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.
2 Effects of Speed The minutes Some folks Save through speed They never even Live to need Burma Shave (Rowsome 1965) The major reason for managing traffic speeds is safety. In this chap- ter, what is known about the relationships among speed, crash inci- dence, and crash severity is reviewed. Individual driver decisions about appropriate travel speeds, however, are guided by more than safety considerations. Thus, the relationship of speed to travel time, fuel use, and other vehicle operating costs is also examined. In addi- tion, driver decisions about speed affect other costs, such as vehicle emissions, which contribute to air pollution in metropolitan areas and to atmospheric changes that may increase the risk of global cli- mate change. These costs, which are briefly reviewed, are borne 36
37 Effects of Speed largely by society as a whole rather than by individual drivers, at least in the United States. The chapter concludes with an assessment of the effects of speed on safety, travel time, and other related costs, and their implications for managing speed. DETERMINATION OF APPROPRIATE DRIVING SPEEDS--MAKING TRADE-OFFS How do people decide how fast to drive? Many factors come into play including the characteristics of the road; the amount of traffic on the road; weather conditions and time of day; the speed limit and its enforcement; the length and purpose of the trip; the vehicle's operat- ing characteristics, such as handling and stopping as well as fuel con- sumption and emissions; and driver-related factors, such as the propensity to take risks and the pleasure associated with driving fast. Taking these and other factors into consideration, drivers face an important trade-off between travel time and safety. By driving faster, travel time is reduced and the destination is reached sooner if the trip is safely completed. However, as discussed later in this chapter, a driver who chooses to drive very fast relative to other traffic or very fast for existing road conditions may increase the probability of being involved in a crash as well as the severity of the crash. A driver can reduce crash probability and severity by driving more slowly, although driving too slowly relative to other traffic may also increase the probability of crash involvement. The theory underlying the trade-off between travel time and safety is discussed in more detail in Appendix A. Conceptually the trade-off is straightforward, but practically one could question whether drivers really trade off safety and travel time when making their trips. Some drivers indicate that this trade-off is not foremost in their mind while traveling; others claim that they are not conscious of making this trade-off. For some drivers in many situations, the choice of driving speed is heavily influenced by speed limits and their enforcement so that the trade-off is, in a sense, made for them. But even in situations where there is little or no speed limit enforcement and many drivers exceed the posted speed limit, few motorists will drive as fast as their vehi-
MANAGING SPEED 38 cles are capable of going. Something other than the fear of speed limit enforcement causes drivers to drive at less than the maximum possible speed. Similarly, when weather conditions such as fog, rain, or snow cause visibility to deteriorate and traction to be reduced, drivers may slow down, often to speeds well below the posted limits. For many drivers faced with these conditions, their choice of a lower speed and increased travel time is almost certainly made with safety in mind. There is reason to believe, therefore, that where speed choice is not constrained by speed limits and their enforcement, the driver does trade off travel time and safety. Even when visibility and weather conditions are good, drivers may still make trade-offs. Rather than making them continuously, how- ever, they may rely on rules of thumb based on driving experience. For example, motorists may well rely on experience with particular roads or types of roads to select a driving speed that has proven to be a reasonable trade-off for them in the past. Only when they encounter new conditions or conditions they face infrequently would they be conscious of explicitly making such a trade-off. In this chap- ter what is known about the key factors affecting drivers' choice of speeds is reviewed. RELATION OF SPEED TO SAFETY The relation of driving speed to safety is investigated first because of the importance that most drivers place on completing their trips safely. The link between speed and safety is complex. Thus an in- depth review of the literature on this topic was commissioned to help shed light on the relationship of speed to crash causation and injury severity. The results of that review, which can be found in its entirety as Appendix B, are summarized in the following sections.1 1 These sections also draw on a second review, discussed more extensively in Chapter 3 and presented in its entirety as Appendix C.
39 Effects of Speed Speed and the Probability of Crash Involvement One of the more widely cited sources of statistics on speed and crashes is the Fatal Analysis Reporting System (FARS) administered by the National Highway Traffic Safety Administration (NHTSA), the federal agency charged with regulating automotive safety. In 1996 NHTSA reported that speeding was a contributing factor in 30 per- cent of all fatal crashes on U.S. highways in that year (NHTSA 1997a, 1). In addition to the 13,000 lives lost in these speeding- related crashes, 41,000 people were reported critically injured at an estimated economic cost to society of nearly $29 billion (NHTSA 1997a, 1).2 Thus speeding is singled out as "one of the most preva- lent factors contributing to traffic crashes" (NHTSA 1997a, 1). These figures must be interpreted with caution. The definition of speeding is broad; for the purposes of coding crash-related informa- tion, speeding is defined as "exceeding the posted speed limit or driv- ing too fast for conditions" (NHTSA 1997a, 1). The determination of whether speeding was involved in a fatal crash is based on the judgment of the investigating police officer; fatal crashes receive a thorough investigation.3 Even if speeding is listed as a contributing factor in a crash, it may not have been the primary cause. Furthermore, and perhaps most important, without knowledge of the incidence of speeding in the driving population, the fatal crash data 2 Economic costs include productivity losses, property damage, medical costs, rehabil- itation costs, travel delay, legal and court costs, emergency service costs, insurance administration costs, premature funeral costs, and costs to employers. They do not include any estimate of the value of lost quality of life associated with deaths and injuries, that is, what society is willing to pay to prevent them. 3 To ensure reporting consistency, FARS analysts, who are state employees contracted and trained by NHTSA, retrieve information about the crash from the police crash report and other sources and put it in a standardized coding format. For each crash, information is recorded at four levels--by crash, vehicle, driver, and person. Speed appears in two places--(a) on the crash-level coding sheet where the speed limit is recorded, and (b) on the driver-level coding sheet where speed-related violations are recorded. Typical violations, noted in the 1996 FARS Coding Manual, include driving at a speed greater than reasonable or prudent or in excess of the posted maximum, tow- ing a house trailer at more than 45 mph (72 km/h), or driving too slowly so as to impede traffic.
MANAGING SPEED 40 cannot be properly interpreted. For example, a recent study suggests that driver compliance with posted speed limits is poor, particularly for limits less than 45 mph (72 km/h) on nonlimited-access highways (Parker 1997, 43). The proportion of those driving above the posted speed limit--hence "speeding" by NHTSA's definition--typically exceeds the share of speeding drivers (approximately 20 percent according to FARS) involved in fatal crashes.4 The literature review attempts to examine the evidence that speeding is linked to the prob- ability of being involved in a crash. Theoretical Issues At least three theoretical approaches link speed with crash involve- ment: (a) the information processing approach, (b) the traffic conflict approach, and (c) the risk-homeostasis motivational approach. The first approach views the driver as an information processor with a limited capacity to process information. As driving speed increases, the rate at which the driver must process information about the highway and its environment increases directly, even though the total amount of information the driver has to process may stay con- stant. At higher speeds there is less time for the driver to process information, decide, and act between the time the information is pre- sented to the driver (e.g., a child is running into the road) and the time when action must be taken to avoid a crash.5 A crash is likely to occur when the information processing demands exceed the atten- tional or information processing capabilities of the driver (Shinar 1978).6 Unexpected events dramatically increase information pro- 4 Note that the 20 percent figure refers to the share of drivers involved in speeding- related fatal crashes as a percentage of drivers involved in all fatal crashes, whereas the 30 percent figure cited earlier refers to the share of speeding-related fatal crashes as a percentage of all fatal crashes. 5 More specifically, as speed increases, the distance covered during the driver's percep- tion-reaction time and the minimum distance required for braking both increase. For a vehicle on a level roadway, minimum braking distance increases with the square of the speed (see glossary definition of braking distance). 6 Although drivers can increase their level of attention and concentration with increasing speed, a heightened level of attention cannot be maintained for long periods because it is fatiguing.
41 Effects of Speed cessing requirements and hence the probability of a crash. This approach leads to the conclusion that "speed kills"; as more drivers increase their speed, the probability of information overload increases along with the potential for crashes. The second approach--the traffic conflict approach--assumes that crash probability is related to the potential for conflict among vehicles traveling in traffic. More specifically, the probability of an individual driver being involved in a multiple-vehicle crash increases as a function of the deviation of that individual driver's speed from the speeds of other drivers. Drivers with speeds much faster or much slower than the median traffic speed are likely to encounter more conflicts (Hauer 1971).7 This relationship leads to the conclusion that "speed deviation kills" and the prediction that on roads with equivalent average traffic speeds, crash rates will be higher on roads with wider ranges of speed. The theory, as formulated, relates only to two-lane rural roads (Hauer 1971, 1). A third approach--the risk-homeostasis motivational approach-- looks at speed and crash involvement from the perspective of driver perception of risk. From this point of view, drivers are neither passive information processors nor reactors to potential traffic conflicts. Rather they adjust their speed according to the risks they perceive (Taylor 1964) to maintain a subjectively acceptable level of risk (Wilde et al. 1985).8 The issue is not the link between speed and crash probability but between actual and perceived risk. Thus, driving 7 The number of conflicts between vehicle pairs is represented by the number of pass- ing maneuvers. The number of passing maneuvers a driver must make increases with his driving speed; the number of times a driver is passed by other vehicles increases as he reduces speed. Hauer (1971) showed that the distributions of the two functions (i.e., the number of times passing and the number of times being passed) have a min- imum at the median traffic speed. The findings relate only to rural roads between intersections (Hauer 1971, 1). 8 There is mixed empirical support for Wilde's risk-homeostasis theory. For example, Mackay (1985) found that British drivers of newer and heavier cars drove faster than drivers of older and lighter cars (with the exception of sport cars), and Rumar et al. (1976) found that drivers with studded tires drove faster than those without such tires on curves in icy but not in dry conditions. O'Day and Flora (1982), however, found that restrained occupants actually had lower impact speeds in tow-away crashes than unrestrained occupants, suggesting that drivers have different risk tolerances.
