|Do people drive
as they live?|
William Tillmann and George Hobbs
To begin with, consider the special case of relating the experimental risk-taking measures to accident records. The most salient feature of traffic accident risk is the risk of physical injury. In our games the participants did incur a risk, but the risk usually consisted of not making any financial gain or losing out on an opportunity for acquiring social recognition. The expected strength of the relationship between the experimental outcomes and accident record would thus depend on the degree of consistency in people's risk-taking tendencies from one type of risk to another.
Some researchers have distinguished four types of risk: physical, financial, social and ethical. Social risk is the risk of incurring social disapproval as a consequence of one's actions, and ethical risk refers to doing something one may feel guilty about afterwards. Positive association between the four types of risk has been observed, but the association is weak. People who take more-than-average risk of one type do take more risk of another type, but only slightly more so. No salient personality trait of all-pervasive risk seeking or risk avoidance has been identified. In other words, no general risk-taking personality trait emerges from the research studies. This is not surprising, for two reasons. First, if people's deviations from optimal risk taking were quite similar in all four types of risk taking, that would have been so obvious that we would have known it all along and it would no longer be a subject for research. Second, a general risk-taking trait would also be highly counterproductive to success in life, since it would mean that people who are less likely to be successful in obtaining satisfaction in one area would also be less likely to obtain satisfaction in another, other things being equal (including skill).
If there is one particular habit that is often believed to be associated with risk-taking personality traits, it must be gambling, especially pathological or compulsive gambling. In everyday language, "to gamble" is almost synonymous with "to take a risk". Yet, the many studies undertaken in order to identify the personality traits that might characterize these people have failed to indicate that gamblers are more-than-average risk takers in other aspects of life as well, and they do not systematically differ from non-gamblers in their personality traits. Even the trait of stimulation seeking--called "sensation seeking" by some--shows no consistent pattern. Some studies find gamblers higher than average on this trait, others lower. The same holds for extraversion.
Just as gambling is the proverbial form of risk taking, the sensation-seeking questionnaire has become a frequently-used instrument for the assessment of individual differences in risk-taking tendency. Maybe this is because the author of this questionnaire defines sensation seeking with specific reference to risk: "the need for varied, novel, and complex sensations and experiences and the willingness to take physical and social risks for the sake of such experiences". Maybe, too, the questionnaire does not quite measure up to the author's definition, because recent research evidence--including his own --seems to indicate that people who get high scores on this questionnaire feel that the personal risks involved in the risky activities they like are less severe than is true for the people who dislike those risky activities and obtain low scores. Similarly, Dutch drivers with high sensation-seeking scores were found to follow more closely to the car in front of them than drivers scoring low on the questionnaire. Their self-reports and physiological measures of arousal, however, indicated that the sensation seekers experienced about the same sensation of risk as those with low scores. Thus, "the seeking of sensation" is largely reduced to "estimating that the risk is low". Although sensation seekers have been reported to be more inclined to engage in apparently risky driving habits like speeding, drinking and driving, and to run up a record of moving violations, various studies found no relation between sensation-seeking scores and traffic accident history.[10,11]Skiers with high sensation-seeking scores have been found to have significantly fewer skiing accidents than sensation avoiders.
A comprehensive review of the literature on the relationship between the habit of gambling and any personality traits of habitual gamblers has this to say: "What little agreement exists suggests a difference in locus of control, with high-frequency gamblers being more external than low-frequency gamblers. However, if such a relationship is a reality, then it is just as likely that the gambling causes the trait as that the trait causes the gambling." For the meaning of "locus of control", see Section 10.3 below. If a habit as risk-riddled as gambling has not been found to be strongly related to personality traits that one associates with a positive inclination toward risk taking, there would seem little hope for finding any marked relationships between accident involvement and personality.
