Better to see Chapter 9 Driver behavior of Traffic Safety  (2004)


Words only (no formatting, figures, tables, or photographs) from 1991 book



Paperback copy of complete unchanged book available from , list price $29.95


Chapter 6.  DRIVER BEHAVIOR (From 1991 book Traffic Safety and the Driver)


                Driver performance, as discussed in Chapter 5, refers to the driver's perceptual and motor skills, or what the driver can do.  Driver behavior refers to what the driver in fact does do.  The example represented in Fig. 5-4 illustrates how the distance covered depends on driver reaction time and vehicle speed.  What is not encompassed in this figure is why one driver chooses to travel at one speed, while another chooses a different speed.  The ability to judge the speed, control the vehicle at that speed, and react to hazards are all in the realm of driver performance.  The speed chosen is in the realm of driver behavior.

                 As driver performance focuses on capabilities and skills, it can be investigated by many methods, including laboratory tests, simulator experiments, tests using instrumented vehicles and observations of actual traffic.  As driver behavior indicates what the driver actually does, it cannot be investigated in laboratory, simulator or instrumented vehicle studies.  As a consequence, information on driver behavior tends to be more uncertain than that about driver performance.

                 The distinction between performance and behavior is one of the most central concepts in traffic safety.  This is because driving is, in Na?"a?"ta?"nen and Summala's [1976] phrase, a "self-paced" task.  That is, drivers choose their own desired levels of task difficulty.  The acquisition of increased skill is likely to lead to an increase in the level of task difficulty, such as driving faster, overtaking in heavier traffic, or
accepting additional secondary tasks like listening to the radio, rather than simply to an increase in safety.  When task difficulty is maintained constant, increases in skill are likely to lead to increases in safety.  Ha?"kkinen [1979] finds that the crash rates of Helsinki bus and streetcar drivers are strongly correlated with a series of performance-measuring tests.  He also finds that the drivers' crash rates are stable over long periods; the time span for the entire study is 1947-1973.  More recently, Lim and Dewar [1989] find higher information processing ability, as measured in laboratory tests, is associated  with lower on-the-job crash rates for bus drivers in Calgary, Canada.  McKenna, Duncan, and Brown [1986] find no statistically significant effect in a similar study of British bus drivers, although the nominal effect is in the same direction.  The basic distinction between the professional drivers in the studies above and drivers in general is that the schedules, and other aspects of driving behavior for professional drivers, is specified.  Constraints militate against such driving being self-paced, so that increased skill can be expected to produce fewer crashes, as is found.  Car driving in a regulated structured environment has much in common with the task of piloting a commercial air liner.  For both tasks, increased skill, knowledge and performance are expected to increase safety.  However, this may not be so for the self-paced task of normal driving.




                 The belief that increased skill would lead to lower crash involvement rates seems to many so intuitively obvious that it should be superfluous even to investigate it.  Such a belief nurtures the view that driver education necessarily increases safety.  It is widely held by driving affectionadoes, especially the racing fraternity, that race drivers have fewer crashes than
the average driver.  For example, in discussing the on-the-road experience of race drivers, an editor of "Road and Track" magazine writes, "I have for many years claimed that the licensed racer is far safer than ordinary chaps, on grounds of practiced skills, mental ability, cognizance of hazards in driving, keen interest in driving well, and so on." [Girdler 1972, p. 98, as cited by Williams and O'Neill 1974]  The belief that superior skills lead to reduced crashes led to the concept of a "Master Driver's License" which would entitle those with proven high driving skills to various privileges denied the average driver.  The National Highway Traffic Safety Administration [1972] at one time addressed harnessing this concept in its efforts to reduce crashes.

                 In order to discover if unusually skilled drivers really did have different on-the-road driving records from the average driver, Williams and O'Neill [1974] obtained names and addresses of national competition license holders from the Sports Car Club of America.  They compared the on-the-road driving records of these license holders (referred to in their paper as racing drivers) in Florida, New York and Texas, to comparison groups of drivers in the same states matched in such characteristics as sex and age. There are minor variations in the details of the matching procedures between the states.

                 The results of the study are summarized in Fig. 6-1, which displays the rate for the racing drivers divided by the corresponding rate for the comparison drivers.  If there were no differences between the groups of drivers, then these ratios would all be close to one, whereas if the racing drivers had lower crash and violation rates, the ratios would be less than one.  What in fact is found is that in all 12 cases studied, the rates for the racing drivers exceed those for the comparison drivers, in most cases by considerable amounts.  Thus, on a per year basis, the racing drivers had substantially more crashes, and more violations, especially speeding violations.  Self-reported estimates of distance of travel indicate that the
racing drivers travelled more than the comparison drivers; however, additional analyses by Williams and O'Neill [1974] indicate that this does not explain all of the difference observed.  What is unambiguous from the study is that the possession of a national competition license is associated with higher crash and violation rates per year.


Fig. 6-1 about here


                 In interpreting the difference between the driving records of the race drivers and the comparison drivers, it is not possible to determine whether the effect flows from the use of the additional skill acquired by the drivers to drive more aggressively, or whether it is simply high-risk drivers who are attracted to racing.  Perhaps, without the additional skills acquired in pursuit of their advanced license, they might have had yet higher crash rates.  The study does show that higher skill levels are not necessarily associated with lower crash rates. 




                 The dependence of various traffic-related quantities on driver sex and age offers an indirect probe of the relationship between skill, behavior and safety.  These variables have the advantage that they are unambiguously known for large numbers of individuals.  Jonah [1986] gives an excellent review of much more material relating to the relationship between crash risk and risk-taking behavior than presented below.


Crash rates


                 Fig. 6-2 shows car driver involvements in severe crashes (sufficient severity to kill an 80 year-old male driver) per unit distance of travel versus sex and age; this figure differs from Fig. 2-10 only in that it refers to car drivers only rather than to all drivers.  At all ages, involvement rates are higher for males than for females.  The higher levels of interest in cars and driving traditionally exhibited by males does not lead to lower crash rates even though it may lead to higher levels of skill and knowledge.  The crash rate at age 40 is about one sixth what it is at age 20.  Although some increase in skill, especially higher level information processing, may contribute to a decreased crash rate with age, it seems implausible that it could generate more than a small fraction of this large effect.  In terms of such performance measures as visual acuity and reaction time, the performance of younger drivers is markedly superior to that of older drivers.  The higher involvement rates of younger, and male, drivers seem more related to how they are choosing to drive, particularly their propensity to take driving risks, than to their abilities at the driving task. 


Fig. 6-2 about here


                 Marital status has been shown to have a large influence on number of crashes per year.  Peck, McBride, and Coppin [1971] find that single males have higher crash rates than married males of the same age, based on analyzing 1961-1963 California data.  An even larger and more systematic effect is found for females -- the unmarried female rate exceeds the married female rate at all ages, typically by a factor of between one and a half and two.  The unmarried female rate is still less than that for the married male. 
Essentially similar patterns are observed for traffic violations.  Part of these differences is almost certainly due to differences in amounts of driving -- in the early 1960's married females probably drove substantially less than unmarried ones.


