Guidelines for the Removal of Traffic Control Devices in Rural Areas Ryan Tenges Center for Transportation Research and Education Iowa State University 2901 South Loop Drive, Suite 3100 Ames IA, 50010 ryant@iastate.edu Reginald R. Souleyrette Center for Transportation Research and Education Iowa State University 2901 South Loop Drive, Suite 3100 Ames, IA 50010 reg@iastate.edu ABSTRACT The overuse of traffic control devices desensitizes drivers and leads to disrespect, especially on low-volume secondary roads with limited enforcement. The maintenance of traffic signs is also a tort liability issue, exacerbated by unnecessary signs. The Federal Highway Administration s (FHWA) Manual on Uniform Traffic Control Devices (MUTCD) and the Institute of Transportation Engineers Traffic Control Devices Handbook provide guidance for the implementation of stop signs based on expected compliance with right-of-way rules, provision of through traffic flow, context (proximity to other controlled intersections), speed, sight distance, and crash history. The direction to stop is left to engineering judgment and is usually dependent on traffic volume or functional class/continuity of the system. Although currently being considered by the National Committee on Traffic Control Devices, traffic volume itself is not given as a criterion for implementation in the MUTCD. Stop signs have been installed at many locations for various reasons that no longer (or perhaps never) met engineering needs. However, no guidance exists for the removal of stop signs at two-way stop-controlled intersections. The scope of this research is low volume gravel intersections in rural agricultural areas of Iowa, where each of the 99 counties may have as many as 300 or more stop sign pairs. This research investigates the relationship between type of control and independent variables (including traffic volume) to determine the type of warrants that may be used for the removal of stop signs. Overall safety performance is examined as a function of a county overuse factor, developed specifically for this study and based on terrain as a proxy for sight distance and various volume warrant levels. Counties with high levels of rural gravel stop controlled intersection crashes may wish to examine the potential for removing stop control in unnecessary locations to promote compliance with remaining control devices. Key words: rural roadways stop signs stop sign removal unpaved roads Proceedings of the 2005 Mid-Continent Transportation Research Symposium, Ames, Iowa, August 2005. 2005 by Iowa State University. The contents of this paper reflect the views of the author(s), who are responsible for the facts and accuracy of the information presented herein.
INTRODUCTION Many rural Iowa two-way stop signs are installed based on policy, general procedure, or an engineering study of geometric and operational factors. Still others may have been placed in response to citizen complaints or studies that were conducted when traffic volumes may have been greater or sight distances smaller than they are today. Sign removal or a change to less restrictive control (i.e., yield signs) may present an opportunity for improved operations and safety through increased respect for traffic control devices. Problem Statement Unnecessary stop signs are expensive to maintain, are a potential liability, and can reduce respect for necessary signs. The expense of maintenance and generally sparse enforcement in rural areas is particularly challenging to local officials. The effect of the overuse of stop control on traffic control disrespect has not been quantified. Furthermore, there are no criteria for the removal of two-way stop signs or the change to less restrictive control. Current practices and the potential for traffic control reduction at two-way rural gravel-to-gravel road intersections in Iowa are investigated. The safety performance of intersections with and without stop signs is compared. Results are intended to serve as a first step in the development of warrants and procedures for elimination of unnecessary stop control. Objectives This study pursued several objectives framed by the following questions: 1. Adjusted for volumes alone, do stop-controlled rural gravel intersections have lower crash rates than uncontrolled intersections? 