Yet, numerous studies have been conducted in hopes of identifying individual characteristics that are related to accident history. Some have met with a degree of success, although it is not uncommon for significant findings in one study to find either no support or even contradictory results in another. Greater accident involvement has sometimes been found for introverts and, in other investigations, for extraverts. Similarly, sensation seekers were sometimes found to have had more accidents, and sometimes sensation avoiders were found to have more. Greater accident involvement has been found for people who tend to attribute the occurrence of important events in their lives to fate or chance rather than to their own doing. The same holds for impulsivity, easy-goingness and low self-control.
Some studies find more accident involvement in people with what is called a "field-dependent perceptual style". But a larger number of investigations found no relationship. This is also true of reaction time, or quick reflexes, and accidents. As noted in Chapter 1, field dependence characterizes people who do not have an analytical style in the way they look at the world that surrounds them, and are more likely to fall victim to erroneous syllogisms. Perhaps not surprisingly, differences in intelligence have not been found to be related to accident history.
Individual differences in attitudes, however, have been found to be related to involvement in past as well as future accidents: drivers with less aggressive, less macho, less authoritarian and more socially-oriented values show safer driver careers. The same holds for people with a history of better school marks for citizenship (diligence and proper behaviour as rated by their school teachers) and stable employment histories. Some of these studies found biographical characteristics, including increased incidence of criminal records, more strongly associated with accidents than personality variables were.
Contrary to what may seem to be suggested here, crash rate has not invariably been found to be associated with traits that are generally considered to be socially undesirable by mainstream society. For instance, accident-involved bus drivers in the USA and in India were found to more frequently display a Type-A personality. This label refers to people marked by a patent display of energy, competitiveness, alertness, ambition and a view that action is urgent because time is short. However, a subsequent study of the same kind of drivers, conducted in Britain, did not find them to be more often involved in accidents, although they more often reported fast driving. To sum up this paragraph: weak, inconsistent and contradictory findings galore!
The facts contradict the notion that most accidents are due to a small minority of people with dangerous, anti-social personalities. To think so, however, seems very tempting. After all, about one-half of all whisky is being drunk by no more than some 5% of all consumers, 100% of all murders are committed by a very small number of people, 100% of all venereal disease is carried by a small proportion of people. Almost 100% of all pianos in a country are found in a small percentage of homes. Moreover, there is evidence that 100% of all automobile accidents in a given year are incurred by less than 10% of all drivers. So, why should the bulk of accidents not be due to a small minority of people?
However, the evidence is clear that this is not the same minority every year. Consider some findings of a study conducted in North Carolina in the USA. Of all drivers who had two accidents in two consecutive years, 87% did not have a traffic accident in the third year. The 13% who did accounted for 1.6% of all accidents in the third year. Although it is true that their accident rate in the third year was about twice that of the average driver, an action even as drastic as deporting drivers with accidents in two consecutive years (or any other action that effectively eliminates them from the roads) would reduce the total accident rate in the third year by no more than 1.6%. Obviously, this offers no justification for the assumption that the accident rate can be reduced substantially by preventing those who have had accidents from further driving.
In passing, we may note that there is also precious little in these North Carolina data to support the beliefs of those who think that cracking down on violators of traffic laws might bring about a major reduction in the accident rate: "If you took all drivers with three or more violations in the past two years [that is, about 1.3% of all drivers] off the highway and kept them off effectively for two whole years, North Carolina would still experience 96.2 per cent of the accidents it would have had anyway. Moreover, of the drivers removed from the highway, 71% would not have been involved in an accident anyway."
The ability, though very limited, to predict future accidents from past accidents or violations might still be useful to insurance companies in their effort to select low-risk customers from among the large numbers of all those who seek insurance, or to set fees in accordance with the level of risk. It would be nearly useless, however, to attempt to predict the accident career of specific individuals and restrict their legal rights accordingly. This fact has long--though not universally--been recognized: "It is not to be expected that within a democracy, the authorities could impose selective treatment on any class of operators, the majority of whom show no need for it."
