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.
3 Managing Speeds: Speed Limits Angels Who guard you When you drive Usually Retire at 65 Burma Shave (Rowsome 1965) The preceding chapter provided evidence of a close link between speed and safety. Speed is directly related to injury severity in a crash, reflecting the laws of physics. The link between speed and the prob- ability of being in a crash is weaker, reflecting the fact that motor vehicle crashes are complex events that can seldom be attributed to a single factor. The evidence presented, however, is sufficiently strong to reaffirm the need for managing speed. In this chapter, one of the primary 77
MANAGING SPEED 78 methods of managing drivers' choice of speed--the imposition of speed limits--is discussed. Speed limits are part of a speed manage- ment system, which involves laws and a process for setting reasonable speed limits as well as enforcement, sanctions, and publicity. The chapter begins with an explanation of why regulatory intervention is justified. After a brief history of speed regulation, the major methods of establishing speed limits are introduced and their strengths and weaknesses summarized. The application of speed limits to different road classes and roadway environments is considered next. A review of what is known about the effect of speed limits on driver behavior and safety follows, drawing heavily on studies of recent changes in speed limits both in the United States and abroad. The chapter ends with a discussion of the implications of these findings for speed limit policies. REGULATING SPEED--A THEORETICAL JUSTIFICATION Drivers continually make choices about appropriate driving speeds, making their own assessment concerning the amount of risk they are willing to bear. Because drivers have a strong incentive to complete their trips safely, one could ask why they should not be left to choose their own travel speeds. There are three principal reasons for regulat- ing drivers' speed choices: (a) externalities,1 that is, the imposition of risks and uncompensated costs on others because of inappropriate speed choices made by individual drivers; (b) inadequate information that limits a motorist's ability to determine an appropriate driving speed; and (c) driver misjudgment of the effects of speed on crash probability and severity. The strongest case for regulatory intervention can be made on the grounds of externalities. Drivers may not take into account the risks imposed on others by their choice of an appropriate driving speed. 1 Externalities are defined as the "effect that occurs when the activity of one entity (a person or a firm) directly affects the welfare of another in a way that is not trans- mitted by market prices" (Rosen 1995, 91).
79 Managing Speeds: Speed Limits For example, drivers who choose to drive very fast relative to other traffic or very fast for existing road conditions in exchange for a shorter trip time may accept the higher risk of death or injury for themselves, but their choice almost certainly increases the risk of death and injury for other road users. Even a single-occupant, single- vehicle crash imposes medical and property damage costs that are not entirely paid for by the driver. Other externalities in the form of higher fuel consumption or higher emissions resulting from higher driving speeds are not directly paid for from current fuel or vehicle operating taxes. Such externalities are the major theoretical justifica- tion for the imposition of speed limits. (Speed limits, of course, are not the only possible regulatory response.) The externalities--partic- ularly the risks to others--may be relatively small on lightly traveled Interstate highways but quite large on streets adjacent to schools or in highly congested areas. The differences in the effects of the exter- nalities are important to consider in setting appropriate speed limits on different types of roads. Regulatory intervention is also justified if drivers are systemati- cally making "wrong" choices because of a lack of information or an inability to understand the information presented to them. (A wrong choice is defined as a choice that is different from the choice drivers would make if they had and understood all the relevant information.) For example, some drivers may not correctly judge the capabilities of their vehicles (e.g., stopping, handling) or anticipate roadway geom- etry and roadside conditions sufficiently to determine appropriate driving speeds. These circumstances may not be as relevant for expe- rienced drivers driving under familiar circumstances, although these drivers can make inappropriate decisions because of fatigue or other factors. Inexperienced drivers, or experienced drivers operating in unfamiliar surroundings, are more likely to underestimate risk and make inappropriate speed choices. For example, even experienced drivers may not make informed choices when faced with entirely new driving circumstances, such as the southerner confronting snow or the easterner confronting a winding mountain road with no shoulders. Another reason for regulatory intervention, which is related to the issues of information adequacy and judgment, is the tendency of some drivers to underestimate or misjudge the effects of speed on
MANAGING SPEED 80 crash probability and severity. For example, drivers may have a good sense of the relationship between driving speed and travel time, but they may not have as good an appreciation of the effect of speed on crash probability and crash severity. As noted in Chapter 1, there is some evidence that drivers systematically overestimate their driving skills and underestimate the risks of driving, particularly at higher speeds. Other drivers may simply be indifferent to speed regulations and will drive as fast as they can, ignoring the risks their speed choices impose on others. The justification for imposing speed lim- its, however, still leaves open the question of how the limits should be set, the topic of the following section. METHODS OF SETTING SPEED LIMITS Brief History of Speed Regulation The idea of regulating the speed of motor vehicle travel has a long history. In fact, the first speed regulations predated the invention of the automobile by some 200 years. The town of Newport, Rhode Island, prohibited the galloping of horses on major thoroughfares in an effort to prevent pedestrian deaths; Boston, Massachusetts, lim- ited horsedrawn carriages to a "foot pace" on Sundays to protect churchgoers (Ladd 1959). In 1901 Connecticut was the first state to impose a maximum speed limit of 8 mph (13 km/h) in cities (Labatut and Lane 1950). A review of early speed legislation suggests that the primary pur- pose of regulating speed was to improve public safety (Parker 1997, 1); another goal was uniformity in state speed regulations (UVC 1967, 436). The Uniform Vehicle Code (UVC), first published in 1926, provided the framework for speed control as it is known today. The original code contained (a) a basic rule requiring motorists to operate at speeds reasonable and prudent for conditions and (b) max- imum general speed limits2 in business and residential districts and other specific situations (e.g., grade crossings, in the vicinity of 2 These maximum limits were established as prima facie limits (UVC 1967, 437).
81 Managing Speeds: Speed Limits schools) (UVC 1967, 428, 437438). The 1934 version of the UVC broadened the maximum speed limits to cover more general situa- tions (e.g., urban districts) and introduced the concept of speed zones, wherein state agencies would determine alternative maximum speed limits at particular highway locations on the basis of engineer- ing and traffic investigations (UVC 1967, 437, 446).3 With the rise of the traffic safety movement during the 1930s, the National Safety Council organized a Committee on Speed Regulation in 1936 to study the speed problem. Its report (NSC 1941) reiterated the framework laid out by the UVC--a basic rule, maximum general speed limits, and authority for establishing speed zones on the basis of a traffic engineering study. The committee rec- ommended that state legislatures adopt uniform speed legislation based on this framework (NSC 1941, 5, 15). Above all it recom- mended a balanced approach toward speed control, directed at speed "too fast for conditions" rather than at speed in excess of some arbi- trary general limit (NSC 1941, 4). At the time the committee was conducting its work, cars were capable of reaching speeds of 80 to 100 mph (129 to 161 km/h), but the general statewide speed limit in many states was 35 to 45 mph (56 to 72 km/h) ( Joscelyn and Elston 1970, 32). With the tremendous growth in motor vehicle travel and further improvements in the highway system and the automobile during the 1930s and 1940s, the motoring public clamored for speeds higher than the maximum posted limits, which were frequently ignored because they were considered below those deemed reasonable by motorists (The American City 1950). In response, traffic engineers began to advocate an approach to setting speed limits (described sub- sequently) that is based on operating speeds as well as other factors. This method attempts to define a safe speed but also accommodates drivers' desire for a reasonable speed. 3 The 1926 UVC had previously recommended that local authorities be empowered to increase speed limits on through highways under their jurisdiction (UVC 1967, 450).
MANAGING SPEED 82 Considerations in Establishing Speed Limits General Speed Limits Versus Speed Zones In any discussion of speed limits, it is important to distinguish between general limits, which apply statewide or even nationwide, and limits in speed zones, which apply to a particular section of road. The former are set by legislation--by state statute, municipal ordi- nance, or Congress (ITE 1992, 347). Typically, general or legislated limits apply to a category of highway (see glossary)--a freeway or an arterial, for example--and reflect the design characteristics of the particular road class. They also differ by area, distinguishing rural from urban or local roads. By definition, general speed limits repre- sent a compromise; they may be well suited for some roads but are either "too high" or "too low" for others (Harwood 1995, 89). Speed limits in speed zones, on the other hand, are established by administrative action and are intended to be determined on the basis of an engineering study (ITE 1992, 347). The limits are tailored for a specific length of road where the general limit is deemed to be inap- propriate. Guidance is abundant on how to conduct the requisite engineering assessment of the traffic, road, and land use conditions that should be considered in establishing an appropriate speed limit in a speed zone (Harwood 1995, 89). Uniformity One might ask why speed zones are not established for every road segment, thereby tailoring speeds to the particular characteristics of the road and the location. Besides creating obvious resource problems because of the requirement to both undertake the necessary engi- neering studies and post signs on each highway section, a system of frequently changing speed limits would create confusion for the driv- er (Harwood 1995, 90). It could encourage a patchwork of different speed limits that may or may not be consistent across road classes and locations (Figure 3-1). The current system of statutory limits with speed zones as exceptions has the merit of encouraging uniformity and consistency of speed limits across a broad range of highways.
83 Managing Speeds: Speed Limits Figure 3-1 Parody of state response to repeal of 55-mph (89-km/h) National Maximum Speed Limit (reprinted with permission of Joe Heller, Green Bay Press-Gazette). Objectives of Speed Limits The primary purpose of speed limits is to enhance safety by reducing the risks imposed by drivers' speed choices. Speed limits enhance safety in two ways. They have a limiting function. By establishing an upper bound on speeds, the objective is to reduce both the probabil- ity and severity of crashes. Speed limits also have a coordinating function--to reduce dispersion in driving speeds (Lave 1985); more uniform speeds are associated with fewer vehicle conflicts. Another function of speed limits, which is related to their coordinating func- tion, is to achieve an orderly flow of traffic and improve traffic flow efficiency. Once established, well-conceived speed limits help deter- mine a reasonable standard for enforcement. Historically, speed lim- its have also been established for energy conservation purposes during times of national crisis.
MANAGING SPEED 84 Those who set speed limits attempt to balance road user safety and travel efficiency, among the many other factors that determine driv- ers' speed choice. Determining the optimal trade-off between these objectives depends, in part, on the function of the road. On limited- access facilities built to move traffic efficiently, greater emphasis may be placed on minimizing travel time without compromising safety. On local roads, where the primary function is access to abutting property, speed limits may be set to accommodate access rather than the efficient movement of traffic (Harwood 1995, 90). Informational Content and Reasonableness Whatever trade-offs are made between safety and travel time in establishing speed limits, posted limits ought to convey information to drivers. According to current practice, the numerical value on the sign advises the motorist of the maximum speed at which a driv- er can lawfully proceed under favorable conditions (e.g., good weather, daylight, and free-flowing traffic). Drivers are expected to reduce their speeds as these conditions change. The maximum speed limit should be related to the actual risk characteristics of the road (e.g., curvature, lane width) if drivers are to perceive the speed limit as credible and if adequate levels of voluntary compliance are to be achieved (Fildes and Lee 1993, 22). State and local governments do not have the resources--nor do they perceive it as a good use of resources--to apprehend and penalize large numbers of out-of-compliance drivers. Routine violation of speed limits by the majority of drivers may breed contempt not only for speed limits but also for other traffic regulations. As a general proposition, then, speed limits should be set at levels that are largely self-enforcing so that law enforcement officials can concentrate their efforts on the worst offenders. This goal, however, may not be achievable on all road classes (e.g., local streets) where lower driving speeds are desir- able but speed compliance is poor. These roads may be candi- dates for other speed management strategies, such as traffic calming.
85 Managing Speeds: Speed Limits Primary Methods of Setting Speed Limits The process of setting speed limits is often viewed as a technical exercise. However, the decisions concerning appropriate limits require value judgments and trade-offs that are appropriately handled by the political process--by Congress in the case of setting national speed limits and by state legislatures and city councils in determining general limits for highways under their respective jurisdictions. In this section, the major methods of determining speed limits are described. The section begins with a description of the methods most appropriate for setting statutory or legislated speed limits and con- tinues with a discussion of the methods appropriate for setting speed limits in speed zones (Table 3-1). Statutory Limits Statutory national speed limits were imposed twice in U.S. history, both during times of national crisis. A federal speed limit of 35 mph (56 km/h) was imposed during World War II. More recently, Congress established the NMSL of 55 mph (89 km/h) during the energy crisis of 1973 to reduce reliance on imported oil. In both cases, the objective was to reduce energy costs rather than transpor- tation costs. Safety benefits and travel time costs were a by-product rather than an intrinsic part of the initial determination of an appro- priate speed limit.4 Following repeal of the NMSL in 1995, most state legislatures acted to raise speed limits on highways subject to the 55-mph (89- km/h) speed limit. Many states reverted to the maximum general speed limits in effect for these highways before the NMSL was enacted. Other states established new speed limits on these high- ways. Legislative decisions typically were accompanied by public input and technical support provided by state departments of trans- portation (DOTs). Before posting speed limit increases, many state 4 Benefit-cost considerations, particularly the trade-offs between loss of life, injury costs, and time savings, were more directly considered and debated when Congress decided to allow states to raise speed limits on rural Interstates in 1987.
