Accident Modification Factors for Traffic Engineering and ITS Improvements (2008)

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.

30 This chapter includes a discussion of the second method- ology used in this study to develop AMFs—analysis-driven expert panels. The chapter includes a listing of the partici- pants in each of the two panels convened, a description of the overall process followed for the two panels with respect to identifying and prioritizing the treatments to be explored, and the results of the panel discussions. Introduction As noted in Chapter 2, one approach to AMF devel- opment/modification involved two analysis-driven expert panels. While other alternative expert panels were discussed during the project planning effort, it was decided by the team and the oversight panel that a more critical need was to assist the research teams from NCHRP Projects 17-26 and 17-29 in developing AMFs needed for the safety prediction tools they were developing for urban/suburban arterials and rural mul- tilane highways. Because of the large number of serious and fatal crashes that occur on two-lane rural roads, much of the past AMF development work has focused on treatments for that roadway class. Literature on AMFs for other roadway classes is limited. It was hoped that expert panels might be able to combine the limited past evaluations specific to these two roadway classes (urban/suburban arterials and rural multilane highways) with modified versions of rural two-lane AMFs to develop the needed estimates. It was also hoped that this option might produce multiple AMFs at a lower cost than the cost of new analyses. Members of the Panels A critical requirement was that the expert panels be analysis- driven. The AMFs derived by the panels were to be based on critical reviews of the existing research literature and on a con- sensus decision that the results from the research literature were robust enough to allow development of an AMF with at least a medium-high level of predictive certainty. Given this orienta- tion, expert panel membership needed to include expert researchers knowledgeable about the AMFs of interest and the strengths and weaknesses of study methods and a group of expert state and local AMF users (i.e., safety engineers) with knowledge of both the specifics of the AMFs needed and of the real-world conditions under which evaluated treatments were probably implemented. The decisions on expert panel mem- bership were made jointly by the Principal Investigators of the three involved NCHRP projects (NCHRP Projects 17-25, 17-26, and 17-29). Members of the expert panels for each proj- ect are shown in Figure 1. Procedures Followed The same procedures were followed for both expert pan- els. Both expert panels met for a 3-day period. Because the number of potential AMFs was greater than could be stud- ied and discussed in this time frame, the AMFs were priori- tized before each meeting. For each meeting, The NCHRP Project 17-25 research team and the Principal Investigator for each companion NCHRP project (17-26 and 17-29) de- veloped two lists of candidate treatments to consider for AMF development—one for roadway segments and one for intersections. These lists were based on the results of the ear- lier-described AMF knowledge matrix and on the specific AMF needs of the companion NCHRP projects. The two lists were sent to each member of the expert panel before the meeting. The expert panel members reviewed each list sepa- rately (intersection and segment treatments) and ranked each variable with respect to level of importance as either primary (P) or secondary (S). Primary variables were those believed to be the most important predictors of safety on the road type in question and in definite need of discussion at the meeting. Secondary variables were those believed to be of less importance with respect to predicting safety, which should only be considered for discussion at the meeting if C H A P T E R 4 Development of New AMFs through Expert Panels

31 time permitted. The primary variables were also ranked from most important (first) to least important (last). These rank- ings were based on each member’s assessment of (1) the per- ceived magnitude of the effect of the variable on safety and (2) the quality and extent of reliable information in the lit- erature on which an AMF could be based. The expert panel inputs were then compiled by the NCHRP Project 17-25 team to develop a final list for discussion. The research team then developed and distributed to each expert panel member a resource notebook. This notebook included the results of the prioritization task, contact infor- mation for all expert panel members, resource materials for each variable/treatment, and pre-meeting assignments for the expert panel members. The resource materials included the following for each AMF under consideration: • The AMF summary material developed in NCHRP Proj- ect 17-25 and described in Chapter 2 of this report. This material included the draft research results digest for those AMFs considered to have high or medium-high levels of predictive certainty and summary pages from the interim report for those of lower quality. • The AMF summary from NCHRP Project 17-27 (Parts I and II of the Highway Safety Manual). This draft summary was completed in 2005 and included assessments of AMFs; a discussion of the studies from which the AMFs were taken or derived; and a discussion of materials reviewed without the recommendation of an AMF or a listing of possible resources that could be reviewed for future AMF develop- ment consideration. • Copies of five cross-sectional studies. These studies in- cluded a number of the high-priority elements. A descrip- tion of the models developed and the variables included were provided for each study. The models could possibly provide additional insight into the direction and magni- tude of the effect of the variables found to be significant. • A draft procedure for adjusting AMF estimates and stan- dard errors. This procedure was developed and applied under NCHRP Project 17-27. In order to ensure that all treatments or variables were ad- equately addressed, the expert panel members were given pre-meeting assignments. The expert panel was divided into three groups, and each group was assigned a subset of the variables to review prior to the meeting and asked to help lead the discussion on those topics. The expert panel was asked to focus on three questions: (1) Do the materials presented include enough quantitative information to potentially develop an AMF for urban and suburban arterials (or rural multilane highways)? The materials provided to the panel included a wide spec- trum of study types (e.g., rigorous before-after studies, simple before-after studies, cross-sectional studies, or less UNC = University of North Carolina. VHB = Vanasse Hangen Brustlin, Inc. Suburban and Urban Arterials Doug Harwood, Midwest Research Institute (Principal Investigator, NCHRP 17-26) David Harkey, UNC Highway Safety Research Center (Principal Investigator, NCHRP 17-25) Dr. James A. Bonneson, Texas Transportation Institute Dr. Forrest Council, VHB Kim Eccles, VHB Dr. Ezra Hauer, University of Toronto (Retired) Dr. Bhagwant Persaud, Ryerson University Stan Polanis, City Traffic Engineer, City of Winston Salem, NC Dr. Raghavan Srinivasan, UNC Highway Safety Research Center Tom Welch, State Transportation Safety Engineer, Iowa DOT Rural Multilane Arterials Dr. Dominique Lord, Texas A&M University (Principal Investigator, NCHRP 17-29) David Harkey, UNC Highway Safety Research Center (Principal Investigator, NCHRP 17-25) Dr. James A. Bonneson, Texas Transportation Institute Dr. Forrest Council, VHB Kim Eccles, VHB Dr. Ezra Hauer, University of Toronto (Retired) Loren Hill, State Highway Safety Engineer, Minnesota DOT Brian Mayhew, North Carolina DOT Dr. Bhagwant Persaud, Ryerson University (representing NCHRP 17-29) Dr. Raghavan Srinivasan, UNC Highway Safety Research Center Dr. Simon Washington, Arizona State University Tom Welch, State Transportation Safety Engineer, Iowa DOT Figure 1. Members of the analysis-driven expert panels.