Observational effects


                 Driver propensity to take risk as indicated by chosen speed and following headway (gap between a driver's vehicle and one in front) was measured in a series of observational studies [Wasielewski 1984; Evans and Wasielewski 1983] in which oncoming cars were photographed from freeway overpasses.  The license-plate number, read from the photograph, was used to extract from state files the driving record, sex, and age of the registered owner.  A photographed driver judged not to differ in age or sex from the registered owner was assumed to be the owner.

                 Fig. 6-3 shows results for Wasielewski's [1984] study, in which the measured variable associated with risk taking is travel speed on a rural two-lane road.  A systematic decline in speed with increasing age is apparent.  Fig. 6-4 shows results from Evans and Wasielewski's [1983] study in which the measured variable associated with risk taking is following headway.  The quantity plotted is the reciprocal of headway, so that, as for the speed case, larger values indicate higher levels of risk taking.  Again, a decrease in the measure of risk taking is apparent with increasing age.  The observational data are insufficient to permit plotting separate relations for males and females.  However, analyses in each of the original papers cited finds higher levels of risk taking associated with male than with female drivers.  A study examining factors present in urban crashes in Leeds, UK, finds, "Driving too
fast was more common for males than for females, and more common for younger drivers than older drivers" [Carsten, Tight, and Southwell 1989].


Fig. 6-3 about here



Fig. 6-4 about here


Sex differences in activity level or risk taking


                 Evidence of clear differences between the behavior of males and females is provided in Fig. 6-5.  What is shown is the number of male pedestrian fatalities per capita divided by the number of female pedestrian fatalities per capita, based on FARS data from 1981 through 1985 and US census data for the same years; there is accordingly very little uncertainty in the data.  At all ages, male pedestrian fatalities per capita exceed those of females.  For the first point plotted, for age from birth to 11 months 31 days, there are 32 male compared to 17 female pedestrian fatalities; for the second point, from age 1 year to age 2 years, there are 144 male compared to 99 female pedestrian fatalities.  These proportionate differences far exceed any difference in numbers in the population (already corrected for in the comparison in Fig. 6-5), and are larger than can be explained by the greater vulnerability of male than female babies to death from the same impact (Chapter 2).


Fig. 6-5 about here


                 It is not possible to infer from Fig. 6-5 whether the differences between the sexes result from differences in exposure, or differences in risk taking
for each exposure.  Either interpretation points to large differences in behavior between the sexes.  Either males cross roads more often than females, or else they are subject to a greater risk per crossing.  Howarth [1985] provides clear evidence based on UK data that boys do in fact have higher risks of being involved in a pedestrian crash per road crossing than do girls. His results (Fig. 6-6) show the number of children involved in pedestrian crashes divided by an estimate of the number of children crossing roads, based on extensive observations of child-pedestrian behavior.


Fig. 6-6 about here


                 It is not possible to determine the extent to which the large systematic differences in Fig. 6-5 reflect intrinsic differences between males and females as distinct from socially induced differences.  It is, for example, possible that higher pedestrian fatality rates for males at age 0, 1, 2, 3, .... 12 reflect different parental treatment of boys and girls.  Although such an explanation is logically possible, I find it implausible for the case of babies.  After the mid-teen years differences in Fig. 6-5 become much larger, being typically a factor of three between age 20 and 60.  These differences could reflect the greater social freedom which society has traditionally afforded males, enabling them to spend more time as pedestrians and also as drivers.  Such differences might diminish, or disappear, as distinctions between the roles of the sexes in society change; there is already evidence of increased female arrests for drunk driving even as male arrests decline [Popkin et al. 1986].  The differences in Fig. 6-5 appear to reflect some greater activity level, and/or greater propensity to take risk, by males compared to females.  Although the data have nothing to do with driving risk, as such, the interpretation in terms of greater activity level or risk taking
on the part of males supports a similar interpretation of the earlier data showing higher driver crash rates for males than for females.




                 Fig. 6-7 (top) shows involvements in severe (sufficient severity to kill an 80 year-old-male driver) single-car crashes per million population.  Single-car crashes are chosen because they are most representative of the behavioral aspects of crash involvement, in that they involve no driver other than the subject driver.  The lower figure shows the number of arrests per thousand population versus sex and age, based on data from FBI Uniform Crime Reports [US Department of Justice 1985].  The data plotted refer to crimes unrelated to driving -- data on such offenses as driving while intoxicated are excluded.  Hence, activity nominally unrelated to anything to do with driving is displayed.  Top and bottom curves both show incidents per capita per year.  One of the reasons why the male crash rate is so much higher than the female crash rate is that males are more exposed to the risk of the crash because they drive greater distances than do females.  Presumably, the reason more males than females are arrested is because they also are exposed more, in that they presumably commit more crimes.  Fig. 6-7 should not be interpreted to mean that males are more likely than females to be arrested per burglary; likewise, it does not imply that males have crash involvements per unit distance of travel greater than those of females in the proportion shown in the figure.


Fig. 6-7 about here


                 The similarity between the two plots suggests that involvement in severe crashes and being arrested for offenses unrelated to traffic both flow in large measure from common explanatory factors.  No one would suggest that the lower arrest rate for 40-year olds compared to 20-year olds occurs because the 40-year olds have at long last learned how to not commit burglaries!  This should invite a parallel caution against interpreting lower crash rates for 40-year-old drivers compared to those for 20-year-old drivers to mean that the 40 year-old drivers have simply learned how to not crash.  It has been facetiously suggested that the crime rate curve is not reflecting decreased involvement in crime with increasing age, but increased skill at avoiding arrest.  The most compelling and common-sense interpretation of the similarity between the two curves in Fig. 6-7 is that involvement in severe crashes and being arrested for offenses unrelated to driving each have explanatory factors in common, and that increasing skill and knowledge is not the major factor in the large decline in crash rates with increasing age. 

                 The type of age dependence shown in Fig. 6-7 seems fairly intrinsic to criminal activity, and also applies to homicide victims [Baker, O'Neill, and Karph 1984].  Gottfredson and Hirschi [1986] write: "The propensity to commit criminal acts reaches a peak in the middle to late teens and then declines rapidly throughout life.  Further, this distribution is characteristic of the age-crime relation regardless of sex, race, country, time, or offense.  Indeed, the persistence of this relation across time and culture is phennomenal."  Although Steffensmeier et al. [1989] document much variability around the claimed pattern (for example, more fraud is committed by 50-year olds than by 20-year olds), the pattern does reasonably well fit a whole range of criminal activity, and also involvement in severe traffic crashes.

                 Direct association between criminal involvement and traffic crashes is provided by Haviland and Wiseman [1974] who compare the driving records of 114
jailed criminals with those of the general population  They find that, compared to average driver, the criminals have 3.25 times as many citations for traffic violations, 5.5 times as many property damage and injury producing crashes, and 19.5 times as many involvements in fatal crashes.  They further write "Criminals who were involved in major traffic offenses were likely to have been involved in major crimes and those involved in minor traffic offenses in a minor crime," suggesting that "the degree of an individual's deviation from societal norms is similar in divergent areas" (p. 432).  A crucial point not addressed by Haviland and Wiseman [1974] is the extent to  which the crimes for which the criminals were in prison were for the same traffic offenses on which their analysis is based.  It is possible that some of the strength of the effects they noted is due to correlating traffic offenses with traffic offenses.