Table 3-1 Characteristics of the Primary Methods of Setting Speed Limits Speed Limit Most Common Relation Appropriateness by Ease of Relation to Method Application to Safety Road Class Implementation Enforcement Statutory General limits Trade-offs among Statutory limits typi- Difficult to achieve Can be difficult to limits safety, travel time, cally are estab- consensus on enforce if limits are set and other objectives lished by road class national limits arbitrarily are politically deter- and sometimes by except during mined location (e.g., times of crisis-- rural) easier to establish at state and local level Optimum General limits or Safety is balanced with Theoretically, should No known practical If implemented, could be speed limits speed zones other objectives be adaptable for application--dif- difficult to enforce (e.g., travel time) to any road class ficult to quantify because socially optimal minimize social key variables speed limits are typically highway transport lower than what individ- costs ual drivers would select Engineering Speed zones Not necessarily a safe May not be as appro- Well-established Helps establish a reason- study speed for all road priate for urban methodology for able target of out-of- method with classes; it depends, roads, particularly determining 85th compliance drivers for speed limits for example, on the residential streets, percentile speed enforcement set near the dispersion of speeds with greater mix of 85th per- between the slowest road users and func- centile speed and fastest drivers tions than major arterials and freeways
Expert Speed zones Helps identify many Probably most useful Complex system to System has been used to system factors, in addition and appropriate for develop, requir- target photo enforce- based to vehicle operating roads in urban ing knowledge- ment (i.e., where recom- approach speeds, that may areas where speed able experts and mended program limit is affect safety limits based solely computer capa- substantially below driv- on 85th percentile bility ers' preferred operating speed may be inap- speeds) propriate Variable speed Freeways Not fully demon- Because of expense, Limited experience Systems are often com- limits strated--some indi- most appropriate in United bined with photo radar cation that more for highest-class States--new tech- enforcement uniform speeds roads with large nologies are being reduce crashes traffic volumes introduced
MANAGING SPEED 88 DOTs conducted engineering surveys of candidate state highways for speed limit increases, examining design speeds, pavement condi- tion, traffic congestion, crash data, and existing travel speeds, among other factors, to determine where the limits could be safely raised. Many states are monitoring driving speeds and crash experience to determine whether further changes in speed limits should be legis- lated. At least one state, New Jersey, has raised speed limits on a trial basis pending the results of a study of the overall effect of the legis- lated increases.5 Statutory speed limits also have a long history at the local level. Many local governments have set speeds by statute or ordi- nance on local roads. In recent years, citizen concerns about speeding, particularly on neighborhood streets, have led to lower speed limits and other measures to manage driver speeds in resi- dential areas. The concept of legislated speed limits has appeal from a policy perspective. The trade-offs between safety, travel time, and other costs implicit in setting speed limits involve value judgments that are often best resolved by the political process. On the other hand, leg- islated speed limits can be arbitrary. The recent NMSL, for example, had been criticized for not appropriately reflecting differences in geography and local traffic conditions. Optimum Speed Limits In the early 1960s Oppenlander proposed a scientifically based pro- cedure for regulating vehicle operating speeds to set speed limits at an optimal level from a societal perspective (Oppenlander 1962). The 5 The legislation, enacted Jan. 19, 1998, raised speed limits to 65 mph (105 km/h) on 400 mi (644 km) of highways but requires the Commissioner of Transportation in consultation with the Attorney General and the authorities (i.e., New Jersey Highway Authority, New Jersey Turnpike Authority, and South Jersey Transportation Authority) to conduct an 18-month study of speeds, crash rates, fatalities, enforcement, air quality, and other issues to evaluate fully the effect of the 65-mph speed limit. They are requested to submit recommendations to the legisla- ture as to whether the number of miles of highways eligible for 65 mph should increase, decrease, or remain the same [P.L. 1997, Chapter 415 (3)].
89 Managing Speeds: Speed Limits method recognized that individual drivers do not always select driv- ing speeds that take into account the risks imposed on others by their choice. For example, driving at high speeds can increase the likeli- hood of a severe crash, which may involve other road users; it also results in added fuel consumption and higher emission levels, costs that are not entirely borne by the individual driver or even other highway users (MacRae and Wilde 1979, 136137). Because of these external costs, the optimum speed for an individual driver is different from the socially optimum speed. Oppenlander's approach attempted to define costs per mile of travel as a function of speed for four cost categories: (a) vehicle oper- ation, (b) travel time, (c) crashes, and (d) service (i.e., comfort and convenience) (Oppenlander 1962, 78). The cost curves were devel- oped from studies of vehicular travel on various types of highways for different traffic situations, travel conditions, and types of motor vehi- cles (Oppenlander 1962, 78). The "optimal speed" was determined by solving for the minimum point on the total cost curve, which repre- sented the minimum social cost of highway transport for a particular set of conditions (Oppenlander 1962, 78). The approach is most appropriate for establishing general speed limits for different road classes. However, it can also be used for setting speed limits in speed zones by adjusting optimal speeds to reflect the specific physical and environmental features of a given highway segment. Marcellis (1962) attempted to apply Oppenlander's theory for dif- ferent types of travel (rural and urban), for different types of roads (two- and four-lane), and for different types of vehicles (passenger cars and commercial vehicles) during daytime and nighttime travel. He found an optimal speed that minimized the cost of traffic movement for each of these conditions.6 The recommended application of his results was the establishment of general speed limits (Marcellis 1962, 1). 6 For example, he found large differences, up to 11 mph (18 km/h), between the opti- mal speeds of passenger cars and commercial vehicles. There were also large differences in optimal speeds by area. In rural areas, the optimal speed for passenger cars was 50 mph (80 km/h); in urban areas optimal speeds decreased with an increase in the num- ber of stops per mile from 41 to 29 mph (66 to 47 km/h). Lesser differences were found for optimal speeds on two- versus four-lane rural highways, and even smaller differences were found between daytime and nighttime optimal speeds (Marcellis 1962, 1, 59).
MANAGING SPEED 90 A more recent variant of this approach ( Jondrow et al. 1982) determines the socially optimum speed by starting with the private optimum speed and then adjusting it to account for external costs. The potential benefit of this approach is that it is based on the driv- er's judgment about the values of trip time, gasoline costs, and, most important, the value of life. It does not impose externally derived val- uations of these parameters. The shortcoming is that it assumes a sin- gle representative driver, which implies that all drivers have the same preferred optimum speed. Although conceptually appealing, optimum speed limits have never been used in practice. One problem, which is typical of most benefit-cost analyses, is the difficulty of quantifying key variables. Considerable work has been done on valuation of travel time as well as on the costs of injury and mortality, but there is no clear consensus on these estimates. Another issue is implementation. In their analysis, MacRae and Wilde (1979) attempted to estimate an optimum national speed limit--a preliminary estimate was between 55 and 60 mph (89 and 97 km/h). However, when con- sideration was given to recommending such a limit, it was not clear that the optimum limit would achieve its goal. Successful imple- mentation of the approach depends on driver compliance and per- ception that the speed limit is reasonable and on the level of enforcement activity. Engineering Study Method The most common method for determining speed limits in a speed zone sets the limit on the basis of an engineering study. The study requires data collection and analysis in the determination of an appropriate limit. The data include measurement of prevailing traffic speeds, crash data, and information on highway, traffic, and roadside conditions not readily apparent to drivers.7 7 Many roadside and roadway features are readily apparent to drivers and have already been taken into account in the speed they choose to drive. The factors that are not so readily apparent and that may warrant a lower speed limit include hidden intersections or driveways, lane drops, and other unexpected conditions.
91 Managing Speeds: Speed Limits A recent survey of state and local governments (Fitzpatrick et al. 1997) found that the 85th percentile speed is the most widely used factor for determining the level at which to set the limit. Other fre- quently considered factors included crash experience, roadside devel- opment, roadway geometry, and maximum speed limits set by state statute or local ordinance (Fitzpatrick et al. 1997, 52). Typically, the speed data--more specifically the 85th percentile speed--provide the first approximation of the speed zone limit (ITE 1992, 348). The limit may be adjusted from this value on the basis of the other factors. Setting the speed limit near the 85th percentile, that is, the speed at or below which 85 percent of drivers operate their vehicles, assumes that most drivers are capable of judging the speed at which they can safely operate (Krammes et al. 1996, 7, 12). The 85th per- centile speed is determined through spot speed studies of free- flowing traffic (i.e., traffic unimpeded by other vehicles), which yield a distribution of speeds from which the 85th percentile is calculated (Krammes et al. 1996, 7) (Figure 3-2).8 The implication for enforce- ment is that no more than 15 percent of motorists will be out of compliance. In practice, typical enforcement tolerances of between 5 and 10 mph (8 and 16 km/h) above posted limits further narrow the enforcement band. As early as 1941, the report of the Committee on Speed Regulation (NSC 1941, 13) advocated determining critical or maxi- mum safe speeds by observing the operating speeds at or below which 80 or 90 percent of drivers travel under normal weather and daylight conditions. Although the method was recommended for establishing 8 At least two additional measures of speed dispersion are available for calculating operating speed as a basis for setting speed limits. The first is the pace speed, which is defined as the 10-mph (16-km/h) range encompassing the greatest percentage of all the speed observations at a particular site. It is described by both the speed value at the lower end of the range and the percentage of all vehicles that are within the range and thus is an alternative indicator of speed dispersion (see glossary). The sec- ond is the skewness of the speed distribution. Research by Taylor (1965) found a strong relationship between the rate of occurrence of crashes and a skewed (i.e., non- normal) speed distribution on rural state highways. Hence, he argued that the appropriate speed for a speed zone should be based on changing the speed distribu- tion from a nonnormal to a normal distribution by a "before" and "after" analysis of the actual speed distribution within the zone (Taylor 1965, 51).
MANAGING SPEED 92 Figure 3-2 Speed distribution showing the 85th percentile speed (Krammes et al. 1996). appropriate speed limits in speed zones, it was also advocated as a way of determining proper values for general speed limits, particu- larly on rural highways.9 A 1960 survey of the member departments of the American Association of State Highway Officials found that the majority set speed limits in speed zones primarily on the basis of the 85th percentile speed, although a few departments used the 90th percentile speed; such factors as design speed, geometric characteris- tics, crash experience, traffic volumes, and development were sec- ondary considerations (Sub-Committee on Speed Zoning 1969 in Joscelyn and Elston 1970, 99). The 85th percentile speed was accepted because traffic engineers often found that this was the upper limit of the 10-mph (16-km/h) pace (Carter 1949 in Joscelyn and Elston 1970, 94). Setting the speed limit near this point would encourage most drivers to travel at more uniform speeds, thus mini- mizing opportunities for vehicle conflict (Baerwald 1953 in Joscelyn 9Speed limits on rural highways in some midwestern states were set at 50 mph (80 km/h), which the committee thought could be raised somewhat for daytime travel. Other states had not posted numerical limits for rural highways because of concerns about the adequacy of enforcement efforts. Speed limits based on operating speeds were perceived to be useful both in determining reasonable speed limits and for enforcement (NSC 1941, 14, 16).
93 Managing Speeds: Speed Limits and Elston 1970, 105). In addition, experience indicated that the 85th percentile speed appeared to be reasonable from a law enforce- ment standpoint (Tennessee Department of Highways 1968 in Joscelyn and Elston 1970, 99). Analytic support for setting speed limits near the 85th percentile speed came from a series of traffic safety studies (Solomon 1964; Cirillo 1968; RTI 1970) whose strengths and weaknesses were dis- cussed in the preceding chapter. The studies found that crash involvement rates on certain road classes were lowest for vehicles traveling in a speed range whose upper bound was about one stan- dard deviation above average traffic speeds, or approximately at the 85th percentile speed.10 Thus, the 85th percentile speed not only rep- resents the upper bound of the preferred driving speed of most driv- ers, but, according to some studies, for some roads it also corresponds to the upper bound of a speed range where crash involvement rates are lowest.11 Setting the speed limit near the 85th percentile speed has great appeal from a behavioral and an enforcement perspective. However, it is less clear that the 85th percentile speed necessarily corresponds to the lowest crash involvement rates on all road classes. The safety ben- efits may well depend on the range of speeds. The narrower the speed dispersion--the less the spread between the average speed and the 85th percentile speed--the greater the safety benefits. This principle was illustrated with the imposition of the NMSL in 1973. The lower speed limit resulted in a considerable narrowing of the spread between the slowest and fastest drivers in 1974, contributing to the substantial reduction in fatalities in that year (Figure 3-3) (Godwin 1988, 25).12 10 The relationship between crash involvement rates and deviation from average traf- fic speeds can also be used to establish minimum speed limits, particularly on lim- ited-access highways designed for high-speed driving. Some states have set minimum speed limits on these highways at one standard deviation below the aver- age traffic speed, or approximately at the 15th percentile speed. 11 According to most of these studies, crash involvement rates are lowest from about the 50th to the 85th percentile speed (Figures 2-1, 2-3). 12 In addition, the fuel shortages of the time caused people to travel less, and, because of high levels of motorist compliance, average speeds declined. Reduced travel and reduced speeds both affected safety.
