What good are laws in the absence of morals?
Horace, Odes[1]
|
| Quid leges sine moribus! What good are laws in the absence of morals? Horace, Odes[1] |
Figure 8.1 refers to a classic study that was conducted in the city of Grand Rapids in Michigan, USA.[2] At the time, this city had just over 200,000 inhabitants. A group of some 6000 drivers who had been involved in accidents and survived were tested on their blood alcohol concentration (BAC). They were also interviewed on their drinking frequency. For the purpose of comparison, some 7600 other drivers were stopped while passing the same sites without accident, and interviewed by the researchers.[2]
As you can see from Figure 8.1, it was found for all drivers that their likelihood of belonging to the accident group was greater as their BAC was higher, regardless of their self-reported drinking frequency. In other words, the higher your BAC, the greater your accident likelihood in traffic--and there is no critical BAC level at which your chances of an accident suddenly rise, as the law in your country may suggest. As long as you remain below the legal limit, your chances of being arrested for driving after drinking will be greatly reduced, but even below that limit, your chances of getting involved in an accident increase with the amount of alcohol taken.
Figure 8.1: Relative probability of crash involvement in subgroups of drivers of different drinking frequency as a function of their blood alcohol concentration (BAC) expressed as percentage weight of alcohol by blood volume.[3]
It is no less important to realize that sobriety is no guarantee of safety. A very large proportion, say, one-half or more, of people killed in accidents are sober, i.e., they do not have a measurable trace of alcohol in their bodies at the time of their accident. So, does it really make sense to believe that the traffic accident rate will go down if laws are successful in reducing the amount of drinking before driving? Yes, it would make sense if we had reason to believe that other behaviours that are relevant to safety would not change. But it would not make sense if we had reason to believe that people would become less cautious in other ways. So, the to-be-expected effect upon safety depends on the way people respond to the legislation. If the legislation increases the desire to be safe, it will reduce the accident rate. If the legislation fails to produce an increase in people's desire to be safe, but merely forbids one way of behaving unsafely, people may obey that legislation, but are likely to behave more unsafely in other ways, and the accident rate will not drop.
Legislation that threatens drivers with severe, swift and consistently enforced penalties for drunk driving may well be heeded in the sense that fewer people will decide to drive after drinking, or they will decide to drive only short distances, or they will attempt to drive more carefully and less conspicuously, or perhaps a combination of the above. Arrests for drunk driving would then be expected to drop, as would the percentage of all drivers with BACs over the legal limit who are killed in accidents.
Such are the potential effects of the legislation, but would it be reasonable to expect that the fatal accident rate in traffic per head of population would actually go down? Perhaps, but only to the extent that the legislation increases the desire to have no accident, and that extent may be very small.
Figure 8.2 shows a marked reduction in the BAC levels of drivers killed in traffic accidents in the USA between 1980 and 1987. There was a very noticeable increase in the percentage of drivers killed who had zero BACs at the time of accident, but there was no commensurate reduction in the traffic death rate per capita. That something similar seems to have happened in Canada may be seen from Figure 8.3, while Table 8.1 offers more detailed information.[4] These observations prompt the suggestion that the drunk accidents have somehow been replaced by sober accidents. Instead of accident reduction, there has been accident metamorphosis.
Figure 8.2: Traffic deaths per capita and changes in BAC levels in drivers killed in the USA 1982-1986.[6]
Figure 8.3: Traffic death rate per capita and changes in BAC in drivers killed, Canada 1973-1986.[7]
By way of example, a crackdown on drunken driving carried out in 1977 in British Columbia may have been counterproductive in that there is a suggestion that it led to an increase in the overall number of road deaths. In that year, without a change in the law regarding drinking and driving, the 2.5 million inhabitants of this part of Canada saw a major and highly visible intensification of BAC enforcement practices by police that was given very prominent attention in the mass media. Conspicuous mobile Blood (breath) Alcohol Testing units (the "BATmobiles") were deployed and located from one high-volume traffic site to another. The total number of times drivers were stopped and checked by the police in 1977 was equivalent to 30% of the total number of vehicles registered in British Columbia! By means of time-series analysis, it has been estimated that the enforcement programme produced an 18% reduction in the number of alcohol-related traffic fatalities, but a 19% increase in the overall number of traffic deaths, alcohol related or not.[8] A Pyrrhic victory: the battle was won but the war was lost; the operation was successful, but the patient died. Is this the price to be paid for well-intentioned but conceptually unsophisticated efforts to reduce the accident rate? J'accuse.
