Many agencies are experimenting with enhanced pavement markings to reduce crashes and/or crash rates (i.e., adding markings to rural two-lane highways, adding wider edge lines, installing specially designed wet markings, etc.). Much of this emphasis has resulted from national programs such as AASHTO’s Safety Plan as described earlier. Other factors, such as increased emphasis on accommodating older drivers, have also inspired agencies to evaluate their marking programs.
Some studies have suggested that the use of markings plays a role in the reduction of specific crash types under certain conditions (6, 7, 8). Run-off-road and opposite-direction crashes are generally overrepresented on our nation’s highways, especially on horizontal curves and at night, when fatal crashes are three to four times more likely to occur. In addition, due to visual and cognitive deficiencies, older and impaired drivers are especially susceptible to these types of crashes. Therefore, crash types that are most likely affected by added markings or enhanced markings (added width or more retroreflectivity) are run-off-road and opposite-direction crashes that occur at night, occur on curves, and involve drivers with reduced visual or cognitive capabilities (e.g., older drivers or impaired drivers).
Early Pavement Marking Safety Studies
The earliest research shows consensus that edge line markings provide crash reductions versus no markings at all. For instance, in 1957 a study was initiated to install edge lines on rural two-lane highways that were at least 20 feet wide (9). A before-after crash study showed a 19 percent reduction in crashes after the installation of the edge lines. In addition, edge lines resulted in a 37 percent reduction in fatalities and injuries, a 63 percent reduction in crashes at access points such as intersections and driveways, and a 35 percent reduction in nighttime crashes. Similarly, another study initiated in 1959 showed that adding edge lines on rural two-lane highways 20 to 26 feet wide and with a minimum average daily traffic (ADT) of 1,000 vehicles per day (vpd) resulted in a 78 percent reduction in fatalities and a 46 percent decrease in the number of crashes at access points (10).
Probably one of the most robust studies to date concerning the safety benefits of adding pavement markings (edge lines in this case) was conceived from the Highway Safety Act of 1973, which established specific highway safety improvement programs including a pavement marking demonstration program that provided 100 percent federal funding for pavement markings on all highways excluding the Interstate Highway System (11). The Surface Transportation Assistance Act of 1978 continued funding for the program through fiscal year 1981.
Although 38 states participated in the demonstration program effort, data from only six states met the minimum criteria for a focused study on pavement marking benefits (resulting in 225 study sites). The minimum criteria for the study sites were that they had to be on two-lane highway sections at least 5 miles in length, have a pavement width of at least 16 feet, have a speed limit of at least 40 mph, and have no other safety improvements except adding pavement markings. The statistical approach used daytime crashes as a control mechanism for regression to the mean. Property damage only (PDO) crashes were not used, and fatal and injury crashes were combined.
Overall, there was a statistically significant 12 percent decrease in nighttime crashes. Adding edge lines resulted in a statistically significant 16 percent decrease in nighttime crashes and a statistically significant 33 percent decrease in low-visibility nighttime crashes.
The results also showed that adding centerlines and edge lines, and adding edge lines to roadways that previously only had centerlines, was most effective on roadways in mountainous and rolling terrain. The reduction in both nighttime and low-visibility nighttime crashes was statistically significant for both mountainous and rolling terrain.
When considering pavement width, several insightful findings appeared. For instance, 22-foot pavements exhibited a 36 percent reduction in nighttime crash rates and a 52 percent reduction in low-visibility nighttime crashes (when edge lines were added to existing centerlines or both centerlines and edge lines were added). For the same group of roadways and for 20-foot pavements, there was a statistically significant 13 percent decrease in nighttime crashes and a statistically significant 23 percent decrease in low-visibility nighttime crashes. For pavement widths of 18 feet or less, there was a statistically significantly 46 percent decrease in low-visibility nighttime crashes.
The report shows that adding edge lines to existing two-lane roadways with centerlines is a cost-effective crash-reducing treatment. A detailed example of North Dakota’s edge line program shows a benefit-cost (B/C) ratio of 23:1, despite an increase in PDO crashes.
