I pulled 1,243,663 truck crashes from MCMIS and sorted them by the hour they occurred. Not the count. The kill rate. Given that a truck crashed, how often did somebody die?
Sadly, one of the things that has been completely overlooked is what were the statistics back when you could split your log. Granted that I think we had more experienced drivers back then as a general rule and drivers back then weren’t afraid to tell a dispatcher to get stuffed when they got tired. A few things changed such as Walmart thinking it’s okay to levy a 500 dollar fine for the load being late even as their distribution center is in Atlanta and subject to the interstate being shut down. The fine is one thing but then refusing to unload the truck for a week is a whole other matter. Apparently might makes right and that is why my wife has been banned from shopping at Walmart for over ten years. Most trucks today are regulated at 62 to 65 mph so maybe speed is not the factor that it once was but perhaps there should be a box on any citation indicating if there’s a governor on the truck. Also, maybe there should be a box indicating whether the other vehicle was cited. Though it might not be totally relevant it might help begin collecting data that would be more helpful. Another factor in night accidents might be visual illusions; truck driver are subject to many of the same visual illusions that pilots have to deal with.
Please explain the acronyms shown in your post - MCMIS, FARS, etc. I’m not a transportation geek but love reading well reasoned essays on a variety of topics.
I didn’t see a notation for solo truck crash vs car/truck crash. Any effect on your numbers? One would expect a multi vehicle crash would leave someone in condition to call it in.
Multi-vehicle goes 2.76 to 6.21 deaths per 100 crashes between 10am and 2am. That's 2.25x. Single-vehicle only goes 2.23 to 3.01. That's 1.35x.
Backwards. If nobody-coming were driving it, the crashes with no witness should be the worst at night. They're the mildest. The night penalty is almost entirely a multi-vehicle effect which points at who the truck is hitting at 2am, not how long it takes to find the wreck. Truck drivers aren't cited more at night. Single-vehicle truck crashes barely get deadlier. Truck-hits-car gets twice as bad. That's the other driver: impaired, wrong-way, and in a car.
Doesn't change the main finding but it does sharpen it. The tired trucker still isn't the story. Running the drunk-driver angle next. FARS has the other driver's BAC, MCMIS doesn't. That's the followup Monday.
That intuitively makes sense. Not surprising the numbers back it up. I’m guessing your source document doesn’t indicate citation to the car driver?
Is there any insight on the foreign driver question? Language, training, etc. ?
I was a career railroader and we would bang on trucks whenever we could. Beat the drum you have…. Lots more to the question and record keeping is sometimes meant to obscure rather than illuminate.
564 of 27,800 fatal truck crashes 2.03%. Mexico 283, Canada 247, other 34. In 72% of fatal truck crashes involving a foreign-licensed driver, the federal government doesnt even know whether that driver was licensed to drive the truck. For American drivers it's 1.1%. That's not a ridiculous gap and it's a border verification fail. Foreign-licensed drivers show up in 2% of fatal truck crashes and are half as likely to be alcohol-involved. The data doesnt say foreign drivers are dangerous but it says nobody checks their credentials. I have a case right now where the driver who killed a father and husband was a Russian licensed CDL holder here illegally.
Part two should be Monday. Part three which touches parking maybe Tuesday.
When traveling I usually drive late at night. Most truck drivers, from what I’ve seen, are doing a good job. The only problem that I’ve noticed is that, unlike during day time city driving, sometimes at night the big rigs don’t slow down when they’re approaching a car driving more slowly ahead of them in their lane. It feels like they’re expecting the car to change lanes.
Our numbers in Australia show in 87% of fatal crashes involving a truck and a third party (car, motorcycle etc) the third party is at fault. Of those, around 30% are deliberate 😢 Keep digging Rob, this data is amazing, and I’ve always wondered about the after-crash effect on fatalities.
Last fall, I was driving a U-Haul truck from Seattle to Tulsa. Part of my route involved I-80 from SLC to North Platte, NE, and it was a little after midnight when I shut down in NE for the night. All along that stretch of I-80, I saw lots of cattle trucks in both directions. The ones heading east were empty - stockyards are in western KS - and those guys were hauling ass. Even going down hills or in crosswinds, they looked like they were running 70 or 80. I stayed in the RH lane and let them pass. God only knows how many of those crash in the middle of the night in the middle of nowhere on that route.
Wondering what the non-truck fatalities show for the same time periods.
Thinking out loud, if the increased fatalities are similar then the cause isn’t necessarily unique to truckers. I’d be surprised if the fatalities are the same or lower.
I'm wondering how much this changes if you split the data as "local day", "local twilight" and "local night".
Is the truck driver less able to *see* the car at twilight/night?
I note this because the deaths seem to start to rise at 6pm, well before most people have a load on. Or so I presume. I should think that people drinking just before that (to be drunk and driving) would mostly continue drinking past that.
This is interesting but it is long on rhetorical declarative statements and very short of a cogent thesis defined by clear exposition. Pith is a tool best used sparingly. In this case it helps undermine whatever is the causation case you must be making. Nonetheless I appreciate the effort.
I think you’ve critiqued the style more than the substance.
Not every article is intended to read like an academic paper. This one is clearly written in a narrative style for a general audience while still laying out its reasoning, its data source, its assumptions and even its limitations.
If you think the causation argument is flawed, I’d be genuinely interested in hearing where you believe the analysis goes wrong. But saying it lacks a “cogent thesis” doesn’t really engage with the actual argument being made.
