Evaluation of Safety Strategies at Signalized Intersections (2011)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

13 This chapter provides a summary of several surveys that were conducted of state and local agencies as part of this project. The intent of this task was to obtain information on the installation of treatments by different agencies, determine their priorities for different treatments, and assess the quality of the available data. Web-Based Surveys The web-based surveys were developed using a tool called Zoomerang. The intent of the web-based survey was to deter- mine if a particular agency had installed a particular treatment, the approximate number of installations, and the agency’s assessment of the importance of knowing the CMF of a par- ticular treatment. Based on the research team’s past experience in conducting surveys of this nature, the research team felt that the shorter the survey, more agencies will respond. Hence, to improve the response rate, the list of treatments was divided into two parts (Part 1 and Part 2) based on an assessment of how often specific treatments are installed in the field and the quality of CMFs that are available from previous research. Part 1 requested information for 23 treatments and Part 2 requested information for 36 treatments. A draft version of the web-based survey was submitted to the NCHRP panel in March 2007. Changes were made to the sur- vey after receiving comments from the NCHRP panel. The surveys were launched in June 2007. The surveys were posted online and an e-mail notification was sent to State DOTs, selected local agencies, and listservs including the Traffic Control Device listserv and the State Safety Engineer listserv. Individual requests were also sent to 130 local agencies. Forty-three agencies responded to Part 1 of the survey, and 33 agencies responded to Part 2 of the survey. Assessment of User Priorities and Development of Short List The results of the survey were summarized and disseminated to the project team. In order to select a recommended and prioritized list of strategies, the research team felt that it was necessary to contact select agencies by telephone to get further information regarding their roadway, traffic, crash data, and installation records. In order to be able to do this efficiently, it was necessary to develop a shorter list of treatments for further consideration. As discussed in Chapter 1, the research team considered the likelihood and importance of evaluating each strategy based on the following: • Extent of the coverage in previous/ongoing work. Chapter 2 provides a summary of the CMFs developed in previous research. • Importance to the user (as identified in the survey response). This is assumed to be a good measure of how often a treatment would be implemented if a sound CMF were developed (i.e., high interest would imply higher future implementation). • Ability to identify crash effects. Strategies that may lead to diversion of traffic and/or have a system-wide effect will require more extensive data collection efforts and hence may not be cost effective. Similarly, treatments that may have a small effect on total crashes may require a significantly large sample of sites to conduct an evaluation and hence may not be cost effective. • Data assessment. This was done based on the responses to the web survey and the research team’s knowledge based on working with HSIS and the FHWA Low Cost Pooled Fund Study. The research team also attempted to examine a measure of “crash harm”—the size of the national crash problem poten- tially affected by each treatment. The first step in this analysis was to assign a primary crash type to each treatment that was being studied (e.g., left-turn crashes for left-turn phasing). The next step was to use the General Estimate System (GES) to calculate the number of crashes per year for each primary crash type that was identified earlier. By multiplying the C H A P T E R 3 Survey of Agencies

