, because the treatment sites were relatively sharp curves, it was not realistic to find similar curves without any type of delineation and/or signing. Also, in many cases, the safety improvements at treatment sites involved upgrading existing advance warning signs, so advance warning signs were present at the treatment sites prior to the safety improvement being implemented as part of the RSIP project. Therefore, it was considered appropriate that the nontreatment sites have the type of delineation and/or signing present at the treated curves prior to the implementation of the RSIP project. Typically, the delineation and signing at the nontreatment sites was older and did not have the retroreflectivity qualities of the newer delineation and/or signing at the treatment sites. Most of the nontreatment sites (18 out of 22) had advance warning signs present, while only a few of the nontreatment sites (6 out of 22) had chevrons present. None of the nontreatment sites had RPMs. Table 12 shows the locations of the curves used as nontreatment sites in the analysis.

Table 12. RSIP Project 36: Location of Nontreatment Sites in Illinois

County Site No. Route Latitude Longitude
Franklin N1 No 9 Blacktop Rd 37.89244 -88.87569
N2 No 9 Blacktop 37.87788 -88.79598
N3 Deering Rd 37.95612 -88.90534
N4 Deering Rd 37.94758 -88.89925
N5 Ruembler Crossing 37.95071 -88.92388
N6 Freeman Spur Blacktop 37.87421 -88.97537
Jackson N1 Mt Joy Rd 37.79177 -89.39721
N2 Stave Mill Rd 37.78477 -89.36497
N3 Stave Mill Rd 37.78126 -89.36036
N4 Town Creek Rd 37.74827 -89.40489
N5 Neunert Rd 37.70826 -89.50093
Randolph N1 CH 10 38.14087 -89.84501
N2 CH 18 38.19713 -89.80132
N3 CH 18 38.19119 -89.79772
N4 CH 2 37.95649 -89.69036
N5 CH 2 37.93442 -89.62605
N6 CH 2 37.99148 -89.74746
N7 CH 5 37.94118 88.67037
N8 CH 1 38.13162 -89.94522
N9 CH 10 38.11039 -89.96031
N10 St Leo's Rd 38.10404 -90.00241
N11 St Leo's Rd 38.09123 -90.00363

Objective

The objective of this evaluation was to quantify the safety effectiveness of combinations of roadway delineation and signing treatments in reducing total crashes and SVROR crashes at horizontal curves on rural two-lane highways. Treatments varied by site and included installation or upgrade of curve warning signs, speed plates, chevrons, and raised pavement markers.

Descriptive Statistics

A total of 41 treatment and 22 nontreatment sites are included in this safety evaluation. Treatments were installed in 2009 in Franklin (15 sites) and Jackson (9 sites) Counties and in 2010 in Randolph County (17 sites). All crashes at each site were located either within the curve (or curves) of interest or within a 0.1-mi buffer zone on each end of the curve since crashes in this buffer area are often related to the curve. Crash and traffic volume data were obtained for years 2004 through 2012 for the analysis; thus the before period was either five- or six-years long and the after period was either two- or three-years long.

The evaluation focused on quantifying the safety effectiveness of the delineation and signing treatments on total crashes (i.e., all collision types combined) and SVROR crashes. Analyses were performed separately for three crash severity levels: Total, FI, and FS crashes.

Table 13 summarizes the treatment site data used in the analysis, separately for each period (before and after) and treatment site. The table provides the length and radius of the curve (or curves), lane width and shoulder width, the average AADT in the specific period, and total (i.e., all collision types combined) and SVROR crash counts, each by severity level (Total, FI, and FS crashes). Table 14 provides similar statistics for the nontreatment sites across the entire study period. Note that Table 13 and Table 14 provide information on lane and shoulder widths for reporting purposes only. This information was not considered in the analysis.

Analysis Approach

The safety effectiveness of the treatment combinations was evaluated using the EB before/after method similar to that discussed in the evaluation of RSIP project 27. Prior to implementing the EB method, the following points were addressed:

1. Selecting an appropriate SPF: It was decided to use the SPF for total severity level (i.e., all severity levels combined) for rural two-lane roads from Chapter 10 of the HSM. The SPF predicts crashes for all collision types combined for tangent sections of rural two- lane roads.

2. Obtaining the proportion of FI and FS crashes for this project (PR2): The proportions of FI and FS severity levels for all collision types combined were calculated from those crashes that occurred on all the nontreatment sites and the before treatment sites combined in the project database (a total of 398 site-years).

3. Obtaining the proportion of target crashes for total, FI, and FS crashes for this project (PR1): The proportions of target crashes (SVROR) to all collision types for total, FI, and FS severity levels were calculated based on those crashes that occurred on all the nontreatment sites and the before treatment sites combined in the project database (a total of 398 site-years).

4. Calibrating the SPFs to the local jurisdiction: Calibration was performed for total crashes using all the nontreatment sites and the before treatment sites combined in the project database (a total of 398 site-years). Since a single base SPF was used, that calibration factor applies to all severity levels considered.

Table 13. RSIP Project 36: Summary Statistics for the Before and After Periods for Treatment Sites in Illinois