MANAGING SPEED 94 40 RURAL PRIMARY PERCENTAGE OF VEHICLES 30 1974 (average = 53.5) 20 1973 (average = 57.1) 10 0 30 35 40 45 50 55 60 65 70 75 80 SPEED (mph) 40 INTERSTATE 40 35 1974 (average = 57.6 mph) PERCENTAGE OF VEHICLES RURAL SECONDARY PERCENTAGE OF VEHICLES 30 30 25 1974 (average = 49.5) 20 20 1973 (average = 65 mph) 15 1973 (average = 52.6) 10 10 5 0 30 35 40 45 50 55 60 65 70 75 80 35 40 45 50 55 60 65 70 75 80 SPEED (mph) SPEED (mph) Figure 3-3 Change in vehicle speed distribution by various road classes, 1973-1974 (TRB 1984, 2627). 1 mph = 1.609 km/h. Also, the 85th percentile speed is not stationary. Monitoring data collected on speeds, including the 85th percentile speed, following the relaxation of speed limits on some rural Interstates in 1987 showed an increase in 85th percentile speeds both for states that raised and states that retained the 55-mph (89-km/h) speed limit (Godwin 1992, 4).13 Data collected by the Insurance Institute for 13 Between 1986 and 1988, 85th percentile speeds changed by 1.7 mph (2.7 km/h) in 55-mph (89-km/h) states and by 3.2 mph (5.1 km/h) in 65-mph (105-km/h) states for which reliable data were available (Godwin 1992, 4). The increase in 85th per- centile speeds in 55-mph states was attributed to higher-speed driving by motorists from other states accustomed to the 65-mph limit (Godwin 1992, 7). It could also have reflected a general relaxation in driver compliance with and enforcement of the 55-mph speed limit. However, because average speeds did not change as much in 55- mph states as 85th percentile speeds, speed dispersion [measured by the estimated standard deviation (85th percentile speed minus average speed)] was larger in these states than in 65-mph states (Godwin 1992, 4).
95 Managing Speeds: Speed Limits Highway Safety (Retting and Greene 1997) suggest that a similar "speed creep" phenomenon may be occurring in the wake of the repeal of the NMSL in 1995. Both 85th percentile speeds and speed dispersion (measured as the speed standard deviation) have increased (Retting and Greene 1997). The key issues of concern from a safety perspective are whether speeds will continue to increase with driver expectations of an enforcement tolerance and what effect these changes will have on crash frequency and crash severity. The findings from a review of studies on this topic are reported later in the chapter. A final concern with setting the speed limit at the 85th percentile speed is that it may not be appropriate for all classes of roads. For example, property access, community concerns, and pedestrian safety are important factors in setting appropriate speed limits on many urban roads, particularly residential streets. Thus, basing speed limits on the 85th percentile speed--a measure of unconstrained free- flowing travel speed--will not be as appropriate on these streets as on major arterial highways where travel efficiency is the primary road function (Harwood 1995, 90). However, compliance with speed lim- its on urban roads is already poor,14 suggesting that setting the lim- its too low may create a greater enforcement burden or demand a greater tolerance for noncompliance. Lowering travel speeds, partic- ularly on residential streets, may require other speed management strategies. Expert SystemBased Approach Several states in Australia have developed an expert systembased approach to setting speed limits in speed zones. Victoria was the first state to embark on this approach in 1987 as the result of its compre- hensive review of all aspects of speed management. The goal was to 14 Research on driver speed behavior on a sample of U.S. roads found that, on the average, 7 out of 10 motorists exceeded the posted speed limit in urban areas. Many of the current speed limits in these areas correspond to the 30th percentile speed (Tignor and Warren 1989, 2).
MANAGING SPEED 96 develop a more uniform and consistent approach to setting speed limits within speed zones (Donald 1994, 284).15 The decisions and judgments required to establish speed limits were thought to be particularly amenable to an expert system approach. Expert systems are computer programs that mimic an expert's thought processes to solve complex problems in a given field (Donald 1994, 287). The problem must have a well-defined knowl- edge base, "experts" must be able to verbalize their knowledge and experience in the form of tasks to be undertaken and decisions to be made, and outcomes must be limited in number and clearly defined (Donald 1994, 287). Development of the Victoria expert system VLIMITS, which was undertaken by the Australian Road Research Board (ARRB), began with field measurements at over 60 sites. Experts then reviewed the field data to elicit decision rules for determining appropriate speed limits for various road classes and traffic conditions. This "expert judgment" was reduced to a personal computer program, which leads the user through a series of question-answer menus that ultimately results in a recommended speed limit for a particular road section. VLIMITS was revised and updated in 1992. At the same time, devel- opment of related versions of the program--NLIMITS and QLIMITS--was begun for use in New South Wales and Queensland, respectively (Donald 1994, 293).16 The ultimate objec- tive is to develop a single countrywide speed zone program. The most recent version of the system takes the user through a five-step process (Figure 3-4), which includes (a) environmental characterization of the area (e.g., urban, rural), (b) roadway and road- side factors (e.g., divided highway, number of lanes), (c) a first approximation of a speed limit based on a and b, (d) special activities (e.g., school zone) or other factors that might modify the final zon- ing (e.g., zone length, adjacent zone speed limits), and (e) 85th per- 15 Similar to the United States, Australia uses general speed limits supplemented by speed zones where the general speed limits are not considered suitable for the par- ticular road and traffic conditions. 16 Development costs for NLIMITS, the more recent system, were $51,800 U.S. (Coleman et al. 1996, 48).
97 Managing Speeds: Speed Limits Resultant Zoning Basic Modifiers Zoning 10 accidents service roads 20 Urban special accesses activities 40 road type length of 50 zone median width Rural QLIMITS, OUTPUT adjacent 85th setback 60 = speed zone zones NLIMITS, percentile recommendation VLIMITS, speed divided schools plus Urban 70 Version 2.2 `flags' of factors fringe number of horizontal needing further lanes alignment 80 consideration access control vertical Rural 90 alignment suitability for fringe 110 km/h 100 interchange spacing 110 Figure 3-4 Overview of the structure of the Australian computer-based speed limit advisor (Donald 1994, 292). 1 km/h = 0.621 mph. centile speed. The output of this process is a recommended speed zone value; specific factors may also be flagged for further consider- ation. The system is programmed not to allow a value higher than the general speed limit established for the particular road class under consideration (Coleman et al. 1996, 48). The expert system approach includes all the factors covered in the engineering study method. The main difference is the process. The expert system approach makes the factors and the decision rules involved in determining an appropriate speed limit more explicit. The computer-based advisor17 is used primarily to assist regional road authorities to determine appropriate speed limits in speed 17System developers have moved away from calling the program an expert system, because it does not "learn" from its previous experience. Rather, the current system
MANAGING SPEED 98 zones. Program results are not intended to be automatically adopted but to provide advice to those who must make the final decision. The system is also used by Local Council Authorities to respond to neighborhood requests, generally for lowering speed limits. Two additional uses in Victoria are by the Royal Automobile Club, pri- marily to respond to member complaints about unreasonable speed limits, and by the Victoria Police Department, who use it as a guide in selecting locations for speed cameras (personal communication, D. Donald, ARRB, Sept. 4, 1997). In practice, on higher-speed roads, the computer advisory system recommends a speed limit that is close to the 85th percentile speed in most cases (Coleman et al. 1996, 48). The system appears to be most useful on roads where the 85th percentile speed is seen as an inappro- priate basis for setting speed limits. Heavily trafficked urban areas with a mix of road users, including bicyclists and pedestrians, and heavy roadside activity (e.g., parking, access to businesses) fall into this category, and here the system--using a common set of criteria-- is likely to recommend a lower speed limit more compatible with the needs of all road users. However, lower speed limits require high lev- els of enforcement to ensure compliance. In Victoria, as in other Australian states, photo radar is heavily used to enforce lower limits (personal communication with D. Donald, ARRB, Sept. 11, 1997). Other Methods of Setting Speed Limits Basic Law Limits Another approach to setting speed limits is to leave it up to the driv- er to determine a reasonable and prudent travel speed. This is the current policy for passenger vehicles in Montana on Interstate high- ways during daylight hours.18 With the repeal of the NMSL, the is hard-coded so that any changes require computer programming input (personal communication with D. Donald, ARRB, Sept. 11, 1997). 18 There are nighttime speed limits for passenger vehicles of 65 mph (105 km/h) on Interstate highways and 55 mph (89 km/h) on all other roads. Heavy trucks must obey 65-mph speed limits day and night on Interstate highways, and 60-mph (97-km/h) speed limits during the day and 55-mph speed limits at night on all other roads.
99 Managing Speeds: Speed Limits state reverted to its former law or "basic rule," which affirms that "vehicles shall be driven in a careful and prudent manner, depending on the conditions at the time and place of operation."19 The issue, of course, is how drivers and law enforcement officials interpret "care- ful" and "prudent" (Figure 3-5). Variable Speed Limits20 Even when speed limits are made explicit, drivers are expected to adjust their speeds on the basis of actual conditions. Variable speed limits offer drivers guidance on appropriate maximum and minimum speed limits on the basis of real-time monitoring of prevailing traf- fic and roadway conditions, using dynamic information displays to inform motorists of the appropriate limits (Parker and Tsuchiyama 1985). Variable message signs, which provide information to motorists about speeds for specific conditions (fog, high crosswinds, work zones), have been in use for some time. Development of a new generation of technologies as part of the Intelligent Transportation Systems program21 has given new impetus to implementation of variable speed limit systems. Variable speed limits are now being used more widely, particularly on motorway systems in some European countries. For example, the Germans have an extensive system of variable speed limits, primarily to manage traffic flow under adverse environmental conditions on the autobahns. The sys- tems have reportedly been successful in reducing crash rates (Coleman et al. 1996, 24). The Dutch (Van den Hoogen and 19 Minimum fines for violation of the basic rule were increased to $70, and the level of enforcement was increased. The number of fines for violations of the basic rule increased by 88 percent to more than 5,700 for the first 9 months of 1996 compared with the same period in 1995 (Montana Department of Transportation and Montana Highway Patrol 1996, 19, 41). 20 Variable speed limits and other speed management approaches are reviewed in detail in Appendix D in the section entitled "Automated Speed Management." 21 The highway-related part of this program, whose primary objective is more effi- cient use of the existing roads, is focused on equipping both vehicles and highways with electronic controls to provide the driver with more real-time information on traffic conditions, among other objectives.
MANAGING SPEED 100 Figure 3-5 Defining a careful and prudent speed. (Reprinted with per- mission of Martin Kozlowski.) Smulders 1994) and, more recently, the British (TRL 1997)22 have introduced variable speed limits on a pilot basis on major motorways. Their primary purpose is to improve traffic flow in congested condi- tions by equalizing speeds in all lanes.23 Preliminary results indicate that, when the variable speed limits are in effect, traffic speeds are 22 Finland has also installed and is in the process of monitoring the effectiveness of a system of variable speed limit signs and message boards on a 9-mi (14-km) exper- imental section of a motorway to warn drivers of ice and other hazardous conditions (Pilli-Sihvola and Taskula 1996). 23 Variable speed limits are most effective, however, before traffic becomes heavily congested. Under heavily congested conditions, the limits are unable to affect stop- and-start driving (TRL 1997).
101 Managing Speeds: Speed Limits more uniform and--in the British pilot--automobile crashes are reduced (Van den Hoogen and Smulders 1994; TRL 1997). The results are promising, but more time is needed to determine whether these improvements can be sustained. Variable speed limit systems are not yet widely in use in the United States, but limited applica- tions are being developed as part of the Intelligent Transportation Systems program.24 Special Situations Speed limits are also developed for special situations. Advisory Speeds Engineers post advisory speeds to help drivers select safe speeds at hazardous locations, such as horizontal curves, intersections, exit ramps, or steep downgrades. The hazardous location warrants a lower speed than the general or posted speed limit, but rather than lower- ing the limit at each such location, traffic engineers post an advisory speed sign instead. Advisory speeds are not legally enforceable except under the basic speed law, which states that motorists must operate at speeds that are reasonable and prudent for conditions. Research suggests that advisory speeds have modest to little effect on driver speeds, particularly for drivers who are familiar with the road (Graham-Migletz Enterprises, Inc. 1996, 5). One reason for poor compliance is that posted advisory speeds are often set unrealistically low; the current criteria for setting advisory speeds on curves, for example, are based on vehicles and tests from the 1930s (Chowdhury et al. 1998, 32). 24 For example, the Nevada Department of Transportation in conjunction with the U.S. Department of Transportation is developing a variable speed limit system that reflects actual traffic speeds and weather conditions on a section of Interstate highway that is frequently subject to adverse weather. Deployment of the system will be accom- panied by a monitoring effort to assess effects on driving speeds and crash experience. A similar system, called Travel Aid, was recently installed at Snoqualmie Pass near Seattle, Washington, to display weather-appropriate speed limits for motorists in an effort to reduce the large number of crashes on this stretch of road (Highway and Vehicle Safety Report 1998, 8).
MANAGING SPEED 102 Nighttime Speed Limits At least four states--Montana, North Dakota, Oklahoma, and Texas--have different nighttime speed limits on certain classes of highways. The report of the Committee on Speed Regulation (NSC 1941) strongly advocated the imposition of nighttime speed limits from a public education perspective to impress on motorists the need to drive more slowly because of poorer visibility (NSC 1941, 18). The higher incidence and severity of crashes at night have also been used to support lower night speed limits (Solomon 1964, 10, 13). Today, more crashes of all types occur during daylight hours--or in dark but lighted conditions--than at night under unlighted conditions or at dawn or dusk (NHTSA 1997, 47). Thus, special nighttime limits are much less common. Moreover, such limits are considered difficult to enforce. Drivers show little inclination to decrease speeds in night- time conditions.25 Finally, lighting of highways, vehicles, and signs has improved. School and Work Zone Speed Limits Special regulatory speed limits are often used in school and work zones. Many jurisdictions establish special speed limits for streets in the vicinity of schools during certain hours of the day in response to the public perception that lower speeds improve safety (Graham- Migletz Enterprises, Inc. 1996, 5). Studies of the effectiveness of school zone limits, however, have generally found poor driver com- pliance, particularly when the limits are set very low, and no relation- ship between pedestrian crashes and the special limits (Graham-Migletz Enterprises, Inc. 1996, 5). In recent years transportation departments have begun to use reg- ulatory speed limits, rather than advisory speed warnings, in work zones. Similar to school zone limits, however, work zone speed lim- its alone26 have not proved very effective in reducing vehicle speeds 25 Data collected on free-flow average and 85th percentile speeds at six sites with speed limits ranging from 25 to 55 mph (40 to 89 km/h) found a 0- to 3-mph (5-km/h) dif- ference for daytime, nighttime, and dawn and dusk driving (Harkey et al. 1990, 44). 26 Several studies found that the presence of law enforcement officers in work zones was effective in reducing motorist speeds (Graham-Migletz Enterprises, Inc. 1996, 6).