MANAGING SPEED 42 at high speeds per se is not dangerous. Rather, the danger comes from driving at a speed inappropriate for conditions, stemming from a misperception of the situational demands or a misestimation of the vehicle's handling capabilities or the driver's skills. This approach would predict that, under most circumstances, drivers who increase speed do not necessarily increase the risk of their crash involvement. Review of Empirical Data Several studies reviewed in this section (Table 2-1), many dating back to the 1960s, have tested the theories about the relationship between driving speed and crash involvement by analyzing actual vehicle speeds and crash data on different classes of roads. Speed is defined in several ways in these studies. It can relate to the speed of a single vehicle or to the distribution of speeds in a traffic stream. In the former case, the term speed deviation is used when referring to the deviation of an individual driver's speed from the average speed of traffic. In the latter case, when referring to the distribution of speeds in a traffic stream, three measures of speed are typically con- sidered: the average speed, the 85th percentile of the speed distribu- tion, and the dispersion in travel speeds. Speed dispersion, in turn, can be quantified by the variance, standard deviation, 10-mph pace, or range (high minus low) of a sample of speed measurements.9 In many studies, the standard deviation is approximated as the 85th per- centile speed less the average speed.10 With several measures of speed, interpreting the results of these studies is often difficult. Validity of the speed measures can also be a problem. For example, it is nearly impossible to obtain a reliable mea- sure of true precrash speeds for crash-involved vehicles because crashes are not planned events. Thus, precrash speeds must be esti- mated, but there is no way of validating their accuracy. In attempting to isolate the effect of speed, many studies assume that everything else remains equal. Of course, crash occurrence and injury severity are 9See definitions in glossary. 10The 85th percentile minus the average speed roughly corresponds to one standard deviation (S), which is the positive square root of the variance (S2).
Table 2-1 Selected Studies of the Relationship Between Speed and Crash Probability Authorship and Date of Road Class and Study Speed Limit Levels Analysis Major Findings Solomon Compared speeds of crash- Found U-shaped relationship between crash involvement and Main rural roads, U.S.; (1964) involved vehicles with speeds travel speeds. Lowest crash involvement rates at speeds three-fourths were two- of non-crash-involved vehi- slightly above average travel speeds. Highest crash involve- lane rural roads with cles ment rates at speeds well above and well below average traf- speed limits of 55 to fic speeds 70 mph (89 to 113 km/h) on 28 out of 35 sections Cirillo Compared speeds of crash- Same finding as Solomon, but crash involvement rates were Rural and urban Interstate (1968) involved vehicles with speeds lower for all travel speeds suggesting importance of roadway highways, U.S.; no of non-crash-involved vehi- geometry to crash probability (i.e., higher design standards speed limits given cles; limited to daytime on Interstate highways than on rural two-lane roads) travel and certain multiple- vehicle crash types (i.e., rear- end and angle collisions and same-direction sideswipe crashes) RTI Compared speeds of crash- Found same U-shaped relationship between travel speed and State and county high- (1970); involved vehicles with speeds crash involvement, but the relationship was less extreme, ways in Indiana with West and of non-crash-involved vehi- particularly at low speeds, when crashes involving turning speed limits greater Dunn cles; separated out crashes vehicles were removed from the analysis than or equal to 40 (1971) involving turning vehicles mph (64 km/h) (continued on next page)
Table 2-1 (continued) Authorship and Date of Road Class and Study Speed Limit Levels Analysis Major Findings Lave Six U.S. highway types-- Analyzed relationship between Speed dispersion significantly related to fatality rates for rural (1985) rural and urban average traffic speed, speed Interstates and rural and urban arterials. After controlling Interstates, arterials, and dispersion (measured as 85th for speed dispersion, average traffic speed not significantly collectors; data from 50 percentile speed minus 50th related to fatality rates for any road type states percentile speed), and two nonspeed measures--traffic citations per driver and access to medical care--on fatality rates Garber Higher-speed roads [i.e., Analyzed relationship between Crash rates increased with increasing speed variance on all road and with average traffic crash rates and average traf- classes. No correlation between crash rates and average traf- Gadiraju speeds of 45 mph (72 fic speed, speed variance, fic speeds when data were disaggregated by road class (1988) km/h) or above], design speed, and posted including rural and speed limits urban Interstates, expressways and free- ways, rural and urban arterials, and rural col- lectors in Virginia
Harkey Rural and urban roads Compared speeds of crash- Found same U-shaped curve as Solomon and Cirillo; crashes et al. with posted speed lim- involved vehicles with speeds limited to weekday, nonalcohol, nonintersection involve- (1990) its of between 25 and of non-crash-involved vehi- ments 55 mph (40 and 89 cles km/h) in North Carolina and Colorado Fildes et al. Two urban arterials with Compared free-flowing travel Found no evidence of Solomon's U-shaped relationship. Those (1991) speed limits of 37 mph speeds and self-reported crash traveling at very fast speeds were more likely to report previ- (60 km/h) and two rural histories of drivers who par- ous crash involvement than those traveling at slower speeds. undivided roads with ticipated in a road safety sur- Self-reported crash involvements were lowest for those trav- speed limits of 62 mph vey eling at speeds below average traffic speeds and highest at (100 km/h), Australia speeds above the average with no advantage at the average Baruya Urban roads with average Analyzed relationship between Both speed level and speed dispersion affected crashes. and traffic speeds ranging personal injury crashes, speed Increased crashes were associated with increasing average Finch from 21 mph (33 km/h) levels, and speed dispersion, traffic speeds. Decreased crashes were associated with reduc- (1994) to 33 mph (53 km/h), defined as the coefficient of tions in speed dispersion at increasing speeds. The net Great Britain variation of the speed distri- effect, however, was an increase in personal injury crashes bution with increasing speeds Kloeden et Speed zones with 37-mph Compared speeds of casualty Found statistically significant increase in probability of al. (1997) (60-km/h) speed limits crash-involved vehicles with involvement in a casualty crash with increasing travel speed in metropolitan speeds of control vehicles above, but not below, the speed limit. Probability of crash Adelaide, Australia traveling in the same direc- involvement at speeds below the speed limit was not statisti- tion, at the same location, cally different from traveling at the speed limit time of day, day of week, and time of year under free-flow- ing traffic in daylight and good weather
MANAGING SPEED 46 influenced by other driver behaviors (e.g., drinking, not using safety belts) and characteristics (e.g., age), vehicle characteristics (e.g., size and weight), and road design (e.g., limited- or nonlimited-access highways). To the extent that these contributory variables are not taken into account, results of the studies must remain highly quali- fied. Correlational Studies This category of studies attempts to determine whether there is a link between speed and crash probability. In the benchmark study con- ducted by Solomon (1964), travel speeds of crash-involved vehicles obtained from police reports were compared with the average speed of free-flowing traffic on two- and four-lane, nonlimited-access rural highways. Solomon found that crash-involved vehicles were overrep- resented in the high- and low-speed areas of the traffic speed distri- bution. His well-known U-shaped curve (Figure 2-1) showed that crash involvement rates are lowest at speeds slightly above average traffic speeds. The greater the deviation between a motorist's speed and the average speed of traffic--both above and below the average speed--the greater the chance of involvement in a crash. The corre- lation between crash involvement rates and deviations from average traffic speed gave rise to the often-cited hypothesis that it is speed deviation, not speed per se, that increases the probability of driver involvement in a crash. Hauer's subsequent theory of traffic conflict (1971) provided a theoretical basis for Solomon's findings.11 Solomon's U-shaped relationship was replicated by Munden (1967) using a different analytic method on main rural roads in the United Kingdom, by Cirillo (1968) on U.S. Interstate highways 11 Some have interpreted these results to suggest that it is as unsafe to drive below as above the average traffic speed. This ignores the fact that drivers involved in a crash at higher speeds are at greater risk of injury than those driving at lower speeds, a rela- tionship that Solomon confirms in his analysis of the relation between speed and crash severity (see subsequent section).
47 Effects of Speed Figure 2-1 Vehicle crash involvement rates as a function of deviation from average traffic speeds (Solomon 1964; Cirillo 1968; Fildes et al. 1991 in Stuster and Coffman 1997, 6). 1 mph = 1.609 km/h. (Figure 2-1),12 and most recently by Harkey et al. (1990) on rural and urban roads posted at speeds ranging from 25 to 55 mph (40 to 89 km/h) in two U.S. states.13 All of the U.S. studies, but most par- 12 Cirillo limited her study to rear-end and angle collisions and same-direction side- swipe crashes involving two or more vehicles traveling between 9 a.m. and 4 p.m. on the assumption that the effect of deviation from average traffic speeds on crash involvement could best be determined by examining crashes involving vehicles travel- ing in the same direction (Cirillo 1968, 71). Thus, head-on, single-vehicle, and pedes- trian crashes were not included. 13 Unfortunately, Harkey et al.'s results cannot be compared with Solomon's results because the former excluded intersection involvements.