2. Can a factor to represent the overuse of stop signs be developed to quantify the effect of stop sign disrespect? 3. As data collection for sight distance for thousands of intersections is cost prohibitive, is it possible to develop a surrogate for the relative number of intersections that would be expected to have sight distance limitations, based on county landcover and topography? 4. Can a combination of volume, overuse, and terrain be used to improve models of the effectiveness of stop control at low-volume intersections? 5. Do certain driver groups (older or younger) have particular problems at rural stopcontrolled or uncontrolled intersections? 6. Are certain types of crashes more or less prevalent at rural stop-controlled and uncontrolled intersections? 7. Can region-specific guidance for the removal/reduction of control at two-way stopcontrolled intersections be developed? BACKGROUND AND LITERATURE REVIEW Installation Warrants Two key references are used for warrants regarding the installation of stop signs: the Federal Highway Administration s (FHWA) Manual on Uniform Traffic Control Devices (MUTCD) and the Institute of Transportation Engineers (ITE) Traffic Control Devices Handbook. Neither of these documents provides warrants for the removal of stop signs. The MUTCD calls for multi- Tenges, Souleyrette 2
way stop signs at intersections with high speeds, sight distance issues, or a crash problem (indicated by five or more reported crashes in a 12-month period that can be corrected by a multiway stop installation), but does not include a daily entering vehicle (DEV) volume warrant. Instead, it has a warrant based upon the average volume for both the major and minor roadway over an eight hour period during the day. This warrant states that a stop sign should be considered at a location where the vehicular volume entering the intersection from the major street approaches (total of both approaches) averages at least 300 vehicles per hour for any 8 hours of an average day, and the combined vehicular, pedestrian, and bicycle volume entering the intersection from the minor street approaches (total of both approaches) averages at least 200 units per hour for the same 8 hours, with an average delay to minor-street vehicular traffic of at least 30 seconds per vehicle during the highest hour. The FHWA conducted a study in 1981 to attempt to establish definitive criteria for the application of two-way stop or yield control at lowvolume intersections. After completing an analysis of variance, the researchers observed a significant increase in crash experience when the volume on the major roadway reached 2,000 vehicles per day. The Uniform Traffic Control Devices Committee is considering proposed language for the third revision of the MUTCD for stop or yield signs used at the intersection of two minor streets, or local roads having more than three approaches, where the combined vehicular, pedestrian, and bicycle volume entering the intersection from all approaches averages more than 2,000 units per day. The proposed language says that stop signs should be considered if engineering judgment indicates that a stop is always required when vehicular traffic volumes on the through street or highway exceed 6,000 vehicles per day. The ITE handbook discourages the use of multi-way stop signs unless the volumes on the major and minor roadways are approximately equal. The ITE handbook also states that the MUTCD warrants should be the main criteria used to determine the locations for stop signs. In several studies, Stokes (2000; 2004) advocates stop sign warrants based on available sight distance and crash history. The studies contend that the majority of crashes associated with rural stop-controlled intersections are not caused by stop sign violations, but rather the driver s inability to judge distance adequately. Stokes concludes that effective solutions to the failure to yield problem should focus on the intersection as a whole rather than simply improving the sight distance or other characteristics of one leg of the intersection. He suggests the use of speed zones and advanced warning signs in