Who Are the Safest Drivers on the Road?

Who Are the Safest Drivers on the Road? You're in for a Surprise

Marc Green

Periodically the news will report a story about an older driver who plowed into another vehicle, pedestrians on the sidewalk or something similar. This immediately raises cries that older drivers should be kept off the roads. Experimental and epidemiological research seemingly supports this view. Shelves of experimental research report that aging produces losses in most sensory, cognitive, and motor abilities. Other studies demonstrate deficits in many driving tasks such as response-time and lane maintenance. These aging deficits suggest that older drivers should be riskier and have more collisions. Many epidemiological studies seemingly confirm this prediction by performing statistical analyses of crash data and finding that older drivers have more accidents per mile/km driven. More detailed research shows that this conclusion is wrong. Older drivers are the safest drivers.

While the experimentally measured deficits are largely true, the apparent crashes/mile metric fails to take two important factors into account. One is the "low mileage bias." People who drive large numbers of miles are typically driving more on limited access freeways and interstates where accidents a relatively rare due to absence of intersections and other complex road designs. Analyses that have corrected collision data for the low mileage bias by matching self-reported travel distance, found no evidence of higher crash involvement by older drivers, regardless of age (e.g., Hakamies-Blomqvist, Raitanen, & O'Neill, 2002; Langford, Methorst, & Hakamies-Blomqvist, 2006). After a suggestion that the self-report data might be unreliable (Staplin, Gish, & Joyce, 2008), the authors used more objective measures and modulated their conclusions somewhat in a later paper (Langford, Koppel, McCarthy, & Srinivasan, 2008), but still found that the bias causes a large overestimation of older driver collision risk. Another study (Alvarez & Fierro, 2008) found among drivers with similar yearly miles driven, 75+ drivers had the lowest crash risk. Evidence of the low mileage bias has been found in several other studies (Langford, Koppel, Charlton, Fildes, & Newstead, 2006; Antin, Guo, Fang, Dingus, Perez, & Hankey, 2017), although it may not apply to rural drivers who seldom encounter demanding roads (Hanson & Hildebrand, 2011). Lastly, Rolison & Moutari (2017) found only a slight rise in the fatal collision rate of the oldest drivers when the data were corrected with a risk-exposure index. The authors further commented that this small rise might be due to a second factor, the "frailty bias." Older drivers are more likely to be injured or killed in a collision so their accidents are more likely to be reported.

The newer findings illustrate important points both about the effects of age on driving and on interpretation of epidemiological studies. The earlier studies selected a misleading exposure metric and exhibited the "low mileage bias" by simply averaging over miles driven and did not consider actual driving habits and differences in driving environments. Older people drive less in general but, more importantly, also make many fewer long trips (between cities, etc) which builds up total mileage. The low mileage bias inflates the percentage of collisions involving older drivers since they travel fewer miles on shorter trips in more hazardous urban conditions (e.g., Planek, 1972). Accidents on interstates are spectacular but relatively low per mile because traffic is lighter, attentional demands lower and quick reaction is less often required. Older drivers are more likely to drive shorter trips within more congested urban areas where overall accident rates are higher and where demands on perception, attention and decision-making are greater. On the other hand, people who limit their driving for any reason, young or old, may be poorer drivers. Poorer drivers may avoid freeways and interstates. It would then be difficult to tell whether the accident rate variation is due to age or to the driving environment. This assumes that driving is primarily a skill-based task, a point that I return to later.

Yet other metrics also say that older motorists are actually the safest drivers. The term "older driver" is problematic because it covers a very wide range of ages. It is more useful to look at collisions per decade of age. Figure 1 shows the results of a study (Braver & Trempel, 2004) that measured fatal and injury crashes using the metric of number of drivers per 100,000. It also decomposed the harmed parties into road user classes.

