3.2 Analyses of five State files: results
NHTSA's earlier evaluations both found that CHMSL initially may have been more effective in collisions involving 3 or more vehicles than in 2 vehicle collisions. The preliminary report, based on CY 1986 data, said that CHMSL reduced the probability of being the lead vehicle in a 3+ vehicle collision by 25.7 percent; of being one of the middle vehicles in such a collision by 17.5 percent; but CHMSL only reduced the probability of being the struck vehicle in a front-to-rear 2 vehicle collision by 13.2 percent [18], pp. 14-18. (However, the sample sizes were small, and those observed differences in effectiveness were not statistically significant.) In NHTSA's evaluation based on CY 1987 data, CHMSL reduced the probability of being a rear-impacted car in a 3+ vehicle collision by 14.6 percent, but only reduced the probability of being the rear-impacted car in a 2 vehicle collision by 9.8 percent; in 9 of the 11 State files analyzed, effectiveness was higher in the 3+ vehicle collisions [19], pp. 28-29. It was hypothesized that "CHMSL are especially effective in preventing chain collisions because they enable a driver to see if the car two vehicles ahead is braking" [19], p. 29.
In the long term, these effects are reversed:
Reduction (%) of Rear Impacts with CHMSL
In 2 Vehicle
Collisions |
In 3+ Vehicle
Collisions |
|
Florida | 6.51 | 2.34 |
Maryland | 3.41 | 2.54 |
Missouri | 5.20 | 3.80 |
Pennsylvania | 5.09 | 2.16 |
Utah | 8.19 | 1.06 |
5-State weighted average | 5.54 | 2.45 |
In the CY 1989-95 data analyzed in this chapter, the effectiveness was higher in 2 vehicle than in 3+ vehicle crashes in each of the five States. In Florida, Pennsylvania and Utah, it was more than double, while in Maryland and Missouri it was slightly higher in the 2 vehicle crashes. The five-State, population-weighted averages: CHMSL reduces the risk of being rear-impacted by 5.54 percent in 2 vehicle collisions and 2.45 percent in 3+ vehicle collisions.
What might have happened between CY 1986-87 and CY 1989-95? It was shown in Chapter 2 that the overall effectiveness of CHMSL dropped in half during those years. That phenomenon has been described as "acclimatization" [17] or a "novelty effect" [7], [9], [31]. In other words, as CHMSL became more common in vehicles on the road, drivers became more accustomed to seeing the lamps. They might "tune out" rather than being spurred to action by the lamps. It could be argued that the acclimatization will be stronger if the precrash situation is more complex. When the vehicle with CHMSL is directly in front of you, it is hard not to see the lamps, but when it is two vehicles ahead of you it becomes easier to "tune them out." Another factor that may have reduced effectiveness in some 3+ vehicle collisions is that the increased popularity of tall vans, sport utility vehicles, and heavily tinted glass may be making it harder for drivers to see through the next vehicle to the one ahead of it.
NHTSA's previous evaluation presented strong evidence that, during 1987, CHMSL were more effective in daylight (13.2 percent) than under reduced lighting conditions (5.6 percent) [19], pp. 34-35. The trend has continued during 1989-95. In our five State files, "daylight" and "reduced light" (night, dawn or dusk) are defined as follows:
Daylight | Reduced Light | |
Florida | light=1 | light=2-5 |
Maryland 1989-92 | light=1 | light=2-6 |
Maryland 1993-95 | light=1 | light=2-4 |
Missouri | light=1 | light=2-4 |
Pennsylvania | light=2 | light=1,3,4,5 |
Utah | light=1 | light=2-5 |
In the five States, CHMSL are, on the average, twice as effective in daylight as at other times, and they are more effective in daylight in each individual State except Utah:
Reduction (%) of Rear Impacts with CHMSL
Daylight | Reduced Light | |
Florida | 5.31 | 3.44 |
Maryland | 3.99 | 1.38 |
Missouri | 5.40 | - 1.21 |
Pennsylvania | 3.82 | 2.33 |
Utah | 6.65 | 7.40 |
5-State weighted average | 4.73 | 2.34 |
NHTSA's earlier evaluation hypothesized that "the driver's field of view is less cluttered by lamps during daylight: CHMSL and the triangle they make with the regular stoplamps are the only lights in front of the driver. At night, street lights, headlamps and taillamps of other vehicles may distract the driver. A second possibility is that regular stoplamps are hard to see during the daytime, because they are not that much brighter than ambient surfaces; thus, the CHMSL (directly in front of the driver) provides extra help by day. Another factor, undoubtedly, is that a substantial percentage of nighttime drivers are insufficiently alert or defensive because of alcohol or other factors; they don't react to the CHMSL at all or not in time" [19], pp. 34-35. Those hypotheses still sound reasonable.