The observed pattern of results may have been due to the following mechanism. Let us assume that the programme was effective in the reducing the BACs of drivers. Those who used to drink and drive prior to the programme now refrained more often, but as their target level of risk had not been lowered, they adapted by driving more, with more passengers in the car, driving faster, less attentively or whatever. The other drivers, those who did not drink and drive before the crackdown, were under the impression that it was very successful in removing the drunks from the roads and now felt less of a need to refrain from driving during the risky hours or "to watch out for the other guy".
Such a scenario is neatly illustrated by some items that appeared one day in a Toronto newspaper. The government in Ottawa had just announced an increase in penalties for drinking and driving. On the front page we were told: "Blitz against drunk driving is paying off. The massive Metro Blitz against drinking drivers appears to have been a success. So far this month, a record 71,718 motorists, 22,000 more than last year, have been stopped by police and the number charged with being impaired is 321, down 150 from last year's 471. `It certainly looks like the lesson is being learned,' a police official said yesterday". . . . On the gloomy side, 11 people were killed on Metro streets and four more on the highways within Metro this month, compared to only four last year." Inside the same issue, a letter to the editor states: "Now that [the government in] Ottawa appears to have picked up the cudgel with harsher penalties for drunk driving it's going to be a pleasure to drive on our roads once more."[9]
Therefore, popular overestimation of the contribution of alcohol to the riskiness of the roads, in combination with the enforcement and mass media activity, may have lulled drivers into an illusion of safety and thus created a short-term increase in the per capita traffic fatality rate. This particular interpretation in terms of the lulling effect (see Section 6.3) may well be speculative. Similar interpretations in the future will likewise remain speculative until programme evaluation research not only looks at the end effect, but also includes the collection of data on the behavioural processes that take place between countermeasure input and accident rate output.
More realistic conditions were created in an experimental study in the Netherlands on the effect of seatbelt wearing on driving style--using a real car on real roads (a 105 km freeway that makes a circular connection between Amsterdam and the cities of Utrecht and Amersfoort). When habitual, "hard-core" non-users of seatbelts complied with the experimenter's request to buckle up, they drove faster than without seatbelts, they followed more closely behind a vehicle in front, they changed lanes at higher speeds and they braked later when approaching an obstacle. A follow-up study over one year showed that these behaviour changes persisted over time.[11]
| 1980 | 1983 | 1986 | 1987 | |
|---|---|---|---|---|
| Road deaths per 100,000 people | 22.5 | 18.2 | 19.1 | 19 |
| Dead drivers per 100,000 people | 12.7 | 10.3 | 11 | 11 |
| Dead passengers per 100,000 people | 5.8 | 4.6 | 4.8 | 4.8 |
| Occupant deaths per 100, 000 people | 18.5 | 14.9 | 15.9 | 15.8 |
| % of all dead drivers wearing seatbelts | 2.5 | 3.2 | 14.6 | 18.2 |
| ditto for dead passengers | 2.1 | 3.9 | 14.4 | 18.3 |
| % of all dead drivers having zero BAC | 33 | 40.8 | 44.9 | 47.1 |
| ditto, BAC between .01 and .05 | 6 | 5.1 | 5.8 | 5.2 |
| ditto, BAC between .06 and .09 | 5.9 | 5.1 | 5.1 | 4.9 |
| ditto, BAC between .1 and.15 | 15.1 | 13.4 | 12.6 | 11.5 |
| ditto, BAC between .16 and .2 | 16.4 | 14.9 | 13.2 | 12.8 |
| ditto, BAC greater than .2 | 23.5 | 20.7 | 18.4 | 18.4 |
This would seem quite plausible, but not so, according to some people in the road safety community. Here are the words of a member of the British parliament, quoted in a popular traffic safety magazine in 1986: "It is interesting to note that the only arguments that have been advanced against the [proposed seatbelt wearing legislation] have been made by the provisional wing of the lunatic fringe of the libertarian lobby."[12] Human understanding is limited; the inclination to display that limitation loudly, unfortunately, is not.
