Effects of daytime running lights on multiple-vehicle daylight crashes in the United States

https://doi.org/10.1016/S0001-4575(01)00013-6Get rights and content

Abstract

Involvements in multiple-vehicle daylight crashes in nine states over 4 years were analyzed for a group of passenger cars and light trucks equipped with automatic daytime running lights. On average, these vehicles were involved in 3.2% fewer multiple-vehicle crashes than vehicles without daytime running lights (P=0.0074).

Introduction

The failure of drivers to either notice or to accurately judge the relative positions of other road users in sufficient time is a contributing factor to many motor vehicle crashes. Countermeasures to prevent this that have been shown to be effective at night are rear and side lighting for vehicles and reflectorization for all road users (Henderson et al., 1983). However, increased vehicle conspicuity can also be effective during the day. Daytime headlight use on motorcycles has been associated with crash reductions of 7–18% (Janoff et al., 1970, Waller and Griffin, 1977, Zador, 1985). Also, a number of states require the daytime use of headlights on all motor vehicles when it is raining or during periods of poor visibility.

Scandinavian countries were the first to require daytime use of headlights for all motor vehicles. Finland mandated winter daytime headlight use on rural roads beginning in 1972. In 1977, Sweden began requiring daytime headlight use during all seasons and on all roads. Norway was the first to require automatic daytime running lights (DRLs), switched on by the ignition, on all new cars beginning in 1985. Evaluations of each of these laws reported reductions in multiple-vehicle crashes, but none of the reductions were statistically significant (Andersson et al., 1976, Andersson and Nilsson, 1981, Elvik, 1993). The methodology of these evaluations has been criticized (Koornstra, 1993, Theeuwes and Riemersma, 1995). Also, it has been suggested that the crash reduction effect may be limited to twilight periods, which are longer at more northern latitudes. Thus there have been some questions as to the potential effectiveness of daytime headlight use in other countries, especially those in southern latitudes.

Fleet studies of passenger vehicles equipped with DRLs have reported significant crash reductions in North America. Crash rates during 1983–84 were compared for 2100 DRL-equipped vehicles and 2200 standard vehicles in corporate fleets operating throughout the United States (Stein, 1985). The DRL-equipped vehicles had 7% fewer daytime multiple-vehicle crashes. In a later study of a government fleet in Canada, there was an estimated 15% reduction in daytime two-vehicle crashes after fitting vehicles with DRLs (Sparks et al., 1993). The effect of DRLs was statistically significant, however, only for crashes occurring during twilight periods.

Canada has required DRLs on all new cars sold since 1989. Arora et al. (1994), using methods similar to those of the Scandinavian studies, reported a statistically significant 8% reduction in two-vehicle, opposing direction, daytime collisions coincident with the legislation. A later evaluation of the Canadian legislation reported a statistically significant 5% reduction in daytime angle and opposing collisions (Tofflemire and Whitehead, 1997).

Williams and Lancaster (1995) summarized the results of 11 studies of DRL effectiveness. They concluded that, although individual studies were open to criticism, daytime running lights were generally associated with small to moderate reductions in multiple-vehicle daytime crashes. Elvik (1996) went even further, combining the results of 17 studies from nine countries in a carefully controlled meta-analysis. He concluded that DRL use reduces the number of multiparty daytime crashes by 10–15%, slightly lower at lower latitudes.

Automatic daytime running lights are not required in the United States, but, when included, they must meet certain restrictions. The National Highway Traffic Safety Administration (NHTSA, 1993) published a rule in 1993 restricting the output of DRLs to a maximum of 7000 candela, approximately one-tenth the intensity of a standard high beam headlamp. This limit is equivalent to the maximum output allowed in Canada, but is more than four times the maximum output allowed in most European countries. Concerns about headlamp glare from both oncoming and following DRL-equipped vehicles have since prompted NHTSA to propose lowering the maximum output to 1500 candela.

Volvo and Saab began installing DRLs on all their new cars sold in the United States in 1995. Volkswagen, Suzuki, and General Motors put DRLs on selected models beginning the same year. Daytime running lights became standard equipment on all new Volkswagens and Suzukis and a number of high-volume General Motors vehicles in 1996. By 1997, DRLs were standard equipment on all new General Motors vehicles.

Researchers at NHTSA, using both nationwide data on fatal crashes and data on nonfatal crashes from four states, evaluated the effects of DRLs over the first few years of US experience (NHTSA, 2000). A number of different methodologies were examined. Depending on the methodology used and the source of data, the estimated effect of DRLs on daytime two-vehicle crashes ranged from a 25% decline to a 27% increase. Most of these individual estimates were not statistically significant. When combined, however, the individual estimates yielded a statistically significant 7% decline.

The NHTSA combined effectiveness estimate has two major faults. First, the individual estimates are required to be statistically independent, but every estimate has at least one companion based on the same set of data (and a different methodology). Second, there is no guarantee that estimates based on different methodologies are measuring the same quantity. Therefore, the NHTSA estimate of a 7% reduction in daytime two-vehicle crashes, while consistent with earlier studies, is not sufficiently justified.

The purpose of the present study was to obtain more precise estimates of the effects of DRLs on multiple-vehicle daylight crashes by examining crashes through 1998 in nine states. The same methodology was used for all states, allowing for the combination of estimates.

Section snippets

Method

Table 1 lists vehicle models that were not equipped with DRLs in model year 1994, but had DRLs as standard equipment beginning in model year 1995. Other than the addition of DRLs, there were no significant design differences between the two model years. There were 765 611 reported sales of these 1995 models, accounting for approximately 5% of new vehicle sales. Table 2 lists vehicle models that were not equipped with DRLs in model year 1995, but had DRLs as standard equipment beginning in model

Results

Multiple-vehicle crash involvements of all 1994–95 model passenger vehicles in the nine study states during 1995–98 are summarized in Table 3. In Florida, for example, vehicles in study group 1 accounted for 11 017 of the 266 208 multiple-vehicle crash involvements. The ratio of daylight to darkness crashes was 3.44 for model year 1994 (without DRLs) and 3.49 for model year 1995 (with DRLs). Thus the odds of a daylight crash were actually slightly higher for the vehicles with DRLs. Overall,

Discussion

Passenger vehicles with DRLs were involved in approximately 3.2% fewer multiple-vehicle daylight crashes than similar vehicles without DRLs. There is some indication that the effect may vary across states, but the data are insufficient at this point to make such a determination. In particular, New Mexico was the only state where DRLs did not indicate a positive effect. Further study is necessary to determine whether this result is due to random variation or some unique characteristic of the

Acknowledgements

The authors wish to acknowledge Laurie Nelson and Laura Ross for gathering and summarizing detailed information on the lighting systems of the study vehicles. This work was supported by the Insurance Institute for Highway Safety.

References (21)

There are more references available in the full text version of this article.

Cited by (0)

View full text