Impact of Shoulder Width and Median Width on Safety (2009)

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.

9Roadway projects where design elements trade-offs are considered typically incorporate a full range of geometric and traffic operational problems, coupled with increasingly restric- tive environmental constraints. These problems may require variation from the normally used guidance values or tradi- tional solutions. Moreover, every project is unique in terms of the geometric conditions, traffic, safety history, purpose and need, project context, community character, and public pri- orities. What is reasonable or may work in one location may not be appropriate in another for any number of technical or context-sensitive reasons. The literature review conducted for this research examined safety implications from geometric element trade-offs, and the findings are presented herein. In addition, NCHRP Synthesis of Highway Practice 299: Recent Geometric Design Research for Improved Safety and Operations presents an extensive literature review on geometric design elements for improving safety and operations (8). The follow- ing section presents first an overview of roadway design issues and then the findings on the effects of specific cross-section elements for multilane highways. Roadway Design Issues The Green Book lacks background information sufficient for understanding the safety and operational implications of combinations of critical geometric features. The recently published Guide for Achieving Flexibility in Highway Design provides some information on these areas, but also lacks any quantifiable relationships for the values of various design elements (5). There are several geometric features that have a greater effect when combined than when considered alone— for example, Zegeer and Deacon (9) showed that the combined lane and shoulder width has a greater impact on the safety level of two-lane rural roadways than does lane or shoulder width alone. At the same time, there are cases where these combi- nations have little or no impact. The same combination of lane and shoulder width has a small to possibly no impact on four-lane roadways. Thus, these relationships and their areas of application must be further examined. Another Green Book topic requiring additional background information for designers centers on the relative importance of various geometric elements on safety. It is apparent that not all geometric elements have the same impact on safety and operational effectiveness, and the selected design value can affect additional elements. For example, the choice of a design speed of 45 mph or less for a road allows the designer to use a smaller minimum curve radius, a narrower clear zone, a shorter vertical curve, and shorter sight distances than those for a higher design speed. Here, the impact is significantly greater than when selecting a single design element to be adjusted. Moreover, roadway elements can exert varying degrees of influence even through a single element. For example, lane width will exert an impact on a two-lane roadway that will be different from that exerted on a four-lane roadway. There- fore, a prioritized list is needed to identify the relative signif- icance of each geometric element. Given the current definition of design speed, it is probably the most critical design element to be selected since it has the potential to impact the values used for almost all other design elements (1, 5). Most studies dealing with safety and speed typically con- sidered speed limit and so little is known about the influence of design speeds on safety. It could be assumed that there is some relationship between design speeds and speed limits, but because of the methods used to establish speed limits in many states, it is not feasible to develop a systematic relation- ship between the two (10). Current highway design approaches emphasize speed as a surrogate for quality and efficiency. This approach is probably reasonable for rural areas where high speeds are frequently desirable, but not for roads in urban or suburban areas. Several studies have examined cross-section elements and attempted to develop models or relationships that could esti- mate safety implications from varying individual components. The work of Zegeer et al. (11–13) identified the relationship C H A P T E R 2 Literature Review

of lane and shoulder width to crashes on two-lane rural roads and quantified these by developing models later included in the Interactive Highway Safety Design Model (IHSDM). A significant and potentially useful conclusion from the literature is that the important element in crash reduction is the total available roadway width. The studies on converting two-lane roads to four-lane roads show that, in general, safety gains are achieved with such conversions (14, 15). The findings of NCHRP Report 330: Effective Utilization of Street Width indicate that there are certain designs for urban arterials where the implementation of strategies that involve the use of narrower lanes has an effect on safety (16). Such strategies include the use of center two-way left-turn lanes or removal of curb parking, and most of these strategies involved projects with restricted right of way and arterials with speeds of 45 mph or less. The study also concluded that even though the use of narrower lanes, when considered alone, may increase specific crash types, the presence of other design features, such as the addition of two-way left-turn lanes, may offset these increases. This study also underscores the potential of interactive effects between various design elements and suggests careful evaluation of the use of narrower-than-typical lanes. A more recent review of safety in geometric design standards by Hauer (6) critically examined the belief that adherence to design standards is directly linked to safe roadways. This review indicated that design guidelines have an inherent safety level, but that little is known about the impacts of their flexibility application in roadway design. Another issue identified by Hauer was the notion that there are two different kinds of safety. One could be called nominal safety and is measured “in reference to compliance with standards, warrants, guide- lines, and sanctioned design procedures” (6). Substantive safety, by comparison, is based on the roadway’s actual safety performance—that is, crash frequency and severity. Designing nominally safe roads does not ensure substantive safe roadways since adherence to values of each guideline does not inherently produce a safe design. Several of the studies examined focused on developing models that investigate and quantify the sub- stantive safety changes from altering design dimensions (17). Another aspect of safety noted by Fambro et al. (18) is the concept that safety is a continuum and not a single yes/no decision. This implies that a change in the value chosen for a particular design element “can be expected to produce an incremental, not absolute change in crash frequency and severity” (17). However, there is a need to better understand the effect on the level of safety from these incremental changes, and such efforts are essential in understanding and quantify- ing the substantive safety of a roadway. This is critical for projects where design flexibility is considered. Stakeholders do not easily accept designs that are considered nominally safe, but require the evaluation of design alternatives that may deviate from the nominal designs. An additional concept that merits attention is that of the presence of a tipping point—the principle that small changes have little or no effect on a system until a crucial point is reached (19). This concept, which has been extensively used in epidemiological research, could also be used in roadway design because of the available flexibility in the values of design elements. It could be hypothesized that safety and operational consequences from altering the values of design elements while remaining within the suggested Green Book values are minimal and, thus, do not create significant problems. Moreover, small departures from these values may have no significant impact, and thus the safety consequence tipping point for any single design value may not be detectable. Highway design typically requires a multi-level assurance by professional engineers that the approved design will not result in unacceptable levels of safety consequence. Projects requiring a design exception could be considered as those that are the farthest from the most desirable design value. The recently completed NCHRP Project 15-22, “Safety Consequences of Flexibility in Highway Design,” found that the small deviations noted in the case studies analyzed indicate that a generally conservative approach is taken when considering values that vary from traditional design (20). Cross-Section Elements The literature review conducted for this research focused on three cross-section elements: lane width, shoulder type and width, and median type and width. This section discusses the findings for these design elements. Several of the find- ings have been cross-referenced with the interim report from NCHRP Project 17-27, “Parts I and II of the Highway Safety Manual,” (21) and NCHRP Web-Only Document 126: Method- ology to Predict the Safety Performance of Rural Multilane Highways (22). Lanes Wider lanes are traditionally associated with higher oper- ating speeds and increased safety. The Highway Capacity Manual (HCM) (23) documents that wider lanes for multi- lane highways result in higher free-flow speeds. On the other hand, very little has been found on the safety implications of wider lanes. It is reasonable to assume that wider lanes may provide additional space to the driver to correct poten- tial mistakes and thus avoid crashes. However, a driver could be expected to adapt to the available space, and the positive safety effects from the wider lanes may be offset by the higher speeds. Most completed research on this topic has focused on the lane width of two-lane, two-way roads, and very little is known of the effect of lane width of multilane rural highways (24). 10

The review conducted by Hauer (24) of studies that attempted to model the effect of lane width on multilane rural highway crashes found no correlation. The same review indicated that there was only one study where lane width was included in the models (25), but these were for freeway facilities. An AMF represents the anticipated change in safety when a particular geometric design element value changes in size. An AMF greater than 1.0 represents the situation where the design change is associated with more crashes; an AMF less than 1.0 indicates fewer crashes. Typically, AMFs are estimated directly from the coefficients of models derived using crash data or expert panels that review current literature and determine the magnitude of the AMF. Estimation of AMFs from models assumes that (1) each AMF is independent since the model parameters are assumed independent and (2) the change in crash frequency is exponential. In practice, AMFs may not be completely independent since changes in geometric design characteris- tics on highways are not done independently (e.g., lane and shoulder width may be changed simultaneously) and the com- bination of these changes can influence crash risk. Nonetheless, experience in deriving AMFs in this manner indicates that the assumptions are reasonable and, with thoughtful model development, the resulting AMFs can yield useful information about the first-order effect of a given variable on safety. A study by Harwood et al. (26) examined AMFs as part of resurfacing, restoration, and rehabilitation (3R) projects. An expert panel adjusted the AMFs developed for two-lane, two-way rural roads to allow for their use in multilane roads, specifically four-lane roads. The factors show no effect for 11-ft lanes and an 8% to 11% increase for 9-ft lanes. These AMFs are summarized in Table 1. The section of the HSM on multilane rural roads devel- oped as part of NCHRP Project 17-27 (21) also proposed AMF values for lane width on rural multilane highways (see Table 2) based on the work of Harwood et al. (26) and Harkey et al. (27) through the deliberations of the joint NCHRP Projects 17-25/17-29 Expert Panel Meeting. Two sets of values were developed from the studies of Miaou et al. (28) and Harkey et al. (27), based on whether the roadway was divided in the presence of a median barrier. These values accounted for the total number of crashes while considering median- related crashes. The recommended values were adjusted from the normal baseline of 30-ft median presented in the report. Most available research has examined this relationship for urban roadways, and some relationship has been found between the lane width and crashes for these roadways. How- ever, these relationships are not applicable for the roadways considered in this research project (which examines multi- lane rural roads only) and therefore are not discussed further. In summary, there is limited past research documenting any effects of lane width on crashes for multilane rural roads. The only study with definitive factors is the new HSM work that is based on an expert-panel approach. Shoulders Shoulders placed adjacent to travel lanes accomplish several functions including emergency stop and pull off, recovery area for driver error, and pavement edge support (1). The use of shoulders to provide an area where a vehicle could stop poses an additional hazard since past research has shown that 11% of fatal freeway crashes are related to vehicles stopped on shoulders (29). There is also evidence that wider shoulders may encourage higher operating speeds because they may communicate to the driver the presence of wider space for correcting errors. Finally, the number of lanes, lane width, and shoulder width are interrelated, and the choice of geo- metric value for each of these elements typically affects the other elements. Most of the research completed to date focuses on two-lane, two-way rural roads (30). An additional problem is that most of the recent studies have analyzed urban or suburban multi- lane highways (rather than rural roads), resulting in an even smaller number of available references for this design element. Hadi et al. (25) examined the effect of shoulder width on crashes on multilane rural highways. Their findings indicated that for four-lane rural divided roads, a small reduction in crashes (1% to 3%) can be attained if the unpaved shoulder is widened by 1 ft. The authors also indicate that the roads with shoulder widths between 10 ft and 12 ft have the lowest crash rates. However, this relationship is present only for unpaved shoulders, and the reduction factor should be used cautiously. Harwood et al. (26) also produced AMFs for multilane highways, again using an expert panel to adjust the AMFs of two-lane rural roads. In this instance, the panel determined that the effect of shoulder width is similar for both multi- and 11 Lane width (ft) 9 10 11 12 Four-lane undivided 1.11 1.06 1.00 0.99 Four lane divided 1.08 1.04 1.00 0.99 Table 1. AMFs for lane width for four-lane highways (21). Lane width (ft) Roadway 9 10 11 12 Undivided 1.13 1.08 1.02 1.00 Divided 1.09 1.05 1.01 1.00 Table 2. AMFs for lane width (22).

