U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590
202-366-4000


Skip to content
FacebookYouTubeTwitterFlickrLinkedIn

Safety

eSubscribe
eSubscribe Envelope

FHWA Home / Safety / Speed Management / Integrating Speed Management within Roadway Departure, Intersections, and Pedestrian and Bicyclist Safety Focus Areas

Integrating Speed Management within Roadway Departure, Intersections, and Pedestrian and Bicyclist Safety Focus Areas

< Previous Table of Contents Next >

Appendix A:

6. ANALYSIS OF THE ROLE OF SPEEDING IN SAFETY-CRITICAL EVENTS USING SHRP 2 DATA

The following summarizes two different types of analyses which were used to assess speeding characteristics of roadway departure, intersection, and bicycle and pedestrian events using the the Strategic Highway Research Program (SHRP) 2 naturalistic driving study (NDS). Baseline and safety critical (crash, near-crash, crash relevant) events were queried or extracted from the InSight Data Access Website (insight.shrp2nds.us).

In the first analysis, the rate of speeding in safety-critical events was compared to speeding in baseline events. In the second analysis, safety-critical events were further evaluated and characteristics of those events summarized.

6.1 DESCRIPTION OF DEFINITIONS

Event characteristics were determined by the Virginia Tech Transportation Institute (VTTI), which reduced events from the NDS data which was collected by SHRP 2. Reduced events are available on the InSight web site which can be accessed by qualified researchers. A set of driver, roadway, and environmental characteristics was reduced for crashes, near-crashes, and crash-relevant events. Hereafter, these events are termed "safety-critical events." Data were also reduced for a set of baseline events. A video clip of the forward roadway view and a narrative description is provided for safety-critical events, and as a result, some additional information can be extracted.

Video is not provided for baseline events, and a narrative is only provided for a limited number of baseline events. As a result, additional information could not be extracted for baseline events.

Relevant definitions for the analyses presented in this memo are provided below in italics as described on the InSight web site (insight.shrp2nds.us).

Currently, the posted or advisory speed is not provided for each event. The posted or advisory speed could be determined from the forward road view clip for some safety-critical events and was described in the narrative for others but was not consistently provided. Speed limit was not identified at all for baseline events. As a result, the definition of "speed-related" is based on VTTI's definitions for driver behavior. Advisory speed is not provided and the definition for exceeded speed limit does not account for drivers traveling over the curve advisory speed.

Exceeded speed limit: Subject vehicle traveling at a speed greater than the posted speed limit (not in a work zone). In Variable Speed Zones, this is relative to the speed limit in effect at the time of the event. Coded when more than 10 mph above posted speed limit.

Exceeded safe speed but not speed limit: Subject vehicle traveling at a speed close to or under the posted speed limit, but still too fast to maintain a safe driving environment given current environmental conditions (e.g., weather, traffic, lighting). (Not in a work zone.)

Ex. during conditions that may require slower speeds such as weather, traffic situation, etc.

Roadway characteristics are derived from several different fields rather than using a common definition such as rural 2-lane arterial or 4-lane urban collector. For instance the field "locality" provides some indication of location (urban versus rural) as well as roadway type using definitions such as open residential or interstate/bypass/ divided highway with no traffic signals. Other roadway characteristics such as number of lanes or presence of median, curve, intersection etc. have to be determined using a range of fields. This makes it difficult to compare by roadway types which are more familiar to traffic engineers (e.g. 4-lane arterial).

Open county: other than the roadway, there is nothing but vegetation visible, road is not an interstate or bypass/ divided road with traffic signals.

Open residential: rural to semi-rural areas with just a few houses.

Moderate residential: area with multiple homes/apartment buildings present.

Business/industrial: any type of business or industrial structure is present but is not as dense as urban.

Urban: higher density where blocks are shorter, streets are a mix of one and two-way, and traffic can include buses and trams.

Interstate/bypass/divided highway with no traffic signals: interstate, bypass, or divided highway where no traffic signals are present.

Bypass/divided highway with traffic signals: bypass or divided highway with traffic signals and no other category description is visible.

Three other types of localities were coded including "church", "playground", and "school", which indicated a church, playground, or school was present. Since this definition is very vague, events coded with these localities were not included in the comparison described in Section 2. The appropriate corresponding locality (i.e. open residential) was extracted for Section 3 to ensure consistency between the types of locations where events took place.