                 Additional evidence that crash involvement is correlated with more general criminal involvement is provided by Sivak [1983], who finds that a state's homicide rate can be used to predict its traffic fatality rate.  Whitlock [1971] finds relationships between road fatalities and homicide deaths, suicide deaths, and total violent death in data for 27 countries, mostly in Europe. 

                 The general association between crash involvement and crime involvement invites the speculation that various factors known to be associated with crime involvement (socioeconomic, race) might also be associated with crash involvement; such variables are unavailable in nearly all traffic crash data files.  However, UK data [Office of Population Censuses and Surveys 1978] suggest that involvement in all types of fatal injuries, including those from driver and pedestrian crashes, is strongly linked to social class; Adams [1985, p. 4] infers from such data, "An unskilled manual worker is four and a half times more likely to be killed in a motor vehicle accident than a self
employed professional, and sixteen times more likely to be killed when walking."  Further evidence of strong relationships between fatality risk and social class is provided by Keeling, Golding, and Millier [1985], who analyze non-natural child deaths in two English communities.  Data for total injury deaths per capita show different rates for different racial groups in the United States, with native Americans having the highest rates and American Asians having the lowest [Baker, O'Neill, and Karph 1984, p. 27].  For pedestrian fatalities per million population, Mueller, Rivara, and Bergman [1987] report rates of 24 for whites, 38 for blacks and 148 for native Americans.

                 Few would suggest that criminal activity flows from insufficient study of ethics, and that instruction in this discipline would much reduce crime.  The associations noted here between criminal activity and crash involvement suggest likewise that lack of knowledge about correct driving procedures is not the primary source of traffic crashes; this observation helps illuminate the lack of demonstrated association between driver education and reduced crash rates.

                 The above comparisons are presented to attempt to gain insight into processes underlying traffic crashes, and to show that they involve much more than inadequate perceptual motor skill performance.  The relationships should not be interpreted to suggest that those involved in traffic crashes generally possess criminal traits.




                 The above findings suggest that broad psychological characteristics of the driver may play central roles in propensity towards crash involvement.  A
number of studies have attempted to gain more specific information on the relationship between personality and driving.


The first study indicating "we drive as we live"


                 One of the earliest studies to examine the relationship between crash involvement and broad psychological characteristics is that of Tillmann and Hobbs [1949].  They compared characteristics of 96 Canadian taxicab drivers who had four or more crashes with a matched (age, sex and driving experience) group of 100 taxicab drivers who had no previous crash record, with the results shown in Table 6-1.


Table 6-1 about here


                 Tillmann and Hobbs [1949, p. 329] conclude:

It would appear that the driving hazards and the high accident record are simply one manifestation of a method of living that has been demonstrated in their personal lives.  Truly it may be said that a man drives as he lives.  If his personal life is marked by caution, tolerance, foresight, and consideration for others, then he would drive in the same manner.  If his personal life is devoid of these desirable characteristics then his driving will be characterized by aggressiveness, and over a long period of time he will have a much higher accident rate than his stable companion.

                 Although the methodology of the Tillmann and Hobbs [1949] study has been criticized on many counts, most recently by Grayson [1990], this study was the first to provide specific evidence of a strong link between broad personality characteristics and crash involvement.  A major deficiency of the study is
that much of the interpretation is based on psychiatric-type interviews conducted while riding in the taxicabs; this procedure is personal and subjective in nature, and given the extreme differences between the groups of drivers, it was not possible for the interviewer to remain unaware of the group to which the driver being interviewed most likely belonged, thus raising the possibility of potential bias.  The comparison is between extremes (some of the high-crash-rate drivers verged on the psychopathic), so it could be argued that the results may not necessarily be applicable to a more moderate degree of crash overinvolvement.


Psychiatric profiles of fatally injured drivers


                 Finch and Smith [1970] applied an imaginative technique to obtain psychiatric profiles of 25 deceased male drivers judged to be at fault in the crashes in which whey were killed in Houston from 1967-1968.  The profiles were produced by conducting in-depth interviews with family members and associates of the deceased.  These profiles were compared to profiles of 25 control subjects selected from the same voter precincts in which the deceased had lived, and matched in such characteristics as age (all were males).  Many criticisms of this study are possible; the information gathering processes were necessarily quite different for the deceased and the control subjects, and the sample sizes are small.  However, the differences found are much larger than any that appear likely to be due to possible biases in the technique.  Fig. 6-8 shows the basic differences in personality factors found between the fatally injured and control drivers.  Only 20% of the fatally injured drivers are found to be free of psychosis or personality disorders, compared to 88% in the control population.  Even the few abnormal personalities amongst the control population are found to be less deviant, and
to have more adequate coping mechanisms which helped compensate for their psychiatric-driving liability.  The study provided evidence supporting Waller's [1967] suggestion that sociopathic personalities are overrepresented in the high-risk driving population, at least amongst fatalities.


Fig. 6-8 about here



Other studies on the relationship between personality and crash risk


                 McGuire [1976] reviews a substantial number of additional studies examining the relationship between driver personality and crash involvement (see also Shinar [1978] and Na?"a?"ta?"nen and Summala [1976]).  In one study, McGuire [1956] administered a paper-and-pencil test to two groups of 67 male subjects in military service; one of the groups reported at least one crash in recent months, whereas the other group reported no crashes throughout their driving careers.  McGuire [1956] summarized his findings as follows: "When compared with the `safe' driver the accident-haver is less mature, less intellectual in his tastes and interests (but not necessarily less intelligent), has a lower aspiration level, expresses `poorer' attitude toward the law and driving, is not as socially well adjusted, and is the product of a less happy childhood."

                 In a later work, McGuire [1972] administered a variety of tests and questionnaires to 2727 Mississippi driver license applicants whose driving records were derived in confidential interviews after a two-year exposure period.  This study, not based on extreme groups, indicated that crashes correlate with having interests which are less intellectually oriented and less aesthetic in nature, and with a tendency not to deny open feelings of
hostility.  Higher crash frequency is associated with increased aggression, prestige seeking, and the seeking of social roles which are oriented towards authority and/or competition in preference to those which emphasize closeness to people and social service.  Those with crash involvements are more likely to have family histories and current family relationships reflecting higher degrees of disruption and conflict.  In a review of more recent literature, Tsuang, Boor, and Fleming [1985] conclude that certain personality characteristics and psychopathology -- such as low tension tolerance, immaturity, personality disorder, and paranoid conditions -- appear to be risk factors for traffic crashes.

                 An indication that avoidance of traffic crashes is related to health-producing habits in general is provided in a study by DiFranza et al. [1986] which finds that smokers have 50% more traffic crashes and 46% more violations than non-smokers.  The additional finding that the excess crashes and violations remained when differences in alcohol consumption, age, driving experience and education are taken into account led the authors to suggest that the differences might reflect more frequent general risk taking by smokers.  Waller [1986] mentions that smoking may induce physiological changes, such as vision impairment, which could play a direct role in increasing crash risk.