103 Managing Speeds: Speed Limits in work zones, and only limited evaluations of their effects on safety have been conducted (Graham-Migletz Enterprises, Inc. 1996, 57). APPLICATION OF SPEED LIMITS Establishing appropriate speed limits depends to a great extent on the class of road involved and its design features, traffic density, geo- graphic location and land use, and to a lesser extent on the type of vehicles using the road. In this section, what is known about each of these factors and their effect on establishing suitable speed limits to encourage appropriate driving speeds is discussed. Roadway Functional Class and Geometric Characteristics Roadway functional class and geometric design features are among the characteristics with the greatest effect on driving speeds and the determination of appropriate speed limits. Research has shown that drivers tend to travel at higher speeds on highways with better geo- metric design characteristics regardless of posted speed limits (Garber and Gadiraju 1988, 2021). Moreover, speed dispersion was shown to decrease when the difference between design speed--a sur- rogate for geometric design characteristics--and posted speed limits is low; the lower the speed dispersion, the lower the crash rates, con- trolling for type of highway (Garber and Gadiraju 1988, 23, 28). Hence, one approach to establishing appropriate speed limits is to design or redesign roads so that their function and design character- istics are more apparent to drivers and more closely aligned with desired motorist driving speeds. The Dutch Government has offi- cially adopted a policy and implementation program known as sus- tainable road safety, one of whose primary goals is prevention of traffic crashes by rationalizing the road system. A long-term approach, the program attempts to distinguish roads more clearly by their primary function (e.g., traffic flow, traffic distribution, access) and to encourage more homogeneous use of each road, preventing large differences in vehicle speed and even separating different types of traffic where necessary (Transport Research Centre 1996, 6). The theory is that a more uniform and predictable traffic system should
MANAGING SPEED 104 enhance motorists' ability to determine appropriate driving speeds, thereby reducing the incidence of speeding and other traffic conflicts that may lead to crashes. Similar efforts to bring road design more closely in line with desired vehicle speeds are under consideration in the United States, although the extent of mileage and the diversity of conditions on U.S. highways will likely preclude as comprehensive an approach as in the Netherlands. Specifically, efforts to identify geometric design charac- teristics that influence motorists' speeds on low-speed urban streets [i.e., below 40 mph (64 km/h)] and on two-lane rural highways-- roads with some of the highest crash rates--are being pursued.27 Traffic Density Traffic density is also a key factor affecting drivers' choice of speeds and the determination of appropriate speed limits. On freeways, for example, speed-flow analyses show that average traffic speeds are rel- atively constant for a wide range of traffic volumes, slowing modestly until conditions reach breakdown levels (TRB 1998, 3-10). On two- lane rural highways, which have more limited capacity and more restricted geometric design features, travel speeds tend to deteriorate more rapidly with increasing traffic volumes. In its report, the Committee on Speed Regulation of the National Safety Council advocated setting speed limits for average traffic con- ditions where the speed limit is prima facie, thereby allowing some leeway for higher speeds when traffic is light and other conditions are favorable (NSC 1941, 15). Where a maximum speed limit is used, however, the recommendation was to set the speed limit for light traffic conditions to avoid an unreasonably low limit (NSC 1941, 15). Today, most speed limits are set for favorable conditions--light traf- fic, dry pavement, and daylight.28 Drivers are expected to exercise 27 These approaches are discussed at greater length in Chapter 5. 28 In fact, policy guides for setting speed limits in speed zones require that speed studies be conducted during off-peak traffic hours during the day in fair weather conditions when traffic speeds are unimpeded.
105 Managing Speeds: Speed Limits judgment and slow down when high traffic volumes, poor weather and visibility, or other adverse conditions are present. By using variable speed limit systems, speed limits can be adapted on a real-time basis to different traffic and environmental conditions (e.g., wet weather, reduced visibility). Although drivers already adapt their speeds to different traffic conditions, variable speed limits can provide more uniform guidance on appropriate speeds for conditions and fine-tune this guidance even on a lane-by-lane basis. Currently, the high cost of variable speed limit systems--between $0.6 million and $1.6 million U.S. per mi ($0.4 million and $1 million U.S. per km)--limits their use to high-volume or high-crash locations where environmental or traffic conditions create large fluctuations in desired speeds (Coleman et al. 1996, 57). Geographic Location and Land Use Determination of the appropriate balance between risk and travel time in setting speed limits will vary by land use as well as road class. For example, on many rural freeways and on some nonlimited-access rural roads, vehicles can travel long distances largely under free-flow- ing traffic conditions, where the opportunities for vehicle conflict (e.g., intersections, restricted sight distance) are limited and enforce- ment is difficult. These conditions suggest that vehicle operating speeds and travel efficiency should be more important in the determi- nation of speed limits for these highways than for other road classes. Urban freeways have many of the characteristics just described. However, peak-period traffic congestion, more entering and exiting traffic, and generally more frequent interchanges suggest the need for somewhat lower speed limits than on rural freeways. Variable speed limits would be appropriate for this road class to manage temporal changes in traffic demands and speeds. Many nonlimited-access rural roads, particularly two-lane roads, have restrictive roadway geometry (e.g., sharp curves). Moreover, drivers may not always be able to anticipate appropriate speeds. Speed limits should reflect these poorer conditions. Appropriate use of speed zones and advisory speed warnings can alert the driver to problem areas or to particularly hazardous locations.
MANAGING SPEED 106 Determining appropriate speed limits on nonlimited-access roads in urban areas poses a more complex problem. Urban roads often have more than one function--access as well as traffic flow. Potential for vehicle conflicts is high because of roadside development and activities, intersecting streets and driveways, parking, traffic signals, and general traffic density. Urban roads also serve a broad range of users, including pedestrians and bicyclists in addition to motor vehi- cles. Thus, speed limits must meet the objectives of this broader group of road users. Speed limits that give priority to travel efficiency will not be valid for all urban streets. As a solution, many European countries have adopted blanket urban speed limits for built-up city areas where access is the primary objective [e.g., 19-mph (30-km/h) zones in the Netherlands, Denmark, and Germany], sometimes in combination with car-free zones in central cities. These speed limit zones appear to have suc- cessfully reduced speeds and crashes within the zones,29 at least when implemented with complementary policies, such as publicity cam- paigns, engineering measures, and increased enforcement. With some exceptions, U.S. cities do not have the density of European town centers. Moreover, compliance with urban speed limits in the United States is already poor (Tignor and Warren 1989). Adoption of blanket limits would be difficult without more enforcement effort or a large enforcement tolerance. Another approach, which was recommended by the 1941 report of the NSC Committee on Speed Regulation, is to establish special higher speed zones for arterial roads whose primary function is the distribution of traffic through a metropolitan area with lower gen- eral limits for specific core business and residential areas (pp. 4950). However, this approach does not resolve the difficulty of enforcing low urban speed limits. Slowing traffic by engineering and traffic measures (e.g., speed humps, lane narrowing, turn prohibi- 29There is some evidence, however, that part of the decrease in crashes is due to the decrease in traffic volumes in the zones and diversion of traffic outside the zones. This diversionary effect needs to be studied further to assess the net safety effects of urban speed zones. See the discussion in Appendix C.
107 Managing Speeds: Speed Limits tions), known as traffic calming, has been advocated as an alternative approach to managing speed, particularly in residential areas.30 Setting appropriate speed limits on roads in rapidly developing urban fringe areas presents a special problem. The driving environ- ment in these areas is complex; traffic volumes and roadside develop- ment create the potential for vehicle conflicts and conditions that frequently approximate more congested inner-city roads. Several studies have shown that drivers--particularly motorists exiting freeways who had been driving at highway speeds--have diffi- culty adapting to the more complex environment and reducing their speed (Várhelyi 1996, 2526; Casey and Lund 1992, 135). However, lowering speed limits in fringe areas to reflect this environ- ment appears to have little effect on average speeds or the uniformity of speeds; rather, it increases the percentage of out-of-compliance drivers (Ullman and Dudek 1987, 45; Thornton and Lyles 1996, 70).31 Vehicle Characteristics Several states have adopted differential speed limits on highways with heavy-truck traffic32 to reflect the different operating charac- teristics of heavy trucks and passenger vehicles (Table 1-1). Differential speed limits are widely used in Europe, where speed governors have been required on all heavy vehicles since January 1994 (ECMT 1996, 32).33 Contrary to the perception of many motorists, passenger vehicles travel, on the average, 1 to 5 mph (2 to 8 km/h) faster than trucks on U.S. roads representing a range of different conditions and speed lim- 30 This approach is discussed in more detail in Chapter 5. 31 As Ullman and Dudek point out, lower speed limits were not accompanied by any increased enforcement or public notification, which could have changed driver com- pliance levels (p. 49). 32 Heavy trucks typically are defined as those weighing at least 26,000 lb (12 000 kg). 33 As of January 1, 1994, heavy goods vehicles over 12 tonnes entering into circula- tion in European Union member states must be fitted with a device that limits speed to 56 mph (90 km/h) and that limits passenger transport vehicles over 10 tonnes to 62 mph (100 km/h) (ECMT 1996, 32).
MANAGING SPEED 108 its (Harkey et al. 1990, 4344). In urban areas, truck speeds are con- sistently about 3 mph (5 km/h) slower than car speeds (Tignor and Warren 1989, 2). Differential speed limits recognize the different performance characteristics of these vehicles by establishing lower speed limits for heavy trucks, although in many cases the speed limit differential is 10 to 15 mph (16 to 24 km/h) (Table 1-1). Advocates of lower speed limits for heavy trucks point to their lower acceleration, more limited maneuverability, and longer stop- ping distances from a given speed34 relative to passenger vehicles. At higher speeds, these features in combination with the heavy weight of large trucks may increase crash probability and most certainly increase the severity of crashes that do occur. Opponents of differen- tial speed limits maintain that the speed differences introduced by lower limits for trucks are likely to increase the potential for vehicle conflicts from lane changes and passing maneuvers and thus increase crash probability. Several studies have been conducted on the effects of differential speed limits on speed and safety on U.S. highways, particularly since 1987 when Congress allowed states to raise the maximum speed limit to 65 mph (105 km/h) on qualifying sections of rural Interstate high- ways.35 States that raised speed limits were faced with the decision of whether to set the limits uniformly for all vehicles. Several studies reported that average truck speeds were lower in states with differential speed limits (Baum et al. 1991a, 6; Harkey and Mera 1994, 54). More important, speed dispersion increased for vehicles of all types on roads with differential speed limits, and these speed differences resulted in more interaction among vehicles and thus greater potential for conflict (Harkey and Mera 1994, 55; Garber 34 An offsetting factor in some situations, however, is the higher position of the truck driver's eyes because of the higher position of the seat in the vehicle, enabling the driver to see farther and thus to begin to brake sooner if needed. 35 On a recent Federal Highway Administration Study Tour for Speed Management and Enforcement Technology, no studies could be found of the effect of differential speed limits on speed and safety in the countries visited, which included the Netherlands, Germany, Sweden, and Australia (Coleman et al. 1996, ix).
109 Managing Speeds: Speed Limits and Gadiraju 1991, 38).36 Changes in speeds associated with differential speed limits, including average speeds and measures of speed dispersion, were statistically significant only for speed limit differentials of at least 10 mph (16 km/h). Interestingly, greater speed differences on highways with differential limits of 10 mph were also associated with a significant reduction in the number of trucks traveling at very high speeds [i.e., greater than 70 mph (113 km/h)] (Baum et al. 1991a, 6; Harkey and Mera 1994, 55). On balance, however, the evidence supports the hypoth- esis that differential speed limits of 10 mph or greater tend to increase speed differences in the traffic stream. The effects of less uniform driving speeds on safety are inconclusive. The studies that examined crash data (Garber and Gadiraju 1991; Harkey and Mera 1994) found that a greater percentage of crashes were rear-end and two-car collisions on roads with differential speed limits.37 But differences between highways with differential speed limits and highways with uniform speed limits in the total number, rate, and sever- ity of crashes were not statistically significant, suggesting no safety advantage to the former (Garber and Gadiraju 1991, 3639; Harkey and Mera 1994, 57). The results are not robust because of methodolog- ical shortcomings in study design (e.g., problems with site selection and representativeness, matching of speed with crash data).38 Thus, a strong case cannot be made on empirical grounds in support of or in opposi- tion to differential speed limits. 36 Harkey and Mera measured speed dispersion by calculating the standard deviation of speed and the coefficient of variation [i.e., standard deviation divided by the aver- age and expressed as a percentage (pp. 12, 22)]. Garber and Gadiraju measured speed dispersion using speed variance--the square of the standard deviation in speed (p. 19). Baum et al. (1991a) measured speed dispersion by observing differences in vehicle speeds at different percentiles--85th, 90th, and 95th--for states with and without differential speed limits. They did not find a statistically significant increase in speed dispersion in states with differential speed limits (p. 6). 37 Garber and Gadiraju (1991) only found significant differences (i.e., at the 95 per- cent confidence level) for two-vehicle, nontruck-nontruck crashes (pp. 3435). 38 For example, Harkey and Mera collected detailed speed data on matched pairs of highways with and without differential speed limits, but the crash data were drawn from a single broad road category--mainline rural Interstates.