Driver distractions were coded and drivers may have included more than distraction. The definitions provided in Section 3 are general but are similar to the language used for coded events.

Impairment was coded and was an assessment of why drivers engaged in a particular behavior. The terms used in Section 3 include:

Drowsy, sleepy, asleep, fatigued: driver exhibits obvious signs of being asleep or tired or is asleep.

Angry: driver exhibits obvious signs of anger.

6.2 COMPARISON OF SPEEDING FOR SAFETY-CRITICAL EVENTS TO BASELINE EVENTS

In the analysis described in this section, the rate of speeding in safety-critical events was compared to speeding in baseline events. This was obtained by querying the number of events that met a particular set of filters (i.e. baseline events on moderate residential roads coded as exceeded speed limit). The following tables compare crash, near-crash, and crash relevant (safety critical) events to baseline events. Both types of events were queried for specific roadway types as defined below. Safety-critical events were also disaggregated by crash type (i.e. roadway departure).

The definition of "exceeded safe speed" was relatively subjective and was highly correlated to roadway surface condition. As a result, for this analysis, events were defined as speeding if they met the criteria of "exceeded speed limit." Additionally only events coded as having dry or wet surface conditions were included.

Speeding versus not speeding is compared for various roadway types (non-intersection) and for intersections. Presence of pedestrians and bicyclists are not indicated in baseline events and are not included in this analysis (see Section 3).

Table 2 shows speeding versus not speeding by various roadway categories. Data are provided for non- intersection baseline and safety-critical events. Safety-critical events may have included rear-ends, near rear-ends, roadway departures, loss of control, etc. Information is compared by roadway types. VTTI defined roadway types. In most cases, these did not correspond to commonly understand roadway types (e.g. rural 2-lane arterial). As a result, roadway types which seemed to be similar were combined.

The percentage of speeding related events is provided next to the number of speeding related events and was calculated as (speeding related events/all events) for a particular category.

The odds of a driver having "exceeded the speed limit" for safety-critical events compared to baselines events for the same roadway type were compared. The 95 percent confidence interval (CI) for the odds are also provided. When the confidence interval contains 1.0, the odds are not statistically significant.

As noted in Table 2, safety-critical events were more likely to be speeding related than for corresponding baseline events. For instance, safety-critical events in general that occurred on two-lane roadways where a curve is present were 6.6 times more likely to be speeding related than baseline events on similar roadways. Around 14 percent of safety-critical events were speeding related while 2.4 percent of baseline events were speeding related.

Table 2: Speeding versus Not Speeding by Roadway Type for All Safety-critical Events
Baseline: 2-lane (open county, open residential, moderate residential, business/industrial) Crash/near crash Odds (CI)
curve (left or right)
not speeding 486 61 6.64 (2.75, 16.01)
speeding 12 (2.4%) 10 (14.1%) 6.64 (2.75, 16.01)
tangent
not speeding 2768 351 2.67 (1.64, 4.34)
speeding 68 (2.4%) 23 (6.1%) 2.67 (1.64, 4.34)

Baseline: Divided and one way (open county, open residential, moderate residential, business/industrial) Crash/near crash Odds (CI)
curve (left or right)
not speeding 97 26 NA
speeding 2 (2.0%) 0 (0%) NA
tangent
not speeding 880 154 3.05 (1.27, 7.31)
speeding 15 (1.7%) 8 (4.9%) 3.05 (1.27, 7.31)

Baseline: Interstate/bypass/divided no signal Crash/near crash Odds (CI)
curve (left or right)
not speeding 353 11 6.02 (1.53, 23.71)
speeding 16 (4.3%) 3 (21.4%) 6.02 (1.53, 23.71)
tangent
not speeding 2032 152 1.24 (0.67, 2.30)
speeding 129 (6.0%) 12 (7.3%) 1.24 (0.67, 2.30)

Baseline: Bypass/divided with signal Crash/near crash Odds (CI)
curve (left or right)
not speeding 56 4 NA
speeding 0 (0%) 0 (0%)  
tangent
not speeding 342 27 3.45 (0.91, 13.13)
speeding 11 (3.1%) 3 (10.0%) 3.45 (0.91, 13.13)

Baseline: Entrance exit ramps Crash/near crash Odds (CI)
not speeding 260 54 1.60 (0.16, 15.72
speeding 3 (1.1%) 1 (1.8%) 1.60 (0.16, 15.72

Table 3 evaluates the types of safety-critical events that are typically roadway departures (roadway departure, sideswipe, opposite direction, animal). Information is shown for curves and tangents separately for two-lane roadways, but were combined for other roadway types due to the small sample size.