A very similar pattern of results has been obtained in several other studies. There is, indeed, a difference in the accident rate between drivers with previous accidents or previous violations, but the difference is so small that it accounts for only a minor fraction of all accidents that happen. From data collected in California, it can be calculated that only about 2% of all accidents in a given year were due to drivers who had experienced accidents in both of the two preceding years. About 87% of these drivers were accident-free in the third year. Not surprisingly, the researchers concluded: "Consequently, it is unrealistic to expect that programs toward reducing accidents by focusing on the accident repeater can effect a large reduction in the total accident picture." Accident countermeasures aimed at accident repeaters and violation repeaters cannot diminish the accident rate by a substantial amount, even if they were 100% effective in reducing to zero the accidents of these individuals in the near future. If complete elimination of their accidents is viewed as unrealistically high, and the accident rate of accident and violation repeaters were reduced to average levels in the population, the reduction in the future accident rate of course, would be even smaller.
That the removal of accident-involved drivers from the road can reduce the future accident rate only by a few measly percentage points was already established some time before the Second World War--at least as far as the USA is concerned--in a publication aptly called "The normal automobile driver as a traffic problem". Yet, cracking down on accident-involved drivers continues to have some political appeal, and this is probably why such action is being proposed from time to time by individuals or advocacy groups in one country or another. But from a public health promotion point of view, this approach is virtually useless. Human history is replete with cases in which an identifiable minority of people is blamed for society's ills. They are made to suffer injustices as a consequence, while the ills are not remedied. The fact is that the accident problem is not located at the far side of society's bell curve. Therefore, accident countermeasures, if they are to have a significant effect upon a nation's accident rate per inhabitant, have to be directed at the driver population as a whole, that is, basically all of us, or at least at large subgroups in the population.
It is evident that risk taking is not a personality trait that is consistent from one situation to another. Similarly, the tendency to have accidents is not consistent from one time period to another. To believe otherwise may well be an example of the "fundamental attribution error". This expression is used by social psychologists to refer to the fact that people typically attribute another person's behaviour to that person's lasting character and not often enough to that person's passing state or the prevailing environmental condition. Steal once and be forever called a thief--hence the use of the term "error", and the error is called "fundamental" because the first and natural impulse of people (psychologists not excepted!) is to make an attribution of this kind. The development of a more considered and controlled opinion does not occur unless a more deliberate and conscious effort is made.
Nevertheless, in a fashion reminiscent of the delta fallacy mentioned in Chapter 1, there seems to be a persistent urge to search for stable personality features that might predict accidents. The dogged search may be due to the occasional success in identifying a relevant factor, however small its influence. Many researchers, this author not excepted, seem to have been thinking (but yours truly no longer so) that a clearer relationship would emerge if only we had more valid tests of attitude and personality and if we had more trustworthy measures of accident involvement. The notion that current measures of accident involvement are not trustworthy is justified, since a considerable proportion of all accidents remains unreported. Maybe this problem could be overcome by looking at professional drivers. Not only do these drivers cover a much greater annual kilometrage under comparable circumstances (which would give personality traits a greater opportunity to become apparent in the safety record), but it is also rather difficult, in the environment in which these people operate, to hide any accidents from the record keepers.
And indeed, a somewhat greater stability in the tendency to incur accidents from one period to another has been observed in professional drivers. These are people (such as taxi drivers and long-haul truckers) who drive very large distances per year under relatively comparable conditions from one year to another. Bus drivers and streetcar drivers in Helsinki, Northern Ireland, Belgrade, Israel and England have been found to show greater-than-zero correlation between the number of accidents they had in one period and in a subsequent period of one or more years. Still, earlier crash involvement correlated no more than about r = +0.30 with accident involvement in the next period.
Translated into everyday arithmetic, this means the following. Suppose you have a sample of 1000 drivers who were accident-free in one period, and a sample of 1000 drivers who were not. Suppose, too, that you predict that all the accident-involved drivers will have accidents again in the second period, and those who were free of accidents would remain so. A correlation r = +0.30 implies that your prediction would turn out to be correct for a little less than 60% of all drivers and wrong for just over 40%. So, you are doing better than chance; 60 - 40 is better than 50 - 50, but not by a whole lot.
