Aggression and Accidents

Last post, I presented research showing that men are more deadly than women when behind the wheel. Researchers presume this is because men have a predisposition toward aggression and thrill-seeking, thanks to the testosterone that helped our male ancestors stalk, struggle and seduce their way to successful gene replication.

This narrative makes the assumption that aggression and sensation-seeking lead to bad driving. Do they?

Marianne Junger, Robert West, and Reinier Timman think so. They examined traffic accidents in the Netherlands, using police data to determine if the reckless driver in each crash had a criminal record. They found that drivers who contributed to accidents were much more likely than the general public to have been guilty of violent crime, vandalism, property crime, and serious traffic violations. Aggression and risk-taking do indeed appear to be associated with bad driving.

The authors attribute the apparent link between lawlessness and risky driving to “a general disregard for the long term adverse consequences of [one’s] actions [which] could be labeled risk-taking, impulsiveness, or lack of self-control.” So, on the admittedly unlikely chance that Dennis Rodman offers you a ride in the near future, think twice.

Is a predisposition to this kind of aggression and thrill-seeking a biological trait? Many researchers believe it is, but other work suggests the issue may be cultural as well. Scholar Michael Sivak has done interesting research on the issue of aggression, thrill-seeking and bad driving, this time on a society-wide basis instead of an individual one.

Sivak examined traffic fatalities in the 50 states plus DC. There is a surprisingly wide gap in accident rates between states. Hard as it may be to believe for those of us who have braved the roads of Boston, the safest state is Massachusetts. Its residents experience about one third the number of fatalities per mile driven as residents of Montana, the most dangerous state.

Sivak tested a number of factors that might cause a place to have more or fewer road fatalities. Three variables came up as statistically significant. The first was the share of the population of drivers who are under age 25. States with proportionately more young drivers tend to have more road fatalities. As anyone who has ever: 1) been a 16-year-old, or 2) been driven by one can tell you, the fact that young drivers have safety issues is not necessarily something we need lots of fancy econometrics to adduce. More on the behavior of young drivers another time.

The other significant variables were homicide rates and accidental death rates; when all else is held equal, both are generally higher in states with lots of deadly traffic accidents. (In fairness to the good people of Montana, their state, with a low murder rate, is an exception.) So our suspected link between aggression (murders), thrill-seeking (non-auto accidental deaths) and fatal crashes appears to be confirmed.

Sivak helpfully updated the original study in 2009. This time, he found that seven variables are important predictors of fatal traffic accident levels on a state-by-state basis:

  • homicide rates (a + relationship with road deaths)
  • the size of the state’s physician population (a -? relationship with road deaths, presumably because more doctors indicates better medical care)
  • the share of older drivers (+ deaths, which I’ll also address another time)
  • gender (more men equals + deaths, confirming what we’ve learned earlier)
  • seat belt usage rates ( – deaths, for obvious reasons)
  • income per capita ( – deaths as income rises)
  • statewide deaths from alcoholism (a variable designed to capture drunk driving levels and a +)

As a package these variables were very powerful predictors of fatality rates. And the single most significant variable was the homicide rate. (Note that this model didn’t include a non-auto accidental deaths variable.)

It is doubtful that fatal traffic accidents are causing drowning in swimming pools or that committing the act of murder somehow causes a person to drive recklessly (except maybe when he’s driving the getaway car).

So there is almost certainly a hidden force at work, which varies across societies and which causes criminal aggression, thrill-seeking, reckless behavior and bad driving. I can only speculate, but perhaps this general incivility has to do with things like the effectiveness of local policing, the local unemployment rate, and particularly local cultural and behavioral norms, for example in terms of attitudes towards authority.

In any event, I’m pleased to say that we in Los Angeles County have an extremely low auto fatality rate, below even that of Massachusetts. Undoubtedly this is because we are a more sophisticated and urbane bunch than those of you in the rest of the country. Or perhaps it’s because traffic here never moves fast enough for our crashes to be fatal.


I suspect the low fatality rate in Los Angeles County is not due to the fact that residents are safer drivers but instead a result of the slow speeds due to rampant gridlock throughout the county.