                 Left-handedness has been shown to be related to shorter life, and to involvement in a whole range of injury-producing events.  Coren [1989] finds that 180 left-handed drivers self-reported 19 traffic crashes in the previous two years compared to 103 reported crashes for 1716 right-handed drivers; that is, the left-handed drivers have a 76% higher crash involvement rate.  Left-handed males have higher rates than right-handed males, and left-handed females have higher rates than right-handed females.  Such an effect could possibly arise because traffic driving on the right (the study was conducted
in British Columbia, Canada) might provide a safety advantage to right-handed drivers.  This could be examined by repeating the study in a jurisdiction in which traffic drives on the left. (There are a number of studies for which replication in jurisdictions driving on the other side of the road could provide illumination; for example, in US traffic, Evans and Frick [1988] find 38% more impacts of high severity from the right side than from the left).  If it turned out that left-handed drivers had lower crash rates in left-drive jurisdictions, this would argue in favor of universal driving on the right, because most drivers are right-handed.  My own guess is that explaining why higher crash risk is associated with left-handedness is as formidable a task as explaining why higher artistic and mathematical talent is associated with left-handedness.  I would expect any performance difficulties to be adequately compensated for -- recall the absence of evidence that monocular drivers have higher crash rates than binocular drivers.


Emotional stress


                 Personality denotes stable character traits that do not change over short time periods.  Emotional stress may produce short or medium term departures from an individual's long term average driving behavior.  McMurray [1970] provides rather clear evidence that the risk of a driver being involved in a crash, or a traffic violation, increases just prior to divorce proceedings.  Finch and Smith [1970] find evidence that drivers killed in crashes are more likely than control drivers to be undergoing periods of personal stress, and also that driving is used as an outlet to handle stress.  Keeling, Golding, and Millier [1985] find a clustering of child pedestrian deaths around the time of the child's birthday, and suggest that the excitement engendered overrides the child's normal caution.




                 Without stating so explicitly, it is assumed in most writing on traffic safety, including this book, that the reason for driving is transportation.  While such a motivation describes much of the use of the automobile, it does not describe all of its use.


Pleasure and thrill seeking motives


                 Na?"a?"ta?"nen and Summala [1976, p 42] discuss "extra motives" that often impinge upon driving.  These include competitiveness, sense of power and control, or more generally, hedonistic objectives -- the pursuit of sensual pleasure for its own sake.  They write "Speed, and especially its acceleration, appears to produce pleasurable excitement even when no specific destination lies ahead and there is no point in haste.....  Driving a car or other motor vehicle affords us basically the same sort of thrills as those experienced on the roller coaster" [p. 46].  Na?"a?"ta?"nen and Summala [1976] further quote extensively from the British study by Black [1966], who interviewed 25 subjects, mainly young people, under hypnosis and not under hypnosis.  When not under hypnosis, responses to traffic-safety related questions were in conformity with accepted good safety practice.  Under hypnosis, the subjects expressed reduced concern about crashes and the consequences of fast driving.  Whitefield [1967] considers that much of the material in Black's [1966] book lacks rigor.  In a treatment not aspiring to technical rigor, Bayley [1986] links the pleasures of fast driving to more traditional pleasures.

                 Extra motives likely play an important role in the enhanced crash rates of younger drivers, especially the higher rates of male drivers.  Although not established by controlled studies, which are bound to be difficult, there are copious anecdotal examples of the use of vehicles to show-off, to attract and impress members of the opposite sex, to provide excitement and to display competitive prowess.  Such use is enshrined in our culture, and figures in many youth-cult movies (Rebel Without a Cause, American Graffiti, to name but two).  Jessor [1984; 1987] discusses an "adolescent problem behavior syndrome" in which cars are used as an outlet for the independence, rebelliousness, and peer acceptance needs of newly licensed adolescents.  Another non-transport use of vehicles is "scutting" -- gaining a ride on the back of a moving vehicle by holding onto it by any means possible.  Casserly and O'Brien [1989] report that this pastime is responsible for more deaths occurring in a hospital in Dublin, Ireland than is any other type of traffic crash involving children.




                 Another non-transportation use of vehicles is suicide.  Suicide in a single-occupant single-vehicle crash provides the most undiscoverable and honorable method of self destruction in motorized societies.  It minimizes guilt in those left behind, and avoids insurance complications.  It appears as an "accident", unrelated to any decisions or deficiencies on the part of the deceased or his or her family or friends.  Not only can the automobile be used for a premeditated suicide, but it may just happen to be available at the instant of a momentary, and perhaps otherwise temporary, impulse towards self destruction.  It is a near-perfect instrument with which to indulge the "death instinct" postulated by Freud.

                 The use of vehicles for suicide unquestionably occurs, and has been discussed in the literature for decades.  Indeed, Tabachnick et al. [1973] author a book "Accident or Suicide? -- Destruction by Automobile".  In the introduction they write: "This book revolves around a specific theory of the etiology of accident -- namely, that in many, perhaps even most accidents, suicide or suicide-like factors are in evidence."  Much of the book is devoted to describing a study in which psychiatric examination and questionnaire data were obtained from three groups of Los Angeles hospital patients; 25 drivers recovering from injuries sustained in severe single-vehicle traffic crashes; 29 recovering from suicide attempts; and 31 recovering from appendectomies.  The most specific finding is of high alcohol use amongst the traffic-crash group, a result to be expected (Chapter 7).  I find little in the 20 tables or extensive discussion that demonstrates convincingly that some specific fraction of traffic crashes were due to suicide; the quoted claim in the introduction is not supported.  The study however does find some similarities between characteristics of those involved in crashes and those attempting suicide.  Additional suggestions of a link between suicides and traffic crashes are provided by Whitlock [1971], who finds that countries with high traffic fatality rates also have high suicide rates, and Bollen [1983], who finds similarities in the day-to-day variations of suicides and traffic fatalities in the US.

                 A novel approach to determining the role of suicide in traffic crashes is that of Bollen and Phillips [1982], who find that extensive news coverage of suicides of famous people is followed by increases in traffic fatalities.  The excess is attributed to suicides, based on findings [Phillips 1979; Bollen and Phillips 1981] that media coverage of suicides leads to increases in suicides in general.  To date, there is no quantitative estimate of the fraction of all traffic fatalities attributable to suicide.  The method of Bollen and Phillips
[1982] may have potential to do this, especially when coupled with the more detailed information about traffic fatalities now available in FARS.  One would expect suicides preferentially to influence single-vehicle, single-occupant fatal crashes, and also, perhaps, pedestrian fatalities.  Indeed, if these increased while other fatalities remained unaffected after major news coverage of suicides, this would provide particularly solid evidence for the approach used by Phillips [1979], and if enough data were combined, might offer a quantitative estimate of the fraction of all traffic fatalities that are suicides.  Such a study might also provide important information on the ability of the media to influence behavior, which has implications for other aspects of traffic safety.  An estimate of the fraction of traffic fatalities attributable to suicide is of considerable interest, because such countermeasures as better occupant protection, improved roads, or increased law enforcement are unlikely to influence this component of traffic fatalities.