County Site
No.		 Curve
Length
(mi)		 Curve
Radius
(ft)		 Lane Width
(ft)		 Shoulder
Width
(ft)		 Before Period After Period
Number of Years AADT Total CrashesSVROR CrashesNumber of Years AADT Total Crashes SVROR Crashes
Total FI FS Total FI FS Total FI FS Total FI FS
Franklin 1 0.310 350 10 2 5 200 0 0 0 0 0 0 3 175 0 0 0 0 0 0
2 0.155 101 11 1 5 850 1 0 0 0 0 0 3 325 0 0 0 0 0 0
3 0.310 1,703 10 2 5 850 2 0 0 2 0 0 3 1,100 1 0 0 1 0 0
4 0.361 1,760 10 2 5 650 2 1 0 1 0 0 3 800 2 1 1 1 0 0
5 0.220 1,105 11 1 5 1,200 4 1 1 1 1 1 3 1,900 1 1 1 1 1 1
6 0.149 1,650 10.5 1 5 1,200 2 1 1 2 1 1 3 1,900 8 2 0 6 2 0
7 0.238 230 11 1 5 900 2 0 0 0 0 0 3 850 2 2 2 2 2 2
8 0.105 1,650 10 1 5 400 0 0 0 0 0 0 3 500 0 0 0 0 0 0
9 0.539 880 10 2 5 900 8 2 2 2 2 2 3 1,000 3 1 1 3 1 1
10 0.113 212 10 1 5 550 3 0 0 1 0 0 3 650 0 0 0 0 0 0
11 0.098 948 12 1 5 950 0 0 0 0 0 0 3 450 0 0 0 0 0 0
12 0.310 358 11 1 5 950 0 0 0 0 0 0 3 800 1 1 0 1 1 0
13 0.153 639 10.5 1 5 1,350 2 1 0 2 1 0 3 1,350 0 0 0 0 0 0
14 0.180 398 10 5 5 1,400 16 4 0 1 0 0 3 1,650 0 0 0 0 0 0
15 0.084 476 1 5 5 1,400 1 0 0 0 0 0 3 1,650 1 0 0 0 0 0
Jackson 1 0.170 952 11 2 5 1,700 1 0 0 0 0 0 3 1,550 0 0 0 0 0 0
2 0.516 352 10 2 5 2,000 10 3 1 3 2 0 3 1,500 3 1 0 3 1 0
3 0.217 1,120 11 2 5 1,950 6 3 1 3 3 1 3 1,800 7 1 0 3 0 0
4 0.196 588 12 1 5 2,550 0 0 0 0 0 0 3 2,350 6 1 0 2 1 0
5 0.083 219 10.5 1 5 400 1 1 0 0 0 0 3 600 2 1 0 2 1 0
6 0.140 81 10 1 5 75 0 0 0 0 0 0 3 50 0 0 0 0 0 0
7 0.180 144 10 1 5 225 0 0 0 0 0 0 3 250 0 0 0 0 0 0
8 0.297 268 10 1 5 225 0 0 0 0 0 0 3 25 2 1 0 2 1 0
9 0.240 683 10.5 1 5 325 2 0 0 0 0 0 3 275 0 0 0 0 0 0
Randolph 1 0.165 453 11 2 6 550 1 1 0 0 0 0 2 500 1 0 0 1 0 0
2 0.128 440 11 2 6 550 2 0 0 0 0 0 2 500 1 0 0 1 0 0
3 0.190 538 11 3 6 550 0 0 0 0 0 0 2 500 0 0 0 0 0 0
4 0.240 510 11 2 6 550 5 2 1 2 2 1 2 650 1 1 1 1 1 1
5 0.190 615 11 3 6 650 4 0 0 2 0 0 2 700 0 0 0 0 0 0
6 0.180 543 11 3 6 650 2 1 1 0 0 0 2 650 0 0 0 0 0 0
7 0.158 796 11 3 6 700 1 1 1 1 1 1 2 650 0 0 0 0 0 0
8 0.152 903 11 3 6 700 2 1 1 1 1 1 2 650 0 0 0 0 0 0
9 0.250 1,471 11 3 6 2,050 5 1 1 3 1 1 2 2,450 2 1 1 0 0 0
10 0.193 1,687 11 3 6 1,150 5 1 1 2 1 1 2 1,650 2 1 0 1 0 0
11 0.194 1,203 11 3 6 1,150 0 0 0 0 0 0 2 1,650 1 1 0 1 1 0
12 0.160 1,290 10.5 2 6 1,150 8 4 2 2 2 2 2 1,275 2 0 0 0 0 0
13 0.310 1,123 10.5 3 6 950 2 0 0 1 0 0 2 1,050 2 1 1 1 1 1
14 0.440 1,626 10 3 6 950 4 2 2 1 1 1 2 1,050 0 0 0 0 0 0
15 0.380 470 9.5 2 6 550 5 2 2 1 2 2 2 600 1 1 0 1 1 0
16 0.290 947 12 4 6 950 0 0 0 0 0 0 2 1,100 0 0 0 0 0 0
17 0.290 888 10 4 6 750 0 0 0 0 0 0 2 1,100 1 0 0 0 0 0
All N/A 9.274 N/A N/A N/A N/A N/A 109 33 18 34 21 15 N/A N/A 53 19 8 34 15 6

Table 14. RSIP Project 36: Summary Statistics for the Entire Study Period forNontreatment Sites in Illinois

County Site
No.		 Curve
Length
(mi)		 Curve
Radius
(ft)		 Lane Width
(ft)		 Shoulder
Width
(ft)		 Number of Years Entire Study Period
AADT Total CrashesSVROR Crashes
Total FI FS Total FI FS
Franklin N1 0.078 260 10 1 8 931 1 0 0 0 0 0
N2 0.136 1,586 9 1 8 438 1 0 0 1 0 0
N3 0.177 682 10.5 2 8 1,000 14 2 0 6 2 0
N4 0.072 425 10.5 3 8 1,000 13 6 3 9 5 2
N5 0.066 87 10.5 0 8 275 4 2 1 2 1 1
N6 0.249 920 10.5 1 8 1,025 5 1 0 2 1 0
Jackson N1 0.083 165 9 1 8 278 0 0 0 0 0 0
N2 0.045 116 9 2 8 519 2 0 0 0 0 0
N3 0.044 85 10 1 8 519 1 0 0 0 0 0
N4 0.263 700 10.5 2 8 1,206 14 4 1 6 3 1
N5 0.199 907 10.5 1 8 306 3 2 2 3 2 2
Randolph N1 0.254 1,007 11.5 2 8 838 1 1 0 1 1 0
N2 0.159 457 10.5 2 8 475 0 0 0 0 0 0
N3 0.149 536 12 2 8 475 0 0 0 0 0 0
N4 0.234 908 11 4 8 688 2 1 0 1 1 0
N5 0.129 1,113 11 3 8 1,450 5 3 1 3 2 1
N6 0.185 859 9.5 3 8 563 0 0 0 0 0 0
N7 0.317 977 9 0 8 381 0 0 0 0 0 0
N8 0.264 716 11.5 2 8 269 1 0 0 1 0 0
N9 0.214 910 11 2 8 250 0 0 0 0 0 0
N10 0.309 792 9.5 1 8 463 4 1 1 1 0 0
N11 0.451 930 9.5 2 8 463 1 0 0 0 0 0
All   4.077 N/A N/A N/A N/A N/A 72 23 9 36 18 7

The HSM SPF for rural two-lane roads for all severity levels combined (total) has the general form:

Equation 3. Predicted crashes per site per year equals AADT times L subscript C times 365 times 10 raised to the power of -6 times e raised to the power of open parenthesis -0.312 closed parenthesis.

