Motostats 2006

[DataDan]

Everyone talks about motorcycle safety stats, but few people know where to find the latest ones and even fewer have the time or interest to figure out exactly what they mean.

The National Highway Traffic Safety Administration (NHTSA), part of US DOT, collects a lot of data about fatal motor vehicle crashes and publishes it annually. Some motorcycle-specific information can be found in the Motorcycle Fact Sheet (750K PDF), more is scattered in the 200-page Traffic Safety Facts Annual Report (1.5MB PDF), and much more sits as raw data in the FARS database (Fatality Analysis Reporting System), of not much use to anyone until it’s compiled into meaningful form. Unfortunately, the motorcycle stats aren’t collected in a single reference, which would be a convenient resource for those interested in the current state of motorcycle safety but who don’t have the time to spend gathering it from multiple sources. Not to mention the value in settling forum arguments.

For several years I’ve collected motorcycle data from NHTSA and other sources and maintained it in tables and charts, mainly for my own interest but also posted on forums occasionally. The 2006 data is now available (though not absolutely final), and I have begun gathering up the bits I’m interested in. Time permitting, I’m going to post some of the data I get and my interpretations in this thread. I’m sure there will a lot of other interpretations and opinions as well.

Here are some of the topics I have in mind. Please post your suggestions, and I'll try to dig something up if it's available.

• fatality rate per registered motorcycle over the past 30 years
  
  
• single- and multiple-vehicle crashes
  
  
• the role of SUVs and other light trucks
  
  
• deaths by age and sex
  
  
• alcohol-involved fatal crashes
  
  
• fatal crash involvement by motorcycle manufacturer
  
  
• helmet use in fatal crashes and the effect of helmet law

I'm not going to include references in individual posts, but I'd be glad to share sources and methods with anyone who's interested. PM me with your questions.

 

[DataDan]

Registrations, Deaths, and Fatality Rate

The talking head gravely proclaims: “Motorcycle Death Toll Soars Again. Tune in at 10.” And then various chin-pulling safetycrats, insurance industry representatives, doctors, and mothers are interviewed, each weighing in with an opinion about motorcycle safety. Of course, this plays much better if it’s edited into video of some horrific motorcycle crash.

Anyone who has read newspapers and watched TV news for a while will take such hysteria with a grain of salt. Chances are the “experts” don’t actually know the facts, or if they do are distorting them for their own purposes. And the news-droid, with the sophistication of a special-ed 8th grader, laps it up, totally unburdened by the hard-boiled skepticism his breed is supposed to possess.

Indeed, annual motorcycle deaths in the US have more than doubled since 1997, a fact trumpeted in more than one headline. But that’s less than half the story. The rest includes the soaring popularity of motorcycling and the historical perspective, which are shown in the attached chart.

Motorcycling in America is in an unprecedented, sustained boom (though probably winding down a bit right now). In the early ‘90s, annual unit sales had dropped to 280,000 a year, off by two-thirds from a decade earlier. And the number of registered bikes fell to 3.8 million in the mid-90s, the lowest count seen in the 30 years DOT has been reporting. But in the past 15 years, sales have rocketed to over 1 million units a year and registrations have hit an all-time high.

As the popularity of motorcycle soared, so have deaths. That’s not surprising. We all know it’s dangerous; that’s why we get trained, read safety-related books, magazine articles, and forum threads, wear good protective gear, and keep our rides in tip-top shape. So the question a thoughtful observer would ask isn’t whether more people are dying—with 75% more people on motorcycles that’s almost a foregone conclusion. The question is whether deaths are increasing disproportionately.

One way to answer that question is by looking at the death rate per registered motorcycle. In fact, the rate has climbed too. From an all-time low of 55 deaths per 100,000 registrations in 1997, it grew to 72 deaths in 2006, a 31% increase in 9 years. Not a good thing, of course, but viewed in historical context, it doesn’t seem particularly alarming.

To get the 31% increase one must cherry-pick the starting rate at its all-time low in 1997. The rate of 72 in 2006 is also about the same as it was in 1989 and 25% below the all-time high of 1978. Is there any reason to believe that the rate should have remained at its all-time low? And how did it get there, anyway? The 40% rate drop from the mid-80s to the mid-90s was pretty extraordinary.

What do you think has driven the motorcycle death rate increase since 1997, and what do you think caused the sharp drop between the ‘80s and the ‘90s?

 

[SVforMe]

• fatal crash involvement by motorcycle manufacturer
  

I'm interested in this juicy tidbit. So, does this mean you use SAS/STAT recreationally?

 

[jrace]

I'm interested in this juicy tidbit. So, does this mean you use SAS/STAT recreationally?