Sadly, one of the things that has been completely overlooked is what were the statistics back when you could split your log. Granted that I think we had more experienced drivers back then as a general rule and drivers back then weren’t afraid to tell a dispatcher to get stuffed when they got tired. A few things changed such as Walmart thinking it’s okay to levy a 500 dollar fine for the load being late even as their distribution center is in Atlanta and subject to the interstate being shut down. The fine is one thing but then refusing to unload the truck for a week is a whole other matter. Apparently might makes right and that is why my wife has been banned from shopping at Walmart for over ten years. Most trucks today are regulated at 62 to 65 mph so maybe speed is not the factor that it once was but perhaps there should be a box on any citation indicating if there’s a governor on the truck. Also, maybe there should be a box indicating whether the other vehicle was cited. Though it might not be totally relevant it might help begin collecting data that would be more helpful. Another factor in night accidents might be visual illusions; truck driver are subject to many of the same visual illusions that pilots have to deal with.
Please explain the acronyms shown in your post - MCMIS, FARS, etc. I’m not a transportation geek but love reading well reasoned essays on a variety of topics.
I didn’t see a notation for solo truck crash vs car/truck crash. Any effect on your numbers? One would expect a multi vehicle crash would leave someone in condition to call it in.
Multi-vehicle goes 2.76 to 6.21 deaths per 100 crashes between 10am and 2am. That's 2.25x. Single-vehicle only goes 2.23 to 3.01. That's 1.35x.
Backwards. If nobody-coming were driving it, the crashes with no witness should be the worst at night. They're the mildest. The night penalty is almost entirely a multi-vehicle effect which points at who the truck is hitting at 2am, not how long it takes to find the wreck. Truck drivers aren't cited more at night. Single-vehicle truck crashes barely get deadlier. Truck-hits-car gets twice as bad. That's the other driver: impaired, wrong-way, and in a car.
Doesn't change the main finding but it does sharpen it. The tired trucker still isn't the story. Running the drunk-driver angle next. FARS has the other driver's BAC, MCMIS doesn't. That's the followup Monday.
That intuitively makes sense. Not surprising the numbers back it up. I’m guessing your source document doesn’t indicate citation to the car driver?
Is there any insight on the foreign driver question? Language, training, etc. ?
I was a career railroader and we would bang on trucks whenever we could. Beat the drum you have…. Lots more to the question and record keeping is sometimes meant to obscure rather than illuminate.
564 of 27,800 fatal truck crashes 2.03%. Mexico 283, Canada 247, other 34. In 72% of fatal truck crashes involving a foreign-licensed driver, the federal government doesnt even know whether that driver was licensed to drive the truck. For American drivers it's 1.1%. That's not a ridiculous gap and it's a border verification fail. Foreign-licensed drivers show up in 2% of fatal truck crashes and are half as likely to be alcohol-involved. The data doesnt say foreign drivers are dangerous but it says nobody checks their credentials. I have a case right now where the driver who killed a father and husband was a Russian licensed CDL holder here illegally.
Part two should be Monday. Part three which touches parking maybe Tuesday.
When traveling I usually drive late at night. Most truck drivers, from what I’ve seen, are doing a good job. The only problem that I’ve noticed is that, unlike during day time city driving, sometimes at night the big rigs don’t slow down when they’re approaching a car driving more slowly ahead of them in their lane. It feels like they’re expecting the car to change lanes.
Our numbers in Australia show in 87% of fatal crashes involving a truck and a third party (car, motorcycle etc) the third party is at fault. Of those, around 30% are deliberate 😢 Keep digging Rob, this data is amazing, and I’ve always wondered about the after-crash effect on fatalities.
Anecdotal, I know, but I’m gonna post it anyway.
Last fall, I was driving a U-Haul truck from Seattle to Tulsa. Part of my route involved I-80 from SLC to North Platte, NE, and it was a little after midnight when I shut down in NE for the night. All along that stretch of I-80, I saw lots of cattle trucks in both directions. The ones heading east were empty - stockyards are in western KS - and those guys were hauling ass. Even going down hills or in crosswinds, they looked like they were running 70 or 80. I stayed in the RH lane and let them pass. God only knows how many of those crash in the middle of the night in the middle of nowhere on that route.
Wondering what the non-truck fatalities show for the same time periods.
Thinking out loud, if the increased fatalities are similar then the cause isn’t necessarily unique to truckers. I’d be surprised if the fatalities are the same or lower.
I'm wondering how much this changes if you split the data as "local day", "local twilight" and "local night".
Is the truck driver less able to *see* the car at twilight/night?
I note this because the deaths seem to start to rise at 6pm, well before most people have a load on. Or so I presume. I should think that people drinking just before that (to be drunk and driving) would mostly continue drinking past that.
This is interesting but it is long on rhetorical declarative statements and very short of a cogent thesis defined by clear exposition. Pith is a tool best used sparingly. In this case it helps undermine whatever is the causation case you must be making. Nonetheless I appreciate the effort.
I think you’ve critiqued the style more than the substance.
Not every article is intended to read like an academic paper. This one is clearly written in a narrative style for a general audience while still laying out its reasoning, its data source, its assumptions and even its limitations.
If you think the causation argument is flawed, I’d be genuinely interested in hearing where you believe the analysis goes wrong. But saying it lacks a “cogent thesis” doesn’t really engage with the actual argument being made.