Figure 3. Equation 3 – General form of HSM SPF for rural two-lane roads for all severity levels combined (total).

where AADT is the average annual daily traffic (veh/day) and Lc is the length of roadway segment in miles.

The value of the overdispersion parameter (k) associated with the SPF for rural two-lane roadway segments is a function of the roadway segment length and is calculated as:

Equation 4. k equals 0.236 divided by L subscript C.

Figure 4. Equation 4 – Overdispersion parameter.

The CMF for horizontal curvature (CMFHC), calibration factors, and crash proportions are used to adjust the base SPF for local conditions as follows for selected severity levels:

Equation 5. Predicted crashes per site per year equals open bracket AADT times L subscript C times 365 times 10 raised to the power of -6 times e raised to the power of open parenthesis -0.312 closed parenthesis closed bracke ttimes PR subscript 1 times PR subscript 2 times CMF subscript HC times C subscript r.

Figure 5. Equation 5 – General form of HSM SPF for rural two-lane roads adjusted for crash type, horizontal curvature, and local conditions.

where PR1, PR2, and Cr are provided in Table 15 for each collision type and severity level.

Table 15. RSIP Project 36: Target Crash Proportions and Calibration Factors Used for Illinois Data

Collision Type Severity Level Number of Site-Yearsa Number of Crashes Proportion of Target Collision Type (PR1)a,b Proportion of Crashes (PR2)a,c Calibration Factor (Cr)a
All Total 398 181 1.00 1.00 6.7
FI 398 56 1.00 0.31
FS 398 27 1.00 0.15
SVROR Total 398 70 0.39 1.00
FI 398 39 0.70 0.31
FS 398 22 0.81 0.15

From RSIP 36 data–control sites and before-period treatment sites.

bProportion of SVROR crashes out of corresponding all–collision crashes.

c Proportion of FI and FS crashes relative to total crashes based on all-collision crashes only.

CMFHC adjusts the crash prediction to horizontal curves and was developed to represent the manner in which crash experience on curved alignments differs from that of tangents. CMF3r is calculated as a function of curve length and curve radius using Equation (10-13) in HSM Chapter 10:

Equation 6. CMF subscript HC equals 1.55 times L subscript C plus 80.2 divided by R minus 0.012 times S all divided by 1.55 times L subscript C.

Figure 6. Equation 6 – CMF for horizontal curves on rural two-lane highways.

By including the horizontal curve CMF in the crash prediction, the analysis approach accounts for the sharpness of each curve in the estimation of the expected safety effectiveness of the delineation and signing treatments.

Analysis Results

The EB before/after method was applied to estimate the safety effectiveness of combinations of roadway delineation and signing treatments. The analyses were based on before and after crash data from 41 treatment curves on rural two-lane roads, crash data from 22 nontreatment curves, and the HSM SPF for rural two-lane road segments. The EB before/after analysis was performed for the following combinations:

The analysis results for all collision types combined are shown in Table 16; those for SVROR crashes are shown in Table 17. The statistics shown in each table for each crash severity are:

Table 16. RSIP Project 36: Safety Effectiveness of Horizontal Curve Delineation and Signing on Total Crashes in Illinois

Crash Severity Number of Treatment Sites Total Curve Length (mi) Safety Effectiveness (%) Standard Error of Treatment Effect (SE, %) Significance
Total 41 9.274 -9.9 12.85 Not significant at 90% CL
FI 41 9.274 6.9 25.25 Not significant at 90% CL
FS 41 9.274 -6.0 33.77 Not significant at 90% CL

Table 17. RSIP Project 36: Safety Effectiveness of Horizontal Curve Delineation and Signing on SVROR Crashes in Illinois

Crash Severity Number of Treatment Sites Total Curve Length (mi) Safety Effectiveness (%) Standard Error of Treatment Effect (SE, %) Significance
Total SVROR 41 9.274 68.7 30.40 Significant at 95% CL
FI SVROR 41 9.274 27.8 33.94 Not significant at 90% CL
FS SVROR 41 9.274 -14.4 35.39 Not significant at 90% CL

The results in Table 16 and Table 17 show two or three prevalent trends. First, the direction of the safety effectiveness values is not consistent across the severity types and, in some cases, is counterintuitive. The analyses of total, FS, and FS SVROR crashes suggest that the delineation and signing treatment combinations resulted in a decrease in crashes, while the analyses of FI, total SVROR, and FI SVROR crashes suggest that the delineation and signing treatments resulted in an increase in crashes. Second, only one of the results (total SVROR) suggests a statistically significant change in crashes occurred at the 90-percent confidence level. None of the other estimates of the safety effectiveness of the delineation and signing treatment combinations were statistically significant at the 90-percent confidence level. Third, the standard error of the treatment effect was large compared to the estimate in all analyses, directly resulting in non-statistically significant results.

Interpretation of the Results

The analysis does not provide reliable estimates of the safety effectiveness of the delineation and signing treatment combinations installed on curves on rural two-lane highways in Illinois as part of the RSIP project. The analysis results vary and in some cases are counterintuitive. A number of reasons may explain the varying trends from the analysis results, including:

  1. Because the safety improvements were installed on relatively short segments of roadway, the overall length of roadways included in the analysis that were improved was relatively small. Thus, the overall number of curves, length of roadway, and associated crash data were insufficient to reliably estimate the safety effectiveness of the delineation and signing treatments installed as part of this project.
  2. Many of the treatment combinations involved upgrading existing advance warning signs.
  3. Because advance warning signs were present in the before condition, it is likely that upgrading signs has a smaller incremental safety effect than adding new signs. Estimates of smaller incremental effects on safety are more difficult to quantify than larger effects.
  4. A wide range of treatment combinations were installed at different curves. Because the same types of treatment combinations were not applied at each site, this likely added to the variability in the results.
  5. Because there was a wide range of treatment combinations installed, it was difficult to determine the appropriate characteristics to use in defining and selecting appropriate nontreatment sites for calibration purposes. Several of the nontreatment sites had advance warning signs present, and a few of the nontreatment sites had chevrons presents. This too could have contributed to the variability in the data. It is possible that the delineation and signing treatments provided drivers with a false sense of security to be able to negotiate the curves at higher speeds, and thus resulted in an increase in crashes; but this does not completely explain why the results for total (all collisions combined) and SVROR crashes differ across the three severity levels (i.e., for some severity levels the results indicate a reduction in crashes, while for other severity levels the results indicate an increase in crashes).