places where drivers on the side road may have difficulty judging the speeds of approaching vehicles. Table 1 presents Stokes recommendation for control based on available sight distance, crash history (three years worth of data), and the major roadway volume. A note at the bottom of the chart indicates that the values should be used in conjunction with MUTCD criteria when assessing the need for stop or yield control. It should be noted that while many intersections are stop controlled, volumes at Iowa rural gravel intersections are typically an order of magnitude lower than the values recommended by Stokes. Tenges, Souleyrette 3
Table 1: Stokes method for determining control type Geometric Studies Several studies report on the impact of geometry on safety of stop-controlled intersections. Stockton (1981) concluded that geometry plays no significant role in safety or operations for choosing the type of control to be used. The report recommended that major roadway volume should be the major factor in the determination of control type and that sight distance has no significant impact on the number of crashes that occur at an intersection. Mounce (1981) completed a study in which he agrees with Stockton, indicating that the choice of control should be strictly based only on the volume of the major roadway, as long as the available sight distance is greater than the minimum required. The study recommended that no control be used on an intersection with a major roadway volume between 0 and 2,000 vehicles, yield control be used with a major roadway volume between 2,000 and 5,000 vehicles, and stop control be used when the major roadway volume is greater than 5,000 vehicles. (The intersection with the highest DEV being investigated in this study is about 600 vehicles per day.) Although Stockton and Mounce do not advocate use of sight distance to determine control type, it should be noted that there are at least two sight distance methods commonly used in the traffic engineering industry. One was included in the Traffic Control Devices Handbook (TCDH) published by the Federal Highway Administration in 1983. However, this method was not included in the handbook when it was updated in 2001. The second can be found in A Policy on Geometric Design of Highways and Streets 2001, Fourth Edition, published by the American Association of State Highway and Transportation Officials (AASHTO). Using the TCDH method, the decision as to whether to use a STOP sign or a YIELD sign is primarily based upon sight distance at the intersection. The method uses a critical approach speed, which is the lowest speed that a motorist would be able to travel and fail to avoid a collision with an approaching vehicle on the cross street. Tenges, Souleyrette 4
The distances to sight obstructions at the intersection under study are measured, parking is considered, and a nomograph is used to determine the critical approach speeds. An intersection with a critical approach speed of less than 10 mph is controlled by a stop sign. An intersection where the critical approach speed is between 10 and 15 mph may be controlled with a yield sign. If the critical approach speed on all approaches is above 15 mph, then the intersection can be left uncontrolled. The TCDH assumes that motorists approaching uncontrolled intersections reduce their speed to account for prevailing conditions at the intersection. Consequently, the use of this method can result in a number of intersections remaining uncontrolled. The method can be useful in urban areas. When Mounce refers to sight distance, he is using the TCDH method. The AASHTO method assumes that motorists approaching uncontrolled intersections reduce their speed by 50% based on observations. The method uses this assumption to determine the length of sight distance necessary along each leg of the intersecting roadway, resulting in a desirable sight distance triangle. If this sight distance triangle exists, then no control is needed. If it does not exist, then control is needed. This method can also be adjusted for grades. The AASHTO method tends to be more suited for rural areas. Use of this method almost always results in control being installed. METHODOLOGY A survey was sent to