two-lane rural roads, so the AMFs could remain the same. The proposed AMFs are presented in Table 3. Further research interest has been placed on shoulder type, which can impact crashes and therefore roadway safety. Again, the focus of work on this topic has concentrated on the two-lane, two-way roads: almost no research has been directed to multilane roads. Rogness et al. (31) used before-and-after crash-rate changes from converting two-lane rural roads with full shoulders to four-lane undivided rural roads without shoulders. The results indicated that for roads with volumes in the 1,000–3,000 vehicles/day range, crashes increased after the conversion. It should be noted here that the study used Texas roadways where, the report indicates, driving on the shoulder on two-lane rural roads is considered acceptable. This fact could impact the findings of their study and therefore not provide any additional understanding of this shoulder-crash relationship. Harwood et al. (32) developed AMFs for the conversion of shoulder types on rural two-lane roads. An expert panel reviewed these factors and determined that they are appropriate for use in both divided and undivided multilane roadways. These estimates, shown in Table 4, were for converting turf or gravel shoulders to paved shoulders and turf shoulders to composite (partially paved) shoulders. Harkey et al. (27) also developed AMFs for rural multilane roadways as part of a study that evaluated traffic engineering and ITS improvements (see Table 5). The study considered undivided roads with more than 2,000 vehicles per day, and the AMFs developed were for roadways where the shoulder related crashes were 35% of the total. Additional procedures are available for roadways with lower volumes or different percentages. For divided highways, the draft HSM uses recommended values from NCHRP Project 17-29 (22), which developed AMFs for shoulder width for rural multilane segments. These AMFs are for paved shoulders and also include the Harkey et al. AMFs for undivided highways (see Table 6). In general, the literature is silent on the relationship between shoulder and safety for multilane rural roads with the exception of the new HSM work. As was the case for the lane width, there is no literature that documents the effect of shoulder width and type on the safety of a roadway seg- ment. Moreover, the new AMFs developed for the HSM are based mainly on an expert-panel approach and on the Harkey et al. work that is itself derived from Zegeer’s work (12, 13). Medians The most important objective for the presence of medians is traffic separation. Additional benefits from medians include the provision of recovery areas for errant maneuvers, accom- modation of left-turn movements, and the provision for emergency stopping. Median design issues typically address the presence of median, along with its type and width. There has been some research completed on these issues and their implications on safety. Hauer (33) conducted a review of studies that investi- gated the effect of medians on rural multilane highway safety levels. This review, which was based on a few studies, did not provide conclusive results on the effectiveness of the presence of medians on safety but did identify the potential for the median to impact safety. One of these studies (34) examined divided and undivided four-lane rural roadways in the context of the safety differences between two-lane and four-lane roadways. The study concluded that the presence of a median had an effect on crashes that was related to the 12 Paved shoulder width (ft; one side) 3 4 5 6 7 8 1.0 0.97 0.95 0.93 0.91 0.90 Table 3. AMF for shoulder width for multilane highways with ADT > 2500 vehicles/day (21). Shoulder width (ft; one side) Treatment 3 4 5 6 7 8 Convert turf to paved 0.99 0.98 0.97 0.97 0.97 0.96 Convert gravel to paved 1.00 1.00 1.00 0.99 0.99 0.99 Convert turf to composite 1.00 0.99 0.98 0.97 0.98 0.98 Table 4. AMFs for shoulder conversion for multilane roadways based on two-lane roads (21). Paved shoulder width (ft) 0 2 4 6 8 1.18 1.11 1.05 1.00 0.95 Table 5. AMFs for paved shoulder width (27).

