From Freakonomics:Killer Cars:

An Extra 1,000 Pounds Increases Crash Fatalities by 47%

The Actual report name is “Pounds that kill: The external cost of vehicle weight.

Well, anyway, the study shows the obvious: If you have a big car that collides with a smaller car, the small car folks are toast.

Well, duh. Yes, all of that’s obvious to anyone who knows the laws of physics. There’s even a famous satire about this problem.

The writers of the paper have an agenda, of course, which is to raise gas taxes so people will  chose a tiny car like the Toyota IQ instead of a minivan or SUV…

Wait a second. This assumes that folks will agree to buy a vehicle that doesn’t have the safety factor of the larger cars. Women often consider safety features as part of their decision on what car to drive, and there are ways to make cars smaller and lighter, and I support government investment into this type of research.

But higher gas prices and higher EPA averages by government mandate? They didn’t work before.  Years ago, when government regulations limiting the gas consumption of cars resulted in smaller cars, the result was that more families  bought SUV’s and minivans. Why? the reason is practical: You simply can’t fit three kids (including two six foot teenagers), the family dog and six bags of groceries into a microcar.

Some of us used to have two cars for that reason: A big safe gas guzzler for family trips, for the teenagers to drive, and for bad weather, and the smaller fuel efficient minicar for short trips by mom or dad to work. But not everyone can afford two cars.

And there is another factor in larger cars: comfort. I used to get a backache driving my small car for more than 30 minutes, and we old folks have problems getting in and out of them…and the larger teenagers had trouble fitting into it the smaller car without their knees hitting their chins.

Reality check: Despite more cars, more miles driven, the fatality rate is flat. which means the fatalities per mile driven has gone down.The reason? Better highways, seat belts, and other safety features in newer cars.

There are ways to make smaller and lighter cars more safe, and if this is done, we will buy them. Side airbags is one example of this. Designing small urban cars with safety features for low speed collisions is another.

Making it the fashion to drive the smaller urban cars in urban areas will help eliminate those rich yuppies driving SUV’s to work, but for those in suburban or rural areas, especially those with kids, micro cars are not the answer.

A second thing that annoyed me in this “scientific” paper is that it might indeed show all the things they claim, but it is unreadable.

Some of the claims are unbelieveable:

Using unique vehicle identifiers (VINs), we determine the curb weight of each vehicle involved in an accident

Uh, you did this for all of them? The article cites a couple hundred thousand accidents in the various tables. Did you actually pull a couple hundred thousand VIN numbers and look them up?

Then there is the jargon:

…. thereby minimizing concerns about attenuation bias induced by measurement error.

The rich set of vehicle, person, and accident observables in the data set allow us to minimize concerns about omitted variables bias. Using these data, we estimate the external effects of vehicle weight on fatalities and serious injuries conditional on a collision occurring…

M.E.G.O…Too much statistical jargon, not enough hard data so I can come to my own conclusions. Usually at this point in the medical literature, I check out the tables. They should have things like average weight and the risk of death. Instead we have numbers that are presumably intelligible to the average statistics PhD but not to the rest of us who only took Statistics 101 in medical school or college.

Well, if I read the tables correctly,  I do actually have concerns on  omitted variables.

One table on accidents involving small trucks does note 10 percent had alcohol involved, but this data is not in the car data in table 2. On that table, you will find that only 5 of 9 of the studied included in their paper have information on “driver characteristics”, only 1 of the 9 studies noted if seat belts were used and only 7 of the 9 papers noted if weather might have contributed to the accident.

And then the authors mention at one point that “each vehicle” is included in the data set twice, “once as the struck vehicle and once as the striking vehicle”. Huh?

But my main problem is that the authors start their paper admitting they did the study to prove their point, not to find the truth in the matter.

At least in medical journals you are supposed to “pretend” you are looking for the truth.  You carefully collect and analyand after showing the data clearly (including tables that show hard data, not statistics) you then say: Voila! I was right.

Science is not about proving an agenda, it is about finding the truth.Papers that cherry pick statistics to prove an agenda are called…”editorials”.


Nancy Reyes is a retired physician living in the rural Philippines.  She blogs at Finest Kind Clinic and Fishmarket.

here, rich folks use SUV’s not only to show off, but because of the many unpaved roads…and another practical reason: It’s harder for the criminals to hijack you if you keep the windows up and the doors locked.

The common folk ride public transportation or use a motorcycle with a side car.

Be Sociable, Share!