MANAGING SPEED 110 EFFECTIVENESS OF SPEED LIMITS How effective are speed limits at managing traffic speeds? Two important indicators of effectiveness are traffic flow efficiency and safety ( Joscelyn and Elston 1970, 106). The effect of speed limits on traffic flow efficiency can be esti- mated by examining speed distributions and flow rates before and after speed limits have been established or changed. For example, a major objective of variable speed limits is to smooth traffic flows as traffic becomes more congested. The effectiveness of variable speed limits could be demonstrated by more uniform traffic flows and more even lane use, and such effects have been found (refer to preceding discussion of variable speed limits). It has not been demonstrated, however, that the changes have resulted in increased capacity. In heavily congested traffic, speed limits probably have little effect on traffic efficiency because most vehicles are traveling below the speed limit. Most effort has been focused on assessing the effect of changes in speed limits on safety; a large body of literature exists on this topic.39 However, empirically establishing a relationship (or the absence of one) presents a difficult task. Studies of behavior change in real- world conditions are inherently messy, and the studies of driver responses to speed limit changes are no exception. The studies must disentangle numerous factors that contribute to driver choice of speed, using data that are often imprecise and general. For example, study data on speed distributions are often highly aggregated. Ideally, speed distribution data must be closely linked with crash data to understand both whether and how actual speed changes--as opposed to changes in speed limits--have affected crash probability and out- comes. Even when speed data are good, isolating the effect of the speed change from all the other factors that affect traffic safety (e.g., changes in traffic volume, alcohol use) to establish a causal link between changes in speed limits, speeds, and crashes is extremely dif- 39 This section draws heavily on a literature survey of the effect of changes in speed limits on speed distributions and highway safety especially commissioned for the study committee, which is presented in its entirety as Appendix C.
111 Managing Speeds: Speed Limits ficult. In addition, few studies have analyzed the effects of alternative enforcement levels in combination with speed limit changes (Finch et al. 1994, 5) to assess how a key determinant of driver speed choice--enforcement--may interact with a change in the speed limit. Finally, coverage can be a problem. Many studies simply exam- ine the direct effects of speed limit changes on those highways where the limits have changed. However, because highways form a network, where a change on one part of the system is likely to affect other sys- tem links, a comprehensive analysis should take into consideration traffic diversion and spillover effects to obtain a complete under- standing of the net safety effects (McCarthy 1994, 355356).40 Recognizing these limitations, the following discussion focuses on the most methodologically sound studies of recent changes in speed limits both in the United States and abroad. Review of U.S. Studies of Changes in Speed Limits Numerous studies of the effects of the imposition of the NMSL on speed and safety were conducted in 1974. These studies are not included here because they were extensively reviewed in an earlier assessment of the effects of lowering speed limits on major highways (TRB 1984). The TRB study of the effects of the 55-mph (89- km/h) speed limit found that the lower limit reduced both travel speeds and fatalities, although driver speed compliance gradually eroded. In recent years, there have been two major changes in speed limits in the United States. In 1987 Congress allowed states to raise the NMSL on qualifying sections of rural Interstate highways to 65 mph (105 km/h) from 55 mph (89 km/h). Forty states raised their limits accordingly and numerous studies were conducted, nationally and at 40 The diversion effect refers to shifts in travel to roads where speed limits have been raised. The spillover effect refers to the adjustments resulting from the diversion effects. For example, if increased speed limits on rural Interstates have diverted traffic from roads with lower speed limits, then the remaining traffic on these lower-speed roads may be able to travel faster. The net effect on safety is ambiguous, however; the reduced traffic on the lower-speed road improves safety, but the higher speed may not.
MANAGING SPEED 112 the state level, to determine the effects of this change on traffic safety. In 1995 Congress repealed the NMSL entirely. As of the writ- ing of this report, 49 states have raised maximum speed limits and many are monitoring the effects on speed and safety. Review of Studies of 1987 Change in NMSL on Rural Interstate Highways This review included both national and state studies of the effect of the speed limit changes. For the most part, it concentrated on stud- ies that examined at least 2 years of postchange experience. Effect of Speed Limit Changes on Driver Speeds Most studies that examined the effect of speed limit changes on speed distributions provided information on several key speed mea- sures, including average traffic speeds, 85th percentile speeds, and speed dispersion (typically defined in these studies as the difference between 85th percentile speeds and average traffic speeds).41 Some of the key national studies reviewed for this report (Table 3-2) found that raising rural Interstate speed limits resulted in higher average and 85th percentile speeds on the affected highways and an increase in speed dispersion of about 1 mph (2 km/h).42 Figure 3-6 shows the National Highway Traffic Safety Administration's (NHTSA's) esti- mate of the change in the distribution of travel speeds on rural Interstate highways between 1986 and 1990 for the 18 states that raised speed limits and continued to monitor speed data (NHTSA 41 This definition of speed dispersion can be traced to the 1970 Research Triangle Institute study, one of the most careful efforts to examine the relationship between crash involvement rates and speed deviations from the average traffic speed by relat- ing the speed of vehicles involved in crashes to the actual distribution of speeds in the traffic stream at the time of the crash. Analysis of the speed data collected for that study found that one standard deviation was approximately 7 mph (11 km/h) from the average traffic speed, following a normal distribution, the same as the 85th and 15th percentile speeds, respectively (West and Dunn 1971, 5455). 42 Average speeds increased on the order of 4 mph (6 km/h) or less for a 10-mph (16-km/h) increase in the speed limit. Eighty-fifth percentile speeds increased by roughly the same magnitude (see Appendix C).
Table 3-2 Summary of Selected National Studies on Speed and Safety Effects of 1987 Speed Limit Changes on Rural Interstate Highways Authorship and Data and Analysis Date of Study Method Major Findings--Speed Major Findings--Safety Garber and 40 states that raised No speed data In 65-mph states, 15 percent median increase in fatalities on Graham 1989 speed limits by rural Interstates among the 40 states, controlling for expo- mid-1988 sure and other variables; 5 percent median increase in Pooled time series fatalities on rural non-Interstates regression (19761988) McKnight et al. Twenty 65-mph 65-mph states: 48 per- 65-mph states: 22 percent increase in fatal crashes on rural 1989 states; eight cent increase in speed Interstates; 1 percent increase in fatal crashes on other 55- 55-mph statesa (measured as speeds mph roads Quasi-experimental > 65 mph) on rural 55-mph states: 10 percent increase in fatal crashes on rural ARIMA models Interstates; 9 percent Interstates; 13 percent increase in fatal crashes on other Before/after analysis increase in speed on 55-mph roads (19821988) other 55-mph roads 55-mph states: 18 per- cent increase in speed on rural Interstates; 37 percent increase in speed on other 55-mph roads (continued on next page)
Table 3-2 (continued) Authorship and Data and Analysis Date of Study Method Major Findings--Speed Major Findings--Safety Baum et al. Forty 65-mph No speed data 65-mph states: 19 percent increase in fatalities on rural 1991b states; eight 55- Interstates relative to other rural roads, taking into mph states account changes in exposure and vehicle occupancy Before/after odds 55-mph states: no effect on odds ratio (changes on rural ratio (19821986 Interstates relative to changes on other rural roads) versus 1989) NHTSA 1989 Thirty-eight 65- 3.0-mph increase in aver- 65-mph states: 35 percent increase in fatalities between mph states; ten age speeds in 65-mph 1986 and 1988; 18 percent increase in fatality rates on 55-mph statesb states rural Interstates Before/after 3.5-mph increase in 85th 55-mph states: 9 percent increase in fatalities between 1986 (19861988) percentile speeds in 65- and 1988; 0 percent increase in fatality rates on rural Regression trend mph states Interstates analysis 0.7-mph increase in esti- (19751988) mated speed standard deviation (85th per- centile minus average speed) in 65-mph states
NHTSA 1992 Thirty-eight 65- 3.4-mph increase in aver- 65-mph states: 27 percent increase in fatalities between 1986 mph states; ten age speeds in 65-mph and 1990; 0 percent increase in fatality rates on rural 55-mph states b states Interstates Before/after 4.1-mph increase in 85th 55-mph states: 3 percent increase in fatalities between 1986 (19861990) percentile speeds in 65- and 1990; 12 percent decline in fatality rates on rural Regression analysis mph states Interstates with comparison 0.7-mph increase in esti- series mated speed standard (19751990) deviation (85th per- centile minus average speed) in 65-mph states Lave and Elias Thirty-eight 65- No speed data 65-mph states: 3 to 5 percent reduction in statewide fatality 1994 mph states; eight rates on average, controlling for the effects of long-term 55-mph states trends, exposure, safety belt laws, and economic factors Before/after analy- sis, 1986 versus 1988 Regression analysis, 19761990, using and extending Garber and Graham's data (continued on next page)
116 Table 3-2 (continued) Note: ARIMA = Autoregressive Integrated Moving Average. See Appendix C for discussion of methodology and more detailed discussion MANAGING SPEED of study results. Equivalences between miles per hour mentioned in the table and kilometers per hour are as follows: mph km/h 0.7 1.1 3.0 4.8 3.4 5.5 3.5 5.6 4.1 6.6 55 89 65 105 a Speed analysis based on nine 65-mph states and seven 55-mph states. b Speed analysis based on eighteen 65-mph states and eight 55-mph states.
117 Managing Speeds: Speed Limits Figure 3-6 Estimated changes in the distribution of rural Interstate travel speeds between the fourth quarter of 1986 and the fourth quarter of 1990 in the 18 states that raised speed limits in 1987 (NHTSA 1992, 13). l mph = 1.609 km/h. 1992, 12).43 It shows a shift toward higher average traffic speeds and a wider speed dispersion, with more vehicles traveling at higher speeds. Actual speed effects, however, differed widely by state, sug- gesting the need for better control of other factors (e.g., weather) that may have affected driver speed changes. The studies manifested two additional limitations. First, those that used comparison groups to examine the speed effects in states that raised the speed limit [65-mph (105-km/h) states] versus those 43 Changes were approximated using average and 85th percentile speed data pro- vided by the 18 states, assuming a normal distribution of travel speeds on rural Interstates (NHTSA 1992, 12). From 1986 to 1990, reported average speeds for the 18-state group increased from 60.6 to 64 mph (98 to 103 km/h); 85th percentile speeds increased from 66.6 to 70.7 mph (107 to 114 km/h); and standard deviations, measured as the difference between 85th percentile and average speeds, increased from 6.0 to 6.7 mph (9.7 to 10.8 km/h).
MANAGING SPEED 118 that did not [55-mph (89-km/h) states] made the critical assumption that the only difference between the two groups of states was the speed limit change. However, an examination of states that retained the lower 55-mph limit shows that the majority (the exceptions are Alaska and Hawaii) are located in the eastern United States, where population density, levels of congestion, and traffic volumes are different from those of the states that raised speed limits. Second, to the extent that the stud- ies examined the effects of raising the speed limit on speed distributions on other roads (i.e., the spillover effects on 55-mph highways in 65- mph states and on speed distributions on Interstate and non-Interstate highways in 55-mph states), the results are mixed.44 Effect of Speed Limit Changes on Safety Conceptually, increased average traffic speeds are associated with greater crash severity. These findings appear to be borne out by national (Table 3-2) and state studies of the safety effects of raising speed limits to 65 mph (105 km/h) on rural Interstate highways. The studies generally found that raising the speed limit led to an increase in both rural Interstate fatalities and fatal crashes. Increased crash severity, even from modest increases in speed levels, is plausible because of the sharp increase in injury severity associated with increased vehicle speeds at impact. Similar to the findings concerning the effects of speed limit changes on driving speeds, the safety effects were not uniform. For example, Garber and Graham, authors of a widely cited study (1989) that attempted to control for many other variables affecting highway safety,45 found a 15 percent overall increase in fatalities on rural Interstate highways for the 40 states that had raised speed limits. 44 For example, McKnight et al. (1989) found that in 55-mph (89-km/h) states, the percentage of drivers exceeding 65 mph (105 km/h) on rural Interstates increased by 18 percent and increased on other 55-mph highways by 37 percent. In 65-mph states, however, there were 48 and 9 percent increases, respectively. Traffic diversion could explain the large difference in the 65-mph states but not the difference in the 55-mph states. 45 The other variables included data on economic performance, seasonal effects, weekend travel, safety belt laws, and a time trend to capture the effect of changes in vehicle miles traveled.
119 Managing Speeds: Speed Limits However, the data for individual states showed that fatalities increased in 28 states and either decreased or were unchanged in 12 states.46 Similar cross-state heterogeneity was found in other studies.47 In addition, the studies reported mixed effects for fatalities and fatal crashes on non-Interstate roads in 65-mph (105-km/h) states. These results, particularly the differences among states, are not surprising given the many differences in state geographic and traffic conditions and the difficulties in modeling the effects in small states with relatively few fatalities. Garber and Graham conclude that the preponderance of the statistical evidence supports a finding of increased fatalities on rural Interstate highways in most states but acknowledge the need to identify and control for other factors that may explain the heterogeneity of effects (Garber and Graham 1989, 1516, 18). The importance of controlling for factors that could affect high- way safety other than speed limits was well illustrated by a series of follow-up studies conducted by Baum et al. (1991b) on the fatality consequences of the 65-mph (105-km/h) speed limits. Their study controlled for changes in vehicle miles traveled and vehicle occu- pancy. Without these adjustments, the study indicated that the odds of a fatality on a rural Interstate in 65-mph states in 1989 increased 29 percent over a base period (1982 to 1986). With the adjustments, the fatality risk increased by 19 percent (Baum et al. 1991b, 171). Similarly, estimates by NHTSA of increased fatalities for states that raised rural Interstate speed limits in 1987 dropped by one-third after travel increases were taken into account (NHTSA 1989, 1). The results illustrate the potential for biased estimates of speed limit effects when significant causal variables are omitted from the analysis. The effects of a change in speed limits may spill over to other roads. Some studies have suggested that these network effects result 46 The increase was statistically significant in 10 states and the decrease in 2 states (Garber and Graham 1989, 15). 47 For example, Baum et al. (1989) found an overall 18 percent increase in fatalities on rural Interstates in 65-mph (105-km/h) states. However, a state-by-state analysis showed that fatality levels decreased in 14 of the 38 states analyzed.