roadway volume (crashes for roads with medians as com- pared with roads without medians exhibited the relation- ship 0.76 × ADT−0.05)1. Another study examined the effect of the median presence in Oregon and also reported crash reductions from the presence of medians (35). The study found that the AMF for median presence is 0.431, showing an agreement with the results of Council and Stewart (34), but a larger magnitude for its effect. Elvik and Vaa (36) also showed a similar finding with separate models for injury and property damage crashes in a meta- analysis of several studies where a median was added. Their AMFs were 0.881 for injury and 0.821 for property damage crashes. The interim report for NCHRP Project 17-27 recom- mended an AMF for the presence of median in the range of 0.85 to 0.50 (21). The contribution of width to the median effect has also been examined. Hauer (33) found that it was not possible to identify AMFs for median width but rather noted three safety trends: (1) cross-median crashes (i.e., opposing vehicles) are reduced with wider medians; (2) median-related crashes increase as the median width increases, with a peak at about 30 ft, and then decrease as the median becomes wider than 30 ft; and (3) the effect of median width on total crashes is questionable. Hadi et al. (25) used negative binomial models to show that the median width has an influence on multilane roadways; these authors produced two models based on the traffic volume range and number of lanes. This is the only study that examined the effect of median width on safety for rural, multilane roads because the several studies reviewed by Hauer (33) and the NCHRP Project 17-27 interim report (21) deal with freeway median width. Table 7, which is taken from the interim report for NCHRP Project 17-27, presents a set of AMFs for the effect of median width on crashes for four-lane rural roadways; these values are based on one study. The HSM section on multilane rural roads developed as part of NCHRP Project 17-29 (22) also proposes AMF values for rural multilane highways (see Table 5 in HSM). Two sets of values were developed based on whether a median barrier was present. These values are based on the studies of Miaou et al. (28) and Harkey et al. (27), and they account for the total number of crashes while considering median-related crashes. The recommended values are summarized in Table 8. Median type has also been examined as it relates to roadway safety. A meta-analysis of several studies conducted by Elvik and Vaa (36) suggests there is an effect due to the type of median used. Their analysis examined the relative effects of concrete, steel, and cable guardrail installations on multi- lane divided highways. The results indicate that the AMF for injury crashes for concrete barriers is 1.15, for steel barriers is 0.65, and for cable is 0.71. The resulting AMF for all crashes for median guardrails is 1.24, indicating that the presence of a median guardrail—and especially a concrete guardrail— has the potential to increase crashes. Thus, designers must carefully consider whether the placement of a median barrier will have an overall positive or negative influence on the safety of a particular roadway segment. A barrier will result in a reduction of cross-median type crashes, but it also has the potential to increase median-related crashes since its absence could allow drivers opportunities to stop their vehicles in the median (37). As Hauer states: “The net effect of placing a barrier in the median is an increase in total accidents; an increase in injury accidents and its effect on the total number of fatal accidents is at present unclear” (33). Fitzpatrick et al. (38) developed AMFs for median barriers on freeways and four- and six-lane rural highways in Texas. For rural highways the influence of the median barrier was examined as a function of the available left shoulder width. The study concluded that for roads with a barrier, increasing the left shoulder width by 1 ft will result in a 1.6% reduction of crashes for both four- and six-lane highways. Other studies have demonstrated that the addition of a barrier could contribute to crash occurrence. Elvik (39) analyzed the results of 32 studies that examined the effect of median barrier presence. His major conclusion was that “ . . . the best current estimates of the effects of median bar- riers are a 30% increase in accident rate, a 20% reduction in the chance of sustaining a fatal injury, given an accident, and a 10% reduction in the chance of sustaining a personal injury, given an accident.” These findings indicate that, in general, crashes can increase, but their severity may decrease. Miaou et al. also noted that crash rates are higher on roadways with median barriers when compared with roads without them and that median barriers present a higher likelihood of vehicle impact (28). A median-type treatment that may be used on multilane rural roads is a two-way left-turn lane (TWLTL). This median type is typically found on rural roads where some develop- ment may be present or anticipated. Such a median treat- ment is often associated with specific types of crashes that are access related, that is, left turns in and out of an access point. An issue of concern in estimating safety impacts from 13 Paved shoulder width (ft) Roadway 0 2 4 6 8 Undivided 1.18 1.11 1.05 1.00 0.95 Divided 1.18 1.13 1.09 1.04 1.00 Table 6. AMFs for paved shoulder width (22). 1The values presented here are those stated in the NCHRP Project 17-27 interim report (21), and they have been adjusted from the original studies.