CHMSL are less effective near traffic signals (including traffic lights, stop signs, flashing lights, yield signs and railroad crossings, but excluding lane markings, advisory signs, speed limit signs, etc.) than at locations away from signals. That was a tentative finding in NHTSA's previous evaluation [19], pp. 35-36, and it is strongly confirmed by the present data. The presence of a traffic signal is defined as follows:
At Signal | Away from Signal | |
Florida | tra_con=3-8 | tra_con=1,2,9 |
Maryland 1989-92 | tra_con=1-4,10,11 | other |
Maryland 1993-95 | tra_con1=Y | tra_con1=N |
Missouri | trfcntl1=1-5 | trfcntl1=6-9 |
Pennsylvania | tra_con=1-6 | tra_con=0,7 |
Utah | tra_con1=1-7 | tra_con1=9,A,B,C,D,E |
In the five States, CHMSL are more effective away from traffic signals:
Reduction (%) of Rear Impacts with CHMSL
At Signal | Away from Signal | |
Florida | 1.88 | 6.66 |
Maryland | 2.28 | 4.49 |
Missouri | .23 | 6.19 |
Pennsylvania | 1.63 | 5.76 |
Utah | 4.70 | 7.82 |
5-State weighted average | 1.75 | 6.07 |
Two hypotheses help explain the results. On the one hand, the signal diverts the driver's attention from the CHMSL of the vehicle directly to the front. Other traffic may also be diverting the driver's attention (since signals are usually near an intersections). On the other hand, the signal itself is sending a message to the driver to be cautious or slow down, reinforcing information that might be sent by the CHMSL. But on the open road, away from signals, the driver is more likely to focus straight ahead, where the next car's CHMSL is.
NHTSA's earlier evaluation suggested that CHMSL were slightly more effective in rural than urban areas [19], pp. 36-37. That trend continues in the three larger States (Florida, Missouri, Pennsylvania). The rural/urban definitions are:
Rural | Urban | |
Florida | rur_urb=1 | other |
Maryland 1989-92 | locality=5 | locality=1-4 |
Maryland 1993-95 | N/A | N/A |
Missouri | pop_grp=12,99 | other |
Pennsylvania | rur_urb=1 | other |
Utah | locality=5,6 | locality=1-4,7-9 |
The Maryland analysis is flawed because rural/urban is not defined in the 1993-95 files. In Utah, the preceding definition makes most crashes "urban." The effectiveness estimates are:
Reduction (%) of Rear Impacts with CHMSL
Rural | Urban | |
Florida | 6.33 | 3.88 |
Maryland (CY 89-92) | - 6.60 | 4.03 |
Missouri | 10.56 | 2.26 |
Pennsylvania | 5.09 | 3.06 |
Utah | 2.12 | 7.30 |
5-State weighted average | 4.62 | 3.57 |
The 5-State average effectiveness is slightly higher in rural than in urban areas, but the difference is in the "noise" range. Rural effectiveness is higher, though, in the three larger States. The working hypothesis here is that rural roads will have fewer diversions (e.g., multilane traffic, parked cars, city lights) of the driver's attention from the CHMSL.
The four preceding analyses show a consistent pattern for CHMSL during CY 1989-95: the simpler the situation, and the fewer the diversions of the driver's attention from the vehicle directly in front, the more powerful the effect of the CHMSL.
CHMSL may be more effective on wet roads than dry roads (although the results are short of conclusive). The codes for road surface condition are:
Dry | Wet | Snowy/Icy | |
Florida | rd_sur1=1 | rd_sur1=2,3,4 | N/A |
Maryland | rd_sur1=2 | rd_sur1=1,5 | rd_sur1=3,4 |
Missouri | rd_sur1=1 | rd_sur1=2,5 | rd_sur1=3,4 |
Pennsylvania | rd_sur1=1 | rd_sur1=2,3 | rd_sur1=4-8 |
Utah | rd_sur1=1 | rd_sur1=2,3,6 | rd_sur1=4,5 |
Florida actually has codes for "snowy" and "icy" but they are too infrequent for a meaningful statistical analysis. The effectiveness estimates are:
Reduction (%) of Rear Impacts with CHMSL
Dry | Wet | Snowy/Icy | |
Florida | 4.10 | 7.51 | . |
Maryland | 5.23 | - .55 | - 1.51 |
Missouri | 3.99 | 5.95 | - 8.29 |
Pennsylvania | 2.17 | 6.15 | 8.68 |
Utah | 6.78 | 7.47 | 2.14 |
5-State weighted average | 3.73 | 5.79 | 2.42 |
The observed effectiveness is higher on wet roads than on dry roads in every State except Maryland. (Even outside Florida, the sample size for snowy and icy roads is too small for a reliable effectiveness estimate.) If CHMSL are indeed more effective on wet roads than on dry, snowy or icy roads, the following explanations may be considered: (1) On dry, snowy or icy roads drivers maintain a distance from the car in front of them commensurate with their stopping capabilities. On wet roads, even if they follow less closely than on dry roads, they do not add distance commensurate with their reduced braking capability. Since the margin of safety is smaller, the reduction in response time with CHMSL provides an extra "edge." (2) Rain, glare, foggy windshields, etc. may make it harder to see conventional brake lights on wet roads, enhancing the advantage of CHMSL.