The authors of the studies mentioned above, in reviewing the evidence, concluded: "This implies that transient factors must play by far the most important role in crash causation"  --emphasis added. As they are saying this, one hears the echo of a conclusion, drawn some twenty years earlier, that "certain people are accident-prone, but sometimes only for short periods of time, and that there are others who are accident-prone over extended periods of time, perhaps for several years or most of their remaining lifetime. Furthermore, different persons are accident-prone for different reasons, and the same person may move in and out of a state of accident-proneness each time because of different circumstances." Among those different circumstances that have been identified in the literature are interpersonal problems, loss of a loved one, job and financial problems, episodes of suicidal thoughts and gestures. One study found that female and male drivers with marital difficulties were more often involved in accidents in the period surrounding the date at which they filed for divorce.
Another factor causing the lack of stability in drivers' accident records is that people may learn from past accidents, other people's as well as their own, as is argued in Section 4.2. Once bitten, twice shy, in other words. A recent accident may be followed by a period of increased caution. This is what we saw in the laboratory (see Chapter 9) and what was also observed in a German study. Drivers who had been injured in accidents rated the risk of road accidents significantly higher and adopted more safety-compatible attitudes and driving behaviours. On the other hand, having had no accident may dull people's alertness and make them more daring, at least until an accident follows. As a result, the accident likelihood of an individual would fluctuate from one time period to another, rather than be stable. The search for personality traits in accident causation appears even less rewarding than the proverbial search for a needle in a haystack. Even if you find one, by the time you do, the haystack will likely no longer be the same.
And then there is the rambling role of chance. Whether a driver error results in a reportable accident or not depends largely on factors unforeseen, if not unforeseeable. These are factors that we usually call chance, or good or bad luck. Drivers losing control of the vehicle or driving through a red light, for instance, may crash or go scot-free, depending on whether there are other vehicles in their path. Even if there are other vehicles, their drivers may perceive the danger and take evasive action so that no accident occurs. This is why even consistent poor driving does not lead to consistent accident careers. We are, in fact, so used to the environment being forgiving in one way or another that, when a crash does occur, this is called "an accident" in everyday language.
We have already noticed that far more than 50% of drivers feel that they are better than average drivers (see Section 4.6). People like to present themselves in a favourable light, not only to others, but also to themselves. Thus, it is to be expected that the very human eagerness to justify oneself will lead people to attribute any accident they had to factors other than their own incompetence, carelessness, inattention, hurry, or what-have-you. So, after an accident they may become more inclined to agree with external-type statements, while before the accident they were actually more internal. Bus and streetcar drivers in Helsinki, with accidents in the past, attributed the occurrence of traffic accidents more often to their tight schedules, fatigue and duration of work shifts than did accident-free drivers.
Performance on tests of skill may, of course, also be influenced when people know or suspect that they are being tested because of their accident history. It has been said that "drivers tend to explain their traffic accidents by reporting circumstances of lowest culpability compatible with credibility". In this effort they may be more or less successful. Witness some excerpts of supposedly true-to-life letters from Canadian drivers to their automobile insurance firms:
Most studies in the current information base are of a retrospective nature: a sample of individuals with valid driver's licences is drawn from the population, or happens to be conveniently available. They are given tests of personality, and other personal characteristics are ascertained, with the inclusion of their past accident career. It is worth noting that the correlations calculated from this information do not include the very people who, in a sense, are the most interesting from the point of view of accident prediction and accident prevention. These are the people who have had a fatal accident or were so severely injured that they cannot participate in investigations, as well as those whose driving behaviour was so deviant that their driving licences have been withdrawn. These people are not included in the samples. Thus, the most dramatic (and possibly the most telling) extreme of the distribution of accidents or violations is missing. This, in turn, may have the effect that the correlations obtained between personal characteristics and accidents underestimate the true strength of association.