MANAGING SPEED 120 in increased fatalities even on roads where speed limits are not changed (Garber and Graham 1989). Others suggest that network effects can offset the adverse effects of higher Interstate speed limits and result in a neutral (McCarthy 1994) or even positive (Lave and Elias 1994) systemwide net safety effect. Lave and Elias suggest two reasons for offsetting effects. First, when speed limits were raised on qualifying rural Interstate highways in 1987, state highway patrols were able to shift their resources from monitoring these highways, as the NMSL had required, to patrolling other less safe roads. Second, some traffic may have diverted from less safe, nonlimited- access roads to the safer, but now higher-speed, rural Interstate highways. Lave and Elias (1994) looked for evidence of these effects. They found general support for increased flexibility in deployment of enforcement resources in police testimony (p. 50) and for the occur- rence of traffic diversion in comparative data on traffic growth by road type in states that had raised or retained rural Interstate speed limits (pp. 5253). Using Garber and Graham's data set and model but substituting statewide fatality rates for Interstate fatalities as the dependent variable, Lave and Elias estimated that the new speed lim- its reduced statewide fatality rates by 3.4 to 5.1 percent (p. 61). Two subsequent comments--Griffith (1995) and Lund and Rauch (1992)48--took issue with the use of statewide data as too broad a measure of network effects and questioned the validity of the traffic volume data. Garber and Graham's analysis had found evidence that spillover effects from higher rural Interstate speed limits on rural non-Interstate highways offset traffic diversion effects in most states. They reported a net median 5 percent increase in rural non-Interstate fatalities in those states in which speed limits had been raised (Garber and Graham 1989, 18). McCarthy (1994) used county-level data to examine diversion and spillover effects as well as direct effects of raising speed limits to 65 mph (105 km/h) on selected rural Interstate highways in California. The author found statistically significant increases in total, fatal, and 48 This comment critiqued an earlier version of the Lave and Elias (1994) article.
121 Managing Speeds: Speed Limits injury crashes in counties where the speed limits had been raised (McCarthy 1994, 362). In counties where speed limits had not been raised, there was no evidence of spillover effects and some evidence of improving highway safety. The net effect across all counties showed no safety decrement. The author concluded that overall safety on California roads had not been compromised by the speed limit increase, but negative effects were experienced in the counties with rural Interstate highways where speed limits had been raised (McCarthy 1994, 362363). The issue of systemwide effects, particularly the selection of an appropriate analysis unit, is an important topic that requires further research and analysis. Review of Studies Following Repeal of the NMSL in 1995 With repeal of the NMSL in 1995, the federal government no longer requires that states monitor driving speeds. However, several states that raised speed limits voluntarily collect speed and crash data on highways where the limits were raised. Reports based on these data have been released in several states and are focused primarily on Interstate highways where speed limits were raised first. NHTSA provided a report to Congress on the effects of the changes in the first year following repeal of the NMSL (NHTSA 1998).49 Finally, the Insurance Institute for Highway Safety (IIHS) released an assessment of the effects of speed limit increases on motor vehicle occupant fatalities during 1996 (Farmer et al. 1997).50 49 The NHTSA study addresses the effects of higher speeds on fatalities (not fatal- ity rates) for three groups of states: (a) 11 states (Arizona, California, Delaware, Illinois, Massachusetts, Montana, Nevada, Oklahoma, Pennsylvania, Texas, and Wyoming) that raised speed limits in late 1995 and early in the first quarter of 1996, (b) 21 states that raised speed limits through the remainder of 1996, and (c) 18 states plus the District of Columbia that did not raise speed limits (NHTSA 1998, v). 50 The IIHS study is focused on 12 states that raised maximum speed limits to at least 70 mph (113 km/h) between Dec. 8, 1995, and April 1, 1996: Arizona, California, Kansas, Mississippi, Missouri, Montana, Nevada, Oklahoma, South Dakota, Texas, Washington, and Wyoming. The comparison group included 18 states that either did not raise maximum speed limits in 1996 or raised them on less than 10 percent of urban Interstate mileage (Farmer et al. 1997, 3).
MANAGING SPEED 122 All of the studies are preliminary. At the time this report was writ- ten, most studies had accumulated only 1 year of data. Crash data, in particular, are limited; information on crash rates are often missing or preliminary, and with only 1 year of data, it is difficult to know whether the reported changes in crashes, crash rates, fatalities, and injuries represent a new trend or simply reflect normal year-to-year variations. Thus, drawing definitive conclusions from these studies is premature. However, the data they provide are provocative and worth a brief discussion here. Effect of Speed Limit Changes on Driver Speeds Most studies of speed limit changes in individual states tracked data on changes in speed, including average speeds and 85th percentile speeds, before and after the new speed limits came into effect.51 A few studies (Retting and Greene 1997; Pezoldt et al. 1997; Davis 1998; Montana Department of Transportation and Montana Highway Patrol 1996) also provided data on speed changes at the high end of the speed distribution [i.e., greater than 70, 75, and 80 mph (113, 121, and 129 km/h)]. Some state studies compared speed parameters on highways on which speed limits had been raised with those that had not. Average speeds typically increased 1 to 3 mph (2 to 5 km/h) despite larger increases in the speed limit--a minimum of 5 mph (8 km/h). The relatively small changes in average speeds compared with the change in the speed limit may reflect poor driver compliance lev- els with the lower limit in effect before the change. Eighty-fifth percentile speeds also generally increased by 1 to 3 mph (2 to 5 km/h). Thus, speed dispersion--at least as measured by the aggregate difference between the 85th percentile and the average speed--remained relatively unchanged 1 year after repeal of the NMSL. Retting and Greene (1997) found somewhat larger increases in speed standard deviations at selected locations (i.e., Riverside, 51 The NHTSA study (1998) did not report speed data. IIHS tracked speed data for selected locations in a separate study (Retting and Greene 1997).
123 Managing Speeds: Speed Limits California, and Houston, Texas) where careful "before" and "after" speed monitoring was conducted.52 A few studies found a large percentage of drivers violating the new speed limits. This suggests that some drivers expect the same enforcement tolerances of 5 to 10 mph (8 to 16 km/h) at the higher speed limits. For example, speed measurements taken on three urban freeways and one urban Interstate in Riverside, California, found that, 1 year after the speed limit was raised to 65 mph (105 km/h), 41 percent of drivers exceeded 70 mph (113 km/h)--up from 29 per- cent immediately before the change (Retting and Greene 1997, 43).53 Thus, there is some evidence that, when speed limits are raised, the distribution of traffic speeds not only shifts rightward with higher average speeds but also outward with a greater dispersion in speeds, at least at the high end of the speed distribution. Data from Montana on speed distributions on Interstate highways before and up to 9 months after the change in speed limits, although prelimi- nary, provide a good illustration of the shifts for the full range of speeds (Figure 3-7). Following the speed limit change, the range in driving speeds widened initially, and average and 85th percentile speeds reportedly increased (Montana Department of Transportation 52 Retting and Greene (1997) found that speed standard deviation had increased from 6.2 to 6.5 mph (10 to 10.5 km/h) on three urban freeways (non-Interstate) and one urban Interstate highway in Riverside immediately before and 12 months after the speed limit was raised to 65 mph (105 km/h) for cars; the limit remained at 55 mph (89 km/h) for trucks (p. 43). The increase was larger for the same time comparison on urban freeways in Houston. Speed standard deviation increased from 5.9 to 6.8 mph (9.5 to 10.9 km/h) on four urban freeways (non-Interstate) and one urban Interstate highway where the speed limit was raised to 70 mph (113 km/h) for cars and to 60 mph (97 km/h) for trucks; lower nighttime speed limits--65 mph for cars and 55 mph for trucks--were in effect (pp. 4344). 53 Retting and Greene (1997) also found that 14 percent of the drivers in Riverside exceeded 75 mph (121 km/h) 1 year after the speed limit was changed, up from 8 per- cent before the change (p. 43). Fifty percent of drivers on urban Interstates and free- ways in the Houston metropolitan area were traveling faster than 70 mph (113 km/h) 1 year after the speed limit was raised to that level, compared with 15 percent imme- diately before the new maximum speed limit took effect (Retting and Greene 1997, 44); 17 percent exceeded 75 mph 1 year after the speed limit change compared with 4 percent immediately before (Retting and Greene 1997, 44).
MANAGING SPEED 124 Figure 3-7 Daytime traffic speed distribution on Interstate highways, Montana, 1995 versus 1996 (Montana Department of Transportation and Montana Highway Patrol 1996, 13). 1 mph = 1.609 km/h. and Montana Highway Patrol 1996, 10, 15). Increased skewness or dispersion in speed distributions has been associated with a higher risk of crash involvement (Solomon 1964; Taylor 1965; Cirillo 1968; Harkey et al. 1990). Higher average speeds and 85th percentile speeds are clearly associated with greater crash severity.
125 Managing Speeds: Speed Limits Effects of Speed Limit Changes on Safety Unfortunately, data to confirm or refute changes in safety attributable to changes in speed limits are presently limited. According to NHTSA, in the first year of experience with higher speed limits, states that increased speed limits experienced approximately 350 more fatalities on Interstate facilities than would have been expected on the basis of historical trends, or about 9 percent above expecta- tions (NHTSA 1998, v).54 NHTSA regards the 9 percent increase as a lower bound.55 The agency noted that the estimated fatality increase follows historical patterns of similar increases associated with raising the speed limit, although the increase was not as large as in 1987 (NHTSA 1998, iii). All states that had increased speed lim- its in 1996 had statistically significant increases in Interstate fatalities compared with those states that had not increased speed limits (NHTSA 1998, 22). Only the "early change" group showed a statis- tically significant upward trend relative to historical trends (NHTSA 1998, 27). NHTSA concluded that without information on increased travel on higher-speed roads, shifts in travel, changes in average and top vehicle speeds, and other traffic safety factors, it was unable to determine how these other factors may have contributed to the increase in Interstate fatalities (NHTSA 1998, v). It is not clear that travel data can ever be collected that would allow changes in exposure to be separated from changes in risk.56 The NHTSA study did not address the issues of diversion and spillover effects to determine net safety effects because the data were preliminary and limited. 54 The study compared fatalities in 1996 with historical trends since 1991 for the three analysis groups (NHTSA 1998, v). 55 The current calculations use total Interstate fatalities for estimating absolute and percentage changes. However, if much less than 100 percent of Interstate mileage was affected by increased speed limits, then the baseline number of fatalities used in the denominator for computing the percentage change would be too large, and the percentage change would be too small. Also, the current analysis did not include any spillover effects on non-Interstate roads (i.e., higher speeds and crashes on these roads), which, if found and linked to raising speed limits, would increase the fatality effect (NHTSA 1998, 30). 56 Estimates of vehicle miles traveled are not tabulated according to posted speed limits. Thus, it is impossible to identify a suitable baseline of highways and travel
MANAGING SPEED 126 IIHS researchers found a larger safety decrement in their analysis of the initial experience with speed limit increases. They reported a statistically significant 16 percent increase in occupant fatalities and nearly a 17 percent increase in fatality rates for a 9-month period in 1996 on Interstate highways and freeways in 12 states that had raised maximum speed limits to 70 mph (113 km/h) by March 1996 (Farmer et al. 1997, 5, 9).57 In contrast, occupant fatalities had increased 4 percent on Interstate highways and freeways in the 18- state comparison group. (Comparable data were not reported for fatality rates.) Spillover effects, however, were small; no statistically significant differences were reported for occupant fatalities on roads other than Interstates and freeways for the two groups (Farmer et al. 1997, 10). This finding is not surprising, because many states raised speed limits on limited-access highways first. Thus, the net safety effect that was attributed to the speed limit increase was a 6 percent increase in total occupant fatalities on all roads combined (Farmer et al. 1997, 10). The IIHS study is limited to a short period. The data were insuf- ficient to analyze the effects of speed limit changes separately for each state. In addition, the comparison group approach assumes comparability between the two study groups, and, without controls for other factors that may have affected fatality rates, the estimates may be biased. Studies from individual states were limited by data availability. NHTSA's review of findings from 10 states58 that provided data con- cluded that there is some evidence of a link between higher speed limits and increases in crashes, but the effects did not follow a con- sistent pattern in all states (NHTSA 1998, 52). In many states, the data on road sections where speed limits were raised as a basis on which to compare fatality outcomes (NHTSA 1998, 1112). 57 A comparison of the effect of speed limit increases on rural and urban Interstate highways and freeways yielded mixed results. Speed limit increases on rural Interstates were associated with a statistically significant 11 percent increase in occu- pant fatalities; no statistically significant changes were found for speed limit increases on urban Interstates and freeways (Farmer et al. 1997, 9). 58 California, Idaho, Iowa, Michigan, Missouri, Montana, Nebraska, New Mexico, Texas, and Virginia.