TWLTLs is the access density since it has the potential to significantly affect the opportunity for crashes. The impact of these treatments has not been extensively evaluated, and their safety gains still require additional verification (38). Hauer (33) estimated that the AMF for most urban and sub- urban TWLTLs ranges from 0.70 to 0.90 based on a review of several studies. These AMFs are for total number of crashes and not the types of crashes associated with the installation of the TWLTL. In summary, the presence of a median has a positive effect on safety, and some AMFs have been developed based on pre- vious studies. The median width has also an impact on road- way safety where wider medians tend to have a larger AMF. Finally, the placement of a barrier is a balancing act because a barrier has the potential to increase median-related crashes but to reduce cross-median crashes. Even though this element has been examined more than the other two elements, several of the reports reviewed indicated that for multilane roadways, additional research is required either to develop new AMFs or to validate existing AMFs. Rural Two-Lane Conversions to Multilane A typical project for rural roadways is the conversion of a two-lane road to a four-lane road with or without a median. Using crash data from four Highway Safety Information System (HSIS) states, Council and Stewart (34) attempted to estimate the safety effects from such conversions on rural roads. The study indicated that safety gains ranging from 40% to 60% were achieved for divided roadways, while smaller gains— approximately 20%—were achieved for undivided roads. These estimates were developed using typical cross sections for each roadway type. The authors cautioned that these findings were based on a predictive model and should be validated with actual before-and-after crash data to provide sound support for the conclusions. Agent and Pigman (40) compared the safety impacts of either (1) converting two-lane rural roads to four-lane roads or (2) realigning two-lane roads. The study examined 49 con- version locations and 24 locations where the two-lane roadway was upgraded with realignment and widening of lanes and shoulders. The study concluded that both conversions to four lanes and upgrades of two-lane roadways reduced crashes after project completion. There was a 56% reduction for converted roadways and a 51% reduction for upgraded two-lane roadways. A comparison to statewide crash rates for each roadway type revealed that converted four-lane roads exhibited crash rates similar to the statewide average, while crash rates of upgraded two-lane roads dropped to approxi- mately one-half the statewide rate for two-lane rural roads. The influence of volume on both upgraded and converted roads was also cited, and the authors acknowledge that additional work is needed to evaluate volume impact and determine which approach—conversion or upgrade—is more appropriate. The important finding of this study is that both approaches improve safety and should be considered as design alternatives. Summary A significant body of research that attempts to quantify the relationships between safety and roadway design elements has been compiled. As previously noted, NCHRP Synthesis of Highway Practice 299 has reviewed and discussed several of these issues at length, and the reader seeking more detailed information is encouraged to review that publication (8). Several studies have focused on two-lane rural roads and have addressed issues relative to lane widths, shoulder widths and types, clear zones, and horizontal and vertical alignments. Even 14 Median width (ft) 10 20 30 40 50 60 70 80 90 1.00 0.91 0.85 0.80 0.76 0.73 0.70 0.67 0.65 Median width (ft) Barrier 15 20 30 40 50 60 70 80 90 With 1.019 1.012 1.000 0.988 0.977 0.967 0.953 0.944 0.935 Without 1.010 1.006 1.000 0.994 0.988 0.983 0.978 0.973 0.968 Table 7. AMFs for median width in four-lane rural non-freeway roads (21). Table 8. AMFs for median width in rural multilane roadways (22).