As noted, roadway departure safety-critical events were more likely to be speeding than baselines events on similar roadways for two-lane curves and interstate/bypass/divided with no signal. The results were not statistically significant for the remaining roadway types.

Table 3: Speeding versus Not Speeding by Roadway Type for Roadway Departure Safety-critical Events
Baseline: 2-lane (open county, open residential, moderate residential, business/industrial) Crash/near crash Odds (CI)
curve (left or right)
not speeding 486 37 7.67 (2.85, 20.63
speeding 12 (2.4%) 7 (15.9%) 7.67 (2.85, 20.63
tangent
not speeding 2768 147 1.66 (0.71, 3.89)
speeding 68 (2.4%) 6 (3.9%) 1.66 (0.71, 3.89)

Baseline: Divided and one way (open county, open residential, moderate residential, business/industrial) Crash/near crash Odds (CI)
not speeding 977 51 1.13 (0.15, 8.64)
speeding 17 (1.7%) 1 (1.9%) 1.13 (0.15, 8.64)

Baseline: Interstate/bypass/divided no signal Crash/near crash Odds (CI)
not speeding 2385 38 3.03 (1.33, 6.90)
speeding 145 (5.7%) 7 (15.6%) 3.03 (1.33, 6.90)

Baseline: Bypass/divided with signal Crash/near crash Odds (CI)
not speeding 56 9 NA
speeding 0 (0%) 2 (18.2%) NA

Baseline: Entrance exit ramps Crash/near crash Odds (CI)
not speeding 260 21 4.13 (0.41, 41.43)
speeding 3 (1.1%) 1 (4.5%) 4.13 (0.41, 41.43)

Table 4 evaluates the types of rear-end safety-critical events. Information was combined for curves and tangents due to the small sample size. Rear-end safety-critical events were more likely to be speeding-related than baseline events for two-lane roadways, and results were not statistically significant for the other roadway types.

Table 4: Speeding versus Not Speeding by Roadway Type for Rear-End Safety-critical Events
Baseline: 2-lane (open county, open residential, moderate residential, business/industrial) Crash/near crash Odds (CI)
curve (left or right)
not speeding 3254 166 3.43 (1.90, 6.18)
speeding 80 (2.4%) 14 (7.8%) 3.43 (1.90, 6.18)

Baseline: Divided and one way (open county, open residential, moderate residential, business/industrial) Crash/near crash Odds (CI)
not speeding 977 108 1.60 (0.46, 5.54)
speeding 17 (1.7%) 3 (2.7%) 1.60 (0.46, 5.54)

Baseline: Interstate/bypass/divided no signal Crash/near crash Odds (CI)
not speeding 2385 119 0.97 (0.44, 2.11)
speeding 145 (5.7%) 7 (5.6%) 0.97 (0.44, 2.11)

Baseline: Bypass/divided with signal Crash/near crash Odds (CI)
not speeding 56 17 NA
speeding 0 (0%) 0 (0%) NA

Baseline: Entrance exit ramps Crash/near crash Odds (CI)
not speeding 260 30 NA
speeding 3 (1.1%) 0 (0%) NA

Table 5 shows events which occurred at an intersection or were intersection-related where the traffic control type was a signal. They are additionally broken down by roadway type. As noted, safety-critical events at signalized intersections were more likely to be speeding-related than baseline events for signalized intersections on business/industrial/urban roadways and when all roadways were combined.

