Cover Image

Not for Sale

View/Hide Left Panel
Click for next page ( 79

The National Academies of Sciences, Engineering, and Medicine
500 Fifth St. N.W. | Washington, D.C. 20001

Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement

Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 78
78 5. Choose best treatment(s) by considering estimated treatment based on this rate per mile, and choose the routes effectiveness, cost and other technical and policy to be treated based on the highest rankings plus other tech- considerations. nical and policy factors. The analyst will then combine the output of the steps An excellent example of location-specific targeting that above with at least two other factors in making a final can be done if crashes are mileposted to specific roadway decision on which treatment(s) to implement the cost of locations involves ramp treatments to prevent or reduce the treatment and other technical and policy considera- truck rollovers. If ramp-related crashes (based on "rela- tions. Unfortunately, there are no good guidelines for how tion to junction") are mileposted, even if just to the inter- to "weight" the different factors. While problem size (total change mainline, the analyst can determine which specific crash cost) and assumed treatment effectiveness are key interchanges and ramps exhibit the largest problem. factors, there may be technical, policy, and cost consider- ations that will remove certain treatments from consider- Note again that the lack of treatment effectiveness data ation even if they are felt to be effective. The analyst will means that the analyst will not be able to verify whether or have to choose the final treatments based on best judg- not a specific set of implemented strategies can be expected to ment. The procedure outlined above will at least ensure meet the established crash-reduction goal. In these cases, the that the major factors in the decision are clearly defined. best that can be done is to proceed in selecting strategies and The output of this step will be one or more chosen treat- target subgroups, times or locations until the available budget ments, with the nature of the treatment defining the spe- for safety improvement has been fully committed. The total cific crash types more likely to be affected. benefit of the selected program will not be forecastable, but 6. Target the chosen treatments to the vehicle types and the success of the program can be determined if a sound crash types where the problem is found. evaluation is conducted after its implementation. In some cases, treatment strategies related to these vehicle Where quantitative estimates or approximations of treat- types will be implemented jurisdiction-wide. In other cases, ment effectiveness can be made, it may be possible to pro- it may be desirable to target the treatment to specific loca- vide estimates of net impact (number of crashes prevented) tions. If targeting is to be done by location, the treatment by multiplying the unit treatment effects by the number of could be targeted to counties, city areas, or routes/streets drivers or roadway segments treated. showing the highest total crash cost or frequency, coupled with the analyst's judgment of potential differences in cost Closure Good Data Produce Better Results between locations and technical and political issues. Most of the strategies in these two guides are related to treating the Choosing treatments and targeting those treatments to the driver or vehicle, rather than the roadway, and this targeting vehicle populations covered in this section is difficult. The to jurisdiction would appear to be the most appropriate. It programs are complex, there is virtually no crash-based would be difficult to target further by crash type or other fac- information on treatment effectiveness for the strategies cov- tors. If the analyst is considering the roadway-related strate- ered in the two guides, and there is limited information on gies, and if the crash data are mileposted, the analyst could program costs. However, choices have to be made given that (1) link crashes to routes and search for the locations of available budgets will always be limited to some degree. It is "clusters" of target crashes for possible treatment, or (2) use hoped that the procedures presented in this section at least a network screening program similar to that described under provide some insight into how such choices can be made. Procedure 2A to identify 1-mile sections with the highest The assumption in this section has been that crash data are crash frequency or total crash cost. The identified windows available, but not necessarily other data such as roadway in- could then be ranked by frequency or total crash cost to ventories. As is obvious in the procedures above, the avail- identify priority locations. The analyst would then correct ability of mileposted crash data will result in improved treat- for "treatment gaps" using the same logic provided in Pro- ment targeting for roadway-related strategies (and perhaps cedure 2A (see Section IV). If the crashes are not mileposted, enforcement strategies), and the availability of linkable (and but there is information available on jurisdiction and route, thus mileposted) inventory data would further increase the the analyst could link crashes to routes within the jurisdic- analyst's ability to both choose treatment strategies and to tion and calculate the total crash cost or number of target target them. For example, inventory data could provide crashes per mile by dividing the sum of the crash costs or the detailed data not found in crash data files on such items as sig- sum of the number of target crashes on that route by route nal timing, intersection layout, and street width, all of which length. The analyst could then rank the potential routes for are related to treatment strategies listed in the guides.