One remedy for this problem is to conduct prospective studies: a sample of drivers is drawn from the general population, for instance, at the moment of licensing or even before. This is a point in time at which no accident, or at least no culpable accident, has yet occurred. Demographic data (age, gender, occupational status and so forth) and personality information are obtained from these people, who are then followed up over a multi-year period. At various points in time after the starting date, information on whether they had an accident or not is gathered and checked for any association with the personal data collected earlier.
Prospective investigations are expensive, in part because they demand large samples. This is because no more than 10 in a hundred drivers can be expected to get involved in an accident in any one year. Thus, if you wish to have a sub-sample of 100 accident-involved drivers at a point in time one year into the study, you have to start with an initial sample size of 1,000. In fact, you need even more, because a general problem of prospective studies is attrition of participants. People move, change names, or they themselves or their accident careers cannot be traced for other reasons.
Another potential problem of prospective studies is that people may change their behaviour as a consequence of participating in an investigation. Jocular minds have called opinion surveys the art of asking people about the opinions they don't have. They may then form an opinion as a consequence of being asked for one. Physicians may question their patients at annual check-ups on their eating, exercising, drinking and smoking habits. Having been asked, people may start thinking about these matters and possibly change their behaviour.
Researchers in Ottawa, Canada, obtained a sample of 1273 Grade 9 and 10 high-school students in an effort to determine whether future traffic accidents could be predicted from information collected from these youngsters, most of whom (more than 86%) did not have a driver's licence at the outset of the investigation although, three years later, 96% did. Participation attrition amounted to 20% in the second year, despite repeated attempts by the investigators to make contact. In the third year, the loss amounted to 30%, and reached 66% in the fourth year, meaning that follow-up data in the fourth year could be collected on only 34% of the original sample. Nevertheless, significant relationships between several personal characteristics and future accident involvement were reported, including that greater accident likelihood was found among those with lesser adherence to traditional social values regarding school and religious worship, as well as greater tolerance for deviance. The individuals with accidents had more liberal attitudes towards alcohol use, drank more regularly, and drank larger quantities per occasion. As well, they indicated more risky driving behaviour, including driving after the use of alcohol or street drugs, and failure to use seatbelts. Among them there were more cigarette smokers, and they reported fewer behaviours conducive to health such as having dental check-ups, a balanced diet, and regular exercise. They also reported a greater frequency of unhealthy behaviours such as eating junk food and not allowing themselves enough sleep. As regards their personality traits, they scored higher on "thrill and adventure seeking" and "experience seeking", which are two of the subscales in the sensation-seeking questionnaire mentioned above.
The researchers reported that combined consideration of all these personal characteristics allowed correct identification of 78% of the study participants as either accident-involved or not accident-involved during the follow-up period.
This finding, however interesting it may seem at first sight, does not detract in any way from our general observation that individual differences in personality characteristics only have a limited bearing on individual differences in accident involvement. This is because it can be calculated from the researchers' report that the percentage of participants who remained accident-free during the follow-up period was 77.6. In other words, if one had predicted that all participants would have been accident-free in the subsequent period of time under study, this prediction would have been correct in 77.6% of the cases. The percentage of correctly predicted cases on the basis of knowing something about the individuals should, of course, be compared with the percentage of correct prediction on the basis of not knowing something about the individuals, that is, on the basis of mere chance. In this study, there was no clear evidence that prediction based on knowledge of individual characteristics was better than prediction ignorant of individual differences.
And even if it had been better, such a finding might give an unrealistically optimistic impression of the true predictive power of the personality and lifestyle characteristics just mentioned. This is because the number of predictor variables that were found to be related to the accident criterion was appreciably smaller than the number that was tested. The implication is that some of the supposedly significant findings may have been due to chance. The larger the number of correlations calculated, the greater the likelihood that the "significant" ones are due to chance. If you find this notion of "significance by chance" a little puzzling, think of tossing a coin. Throwing heads five times in a row has a small chance likelihood, namely 0.5 x 0.5 x 0.5 x 0.5 x 0.5, which is about 3%. If you stopped tossing at that point, you would have a "significant" finding and might suspect that the coin is not balanced. But toss each of 100 perfectly balanced coins five times and you may expect five heads to turn up in about three of the 100 coins; that is three cases of "significance by chance". Now toss each of the coins five times again. Once more, you will expect to see the five-heads pattern in about three of the coins, but not in the same three. Hence, when one personality characteristic among several tested emerges as "significantly" related to accidents, one should not consider this a reliable finding until the same characteristic does so again in further investigations.