all 99 Iowa county engineers. Respondents were asked the following: Number of uncontrolled intersections, all-way stops, and yield-controlled intersections that used by the county Criteria used for installation of stop signs, including references used to determine warrants (MUTCD, ITE, etc.) Whether an engineering study was completed prior to installation Whether any formal or general policy exists for stop sign usage (e.g., placing a stop sign at all intersections, only at intersections with sight distance issues, only at intersections with paved roadways, etc.) A stop sign database has also been created for 19 counties. A crash analysis was conducted using ten years of crash data (1994 2003). The average number of crashes, crash rate, average severity loss, and cost per intersection was computed and summarized for each county and stratified by the number of intersection legs and type of control. An example county summary is shown in Table 2. All data were pooled to create a statewide summary table, shown in Table 3. Tenges, Souleyrette 5
Table 2: Crash analysis in Story County Tenges, Souleyrette 6
Table 3: Statewide crash analysis Initial Analysis Statewide stop-controlled intersections have a slightly lower average crash rate per million entering vehicles, as well as a lower average cost per crash. On the surface, it would appear that savings could be realized by switching all of the uncontrolled intersections to stop-controlled intersections. Crashes saved by changing to stop control = (0.05*68*365*10*3829)/1000000 (1) = 47.5 crashes To calculate the expected savings, the cost of the crashes after the conversion (256 crashes x average cost per crash for stop-controlled intersections, $50,000) is subtracted from the cost of the crashes before the conversion (304 crashes x average cost per crash for uncontrolled Tenges, Souleyrette 7
intersections, $70,000). The savings would be approximately $8.5 million, not including the cost of installing and maintaining stop signs, which may be on the order of $100/sign x 2 x 3829 intersections = $770,000. The relationship between crashes and volume that one would expect to observe can be seen in Figure 1. Each point represents an intersection where the uncontrolled intersections are expected to have a higher number of crashes per average volume than stopcontrolled intersections. Figure 1. Relationship between crashes and volume Regression Logistic regression was used to determine the probability of one or more crashes occurring during 10 years based on type of control and DEV. Figure 2 shows the regression equation and resulting probabilities. Type of control is treated as a dummy variable (-1 for stop controlled, +1 for no control). Below 100 DEV, uncontrolled intersections have a lower probability of crashes than stop-controlled intersections. Sight distance is not included as a variable in this analysis. Tenges, Souleyrette 8
Figure 2: Probability of a crash occurring in 10 years based on type of control and DEV Overuse Factor The analysis presented above makes the simple assumption that conversion of all intersections to stop control would result in a reduction of crashes. The procedure does not take into account the potential for increasing disrespect for signs or any underlying factors that may make the previously uncontrolled intersections different from their controlled counterparts. Therefore, an overuse factor was developed that is a function of how many stop signs a county has that are unwarranted (expressed as a percentage). To determine the number of unwarranted intersections per county (on a volume basis alone), a volume warrant was chosen, as indicated in Table 4, and the corresponding percentage of intersections exceeding that warrant were used in the following equation to determine the percentage of unwarranted stop-controlled intersections for that county: Overuse factor (based solely on volume) = (Total # of stop int warranted # number of stop int) / total # of stop int (2) The chart below shows a sensitivity analysis for an overuse factor based on volume alone. Volume warrants ranged from 50 to 200 vehicles per day. After the number of intersections meeting each volume warrant was determined, an overuse factor for each county was calculated. Tenges, Souleyrette 9