The age or sex of the driver of the rear-impacted vehicle ought to have little direct influence on the effectiveness of CHMSL, since the actions of this driver are, for the most part, not directly responsible for causing the collision. But the age or sex of the other driver, the one in the frontally impacting vehicle in a front-to-rear collision, could be relevant. On the one hand, CHMSL might be more effective when the other driver is young and/or male: since young/male drivers tend to be more aggressive, and have less of a safety margin in their following distance, they might obtain an extra "edge" from the improved response time with CHMSL. On the other hand, older and/or female drivers, because of their shorter seated height, might have more trouble seeing conventional brake lights and get an extra advantage from CHMSL. Additionally, older drivers have generally slower response and reaction times and might be helped more than others by the visual cues from CHMSL. The five State files, however, show little difference in CHMSL effectiveness by age or gender of the other driver. In other words, the factors suggested above are negligible or cancel one another out. (The analyses are limited to 2 vehicle crashes, since it is often not easy to determine the "striking" vehicle in a 3+ vehicle crash.)
Reduction (%) of Rear Impacts with CHMSL
Other Driver
Age 15-24 |
Other Driver
Age 25-54 |
Other Driver
Age 55+ |
|
Florida | 7.29 | 4.98 | 10.88 |
Maryland | - 4.14 | 7.32 | 5.73 |
Missouri | 5.49 | 5.99 | 7.78 |
Pennsylvania | 7.99 | 5.63 | - .41 |
Utah | 8.97 | 7.05 | 10.45 |
5-State weighted average | 5.84 | 5.74 | 6.08 |
(Note that the effectiveness within each age group is just slightly higher than the average effectiveness estimate in all 2-vehicle crashes, 5.54 percent. The other driver's age was unreported in a moderate proportion of these crashes, and in that group [not shown above] the observed effectiveness estimate happened to be lower than average.)
Reduction (%) of Rear Impacts with CHMSL
Other Driver
Male |
Other Driver
Female |
|
Florida | 6.60 | 6.36 |
Maryland (CY 89-92) | 5.60 | 1.20 |
Missouri | 6.09 | 6.35 |
Pennsylvania | 4.70 | 4.04 |
Utah | 8.79 | 7.54 |
5-State weighted average | 5.88 | 4.98 |
NHTSA's evaluation of 1987 data found that CHMSL were at least as effective in towaway as nontowaway crashes and probably more effective in preventing injury crashes than property-damage-only (PDO) crashes [19], pp. 25, 26, 28. Those were encouraging results for anyone who feared that CHMSL might only be effective in the lowest severity crashes. The 1989-95 data more or less continue those trends, showing perhaps slightly higher effectiveness in towaways than nontowaways and nearly equal effectiveness in PDO and injury crashes. Crash damage severity is defined in every State except Utah; the codes are:
Towaway | Nontowaway | |
Florida | sev_dam=1 | other |
Maryland 1989-92 | dam_sev=1 | other |
Maryland 1993-95 | dam_sev=4,5 | other |
Missouri | towaway=Y | other |
Pennsylvania | towaway=1 | other |
In Florida, "towaway" means any vehicle involved in the crash was towed; in the other three States, "towaway" means the rear-impacted car was towed. The effectiveness estimates are:
Reduction (%) of Rear Impacts with CHMSL
Towaway | Nontowaway | |
Florida | 6.84 | 3.63 |
Maryland | 2.93 | 3.07 |
Missouri | 5.32 | 3.55 |
Pennsylvania | 2.81 | 2.03 |
4-State weighted average | 4.70 | 3.00 |
Effectiveness is higher in towaway crashes in the three larger States (Florida, Missouri, Pennsylvania) and about the same in Maryland. The 4-State weighted average is somewhat higher in towaway crashes.
The definitions of an "injury crash" and a "property-damage-only" crash are:
Injury Crash | Property Damage Only | |
Florida | severity=2-6 | other |
Maryland | severity=2-5 | severity=1 |
Missouri | severity=1,2 | severity=3 |
Pennsylvania | severity=1-4 | severity=6 |
Utah | severity=2-5 | severity=1 |
In all five States, "injury" crashes are those in which any occupant is injured or killed, an occupant of the rear-impacted vehicle and/or an occupant of any other vehicle. In a PDO crash, no occupant of any vehicle is injured or killed. The effectiveness estimates are:
Reduction (%) of Rear Impacts with CHMSL
Injury Crashes | Property Damage Only | |
Florida | 4.82 | 4.65 |
Maryland | 4.65 | 1.49 |
Missouri | 4.76 | 3.77 |
Pennsylvania | 2.10 | 4.89 |
Utah | 7.09 | 6.42 |
5-State weighted average | 4.01 | 4.27 |
In four States, effectiveness is higher in preventing injury crashes, but in populous Pennsylvania, the situation is reversed. The 5-State average effectiveness is virtually identical in PDO and injury crashes. It may be concluded that CHMSL are effective across a fairly wide spectrum of crash severities. Intuitively, that makes sense. If CHMSL provide an extra "edge" of quicker response, they ought to reduce injury crashes to PDO crashes, and PDO crashes to fender benders just as well as they reduce fender benders to "close calls."