MANAGING SPEED 48 ticularly Solomon's, have been criticized for their dependence on crash reports14 for the precrash speeds of the crash-involved vehicles, which could bias the results (White and Nelson 1970, 67).15 Solomon's study has also been criticized for unrepresentative com- parative traffic speed data,16 lack of consistency between the crash and speed data,17 and mixing of crashes of free-flowing with slowing vehicles, which could explain high crash involvement rates at low speeds.18 The Research Triangle Institute (RTI) together with Indiana University (RTI 1970) addressed several of these issues by using speed data based, in part, on traffic speeds recorded at the time of the crash.19 They examined crashes on highways and county roads with speed limits of 40 mph (64 km/h) and above and found a similar but less pronounced U-shaped relationship between crash involvement 14 Solomon's study relied on estimates by the crash-involved driver or by the police or other third parties contained in police reports (Solomon 1964), which are frequently criticized as unreliable. 15 The authors demonstrated mathematically that errors in estimating speeds of the crash-involved vehicles could result in overestimates at the extreme speed deviations and underestimates in the middle speed interval, resulting in the U shape (White and Nelson 1970, 7071). 16 Spot speed surveys were taken at one location for each highway section, which state highway department engineers selected as being representative of the average traffic speed for the entire section. Crashes, however, occurred at many different locations along a section where average traffic speeds may or may not have been comparable with those at the speed survey location (Kloeden et al. 1997, 10). 17 The speed observations were made over a 12-month period ending in 1958, whereas the crash data were gathered over a 4-year period, also ending in 1958 (Solomon 1964, 7). Although expansion procedures were used and cross checks made to extend the speed data, the data collection issues remain troubling. 18 It was argued that if slowing or turning vehicles had been removed, the crash- involvement rates for vehicles moving at slow but free-flow speeds would have been lower (Accident Reconstruction Journal 1991, 16). 19 A system of on-line digital computer and magnetic loop detectors was installed in the pavement of an Indiana highway, which computed vehicle headways, speeds, lengths, and volumes (West and Dunn 1971, 52). By tracing vehicle trajectories and speeds between detectors, changes in speed fluctuations resulting from a crash could be identified and the crash-involved vehicles pinpointed (West and Dunn 1971, 53). The data from this instrumented highway comprised about half of the sample of crash involvements investigated in the study (Cowley 1987, 11).
49 Effects of Speed and speed. Thus, the RTI study appears to confirm the critical role of deviation from average traffic speeds for crash-involved vehicles. Deviation from average traffic speeds, however, is not the only factor linking speed with crash involvement. It does not explain, for example, the significant fraction of speeding-related driver involvements in fatal crashes involving only one vehicle--nearly 50 percent in 1996.20 In fact, when Solomon's data are disaggregated by crash type, the U-shaped relationship is only fully replicated for one crash type--nighttime head- on collisions (Cowley 1980 in Cowley 1987, 9) (Figure 2-2). Several studies have provided alternative explanations for the high crash involvement rates found by Solomon at the low end of the speed distribution, whereas others have simply not found the associ- ation. For example, West and Dunn (1971) investigated the relation- ship between speed and crash involvement, replicating Solomon's U-shaped relationship. However, when crashes involving turning vehicles were removed from the sample, the U-shaped relationship was considerably weakened--the curve became flatter--and the ele- vated crash involvement rates that Solomon had found at the low end of the speed distribution disappeared; crash involvement rates were more symmetric above and below mean traffic speeds (Figure 2-3).21 West and Dunn's analysis supports the conclusion that the character- istics of the road--numerous intersections or driveways on undivided highways, for example--are as responsible for creating the potential for vehicle conflicts and crashes as the motorist's driving too slowly for conditions.22 20 According to FARS 1996 data, nearly 70 percent of speeding-related fatal crashes involved a single vehicle. The lower driver involvement figure is used here to be con- sistent with Solomon's definition of crash involvement. 21 Solomon, too, found that rear-end and angle collisions accounted for a substantial proportion of total crash involvements at lower speed ranges, suggesting the presence of stopping and turning vehicles, even though the study sections had been selected so that crossroads and driveways were at a minimum (Solomon 1964, 36). 22 The one exception to this finding is the analysis conducted by Harkey et al. (1990) relating crash involvement to deviation from average traffic speeds on lower-speed rural and urban roads. Crash involvements were limited to weekday, nonalcohol, and nonin- tersection crashes. The analysis shows the same U-shaped curve as Solomon's even though intersection involvements were excluded (Harkey et al. 1990, 48). This study, however, suffered from the precrash speed measurement problems mentioned earlier.
Figure 2-2 Vehicle crash involvement rates by crash type (Cowley 1980 in Cowley 1987). Disaggregation of Solomon data for nonlimited-access rural highways. 1 mph = 1.609 km/h.
51 Effects of Speed Figure 2-3 Vehicle crash involvement rates including and excluding turning vehicles (West and Dunn 1971, 5354). 1 mph = 1.609 km/h. A more recent Australian study (Fildes et al. 1991), which exam- ined crash involvement rates as a function of speed on urban arteri- als as well as on two-lane rural roads,23 found no evidence of the U-shaped relationship. Crash involvement rates rose linearly as a function of speed; crash involvements were lowest at speeds below average traffic speeds and highest at speeds above the average with no advantage at the average (Fildes et al. 1991, 60) (Figure 2-1). Furthermore, the researchers did not find evidence of very low-speed driving that had been apparent in both the Solomon and Cirillo data (Fildes et al. 1991, 60). The results are based on small sample sizes 23Posted speed limits were 37 mph (60 km/h) on the urban arterials and 62 mph (100 km/h) on the rural roads (Fildes et al. 1991, 56).
MANAGING SPEED 52 and self-reported crash involvement, although Shinar notes that there is little reason to believe that slow-moving drivers would underreport their crashes.24 The findings point to a linear and posi- tive association between crash probability and the speed of crash- involved vehicles. A very recent Australian study (Kloeden et al. 1997) that exam- ined the relationship between speed and the probability of involve- ment in a casualty crash lends support for some of the results reported earlier by Fildes et al., at least for speeds above the average speed of traffic. Using a case control approach, the authors compared the speeds of cars involved in casualty crashes25 (the case vehicles) with the free-flowing speeds of cars not involved in crashes but trav- eling in the same direction at the same location, time of day, day of week, and time of year (the control vehicles) (Kloeden et al. 1997, i). Data collection was focused on weekday, daylight crashes--to exclude most alcohol-related crashes--in speed zones with a 37-mph (60-km/h) speed limit in the Adelaide metropolitan area (Kloeden et al. 1997, i). Precrash speeds were determined using crash reconstruc- tion techniques (Kloeden et al. 1997, 30). The data showed a steady and statistically significant increase in the probability of involvement of the case vehicles in a casualty crash with increasing speed above, but not below, the 37-mph speed limit, which roughly approximated the average traffic speed. The risk approximately doubled with each 3-mph (5-km/h) increase in speed above the limit (Kloeden et al. 1997, 38).26 The probability of casualty crash involvement at speeds below 37 mph was not statistically different from the probability at the speed limit (Kloeden et al. 1997, 38). The absence of a significant association between speed and crash involvement at speeds below the average traffic speed may well be the result of the study design. The analysis excluded all but injury crashes; crashes at lower speeds tend to be less severe. In addition, case vehicles were confined to those 24 See discussion of this report in Appendix B. 25 Casualty crashes are defined as crashes that involve transport of at least one person from the crash scene by an ambulance. 26 In contrast to the results reported by Fildes et al., the relationship between speed and crash involvement above the speed limit is nonlinear.
53 Effects of Speed with free-flow speeds prior to the crash, thus excluding the speeds of slowing vehicles that may have "caused the crash." Several studies have attempted to analyze the relationship between crash involvement and measures of the distribution of speeds in a traffic stream, thereby avoiding the problem of estimating the pre- crash speeds of individual vehicles. On the basis of data from 48 states, Lave (1985) developed models for a range of road classes (e.g., Interstates, arterials, collectors) to investigate the relationship between average traffic speed, speed dispersion, and fatality rates, attempting to hold constant some of the other factors that affect highway fatality rates using standard statistical techniques.27 He found that speed dispersion was significantly related (in a statistical sense) to fatality rates for rural Interstates and rural and urban arte- rials (Lave 1985, 1162).28 After controlling for speed dispersion, average traffic speed was not found to be significantly related to fatal- ity rates for any road type (Lave 1985, 1162). A series of analyses spawned by Lave's study, many of which contained a larger set of explanatory variables (e.g., driver age, alcohol use), confirmed the importance of speed dispersion to fatality rates but also found that average traffic speed is an important determinant.29 None of the studies discussed in this paragraph examined differences in roadway design features or traffic levels within road class, which could affect traffic speeds and crash rates. 27 Lave defined speed dispersion as the difference between the 85th percentile speed and the average traffic speed. These measures are aggregated over the period used to collect the speed data and thus may not reflect the distribution of traffic speeds at the time of the crashes. The other variables included a measure of enforcement--traffic citations per driver--and access to medical care. 28 Speed dispersion, measured as the difference between the 85th percentile and aver- age traffic speed, was statistically significant at the 5 percent level of confidence for the models for these road classes but not for the others. With the exception of rural Interstates, where the variables in the models explained 62 percent of the variation in fatality rates for 1981 and 52 percent for 1982, the variables in the models for the other road classes explained one quarter or less of the variation in the fatality rate (Lave 1985, 1161). 29 The relevant studies are those by Levy and Asch (1989), Fowles and Loeb (1989), Snyder (1989), Lave (1989), and Rodriguez (1990).
