Featured Post

Part 1: Mark Baumer Reflection: Impounding Vehicles & Immigrant Rights

This is part of a multi-part reflection I've been doing following the death of my friend, Mark Baumer . There's nothing graphic i...

NSC and NHTSA Use Different Metrics for Vehicle Deaths

I've changed the headline for this article from "NY Times Under-reports Traffic Death Increase?" to "NSC and NHTSA Use Different Metrics" to better reflect what I've found about this statistic. See updates below for full accounting of data.

It's totally possible that I'm misunderstanding something-- in fact, let's go ahead and say that it's likely that I'm misunderstanding something -- but as far as I can tell the New York Times and Streetsblog both under-reported the percentage increase in U.S. traffic deaths that occurred between 2014 and 2016. 

The claims are from a report from the National Safety Council. Quoting from the Grey Lady:
"[M]ore Americans are dying on roads and highways than in years, and the sudden and sharp increase has alarmed safety advocates. 
The latest batch of bad news arrived Wednesday in traffic fatality estimates released by the National Safety Council, a nonprofit organization that works closely with federal auto-safety regulators. According to its estimates, 40,200 people died in accidents involving motor vehicles in 2016, a 6 percent rise from the year before. 
If the estimates are confirmed, it will be the first time since 2007 that more than 40,000 people have died in motor vehicle accidents in a single year. The 2016 total comes after a 7 percent rise in 2015 and means the two-year increase — 14 percent — is the largest in more than a half a century." (my emphases)
The 14% increase figure only makes sense if you calculate off of 2015 figures. 35,092 people died in U.S. traffic crashes in 2015. If you divide the 2016 estimate (40,200) by the 2015 figure (35,092) you get a figure close to 14% (although normal rounding practices would state that as 15% since it's 14.55%).

But we're not measuring a figure for 2017, but 2016. So a two-year stretch has to go back to 2014 numbers. According the NHTSA, there were 32,675 deaths in 2014. That would mean that we've had a 23% increase in traffic fatalities.

It's totally possible I'm missing something, and that the New York Times is right. But I think someone just made the totally understandable mistake of calculating off of the current year (hey, I've done worse).

Update: (See below) In fact, speaking of how easily such mistakes can be made, I did a similar thing in an earlier version of this article. I calculated the difference between 2016, on one hand, and 2015 and 2014 respectively, when I should have tabulated the data using only the year-over-year increase. In other words, I should have added (40,200-35,092)=5,108 to (35,092-32,675)=2,417 not (40,200-32,675)=7,525, because the 2015 numbers didn't rise all the way to 40,200, but just to 35,092. Combining the two figures gets us to an additional 7,525 deaths, not an additional 12,633. That's 2.5 September 11ths, not 4.2. Still staggering.
The additional deaths from 2015 and 2016 over 2014 figures amount to 12,633 7,525 people. That's 4.2 2.5 September 11ths. Of course, we don't experience a daily trickle of deaths the same way we do a sudden burst of them, but these deaths are not "accidents" in that policies we've implemented have led to their greater likelihood. 
Further Update: Maureen Vogel of National Safety Council got back to me to say that the difference between the percents I found and theirs may be the result of using different standards of measurement. See below:
Hi James,
Thank you for your interest in our motor vehicle fatality estimates. I believe the 23% increase you calculated is a result of you comparing the 2014 National Highway Traffic Safety Administration’s (NHTSA) fatality estimate (32,744) with the National Safety Council’s 2016 preliminary estimate (40,200). NHTSA’s estimates are not fully comparable to the National Safety Council’s estimates. The National Safety Council’s 2014 fatality estimate is 35,398. The percent change from 35,398 to 40,200 is approximately 14%. Attached is a page from our publication, Injury Facts, that describes the differences between the NHTSA and National Safety Council’s motor vehicle fatality estimates.

Thanks again for reaching out, and we hope this helps clarify. Have a safe day.
The NSC sent a PDF with an explanation of how the statistical analyses differ between the two organizations:

The National Safety Council (NSC) and the National Highway Traffic Safety Administration (NHTSA) count motor-vehicle crash deaths using somewhat different criteria. NSC counts total motor-vehicle-related fatalities – both traffic and nontraffic – that occur within one year of the crash. This is consistent with the data compiled from death certificates by the National Center for Health Statistics (NCHS). NSC uses NCHS death certificate data less intentional fatalities as the final count of unintentional deaths from all causes. NHTSA counts only traffic fatalities that occur within 30 days of the crash in its Fatality Analysis Reporting System (FARS). This means that the FARS count omits about 800 to 1,000 motor-vehicle-related deaths each year that occur more than 30 days after the crash. Nontraffic fatalities (those that do not occur on public highways; e.g., parking lots, private roads, and driveways), which account for 900 to 1,900 deaths annually, also are omitted. By using a 30-day cutoff, NHTSA can issue a “final” count about eight months after the reference year. 
This edition of Injury Facts includes the 2013 and 2014 NCHS final counts by cause of death including motor-vehicle crashes.The graph below shows the NCHS death certificate counts of unintentional motor-vehicle deaths through 2014 compared to the NHTSA FARS counts of traffic deaths. 
It's an interesting coincidence that the difference between the NHTSA 2015 numbers and 2016 NSC estimate are 14%, as well as that being the difference between the 2014 NSC and 2016 NSC number. I guess you learn something new everyday.

By the way, if we take the 40,200 estimate and the 35,398 estimates that are apples-to-apples comparisons for NSC-to-NSC death statistics, we get a final result of 1.6 September 11ths of deaths resulting just from this increase in traffic fatalities from 2014-2016 (Take 40,200-35,398= 4,802 and then divide that by 2,996 to get 1.602. The September 11th number actually includes the hijackers, so you'd get a slightly higher number if you wanted to do a calculation of "innocent victims" caused by this change)

1 comment:

  1. Dumb question: why is the body count going up at all?

    I thought that we lived in an era of technological explosion where the body count should of course go down every single year.

    Am I correct in saying that non-drivers, or can we call them national sacrifice victims, are 1/4 of the body count?

    Yours anonymously,
    Paul Klinkman