The lack of reliable estimates of the safety effectiveness of the delineation and signing treatments based on this study does not indicate that the treatment combinations are ineffective at reducing crashes at horizontal curves on rural two-lane roads, but rather that there is insufficient evidence at this time to determine their effectiveness.

RSIP Project 37

Agency: Louisiana Department of Transportation and Development (La DOTD)

Focus of Evaluation: Improved Signing and Pavement Markings at Intersections

Project Background

LaDOTD developed a strategic plan to reduce crashes at high-crash intersections throughout rural Louisiana. Low-cost safety improvements were targeted for installation at 89 stop- controlled intersections and 15 signalized intersections. At the stop-controlled intersections, the primary safety improvements included oversized stop signs, oversized intersection warning
signs, route signs, junction auxiliary signs, and new stop bars. At the signalized intersections, the primary safety improvements included installation of back plates for the signal heads, 12-in LED lens, retiming of clearance intervals, elimination of flashing operation during night conditions, intersection warning signs, and route marker signs. The total cost of the safety improvements implemented under the RSIP was $1,000,653. All of the safety improvements were completed during 2010.

This safety evaluation focused on estimating the safety effectiveness of treatments implemented at rural stop-controlled intersections in reducing intersection and intersection-related crashes. Of the 89 stop-controlled intersections improved as part of the RSIP project, 36 treatment sites were included in the safety evaluation based on the geographical locations of the improved intersections and availability of treatment type, traffic volume, and crash data for the analysis. Table 18 shows the locations of the treated intersections and the type of improvements installed.

Seven nontreatment sites included in the analysis had similar characteristics to the treatment sites, but none of the sites had oversized signs, and only one site had route marker signs present on the major road.

Objective

The objective of this evaluation was to quantify the safety effectiveness of combinations of treatments installed in reducing total crashes and target collision types including angle, rear-end, and turning crashes at rural stop-controlled intersections. Treatments varied by intersection and included oversized stop signs, oversized intersection warning signs, route signs, junction auxiliary signs, and new stop bars.

Descriptive Statistics

A total of 36 treatment and 7 nontreatment sites are included in the safety evaluation.

All treatments were installed in 2010. Crash and traffic volume data were obtained for years 2006 through 2012 for analysis.

For all treatment sites, the before period is from 2006 through
2009. The after period consists of two years–2011 and 2012.

All crash types were considered in the analysis, as well as target crashes which included rear- end, right-angle, and turning crashes. Analyses were performed separately for two crash severity levels: total and FI. No severity information was available to identify FS crashes.

Table 19 summarizes the treatment site data used in the analysis, separately for each period (before and after) and treatment intersection. The table provides the average major- and minor- route AADTs in the specific period, and crash counts (total and FI crashes; all collision types combined) for each intersection. Table 20 provides similar statistics for total and FI crashes (all collision types combined) for the nontreatment intersections across the entire study period. Similarly, Table 21 summarizes target crash counts for the treatment sites in the before and after periods, and Table 22 summarizes target crash counts for the nontreatment sites across the entire study period.

Table 18. RSIP Project 37: Location of Treatment Sites in Louisiana and Types of Safety Improvements

Site No. Number of Legs Roadway Type Major Route Minor Route Major Route Minor Route
Intersection Warning Signs Route Marker Signs Intersection Warning Signs Stop Sign Stop Bar
1 3 2-lane LA 1 LA 3170 na New Oversized Oversized double stop signs New
2 3 Multilane US 61a LA 626 na na Oversized Oversized na
3 3 Multilane US 61 LA 628 na New Oversized na New
4 3 Multilane LA 3127 LA 3160 na New Oversized na New
5 3 Multilane LA 3127 LA 3142 Na New Oversized na New
6 3 2-lane LA 18 a LA 20 Oversized New Oversized na New
7 3 2-lane LA 20 LA 643 Oversized New Oversized Oversized  
8 3 2-lane LA 20 LA 644 Oversized New Oversized Oversized New
9 3 2-lane LA 3125 LA 3214 Oversized New Oversized Oversized New
10 3 2-lane LA 182 LA 358 Oversized double advanced warning in one direction New Oversized Oversized na
11 3 Multilane US 190 LA 103 SB na New Oversized na na
12 4 2-lane LA 76 LA 411 na New Oversized Oversized na
13 3 2-lane LA 20 LA 307 Oversized advanced warning with LED beacon in both directions New Oversized Oversized New
14 3 2-lane LA 84 LA 459 Oversized New Oversized Oversized New
15 4 2-lane US 165 a LA 124 na New Oversized Oversized New
16 3 2-lane LA 16 LA 22 Oversized New Oversized Oversized na
17 3 2-lane LA 16 LA 444 na New Oversized Oversized na
18 3 2-lane LA 16 LA 42 na New Oversized Oversized na
19 3 2-lane LA 16 LA 447 Oversized advanced warning with LED beacon in both directions New Oversized Oversized New
20 4 2-lane LA 42 LA 63 Oversized New na Oversized New
21 4 2-lane LA 43 a LA 442 Oversized New Oversized Oversized New
22 4 2-lane LA 10 LA 67 na New Oversized Oversized na
23 3 2-lane US 167 LA 748 Oversized New Oversized Oversized New
24 3 2-lane LA 85 LA 674 na New Oversized Oversized New
25 3 Multilane LA 70 LA 3120 Oversized double advanced warning New na Oversized na
26 3 2-lane US 71 a LA 1177 na New Oversized Oversized na
27 4 2-lane LA 10 a LA 67 na New na Oversized na
28 3 2-lane LA 22 a LA 70 na New na Oversized na
29 4 2-lane LA 44 LA 941 na New Oversized Oversized na
30 4 Multilane US 190 LA 741 Oversized (one direction) New Oversized Oversized na
31 3 2-lane LA 31 LA 355 na New na na na
32 4 2-lane LA 182 a Duchamp RD na New Oversized Oversized na
33 3 2-lane LA 10 LA 25 na New Oversized Oversized na
34 3 2-lane LA 25 a LA 430 Oversized New Oversized Oversized na
35 4 2-lane LA 25 LA 438 Oversized New Oversized Oversized na
36 4 2-lane LA 62 a LA 436 Oversized New Oversized Oversized na

a Overhead flashers present facing all directions/approaches.

na = information not available.