Consider a recent prospective study that was carried out on male Finnish army conscripts. Two personality traits--boldness and the tendency to be trusting--among the sixteen traits measured when they were enlisted, showed significant relation to accident occurrence during their military service. Do these results justify any confidence in the notion that individual differences in accident involvement can be predicted on the basis of personality? You would probably wish to suspend judgement until you know how these traits stand up in subsequent validation studies, especially since the study found no significant correlation for any of the six other traits which, in various earlier studies, had been reported to be associated with traffic accidents. Your confidence would probably be further reduced by the fact that individual differences in the accident rate of the soldiers before induction into the armed forces bore no relation to their accident involvement while in military service. This reflects zero-order reliability of individual differences in the accident rate from one period to another. There was no relationship between the number of traffic fines obtained before and during military service, or between the number of traffic fines incurred by the men before they entered the service and while they were in the army, and their pre-army violation rate was not related to accidents while in the service. In short, there was no connection between violations and accidents, and both were unstable over time.
On the other hand, the men who had more than the average of 64,000 km of driving experience before being enlisted were found to be less likely to have an accident during their military service. This is a finding that deserves more general credence.
In the last several years, in Ontario, about 16% of all licensed drivers were between 16 and 24 years of age, yet they accounted for about 30% of all fatalities. They were thus 1.9 times more likely to be killed in road accidents than drivers on average. This overrepresentation, although it varies in size from country to country and from time to time, is an international phenomenon. It holds true both per km driven as well as per person. There is also some indication that elderly drivers have more accidents per km driven than those of middle age, but not necessarily more per person-year. This is because they tend to drive much less, thus reducing their exposure to accident risk. Furthermore, males are more often involved in accidents as compared to female drivers, both per km driven and per person-year.
It takes novice drivers some five to seven years for their accident rate to drop to the average accident rate of the driver population as a whole, and the younger the driver at the time of licensing, the longer this period. This implies that the overrepresentation of novice drivers in the accident statistics is due to two different factors, immaturity and inexperience. In Ontario data, for instance, the overrepresentation of 16-year-old males during their first year of driving is about twice as great as the overrepresentation of male drivers aged 30 or older during their first year of driving. In their fourth year of driving, male drivers who obtained their licence when they were 30 years or older had about the same accident rate per 1000 drivers as the average of all male drivers in the population. In contrast, males who had acquired their licence when they were 16 incurred accidents at a rate about 40% higher during their fourth year of driving than was true for all males drivers in the population.
This raises two questions: what is it about being a late teen, or a sprouting adult, that makes young people more likely to have accidents, and why is inexperience related to greater danger? Before attempting to answer these questions, let us first consider what requirements a driver must fulfil to prevent a potential accident from occurring.
Obviously, (1) the driver must be awake. But to be awake is not enough. The driver may be fully awake, but inattentive to the driving task and paying attention to other things, for instance, the cellular telephone, the radio or conversation with a passenger. Thus, (2) the driver must be attentive to the traffic situation. But paying attention to the traffic situation is not enough, because the driver may not have the sensory abilities (vision and hearing, among other things) to clearly perceive the danger-relevant aspects of the traffic situation. Thus, (3) the driver must have the necessary sensory abilities. But having the necessary sensory acuity is not enough, because the driver may fail to be aware of the amount of risk that is contained in the traffic situation he or she perceives. Risk perception, like the perception of beauty, is a product of experience and reasoning, however rudimentary, intuitive or subconscious these may be. Thus, (4) the driver has to be able to infer the amount of accident risk that is contained in the traffic situation. But risk recognition is not enough, because for the driver to be motivated to take action to reduce the accident risk, the amount of risk acknowledged must be greater than the level of risk the driver is willing to accept. So, (5) the risk must be greater than the driver is willing to tolerate. But the wish to reduce the risk is not enough, because the driver may or may not have the ability to decide what should be done in order to reduce the risk. Therefore, (6) the driver has to have the necessary decision-making skill for risk reduction. But to have the decision-making skill is not enough, because even the driver who knows what ought to be done may not have the vehicle-handling skill to carry out the necessary manoeuvre. Hence, (7) the driver has to have the necessary vehicle-control skill. Moreover, all these conditions must be fulfilled at a point in time at which an imminent accident can still be averted.