127 Managing Speeds: Speed Limits data appeared to confirm IIHS's finding of a significant increase in crash severity on roads where speed limits were raised. In California, for example, fatal crashes and fatal crash rates increased on freeways where speed limits had been raised to 65 and 70 mph (105 and 113 km/h).59 In Texas the effect was confined to fatal and serious injury crashes on urban Interstates where the speed limit was raised from 65 to 70 mph.60 Idaho had sharp increases in speed-related crash rates on urban Interstates where speed limits were raised to 70 mph but no corresponding increase on rural Interstates where they were raised to 75 mph (121 km/h) (Idaho Department of Transportation 1997). Data from New Mexico are interesting for what they reveal about the importance of enforcement to driver compliance with speed lim- its. On two rural Interstate highways where speed limits were raised from 65 to 75 mph (105 to 121 km/h), speeds [i.e., average speeds, 85th percentile speeds, and the percentage of drivers exceeding 80 mph (129 km/h)] increased and so did crash frequency and severity.61 The increase in incapacitating injuries in multiple-vehicle crashes-- mainly rear-end and sideswipe crashes--was linked to increases in 59 Fatal crash rates on California freeways increased 22 percent [from 0.4 to 0.5 per 100 million vehicle-mi (100 MVM) (0.2 to 0.3 per 100 million vehicle-km); from 330 to 403 fatal crashes] 11 months after speed limits had been raised from 55 to 65 mph (89 to 105 km/h) compared with the same 11 months before the new limits were introduced. Similarly, fatal crash rates on freeways increased 12 percent [from 1.5 to 1.7 per 100 MVM (0.93 to 1.1 per 100 million vehicle-km); from 165 to 185 fatal crashes] 11 months after speed limits had been raised from 65 to 70 mph (105 to 113 km/h) compared with the same 11 months before the new limits were introduced (California Department of Transportation, provisional data as of Dec. 31, 1996). 60 Average monthly serious crash frequencies (defined as crashes in which at least one person was killed or suffered an incapacitating or nonincapacitating injury) increased from 36 ( Jan.Sept. 1995) to 52 ( Jan.Sept. 1996). Serious crash rates increased from 13.6 to 18.8 per 100 MVM (8.5 to 11.7 per 100 million vehicle-km) for the same periods (Pezoldt et al. 1997, 21). Increases in serious crash frequencies and rates were statistically significant. A subsequent review by the Texas Department of Transportation (unpublished data, May 16, 1997), however, attributed most of the increases to severe winter weather and other factors (e.g., drunk driving, fatigue), largely unrelated to the higher speed limit. 61 Tow-away crashes increased by 29 percent, injuries by 31 percent, incapacitating injuries by 44 percent, and fatalities by 50 percent. All of the increases are statisti- cally significant (Davis 1998, 1).
MANAGING SPEED 128 speed dispersion (Davis 1998, 2, 1617). This experience was in sharp contrast to another rural Interstate--I-10--where speed limits had also been raised but where speeds remained relatively constant and injury crashes and crash severity showed a slight decline (Davis 1998, 1). The major differences were attributed to rigorous enforce- ment and the high percentage of heavy-truck traffic on I-10, which tended to keep all vehicle speeds lower.62 Review of Studies of Changes in Speed Limits on Nonlimited-Access Highways Most U.S. studies have focused on changes in speed limits on lim- ited-access highways. A recent study (Parker 1997), however, exam- ined the effect of changes in speed limits--both increases and decreases--in short speed zones [typically less than 2 mi (3 km)] on rural and urban nonlimited-access highways. Changes in driving speeds and crash experience at these sites were compared with closely matched comparison sites where speed limits remained constant.63 The study found that changing posted speed limits had little effect on driving speeds. Specifically, a review of before and after speed data at the selected sites revealed that differences in average speeds, stan- dard deviations of speeds, and 85th percentile speeds were generally less than 2 mph (3 km/h) and were not related to the amount the posted speed limit was changed (Parker 1997, 85).64 Part of the explanation may lie in the fact that the speed limit changes--at least increases in the speed limit--simply rationalized the speeds that drivers were already driving. In fact, where speed limits were raised 62 The Doña Ana County Sherriff 's Office issued more than 1,000 citations for speeding under a grant from the Traffic Safety Bureau during the time the speed data were being recorded on I-10 (Davis 1996, 7). 63 The comparison sites could not be randomly drawn from the same population or source, but every effort was made to match as closely as possible the geometric, vol- ume, and speed characteristics of the sites where the speed limits had been changed (Parker 1997, 9). 64 The researchers noted that the changes were statistically significant, primarily because of large sample sizes, but "not sufficiently large to be of practical signifi- cance" (Parker 1997, 87).
129 Managing Speeds: Speed Limits by 10 to 15 mph (16 to 24 km/h),65 there was a fourfold increase in driver compliance levels (Parker 1997, 46). Conversely, where speed limits were lowered, compliance levels declined sharply; drivers appeared to ignore the new, lower speed limits at these sites (Parker 1997, 46). The author concluded that changing posted speed limits alone--without additional enforcement, educational programs, or other engineering measures--has only a minor effect on driver behavior (Parker 1997, 87). Not surprisingly, with such small speed changes, Parker found no evidence of changes in total crashes or fatal and injury crashes when posted speed limits were raised or lowered (Parker 1997, 86). The study findings, however, cannot be generalized to all nonlimited- access roads because the site selection process was not random.66 The lack of observed changes in driver behavior raises the concern that, if the planned speed limit changes simply legalized existing behavior, the results could be significantly biased in favor of the finding that the speed limit changes had little effect on driver behavior and thus offer little insight into the independent effect of a change in speed limits on the distribution of driving speeds.67 Nevertheless, Parker's results were confirmed in another recent study of speed limit changes for a range of road types, mainly nonlimited-access state highways (Agent et al. 1997). Data were col- lected on speeds and crashes at more than 100 speed zones in Kentucky where speed limits had been changed. In most cases, the speed limit was lowered to near 35th percentile speeds; the predominant change was from 55 to 45 mph (89 to 72 km/h). The study found modest changes in 85th percentile speeds--less than the change in the speed limit itself--whether the speed limit was raised or lowered (Agent et al. 1997, 12). Where 85th percentile speeds before the change were high relative to the new limit, modest 65 "Before" speed limit levels ranged from 20 to 50 mph (32 to 80 km/h) (Parker 1997, 9192). 66 It should be noted that safety issues and legal concerns are likely to preclude any experimental design that involves random site selection for speed limit changes. 67 For a more detailed discussion of the Parker study, see the section on Posted Speed Limits and Speeding Behavior in Appendix C.
MANAGING SPEED 130 reductions in speed were recorded but were accompanied by a high rate of noncompliance (Agent et al. 1997, 1213). The authors con- cluded that motorists will drive at what they consider an appropriate speed regardless of the speed limit (Agent et al. 1997, iii). Not sur- prisingly, given the small changes in speed, no statistically significant changes were observed in the total number of crashes or fatal or injury crashes (Agent et al. 1997, 16). The study exhibits many of the same limitations of the Parker study, mainly nonrandom selection of sites, that limit generalization of results. In addition, variability in data collection techniques for speed measurement may have affected the reliability of results. However, both studies suggest the need for reasonable speed limits and the difficulty of changing driver behavior where drivers perceive that an appropriate speed is other than the posted speed limit. Review of International Experience on the Effects of Changes in Speed Limits In addition to the U.S. research on the relationship between changes in speed limits and highway safety, a number of international studies have examined this issue.68 International studies of the effects of changes in speed limits on low-speed roads are numerous and were summarized briefly earlier in the chapter. This section focuses on a more limited number of recent studies of speed limit changes on high- speed roads. In contrast to the United States, where most studies have evaluated the speed and safety effects of raising speed limits on limited-access highways, international studies have primarily examined the effects of reductions in speed limits. Studies of speed limit reductions in Sweden (Nilsson 1990; Johansson 1996), the Netherlands (Borsje 1995), Victoria, Australia (Sliogeris 1992), and Finland (Salusjärvi 1981) all reported results that are a mirror image of those found in the United States. Lower speed limits resulted in lower average speeds, although 68 This section also draws heavily on the review commissioned for this study, which is presented in its entirety as Appendix C.
131 Managing Speeds: Speed Limits the changes were typically less than the absolute reduction in the speed limit. Lower speed limits were also associated with reduced crash inci- dence and, in some cases, with reduced crash severity. Many of the studies, however, do not control for the potentially confounding effects of other policies undertaken at the same time as the speed limit change (e.g., public information campaigns, increased levels of enforcement) or other factors that may have affected highway safety (e.g., changes in amount of travel affecting exposure levels, safety belt legislation). Most studies failed to consider systemwide effects of speed limit changes to determine net safety effects. Any generalization of the results to the United States, of course, must be mindful of differences in highway networks, driving environment, and driving culture (there is more legal high-speed driving in European countries). SUMMARY The potential adverse consequences of speeding, particularly the risks imposed on others from an individual driver's speed choice, are suffi- cient reason for regulating speed. Speed limits, one of the oldest methods of managing speeds, are intended to enhance safety by establishing an upper bound on speed to reduce both the probability and the severity of crashes. They also have a coordinating function; the intent is to reduce dispersion in driving speeds and thus reduce the potential for vehicle conflicts. Numerous methods are available for setting speed limits, ranging from legislated limits on broad road classes, to limits in speed zones determined on the basis of an engineering study, to limits established by local ordinance on residential streets. Whatever method is used, speed limits reflect implicit trade-offs among road user safety, travel efficiency, and practicality of enforcement. The trade-offs vary by roadway functional class and environment, reflecting in part different levels of risk associated with driving on different roadway types. Setting speed limits that give priority to travel efficiency, for example, may be appropriate on rural freeways where vehicles travel long distances under free-flowing traffic condi- tions with little likelihood of conflict with other road users and where the ability to enforce speed on extensive road mileage is limited. A
MANAGING SPEED 132 maximum speed limit is probably necessary, however, because of the cause-and-effect relationship between high speeds and crash severity. Speed limits that give priority to travel efficiency are less likely to be appropriate in urban areas, where the roads must be shared with a broad range of users, including vulnerable pedestrians and bicyclists, and where roadside development increases opportunities for vehicle conflict and raises the probability of an unexpected event. Because driver com- pliance with urban speed limits has been poor, alternative methods for managing and enforcing speeds may be necessary in these areas. Most roads and roadway conditions fall between these extremes. Appropriate use of speed zones can help establish speed limits suit- able for conditions. The effects of speed limits, particularly on safety, have been stud- ied extensively. U.S. experience with raising speed limits on qualified sections of rural Interstate highways in 1987 suggests that higher speed limits resulted in higher average and 85th percentile speeds and modest increases in speed dispersion. Higher speeds are linked unequivocally with increased injury severity in a crash. Indeed, the most methodologically sound studies found that higher speeds led to increased fatalities and fatal crashes on rural Interstates in most states. The studies were less clear about the absolute size of the safety decrement, the extent and direction of any network effects, and the role of enforcement in encouraging driver compliance with new speed limits. Preliminary data are available for speed and safety changes, pri- marily on limited-access highways, in the first year following repeal of the NMSL in 1995. Average and 85th percentile speeds rose less than increases in the posted limit, reflecting, in part, poor driver compliance with lower speed limits in effect before the change. Speed dispersion increased in some states but not in others, in part depending on what measure of speed dispersion was used. Monitoring studies show some evidence of more high-speed driving at levels that exceed the new speed limits, suggesting that at least some drivers expect the same enforcement tolerance as at the lower speed limits. Although the findings are not consistent across all states, most studies indicated an increase in fatalities on highways where speed limits were raised. Only one study examined possible
133 Managing Speeds: Speed Limits system effects, finding modest spillover effects. The results of these studies must be considered preliminary because they are generally based on 1 year of data or less. The available studies of speed limit changes in speed zones on nonlimited-access highways in the United States found little change in speeds or crashes when speed limits were raised to near the 85th percentile speed or lowered to a limit well below the 85th percentile. Although methodological problems limit the generalization of study results, the findings suggest the difficulty of changing driver behav- ior merely by changing the sign. Speed limits are likely to have more effect if the majority of driv- ers perceive the limits as reasonable and related to the risks of driv- ing, including risks to other road users. The increased probability of a severe crash is sufficient reason to impose maximum speed limits and direct enforcement at motorists who drive well in excess of the speed limit. Speed limits alone, without enhanced enforcement or innovative engineering measures, are insufficient to achieve compli- ance with posted speed limits on many roads. In most cases, enforce- ment is critical to ensure driver compliance. Appropriate enforcement strategies--the subject of the next chapter--can help persuade drivers that speed limits are in fact legal limits and not sim- ply guidance on appropriate driving speeds. REFERENCES ABBREVIATIONS ECMT European Conference of Ministers of Transport ITE Institute of Transportation Engineers NHTSA National Highway Traffic Safety Administration NSC National Safety Council RTI Research Triangle Institute TRB Transportation Research Board TRL Transport Research Laboratory UVC Uniform Vehicle Code Agent, K.R., J.G. Pigman, and J.M. Weber. 1997. Evaluation of Speed Limits in Kentucky. KTC-97-6. Kentucky Transportation Center in cooperation with the Kentucky Transportation Cabinet and the Federal Highway Administration. April, 90 pp.