Table 19. RSIP Project 37: Summary Statistics for the Before and After Treatment Periods for Treatment Intersections in Louisiana–All Collision Types

Site No. Major Route Minor Route Before Period: 4 years (2006-2009) After Period: 2 years (2011-2012)
Average AADTmajor Average AADTminor Total Crashes FI Crashes Average AADTmajor Average AADTminor Total Crashes FI Crashes
1 LA 1 LA 3170 2,287 6,151 1 0 1,672 5,680 0 0
2 US 61 LA 626 32,126 3,734 19 7 31,440 3,084 2 0
3 US 61 LA 628 26,704 1,305 2 1 27,374 1,774 1 0
4 LA 3127 LA 3160 13,145 2,083 6 2 13,743 1,851 1 0
5 LA 3127 LA 3142 6,544 5,471 5 1 7,586 5,441 2 0
6 LA 18 LA 20 2,679 6,129 4 0 2,977 4,435 1 1
7 LA 20 LA 643 6,654 3,461 6 1 7,381 3,148 0 0
8 LA 20 LA 644 6,654 4,234 16 5 7,381 4,024 9 5
9 LA 3125 LA 3214 4,618 2,908 4 1 4,917 2,673 0 0
10 LA 182 LA 358 25,476 2,493 6 2 21,358 1,926 1 1
11 US 190 LA 103 SB 16,908 7,205 1 1 17,030 6,586 0 0
12 LA 76 SIDNEY (LA 411) 732 2,620 4 3 701 2,855 0 0
13 LA 20 LA 307 6,654 2,748 7 1 7,381 2,757 2 0
14 LA 84 LA 459 4,142 1,051 8 3 3,608 859 0 0
15 US 165 LA 124 4,977 961 7 4 4,425 784 0 0
16 LA 16 LA 22 6,450 7,041 3 2 7,106 7,892 1 1
17 LA 16 LA 444 5,104 2,790 2 0 5,627 2,972 0 0
18 LA 16 LA 42 5,104 5,273 4 2 5,627 5,956 0 0
19 LA 16 LA 447 4,992 6,606 19 9 5,623 6,173 3 1
20 LA 42 LA 63 3,414 4,723 3 1 4,603 6,167 0 0
21 LA 43 LA 442 5,407 2,709 6 5 5,610 2,704 1 0
22 LA 10 LA 67 (E JCT) S. 3,073 4,515 6 3 5,590 4,217 0 0
23 US 167 LA 748 15,438 1,857 6 2 17,133 2,068 0 0
24 LA 85 LA 674 596 3,257 5 1 641 3,627 1 0
25 LA 70 LA 3120 17,375 2,748 1 1 21,422 3,138 4 3
26 US 71 LA 1177 4,155 328 0 0 4,533 291 0 0
27 LA 10 LA 67 (W JCT) N. 4,124 9,162 4 2 4,613 8,784 1 0
28 LA 22 LA 70 16,241 16,188 114 34 18,438 16,315 41 14
29 LA 44 LA 941 9,333 3,470 18 9 10,362 3,188 1 0
30 US 190 LA 741 12,718 1,318 12 9 12,842 1,102 7 2
31 LA 31 LA 355 2,039 4,008 3 0 2,151 4,497 2 2
32 LA 182 DUCHAMP RD 12,500 3,519 19 11 13,278 3,856 18 9
33 LA 10 LA 25 4,648 9,922 2 2 3,311 7,228 0 0
34 LA 25 BENE (LA 430) 4,984 5,166 4 1 3,465 3,665 0 0
35 LA 25 LA 438 2,479 805 7 3 1,888 724 0 0
36 LA 62 LA 436 2,185 1,149 3 1 2,234 978 1 0
Total N/A N/A N/A N/A 337 130 N/A N/A 100 39

Table 20. RSIP Project 37: Summary Statistics for the Entire Study Period for Nontreatment Intersections in Louisiana–All Collision Types

Site No. Number
of Legs		 Roadway
Type		 Major
Route		 Minor
Route		 Entire Study Period: 6 years (2006-2012 excluding 2010)
Average
AADTmajor		 Average
AADTminor		 Total
Crashes		 FI
Crashes
N1 3 Multilane US 165 LA 112 5,956 997 5 2
N2 3 2-lane LA 29 LA 1161 7,239 2,613 18 7
N3 3 Multilane LA 3127 LA 3141 6,891 1,803 6 3
N4 3 2-lane LA 31 LA 354 2,076 926 5 3
N5 3 2-lane LA 31 LA 341 3,842 1,877 6 3
N6 3 2-lane LA 31 LA 351 6,618 3,029 4 2
N7 3 Multilane US 71 US 167 3,841 1,175 2 2
Total N/A N/A N/A N/A N/A N/A 46 22

Table 21. RSIP Project 37: Summary Statistics for the Before and After Treatment Periods for Treatment Intersections in Louisiana–Target Crashes