This simple logic might suggest that there are as many as seven different factors involved in accident avoidance, but this is a misconception since these factors do not operate independently of one another. If a person's decision-making or vehicle-handling skills are poor, that person's level of perceived risk should be high, and if it is not, this reflects overconfidence in one's skills. The same holds for sensory abilities. To be colour-blind or hard of hearing does not imply a significant increase in accident liability, provided the driver considers these handicaps in the estimation of risk. Poor night vision will not increase a person's accident risk unless the person is unaware of it or is willing to accept high levels of accident risk. Deficiencies in skills and sensory functioning, other things being equal, can increase a person's accident likelihood only to the extent that these deficiencies are being underestimated by the person in question, and thus lead to an inappropriately low level of perceived risk.
Poor skill will not enhance a person's accident risk if that person is fully aware of his or her poor skill, because--risk acceptance level being the same--that person is less likely to engage in manoeuvres he or she cannot handle very well. It is not surprising, therefore, that sensory abilities and other driving skills have generally been found to show no association, or weak association, with accident involvement.[35,36,37] In Section 6.3 we noticed that better-than-average driving skill--presumably both on the level of decision making and vehicle handling--may sometimes be associated with greater accident likelihood, because the driver is being lulled into an illusion of safety.
Inattentiveness to the driving task implies either that the driver estimated the risk of accident as very low and permitted paying more attention to other things, or that the driver considered other things to be at least temporarily more important than safety, thereby accepting a higher level of accident risk.
Finally, lapses in wakefulness during driving will not occur unless the driver underestimated the chances of falling asleep behind the wheel or accepted them. Here again, the underlying cause is either risk underestimation or increased risk acceptance.
It may thus be concluded that the seven factors in our original analysis shrink to only two that are truly relevant to increased accident likelihood. It may further be argued that individuals with a very low risk acceptance will be motivated to ensure that they do not underestimate risk and that, therefore, only one dominant factor remains: risk acceptance.
We now will attempt to explain why inexperienced drivers have more accidents than experienced drivers. Obviously this is not because of their lower level of skill per se. In principle, unskilled people can reduce accident risk by choosing manoeuvres that match their level of skill in driving, and by reducing their exposure. However, in practice, this is not so. They cannot fully adjust their driving manners to their driving skill, because they operate as a minority in a road system in which most drivers are experienced. There are strong forces at work that compel inexperienced drivers to drive at a certain speed and at a certain following distance, and to do other things similar to what the more experienced majority does. Thus, in order to acquire experience, they have to drive above their own level of competence and comfort, and that is why they experience more risk when driving.[38,39] Their elevated experience of risk corresponds with the increased risk they incur.
The point made here is neatly illustrated by the case of "Mrs. Cautious Driver". This is the name we will give to a lady whose problems were presented at the Grand Rounds in Psychiatry in the Kingston General Hospital in Ontario in the mid-'70s. She suffered a serious nervous breakdown and an acute case of driving phobia. This is her story.
In the course of four years she experienced four traffic accidents. This is a rare occurrence, with a frequency of less than one in 10,000 drivers. In none of these accidents was she at fault. That makes her predicament a much rarer occurrence still. In all cases she was driven into at an intersection by another car. She was an extremely cautious driver in the sense that she drove her station wagon well below the speed limit on four-lane highways and she always buckled her seatbelt in a period before this was compulsory by law. At stop and yield signs, she was in the habit of waiting very long before she would accept a gap wide enough to her liking and go ahead. In such circumstances she would sometimes brake and stop again. Three times she was rear-ended in this situation, the third time by a police patrol car.