Table 4. Table of overuse factors based on volume alone Tenges, Souleyrette 10 Volume Warrant Values County 50 100 150 200 Total # of # of stop intersections controlled intersections % warranted on volume % overuse on volume alone alone 50 100 150 200 50 100 150 200 Number of intersections meeting warrant Adams 149 17 4 2 361 252 41% 5% 1% 1% 41% 93% 98% 99% Boone 282 60 18 6 379 94 74% 16% 5% 2% -200% 36% 81% 94% Bremer 224 77 39 14 268 113 84% 29% 15% 5% -98% 32% 65% 88% Calhoun 237 50 16 5 379 140 63% 13% 4% 1% -69% 64% 89% 96% Carroll 343 144 46 21 401 245 86% 36% 11% 5% -40% 41% 81% 91% Cedar 313 142 58 29 396 167 79% 36% 15% 7% -87% 15% 65% 83% Cerro Gordo 282 89 23 8 359 208 79% 25% 6% 2% -36% 57% 89% 96% Cherokee 183 33 9 6 350 93 52% 9% 3% 2% -97% 65% 90% 94% Clay 198 40 10 3 332 49 60% 12% 3% 1% -304% 18% 80% 94% Emmet 90 10 4 3 183 92 49% 5% 2% 2% 2% 89% 96% 97% Henry 263 130 51 23 346 181 76% 38% 15% 7% -45% 28% 72% 87% Madison 347 166 101 55 496 143 70% 33% 20% 11% -143% -16% 29% 62% Montgomery 183 54 8 1 349 208 52% 15% 2% 0% 12% 74% 96% 100% Osceola 121 19 4 0 219 98 55% 9% 2% 0% -23% 81% 96% 100% Pocahontas 201 26 3 0 325 66 62% 8% 1% 0% -205% 61% 95% 100% Sac 269 54 16 5 379 88 71% 14% 4% 1% -206% 39% 82% 94% Story 253 98 33 15 328 12 77% 30% 10% 5% -2008% - - -25% 717% 175% Washington 337 151 61 24 483 398 70% 31% 13% 5% 15% 62% 85% 94% Woodbury 291 125 81 57 489 368 60% 26% 17% 12% 21% 66% 78% 85%
Figure 3 shows the expected relationship between counties with higher and lower overuse factors. Figure 3. Expected relationship between crashes and volume (not accounting for sight distance) While an overuse factor may be used to estimate the number of unwarranted stop-controlled intersections based on volume, it does not account for the use of stop signs warranted based on sight distance issues. Terrain Factor To attempt to account for the intersections with sight distance issues, a terrain factor has been developed. The overuse factor (based on volume alone) can be adjusted by this terrain factor to more appropriately represent the effect of stop sign overuse. This factor was qualitatively created based on topography and land cover on a county by county basis. The terrain factor was created using a United States Geographical Services (USGS) land cover and shaded relief map for the state of Iowa. Each county was given a value of one to three (one = low, flat land, agriculture cover; three = high, river land, forest cover) based on topography and land cover. See Figures 4 through 6 below. Figure 4. USGS land cover map Tenges, Souleyrette 11
Figure 5. USGS shaded relief map Figure 6. Maps showing values for each county based on topography and land Warranted Intersections Based on Volume and Terrain Volume and terrain are incorporated in the following equation to determine the fraction of warranted stop-controlled intersections for each county: Tenges, Souleyrette 12
Fraction of warranted stop int = 1 (1 volume fraction) x (1 terrain fraction) (3) An adjusted overuse factor was calculated as follows: Overuse factor (based on volume and sight distance) = (Total # of stop int warranted # number of stop int) / total # of stop int (4) The development of the overuse factor based on both volume and terrain is seen below in Tables 5 through 7. Volume warrants again ranged from 50 to 200 vehicles per day, and these were combined with three sets of terrain factors to comprise 12 inputs for regression and sensitivity analysis. Tenges, Souleyrette 13
Tenges, Souleyrette 14 Table 5. Development of the overuse factor (1) Terrain values County Landcover Topography Warranted stops based on both volume and terrain Overuse factors based on volume and terrain Terrain factor 50 100 150 200 50 100 150 200 Adams 0.7 0.7 0.49 253 186 179 178 0.00 0.26 0.29 0.29 Boone 0.5 0.5 0.25 306 140 108 99-2.26-0.49-0.15-0.06 Bremer 0.5 0.5 0.25 235 125 96 78-1.08-0.10 0.15 0.31 Calhoun 0.5 0.5 0.25 273 132 107 99-0.95 0.06 0.24 0.30 Carroll 0.5 0.5 0.25 358 208 135 116-0.46 0.15 0.45 0.53 Cedar 0.7 0.7 0.49 354 266 224 209-1.12-0.60-0.34-0.25 Cerro Gordo 0.5 0.5 0.25 301 157 107 96-0.45 0.25 0.49 0.54 Cherokee 0.5 0.5 0.25 225 112 94 92-1.42-0.21-0.01 0.01 Clay 0.5 0.5 0.25 232 113 91 85-3.72-1.31-0.85-0.74 Emmet 0.5 0.5 0.25 113 53 49 48-0.23 0.42 0.47 0.48 Henry 0.7 0.7 0.49 304 236 196 181-0.68-0.30-0.08 0.00 Madison 0.9 0.7 0.63 441 374 350 333-2.08-1.61-1.45-1.33 Montgomery 0.5 0.7 0.35 241 157 127 123-0.16 0.24 0.39 0.41 Osceola 0.5 0.5 0.25 146 69 58 55-0.48 0.30 0.41 0.44 Pocahontas 0.5 0.5 0.25 232 101 84 81-2.52-0.53-0.27-0.23 Sac 0.5 0.5 0.25 297 135 107 99-2.37-0.54-0.21-0.12 Story 0.5 0.5 0.25 272 156 107 93-21.65-11.96-7.90-6.77 Washington 0.7 0.7 0.49 409 314 268 249-0.03 0.21 0.33 0.37 Woodbury 0.5 0.7 0.35 360 252 224 208 0.02 0.31 0.39 0.43
