Another example of New York journalism gone awry!

The January 12th issue of the New Yorker published an anti-SUV article (Requires Adobe Acrobat reader):

Download gladwell_suv_040112.pdf

I'm not particularly pro or anti-SUV - I certainly think that if I were to argue against them, there are much better arguments than those made by Mr. Gladwell.

Given that there were glaring errors in the article, I did my civic duty and wrote a flaming letter to the editor and subsequently wrote the author as well. They never responded or even corrected their unbelievably simple math mistakes made in public.

Here's my letter to the editor:

The blaring “Road Killers – Malcolm Gladwell on the S.U.V. culture” in the January 12th issue is certainly attention getting – it’s too bad that the New Yorker apparently performed no modicum of review on this article else you, dear editor, would fain have published such drek.

Mr. Gladwell, despite your insistence, there is nothing here. You’d have us understand that the only reasons people buy SUVs is so they can be “high above the ground … look down on other drivers … see if someone was lurking behind or beneath [the vehicle] … [and] drive it up on someone’s lawn with impunity” – all in some self-induced haze of misbegotten safety. What happened? Did someone in an SUV run over your poodle? If these are really the reasons, why isn’t everyone driving 18-wheelers? They certainly make an even better match for your humors. Of course, no one who has an SUV acquired it because it can drive places that other vehicles cannot, or that it can carry a few more passengers than a Boxster, or that it fits with their recreational desires, or that they have a towing application, or that they must regularly confront hazardous weather conditions. No, no, these people of the “SUV culture” move along in some hulking drool, murmuring “high above, look down, see lurking, drive up …. ” in some Feldman-esque fashion – it must be where they got the idea for that Buick commercial.

In your prolonged search for small differences that mean big things, you must have decided that SUVs are bad. After all, you present a tablature – pseudo-science proof - which shows that people die within all kinds of vehicles. Also people think that SUVs are safe and because they think that they must be unsafe –they aren’t as maneuverable as a Boxster so they must not be safe. Please!

The fact is that your table demonstrates no causality of any kind in relation to what kind of vehicle is involved in fatality accidents. The only thing that it does show is that the difference between the vehicles, statistically, is so small that it hardly makes any difference. If we took the ‘best’ versus the ‘worst’ in the table there is a difference between them of almost two parts in ten-thousand – which in this context is nearly indistinguishable.

To be a little less callous about the loss of human life let’s compare the statistical difference in the ‘safety statistics’ (though a subcompact is apparently the next worse offender as the bad old pickup), with… let’s see, I know: the game of golf. Let’s take the statistics for the PGA tour – as good a statistical examination as any. For 2003 the best golfer in scoring stats was – surprise – Tiger Woods. The 33rd best golfer (or let’s say ‘worst’ so we are matching up with the 33 vehicles in the proffered table) was K. J. Choi. For the entire year, it took Tiger, on average, a little less than two strokes per round to get around the course than it did Mr. Choi (68.41 to 70.35). Pretty close, huh? Their performance in scoring is separated by about 2.8%. Of course Tiger took home 334% more dollars than did Mr. Choi in tournament winnings and probably an inconceivable amount more in sponsorship dollars. Little differences can mean a lot – very true. It wouldn’t be statistically commensurate of course to pit these gentleman’s totals of about 12,000 strokes for the year against the almost 3 trillion miles that Americans drove last year – nor does a comparison of the 33 best golfers in the world match with every grandmother, 16 year old, and taxi driver that drives a vehicle, but just for the benefit of the doubt let’s compare the statistical differences. Tiger and K. J. Choi -pretty close right? The difference between them in their scoring skill is 155 times larger than the largest difference in Mr. Gladwell’s table. That’s two orders of magnitude. If Tiger and Mr. Choi were separated in their scoring skill by the amount that the vehicles are in the table their score would be indistinguishable from each other. To argue that the statistical difference between the vehicles in the table is meaningful is simply vacuous.

Mr. Gladwell not only argues it, he opines about the “extraordinary performance of some subcompacts” evidenced by his table. Let’s see, that would be roughly equivalent to a golf scoring average of 68.41 compared to one of 68.4107. What an extraordinary difference!