MANAGING SPEED 54 A related study by Garber and Gadiraju (1988) found, as Lave had, that average traffic speeds are not significantly related to fatality rates. Garber and Gadiraju examined the relationship between crash rates, speed dispersion,30 average traffic speed, and other measures that influence speed--design speed and posted speed limits--on sev- eral different classes of roads in Virginia.31 They found that crash rates declined with an increase in average traffic speeds when data for all road classes were combined (Garber and Gadiraju 1988, 26). The correlation disappeared when the data were disaggregated by road class, suggesting that the aggregated analysis simply reflected the effects of the different design characteristics of the roads being stud- ied (e.g., lower crash rates on high-speed Interstates). When crash rates were modeled as a function of speed dispersion for each road class, however, crash rates increased with increasing speed dispersion (Garber and Gadiraju 1988, 28).32 The minimum speed dispersion occurred when the difference between the design speed of the high- way, which reflects its function and geometric characteristics, and the posted speed limit was small [i.e., 10 mph (16 km/h)] (Garber and Gadiraju 1988, 2325). Evidence by Road Class The studies just reviewed suggest that the type of road may play an important role in determining driver travel speeds and crash proba- 30 Garber and Gadiraju quantified speed dispersion using speed variance as the mea- sure. Similar to Lave's treatment, this measure is based on aggregate data, which may or may not correspond to the distribution of traffic speeds and speed variance at the time of the crashes. 31 They examined higher-speed roads [i.e., average traffic speeds of 45 mph (72 km/h) or above], including rural and urban Interstate highways, expressways and freeways, rural and urban arterials, and rural collectors (Garber and Gadiraju 1988, 15). 32 Kloeden et al. (1997) point out, however, that the relationship between crash rates and speed dispersion could also reflect different design features of the roads being studied. For example, better-designed roads have lower crash rates because provision is made for overtaking and turning vehicles (or is not an issue on Interstate highways and freeways), thereby mitigating the circumstances that lead to speed dispersion (e.g., pla- toons forming behind slow-moving vehicles) (p. 23).
55 Effects of Speed bility. Thus, what is known about speed and crash probability by road class was also examined. Limited-Access Highways Most studies have focused on high-speed roads. By design, limited- access highways provide the least opportunity for vehicle conflicts and thus should have the lowest crash rates of all road classes. Cirillo's study (1968)--the only study focused specifically on limited- access Interstate highways--bears out this judgment. Crash involve- ment rates were lower across the board than those reported for other road types, except at very low speeds (Figure 2-1). Cirillo found, as Solomon had before, an association between crash involvement rates and deviation from average traffic speeds even on Interstate high- ways. More specifically, crash involvement rates were higher in the vicinity of interchanges where differences in vehicle speeds were greatest and, thus, the potential for vehicle conflicts was highest (Cirillo 1968, 75). Not surprisingly, the effect was greater near inter- changes on urban Interstate highways because of higher traffic vol- umes, making merging and diverging more difficult, and because of more complex and less adequate design of some urban interchanges (Cirillo 1968, 75). This finding points to the effect of traffic density as well as speed dispersion on crash rates. Two other studies reinforce the importance of traffic speed disper- sion to crash involvement on Interstate highways. Lave (1985, 1162) found a statistically significant relationship between increasing traf- fic speed dispersion and fatality rates on rural but not on urban Interstate highways. Garber and Gadiraju (1988, 2829) found that crash rates increased as traffic speed dispersion increased on both rural and urban Interstate highways. Nonlimited-Access Rural Highways The potential for vehicle conflicts is considerably greater on undi- vided highways, particularly high-speed nonlimited-access highways. Vehicles entering and exiting the highway at intersections and drive- ways, and passing maneuvers on two-lane undivided highways, increase the occurrence of conflicts between vehicles with large speed differences and hence increase crash probability. Solomon's study
MANAGING SPEED 56 (1964) provides strong evidence for these effects on two- and four- lane rural nonlimited-access highways. High crash involvement rates are associated with vehicles traveling well above or below the average traffic speed; at low speeds, the most common crash types are rear- end and angle collisions, typical of conflicts at intersections and driveways (Solomon 1964, 36). West and Dunn's analysis (1971) pin- pointed the important contribution of turning vehicles to crash prob- ability on these highways. When turning vehicles were excluded from the analysis, crash involvement rates at low speeds were not as high as those found by Solomon (Figure 2-1); they were more symmetric with crash involvement rates at high speeds (Figure 2-3). The study by Fildes et al. (1991) showed a gradual increase in crash probability for vehicles traveling above, but not below, average traffic speeds on two-lane rural roads (Figure 2-1). The previously cited studies by Garber and Gadiraju (1988) and Lave (1985) provide additional support for the contribution of speed dispersion to traffic conflicts and crash involvements on rural nonlim- ited-access highways. Garber and Gadiraju (1988, 2830) found a high correlation between increasing speed dispersion and crash rates on rural arterial roads, but the model included only these two variables. Lave's rural arterial model, which attempted to control for more vari- ables, found a weak but statistically significant relationship between traffic speed dispersion and fatality rates for only 1 year of data (wider dispersions were associated with higher fatality rates) (Lave 1985, 24). Neither study found any significant relationships between average traffic speeds and crash or fatality rates for this road class. Solomon's study provides some support for the role of speed per se in crash involvement on high-speed, nonlimited-access rural highways. He found that the percentage of single-vehicle crashes, which are more common on high-speed roads generally (NHTSA 1997b, 51), increased sharply as a function of the speed of the crash- involved vehicles (Solomon 1964, 36).33 33 Single-vehicle involvements represented a small proportion of all crash involve- ments at lower speeds, but they increased sharply at speeds in excess of 50 mph (80 km/h). At speeds exceeding 70 mph (113 km/h), they accounted for up to half of all crash involvements (Solomon 1964, 36).
57 Effects of Speed Together, these studies suggest that speed dispersion, created in part by the characteristics of rural nonlimited-access highways, con- tributes significantly to increased crash probability for this road class. The level of speed also appears to affect crash probability for certain crash types, such as single-vehicle crashes. Urban Roads and Residential Streets This category encompasses a wide variety of situations, from high- speed urban arterials to low-speed local streets. In theory, traffic speed dispersion and the potential for vehicle conflicts are likely to be high on urban roads, particularly on heavily traveled urban arterials. The highest levels of driver noncompliance with speed limits are in urban areas where an average of 7 out of 10 motorists exceed posted speed limits (Tignor and Warren 1990, 84). Numerous intersections, high levels of roadside activity, high traffic volumes, and insufficient following distances in congested traffic all contribute to increased crash probability. Offsetting these effects to some extent is the fact that congestion tends to reduce driving speeds, thus lessening the severity of the crashes that do occur. Some studies of the relationship between speed and crash proba- bility on urban arterials found a link between speed deviation and crash involvement for vehicles that travel at speeds well above aver- age traffic speeds. The primary evidence comes from the two Australian studies--Kloeden et al. (1997) and Fildes et al. (1991). Neither study, however, found that crash probability increased for those traveling below average traffic speeds. In fact, Fildes et al. found that crash involvement rates were lower for vehicles traveling below average traffic speeds, providing support for the importance of speed itself to crash probability. Lave found a low correlation between his measure of speed dis- persion and fatality rates in his model for urban arterials and reported that the correlation was statistically significant for only 1 of 2 years of data (Lave 1985, 1162).34 A small study of vehicle-pedestrian 34The model only explained about 17 to 18 percent of the variation in fatality rates (Lave 1985, 1162).
MANAGING SPEED 58 crashes at an urban intersection in Helsinki, for which vehicle speeds were actually videotaped, found that the majority of crash-involved drivers (8 of 11) were driving faster (30 mph or 48 km/h) than the average traffic speed (24 mph or 39 km/h) or the speed limit (25 mph or 40 km/h), thus also providing some confirmation of the role of speed deviation in urban crashes (Pasanen and Salmivaara 1993). Finally, a recent study of traffic speeds and personal injury crashes on urban roads in Great Britain, which classified roads by speed-related variables, found that measures of speed dispersion and speed levels have counterbalancing effects (Baruya and Finch 1994, 228).35 Crashes increase with the average speed of traffic,36 but at higher speeds, the dispersion in speeds is less, thereby reducing crash involvements. The net effect, however, is negative; the effect of increasing crash involvement with higher speeds appears to over- whelm any reduction in crash involvement from more uniform travel speeds (Baruya and Finch 1994, 229). The results of these studies suggest but do not prove that speed dispersion plays a role in crash probability on urban streets, particu- larly on urban arterials, and that many other factors, including speed itself, affect crash occurrence. Unfortunately, no studies that examine the relationship between speed and crash probability on residential streets could be found. Causal Studies The correlational studies are useful for identifying speed-related vari- ables associated with crash probability. However, they fail to establish a cause-and-effect relationship. In another type of study, generally referred to as clinical studies, detailed analyses of individual crashes 35 The roads were categorized using nonhierarchical cluster analysis into four groups with average traffic speeds ranging from 21 to 33 mph (34 to 53 km/h). Unfortunately the speed data were collected in 1992 and 1993, whereas the crash data were collected from 1983 to 1988 (Baruya and Finch 1994, 220221). 36 Geometric design differences among the different roads studied did not appear to play a significant role in the model as a correlate of crash frequency (Baruya and Finch 1994, 228).