Table 6. Development of the overuse factor (2) Tenges, Souleyrette 15 Terrain values Warranted stops based on both volume and terrain Overuse factors based on volume and terrain County Landcover Topography Terrain factor 50 100 150 200 50 100 150 200 Adams 0.6 0.6 0.36 225 141 133 131 0.11 0.44 0.47 0.48 Boone 0.4 0.4 0.16 298 111 76 66-2.17-0.18 0.19 0.30 Bremer 0.4 0.4 0.16 231 108 76 55-1.04 0.05 0.33 0.52 Calhoun 0.4 0.4 0.16 260 103 74 65-0.86 0.27 0.47 0.54 Carroll 0.4 0.4 0.16 352 185 103 82-0.44 0.24 0.58 0.67 Cedar 0.6 0.6 0.36 343 233 180 161-1.05-0.40-0.08 0.04 Cerro Gordo 0.4 0.4 0.16 294 132 77 64-0.42 0.36 0.63 0.69 Cherokee 0.4 0.4 0.16 210 84 64 61-1.26 0.10 0.32 0.34 Clay 0.4 0.4 0.16 219 87 62 56-3.48-0.77-0.26-0.14 Emmet 0.4 0.4 0.16 105 38 33 32-0.14 0.59 0.65 0.65 Henry 0.6 0.6 0.36 293 208 157 139-0.62-0.15 0.13 0.23 Madison 0.8 0.6 0.48 419 324 291 267-1.93-1.27-1.03-0.86 Montgomery 0.4 0.6 0.24 223 125 90 85-0.07 0.40 0.57 0.59 Osceola 0.4 0.4 0.16 137 51 38 35-0.39 0.48 0.61 0.64 Pocahontas 0.4 0.4 0.16 221 74 55 52-2.35-0.12 0.17 0.21 Sac 0.4 0.4 0.16 287 106 74 65-2.26-0.20 0.16 0.26 Story 0.4 0.4 0.16 265 135 80 65-21.08-10.23-5.68-4.42 Washington 0.6 0.6 0.36 390 271 213 189 0.02 0.32 0.47 0.52 Woodbury 0.4 0.6 0.24 339 212 179 161 0.08 0.42 0.51 0.56
Table 7. Development of the overuse factor (3) Tenges, Souleyrette 16 Terrain values Warranted stops based on both volume and terrain Overuse factors based on volume and terrain County Landcover Topography Terrain factor 50 100 150 200 50 100 150 200 Adams 0.6 0.6 0.36 225 141 133 131 0.11 0.44 0.47 0.48 Boone 0.3 0.3 0.09 291 89 50 40-2.09 0.06 0.46 0.58 Bremer 0.3 0.3 0.09 228 94 60 37-1.02 0.17 0.47 0.67 Calhoun 0.3 0.3 0.09 250 80 49 39-0.78 0.43 0.65 0.72 Carroll 0.3 0.3 0.09 348 167 78 55-0.42 0.32 0.68 0.77 Cedar 0.6 0.6 0.36 343 233 180 161-1.05-0.40-0.08 0.04 Cerro Gordo 0.3 0.3 0.09 289 113 53 40-0.39 0.46 0.74 0.81 Cherokee 0.3 0.3 0.09 198 62 40 37-1.13 0.34 0.57 0.60 Clay 0.3 0.3 0.09 210 66 39 33-3.29-0.35 0.20 0.33 Emmet 0.3 0.3 0.09 98 26 20 19-0.07 0.72 0.78 0.79 Henry 0.6 0.6 0.36 293 208 157 139-0.62-0.15 0.13 0.23 Madison 0.9 0.6 0.54 427 344 314 293-1.99-1.41-1.20-1.05 Montgomery 0.3 0.6 0.18 213 107 69 64-0.02 0.49 0.67 0.69 Osceola 0.3 0.3 0.09 130 37 23 20-0.32 0.62 0.76 0.80 Pocahontas 0.3 0.3 0.09 212 53 32 29-2.21 0.20 0.52 0.56 Sac 0.3 0.3 0.09 279 83 49 39-2.17 0.05 0.45 0.56 Story 0.3 0.3 0.09 260 119 60 43-20.65-8.89-3.96-2.60 Washington 0.6 0.6 0.36 390 271 213 189 0.02 0.32 0.47 0.52 Woodbury 0.3 0.6 0.18 327 191 154 135 0.11 0.48 0.58 0.63
Equation 3 takes both volume and a proxy for sight distance warrants into account and is expected to result in a relationship such as the one depicted in Figure 7. Each point represents one county and the counties with higher adjusted overuse factors are expected to have a higher average crash rate. Figure 7. Expected effect of stop sign overuse Age Group Analysis An age group analysis was completed to determine whether older or younger drivers have more problems at rural stop-controlled and uncontrolled intersections. Tables 8 and 9 show the involvement of younger (less than or equal to 19 years) and older (greater than or equal to 65 years) drivers in all crashes and in all multi-vehicle crashes in Iowa during 2003. All multi-vehicle crashes in the study are rural gravel stopcontrolled and uncontrolled intersections between 1994 and 2003. Table 8 shows that the representations of older and younger drivers are approximately equal statewide, but at stop-controlled gravel intersections, there is a significant