Celling Out

I wanted to comment on Mitch’s Over Celling post, specifically Jonathan’s last comment, but the comment mushroomed, so I’ll turn it into a post. There is a very interesting report from the NHTSA and a group of VA Tech academics that put cameras in cars, recording the behavior (and consequences of that behavior) of 241 drivers over the course of a year. The consequences were grouped into the categories crashes, near-crashes, and “incidents” They take quite a bit of space to define “incident”, so if you’re interested, go take a gander at the report itself. If you look at this data, especially Figure 11-9 on page 293, I think the conclusion is pretty inescapable that cell phones are a major contributing factor to vehicle accidents.

However, I come away from this report agreeing with Mitch that banning cell phone use while driving is not really helpful. The reason is that this data is very poorly presented, giving me the feeling that the NHTSA is skewing the presentation because the political bosses wish to further the cause of banning cell phones. Upon reading the report, I come away with a striking impression. If we just look at the chart on the top of page 29, depicting the number of incidents, the bottom two cells in the table provide a staggering total. Because the data is presented poorly, I’m going to have to make some assumptions, and I’ll be conservative. The bottom two boxes are labeled 151-200 incidents (over the year) and 200+ incidents. Let’s say that the 29 drivers in the 151-200 incident box all fell in the 151 incident category. That makes for 4379 incidents. If the 9 drivers in the 200+ incident category all fell in at 201 incidents, that would make their total 1809. I’m sure the averages were higher than that, but the minimum total number of incidents from this batch of 38 morons alone was 6188, out of 8295 for all 241 drivers. That is a whopping 74.6% of the incidents caused by 15.8% of the drivers. (pretty close to the 80/20 rule) That’s at the minimum – I have a feeling that the bozo brigade we’re talking about here accounts for much more than three quarters of the incidents, because the mean number of incidents in each box is likely to be a lot closer to the median than I’ve assumed here.

It is a huge deficiency of this report that those segments are not identified, quantified and addressed in detail. In addition, the cross-correlation between the number of incidents and number of near-crashes and crashes is not addressed. In other words, do some drivers have a lot of minor incidents that just make the cut, but don’t really do anything totally insane, while others do insane stuff once in a while – often enough to rack up multiple crashes and near-crashes, but not often enough provide huge incident totals? (An example might be people who drive OK on their own time, but are chronically late for work, making them insane on the morning commute.) Those statistics need to be published. But by so doing, the NHTSA would put the blame on people, specifically about 15% of drivers who are incurable idiots, and shove technological factors down on the list of things to deal with to improve traffic safety. It’s easy to ban cell phones and then measure ticket rates, it’s a lot harder in our legal climate to find the morons, tell them unequivocally that they are morons, and then revoke their driving privileges. The bureaucrats in the NHTSA are happy to make more rules that they can be measured (favorably) against, but less happy to tell the constituents of their political bosses that they don’t have enough brains in their head to successfully pilot an automobile on the roadways with the rest of us.

The authors do make a big deal out of 4 drivers who happened to be in the worst 95th percentile in two out of the three measures. They summarize their statistics on page 30. That’s great, as far as it goes, those 4 people account for 790 incidents, or 9.5% of the total. Taking those four people off of the road would be a service to humanity. But the report ignores the other bozos. Turning an “incident” into a crash or near-crash is in some sense a matter of luck – the physical conditions of the road, and the alertness and skill of the other driver involved. With those four people you can see a direct relationship between incidents and near crashes – the more incidents, the more near crashes, although the relationship is not linear because of the small numbers we are dealing with, here. The relationship is not as clear with crashes because of the even smaller numbers. However, my point stands – the more risky situations caused by a driver’s behavior, the higher the probability of a crash.

Those who are in the 90th percentile or higher for incidents are all menaces in my book. You’ve all seen them. Recently I watched a car roll through a stop sign without stopping, cutting me off in the process (my road did not have a stop sign). This driver proceeded to the next intersection, turned left without braking or signaling, and in so doing cut off a car coming in the other direction. Four moving violations and two “incidents” in less than a quarter mile and less than 5 minutes. That right there is one of your “outliers”, and a cellphone was not even involved, as far as I could see.

Banning cell phone use in cars looks a lot like gun control to me. Ban legitimate possession and incidents will go down somewhat, but only somewhat, because scofflaws will just ignore the law, and people will find other weapons with which to kill one another. Car accident rates did not shoot up after the widespread use of cellphones – the phones are just the current distraction of choice for these morons – they were bad drivers before cells became widespread, and they’ll continue to be bad drivers if cells are banned. Even if we banned just this segment from using phones, the next three causes of driver inattention in the survey accounted for more incidents than cell phones, although cell phones were far and away the leading cause of inattention that led to incidents.

Let’s leave aside the fact that these horrible drivers will probably use cell phones illegally. They have poor driving habits from the git-go, and they’ll run into someone while distracted by eating a cheeseburger if their cells are banned. Finding these drivers and revoking their licenses would do more for traffic safety than a ban on telephones, eating, fiddling with the radio, and carrying children in the car (the top distractions in the NHSTA report) combined. I like my legal system to take the approach of good parents everywhere: “make as few rules as possible, and don’t take no for an answer”. When I was in college, there was an engineering frat that had been busted for drugs back in the 70s. As part of the deal to remain certified by my school, some 15 years later that frat still had a cap on the number of yearly pledge offers so that the total house size stayed below a certain number of members. That’s what I’m talking about. Find these people, revoke their licenses, and don’t give them back. Don’t take no for an answer, don’t take “but I’ve got to get to work” for an excuse. That would make the time I’m forced to spend on I-95, I-81, and I-76 a lot more bearable.

The problem is finding these people. That issue is really highlighted by the four outliers singled out in the VA Tech report. Some people get lucky and leave behind a trail of incidents and near-crashes without having a crash recorded that would alert the authorities to the incompetence. Take a look at the table on page 30. Driver number 204 had equipment problems that resulted in a low number of miles (2603) recorded for the year, yet still managed to rack up 60 incidents and one crash in those few miles, but driver 308 truly seemed to have been lucky, getting only one crash out of 171 incidents and 19 near-crashes. No one can be lucky forever, though, and I bet we’d see a crash or two the next year for driver 308, whereas a good driver who got unlucky in the study year would show none the next year.

Given my libertarian leanings, I’d like to see a private solution to the problem. If insurance companies were to look at people with high crash rates over time (say, one per year for 3 years, or more than 2 in any given year), and force those people to install this device for one year, the insurer could monitor for behavior like that demonstrated by our four friends on page 30. I’d bet the VA Tech people would be happy to subcontract this work to pay for their research. The behavior over the monitoring period could then be reflected in the driver’s insurance rates, with rates for the four members of the pinhead platoon up there being somewhere in the range of $20,000 per year.

We need to get a handle on this before the boomers get much older, because this problem will get worse with time. The number of elderly drivers is going to mushroom over the next decade, and the problems caused by people sporting Florida plates in the Mid-Atlantic and New England states are pretty obvious every time I hit the interstate. Skills only deteriorate with age. It’s interesting, and quite frightening, to note that none of the four worst drivers in the NHSTA report was over 44. God help those of us on the road when they turn 65.*

* I realize that some of the problem with younger males is often reckless driving that tends to calm down with age, but the report shows that inattention is a much greater threat than recklessness, and that only gets worse with age.