                 Although there is a large body of evidence showing that behavior is of crucial importance in traffic safety, most of the evidence is indirect, non-quantitative, and subject to differing interpretations.  There are intrinsic barriers that conceal the type of direct information one would ideally like. One problem is that crash rates vary by large amounts based on specific known factors, such as age and alcohol use.  Hoxie [1985] reports car occupant fatality rates differing by a factor of 200, while the hypothetical "high-risk" driver in Evans, Frick and Schwing [1990] has a fatality risk 1000 times that of the "low-risk" driver (differences in occupant protection contribute part of the effect).

                 The largest and most unmistakable behavior effects (as distinct from demographic, etc. factors) are obtained in comparing extreme cases, such as comparing taxi drivers with four or more crashes to those with zero crashes [Tillmann and Hobbs 1949] or fatally injured drivers [Finch and Smith 1970] to more typical drivers.  Since the comparison is between such extremes, bordering on comparing psychopathic to normal drivers, it could be argued that the results may have no validity when interpolated to the more moderate degrees of overinvolvement which contribute to most traffic crashes.  This problem has similarities to the dose-response problem in toxicology.  Does a large, easily measured, deleterious effect associated with a massive dose of exposure to some substance support the inference that (say) one tenth of the dose would still produce some deleterious effect, such as about one tenth the effect of the initial dose?  Or is there some threshold below which the substance produces no deleterious effect?  Given that most crashes involve drivers not at the fringes of society, it is important to know whether moderate variations within the normal ranges of behavior can explain variations in crash rates.  As in the toxicology case, the smaller the dose, the more difficult it is to measure the response.

                 There are methodological problems inherent in comparing those with moderately above average crash rates to those with below average crash rates.  Let us assume an average crash rate of one crash per ten years, or 0.1 per year.  In a seven-year period, a "better than average" driver therefore expects to have no crashes, while one crash, which exceeds the expected average number of 0.7, nominally indicates a "worse than average" driver.  Even a driver whose behavior is such as to double crash risk would still have a 25% chance of having no crashes in a seven-year period (assuming a Poisson distribution).  On the other hand, a driver whose crash rate is half the average would still have a 30% chance of having one or more crashes in a seven
year period.  The specific mix of drivers, by propensity to crash, in a sample of crash-free drivers and a sample of drivers with one or more crashes, depends on the distribution of propensity to crash in the population, which is difficult to determine.  What is, however, very clear is that the crash-free population contains many drivers with above average propensity to crash, and the crash-having population contains many with below average propensity to crash.  Thus any relationship between driver characteristics and propensity to crash is going to be difficult to discover in a study based on comparing the personal characteristics of the drivers assigned to each these two groups.  Accordingly, even if crash risk were strongly related to personality factors, such relationships would manifest themselves clearly only in groups of drivers with crash rates many times the average.

                 There does not appear to be any way out of this dilemma.  Even if data were available for periods of many decades, the propensity to crash may vary and therefore introduce other unwanted sources of variation.  The best that probably can be done is to evaluate judgmentally the considerable body of evidence.  My own intuition is that the dose-response relationship is not of a threshold type, so that the finding that large deviations in certain directions from normal behavior are associated with large increases over normal crash risk strongly suggests that small deviations in the same direction from normal behavior are likely to be associated with important increases in crash risk.  Such a conclusion is supported weakly by research on non-extreme comparisons.  However, as discussed above, even if such associations are in fact strong, they would appear diluted in any study seeking correlations.

                 I see no reason to dissent from McGuire's [1976] conclusion, based on summarizing the extensive literature on the relationship between personality factors and traffic crashes, "It may be said that highway accidents are just
another correlate of being emotionally unstable, unhappy, asocial, anti-social, impulsive, under stress and/or a host of similar conditions under other labels."




                 Among the factors contributing to driver speed choice is a systematic underestimation of the probability that they will be killed [Lichtenstein et al. 1978].  Another is that speed is desired for its own sake, for sensuous pleasure rather than just for such utilitarian motives as saving time.  Noguchi [1990] examines various aspects of driver's speed choice by having interviewers (a psychologist, a sociologist, and others) conduct in-depth interviews with 26 Japanese drivers selected as possessing average driver characteristics.  The subjects were asked to imagine themselves driving on a highway or on a motorway with speed limits of 50 km/h and 100 km/h, respectively, and each without traffic jams.  They were asked at what "actual" speed they would drive, and also what they considered to be "economic", "safe", and "pleasant" speeds, with the results shown in Fig. 6-9.  Their placing the safe speed at above the speed limit is consistent with other parts of the same study which find that subjects judged speed limits to be too low.  The chosen speed appears to represent a compromise between competing desires for pleasure and safety.


Fig. 6-9 about here


                 Chosen speed has a large influence on the risk of crash, injury and death.  Nilsson [1982; 1990] examined changes in crashes and casualties associated with changes in travel speeds following increases or decreases in speed limits
in Sweden.  In many cases, speeds on roads with changed speed limits were compared to speeds on similar roads which did not have their limits changed.  Nilsson finds that if mean or median speeds change (upwards or downwards) to new values, vb, from prior values, va, the following relations estimate approximately the ratio of events before and after the speed change:


                  (Crash rate after)/(Crash rate before) =  vb/va          Eqn 6-1


    (Injury crash rate after)/(Injury crash rate before) = (vb/va)2    Eqn 6-2


(Fatality crash rate after)/(Fatality crash rate before) = (vb/va)4   Eqn 6-3


These relationships have plausible physical interpretations.  Crash rate is clearly related to speed, so the simplest relationship, that of proportional­ity reflected in Eqn 6-1, is a natural starting point.  If, when a crash occurs, the risk of injury is proportional to speed, then injury crash rate will be proportional to speed squared.  If the probability that an injury proves fatal is proportional to the energy dissipated in the crash (which is proportional to the speed squared), then the probability of a fatal crash is proportional to the fourth power of the speed.

                 The plausibility of these relationships is supported further by increases in fatality rates when rural Interstate speed limits in the US were increased from 55 mph to 65 mph in 1987.  Miller [1989] reports that this legislative change led to an increase in average speeds from 60.8 mph to 62.2 mph.  Substituting into Eqn 6-3 gives (62.2/60.8)4 = 1.10; that is, a 10% increase in fatal crashes is estimated.  Freedman and Esterlitz [1990] estimate a larger increase of "nearly 3 mph"; if we interpret this to mean a 2.5 mph increase over an initial 60 mph we compute an 18% fatality rate increase. 
These predictions are in reasonable agreement with observed fatality increases associated with the speed limit change of 16% [National Highway Traffic Safety Administration 1989], 15% [Garber and Graham], 15% [Baum, Lund, and Wells 1989], and 11%, by inference, from McKnight and Klein [1990].

                 When the 55 mph nationwide speed limit was introduced in the US in 1974, average speeds on rural Interstates, which previously had maximum speed limits of 70 mph, decreased from 63.4 mph to 57.6 mph (Federal Highway Administration data used in Evans [1987b]).  Substituting these values into Eqn 6-3 estimates a reduction in the fatal crash rate of 32%; the observed reduction is 34% (Fig. 13-3).

                 The above discussion ignores distinctions between fatalities and fatal crashes, and in some cases between fatalities and fatality rates.  Such differences are minor relative to the approximate nature of the discussion, and the exclusive focus on percent changes.