59 Effects of Speed are conducted to determine the contribution of causal factors, such as speed, to crash occurrence. These studies enable more definitive statements to be made about the contribution of speeding to crash involvement. Their primary failing is the absence of any adjustment for exposure, that is, for any measure of the incidence of speeding in the general driving population relative to the crash-involved driver. Without such a measure, it is difficult to gain perspective on the rel- ative importance of speeding as a highway safety problem. Results of Clinical Studies The role of speeding as a crash cause was probably first analyzed in a detailed and comprehensive manner in Indiana University's Tri- Level Study (Treat et al. 1977). Speed was defined as causal if it met two conditions: (a) it deviated from the "normal" or "expected" speed of the average driver for the site conditions, and (b) it "caused" the crash, that is, the crash would not have occurred had the speed been as expected. On the basis of this definition, the study estimated "excessive speed" to be a definite cause in 7 to 8 percent of the crashes and a probable cause in an additional 13 to 16 percent of the crashes.37 Speed was identified as the second most common factor contributing to crash occurrence, second only to "improper lookout" (i.e., inattention) (Treat et al. 1977 in Bowie and Walz 1994, 32). Bowie and Walz (1994) integrated three large data files to obtain more reliable estimates of the role of speed in crash causation.38 Although they were based on different data sets and methodologies, the three sources yielded similar estimates, with "excessive speed" reported as being involved in approximately 12 percent of all crashes and more than 30 percent of fatal crashes (Bowie and Walz 1994, 31). 37 The crashes dated from 1970 to 1975 and were confined to state, county, and munic- ipal roads in Monroe County, Indiana (Treat et al. 1977). 38 These files included the census of all fatal crashes from FARS; 1 year of data from all police-reported crashes from six states in NHTSA's Crash Avoidance Research Data File (CARDfile), which had been specifically developed to analyze the factors involved in crash causation; and a subset of the crashes analyzed in depth from the Tri- Level Study. The full range of road classes was represented.
MANAGING SPEED 60 A more recent study of fatal crashes,39 analyzed for crash causa- tion and crash avoidance opportunities, found that aggressive driving, excessive speed, and loss of control were involved in 19 percent of those crashes (Viano and Ridella 1996, 132). The second most fre- quently cited crash cause--responsible for 11 percent of the fatal crashes--was labeled "rocket-ship." It involved single-vehicle, frontal-impact crashes with the "vehicle leaving the road at a very high speed." Because the analysis was confined to fatal crashes of belted occupants--and unbelted occupants are more highly repre- sented in fatal crashes--the percentage of speeding-related fatal crashes in the population at large is certain to be higher. Crash data from police crash reports from 1991 to 1995 were examined in a recent Canadian study (Liu 1997) to determine the role of speed in crashes. A speed-related crash was defined as one in which the driver was reported by the investigating officer to be both "exceeding the speed limit and driving too fast for conditions" (Liu 1997, 67). Although the definition is conservative, it is appropriate because police reports are not as reliable as professional, in-depth crash investigations. On the basis of this definition, Liu found that speed was a causal factor in 9 to 11 percent of all crashes and 12 to 15 percent of all casualty crashes (Liu 1997, 67). Despite different data files, different definitions of speeding and excessive speed, and different and often subjective techniques for making judgments about crash causation, the studies consistently found that speeding or excessive speed contributes to a relatively small but significant percentage of all crashes and a higher percent- age of more severe crashes. Behavioral Data on Speeding Relatively little is known about the behavioral aspects of speeding, at least in the United States. Some data are available from FARS about 39 The crashes actually occurred during an 18-month period from 1985 to 1986 throughout the United States, during which time the insurance industry undertook an incentive program to increase safety belt use by providing a $10,000 insurance policy in case of death while restrained in a crash. The cases were well documented by insur- ance adjusters and safety engineers (Viano and Ridella 1996, 125).
61 Effects of Speed a subset of the driving population: those involved in fatal speeding- related crashes. Speeding appears to be linked with other driver char- acteristics and behaviors. For example, young drivers (ages 15 to 20) are overinvolved40 in speeding-related fatal crashes (NHTSA 1997a, 2). Moreover, a high percentage of youthful drivers involved in speeding-related fatal crashes were also intoxicated and not wearing their safety belts at the time of the crash (NHTSA 1997a, 2, 5). Speeding-related fatal crashes also appear to be largely a phenom- enon of single-vehicle crashes. In 1996 single-vehicle fatal crashes represented 68 percent of all speeding-related fatal crashes according to FARS data. The next highest percentages were head-on (12 per- cent) and angle (10 percent) crashes. Single-vehicle fatal crashes are also highly associated with alcohol use, unbelted drivers, and night- time driving (NHTSA 1997b, 56). Speeding-related fatal crashes are also linked with road class. In 1996, for example, 17 percent of all speeding-related fatal crashes occurred on Interstate highways and freeways, according to FARS data. Fifty-four percent occurred on non-Interstate rural roads-- approximately equally divided between primary arterials and major collectors and other rural roads. The remaining 29 percent occurred on non-Interstate urban roads, the majority on local urban streets. Road design appears to play a role in the link between speed and fatal crash involvement, but making a definitive connection depends on knowing the incidence of speeding by road class, data for which are unavailable. Summary Although the evidence is not conclusive, speed appears to contribute to crash occurrence. Theory, empirical data drawn from correlational studies, and causal analyses of crashes provide evidence that both speed and speed dispersion are associated with crash involvement. Crash involvement rates rise as a function of speed for certain crash 40Overinvolvement is based on the percentage of young drivers in the population, not on the exposure of young drivers.
MANAGING SPEED 62 types, such as single-vehicle crashes. Deviation from the average traf- fic speed is also associated with crash involvement. At high speeds, deviation from average traffic speeds not only increases crash proba- bility but also the risk of a severe crash because of the close link between speed and injury severity discussed in the following section. At lower speeds, roadway characteristics--the presence of intersec- tions, turning vehicles, and the presence of pedestrians and bicy- clists--create the potential for conflict and crash involvement, but crashes may be less severe. Limited data are available to analyze speed-safety relationships by road class. Deviation from average traffic speeds appears to play a role in crash involvement on Interstate highways, particularly near inter- changes on urban Interstates, and to a greater extent on rural non- limited-access highways where high vehicular speeds and poorer road design combine to increase crash probability. Less is known about the role of speed and speed dispersion on urban roads. Given the charac- ter of many urban streets--numerous intersections, roadside activity, and the presence of pedestrians and bicyclists--the potential for con- flict is great, but congestion often restrains speed and lessens crash severity. The studies that have examined these relationships suggest that speed dispersion does indeed play a role in crash probability on urban streets, particularly on urban arterials, but that many other fac- tors, including speed itself, are likely to affect crash probability. Very little is known about the role of speed and speed dispersion on resi- dential streets. The clinical studies are unanimous in their finding that "excessive speed," that is, driving too fast for conditions, contributes to a signif- icant share of all crashes and a higher share of severe crashes. As the following section shows, the evidence for the effect of speed on crash severity is far more conclusive. Speed and Crash Severity The relationship between speed and crash severity is more straight- forward than the link between speed and crash probability. Once a crash has occurred--a vehicle has hit another vehicle or a stationary object--the vehicle undergoes a rapid change in speed. The vehicle
63 Effects of Speed decelerates rapidly but vehicle occupants continue to move at the vehicle's speed prior to impact until they are stopped in a second col- lision by striking the interior of the vehicle, by impact with objects external to the vehicle if ejected, or by being restrained by a safety belt or an airbag that deploys (Evans 1991, 247). The greater the speed at which occupants must absorb the energy released by the vehicle at impact, the greater the probability and severity of injury. The vehicle's rapid velocity change in a crash, which is often referred to as Delta-V, is thus an important measure of crash sever- ity. The probability that a crash will result in an occupant injury increases nonlinearly with impact speed. The energy released at impact, in turn, is determined by the speed at which the vehicle was traveling at the time of the crash. The power relationship between impact speed and the energy released in a crash--the energy release is proportional to the square of the impact speed--is responsible for the sharp rise in injury probability for the vehicle occupants.41 For example, an 18 percent increase in impact speed in a collision--from 55 to 65 mph (89 to 105 km/h)--results in nearly a 40 percent increase in the energy that must be absorbed by the vehicle occu- pants. Actual effects may differ because several factors can mitigate the duration and rate of the deceleration and hence the injury to the vehicle occupants. These measures include vehicle mass (the greater the vehicle weight relative to the weight of the other vehicle involved in a collision, the less the energy that must be absorbed and the injury to the occupants of the heavier vehicle),42 the energy-absorbing char- acteristics of the vehicle other than the mass, and the restraints on the vehicle occupants--safety belts and airbags--which enable them to "ride down" the impact forces (Evans 1991, 221). Solomon's 1964 study investigated the relationship between speed and crash severity in real-world crashes. Using three measures of crash severity--deaths, injuries, and property damage per involve- ment--the study showed that the higher the speed, the greater the 41 The equation that describes the release of kinetic energy as it relates to vehicle mass and speed is as follows: kinetic energy = 0.5 mass (velocity)2. 42 Vehicle mass, however, is less relevant when the vehicle strikes a fixed object.