Table 5: Speeding versus Not Speeding at Signalized Intersections by Roadway Type
Baseline: Business/industrial/urban Crash/near crash Odds (CI)
not speeding 747 332 7.65 (1.36, 33.62)
speeding 2 (0.3%) 6 (1.8%) 7.65 (1.36, 33.62)

Baseline: Moderate residential/open country/open residential Crash/near crash Odds (CI)
not speeding 142 40 NA
speeding 0 (0%) 3 (7.0%) NA

Baseline: Divided highway with traffic signal Crash/near crash Odds (CI)
not speeding 38 15 NA
speeding 1 (2.6%) 0 (0%) NA

Baseline: All Crash/near crash Odds (CI)
not speeding 927 384 7.34 (1.95, 26.90)
speeding 3 (0.3%) 9 (2.3%) 7.34 (1.95, 26.90)

Table 6 shows events which occurred at an intersection or were intersection-related where the traffic control was a stop sign. They are additionally broken down by roadway type. It is unknown whether events which occurred on the major approach for a two-way, stop-controlled intersection were coded with traffic control as being "stop sign" or "no-control." Safety-critical events at stop signs on business/industrial/urban roadways and for all roadways combined were more likely to be speed-related.

Table 6: Speeding versus Not Speeding at Stop-controlled Intersections by Roadway Type
Baseline: Business/industrial/urban Crash/near crash Odds (CI)
not speeding 105 41 2.56 (0.16, 41.92)
speeding 1 (0.9%) 1 (2.4%) 2.56 (0.16, 41.92)

Baseline: Moderate residential/open country/open residential Crash/near crash Odds (CI)
not speeding 225 42 NA
speeding 0 (0%) 2 (4.5%) NA

Baseline: Divided highway with traffic signal Crash/near crash Odds (CI)
not speeding 0 83 NA
speeding 0 (0.3%) 3 (3.5%) NA

Baseline: All Crash/near crash Odds (CI)
not speeding 330 83 11.93 (1.22, 116.14)
speeding 1 (0.3%) 3 (3.5%) 11.93 (1.22, 116.14)

Table 7 shows events which occurred at an intersection or were intersection-related where the traffic control was a yield sign. They are additionally broken down by roadway type. No safety-critical events were speeding-related, so the odds could not be calculated.

Table 7: Speeding versus Not Speeding at Yield-controlled Intersections by Roadway Type
Baseline: All Crash/near crash Odds (CI)
not speeding 16 14 NA
speeding 2 (11.1%) 0 (0.0%) NA

6.3 TYPOLOGY OF SAFETY-CRITICAL EVENTS

In the analysis described in this section, each crash and near-crash were reviewed. Characteristics such as roadway type, crash type, etc. were compared against what was coded, and adjustments were made when a different set of characteristics seemed more appropriate.

Time series data are provided and speed before the beginning of each event was coded. As noted in section 2, speed limit was not provided as an attribute. In some cases speed limit or advisory speed was available from the narrative and in some cases it could be viewed from the forward view. However, it could not be determined in a sufficient sample size to conduct additional analyses. Although the posted/advisory speed was not universally available, the actual speed of the vehicle just prior to the safety-critical event was extracted from the time series data.

The sections below summarize safety-critical events by roadway type.

6.3.1 Non-intersection

Rear-end or near rear-end

Thirteen coded as business/industrial, moderate residential, open country, open residential:

Twenty-two coded as bypass/divided highway with traffic signal or interstate/bypass/divided no signals:

Roadway Departure (includes typical ROR, animal in roadway, head-on, sideswipe)

Seventy-five coded as business/industrial, moderate residential, open country, open residential:

Sixteen coded as bypass/divided highway with traffic signal or interstate/bypass/divided no signals:

Eleven coded as exit/entrance ramp:

6.3.2 Signalized Intersection

Rear-end or near rear-end

Fourteen coded as business/industrial, moderate residential, open residential, bypass/divided highway

Roadway Departure (includes typical ROR, loses control, sideswipe)

Twenty-two coded as business/industrial, moderate residential, open residential, bypass/divided highway:

6.3.3 Stop or Yield Controlled Intersection

Rear-end or near rear-end

None were coded as rear-end or near rear-end.

Roadway Departure (includes typical ROR, loses control, sideswipe)

Four coded as business/industrial, moderate residential, open residential, bypass/divided highway:

Broadside

One coded as near-broadside (nighttime, dry), where speed just before the event was 44 mph.

6.3.4 Roundabout/Traffic Circle

Rear-end or near rear-end

One coded as moderate residential (exceeded speed limit, dry roadway, daytime) at 19 mph before incident.

Roadway Departure (includes typical ROR, loses control, sideswipe)

Six coded as business/industrial, moderate residential, open residential:

6.3.5 Pedestrian

No speeding or speed-related events that involved pedestrians were reported. Since pedestrian and bicyclist safety is of particular concern, all pedestrian crashes and near crashes involving a subject driver were assessed and relevant characteristics summarized below.