How can we explain why some people went so far as to accuse others of lunacy in a case like this? First, there is the evidence they see: drivers who are wearing their seatbelts are much more likely to survive a crash than those who are not. This evidence is reliable; it has been produced in many studies in many different countries. Secondly, they were probably unaware that it does not necessarily and logically follow from this evidence that more people would survive traffic accidents if all drivers were compelled by law to use the seatbelt. That would logically follow only if all other relevant factors, including road-user behaviours, remained the same. They did not consider that habitual non-users of the seatbelt might alter their driving style as a consequence of being compelled to buckle up. They did not consider the possiblity of behavioural compensation for changes in risk. Thirdly, they may simply have been blinded by their zeal to do something quick and easy for safety.
But their lack of awareness of compensatory behaviour is astounding because these very people are the among the first to proclaim that they would refuse to drive if they did not have a seatbelt in their car. A small amount of introspection should have been sufficient to make them realize that they themselves are subject to the phenomenon of risk compensation, in that they are willing to expose themselves to the dangers of traffic only if they have assured themselves of a degree of protection. They must surely know that the protection is only partial, because even buckled-up drivers get killed in accidents.
To say that one is willing to drive provided one has a crashworthy car, a seatbelt, collision and liability insurance, and so forth, is to express the effect of risk compensation upon one's behaviour. It is of more than passing interest to note that, in the just-mentioned Dutch seatbelt study, there was also a sample of habitual seatbelt wearers. These people were happy to comply with the experimenter's request to drive the 105 km route with the seatbelt on, but all of them refused to do so with the seatbelt unbuckled. To refuse sexual intercourse unless protected by a condom, to refuse to go skiing unless a first-aid station is nearby, to refuse to stay in a hotel unless it is equipped with smoke alarms, to refuse anything unless at least partial protection from disaster is provided, is to tacitly admit to the essence of risk homeostasis theory.
The people we were speaking of above may not have been much inclined to introspection, but had they been more attentive to the already existing evidence regarding the effects of seatbelt legislation, they would have had another reason for doubting the law's effectiveness. Dr. John Adams of University College, London, UK, had already published his much-discussed analysis of the trends in traffic fatality rates in countries with and without seatbelt-wearing laws.[13] Figure 8.4 summarizes his findings.
This figure clearly shows that the fatal traffic accident rate in the countries that introduced seatbelt legislation dropped to levels well below of what had been experienced before. We should, of course, be warned that the economic juncture might have something to do with this (see Section 5.4). But what this figure shows, too, is that traffic fatalities also decreased in countries without such legislation. In fact, the drop was even somewhat greater in the latter. Could this possibly have been due to the lulling effect discussed in Section 6.3? Could it be due to the fact that, in countries in which seatbelt-wearing became mandatory, the public was told over and over again in mass media campaigns that "seatbelts save lives"? In other words, could it be due to the public coming to believe that wearing the seatbelt would give a greater safety advantage than it actually does?
The answers to these questions may be uncertain, but surely these findings should have been taken into account by British lawmakers in the mid-eighties and again by American legislators several years later. Although British and American legislatures discussed and introduced seatbelt legislation some 10 to 15 years later than did continental Europe, apparently very little had been learned by the latecomers. Is it true, then, that what we learn from history is that we learn very little from history? Even so, there is hope that those who realize this will escape the doom of this predicament and rise to a level of understanding from which they will be able to take measures towards real progress.
Figure 8.4: Indices of annual road deaths in countries with and without seatbelt wearing laws. Dots indicate the dates at which legislation came into effect.[14]
The very few experimental studies that have been carried out do not support these general beliefs about the effectivenes of enforcement. That is the opinion of a team of researchers who tried to determine whether there were any effects on the rate of traffic accidents caused by naturally occurring variations in enforcement activity, after these had taken place. Some time ago, the police department of the city of Nashville, Tennessee, decided to crack down on moving violations in traffic. The number of charges rose to 52% above normal. At the same time, there happened to be a salary dispute between the police force and the city, and this was the cause of the next variation in enforcement activity. The heightened enforcement activity was interrupted by a tactic that the police officers undertook in order to strengthen their negotiating position in the wage dispute: they reduced the ticketing activity to as little as 36% of what was normal. Some time later, the dispute was resolved and the number of charges laid for moving violations returned to the usual level.