Maybe they just have really low blood sugars? All these guys need to eat more cookies, the world would be a much safer place.


What about a correlation to how rural or urban the population of a state is?


Traffic in Massachusetts, like that in LA County, tends to be at relatively low speeds because of congestion. The speed factor may be far more important than the other issues mentioned--and would account for the low murder rate and high traffic mortality rate in Montana.


Yeah, I live in Montana. I'm pretty confident two of the big causes of fatalities we have are drunk driving (we are one of the worst in the nation) and horrible road conditions. In the winter our crazy mountain roads get coated with ice and snow and make for great fun.

So again, driverless cars FTW!


I drive through Boston everyday to/from work, and I seriously doubt at any given time I was going fast enough to kill anyone.

John B. Chilton

Like first commenter my guess is "the hidden force at work" is the average highway speeds. Low average tends to mean congestion due to high volume -- e.g. your own example of L.A.

As to Montana, there's little congestion and there are long distances to travel to get anywhere. If I'm not mistaken folks drive well over the posted speed limit.


Your last sentence reads like a throwaway, but I wonder if there's more to it.

Has anyone looked at the proportion of highway miles driven to local road miles driven?

Based on other things you've written on the subject, I'd be stunned if that weren't a factor. And it feels intuitive that people do more highway driving in Montana than they do in Massachusetts.


ScottF is on the right track. Fatal and casualty collision rates are highest on rural roads. I would imagine the states with the highest fatality rates per mile driven are also those with the highest proportion of miles driven on rural roads.

Jeff W

Interesting to note that being elderly is a risk factor, even though you'd expect elderly drivers to exhibit lower levels of aggression. The implication is that accidents might cluster into types. For instance, the elderly are far more prone to runaway-accelerator accidents (as I address in my blog, I've heard that they also, like 16-year-olds, are prone to drive off of straight, level roads, a type of accident that other demographics are good at avoiding.


I would think that some of the male lethality could be attributed to the size of vehicles driven by male drivers. You are more likely to find a man behind the wheel of a semi truck, larger delivery vehicle, or just a plain old truck. When one of these vehicles his another there is more likely to be a death then if a car hits another car.

This is not to say that males are not prone to thrill seeking, just that the number may be inflated by the cultural bias in the drivers of heavy vehicles.



Don't for get about deer and other wild life. Deer kill more people with their vehicle assisted suicided then any other animals in north america cause the death of humans by any other means. A 180lb weight placed 3.5 feet above the road surface is pretty lethal when hit by a car traveling at 65 miles an hour.

ray bans on my face

Figure in individual I.Q. level and reaction time into the analysis and you'll have a more complete answer. I guarantee both I.Q. level and reaction time have a profound effect of the outcomes that you're finding, because they're the ultimate psychological component behind all of your statistically significant variables.

Joel Upchurch

Unless Dr. Sivak controls for the urban/rural factor, I doubt that his study has much validity. 56% of traffic fatalities happen in rural areas, while only 23% of the population live in rural areas ( Montana only had 18 urban traffic fatalities compared to 211 rural ones in 2008 ( There also should be controls for how much rural driving is done on Interstate highways, since Interstates are safer than other roads (

It also occurs to me that traffic fatalities and homicide rates will both correlate to the availability of Emergency Medical Services (EMS). Much of the decrease in homicide rates in recent decades can be attributed to improved trauma care rather than decreases in violence.


Have the researchers looked at the types of cars involved in traffic fatalities? If SUVs are involved, is the risk of fatality greater?


LA also has a sizable transit-dependent population, whereas in Montana I assume most people drive most places. Buses don't usually get into fatal accidents, and although our Blue Line is known for pedestrian fatalities, it's still a much lower rate than traffic fatalities. It would be interesting if the researchers controlled for the portion of the population in question who are arguably not at significant risk for causing a fatal traffic accident because they're simply not behind the wheel. This would best be examined while holding urban/rural constant - IE examine the traffic fatalities of several large, urban cities with different mode shares.