MANAGING SPEED 64 fatalities, injuries, and property damage (Solomon 1964, 1114). Injury severity levels at high speeds were much greater than at lower speeds. For example, up to about 45 mph (72 km/h), 20 to 30 persons were injured and about 1 person killed per 100 crash-involved vehi- cles (Solomon 1964, 12). At 65 mph (105 km/h), 70 persons were injured and 6 persons killed per 100 crash-involved vehicles. The rate increased dramatically at very high speeds.43 Of course, Solomon's study was conducted before federal safety standards were introduced for motor vehicles. Consequently, fatality and injury rates are lower in absolute terms today. However, the association between higher speeds and higher crash severity levels that Solomon found has been borne out in subsequent studies. Several other researchers have confirmed the consistent relation- ship between speed and injury severity in crashes. Using data from the National Crash Severity Study, an intensive investigation of approximately 10,000 crashes from 1977 to 1979, O'Day and Flora (1982) found that the probability of a fatality increased dramatically with Delta-V. A driver crashing with an impact speed of 50 mph (80 km/h) was twice as likely to be killed as one crashing with an impact speed of 40 mph (64 km/h) (O'Day and Flora 1982 in TRB 1984, 39). At impact speeds above 50 mph, the probability of death exceeded 50 percent. Using NHTSA's National Analysis Sampling System (NASS), which contains data on a nationally representative sample of police- reported crashes of all severity levels, Joksch (1993) also found a very consistent relationship between Delta-V and the probability of death for drivers involved in car-to-car collisions. Fitting curves to crash data from 1980 to 1986 with known and estimated Delta-Vs, he obtained very similar functions: the probability of a fatality is related to Delta-V to the fourth power ( Joksch 1993, 104). 43 For example, at speeds of 73 mph (117 km/h) and greater, nearly 130 persons were injured and 22 persons were killed per 100 crash-involved vehicles (Solomon 1964, 12). The fatality estimates should be interpreted with care because of the small numbers of crash involvements. However, the same trend is evident for the injury data, where the sample is larger.
65 Effects of Speed Bowie and Walz (1994) examined the crash severity relationship for nonfatal injuries using somewhat more recent NASS data (from 1982 to 1989) and the Abbreviated Injury Scale (AIS). The AIS system rates injury levels from 1 (for a minor injury) to 6 (for an injury that is not currently survivable). The results (Figure 2-4) showed a dramatic increase in injury severity as Delta-V increased, confirming that real- world crash experience follows the laws of physics (Bowie and Walz 1994, 34). Combining several different crash files, the authors also compared injury severity levels with the distribution of injuries in speeding-related crashes. They found that the share of speeding- related crashes increased with increasing injury level. Ten percent of 70 60 AIS 2+ 50 AIS 3+ Level of injury 40 30 20 10 0 1 to 10 11 to 20 21 to 30 31 to 40 40 to 50 Over 50 Delta V (mph) Figure 2-4 Injury rates by crash severity, NASS (19821986) and Crashworthiness Data System (19881989) (Bowie and Walz 1994, 33). Rates are based on the number of vehicle occupants at a known Delta-V level injured at a specific AIS level divided by the total number of vehicle occupants involved in crashes at that level of Delta-V times 100. AIS 2+ injuries range from moderate to fatal; AIS 3+ injuries range from serious to fatal. Data are limited to tow-away crashes involving passenger cars and light-duty trucks. 1 mph = 1.609 km/h.
MANAGING SPEED 66 noninjured vehicle occupants, 17 percent of occupants sustaining inca- pacitating injuries, and 34 percent of fatally injured occupants were involved in speeding-related crashes (Bowie and Walz 1994, 34). The relationship between speed and crash severity is perhaps most dramatically demonstrated for vehicle crashes with pedestrians, the most vulnerable road users. The study of vehicle-pedestrian crashes in Helsinki (Pasanen and Salmivaara 1993) showed that the risk of death for a pedestrian increased rapidly from very low speeds (15 mph or 24 km/h) to about 50 mph (80 km/h), where death was almost certain (Pasanen and Salmivaara 1993, 308). A European review of several studies of vehicle-pedestrian crashes confirmed these results. It concluded that 5 percent of pedestrians are likely to die if they are struck by a vehicle traveling at 20 mph (32 km/h) and that risk levels rise sharply with speed--to a 45 percent probability of fatality for the pedestrian at 30 mph (48 km/h) and an 85 percent probability of fatality at 40 mph (64 km/h) (ETSC 1995, 11). In summary, all of the studies that have investigated the link between vehicle speed and injury severity have found a consistent relationship. As driving speed increases, so does the impact speed of a vehicle in a collision. Increased impact speed, in turn, results in a sharp increase in injury severity because of the power relationship between impact speed and the energy released in a crash. RELATIONSHIP OF SPEED TO TRAVEL TIME In addition to safety, travel time is a major factor affected by speed that influences drivers' choice of an appropriate driving speed. The importance and cost of travel time as a function of speed were amply illustrated by the recent experience of the 55-mph (89-km/h) National Maximum Speed Limit (NMSL). A review of the NMSL (TRB 1984) estimated that in 1982 motorists were spending about 1 billion extra hours on highways posted at 55 mph because of slower driving speeds compared with speeds on these highways in 1973, the year before the NMSL was enacted (TRB 1984, 120). Most of this additional travel time was expended by passengers in personal vehi- cles (TRB 1984, 119). Frequently it involved small increments in travel time for individual trips.
67 Effects of Speed Of course, any analysis of the time cost of travel has to take into account the cost savings from reduced crashes and averted fatalities and serious injuries from driving at lower speeds. When travel time costs were compared with estimated lives saved and serious injuries averted by the 55-mph (89-km/h) travel speed, the time cost worked out to about 40 years of additional driving time per life saved and serious injury avoided (TRB 1984, 120). The average remaining life expectancy of motor vehicle crash victims in 1982 was about 41 years. Thus, the number of years of extra driving time closely approximated the number of years of life saved.44 Although the study committee concluded that making a comparison between the value of a year of life and the value of a year of driving time was not meaningful, it did provide one framework for assessing the central trade-off between travel time and safety involved in the decision to retain or relax the 55-mph speed limit (TRB 1984, 120). Travel time costs are not equally distributed either by road type or road user. For example, the 55-mph (89-km/h) NMSL exacted the highest travel time costs for users of rural Interstate highways. At the time of the introduction of the NMSL, these highways had the high- est speeds, among the lowest crash rates, and carried the majority of long-distance travel, particularly commercial travel. Lowering speeds on these roads was estimated to cost motorists and truckers alike 100 years of additional driving per life saved--about four times as much as on all other affected roads (TRB 1984, 123). The travel time costs to motorists on other road classes were estimated to have much smaller effects, in part a reflection of the role of congestion and roadway geometry in limiting travel speeds on these nonlimited-access high- ways.45 Given these results, it was not surprising that the relaxation of the NMSL first occurred on rural Interstate highways. 44 A more recent analysis of the time-safety trade-off of raising speed limits on quali- fied sections of rural Interstate highways in 1987 found that the 65-mph (105-km/h) limit cost at least as much time as it saved when the years lost to deaths, injuries, and travel delays were compared with the travel time saved (Miller 1989, 73). 45 The comparable figures were 31 years of driving per life saved on urban Interstate highways and freeways, 28 years on rural arterials, and 14 years on rural collectors (TRB 1984, 123).
MANAGING SPEED 68 Travel time costs also tend to be unevenly distributed by road user. Most of the additional travel time attributed to the NMSL, for example, was borne by motorists engaged in personal travel. However, the value of this travel depends on trip purpose and length. For example, more highly valued work-related travel is relatively insensitive to changes in speed limits and accounts for a sizeable share of all local personal vehicle travel--most recently estimated at 32 percent in 1990 (FHWA 1992). However, commuting trips typi- cally are short--the average trip length is about 11 mi (18 km)--and average trip time is about 22 min (VNTSC 1994). Thus, slower speeds generally result in adding small time increments. For many work trips, congestion is likely to have more effect on driving speeds and travel time than are reductions in speed limits. Most personal travel (68 percent in 1990) is for shopping, family and other personal business, and social and recreational purposes. Because many of these trips are discretionary and do not have the same economic purpose as work travel, the time value of these trips is lower than for work travel, and, by extension, the incremental cost of reduced driving speeds from lower speed limits is also lower. Fortunately, most of these trips are short. Particular groups of road users--commercial truckers and other business travelers--may be more adversely affected by reduced driv- ing speeds attributable to lower speed limits. These groups drive more miles than the average motorist and often use high-speed roads. The economic cost of increased travel time for these user groups, particularly from lost productivity, can be substantial.46 RELATION OF SPEED TO FUEL USE AND OTHER VEHICLE OPERATING COSTS The primary motivation for the NMSL was to conserve energy by reducing driving speeds. Today, because of low fuel prices, driver con- cern for fuel economy plays a much smaller role in determining appropriate driving speeds. 46 In the case of the NMSL, however, the lower speed limit did have some benefits for truckers, such as lower fuel and maintenance costs.