though these are the general areas of interest for this research, there is a lack of information regarding any association between typical and other than typical design values for several design elements. To some degree, the design elements selected for further examination in this research have the potential to affect safety. The degrees of influence vary by design element and application and often are specific to a set of roadway conditions. There are current parallel efforts under way to address the quantification of the safety and operational impacts from design element trade-off. Specifically, such models exist for two-lane rural highways, and similar models will be developed in the near future for multilane highways. The most directly applicable lesson from the literature is that values for design elements can be varied. Most research has been directed to the task of evaluating specific design elements, without considering the effects when multiple elements are varied in combination. An additional issue that has not been discussed extensively is the potentially opposite effects that selected values for design elements can impart. For example, wider shoulders have shown the potential to improve safety. On the other hand, they also have the potential to encourage increased operating speeds that, in turn, can lead to increased crash severity. A similar counterbalancing potential was noted for the presence and type of barrier in medians. Therefore, design decisions and countermeasure applications should consider the types of crashes associated with the modification and then determine the appropriate design element. A summary of the literature reviewed and pertinent findings relative to the objectives of this research project are presented in Table 9. 15 Table 9. Summary of literature review. Reference Element Results Comments Harwood et al. 2003 (26) Lane width AMF for lane width Lane width (ft) 9 10 11 12 Four-lane undivided 1.11 1.06 1.00 0.99 Four-lane divided 1.08 1.04 1.00 0.99 AMF for lane width is based on rural two-lane roads and from expert panel recommendation Lord et al. 2008 (22) Lane width AMF for lane width Lane width (ft) Roadway 9 10 11 12 Undivided 1.13 1.08 1.02 1.00 Divided 1.09 1.05 1.01 1.00 AMF for undivided is expert panel based in the HSM; divided is based on models Harwood et al. 2003 (26) Shoulder width AMF for shoulder width Paved shoulder width (ft; one side) 3 4 5 6 7 8 1.0 0.97 0.95 0.93 0.91 0.90 AMF for shoulder width is based on rural two-lane roads and from expert panel recommendation Harwood et al. 2000 (32) Shoulder type AMF for shoulder conversion Shoulder width (ft; one side) Treatment 3 4 5 6 7 8 Turf to paved 0.99 0.98 0.97 0.97 0.97 0.96 Gravel to paved 1.00 1.00 1.00 0.99 0.99 0.99 Turf to composite 1.00 0.99 0.98 0.97 0.98 0.98 AMF for shoulder conversion is based on rural two-lane roads and from expert panel recommendation Harkey et al. 2008 (27) Shoulder width AMF for paved shoulder width Paved shoulder width (ft) 0 2 4 6 8 1.18 1.11 1.05 1.00 0.95 AMF is developed from expert panel evaluating ITS improvements Lord et al. 2008 (22) Shoulder width AMF for paved shoulder width Paved shoulder width (ft) Roadway 0 2 4 6 8 Undivided 1.18 1.11 1.05 1.00 0.95 Divided 1.18 1.13 1.09 1.04 1.00 AMF is from expert panel for paved shoulders; recommended in the HSM. (continued on next page)

16 Lord et al., 2008 ( 22 ) Median width AMF for me dian width Median width (ft) Barrier 15 30 50 70 90 With 1.019 1.000 0.877 0.953 0.935 Without 1.010 1.000 0.988 0.978 0.968 Based on expert panel and reco mme nded in th e HSM Hauer 2000 ( 33 ) TWLTL AMF range for presence 0.70 to 0.90 Reviewing previous studies Elvik 1995 ( 39 ) Median presence Estim ated increase 30% for all crashes Based on prior studies for roads with barriers Fitzpatrick et al., 2008 (38) Median and left shoulder Roads with median, increasing left shoulder by 1 ft will result in 12% reduction in crashes at 4- and 6-lane highways AMF developed for roadways in Texas Council & Stewart 1999 ( 34 ) Median presence Crashes for roads with medians 0.76xADT -0.05 Based on study of converting 2-to 4-lane roads Strathm an et al. 2001 ( 35 ) Median presence AMF for roads with me dians 0.46 Larger than Council and Stewart but consistent trend; all crashes Elvik and Vaa 2004 ( 36 ) Median presence AMF for all crashes for roads with me dians 0.88 AMF for property dam age crashes on roads with me dians 0.82 Based on meta-analysis of several prior studies iTrans 2005 ( 21 ) Median presence AMF range 0.50–0.85 General statem ent by review of prior studies; difficult to be precis e iTrans 2005 ( 21 ) Median width AMF for me dian width Median width (ft) 10 20 30 050 70 90 1.00 0.91 0.85 0.80 0.70 0.65 AMF for shoulder width is based on rural two-lane roads and from expert panel reco mme ndation Elvik and Vaa 2004 ( 36 ) Median type AMF for me dian guardrails: 1.24 all crashes AMF for concrete barriers: 1.15 injury crashes AMF for steel barriers: 0.65 injury crashes AMF for cable barriers: 0.71 injury crashes Based on meta-analysis of several prior studies Table 9. (Continued).