More Recent Pavement Marking Safety Studies
Much has changed since the first pavement marking studies, including vehicle design, vehicle speeds, and traffic volumes. A more recent study, and one of the most often cited pavement marking safety studies, was published by Miller in 1991 (12). Using crash statistics and cost estimates from that time, Miller determined that even on rural two-lane roads with an ADT as low as 500 vpd, edge lines yield a B/C ratio of 17:1. On average, Miller showed a B/C ratio of 60:1, noting that the B/C ratio increases with traffic volumes and the urban ratio is twice the rural ratio. Miller further concluded that edge lines would be justified on two-lane rural roadways if an average of one non-intersection crash occurs annually every 15.5 miles.
Miller included a meta-analysis of studies with pavement marking safety numbers. He found, using studies deemed credible, an average crash reduction of 21 percent that could be attributed to pavement markings. One of the reports he reviewed was by Bali et al. (13), who examined delineation treatments on rural two-lane highways. This was a 10-state study including more than 500 sites. Their study found that adding edge lines and centerlines reduced crashes by 36 percent. Adding edge lines to existing centerlines reduced crashes by 8 percent. Using the Bali et al. data, Miller produced a B/C ratio for adding edge lines to rural two-lane highways as a function of ADT (see Figure 1).
FIGURE 1 Benefit-Cost Ratio for Adding Edge Lines on Two-Lane Highways.
Even more recently, several meta-analysis efforts have focused on estimating crash reduction factors expected for specific countermeasures. These documents include comprehensive literature reviews for various pavement marking applications by factors such as crash type, crash severity, and lane volume (14, 15). A summary of crash reduction factors for various pavement marking countermeasures is shown below.
Countermeasure | Crash Reduction Factor |
---|---|
• Install lane lines to multilane urban roads | 18 |
• Install centerlines | –1 to 36 |
• Install centerlines and edge lines | –3 to 24 |
• Install edge lines | 4 to 66 |
Safety of Wider Pavement Markings
Naïve before-after crash studies conducted in Virginia and New Mexico in 1987 and 1988 suggest that wider lines have no safety benefit in terms of reducing crashes (6, 16). However, these studies were hampered by insufficient data and lack of experimental control. In an ongoing Federal Highway Administration (FHWA) study, commenced in 2006, researchers are taking a much more extensive look at the safety of wider pavement markings (17). As part of the current study, a nationwide survey was conducted to identify states that have wider pavement markings (wider than 4 inches) on all or some of their highways. For those states using wider pavement markings, follow-up phone surveys were used to determine if:
- the locations (by route number and linear reference) could be determined;
- the use was extensive on roadway segments (i.e., not small spot treatments);
- the date of installation was known; and
- sufficient crash, traffic, and roadway data existed.
The convergence of all the necessary criteria was rare, but three states were identified as having the required information—Michigan, Illinois, and Kansas. To date, the researchers have focused their efforts on rural two-lane highways in Illinois and Michigan. The total numbers of crashes by severity along with single-vehicle and opposite-direction crashes have been disaggregated by time of day, driver age, and weather. The widespread use of wider lines in these states minimizes the concern of selection bias or regression to the mean.
In Illinois, data screening reduced the rural two-lane data set to 3,973 segments (1,817 miles) consisting of 3,224 segments (1,511 miles) with 4-inch edge lines and 749 segments (306 miles) with 5-inch edge lines.
TABLE 1 shows the estimates of the negative binomial regression model coefficients. The regression coefficient for edge line width was negative and statistically significant at a = 0.05, which indicates a positive safety effect of wider edge lines (i.e., a smaller number of crashes is associated with wider edge lines). It can also be observed that the signs of the coefficients for lane width, shoulder width, log of annual average daily traffic (AADT), and presence of horizontal curve (whenever they are statistically significant) are consistent with intuition.