Dude. His username is DataDan. That's practically superpowers right there.

-jim

 

[xgambit]

and i thought this was a thread about yeast infection :wtf

 

[donoman]

Instead of comparing vs. registrations, can you find the # of licensed riders? The reason I ask is that many people own multiple motorcycles. Of course, some licensed riders no longer ride.

 

[Nemo Brinker]

Fabulous information, Datadan, htanks for using your superpowers here! Any way you can pull out deaths per miles ridden?

 

[DataDan]

Nemo Brinker wrote: Any way you can pull out deaths per miles ridden?

I could post a chart, but it wouldn't be meaningful.

US DOT estimates of annual motorcycle miles are crap because states aren't required to report it, and some don't. For 2003, DOT estimated 1800 miles per registered motorcycle, while the Motorcycle Industry Council (trade group) estimated over 3000 miles.

One action item that came out of a 2007 NTSB (part of DOT) motorcycle safety forum is for DOT to find a way to publish more accurate motorcycle mileage estimates.

Since annual mileage per bike probably doesn't change much year-over-year, the rate per registration is a reasonably good way to view the trend in the short term, IMHO. However, lack of mileage estimates makes a comparison between bikes and cars a problem. But then there are lots of other reasons that the car/bike comparison is difficult.

donoman wrote: Instead of comparing vs. registrations, can you find the # of licensed riders?

US DOT doesn't report motorcycle licenses. In fact, I don't think they're required in all states. California does report them, and the number of licensed motorcycle riders is about 50% higher than the number of registered motorcycles. So rate per licensed rider wouldn't be able to show the effect of multiple-bike riders.

 

[TTTom]

...To get the 31% increase one must cherry-pick the starting rate at its all-time low in 1997. The rate of 72 in 2006 is also about the same as it was in 1989 and 25% below the all-time high of 1978. Is there any reason to believe that the rate should have remained at its all-time low? And how did it get there, anyway? The 40% rate drop from the mid-80s to the mid-90s was pretty extraordinary.

What do you think has driven the motorcycle death rate increase since 1997, and what do you think caused the sharp drop between the ‘80s and the ‘90s?

It would be interesting to compare this data with the various changes in state helmet laws. There was a push to enact helmet laws all around the country after the first big US motorcycle boom. Since then the trend has gone the other way and a lot of states have rolled back their helmet laws.

 

[bikecraft]

-deaths by engine CC size

 

[LS1Bandit]

Since annual mileage per bike probably doesn't change much year-over-year,

Out of curiosity - do you have anything to support that? It would seem that you're potentially throwing out a clue to the increased rate. If people are riding their motos more, then that might explain the increased death rate. Maybe as gas prices continue to go up people use their motos for commuting more and thus their mileage goes up ....

 

[DataDan]

Out of curiosity - do you have anything to support [DataDan's assertion that average annual mileage per bike doesn't change much year-over-year]? It would seem that you're potentially throwing out a clue to the increased rate. If people are riding their motos more, then that might explain the increased death rate. Maybe as gas prices continue to go up people use their motos for commuting more and thus their mileage goes up ....

Motorcycle Industry Council owner surveys showed just a 1% change in annual mileage per bike from 1998 to 2003 (the two most recent). But those figures nearly doubled the annual miles from the 1990 survey.

That's the support for my guess that miles-per-bike doesn't change much in the short-term. Plus, just a sense that when we're talking about 5 million motorcyclists, population behavior tends to change slowly.

 

[DataDan]

Gray-Haired Hooligans

Sometime within the next few months we’ll be discussing the latest news release about motorcycle safety. It will probably come from government bureaucrats, the insurance industry, or the medical profession, and it will be picked up by the Associated Press and distributed far and wide to a front porch, TV, or website near you. The article’s grabber will be about the increase in motorcycle deaths, which have more than doubled in the past ten years. Various experts will be quoted, each blaming the rising death toll on a favored demon summoned up from the statistical ooze. Repealed helmet laws, big cruisers, sportbikes, and alcohol may be cited. But one of the most popular culprits will be “older” riders. What could be more obvious? Feeble-minded, feeble-bodied old coots trying to recapture lost youth, take up motorcycling when they should be playing shuffleboard at the Senior Center and counting down the days ‘til their octuple bypass.

Problem is, older riders aren’t the problem. At least not in the way the experts would have us believe. They’re the ones dying on motorcycles these days because they’re the ones who happen to be riding motorcycles these days. And bikes are dangerous, just as they were when they were invented 100 years ago, when T.E. Lawrence died on his Brough Superior, and when Brando portrayed the archetypical hard-ass biker dude in The Wild One. Truth be told, risk is part of the attraction. A motorcycle demands a rider’s best if he is to survive, and our over-padded, over-regulated, warning-labeled modern world doesn’t offer a lot of challenges like that. Casualties, while tragic, are also inevitable.