                 The observation that fatalities per unit distance of travel on German Autobahns are similar to those on US Interstate freeways has led to questioning of the safety effects of speed.  Lenz [1990] reports that average speeds on Autobahns have been increasing at about 1 km/h per year, reaching 132 km/h, or 82 mph, in 1988.  Substituting into Eqn 6-3, and assuming 60 mph for US Interstates, estimates a German fatality rate 3.5 times that for the US.  Such simple comparisons of rates between jurisdictions are to be approached with extreme caution, a point also made by O'Neill [1986], because there are many other important factors, including the four below.  First, car ownership is focused on a narrower economic segment of society in Germany than in the US; there are indications of strong relationships between socioeconomic status and crash rate.  Second, German anti-drunk driving laws are stricter.  Third, the minimum age for a German driving license is 18, compared to, generally, 16 in the US; even at 18, German teenagers have less opportunity to
drive than do American youths.  Fourth, safety belt use in Germany is about 95%, compared to about 45% in the US.  Apart from the fourth factor, which generates a fatality ratio of 1.4 (Chapter 10), these factors are difficult to quantify.  However, it is not implausible that their combined effect is a factor of 3.5, and that increases in speed are cancelling an expected large reduction in fatality risk.

                 While the discussion has been in terms of average speed, the majority of harm is done by those traveling at speeds above the average.  However, the fraction of drivers exceeding any speed is strongly linked to average speeds; for example, when the speed limits on US rural Interstate freeways increased from 55 mph to 65 mph, the proportion of cars exceeding 70 mph nearly doubled [Freedman and Esterlitz 1990]. 


Speed variance


                 The importance of variability around average speeds has been recognized as important in safety, especially since Solomon [1964] and Cirillo [1968] used crash data to show that the driver travelling at close to the average speed has a lower crash risk than drivers travelling at higher or lower than average speeds.  Hauer [1971] derives such a U-shaped relationship by considering crashes proportional to the number of times a vehicle is overtaken or overtakes, which is minimum when travelling at the average speed.  Intuitively, identical vehicles driven by identical drivers at identical speeds would provide optimum safety for a given average speed.

                 From time to time there are claims that speed variance is more important than speed, based on the above findings and the truism that vehicles travelling at identical speeds in the same direction cannot crash into each other.  It is even periodically claimed that speed as such does not affect
safety, and that instead of speed limits the goal should be to require all traffic to move at the same speed.  While it is true that vehicles travelling slower than average have above average crash rates, the reason they are travelling slower is because the driver chooses to travel slower, most likely because of driver or vehicle inadequacies.  If this driver is encouraged or compelled to speed up to the average speed, an increase in crash risk seems more likely than a decrease.  Even if slower than average drivers have crash rates similar to faster than average drivers, their injury and fatality rates will still be substantially less.  Fig. 3-3 shows that 45% of car occupants killed are killed in single-car crashes; speed variance is irrelevant unless vehicles travelling at different speeds interact in some way with one another.  Only about 5% of fatally injured car occupants are killed as a result of any type of rear impact.  Thus speed variance as such can play, at most, only a minor role in fatalities, and contrary suggestions based on regression analyses [Appendix C, p. 200, of Transportation Research Board 1984] provide another example of how just about any desired conclusion can be coaxed out of data using such methods.  It is because of the large influence of speed on casualties that Summala [1985] concludes that control of driver speed should be the primary focus of traffic safety interventions.




                 Driver education is, in some regards, one of the most successful educational endeavors.  Students who cannot drive take a short course, at the completion of which they can drive.  While safety is a primary consideration in all aspects of such training, much of the beginning driver's attention must be devoted to acquiring driving skills, and on learning the rules of the road.  In Chapter 5, and in this chapter, we find no convincing evidence that driver
education, or increased driving skill and knowledge, increase safety.  Although driver education speeds up the process of learning driving skills, the main way such skills are acquired and honed is through direct feedback.  Drivers learn, for example, to execute right turns correctly only after many errors such as riding over the curb, cornering too wide, or entering too fast or slow.  With minimal instruction most people could probably learn to drive by trial and error.

                 In contrast, safety cannot be learned by direct feedback, but requires absorption of accumulated knowledge and experience of others.  The main reason people buy smoke-detector fire alarms is not that their last house burned down; similarly, experiencing crashes is an ineffective way to learn how to avoid them.  The difference between skillful driving and avoiding crashes is so basic as to suggest a course focused more specifically on safety, perhaps with a title like "Safe-driving education".  Such a course would acknowledge that all the students know how to drive, are good drivers, and are getting better all the time as they continue to practice.  In the US, such a course might be appropriate in the last year of high school.

                 While there are many specific items (Chapter 12) that could be covered in such a course, a main thrust would be class discussions of broader safety issues building on growing interest in a host of other hazards which receive widespread mass media coverage.  While the interest in reducing risks seems widespread, the public has acquired a distorted view that many highly improbable risks are much greater than they are, while at the same time largely discounting much more likely hazards [Lewis 1990].  The US population seems subject to frequent bouts of extreme fear over the astronomically improbable, or even the impossible.

                 In contrast, the possibility of commonplace events, such as a child, without warning or good cause, running onto a roadway is dismissed. 
Observational research in the UK shows that motorists behave as if they consider it the responsibility of the child, not the driver, to avoid child- pedestrian crashes [Howarth 1985].  Motorists have passed hundreds of children who have not run out in front of their vehicles, so experience has taught them that such events do not occur.  Traffic fatality data leave little doubt that such "unexpected" events not only happen, but happen often.  FARS data (Fig. 2-15) show that more than 100 six-year-old pedestrians are killed per year (and, of course, vastly more are injured).  Most drivers have not sufficiently considered or discussed the harm that they may cause others even while driving legally, and at their normal safety margins and levels of risk-taking.

                 If drivers adopted safer driving practices, and allowed larger margins of safety, by the end of their driving careers it would not have made much difference in most cases; this follows from the statistical nature of crashes, which are rare events.  But such a behavior change would spare large numbers of individual drivers the burden of having to claim, with legal correctness, that the six-year-old child was killed because it was the child's fault.  The comment, "There was nothing I could do to prevent it," might be the subject of fruitful abstract discussion about hypothetical crash scenarios.  Drivers who take extra care when in the proximity of pedestrians, bicyclists and motorcyclists, greatly reduce the probability that they will be the legally innocent and physically unharmed agent of some fellow human's death or permanent injury.

                 It seems to me that there ought to be an opportunity at least to point out that if a young person has some idealistic sentiment in the direction of helping others, the most immediate expression of it can be in taking rather obvious steps to halve or quarter the chances that he or she will kill or permanently injure some innocent bystander in a crash.  This is an area in which the young person has the chance to directly control events.  The
irresponsible use of vehicles, rather than being cool or admirable, might be viewed as a self-indulgence, such as polluting the air to make money, in which the innocent pay for the behavior of the guilty.  Perhaps appealing to the altruism of good driving might be more effective than appealing to the more traditional, and selfish, motive of self-preservation.  It might be an uphill battle to persuade young people that they should not admire or emulate other young people who use automobiles to risk their own lives in the pursuit of pleasure and the defiance of authority.  It may possibly be more fruitful to persuade them that risking the lives of innocent bystanders, especially young children, is hardly admirable conduct.  Education focused more on conveying understanding about why people drive the way they do, and the potential consequences of such behavior to themselves and to others, might make a contribution to safer driving.