Site No. Major Route Minor Route Before Period: 4 years (2006-2009) After Period: 2 years (2011-2012)
Average AADTmajor Average AADTminor Total Crashes FI Crashes Average AADTmajor Average AADTminor Total Crashes FI Crashes
1 LA 1 LA 3170 2,287 6,151 1 0 1,672 5,680 0 0
2 US 61 LA 626 32,126 3,734 9 5 31,440 3,084 1 0
3 US 61 LA 628 26,704 1,305 2 1 27,374 1,774 0 0
4 LA 3127 LA 3160 13,145 2,083 2 1 13,743 1,851 0 0
5 LA 3127 LA 3142 6,544 5,471 3 1 7,586 5,441 0 0
6 LA 18 LA 20 2,679 6,129 1 0 2,977 4,435 1 1
7 LA 20 LA 643 6,654 3,461 2 0 7,381 3,148 0 0
8 LA 20 LA 644 6,654 4,234 9 3 7,381 4,024 6 2
9 LA 3125 LA 3214 4,618 2,908 2 1 4,917 2,673 0 0
10 LA 182 LA 358 25,476 2,493 2 0 21,358 1,926 1 1
11 US 190 LA 103 SB 16,908 7,205 0 0 17,030 6,586 0 0
12 LA 76 SIDNEY (LA 411) 732 2,620 2 1 701 2,855 0 0
13 LA 20 LA 307 6,654 2,748 6 1 7,381 2,757 1 0
14 LA 84 LA 459 4,142 1,051 5 2 3,608 859 0 0
15 US 165 LA 124 4,977 961 5 3 4,425 784 0 0
16 LA 16 LA 22 6,450 7,041 0 0 7,106 7,892 0 0
17 LA 16 LA 444 5,104 2,790 0 0 5,627 2,972 0 0
18 LA 16 LA 42 5,104 5,273 3 1 5,627 5,956 0 0
19 LA 16 LA 447 4,992 6,606 6 5 5,623 6,173 3 1
20 LA 42 LA 63 3,414 4,723 1 0 4,603 6,167 0 0
21 LA 43 LA 442 5,407 2,709 5 4 5,610 2,704 0 0
22 LA 10 LA 67 (E JCT) S. 3,073 4,515 3 1 5,590 4,217 0 0
23 US 167 LA 748 15,438 1,857 4 2 17,133 2,068 0 0
24 LA 85 LA 674 596 3,257 1 0 641 3,627 0 0
25 LA 70 LA 3120 17,375 2,748 0 0 21,422 3,138 2 1
26 US 71 LA 1177 4,155 328 0 0 4,533 291 0 0
27 LA 10 LA 67 (W JCT) N. 4,124 9,162 3 2 4,613 8,784 1 0
28 LA 22 LA 70 16,241 16,188 81 25 18,438 16,315 37 14
29 LA 44 LA 941 9,333 3,470 10 6 10,362 3,188 1 0
30 US 190 LA 741 12,718 1,318 10 8 12,842 1,102 7 2
31 LA 31 LA 355 2,039 4,008 0 0 2,151 4,497 1 1
32 LA 182 DUCHAMP RD 12,500 3,519 16 8 13,278 3,856 14 9
33 LA 10 LA 25 4,648 9,922 0 0 3,311 7,228 0 0
34 LA 25 BENE (LA 430) 4,984 5,166 4 1 3,465 3,665 0 0
35 LA 25 LA 438 2,479 805 6 2 1,888 724 0 0
36 LA 62 LA 436 2,185 1,149 2 1 2,234 978 0 0
Total N/A N/A N/A N/A 206 85 N/A N/A 76 32

Table 22. RSIP Project 37: Summary Statistics for the Entire Study Period for Nontreatment Intersections in Louisiana–Target Crashes

Site No. Major Route Minor Route Entire Study Period: 6 years (2006-2012 excluding 2010)
Average
AADTmajor		 Average
AADTminor		 Total
Crashes		 FI
Crashes
N1 US 165 LA 112 5,956 997 2 1
N2 LA 29 LA 1161 7,239 2,613 13 6
N3 LA 3127 LA 3141 6,891 1,803 3 2
N4 LA 31 LA 354 2,076 926 2 1
N5 LA 31 LA 341 3,842 1,877 3 2
N6 LA 31 LA 351 6,618 3,029 3 2
N7 US 71 US 167 3,841 1,175 1 1
Total N/A N/A N/A N/A 27 15

Analysis Approach

The 43 Louisiana intersections were located on either rural two-lane roads or multilane highways and had either 3 or 4 legs; their breakdown by those characteristics is shown in Table 23.

Table 23. RSIP Project 37: Breakdown of Louisiana Intersections by Facility Type and Number of Legs

Intersection Type Facility Type and Number of Legs All Intersections
Rural Two-Lane Roads Rural Multilane Roads
3 Legs 4 Legs 3 Legs 4 Legs
Treatment 19 10 6 1 36
Nontreatment 4 0 3 0 7
All Intersections 23 10 9 1 43

Based on the site distribution by facility type and number of legs shown above, the safety evaluation was performed for the following three combinations:

No safety evaluation was performed based on the single four-leg intersection on a multilane highway (Site No. 30).
The safety effectiveness of the treatment combinations was evaluated using the EB before/after method similar to that discussed in the evaluation of RSIP project 27. Prior to implementing the EB method, the following points were addressed:

1. Selecting appropriate SPFs: The SPFs for intersections on rural two-lane roads and multilane highways from Chapters 10 and 11 of the HSM were selected for total and FI crashes. The coefficients of these SPFs vary by facility type and number of intersection approach legs.

2. Obtaining the proportion of target crashes (PR1) relevant to RSIP Project 37 evaluation: The proportions of both total and FI severity levels for the target crashes (rear-end, right- angle, and turning crashes) were calculated based on the total and FI crashes (all collision types combines and target crashes only) that occurred on all nontreatment sites and the before treatment sites combined in the project database. These proportions were calculated separately for each facility type and intersection number of legs, with one exception.

3. Obtaining the proportion of FI out of total crashes (PR2) for this project for two-lane roads, separately for three-leg and four-leg intersections: Where needed, the proportion of FI out of total crashes (all collision types combined) was calculated from the total and FI crashes that occurred on all the nontreatment sites and the before treatment sites combined in the project database (a total of 125 site-years for three-leg intersections and 52 site-years for four-leg intersections on rural two-lane roads).

4. Calibrating the SPFs to the local jurisdiction: Calibration was performed separately for total and FI crashes (all collision types combined), facility type, and number of intersection legs, using all the nontreatment intersections and the before treatment intersections combined in the project database.

The HSM SPFs for intersections on rural two-lane and rural multilane roads for total and FI severity levels have the general form:

Equation 7. Predicted crashes per year equals exponent open bracket a plus b times open parenthesis logarithm base ten of AADT subscript Major closed parenthesis plus c times open parenthesis logarithm base ten of AADT subscript Minor closed parenthesis closed bracket.

Figure 7. Equation 7 – Equation 6. General form of HSM SPF for intersections on rural two-lane and rural multilane roads for total and fatal and all injury severity levels.

where a, b, and c are regression coefficients shown in Table 24. These coefficients apply to base conditions and vary by facility type and number of intersection legs, separately for each severity level (total and FI).

CMFs, calibration factors, and proportions of target crashes are then used to adjust for local conditions as follows:

Equation 8. Predicted crashes per year equals open brace exponent open bracket a plus b times open parenthesis logarithm base ten of AADT subscript Major closed parenthesis logarithm base ten of AADT subscript Minor closed parenthesis closed baracket closed brace times PR subscript 1 times PR subscript 2 time CMF subscript Combined times C subscript r.

Figure 8. Equation 8 – General form of HSM SPF for intersections on rural two-lane and rural multilane roads adjusted for crash type, combined CMFs, and local conditions.

where PR1, PR2, and Cr are provided in Table 24 for combination of facility type, number of legs, and severity level; and CMFCombined is provided in Table 25. Table 26 provides the combined crash modification factors for nontreatment sites.