Midblock Crosswalks

Six events involved pedestrians crossing in a midblock cross walk. The subject driver went around vehicles stopped at the cross walk and nearly struck pedestrians in two instances. Both were during the day, one on wet roads and one on dry. One driver was moving or reaching for an object and the other driver had no coded distractions. Both were traveling almost 20 mph.

Three of the six events involved drivers nearly striking pedestrians in midblock crosswalks (speeds were 23 to 34 mph). All three occurred during the day on dry roads. One driver was interacting with a passenger and one driver was looking down or engaged in an external distraction. One event occurred during the nighttime with the driver traveling at 27 mph along a business/industrial road. The crosswalk is midblock, has on pavement "Ped Xing" markings, flashing beacons, and a painted crosswalk. The driver was engaging with a passenger and engaged in an external distraction. The driver failed to noticed the pedestrian and braked hard to avoid striking the person.

Midblock but not Crosswalk Crossings

Eleven events involved a pedestrian crossing midblock but not in a cross walk. A child darted into the street in three of those events. All were residential/moderate residential with driver speeds between 17 and 19 mph. It was daytime for two events and dawn/dusk for one. The roads are dry for two events and wet for one. All three drivers were engaged in a distraction (personal hygiene, interacting with rear seat passenger, and using cell).

Eight of the 11 events involve a pedestrian crossing the street or entering the street in front of the driver and nearly being struck. In all cases the roads were dry or wet. Two drivers were texting, one was interacting with a passenger, and one driver was coded as drowsy. The roadways are urban, moderate residential, or business/ industrial.

Intersection Straight

Four pedestrians crossed perpendicular to the driver who was traveling straight during the green phase at a signal. In these cases, the driver had the right of way. Driver speeds just prior to the event ranged from 7 to 32 mph, and roadways were urban or business/residential. Two occurred at night with lighting and two occurred during the day with three coded as having dry roads and one with wet roads. One driver was engaged in personal hygiene and an external distraction, and one driver was talking or browsing on a cell phone.

Three events occurred at a stop-controlled intersection or driveway with the pedestrian crossing in front of the driver on the roadways in areas coded as being moderate residential. Speeds just before the event were 0 to 12 mph. Two events occurred during the day with one occurring at night with lights, and all were on dry roads. One driver was texting and another driver was talking on a cell phone.

Intersection Left Turning

Nine events involved a driver nearly striking a pedestrian while turning left, with the pedestrian crossing being parallel to the driver's original travel path. Five were at signals, two were at locations with stop control, and one occurred at the entry to a driveway. Two drivers were interacting with front seat passengers, one was dialing a cell phone, and one was looking down at a cell phone. The roadways were moderate resident, urban, or business/ industrial and speeds just prior to the event ranged from 6 to 18 mph. In six cases the roadways were mostly dry and in two they were wet.

Intersection Right Turning

Drivers were turning right and nearly struck a pedestrian in seven cases that occurred at urban, moderate residential, or business/industrial roadways. Four were at signals, two at stop controlled approaches, and one was turning into a driveway. The majority (6 cases) occurred during the day on dry roads with one event occurring at night with wet roads. One driver was talking/singing to self and engaged in an external distraction while the remaining drivers were not coded as having distractions.

6.3.6 Bicyclist

Only two speed-related events were reported. Since pedestrian and bicycle safety is of particular concern, all bicycle crashes/near crashes involving a subject driver were assessed and relevant characteristics summarized below.

Twenty safety-critical events involving bicyclists were reported.

Midblock but not Crosswalk Crossings

In one near crash, the subject driver exceeded the posted speed limit by 5 to 10 mph along an urban roadway. A bicyclist crossed the roadway midblock in an area with a cross-walk and railroad crossing. The driver nearly struck the bicyclist. Just prior to the event, the driver was traveling 39 mph. The driver was listed as "dancing."

Edge of Roadway

In another scenario the driver is listed as exceeding a safe speed but not the speed limit, driving 39 mph during the daytime. The driver nearly struck a bicyclist traveling along the edge of the roadway (listed as business/ industrial). One other event listed the driver nearly striking a bicyclist traveling on the shoulder along a rural roadway while traveling at 40 mph. In both situations, it was dark and bicyclist was difficult to see. In one event the driver was talking/singing to self, but no distraction is listed for the other event.