An earlier post showed that US motorcycle deaths have indeed more than doubled in the past 10 years and that the death rate per registered motorcycle has increased too. Breaking down deaths into age groups, riders 40 and up have accounted for most of the additional 2700 deaths in 2006 over 1997, and the 40-49 and 50+ age groups each account for nearly as many deaths as the 20-29 group (see attachment 1).

Experts commenting on this change, digging deep for an explanation, will assert that older riders crash because they are unfit to ride motorcycles. Diminished strength, agility, vision, and reactions are often mentioned. What will be ignored in the rush to identify a cause is the abundant evidence that older motorcyclists are less likely to crash than younger riders. Hurt found: “Motorcycle riders between the ages of 16 and 24 are significantly overrepresented in accidents; motorcycle riders between the ages of 30 and 50 are significantly underrepresented.” Similarly, MAIDS (the European study published in 2004) found: “Riders between 18-21 and 22-25 were over-represented, while riders between 41 and 55 were under-represented in the accident population.” The same relationship between older and younger riders is found when US fatality rates per owner by age group are compared. As attachment 2 shows, older riders have a much lower death rate than younger riders. The growing number of 40+ riders among those killed in crashes is due not to imagined infirmity but to their growing numbers in the riding population.

The changing age distribution of US motorcyclists also helps to explain the long-term trend in the fatality rate described in an earlier post. Deaths per 100,000 registered motorcycles dropped from 88 in the mid ‘80s to 55 in the mid ‘90s and have since crept back up to 72 in 2006. Authorities commenting on this trend focus on the recent increase, but an understanding of the earlier decrease is also necessary to get the big picture. Attachment 3 shows that the falling death rate from 1986 to 1997 tracked a plummeting percentage of under-30 riders in the population while the percentage of 40+ riders grew. Because of the wide gap in age-group death rates seen in attachment 2, the change in rider age distribution significantly affected the overall death rate. The combination of fewer <30 riders and more 40+ riders pushed the fatality rate downward.

But age distribution alone can’t fully explain the drop in the death rate through the mid ‘90s, nor can it explain the subsequent increase. To complete the picture it’s necessary to consider also riding experience, another significant factor in crash risk. Hurt found that riders with less than 6 months experience were more likely to crash than the riding population average and riders with over four years were less likely to crash. MAIDS found that “riders who have less than 6 months experience on any motorcycle are more likely to be in an accident when compared to the riding population,” and “riders with a great deal of riding experience (i.e. over 98 months) were found to be less likely to be in an accident.”

An increase in average experience and consequent decrease in risk were the silver lining in the decline of motorcycling that began in the early ‘80s and ended at the sport’s low point in the mid ‘90s. Fewer new riders—with zero experience—were taking up the motorcycling and lowering the average, while those who stuck to it gained experience and raised the average. Median rider age climbed from 27 in 1985 to 32 in 1990 to 38 in 1998, due to the simultaneous effects of older riders taking up the sport and continuing riders gaining both years and experience. The combined effects of an aging riding population with more experience led to the all-time low rider fatality rate in 1997.

The mirror image of the slumping industry / decreasing risk relationship was an increase in average population risk when motorcycling rebounded in the mid ‘90s. More new riders were taking up the sport, lowering experience and raising risk in the population. As shown in attachment 3, the rider increase was almost exclusively in the 40+ age group, and the increase in the fatality rate for those riders from 1998 to 2003 seen in attachment 2 is the result. These older riders are dying more frequently—though still well below the rate for riders under 40—not because they are old but because they are newbies.

The riding season safety drumbeat is just around the corner. When it arrives, try to keep the safetycrats’ message in perspective. If you’re an older rider, the news isn’t as dire as it seems. If you’re a younger rider, don’t interpret the emphasis on aging riders as letting you off the hook. We all need to keep our heads in the game at all times, no matter how old we are.

 

[DataDan]

Fatal crash involvement by manufacturer

A question sometimes asked is how fatalities break down by type of motorcycle. Who’s more at risk, the cruiser rider or the sportbiker? Actually, the subtext of the question is more about riders than motorcycles. And it’s more of an assertion than a question: Those who ride my kind of bike are surely safer than those who ride the other kind.

Unfortunately, DOT’s data doesn’t make that kind of information easily available. Unlike cars in fatal crashes, which are identified by make and model, motorcycles are identified only by make and displacement group—and the displacement groups aren’t very helpful.