Note that PR2 is equal to 1 for total and FI crashes in those cases where a specific SPF is provided in the HSM for that severity level. The CMFcombined is the product of the CMFs from Chapters 10 and 11 of the HSM for intersection lighting, skew angle, number of major-road left- turn lanes, and number of major-road right-turn lanes for a particular intersection.

Table 24. RSIP Project 37: SPF Coefficients, Target Crash Proportions, and Calibration Factors Used for Louisiana Intersection Data

Facility Type Number of Legs Number of Site-Yearsa Severity Level Intercept (a)b log10AADTMajor Coefficient (b)b log10AADTMinor Coefficient (c)b Overdispersion Parameterb Proportion of Target Collision Type
(PR1)a,c		 Proportion of FI/Total Crashes (PR2)a,d Calibration Factor (Cr)a
Two-Lane Road 3 125 Total -9.86 0.79 0.49 0.54 0.60 1 0.70
125 FI -9.86 0.79 0.49 0.54 0.64 0.33 0.70
4 52 Total -8.56 0.6 0.61 0.24 0.69 1 0.39
52 FI -8.56 0.6 0.61 0.24 0.67 0.55 0.39
Multilane Highway 3 48 Total -12.526 1.204 0.236 0.46 0.47 1 0.64
48 FI -12.664 1.107 0.272 0.569 0.60 1 0.72
4 6 Total -10.008 0.848 0.448 0.494 0.62 1 1.25
6 FI -11.554 0.888 0.525 0.724 0.75 1 2.15

Calculated from RSIP Project 37 data.

b From HSM, Chapters 10 and 11.

c Target crash types include: rear-end, right-angle, and turning crashes only.

d Crash types include all collision types.

Table 25. RSIP Project 37: Combined Intersection CMFs Used for Louisiana Treatment Intersections

Site No. Roadway Type Number of Legs Lighting Skew Angle Number of Major-Road Left-Turn Lanes Number of Minor-Road Right-Turn Lanes Combined CMFa
Total Crashes FI Crashes
1 2-lane 3 No 0 1 0 0.6 0.6
2 Multilane 3 Yes 45 1 0 0.7 0.6
3 Multilane 3 Yes 0 1 0 0.5 0.4
4 Multilane 3 No 0 1 0 0.6 0.5
5 Multilane 3 No 0 1 0 0.6 0.5
6 2-lane 3 Yes 0 0 0 0.9 0.9
7 2-lane 3 Yes 45 0 0 1.1 1.1
8 2-lane 3 Yes 15 0 0 1.0 1.0
9 2-lane 3 No 0 0 0 1.0 1.0
10 2-lane 3 No 0 1 0 0.6 0.6
11 Multilane 3 Yes 0 1 0 0.5 0.4
12 2-lane 4 Yes 15 0 0 1.0 1.0
13 2-lane 3 Yes 60 0 0 1.2 1.2
14 2-lane 3 Yes 0 0 0 0.9 0.9
15 2-lane 4 Yes 0 0 0 0.9 0.9
16 2-lane 3 Yes 0 0 1 0.8 0.8
17 2-lane 3 No 0 0 0 1.0 1.0
18 2-lane 3 No 30 0 1 1.0 1.0
19 2-lane 3 No 30 0 1 1.0 1.0
20 2-lane 4 No 0 0 0 1.0 1.0
21 2-lane 4 Yes 0 0 2 0.7 0.7
22 2-lane 4 Yes 0 0 0 0.9 0.9
23 2-lane 3 No 0 0 1 0.9 0.9
24 2-lane 3 No 0 0 0 1.0 1.0
25 Multilane 3 Yes 0 1 1 0.4 0.3
26 2-lane 3 Yes 30 0 1 0.9 0.9
27 2-lane 4 Yes 0 0 0 0.9 0.9
28 2-lane 3 Yes 0 0 0 0.9 0.9
29 2-lane 4 Yes 0 0 0 0.9 0.9
30 Multilane 4 Yes 0 2 0 0.5 0.4
31 2-lane 3 Yes 45 0 0 1.1 1.1
32 2-lane 4 Yes 0 0 0 0.9 0.9
33 2-lane 3 No 0 0 1 0.9 0.9
34 2-lane 3 Yes 0 0 1 0.8 0.8
35 2-lane 4 Yes 0 0 0 0.9 0.9
36 2-lane 4 Yes 0 0 2 0.7 0.7

aCombined CMF to account for lighting, skew angle, number of major-road left-turn lanes, and number of minor-road right-turn lanes

Table 26. RSIP Project 37: Combined Intersection CMFs Used for Louisiana Nontreatment Intersections

Site No. Roadway Type Number of Legs Lighting Skew Angle Number of Major-Road Left-Turn Lanes Number of Minor-Road Right-Turn Lanes Combined CMFa
Total Crashes FI Crashes
N1 Multilane 3 No 0 1 0 0.6 0.5
N2 2-lane 3 Yes 15 0 0 1.0 1.0
N3 Multilane 3 No 0 1 0 0.6 0.5
N4 2-lane 3 No 30 0 0 1.1 1.1
N5 2-lane 3 Yes 0 0 0 0.9 0.9
N6 2-lane 3 Yes 20 0 1 0.8 0.8
N7 Multilane 3 Yes 0 1 1 0.4 0.3

aCombined CMF to account for lighting, skew angle, number of major-road left-turn lanes, and number of minor-road right-turn lanes

Analysis Results

The EB before/after method was applied to estimate the safety effectiveness of the combination of improvements at rural stop-controlled intersections. The analyses were based on before and after crash data from 35 treatment intersections, crash data from 7 nontreatment intersections, HSM SPFs for three-leg intersections on rural two-lane roads and multilane highways, and HSM SPFs for four-leg intersections on rural two-lane roads. The EB before/after analysis was performed for the following combinations:

Target crashes included rear-end, right-angle, and turning crashes.