32 Treatment Review PanelA AMFDevelopedB Comments C Intersection Treatments Intersection Skew Angle U/S Arterials No Left-Turn Lanes U/S ArterialsRML Highways Yes No Right-Turn Lanes U/S ArterialsRML Highways Yes No Intersection Medians U/S Arterials No Signal Installation RML Highways No Developed by separate 17-25 analysis Left-Turn Phasing U/S Arterials RML Highways Yes No Combined findings from two studies; further developed by separate 17-25 analysis Signal Change Interval U/S Arterials No Signal All-Red Interval U/S Arterials No Signal Cycle Length U/S Arterials No Right-Turn-On-Red (effects on vehicle crashes) U/S Arterials Yes Further developed by additional 17-26 analysis Right-Turn-On-Red (effects on pedestrian crashes) U/S Arterials No Possible development in 17-26 Driveways Near Intersection U/S Arterials No Intersection Sight Distance U/S Arterials No Curb Parking Near Intersection U/S Arterials No Intersection Lighting U/S ArterialsRML Highways Yes Yes Further developed by additional 17-25 and 17-26 analysis Red-Light Cameras U/S Arterials Not needed Available from prior study (26) Left-Turn Channelization U/S ArterialsRML Highways Not needed No Right-Turn Channelization U/S ArterialsRML Highways No No Approach Speeds U/S Arterials No Roundabouts U/S Arterials Not needed Available from prior study (8) Roadway Segment Treatments Number of Lanes U/S Arterials No Lane Width U/S Arterials No Shoulder Type U/S ArterialsRML Highways No Yes Shoulder Width U/S ArterialsRML Highways No No Add a Median U/S Arterials No Median Width U/S ArterialsRML Highways No No Developed by separate 17-25 analysis Two-Way Left-Turn Lane (TWLTL) U/S Arterials Yes Horizontal Curvature U/S Arterials No Possible development in 17-29 RML Highways No Sideslope RML Highways Yes Clearzone Width RML Highways No Roadside Fixed Objects U/S ArterialsRML Highways No No Developed by additional 17-26 analysis Driveways/Access Points U/S ArterialsRML Highways No No 17-26 analysis produced separate models for driveway accidents Speed Limits/Zoning U/S ArterialsRML Highways No No Developed by separate 17-25 analysis On-Street Parking U/S Arterials Yes Segment Lighting U/S Arterials Yes Further developed by additional 17-25 and 17-26 analysis Rumble Strips U/S ArterialsRML Highways Not Needed No Miscellaneous Treatments Pedestrian/Bicycle Treatments U/S Arterials No Some being developed in 17-26 Far-Side vs. Near-Side Bus Stops U/S Arterials No A Indicates which panel considered the treatment to be a priority, reviewed the relevant research, and discussed the potential for AMF development; U/S Arterials = urban/suburban arterials and RML Highways = rural multilane highways. B “Yes” indicates that consensus was reached at the expert panel meeting on an acceptable AMF from the critical review and discussion of prior research. C Numbers in “Comments” refer to NCHRP projects. Note that the 17-26 and 17-29 teams were involved in their respective panels. Table 17. Treatments reviewed and AMFs developed by expert panels.

vigorous data assessments). In many cases, the materials and existing AMFs were related to rural two-lane roads. Expert panel members were asked to assess whether or not the material was sufficient for the specification of an AMF for an urban/suburban arterial or rural multilane highway. (2) If an AMF can be developed from the material pro- vided, what is the magnitude of the effect and to what types and severities of crashes does it apply? (3) Are there other studies that are not included in the existing set of materials that should be discussed at the meeting? The project team completed all logistics for the meeting, the expert panel members prepared for the meeting, and the meeting was held at the University of North Carolina (UNC) Highway Safety Research Center facilities. The project team recorded detailed notes of the ensuing discussions and con- tinually displayed both the notes and possible findings. Final decisions were then made by expert panel consensus. Results The suburban/urban arterial expert panel reviewed research materials on 14 segment treatments, 19 intersection treat- ments, and 2 miscellaneous treatments. They reached consen- sus on acceptable AMFs for eight of these and, after discussion, found three to be unnecessary for this class of roadways (see Table 17). There were four additional AMFs that were devel- oped by the NCHRP project research teams through further analysis after the meeting. The rural multilane arterials expert panel reviewed material on 10 segment treatments and 7 inter- section treatments and developed AMFs for 3 of these. In ad- dition, both expert panels recommended reanalysis of the data from the Elvik et al. study (74) to validate or modify the AMF for speed change versus crash frequency change. Since neither expert panel could reach consensus regarding an AMF for median width, they also recommended an additional analysis for this treatment. Both of these efforts were described in Chapter 3 of this report. The details on all AMFs resulting from these expert panels are presented in Chapter 5 of this report. 33