Variable | Total Crashes | Fatal Injury | PDO | Day | Night | Daytime Fatal Injury | Nighttime Fatal Injury |
---|---|---|---|---|---|---|---|
Intercept | –5.0743 | –5.9166 | –5.5362 | –7.0307 | –4.8645 | –7.0871 | –6.1145 |
Edge line width | –0.0403 | –0.3295 | 0.0326 | –0.1995 | 0.0164 | –0.3986 | –0.2021 |
Lane width | –0.0771 | –0.0966 | –0.0733 | –0.1090 | –0.0580 | –0.1120 | –0.0720 |
Shoulder width | –0.0121 | –0.0339 | –0.0073 | –0.0320 | –0.0014 | –0.0391 | –0.0316 |
Log AADT | 0.8489 | 0.9607 | 0.8269 | 1.1263 | 0.6651 | 1.1064 | 0.7453 |
Curve presence | 0.3146 | 0.6294 | 0.2038 | 0.3156 | 0.3375 | 0.3910 | 0.9120 |
Dispersion | 0.4238 | 0.5681 | 0.4446 | 0.5793 | 0.4133 | 0.8003 | 0.3184 |
Pearson chi-square/DF | 1.2993 | 1.2655 | 1.2287 | 1.4470 | 1.1219 | 1.3075 | 1.0695 |
Variable | Wet | Wet Night | Single Vehicle | Single Vehicle Wet | Older Driver | Opposite Direction | Fixed Object |
Intercept | –7.7794 | –7.5721 | –3.5669 | –5.7371 | –7.0303 | –14.7968 | –4.6039 |
Edge line width | –0.2539 | –0.1727 | –0.0044 | –0.2196 | –0.1127 | 0.2422 | –0.2865 |
Lane width | –0.0608 | –0.0270 | –0.0511 | –0.0146 | –0.0657 | –0.1174 | –0.0528 |
Shoulder width | –0.0192 | 0.0015 | –0.0072 | –0.0149 | –0.0167 | –0.0202 | –0.0655 |
Log AADT | 0.9893 | 0.7668 | 0.5547 | 0.6005 | 0.9307 | 1.5208 | 0.6926 |
Curve presence | 0.4410 | 0.5484 | 0.4165 | 0.5715 | 0.1958 | 0.6133 | 0.7616 |
Dispersion | 0.7240 | 0.6887 | 0.4068 | 0.7235 | 0.5613 | 0.4919 | 0.5084 |
Pearson chi-square/DF | 1.1055 | 1.0854 | 1.1337 | 1.0966 | 1.2856 | 1.2123 | 1.2565 |
Note: Significant (at [omitted symbol] =0.05) effects are shown in bold.
For Illinois, raised reflective pavement markings (RRPMs) are used statewide, and rumble strips are used on interstates statewide. It needs to be noted, however, that the information on additional delineation and guidance measures (other than RRPMs and rumble strips) and on the roadway curvature was not available and could not be incorporated into the analysis. Therefore, the above observations are based on the assumption that the effects of the variables not in the database, such as those additional delineation/guidance measures and the roadway curvature, are the same (or averaged out) for the segments with and without wider edge lines.
In Michigan, before-after evaluations were conducted with 3 years (2001~2003) of before and 2 years (2005~2006) of after data obtained from 386 rural two-lane segments corresponding to 1,223 miles of rural two-lane roadways. Although regression-to-the-mean bias is not expected to be present (because wider lines were installed statewide in 2004), the empirical Bayes before-after evaluations (see, e.g., Hauer [18]) were employed to account for the remaining sources of bias such as differences between before and after study periods in traffic volumes, weather, vehicle fleet, driver characteristics, economic conditions, reporting practice, etc.
The empirical Bayes before-after evaluations (using the Illinois segments with 4-inch edge lines as the reference sites) resulted in the following crash reduction estimates: total (5.8 percent), fatal and injury (24.6 percent), PDO (3.9 percent), daytime (10.9 percent), nighttime (3.6 percent), daytime fatal and injury (28.7 percent), nighttime fatal and injury (39.5 percent), wet (30.9 percent), wet night (33.2 percent), single vehicle (1 percent), single vehicle wet (27.6 percent), single vehicle night (0.9 percent), and opposite direction (39.3 percent). All of these crash reduction estimates but nighttime, single-vehicle, and single-vehicle night crashes, were statistically significant at the 95 percent level. The researchers are finalizing their rural two-lane analyses with the addition of data from the Kansas Department of Transportation.