Unwittingly, Mr. Gladwell delivers a round riposte to his own argument (even with employing poor math in his favor) in his diatribe against the attention that everyone gave Bridgestone Firestone. At the end of his treatise, he states: “That sounds like a lot [the number of fatalities associated with Firestone tire failure], until you remember that the total number of tires supplied by Firestone to the Explorer … was fourteen million, and that the average life span of a tire is forty-five thousand miles. The allegation … amounts to the claim that its tires failed with fatal results two hundred and seventy one times in the course of six hundred and thirty billion vehicle miles”. Hello! Mr. Gladwell, that’s not vehicle miles – that’s TIRE miles you are referring to. That’s not “six hundred and thirty billion vehicle miles” it’s “one hundred and fifty seven billion, five hundred million vehicle miles.”

So what? According to the NHTSA Traffic Safety Facts 2002 (http://www-nrd.nhtsa.dot.gov/pdf/nrd-30/NCSA/TSFAnn/TSF2002EE.pdf) report we had 42,815 traffic deaths in 2002 with a total of 2,829,645,000,000 (almost 3 trillion) vehicle miles driven. That means at current levels (which are reduced from the averages of times past) we’d expect to see 2363 fatalities in 157,500,000,000 miles (or 9450 in 630,000,000,000 miles). Mr. Gladwell takes everyone to task for having a focus on identifying 271 deaths out of an expected 2363 as being directly attributable to manufacturing defect. Let’s see – statistically that’s about 11.5% of expected deaths for the mileage involved. That’s just shame, shame on all of our institutions for paying so much attention. After all, that’s only 646 times the statistical variation in Mr. Gladwell’s table! Even if we were to accept his comedy of horrors (14 million vehicles, each riding along on one tire for 45,000 miles!) the statistical difference between what he argues as unwarranted attention and the variation in his table exceeds even the golf example above.

Mr. Gladwell, you really must offer Gilda Radner’s famous refrain. An honorable man would offer a public retraction – or at least a promise not to practice pseudo-science or math in public places for five to ten years.

An examination of the NHTSA report does provide a number of insights relative to Mr. Gladwell’s arguments. Out of fatality accidents in 2002, over 13% involved non-motorists. Apparently something needs to be done about the maneuverability of pedestrians – they don’t seem to be able to avoid all those motor vehicles very well. The statistics also belie Mr. Gladwell’s arguments about vehicle maneuverability – in the report, fully 70% of fatality accidents are classified as ‘moving in a straight line’ – and this in a list that includes turning left, and right, veering one way or another, attempting to avoid other vehicles or obstacles, etc. Apparently, SUVs (classified within the Light Trucks category), in terms of fatality data, are a little safer to drive than passenger cars (of course 18-wheelers are much, much safer – at least 2/3 safer. That couldn’t have anything to do with driving skill could it? Watch for grass roots legislation sponsored by Mr. Gladwell against citizens buying truck cabs for personal use.)

Additionally, the state in which you live had much more statistical impact (for all states between 2001 and 2002 - except Texas) on traffic fatalities than what kind of vehicle you drive – and far more variation than Mr. Gladwell’s table... error standards for the collection and reporting of data for this report far exceeds (by at least two orders of magnitude) the variation that Mr. Gladwell wants us to believe is so important in his table… In 2002, almost 3,000,000 people were injured in some kind of motor vehicle event. That’s about 70 times the number of fatalities. Perhaps that’s worth some study as well.

Of course there are really glaring statistics in the NHTSA report. After a brief reading, one could easily produce a headline like: “Road Killers: 41% of fatal vehicle crashes in 2002 were caused by drunk drivers. Of those deaths, drunks driving between 9pm and 6am, primarily male, caused more than 2/3 of them.” Now, there’s a Tipping Point for you Mr. Gladwell. Perhaps it wouldn’t rank a frontispiece in the New Yorker – but at least it would be true.

UPDATE: What is it, a year and counting? No correction or retraction from the New Yorker or Gladwell so far. Michelle Malkin decries an anti-SUV bias headline today, January 31, 2005.