number of crashes of a particular type with the average cost for that type of crash (based on Council et al., 2005), we can get the crash harm associated with that particular crash type. However, in attempting to do this, we realized that in order for this crash harm analysis to be useful, we would need to include measures of both the expected size of the effect (before the evaluation) and either the proportion of signal- ized intersections or the proportion of signalized-intersection crashes that might be affected by each treatment to be assessed. For example, while left-turn phasing and split phasing can both be targeted to reducing crashes involving left-turning vehicles, they might affect a different proportion of the left-turn crashes, and more importantly, they might be only suitable for use at a different proportion of signalized intersections. Since there is no national inventory of signalized inter- sections, it was not possible to develop the needed estimates. Thus, we assumed that the inputs from users concerning treatment priorities give some indication of the size of the remaining signalized-intersection problem to be solved in their jurisdiction—the “problem size” for them. Based on the project team’s review and assessment of the first three of these four aspects (i.e., the data assessment required follow-up phone interviews with the states), the list of potential treatments was narrowed down to the following: • 17.2 A1: Split phasing • 17.2 A1: Adding protected left-turn phasing* • 17.2 A2: Modifying the change interval* • 17.2 A3: Restricting or eliminating turns at the intersection • 17.2 A7: Remove unwarranted signals • 17.2 B1: Adding left-turn lanes* • 17.2 B1: Lengthening left-turn lanes • 17.2 B2: Improving right-turn channelization • 17.2 B4: Modify intersection skew • 17.2 C2: Improve sight distance • 17.2 D1: Advance Warning Signs for Red Signal • Improvements in signal visibility and conspicuity including – 17.2 D2: Backplates – 17.2 D2: Adding reflective sheeting to backplates – 17.2 D2: Increase signal head size to 12 inches* – 17.2 D2: Installing louvers and visors – 17.2 D2: Installing additional signal heads* – 17.2 D2: Installing far side left-turn signals Note that the treatments designated with asterisks in the previous list have existing CMFs which have been judged to be of at least “medium high” predictive certainty. However, they continued to be included on the potential treatment listing since (1) they are rated highly in the survey, and (2) the research team feels that the existing CMFs could be further improved depending on the availability of data, e.g., to provide variable CMFs for different implementation circumstances. Following is a list of treatments that were rated by the survey respondents among the top 15 in terms of importance, but were not selected in the short list: • Coordinate signals along routes or corridors. The research team felt that this treatment is mainly used to improve traffic flow and reduce delays and not specifically imple- mented to improve safety. For this reason, it was felt that the development of an improved safety-based CMF would not significantly affect use of this treatment, and that a true evaluation would need to trade-off safety findings and delay findings, something beyond the scope of this project. In addition, the evaluation of this treatment is expected to be difficult since (1) finding coordinated systems that had no coordination before might be difficult, and (2) the coordi- nated system might change over time, making it difficult to clearly specify a treatment corridor. • Delineate turn path inside an intersection. The research team felt that the effects of this treatment on safety are small and hence would require a substantial number of sites to statistically detect this expected change in safety, and hence, research would not be cost-effective. • Utilizing crossing guards for school children. The expo- sure to this treatment is limited (i.e., during school open- ings, closings, and during lunch), reducing the number of potential crashes for study. Thankfully, pedestrian crashes are rare events. Hence, hundreds if not thousands of sites would be needed in order to do an effective study. In addi- tion, conducting an evaluation will require data on pedes- trian crossing volumes that the research team found very difficult to find based on previous studies that have been conducted. Hence, the research team felt that allocating re- sources to evaluate this treatment will not be an efficient use of the project budget. Phone Calls to Selected Agencies Based on survey information on available data, the follow- ing agencies were initially contacted for follow-up telephone interviews: • City of Grand Junction, CO; • City of Memphis, TN; • City of Overland Park, KS; • City of San Diego, CA; • City of Scottsdale, AZ; • City of Sparks, NV; • City of Tempe, AZ; • Broward County DOT, FL; • Lee County DOT, FL; • Lexington-Fayette Urban County Government, KY; • Washtenaw County Road Commission, MI; 14

• Arkansas State Highway and Transportation Dept; • Hawaii DOT; • Kansas DOT; • Minnesota DOT; • Missouri DOT; and • South Carolina DOT. These agencies were selected because they indicated they had a sizeable amount of installations of one or more of the treat- ments of interest and they did not indicate any problems with their crash data in their survey response. The project team inter- viewed representatives from each of the agencies that responded to the follow-up contact (indicated in bold in the previous list). A series of questions were asked regarding the installation data, crash data, traffic data, roadway data, and individual treat- ments. Examples of the questions asked included: • Do you keep records of installation for these treatments? If yes, what is the format? • Who maintains the crash data? • What years of crash data are available? • How are the locations of crashes referenced in the data? • In what format are the traffic data maintained? • In what format are the traffic counts presented? Raw data? ADT? AADT? • Are turning movement counts available? • What source of information is available on roadway/ intersection geometry? • Are there any known problems with any of the data? • Was the protected left-turn phasing used when the inter- section was built or was it added after the intersection was already operational? • Are the advanced warning signs dynamic or static? 15