overrepresentation of younger drivers (about 35% of crashes and 20% of drivers) as opposed to older drivers (about 14% of crashes and 7% of drivers). These differences could be explained if inexperienced drivers have more difficulty judging distances than older drivers. Table 9 shows that there is a significant overrepresentation of younger drivers at uncontrolled gravel intersections (about 30% of crashes and 18% of drivers) as opposed to older drivers (about 18% of crashes and 9% of drivers). Overall, younger drivers are also overrepresented at study area gravel intersections, (about 33% of crashes and 19% of drivers) as opposed to older drivers (about 16% of crashes and 8% of drivers). Tenges, Souleyrette 17
Table 7. Age group analysis, stop-controlled intersections Tenges, Souleyrette 18
Table 8. Age group analysis, uncontrolled intersections Crash Type Analysis A crash type analysis was completed based the Iowa crash database and validated using crash narratives available from the crash reports. A summary table of this information can be seen in Table 10. As expected, the most prominent contributing circumstance for uncontrolled intersections is failure to yield right of way. The contributing circumstances from the crash data as well as the reports are very similar, but 28 crashes did not have a contributing circumstance from at least one of the drivers involved in the crash. This may mean that the driver did not provide that information on the report used to create the crash data or that the reporting officer did not record any contributing circumstance. The most prominent crash type for uncontrolled intersections is a broadside/right-angle crash. Tenges, Souleyrette 19
Failure to yield right of way was the most prominent contributing circumstance at stop-controlled intersections as well. Another common contributing circumstance for stop-controlled intersections was failing to stop at the stop sign. This suggests a level of disrespect for stop signs in rural areas. Table 9. Crash type analysis CONCLUSIONS Use of Overuse Factors in Regression The ranges of overuse factors were incorporated into the regression analysis in an attempt to quantify the effect of traffic control device overuse and disrespect and improve the regression results. However, the use of these factors did not improve the models, indicating either no effect of county-wide overuse of stop signs, or, as is more likely, an error in the construction of the overuse factors themselves, as no sitespecific sight distance information was available (i.e., the proxies did not work). Removal Techniques Although not currently supported by the models developed in this research, removal of unneeded stop signs may be considered by operating authorities. Should this be the case, procedures need to be developed. ITE suggests procedures for the removal of traffic signals that may also be useful for removal/conversion of stop signs. ITE recommends the distribution of newsletters prior to the activation/removal to inform the public of the future change. ITE also recommends that signal ahead Tenges, Souleyrette 20
signs be installed/removed to warn drivers just before conversion actually takes place. The final ITE recommendation involves the distribution of news releases to all local newspapers, radio, and television stations of the impending change to ensure that all potential users of the intersection are aware of the conversion. The FHWA has published several techniques for the removal of multi-way stops with minimum hazard. These techniques occur in three phases: pre-conversion phase, conversion phase, and the post-conversion phase. These techniques are as follows: 1. Pre-conversion phase Conduct traffic engineering studies Publicize impending conversion (radio, newspapers, and bills) Post notice signs beneath stop signs 30 days prior to conversion 2. Conversion phase Remove unwarranted signs Remove notice signs Install caution signs beneath stop signs Remove/install stop ahead signs as appropriate Improve sight distance 3. Post-conversion phase Monitor intersection Police enforcement Remove caution signs 90 days after conversion Urban Extension The overuse of stop control is also a problem in urban areas. To attempt