69 Effects of Speed The most recent study of fuel efficiency (West et al. 1997 in Davis 1997, 3-50), based on a small sample of 1988 to 1995 model year auto- mobiles and light-duty trucks, shows a clear relationship between fuel economy and driving speed. Under steady-state, cruise-type driving con- ditions, fuel economy peaks at about 55 mph (89 km/h) and then declines at higher speeds, reflecting primarily the effect of aerodynamic drag on fuel efficiency (Figure 2-5).47 At lower speeds, engine friction, tires, and accessories (e.g., power steering) reduce fuel efficiency (TRB 1995, 63). Fuel efficiency also varies as a function of vehicle class. Sport utility vehicles, minivans, and pickup trucks--which represent a growing share of the U.S. passenger vehicle fleet--have poorer fuel economy, on Figure 2-5 Fuel economy as a function of speed, model year 19881995 automobiles and light-duty trucks (Davis 1997, 3-51). 1 mph = 1.609 km/h; 1 gal = 3.8 L. 47 Data on fuel economy as a function of speed for heavy trucks are older and more sparse. The available information suggests that fuel economy for heavy-duty diesel trucks declines sharply at speeds above about 50 mph (80 km/h), largely because of the effect of aerodynamic drag (TRB 1995, 125).
MANAGING SPEED 70 the average, than all but the heaviest automobiles for a wide range of speeds (Davis 1997, 3-52). Similarly, their fuel economy peaks at lower speeds, on the average, than does that of most passenger vehicles. Other vehicle operating costs, such as tire wear, are also likely to increase as a function of speed. Relative to fuel costs, however, these other operating costs are small; speed-related changes in their costs are not readily discernible by the average driver. Thus, they are not likely to affect motorists' choice of appropriate driving speeds. RELATION OF SPEED TO EMISSIONS Speed is clearly linked with vehicle emissions that contribute to pol- lution of the atmosphere, particularly to the degradation of metro- politan air quality. According to current models, volatile organic compounds (VOCs)--an ozone precursor--and carbon monoxide (CO) are highest at very low speeds associated with heavily con- gested stop-and-go traffic and rise again with high-speed, free-flow highway driving (TRB 1995, 4952). At high speeds, increased power demands on the engine cause CO and VOC emissions to increase, but at exactly what speed this occurs and by how much emissions are increased are unclear (TRB 1995, 122). Emissions of oxides of nitrogen (NOx), another ozone precursor, are thought to increase gradually at speeds well below free-flow highway driving, but again there is considerable uncertainty about the speeds at which this increase begins and the rate of increase (TRB 1995, 122). Data on emissions of heavy trucks as a function of speed are far more limited. The available data suggest that exhaust emissions of VOC and NOx from heavy-duty diesel vehicles rise at high speeds (TRB 1995, 122). Detailed data on diesel particulate emissions as a function of speed are unavailable. This is particularly troubling because particulate concentrations pose a significant health risk, and heavy-duty diesel vehicles are the primary source of highway vehicle particulate emissions (TRB 1995, 129). In addition to being a source of pollutants that degrade metro- politan air quality, transportation in general and motor vehicles in particular are the largest source of carbon dioxide (CO2) emissions, one of the principal greenhouse gases associated with global warm-
71 Effects of Speed ing.48 In 1994, motor vehicles accounted for about one-quarter of all U.S. CO2 emissions (TRB 1997, 83). The United States, in turn, is the largest emitter of CO2, accounting for one-quarter of global emissions (TRB 1997, 84). CO2 emissions--a by-product of any engine that burns fossil fuels--are closely linked with fuel economy and thus speed. At high speeds, where fuel economy is poor, vehicles emit more CO2. Vehicle speeds are also associated with noise; noise levels rise at higher vehicle speeds. Sonic pollution is of greatest concern to those living near freeways and on residential streets with higher-speed traffic. Despite the link between driving speeds and adverse environmen- tal effects, U.S. drivers do not directly pay for the costs that this pol- lution imposes on society.49 Thus they are not apt to consider environmental costs in their choice of an appropriate driving speed. SUMMARY In this chapter, the role of speed has been considered as it relates to the major factors motorists take into account in determining appro- priate driving speeds. The relation of speed to safety--a major con- cern for most drivers--is complex. Driving speed is clearly linked with crash severity. Injury severity in a crash rises sharply with the speed of the vehicle in a collision, reflecting the laws of physics. At equivalent impact speeds, injury severity for pedestrians, the most vulnerable of road users, is dramatically greater than for vehicle occu- pants. Furthermore, the incidence of speeding as a contributing fac- tor in crashes is higher the more severe the crash. The strength of the relationship between speed and crash severity alone is sufficient grounds for managing speed. 48 Unlike most other vehicle emissions, CO2 is not toxic. Along with other greenhouse gases, its effect in the upper atmosphere is to trap heat and warm the earth; hence the term greenhouse effect. 49 Drivers do pay for the cost of pollution controls on vehicles, emission inspections, and improved fuels.
MANAGING SPEED 72 Speed is also related to the probability of being in a crash, although the evidence is not as compelling. Theory, the results of empirical studies, and clinical analyses of crash causation all link speed with crash probability. However, crashes are complex events, and isolating the effect of speed from all the other factors that con- tribute to crash probability to establish causality unequivocally is not practicable. Moreover, the concept of speed itself is complex. Crash involvement has been associated with the dispersion in traffic speeds--in particular, with the deviation of an individual driver's speed from the average speed of traffic at both higher and lower speeds than the average. Those who drive at high speeds, well above the average speed of traffic, pose the greatest safety concern to them- selves and others because of the clear link between speed and crash severity. Crash involvement has also been associated with a driver's speed of travel. For example, single-vehicle crash involvement rates have been shown to rise with travel speed. The relationships among speed, speed dispersion, and crash prob- ability also appear to vary by road class. However, data are limited for many road types, and thus the observations that can be drawn are suggestive rather than conclusive. Speed dispersion poses an impor- tant safety concern on high-speed, nonlimited-access highways, such as rural, two-lane, undivided highways; wider speed dispersions are associated with higher crash involvement rates. Crash probability is also associated with speed dispersion on Interstate highways, partic- ularly on urban Interstates near interchanges. The potential for vehi- cle conflict is high on most urban streets, where pedestrians and parked vehicles augment normal vehicle conflicts. On these roads, however, lower driving speeds reduce injury severity if a collision occurs. Vehicle-pedestrian crashes are an exception, because pedes- trian injuries tend to be severe even at low impact speeds. Both speed and speed dispersion appear to play a role in crash likelihood on urban arterials; speed deviation above average traffic speeds and higher speeds in general are closely linked with crash probability on these roads. Little is known about the relationship between safety and speed on residential streets. Crash probability also varies by crash type. Speed dispersion is a contributing factor in the occurrence of multiple-vehicle rear-end
73 Effects of Speed and angle collisions, particularly for those driving well below average traffic speeds. Driving at high speeds is associated with a greater inci- dence of single-vehicle crashes. Travel time is another major factor affected by speed that influ- ences motorists' selection of an appropriate driving speed. However, travel time costs are not equally distributed either by road class or by road user. The highest travel-time costs occur on high-speed roads, particularly Interstate highways and freeways, where speed regula- tion, if enforced, can increase driving time under free-flowing traffic conditions. Commercial truckers and business travelers are heavy users of these types of roads and typically drive more miles than the average motorist. Consequently, the economic cost of increased travel time and lost productivity from speed reduction measures can be sub- stantial for these road users. Currently, fuel and other vehicle operating costs play a relatively minor role in motorists' selection of an appropriate driving speed. The relationship between speed and fuel use is unambiguous--fuel economy is inversely related to driving speeds above about 55 mph (89 km/h) for passenger vehicles, on the average, and at somewhat lower speeds for light and heavy trucks. After more than a decade of low fuel costs, however, drivers have little incentive to consider fuel costs in their choice of speed. Driving speed is clearly linked with vehicle emissions that contribute to metropolitan air pollution and emissions of CO2, a greenhouse gas closely associated with global warming. High driving speeds are also associated with noise pollution. U.S. drivers, however, have never directly paid for these costs. Thus, at present, the choice of an appropri- ate driving speed is not affected by consideration of environmental costs. These findings have several implications for managing speed. First, the unambiguous relationship between speed and crash severity alone is sufficient justification for controlling driving speeds. Second, if they are enforced, speed limits--the most common method of managing speed--can help restrict travel speeds, particularly at the very high speeds where the injury consequences of crashes are the greatest. Third, deviation of driving speeds from the average speed of traffic is associated with crash involvement. Thus, speed limit policies should attempt to minimize speed dispersion.