Even analysis by make is a problem because there are no exposure measures to compare to. One can’t look up registrations by make and calculate a fatal crash involvement rate per registered motorcycle. So the meager knowledge that can be gained comes from looking at trends over time. And that’s what the chart below shows.

Considering the fatality rate increase over the past 10 years shown in an earlier post, is there anything that points to a particular brand of motorcycle driving the trend?

From 1994 to 2000, Harley involvement in fatal crashes increased considerably, growing from 21% to 31%. That’s the time when Harley sales were growing fastest, and they probably attracted inexperienced new riders. But after that, the HD contribution has leveled off, so their contribution to the subsequent increase isn’t out of proportion to that of other manufacturers.

After 2000, Suzuki has increased from 14% to 18%, which might reflect the success of the GSX-R series and, of course, the “fastest bike on the planet”, the Hayabusa. These motorcycles are wildly popular with sportbike riders who may be pushing rates upward.

The real puzzler is the declining percentage of Hondas among bikes involved in fatal crashes, which dropped from 30% to 20% in the period charted. As of 2003 they still held the #1 plate in market share, so declining sales don’t seem to be the answer.

The “other” category in the chart includes motorcycles that NHTSA reports as other and unknown as well as BMW, Ducati, and Triumph, whose percentages are too small for this chart. One noteworthy detail is a sharp increase in the number of Ducatis in fatal crashes. From the low- to mid-teens in 1999 to 2004, it jumped suddenly to 33 in 2005 and 27 in 2005. This scrap of evidence is too small to be considered significant, but it hints at increasing risk in the high-end sportbike segment.

 

[shraz]

interesting graphs.

 

[SpikedLemon]

Have you sufficient data to determine cause of accident?
Single Vehicle vs. multi-vehicle?
if Multi-vehicle: who was at fault?

Or: rate of speed in respect to the accident rate? (hurt report's data suggested that more accident occurred at low speeds <30mph) ... Can that be correlated back to engine displacement (does engine displacement actually increase the death rate / accident rate)

 

[vaara]

So if they'd just raise the motorcycle licensing age to 40, the fatality rate would be cut in half! :thumbup

Way, way, way OT, but DataDan: do you have any source for motorcycle sales and/or registrations by model and manufacturer? I've been web-snooping for this data, but haven't found anything.

 

[Eisernkreuz]

Harley and Honda riders seem to be the problem :laughing No wonder my bike's insurance is $8.50 a month :rofl

 

[budman]

:applause

Great stuff Dan as always.

:smoking

 

[DataDan]

SpikedLemon wrote: Have you sufficient data to determine cause of accident?
Single Vehicle vs. multi-vehicle?
if Multi-vehicle: who was at fault?

Or: rate of speed in respect to the accident rate? (hurt report's data suggested that more accident occurred at low speeds <30mph) ... Can that be correlated back to engine displacement (does engine displacement actually increase the death rate / accident rate)

Single vs. multiple is available, as well as types of other vehicles in 2-vehicle crashes. That will be in an upcoming post.

Circumstances of a particular crash (e.g., motorcycle going straight hit oncoming left-turner who failed to yield to motorcycle) can be inferred by looking at details in the database. But there's no easy way to identify causes in a query on, say, 1000 crashes. I compiled data on Central Coast deaths over the past three years via the FARS database and news reports, and I was able to construct narratives, describe the bikes and riders, and decide who was at fault. But that was 35 crashes and it took many hours. Even the Bay Area would be a much bigger job than that.

Speed isn't very consistently reported. Even if it were, there's the problem of speed when. Hurt's thorough investigations identified both crash speed and pre-crash speed. I.e., how fast the motorcycle (and other vehicle) was traveling before the crash scenario unfolded and at impact. On the FARS database, there's only one speed data element per vehicle, and it's not real clear what is reported. My only experience at analyzing by vehicle speed was fatal California intersection crashes in 2004. I found 37 where motorcycle speed was reported, in 16 the rider was speeding (> speed limit) and in 11 of those speed was >= limit + 15. I certainly didn't try to draw a conclusion from that. I was just wanted a sense of how often speed contributes to intersection crashes.

In the online database, engine displacement coding is really lame. Basically it's up to 500, 500-750, and 750+. Some of the research that NHTSA publishes reports displacement in much more detail, but I don't know how to get that data.

vaara wrote: Way, way, way OT, but DataDan: do you have any source for motorcycle sales and/or registrations by model and manufacturer?

Sales are reported by the Motorcycle Industry Council in a publication they call the Statistical Annual. You'll find ordering info at their website, but the prices they charge are outrageous.