                 The problem of traffic crashes is much more one of drivers doing things that they know they ought not to do, than of not knowing what to do.  This does not imply that it is simply an ethical question, with the goal being to change vice into virtue; the history of the human race offers little hope of progress towards this goal.  Characterizing the problem in simplistic good versus evil terms has led some to conclude that nothing can be accomplished by education or media campaigns because "You can't change human nature."  Reducing traffic crashes requires a change not in human nature, but in behavior.  The evidence is overwhelming that norms of behavior regarding many factors important in public health change, including, in the traffic safety area, increased use of safety belts and motorcycle helmets, and decreased drunk driving.  More dramatic is the decline in smoking.  Over the last two decades smoking has gradually become more associated with chemical dependency and negative qualities rather than with glamor and success.  A large change in traffic safety could occur if the life threatening use of vehicles became more
associated with immaturity and failure than with glamor and excitement.  Education that places safe and considerate driving more in a public health context [Sleet 1984] may have potential, even though it is bound to be difficult to dissuade young people from doing what gives them pleasure.




                 Racing drivers, young drivers, and male drivers, the very groups with the highest levels of perceptual-motor skills and interest in driving, are groups which have higher than average crash involvement rates.  This demonstrates that increased driving skill and knowledge are not the most important factors associated with avoiding traffic crashes.  What is crucial is not how the driver can drive (driver performance), but how the driver does drive (driver behavior).

                 The overinvolvement of youth, and males, in traffic crashes is quantitatively similar to the overinvolvement of youth, and males, in criminal activity unrelated to traffic.  Observations of actual drivers show higher levels of risk taking, as indicated by higher speeds and closer following distances, to be associated with young, and male, drivers.  At all ages, males have larger pedestrian fatality risk than females of the same age, suggesting an intrinsic greater activity level, or propensity to take risk, which leads to greater involvement in traffic crashes.

                 Many studies provide evidence supporting the general contention that people drive as they live.  Involvement in traffic crashes is correlated with being emotionally unstable, unhappy, asocial, anti-social, impulsive, aggressive, and under stress.  Educational initiatives which attempt to increase understanding about why people drive the way they do, and the potential consequences of such ongoing behavior to themselves, and more
especially to others, such as young children, might make a contribution to safer driving.




Adams, J.G.U.  Risk and freedom - the record of road safety regulations. Nottingham, UK: Bottesford Press; 1985.

Baker, S.P.; O'Neill, B; Karph, R.S.  The injury fact book. Lexington, MA: Lexington Books; 1984.

Baum, H.M.; Lund, A.K.; Wells, J.K.  The mortality consequences of raising the speed limit to 65 mph on rural Interstates. American Journal of Public Health 79:1392-1395; 1989.

Bayley, S.  Sex, drink and fast cars. London, UK: Farber and Farber; 1986.

Black, S.  Man and motor cars: an ergonomic study. New York, NY: W.W. Norton; 1966.

Bollen, K.A.  Temporal variations in mortality: a comparison of U.S. suicides and motor vehicle fatalities, 1972-1976. Demography 20:45-59; 1983.

Bollen, K.A.; Phillips, D.P.  Suicidal motor vehicle fatalities in Detroit: a replication. American Journal of Sociology 87:404-412; 1981

Bollen, K.A.; Phillips, D.P.  Imitative  suicides: a national study of the effects of television news stories. American Sociological Review 47:802-808; 1982.

Carsten, O.M.J.; Tight, M.R.; Southwell, M.T.  Urban accidents: why do they happen?  Basingstoke, UK: AA Foundation for Road Safety Research; 1989.

Casserly, H.B.; O'Brien, T.  Scutting: a dangerous pastime? Injury 20:337-338; 1989.

Cirillo, J.A.  Interstate system accident research study II, interim report II. Public Roads 35:71-75; 1968.

Coren, S.  Left-handedness and accident-related injury risk. American Journal of Public Health 79:1040-1041; 1989.

DiFranza, J.R.; Winters, T.H.; Goldberg, R.J.; Cirillo, L.; Biliouris, T. The relationship of smoking to motor vehicle accidents and traffic violations. New York Journal of Medicine 86:464-466; 1986.

Evans, L.  Young driver involvement in severe car crashes. Alcohol, Drugs, and Driving 3(3-4):63-78; 1987a.

Evans, L.  Factors controlling traffic crashes. Journal of Applied Behavioral Science 23:201-218; 1987b.

Evans, L; Frick, M.C.  Seating position in cars and fatality risk. American Journal of Public Health 78:1456-1458; 1988.

Evans, L.; Frick, M.C.; Schwing, R.C.  Is it safer to fly or drive? -- a problem in risk communication. Risk Analysis 10:239-246; 1990.

Evans, L; Wasielewski, P.  Risky driving related to driver and vehicle characteristics. Accident Analysis and Prevention 15:121-136; 1983.

Finch, J.R.; Smith, J.P.  Psychiatric and legal aspects of automobile fatalities. Springfield, IL: Charles C. Thomas; 1970.

Freedman, M.; Esterlitz, J.R.  The effect of the 65 mph speed limit on speeds in three states. Arlington, VA: Insurance Institute for Highway Safety; January 1990.

Garber, S; Graham, J.D.  The effects of the new 65 mile-per-hour speed limit on rural highway fatalities: a state by state analysis. Accident Analysis and Prevention 22:137-149; 1990.

Girdler, A.  About the sport. Road and Track 23(6):97-98; 1972.

Gottfredson, M.; Hirschi, T.  The true value of lambda would appear to be zero: an essay on career criminals, criminal careers, selective incapacitation, cohort studies and related topics. Criminology 24:213-234; 1986.

Grayson, G.  Individual and social factors in accident liability: a review of the literature. Paper presented to International Symposium on Driving Behaviour in a Social Context, Paris, France; 16-18 May 1989. To be published in proceedings of the meeting (1990).

Ha?"kkinen, S.  Traffic accidents and professional driver characteristics: a follow-up study. Accident Analysis and Prevention 11:7-18; 1979.

Hauer, E.  Accidents, overtaking and speed control. Accident Analysis and Prevention 3:1-12; 1971.

Haviland, C.V.; Wiseman, H.A.B.  Criminals who drive. American Association for Automotive Medicine, 18th Annual Proceedings, Toronto, Ontario, p. 432-439; 12-14 September 1974.

Howarth, C.I. Interactions between drivers and pedestrians: some new approaches to pedestrian safety. In: Evans, L.; Schwing, R. C., editors.  Human behavior and traffic safety. New York, NY: Plenum Press, p. 171-182; 1985.

Hoxie, P.  Transportation fatalities: a risk assessment. SAE paper 851225. Warrendale, PA: Society of Automotive Engineers; 1985.

Jessor, R.  Adolescent development and behavioral health. In: Matarazzo, J.D.; Weiss, S.M.; Herd, J.A.; Miller, N.E., editors. Behavioral health: a handbook of health and disease prevention. New York, NY: John Wiley; 1984.