Intersection Straight

The remaining bicycle-related events occurred at intersections. Six of the events occurred at a signalized intersection when the bicyclist crossed the road perpendicular to the driver, who was traveling straight during the green phase with the driver nearly striking the bicyclist. In these cases, the driver had the right of way. Driver speeds just prior to the event ranged from 4 to 30 mph. Five occurred during the day with one occurring at dawn/ dusk. All were on dry roads. One driver was interacting with a passenger, one was coded as having a cognitive distraction, and one was coded as having "internal eye glance." No distractions are coded for the other events. Two events occurred at stop-controlled intersections with the bicyclist crossing the road perpendicular to the driver, who had the stop sign. One occurred at night with lighting present (traveling 7 mph prior to event) and one occurred during the day (traveling 14 mph). One driver was listed as talking/singing to self while the other has not reported distraction. In both cases the driver appears to execute a rolling stop.

Intersection Left

Six drivers were turning left at a signal (5 cases) or stop sign (1 case) and nearly struck bicyclists who were crossing the adjacent roadway (parallel to the drivers' original travel paths). Drivers were traveling 7 to 19 mph just prior to the event. Five occurred during the day, with one occurring at night with lighting. All were on dry roads. One driver was texting and one was looking at something outside the vehicle.

Intersection Right

Two drivers were turning right; one at a signal and one from a driveway, and nearly struck bicyclists who were crossing the adjacent roadway (parallel to the drivers' original travel paths'). The drivers were traveling 4 to 7 mph just prior to the event. Five occurred during the day with one occurring at night with lighting. All are on dry roads. One driver was holding a cell phone and talking, the other had no distraction.

Red Light Running

One driver ran a red light and nearly struck a bicyclist is in the intersection. The vehicle was traveling 58 mph prior to the event, which occurs in a business/industrial area during the daytime on dry roads. No driver distractions were coded.

6.4 STUDY LIMITATIONS

The analyses presented in the following sections were based on data from drivers observed in the SHRP 2 NDS and may not be representative of all drivers.

Data can only be queried on the InSight website unless a data-sharing agreement and additional resources are provided to download event information, which was beyond the scope of this project. As a result, analyses were based on those that could be obtained from manual reduction of information or by query of the available attributes.

The major limitation to the analyses is that speed limit was not available for individual events which limited the ability to assess speeding. With additional resources, speed limit could be obtained by requesting GPS coordinates for baselines and near-crash events.

Although there were some limitations due to project resources and the type of data there were available, these analysis provide valuable insight into the role of speeding in safety-critical events.

6.5 RECOMMENDATIONS AND SUGGESTIONS FOR FUTURE RESEARCH

In order to fully utilize the SHPR 2 NDS data to investigate the relationship between speed and crash risk, several recommendations are offered.

First, the major limitation to this study was the lack of speed limit information. As a result, it would be very helpful if speed limit and advisory speed (when present) were coded for all crashes and near-crashes and were provided as attributes in the event detail table. This would allow speeding to be determined.

Second, although baseline events were provided in the data available on the InSight website, speed limit data were not provided, and little information is available about where the event actually occurred. Additionally, as noted in Section 6.1, roadway types used in the InSight website do not correspond well to traditional definitions used by traffic engineers. Since a forward view was not provided, there was no manner in which roadway type could be coded. As a result, it is difficult to compare safety-critical events to normal events in a meaningful manner. Providing speed limits for the baseline and a forward snapshot showing roadway location would be very helpful in conducting speed related analyses.

The SHRP 2 NDS dataset does provide a unique way to evaluate the impact of speeding. If speed limit for safety-critical events could be obtained, a reasonably simple analysis could be conducted to identify the role of speeding in crashes and near crashes using the "Event Detail Table" in the InSight website.

If additional resources were available, an ideal research project would identify roadways where safety-critical events had occurred and then select normal driving events for the same roadways so that the relationship between speeding and crash risk could be determined.

Page last modified on April 29, 2016
Safe Roads for a Safer Future - Investment in roadway safety saves lives
Federal Highway Administration | 1200 New Jersey Avenue, SE | Washington, DC 20590 | 202-366-4000