MANAGING SPEED 134 Baerwald, J.E. 1953. Indiana Traffic Speeds 19421952. Joint Highway Research Project. No. 73. Purdue University, Lafayette, Ind., March. Baum, H.M., A.K. Lund, and J.K. Wells. 1989. The Mortality Consequences of Raising the Speed Limit to 65 MPH on Rural Interstates. American Journal of Public Health, Vol. 79, No. 10, Oct., pp. 13921395. Baum, H.M., J.R. Esterlitz, P. Zador, and M. Penny. 1991a. Different Speed Limits for Cars and Trucks: Do They Affect Vehicle Speeds? In Transportation Research Record 1318, Transportation Research Board, National Research Council, Washington, D.C., pp. 37. Baum, H.M., J.K. Wells, and A.K. Lund. 1991b. The Fatality Consequences of the 65 MPH Speed Limits, 1989. Journal of Safety Research, Vol. 22, No. 4, Winter, pp. 171177. Borsje, J.F. 1995. The Effects of the 1988 Differentiation of Speed Limits in The Netherlands. No. 2A, Part 4. Proc., Road Safety in Europe and Strategic Highway Research Program (SHRP), Lille, France, Sept. 26-28, 1994, pp. 2436. Carter, F.M. 1949. Speed Zoning. Proc., First California Institute on Street and Highway Problems, Berkeley, Calif., Jan. 31-Feb. 2, pp. 151156. Casey, S.M., and A.K. Lund. 1992. Changes in Speed and Speed Adaptation Following an Increase in the National Maximum Speed Limit. Journal of Safety Research, Vol. 23, pp. 135146. Chowdhury, M.A., D.L. Warren, H. Bissell, and S. Taori. 1998. Are the Criteria for Setting Advisory Speeds on Curves Still Relevant? ITE Journal, Vol. 68, No. 2, Feb., pp. 3245. Cirillo, J.A. 1968. Interstate System Accident Research--Study II--Interim Report II. Public Roads, Vol. 35, No. 3, 1968. Coleman, J.A., et al. 1996. FHWA Study Tour for Speed Management and Enforcement Technology. FHWA-PL-96-006. Federal Highway Administration, U.S. Department of Transportation, Feb., 83 pp. Davis, J.W. 1996. The Effects of Higher Speed Limits in New Mexico. Draft. Grant 1-TR- 97-01. University of New Mexico, Jan. Davis, J.W. 1998. The Effects of Higher Speed Limits in New Mexico. Grant 98-TR-01- 01. Division of Government Research, University of New Mexico, Jan., 19 pp. Donald, D. 1994. Using Expert System Technology to Set Appropriate, Consistent Speed Limits--The Australian Experience. In Proc., Third International Conference on Safety and the Environment in the 21st Century: Lessons from the Past, Shaping the Future, Tel Aviv, Israel, Oct., pp. 284295. ECMT. 1996. Speed Moderation. Organisation for Economic Cooperation and Development. OECD Publications Service, Paris, 86 pp. Farmer, C.M., R. A. Retting, and A.K. Lund. 1997. Effect of 1996 Speed Limit Changes on Motor Vehicle Occupant Fatalities. Insurance Institute for Highway Safety, Arlington, Va., Oct., 12 pp. Fildes, B.N., and S.J. Lee. 1993. The Speed Review: Road Environment, Behaviour, Speed Limits, Enforcement and Crashes. Monash University Accident Research Centre, Victoria, Australia, Sept., 146 pp.
135 Managing Speeds: Speed Limits Finch, D.J., P. Kompfner, C.R. Lockwood, and G. Maycock. 1994. Speed, Speed Limits and Accidents. Project Report 58. S211G/RB. Transport Research Laboratory, Berkshire, England, 24 pp. Fitzpatrick, K., R.A. Krammes, and D.B. Fambro. 1997. Design Speed, Operating Speed and Posted Speed Relationships. ITE Journal, Vol. 67, No. 2, Feb., pp. 5259. Garber, N.J., and R. Gadiraju. 1988. Speed Variance and Its Influence on Accidents. University of Virginia, Charlottesville, July, 56 pp. Garber, N.J., and R. Gadiraju. 1991. Impact of Differential Speed Limits on Highway Speeds and Accidents. University of Virginia, Charlottesville, Feb., 52 pp. Garber, S., and J.D. Graham. 1989. The Effects of the New 65 Mile-Per-Hour Speed Limit on Rural Highway Fatalities: A State-By-State Analysis. DOT-HS-807- 452. National Highway Traffic Safety Administration, U.S. Department of Transportation, July, 34 pp. Godwin, S.R. 1988. Implications of Raising the U.S. Maximum Speed Limit to 65 mph. VTI Rapport 332A. Proc., Roads and Traffic Safety on Two Continents, Gothenburg, Sweden, Sept. 911, 1987, pp. 2143. Godwin, S.R. 1992. Effect of the 65 m.p.h. Speed Limit on Highway Safety in the U.S.A. Transport Review, Vol. 12, No. 1, pp. 114. Graham-Migletz Enterprises, Inc. 1996. Procedure for Determining Work Zone Speed Limits. Research Results Digest, No. 192, National Cooperative Highway Research Program, Sept., 43 pp. Griffith, M.S. 1995. Discussion of "Did the 65 mph Speed Limit Save Lives?" (with response by C. Lave). News, Queries and Comment, Accident Analysis and Prevention, Vol. 27, No. 1, pp. 137140. 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. Harkey, D.L., and R. Mera. 1994. Safety Impacts of Different Speed Limits on Cars and Trucks. FHWA-RD-93-161. The Scientex Corporation, Charlotte, N.C., May, 67 pp. Harwood, C.J. 1995. Criteria for Setting General Urban Speed Limits. Road and Transport Research, Vol. 4, No. 2, June, pp. 8894. Highway and Vehicle Safety Report. 1998. Speed Limits Change with Weather in Seattle. Vol. 24, No. 8, Jan. Idaho Department of Transportation. 1997. Speed and Accident Report for MayOctober, Six-Month Report Comparing MayOctober for 1991 Through 1996. Jan. 8. ITE. 1992. Traffic Engineering Handbook: Fourth Edition ( J.L. Pline, ed.). Prentice- Hall, Englewood Cliffs, N.J. Johansson, P. 1996. Speed Limitation and Motorway Casualties: A Time Series Count Data Regression Approach. Accident Analysis and Prevention, Vol. 28, No. 1, pp. 7387. Jondrow, J.M., M. Bowes, and R.A. Levy. 1982. Optimal Speed Limit: A New Approach. In Transportation Research Record 887, Transportation Research Board, National Research Council, Washington, D.C., pp. 12. Joscelyn, K.B., and P.A. Elston. 1970. Maximum Speed Limits--Volume II, The Development of Speed Limits: A Review of the Literature. FH-11-7275. Indiana University, Bloomington, Oct., 146 pp.
MANAGING SPEED 136 Krammes, R.A., K. Fitzpatrick, J.D. Blaschke, and D.B. Fambro. 1996. Speed: Understanding Design, Operating, and Posted Speed. Report 1465-1. Texas Transportation Institute, College Station, March, 16 pp. Labatut, J., and W. Lane. 1950. Highways in Our National Life: A Symposium. Princeton University Press. Ladd, W.D. 1959. Organizing for Traffic Safety in Your Community. Charles C. Thomas, Springfield, Ill. Lave, C. 1985. Speeding, Coordination, and the 55-mph Limit. The American Economic Review, Vol. 75, pp. 11591164. Lave, C., and P. Elias. 1994. Did the 65 MPH Speed Limit Save Lives? Accident Analysis and Prevention, Vol. 26, No. 1, pp. 4962. Lund, A.K., and W.J. Rauch. 1992. On the Contrary! A Comment on Lave and Elias's Question "Did the 65 mph Speed Limit Save Lives?" Insurance Institute for Highway Safety, Arlington, Va., 5 pp. MacRae, D., Jr., and J.A. Wilde. 1979. The Optimum Speed Limit. In Policy Analysis for Public Decisions, Duxbury Press, Mass., pp. 133156. Marcellis, J.C. 1962. An Economic Evaluation of Traffic Movement at Various Speeds. Vehicular Speed Regulation Research Project, Department of Civil Engineering, University of Illinois, Sept. McCarthy, P.S. 1994. An Empirical Analysis of the Direct and Indirect Effects of Relaxed Interstate Speed Limits on Highway Safety. Journal of Urban Economics, Vol. 36, pp. 353364. McKnight, A.J., T.M. Klein, and A.S. Tippetts. 1989. The Effect of the 65 MPH Limit on Speeds and Accidents. DOT-HS-807-463. National Public Services Research Institute, Landover, Md., National Highway Traffic Safety Administration, U.S. Department of Transportation, Aug., 48 pp. Montana Department of Transportation and Montana Highway Patrol. 1996. Data Reference Related to Montana Speed Laws. Dec. NHTSA. 1989. Effects of the 65 mph Speed Limit Though 1988: A Report to Congress. U.S. Department of Transportation, Oct. NHTSA. 1992. Effects of the 65 mph Speed Limit Through 1990: A Report to Congress. U.S. Department of Transportation, May. NHTSA. 1997. Traffic Safety Facts 1996. DOT-HS-808-649. U.S. Department of Transportation, Dec. NHTSA. 1998. The Effect of Increased Speed Limits in the Post-NMSL Era. Report to Congress. DOT-HS-808-637. NRD-31. U.S. Department of Transportation, Feb. Nilsson, G. 1990. Reduction in the Speed Limit from 110 km/h to 90 km/h During Summer 1989. VTI Rapport 358A, Swedish Road and Traffic Research Institute, 36 pp. NSC. 1941. Speed Regulation. Committee on Speed Regulation, Chicago, Ill., July, 62 pp. Oppenlander, J.C. 1962. A Theory on Vehicular Speed Regulation. Bulletin 341, Highway Research Board, National Research Council, Washington, D.C., pp. 7791. 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.
137 Managing Speeds: Speed Limits Parker, M.R., Jr., and K.H. Tsuchiyama. 1985. Methods for Reducing Large Speed Differences in Traffic Streams, Volume 1 - Inventory of Methods. FHWA/RD-85/103. Martin R. Parker & Associates, Inc., Canton, Mich., Aug., 124 pp. Pezoldt, V.J., R. Q. Brackett, and D.E. Morris. 1997. The Impact of the Change in Speed Limits on Texas Highways: A Nine-Month Evaluation. Texas Transportation Institute, Texas A&M University System, College Station, Feb. 19. Pilli-Sihvola, Y., and K. Taskula. 1996. Mustaa Jäätä and Finland's Weather- Controlled Road. Traffic Technology International 1996, UK and International Press, Surrey, United Kingdom, pp. 204206. Retting, R.A., and M.A. Greene. 1997. Traffic Speeds Following Repeal of the National Maximum Speed Limit. ITE Journal, Vol. 67, No. 5, May, pp. 4246. Rosen, H.S. 1995. Public Finance (4th edition). Richard D. Irwin, Inc. 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. Salusjärvi, M. 1981. The Speed Limit Experiments on Public Roads in Finland. Road and Traffic Laboratory, Technical Research Centre of Finland, Dec., 133 pp. Sliogeris, J. 1992. 110 Kilometre Per Hour Speed Limit--Evaluation of Road Safety Effects. GR 92-8. Vic Roads, Kew, Victoria, Australia, Aug., 27 pp. 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. Sub-Committee on Speed Zoning. 1969. Resolution of the Annual Meeting of the American Association of State Highway Officials. Taylor, W.C. 1965. Speed Zoning, A Theory and Its Proof. Traffic Engineering, Vol. 34, No. 12, Jan., pp. 1719, 4850. Tennessee Department of Highways. 1968. A Summary of Practices in Establishing Speed Zones. Traffic Engineering Division, June. The American City. 1950. What Do Speed Limits Mean? Vol. 65, No. 11, Jan., p. 121. Thornton, M., and R.W. Lyles. 1996. Freeway Speed Zones: Safety and Compliance Issues. In Transportation Research Record 1560, Transportation Research Board, National Research Council, Washington, D.C., pp. 6673. Tignor, S.C., and D.L. Warren. 1989. Speed Zoning in America: Some Preliminary Research Results. Federal Highway Administration, U.S. Department of Transportation, June. Transport Research Centre. 1996. Toward Safer Roads: Opportunities for a Policy to Bring About a Sustainably Safe Traffic System. Ministry of Transport and Public Works, The Netherlands, 16 pp. TRB. 1984. Special Report 204: 55: A Decade of Experience. National Research Council, Washington, D.C., 262 pp. TRB. 1998. Special Report 209: Highway Capacity Manual. Third Edition, 1997 Update. National Research Council, Washington, D.C. TRL. 1997. M25 Controlled Motorways Monitoring: Executive Summary. Crowthorne, Berkshire, England, Feb., 4 pp. Ullman, G.L., and C.L. Dudek. 1987. Effects of Reduced Speed Limits in Rapidly Developing Urban Fringe Areas. In Transportation Research Record 1114, Transportation Research Board, National Research Council, Washington, D.C., pp. 4553.
MANAGING SPEED 138 UVC. 1967. Rules of the Road with Statutory Annotations. Van den Hoogen, E., and S. Smulders. 1994. Control by Variable Speed Signs: Results of the Dutch Experiment. Conference Publication 391, Road Traffic Monitoring and Control, April 2628, IEE, pp. 145149. Várhelyi, A. 1996. Dynamic Speed Adaptation Based on Information Technology--A Theoretical Background. Bulletin 142. Lund University, Lund Institute of Technology, Sweden, 220 pp. West, L.B., Jr., and J.W. Dunn. 1971. Accidents, Speed Deviation and Speed Limits. Traffic Engineering, Vol. 41, No. 10, pp. 5255.