Jessor, R.  Risky driving and adolescent problem behavior: an extension of problem-behavior theory. Alcohol, Drugs, and Driving 3(3-4):1- 11; 1987.

Jonah, B.A.  Accident risk and risk-taking behaviour among young drivers. Accident Analysis and Prevention 18:255-271; 1986.

Keeling, J.W.; Golding, J.; Millier, H.K.G.R.  Non-natural deaths in two health districts. Archives of Disease in Childhood 60:525-529; 1958.

Lenz, K.H.  Effects of speed. IATSS Research -- Journal of International Association of Traffic and Safety Sciences 14(?):??-??;1990. (in press)

Lewis, H.W.  Technological risk. New York, NY: Norton; 1989.

Lichtenstein, S.; Slovic, P.; Fishoff, B.; Layman, M; Combs, B.  Judged frequency of lethal events. Journal of Experimental Psychology: Human Learning and Memory 4:551-577; 1978.

Lim, C.S.; Dewar, R.E.  Driver cognitive ability and traffic accidents. University of Calgary, Calgary, Alberta, Canada; 1989.

McGuire, F.L.  The safe-driver inventory: a psychological test for selecting the safe automobile driver. U.S. Armed Forces Medical Journal 7:1249-1264; 1956.

McGuire, F.L.  A study of methodological and psycho-social variables in accident research. JSAS Catalog of Selected Documents in Psychology. Ms. no. 195; 1972.

McGuire, F.L.  Personality factors in highway accidents. Human Factors 18:433-442; 1976.

McKenna, F.P.; Duncan, J.; Brown, I.D.  Cognitive abilities and safety on the road: a re-examination of individual differences in dichotic listening and search for embedded figures. Ergonomics 29:649-663; 1986.

McKnight, A.J.; Klein, T.M.  The relationship of the 65 mph limit to speeds and fatal accidents. Paper 890711 presented to the 69th Annual Meeting of the Transportation Research Board, Washington. DC; 7-11 January 1990.

McMurray, L.  Emotional stress and driving performance: the effects of divorce. Behavioral Research in Highway Safety 1:100-114; 1970.

Miller, T.R.  65 mph: does it save time. Association for the Advancement of Automotive Medicine, 33rd Annual Proceedings, Baltimore, MD, p. 73-90; 2-4 October 1989.

Mueller, B.A.; Rivara, F.P.; Bergman, A.B.  Factors associated with pedestrian-vehicle collision injuries and fatalities. Western Journal of Medicine 146:243-245; 1987.

Na?"a?"ta?"nen, R. and Summala, H.  Road-user behavior and traffic accidents. Amsterdam, Netherlands: North Holland; 1976.

National Highway Traffic Safety Administration.  Transcript of proceedings: Workshop on Advanced Driver Education Techniques, Washington, DC; January 1972.

National Highway Traffic Safety Administration.  Interim report on the safety consequences of raising the speed limit on rural Interstate highways. Washington, DC; 1989.

Nilsson, G.  The effect of speed limits on traffic accidents in Sweden. VTI Report No. 68. Linkoping, Sweden: National Road and Traffic Research Institute, S-58101, p. 1-10; 1982.

Nilsson, G.  Personal communication describing results published in Swedish; publication in English is planned; 1990.

Noguchi, K.  In search of optimum speed: from the users viewpoint. IATSS Research -- Journal of International Association of Traffic and Safety Sciences 14(?):??-??;1990. (in press)

Office of Population Censuses and Surveys.  Information from the Registrar General's decennial supplement on occupational mortality, 1970-1972, Series DS No. 1. London, UK: Her Majesty's Stationery Office; 1978.

O'Neill, B.  Autobahns vs. U.S. Interstates. Insurance Institute for Highway Safety, Status Report 21(13):2-2; 1986.

Peck, R.C.; McBride, R.S.; Coppin, R.S.  The distribution and prediction of driver accident frequencies. Accident Analysis and Prevention 2:243-299; 1971.

Phillips, D.P.  Suicide, motor vehicle fatalities, and the mass media: evidence towards a theory of suggestion. American Journal of Sociology 84:1150-74; 1979.

Popkin, C.L.; Rudisill, L.C; Geissinger, S.B.; Waller, P.F.  Drinking and driving by women. American Association for Automotive Medicine, 30th Annual Proceedings, Montreal, Quebec, p. 1-14; 6-8 October 1986.

Shinar, D.  Psychology on the road -- the human factor in traffic safety. New York, NY: John Wiley; 1978.

Sivak, M.  Society's aggression level as a predictor of traffic fatality rate. Journal of Safety Research 14:93-99; 1983.

Sleet, D.A. Reducing motor vehicle trauma through health promotion programming. Health Education Quarterly 11:113-125; 1984.

Solomon, D.  Accidents on main rural highways related to speed, driver, and vehicle. Washington, DC: Federal Highway Administration, US Department of Transportation, July 1964.

Steffensmeier, D.J.; Allan, E.A.; Harer, M.D.; Streifel, C.  Age and the distribution of crime. American Journal of Sociology 94:803-31; 1989.

Summala, H.  Modeling driver behavior: a pessimistic prediction?  In: Evans, L.; Schwing, R.C., editors. Human behavior and traffic safety. New York, NY: Plenum Press, p. 43-61; 1985.

Tabachnick, N; Gussen, J; Litman, R.E.; Peck, M.I.; Tiber, N.; Wold, C.I. Accident or suicide? Destruction by automobile. Springfield, IL: Charles C. Thomas; 1973.

Tillmann, W.A.; Hobbs, G.E.  The accident-prone automobile driver. American Journal of Psychiatry 106:321-331; 1949.

Transportation Research Board. 55: a decade of experience. TRB Special Report 204. Washington, DC: National Research Council; 1984.

Tsuang, M.T; Boor, M.; Fleming, A.A.  Psychiatric aspects of traffic accidents. American Journal of Psychiatry 142:538-546; 1985.

US Department of Justice, Federal Bureau of Investigation. Washington, DC: Uniform crime reports: crime in the United States; 28 July 1985.

Waller, J.A.  Identification of problem drinking among drunken drivers. Journal of the American Medical Association 200:114-120; 1967.

Waller, J.A.  On smoking and drinking and crashing. New York Journal of Medicine 86:459-460; 1986.

Wasielewski, P.  Speed as a measure of driver risk: observed speeds versus driver and vehicle characteristics. Accident Analysis and Prevention 16:89-103; 1984.

Whitefield, W.  Review of "Man and Motor cars" by S. Black. Ergonomics 10:363-364; 1967.

Whitlock, F.A.  Death on the road: a study of social violence. London, UK: Tavistock; 1971.

Williams, A.F.; O'Neill, B.  On-the-road driving records of licensed race drivers. Accident Analysis and Prevention 6:263-270; 1974.



Table 6-1. Relative frequency of contact with social agencies for 96 taxicab drivers with four or more crashes and for 100 with zero crashes. Table entries are percents.



                     Adult   Juvenile  Public   Social   Credit   At least

                     court   court     health   service  bureau   one agency



   Four or more

                      34       17        14       18        34        66






                       1        1         0        1         6         9






Data from Tillmann and Hobbs [1949]