to quantify the differences between the uses of a stop-controlled intersection and an uncontrolled intersection, a similar study will be completed on two cities: Ames and West Des Moines, Iowa. A video log of all the intersections in the City of Ames has been created as part of this project. Along with the help of a data collection program, this video log was used to create a traffic control database for the City of Ames. (A database for the City of West Des Moines had already been developed previously.) These databases will make it possible to compare the safety at controlled intersections to that of uncontrolled intersections. An analysis similar to that of the rural analysis will be completed. Tenges, Souleyrette 21
REFERENCES Celniker, S. 1989. All-Way Stops: A New Policy. Transportation Research Record 1244. Washington, DC: TRB, National Research Council, pp. 17 22. Cottrell, B. 1997. Using All-Way Stop Control for Residential Traffic Management. Transportation Research Record 1605. Washington, DC: TRB, National Research Council, pp. 22 27. Do traffic signals ever become too unwarranted? 1981. American City and County 96/11.81, pp. 25 29. Eck, R. and J. Biega. 1988. Field Evaluation of Two-Way Versus Four-Way Stop Sign Control at Low- Volume Intersections in Residential Areas. Transportation Research Record 1160. Washington, DC: TRB, National Research Council, pp. 7 13. Fitzpatrick, K., K. Balke, D.W. Harwood, and I.B. Anderson. 2000. Accident Mitigation Guide for Congested Rural Two-Lane Highways. NCHRP Report 440. Washington, DC: Transportation Research Board. Institute of Transportation Engineers. 2001. Traffic Control Devices Handbook. Washington DC: Institute of Transportation Engineers. Institute of Transportation Engineers Public Agency Council Committee PAC-101-03. 2003. Guidelines for the Activation, Modification, or Removal of Traffic Signal Control Systems, An ITE Proposed Recommended Practice. Washington, DC: Institute of Transportation Engineers. JHK and Associates and Wagner-McGee Association. 1980. User Guide for Removal of Not Needed Traffic Signals. Implementation Package FHWA-IP-80-12. Federal Highway Administration, U.S. Department of Transportation. Ligon, Claude M., Carter, E.C., H.W. McGee. 1984. Removal of Multiway Stop Signs with Minimum Hazard, Volume 1: Technical Report. FHWA/RD-84/084. Federal Highway Administration, U.S. Department of Transportation. Ligon, Claude M., Carter, E.C., H.W. McGee. 1984. Removal of Multiway Stop Signs with Minimum Hazard, Volume 2: Recommended Procedures. FHWA/RD-84/085. Federal Highway Administration, U.S. Department of Transportation. Lovell, J. and E. Hauer. 1986. The Safety Effect of Conversion to All-Way Stop Control. Transportation Research Record 1068. Washington, DC: TRB, National Research Council, pp. 103 107. Mounce, J.M. Driver Compliance with Stop Sign Control at Low-Volume Intersections. 1981. Transportation Research Record 808. Washington, DC: Transportation Research Board. Neudorff, L.G. Relationship Between the Removal of Traffic Signals and Intersection Accidents. Transportation Engineering March 1978, pp. 16 20. Savage, W.F. 2001. Traffic Calming Through Traffic Signal Removal-City of Leslie. Washington, DC: Institute of Transportation Engineers. Smith, D. 2002. Handbook of Simplified Practice for Traffic Studies. Iowa DOT Project TR-455, CTRE Project 01-80. Ames, IA: Center for Transportation Research and Education, Iowa State University. Stockton, W. R., R.Q. Brackett, and J.M. Mounce. 1981. Stop, Yield, and No Control at Intersections. FHWA/RD-81/084. Federal Highway Administration, U.S. Department of Transportation. Stokes, R. 2000. Analysis of Rural Intersection Accidents Caused by Stop Sign Violation and Failure to Yield the Right-of-Way. Report No. K-TRAN: KSU-98-6. Manhattan, KS: Kansas State University. Stokes, Robert. 2004. Effectiveness of Two-way Stop Control at Low-Volume Rural Intersections. Report No. K-TRAN: KSU-99-5. Manhattan, KS: Kansas State University. U. S. Department of Transportation Federal Highway Administration. 2001. Manual on Uniform Traffic Control Devices. Washington, DC: Federal Highway Administration, U. S. Department of Transportation. Tenges, Souleyrette 22