So, there were four periods in which the frequency of charges was first 100%, then moved upward to 152%, then downward to 36% and finally back to 100%. The slowdown action of the police was widely publicized in the mass media: radio, television and newspapers. The frequency of accidents with property damage, personal injury and fatalities was tracked over these periods.
Did the variations in enforcement have any consequences for the frequency or severity of accidents occurring? Not so, according to the investigators, who concluded that "the present retrospective analysis of police traffic enforcement shows that wide variations in the overall levels of enforcement have no immediate measurable impact on the frequency or severity of traffic accidents, even when these interventions are highly publicized".[15] It may be that a much greater intensification of police enforcement would have a measurable effect, but it is questionable that it would last. Suppose it were physically, financially and politically possible to increase enforcement by a factor of ten or more. This would be expected to lead to a reduction in the rate of traffic violations. Suppose also that it would lead to an increase in the perceived cost of an accident, because associated violations of the highway code would be more likely to be noticed by the authorities, with all the unpleasant consequences thereof. We can assume that a noticeable reduction in traffic accidents would occur.
But as the rate goes down and the accident problem is reduced, the general public's concern for social problems other than road safety will probably increase. Issues other than traffic (violence, theft, vandalism, drug abuse) would become more salient in the political arena and the police forces would no longer receive the pressure or support necessary for maintaining the enforcement rate for driving offences. Traffic enforcement activity would diminish again; the driving public would discover this and become more inclined to violate traffic rules and regulations.
A gradual return to the original accident rate would be the end result. This is not to say that a short-term drop in the accident rate could not be worthwhile, but the fact that the reduction is bound to be temporary should be recognized. Here again, it becomes apparent that the crucial variable for lasting reduction in the accident rate per person is the level of demand for safety in the population at large.[16] Crackdowns are aptly called crackdowns: there is a sudden eruption of activity, but that will be followed by a let-up, leading to a return to the previous equilibrium between the rate of violation by road users and the surveillance intensity of the police. No upheaval lasts for long, as the Icelanders say.
Japan's fatal traffic accident rate per head of population is reported by Professor Koshi, at the University of Tokyo,[19] to have been reduced by one-half in the period from 1970 to 1983. Part of this may have been due to the energy crisis and a slacking economy, which is known to be associated with a reduced accident toll, as we have seen in Section 5.4. In fact, the average rate of total unemployment (part-timers being considered as employed) virtually doubled from the ten-year period prior to the oil crisis in 1973, to the ten-year period following the crisis.
Moreover, Japanese authorities have taken a number of deliberate measures that might well have had a major reducing effect upon the target level of risk in the actual and potential driving population. Here are some observations in Prof. Koshi's publication:
To an external observer, these measures would seem draconian enough to reduce people's appetite for driving in general, and for driving in a risky manner in particular. Although it may not be possible to determine to what quantitative extent the level of target risk has been reduced by the various factors above, it would seem that the accident toll can be drastically reduced by severely punitive legislative and enforcement measures, provided that such measures are sustained over time, and further helped along by a weakening economy.
Thus, there would seem to be no justification for interpreting the Japanese experience as "totally at variance with risk homeostasis theory". On the contrary, the suggestion that it is compatible with it appears to be supported by some additional calculations on the data published by Koshi. The death rate per billion km driven (motorcycle and moped kilometrage being excluded) fell on average by 11% per year between 1966 and 1982, while the motorized kilometrage rose by an average of 8% from year to year in that same period.[20] The product-moment correlation between the two annual rates amounts to r = -0.97. In other words, those years that were marked by relatively large decreases in the death rate per km, were also marked by relatively large increases in kilometrage per capita. This pattern of findings, which can also be observed in American, British and Canadian data,[21] seems to agree surprisingly well with the longitudinal deduction from risk homeostasis theory: within a country there is an inverse relationship between the changes from year to year in the accident rate per km driven and the changes in the motorized kilometrage per head of population (see Table 5.1 and Section 5.2).
[TABLE OF CONTENTS] [PREVIOUS CHAPTER] [NEXT CHAPTER]