MANAGING SPEED 74 REFERENCES ABBREVIATIONS ETSC European Transport Safety Council FHWA Federal Highway Administration NHTSA National Highway Traffic Safety Administration RTI Research Triangle Institute TRB Transportation Research Board VNTSC Volpe National Transportation Systems Center Accident Reconstruction Journal. 1991. Is Speed Variation, Not Speed Itself, the Real Problem? Vol. 3, No. 6, Nov.Dec. Baruya, A., and D.J. Finch. 1994. Investigation of Traffic Speeds and Accidents on Urban Roads. Proc., Seminar J, PTRC European Transport Forum, University of Warwick, England, pp. 219230. Bowie, N.N., and M. Walz. 1994. Data Analysis of the Speed-Related Crash Issue. Auto and Traffic Safety, Vol. 1, No. 2, Winter, NHTSA, U.S. Department of Transportation, pp. 3138. Cirillo, J.A. 1968. Interstate System Accident Research--Study II--Interim Report II. Public Roads, Vol. 35, No. 3. Cowley, J.E. 1980. A Review of Rural Speed Limits in Australia. Report CR 20. J.E. Cowley & Associates, Melbourne, Australia. Cowley, J.E. 1987. The Relationship Between Speed and Accidents: A Literature Review. J.E. Cowley & Associates, Melbourne, Australia, March. Davis, S.C. 1997. Transportation Energy Data Book: Edition 17. ORNL-6919. Oak Ridge National Laboratory, Oak Ridge, Tenn., Aug. ETSC. 1995. Reducing Traffic Injuries Resulting from Excess and Inappropriate Speed. Brussels, Belgium, Jan., 35 pp. Evans, L. 1991. Traffic Safety and the Driver. Van Nostrand Reinhold, New York, 405 pp. FHWA. 1992. 1990 Nationwide Personal Transportation Survey: Summary of Travel Trends. FHWA-PL-92-027. U.S. Department of Transportation, March. Fildes, B.N., G. Rumbold, and A. Leening. 1991. Speed Behavior and Drivers' Attitude to Speeding. Report 16. Monash University Accident Research Center, Monash, Victoria, Australia, June. Fowles, R., and P.D. Loeb. 1989. Speeding, Coordination, and the 55-mph Speed Limit: Comment. The American Economic Review, Vol. 79, pp. 916922. Garber, N.J., and R. Gadiraju. 1988. Speed Variance and Its Influence on Accidents. University of Virginia, Charlottesville, July, 56 pp. Harkey, D.L., H.D. Robertson, and S.E. Davis. 1990. Assessment of Current Speed Zoning Criteria. In Transportation Research Record 1281, Transportation Research Board, National Research Council, Washington, D.C., pp. 4051. Hauer, E. 1971. Accidents, Overtaking and Speed Control. Accident Analysis and Prevention, Vol. 3, pp. 113. Joksch, H.C. 1993. Velocity Change and Fatality Risk in a Crash. Accident Analysis and Prevention, Vol. 25, pp. 103104.
75 Effects of Speed Kloeden, C.N., A.J. McLean, V.M. Moore, and G. Ponte. 1997. Travelling Speed and the Risk of Crash Involvement. Vol. I--Findings. NHMRC Road Accident Research Unit, The University of Adelaide, Australia, Nov., 60 pp. Lave, C. 1985. Speeding, Coordination, and the 55-mph Limit. The American Economic Review, Vol. 75, pp. 11591164. Lave, C. 1989. Speeding, Coordination, and the 55-mph Limit: Reply. The American Economic Review, Vol. 79, pp. 926931. Levy, D.T., and P. Asch. 1989. Speeding, Coordination, and the 55-mph Speed Limit: Comment. The American Economic Review, Vol. 79, pp. 913915. Liu, G.X. 1997. Identification and Analysis of Speed Related Accidents on Highways. Proc., Canadian Multidisciplinary Road Safety Conference X, Toronto, Ontario, Canada, June 811, pp. 6574. Mackay, M. 1985. Seat Belt Use Under Voluntary and Mandatory Conditions and Its Effects on Casualties. In Human Behavior and Traffic Safety (L. Evans and R.C. Schwing, eds.), Plenum Press, New York, pp. 259277. Miller, T.R. 1989. 65 MPH: Does It Save Time? The Urban Institute, Washington, D.C. In 33rd Annual Proc., Association for the Advancement of Automotive Medicine, Baltimore, Md., Oct. 24, pp. 7390. Munden, J.W. 1967. The Relation Between a Driver's Speed and His Accident Rate. Report LR 88. Road Research Laboratory, Crowthorne, England. NHTSA. 1997a. Traffic Safety Facts 1996: Speeding. U.S. Department of Transpor- tation, 6 pp. NHTSA. 1997b. Traffic Safety Facts 1996. DOT-HS-808-649. U.S. Department of Transportation, Dec., 192 pp. O'Day, J., and J. Flora. 1982. Alternative Measures of Restraint System Effectiveness: Inter- action with Crash Severity Factors. SAE Technical Paper Series No. 820798. Warren- dale, Pa. Parker, M.R., Jr. 1997. Effects of Raising and Lowering Speed Limits on Selected Roadway Sections. FHWA-RD-92-084. Martin R. Parker & Associates, Inc., Wayne, Mich., Jan., 175 pp. Pasanen, E., and H. Salmivaara. 1993. Driving Speeds and Pedestrian Safety in the City of Helsinki. Traffic Engineering and Control, Vol. 34, No. 6, pp. 308310. Rodriguez, R.J. 1990. Speed, Speed Dispersion, and the Highway Fatality Rate. Southern Economic Journal, Oct., pp. 349356. Rowsome, F., Jr. 1965. The Verse by the Side of the Road. The Penguin Group. RTI. 1970. Speed and Accidents, Vol. II. Final Report, Summary and Conclusions, May. Rumar, K., U. Bergrund, P. Jernberg, and U. Ytterbom. 1976. Driver Reaction to a Technical Safety Measure. Human Factors, Vol. 18, pp. 443454. Shinar, D. 1978. Psychology on the Road: The Human Factor in Traffic Safety. Wiley, New York. Snyder, D. 1989. Speeding, Coordination, and the 55-mph Limit: Comment. The American Economic Review, Vol. 79, pp. 922925. Solomon, D. 1964. Accidents on Main Rural Highways Related to Speed, Driver, and Vehicle. Bureau of Public Roads, U.S. Department of Commerce, July, 44 pp.
MANAGING SPEED 76 Stuster, J., and Z. Coffman. 1997. Synthesis of Safety Research Related to Speed and Speed Management (unpublished draft). Anacapa Sciences, Inc. Santa Barbara, Calif., FHWA, U.S. Department of Transportation, 56 pp. Taylor, D.H. 1964. Drivers' Galvanic Skin Response and the Risk of Accidents. Ergonomics, Vol. 7, pp. 439451. Tignor, S.C., and D.L. Warren. 1990. Driver Speed Behavior on U.S. Streets and Highways. ITE Compendium of Technical Papers, 60th Annual Meeting, Orlando, Fla., Aug. 5-8, pp. 8386. TRB. 1984. Special Report 204: 55: A Decade of Experience. National Research Council, Washington, D.C., 262 pp. TRB. 1995. Special Report 245: Expanding Metropolitan Highways: Implications for Air Quality and Energy Use. National Research Council, Washington, D.C., 387 pp. TRB. 1997. Special Report 251: Toward a Sustainable Future. National Research Coun- cil, Washington, D.C., 261 pp. Treat, J.R., N.S. Tumbas, S.T. McDonald, D. Shinar, R.D. Hume, R.E. Mayer, R.L. Stansifer, and H.J. Castellan. 1977. Tri-Level Study of the Causes of Traffic Accidents. Volume I: Causal Factor Tabulations and Assessment. DOT-HS-805-085. National Highway Traffic Safety Administration, U.S. Department of Transportation. Viano, D.C., and S. Ridella. 1996. Crash Causation: A Case Study of Fatal Accident Circumstances and Configurations. No. 960458. Society of Automotive Engineers, International Congress and Exposition, Detroit, Mich., Feb. 2629, pp. 123134. VNTSC. 1994. Journey-to-Work Trends in the United States and Its Major Metropolitan Areas, 19601990. FHWA-PL-94-012. U.S. Department of Transportation. West, B.H., R.N. McGill, J.W. Hodgson, S.S. Sluder, and D.E. Smith. 1997. Development and Verification of Light-Duty Modal Emissions and Fuel Consumption Values for Traffic Models. Federal Highway Administration, U.S. Department of Transportation, April. West, L.B., Jr., and J.W. Dunn. 1971. Accidents, Speed Deviation and Speed Limits. Traffic Engineering, Vol. 41, No. 10, pp. 5255. White, S.B., Jr., and A.C. Nelson, Jr. 1970. Some Effects of Measurement Errors in Estimating Involvement Rate as a Function of Deviation from Mean Traffic Speed. Journal of Safety Research, Vol. 2, No. 2, June, pp. 6772. Wilde, G.J.S., S.P. Claxton-Oldfield, and P.H. Platenius. 1985. Risk Homeostasis in an Experimental Context. In Human Behavior and Traffic Safety (L. Evans and R.C. Schwing, eds.), Plenum Press, New York, pp. 119149.