The analysis results for three-leg stop-controlled intersections on rural two-lane roads are shown in Table 27; those for four-leg stop-controlled intersections on rural two-lane roads are shown in Table 28; and those for three-leg stop-controlled intersections on multilane highways are shown in Table 29. The statistics shown in each table for each crash severity are:

Table 27. RSIP Project 37: Safety Effectiveness of Combination Intersection Improvements on Total and Target Crashes in Louisiana–Three-Leg Stop-Controlled Intersections on Rural Two-Lane Roads

Collision Type Crash Severity Number of Treatment Intersections Safety Effectiveness (%) Standard Error of Treatment Effect (SE, %) Significance
All Total 19 -67.4 4.3 Significant at 95% CL
FI 19 -56.3 8.9 Significant at 95% CL
Target Total 19 -30.3 10.3 Significant at 95% CL
FI 19 -13.2 20.2 Not significant at 90% CL

Note: Target crash types include: rear-end, right-angle, and turning crashes only.

Table 28. RSIP Project 37: Safety Effectiveness of Combination Intersection Improvements on Total and Target Crashes in Louisiana–Four-Leg Stop-Controlled Intersections on Rural Two-Lane Roads

Collision Type Crash Severity Number of Treatment Intersections Safety Effectiveness (%) Standard Error of Treatment Effect (SE, %) Significance
All Total 10 -52.8 10.3 Significant at 95% CL
FI 10 -63.9 12.2 Significant at 95% CL
Target Total 10 -32.9 17.3 Significant at 90% CL
FI 10 -26.3 25.1 Not significant at 90% CL

Note: Target crash types include: rear-end, right-angle, and turning crashes only.

Table 29. RSIP Project 37: Safety Effectiveness of Combination Intersection Improvements on Total and Target Crashes in Louisiana–Three-Leg Stop-Controlled Intersections on Multilane Highways

Collision Type Crash Severity Number of Treatment Intersections Safety Effectiveness (%) Standard Error of Treatment Effect (SE, %) Significance
All Total 6 -54.9 14.5 Significant at 95% CL
FI 6 -66.1 19.8 Significant at 95% CL
Target Total 6 -52.5 28.0 Significant at 90% CL
FI 6 -71.0 29.3 Significant at 95% CL

Note: Target crash types include: rear-end, right-angle, and turning crashes only.

Interpretation of Results

For three-leg stop-controlled intersections on rural two-lane roads, the analysis results indicate that the combinations of intersection improvements reduced all types of intersection and intersection-related crashes for both total and FI severity levels. The results show a 67-percent reduction in total crashes and a 56-percent reduction in FI crashes, both statistically significant at the 95-percent confidence level. For target crashes, the results show a 30-percent reduction in total crashes, statistically significant at the 95-percent confidence level. The results indicate a 13- percent reduction in target FI crashes, however, this reduction is not statistically significant at the 90-percent confidence level. Looking at all severity levels (i.e., total), the results indicate a higher percent reduction in all crash types combined (67-percent reduction) as compared to target crashes (30-percent reduction).

The safety effectiveness estimates differ between three- and four-leg stop-controlled intersections, but the general trends are the same. The results for FI target crashes were not statistically significant, and the safety effectiveness estimates for all severity levels (total) indicate a higher percent reduction in all crash types combined compared to target crashes.

For three-leg stop-controlled intersections on rural multilane roads, the results indicate that the combinations of intersection improvements reduced all types of intersection and intersection- related crashes and all target crashes for both total and FI severity levels; however, only six treatment sites were available for analysis, so these results should be interpreted with caution.
When comparing the combination of improvements implemented at the treatment sites, the primary treatment that likely has the greatest impact on alerting drivers to the presence of an intersection is the installation of oversized signs, whether they be oversized advance warning signs or oversized stop signs. The other types of treatments evaluated, including the installation
of new route marker signs and new stop bars, probably could be considered secondary treatments to the oversized signs. From the analysis results, the incremental effect of adding oversized advanced warning signs and/or oversized stop signs cannot be distinguished from the incremental effects of adding new route marker signs and stop bars. However, based on the characteristics of the treatment, it is likely that the oversized advanced warning signs and/or oversized stop signs contributed more to the reduction in crashes that was found as compared to the impact of the secondary treatments including new route marker signs and stop bars.

Summary

Detailed quantitative analyses were performed to estimate the safety effectiveness of safety improvements implemented as part of three separate RSIP projects, including RSIP Projects 27, 36, and 37. The analyses focused on estimating safety effectiveness of the following treatment combinations:

Two of the three detailed quantitative analyses (RSIP Project 27 and RSIP Project 37) found a statistically significant reduction in crashes due to installation of the safety improvements.

The results indicate that the dual application of centerline and shoulder rumble strips is effective at reducing total and FI target crashes, which included SVROR, sideswipe-opposite direction, and head-on crashes. The safety effectiveness of the dual application of centerline and shoulder rumble strips on target crashes on rural two-lane roads is estimated as follows:

The results also indicate that improved signing and pavement markings at stop-controlled intersections on rural two-lane roadways and multilane highways are effective at reducing crashes. The combination of signing and pavement markings that were evaluated included installation of oversized advance warning signs, oversized stop signs, new route marker signs, and new stop bars. Based upon the characteristics of the treatments, the oversized advanced warning signs and oversized stop signs are considered the primary treatments, and the new route marker signs and stop bars are considered the secondary treatments. The most reliable estimates of safety effectiveness of the improved signing and pavement markings treatments at stop- controlled intersections for all intersection and intersection related crashes and for target crashes which included rear-end, right-angle, and turning crashes are as follows:

Similar trends were seen in the safety effectiveness of the signing and pavement marking improvements at three-leg stop-controlled intersections on multilane roadways; but the analysis only included data from six treatment sites so the results are not considered as reliable.

The analysis results did not provide reliable information to quantify the safety effects of the delineation and signing treatment combinations installed on horizontal curves on rural two-lane roads. Treatment combinations varied by site and included installation or upgrade of curve warning signs, speed plates, chevrons, and RPMs. The lack of reliable estimates on the safety effectiveness of the delineation and signing treatments does not indicate that the treatment combinations are ineffective at reducing crashes at horizontal curves on rural two-lane roads, but rather that there is insufficient evidence at this time to determine their effectiveness.

 

< Previous Table of Content Next >

 

Return to top

Page last modified on June 12, 2014
Safe Roads for a Safer Future - Investment in roadway safety saves lives
Privacy Policy | Freedom of Information Act (FOIA) | Accessibility | Web Policies & Notices | No Fear Act | Report Waste, Fraud and Abuse
U.S. DOT Home | USA.gov | WhiteHouse.gov

Federal Highway Administration | 1200 New Jersey Avenue, SE | Washington, DC 20590 | 202-366-4000