Figure 1 Deaths (left) and injuries (right) per 100,000 drivers of various age groups. (After Braver & Trempel, 2004)

The left graph shows fatal crashes. The death rate road users outside the older driver's vehicle falls continuously up to age 74+ where there is a very slight uptick. The data for drivers exhibit a more pronounced rise. This is almost certainly due to the frailty bias. There is a lesser uptick for passengers but this is likely also the effect of frailty because elderly drivers are more likely to have older passengers such as an elderly spouse. In sum, the danger to other vehicle occupants and pedestrians is very low. The data for injuries tell a similar story: drivers become safer as they grow older. That is, the safest drivers on the road are older drivers. There would be more justification for banning drivers under age thirty than for banning 75+ old drivers.

Research using additional metrics reaches a similar conclusion. A Highway Loss Institute Study (2012) found that claims per 100 insured vehicle years bottomed out at ages 60-64 and that even drivers aged 75-79 had lower claims rates than drivers below the age of thirty. Compared to the 75-79 age group, teenage drivers had 65 percent greater rates and 20-24 year-old drivers had 50 percent greater rates. Similarly, Regev, Rolison, & Moutari (2018) equated drivers of different age decades for number of trips and number of licenses. Drivers aged 70+ had the lowest fatal and non-fatal crash risk except for drivers aged 60-69 and 17-20 who surprisingly had very low corrected rates. If the definition of "older driver" rises beyond the age of 75-80, then there is some evidence that risk increases. However, some studies find an uptick in risk at 75+ and others only at 80+. In sum, "older drivers" in their 60's are the safest drivers on the road. Older drivers up to 75 are only slightly riskier. It isn't until age 75+ where a significant risk increase begins to appear and this mail simply be a frailty effect. Even this depends on operational definitions since there are metrics which push up the age where risk begins to grow. Drivers 85+-years-old had lower claims than drivers under age 25 (Highway Loss Institute Study, 2012). There is less risk in a one-year license renewal for a 70-year-old than for a 40-year-old (Evans, 2000).

These results go against typical stereotypes of the older drivers. The news reports highlight instances of older drivers having collisions giving the false impression that it is a common occurrence. This is what might be expected based on the experimental demonstration of sensorimotor losses in aging. Yet much like the failed predictions that talking on cell phone increases crash risk, the behavior deficits of older drivers have not translated to higher risk on the road. Why not?

The reasons are similar. Experimental research does not translate simply to the real-world and epidemiological models rely on many assumptions. Older drivers tend to restrict their driving to safer conditions, e.g., daylight, low traffic volume, etc., or they simply don't drive at all. However, the major reason is that driving is not primarily a skill-based task so losing skill is not a major handicap. If it were a skill-based task, then race car drivers would be the safest drivers, but they are actually the riskiest (Williams & O'Neill 1974). Instead, driving is a self-regulated task where drivers choose their speed, inter-vehicle spacing, gap size, etc. Older drivers travel slower, leave longer spacing and wait for larger gap, etc. The general name for such behavior is "risk compensation" because the older driver trades off speed and efficiency against diminished sensorimotor skill. In research studies, moreover, older drivers typically must follow the experiment protocols and are not allowed to self-regulate. This is also one of the reasons that research studies erred in predicting that drivers talking on cell phones would have a higher crash rate. Confirmation bias is a powerful thing. Researchers tend to find what they are looking for.

While older drivers may overall be the safest drivers, there are likely specific situations in which they are not. They may be less likely to avoid a collision when the event is unexpected (a pedestrian runs across in front of the car) and where keen vision (night) and fast response time is necessary. The lesson is that there are two distinct ways of defining safety and risk. One is "general," safety and risk aggregated over all events. The other is "specific risk," the risk in a given situation. An analogous distinction exists in other fields such as finance which distinguishes between "specific (or unsystematic) risk" and "systematic risk." Older drivers may be less safe when risk is defined unsystematically, but they present the lowest systemic risk.

A division of 2057949 Ontario, Inc.
Copyright © 2024 Marc Green, Ph. D.
Home Page
Contact Us