Mrs. Cautious Driver presents an interesting paradox. She was very careful indeed, in fact, so careful that her behaviour was rather unpredictable to other drivers. This made her liable to having accidents. On the other hand, if everybody were to behave as cautiously as she, there would be fewer crashes.
Greater risk, and acceptance thereof, is inevitable if one wishes to become experienced. Experience must be bought; accidents are part of the price. Nobody can expect to be able to learn to play the violin and perform a piece at the required tempo without making many mistakes in the learning process. There is no royal road to learning.
The number of mistakes made by the novice violinist and the inexperienced driver could be reduced if they were allowed to perform at a reduced pace, solo, and in concert with other inexperienced persons. In other words, there are reasons for assuming that the accident rates of novice drivers would have been lower if there were only novice drivers making use of the roads. It is not their inexperience per se, but the experience mix in the collective of road users that makes the inexperienced driver more prone to having accidents.
This view is supported in findings obtained by the method of verbal risk ratings described in Section 3.3. Drivers of different levels of experience orally express, on a numerical rating scale, the level of risk they perceive while driving. An observer sitting in the front passenger seat, who may be either experienced or inexperienced as a driver, makes observations using the same scale. The experimenter does not allow any communication between driver and observer and keeps a record of their independent risk estimates.
It is found that inexperienced drivers' danger ratings tend to agree more with other inexperienced drivers' ratings than with those of experienced drivers. Similarly, ratings by experienced drivers do not agree with those by inexperienced observers as much as with experienced ones. In simpler words: drivers of similar levels of experience show greater similarity in the amount of risk they perceive. In a situation of traffic conflict, an experienced driver will, therefore, be better able to predict what another experienced driver will do to avoid a collision. Similarly, inexperienced drivers will predict the reactions of other inexperienced drivers more accurately than they will predict those of experienced ones.
The implication is that the overrepresentation of novice drivers in accident statistics--insofar as it is due to lack of experience--might be reduced by techniques that speed up the process of learning to perceive the risks of the road in the same manner as experienced drivers do. The current research interest in risk-perception skills[40,41] may thus lead to the development of new didactic methods.
In order to explain the overrepresentation of novice drivers--especially male--in relation to their youthfulness, a variety of factors may be surmised. Young people have higher stimulation-seeking scores, and we know that such scores are associated with a tendency to view the risk in "risky" activities as rather low (see Section 10.1). Moreover, young people tend to have fewer responsibilities to others; they are less likely to be married and have children. and they have fewer accomplishments. Thus, apart from the potential loss of a few more years of life, they have less to lose by taking risks. These are factors that may be assumed to have an increasing effect on the level of risk they are willing to accept. At the same time, they have more to gain from risky behaviour. For example, by showing bravado they may gain prestige among their peers. Furthermore, the general culture expects them to be daring and venturesome: young colts will canter. They often drive cars that are not their own. Thus, they would seem to lose less and gain more from risky conduct. Similar factors may well explain why men have more traffic accidents, per person-year as well as per km driven, than women.
Returning, finally, to the motto at the heading of this chapter, should we conclude that people drive as they live? The original study which affirmed this "has serious methodological shortcomings". For one thing, the drivers were questioned by interviewers who knew their accident history and thus could keep asking questions until they received answers which satisfied their need to explain the interviewee's accident record. This has led to a gross overestimation of personality and lifestyle factors in the causation of accidents. In other words, an association does exist, but it is so tenuous that it would seem unwise to focus upon it in prevention efforts that wish to achieve more than accident reduction by just a few percentage points. It may be more promising to attempt to alter the driving style of the population as a whole, or a demographic subgroup, and to do this by focusing on the level of risk people are willing to accept. This is the topic of the next two chapters.
[TABLE OF CONTENTS] [PREVIOUS CHAPTER] [NEXT CHAPTER]