Digging Into Traffic Safety Data

October 23, 2015

For every 1 minute of travel time saved by speeding, roughly 2.5 minutes of human life are lost in speeding-related accidents

This stat is from a prior post, the cost-benefit trade-off of speeding. Here is an explanation of how I reached those figures, along with some more thoughts / facts about the science of speeding.

These numbers are based on two separate calculations.

  1. How many years of human life have been lost in speeding-related accidents?
  2. What is the total amount of time Americans have saved by speeding?

1. Estimating the loss of life in years

To estimate the amount of life lost due to speeding-related accidents, I looked to the Dept. of Transportation’s FARS (Fatality Analysis Reporting System) database, which records every fatal traffic accident going back to the 70’s, including detailed information about every vehicle and each individual involved.

Looking at the last 10 years of available data (2004-2013), I identified all the individuals who died in accidents where speeding was flagged as a contributing factor. There were about 115,000 of them, roughly one-third of all traffic fatalities. For comparison, alcohol was a contributing factor in about 124,000 traffic fatalities.

speeding deaths by year

116k individuals have lost their lives in speeding related accidents, 2004-2013

To estimate the years of life lost by each of these 116,000 people, I used life expectancies published by the SOA.

The FARS dataset includes gender and the exact age of each individual at the time of death, so each one can be matched up with a fairly precise life expectancy. Summing the life expectancies of all 116,000 individuals over the 10 year period, the total comes out to about 5.2 million.

From 2004 to 2013, 5.2 million years of life were lost due to speeding-related accidents

 

2. Estimating the amount of time saved by speeding

To estimate the other side of the equation, the total amount of time saved by speeding, I started with the Dept. of Transportation’s HPMS (Highway Performance Monitoring System) data, which includes an estimate the number of annual vehicle miles traveled in the U.S., broken down by type of road (e.g. freeway, state highway, local road, etc). For each type of road, I assumed an average speed limit, which is somewhat arbitrary. But in the end it doesn’t make much difference for the final calculation.

Based on total number of miles driven and the assumed speed limit for each of those miles, I calculated the total driving time if everyone drove exactly at the speed limit. That number comes out to 84,120,493 years.

To see how that number changes when people speed, I used the data from this report by the Insurance Institute for Highway Safety. Here are the figures from their report in terms of variation from the average speed.

speeding prevalence by speed limit

In rough terms:

  • 10% of vehicles drive at least 5 mph faster than the average traffic speed.
  • 5% of vehicles drive at least 10 mph faster than the average.
  • 1% of vehicles drive at least 20 mph faster than the average.

These figures are consistent with other reports that have estimated speeding behavior. Assuming these rates of speeding, the total time that Americans spent driving drops to 82,545,028 years. (vs. 84,120,493 years if everyone drove at the speed limit). The difference is 1,575,465 years. Then to account for the presence of passengers, I multiplied this number by 1.38 (0.38 is the average number passengers per vehicle).

From 2004 to 2013, 2.17 million years of travel time was saved by speeding

5.2 million years in life lost vs 2.17 million years of travel time saved is a ratio of about 2.5 to 1, the stat quoted at the top.

Here is the original post about these figures: The Science of Speeding

If you feel like digging into numbers yourself, the data is posted on the get the data page.

get the data button

 

 

 

Primary data sources:

 

My latest project, Elementus, aims to bring transparency to the cryptocurrency market. Check out our blog for some crypto-related data visualizations.

 

Follow Metrocosm

Max Galka

I'm an NYC-based entrepreneur (my newest project: Blueshift) and adjunct instructor at UPenn. I'm fascinated by data visualization and the ways that data is transforming our understanding of the world. I spend a lot of time with my face buried in Excel, and when I find something interesting I write about it here and also as a Guardian Cities and Huffington Post contributor.
More about my background

Latest posts by Max Galka (see all)

Related Posts

  • Pingback: Speeding – For Every Minute Saved, 2.5 Minutes of Human Life Are Lost in Accidents | Metrocosm()

  • PA

    The biggest issue with this post (and the preceding one) is that you use the term “speeding” to mean “deviation from the average vehicle speed”, while the legal and, I’d argue, common definition is “deviation from posted speed limit.” This is particularly confusing in section 2 of this post where you talk about a speed limit (but actually mean an average speed).

    This is problematic because it changes the behavioral implications of the article. If people read your articles with the common definition and start sticking rigorously to the posted speed limit, they will push themselves down the Soloman Curve and actually be in greater danger than if they traveled the modal speed. They would avoid this conclusion if they carefully read the text below the Soloman curve, but all your headings would lead them astray.

    Beyond that, what speed do you think is actually safest (average vs. median vs. mode) and how would you recommend recognizing it in practice?

    • http://metrocosm.com Max Galka

      Thought I made the distinction pretty clear in the other post (e.g. the header “Driving fast is not necessarily dangerous. Driving faster than everyone else on the road is.”), but I see your point on this one, particularly the section you mention. Will see how I can clarify the language.

      My interpretation is that the risk is not from speed per se, but from passing other vehicles. So in practice, I think the safest would be to focus less on speed and more on driving along with the cars nearest to you.

      Though technically, my guess is that the mean speed is probably slightly safer than the median or the mode, based on the Solomon Curve. The mean should be a little faster than the median, and that’s my guess as to why the lowest point in the curve is above 0 (its x-axis is median speed).

      Any other theories?

      • PA

        I think that’s right. The speed that minimizes passing should be the mean speed, which would be consistent with your hypothesis that passing (or interaction with other cars) is actually what causes accidents. Here is a link to a review of studies involving speed dispersion, which typically find it is a good indicator of fatality rate: http://www.nap.edu/read/11387/chapter/11#307
        Situations with high speed dispersion would be predicted to have high passing/car interaction rates.

        As far as how to recognize it in practice (the key to using this information), I’m not sure exactly how to do that. If you pass cars as often as you are passed (keep a differential count), then you are traveling at the median speed. If you fall in line behind a big group and just follow along, you are traveling at the local modal speed.

        However, the modal strategy raises an interesting point. Traveling at the median or (if you could recognize it) mean speed necessitate more frequent merging strategies than modal driving. Merging strikes me as a high risk maneuver, which makes it difficult for me to believe that rigorously going the average (or median) speed is actually the best driving strategy. However, someone who rigorously goes the local modal speed would (on average) go approximately the mean speed while minimizing merging/interactions.

        Ergo, my recommendation would actually be to just follow the largest group of people (local mode speed), as it will minimize the number of times you or others attempt to merge in/out near you.

        • http://metrocosm.com Max Galka

          Thanks. Good resource. Looks like all the research agrees on that point.

          Think your argument is probably right that the mode speed would be the safest, at least in theory. In practice, I would guess that it comes down to what the two or three nearest cars are doing in that moment.

  • Pingback: Map: Every Fatal Traffic Accident in the U.S. 2004-2013 | Metrocosm()

  • Kidd Canseco

    I found the same issue as PA. You underestimate time saved by speeding because you compare it to the average speed traveled; often well above the speed limit. The time saves versus the speed limit is actually much greater.

    Also, the vast majority of “speed-related” fatalities are actually drunks, drunks that were not wearing their seat belts, or speeders so stupid they were not wearing their seat belt. The Dept. of Transportation has done a study where speed is only a true causal factor in less than 5% of all collisions. This coincides closely with other studies once you take out the BS the NHTSA lumps into “speeding” to pad and propagate the “speed kills” myth.

Follow

Get the latest posts delivered to your mailbox: