Motostats 2018

[DataDan]

In a hurry? Jump to the tl;dr thread summary post.


I'm going to use this thread to post information that may have gotten lost in the shuffle of current events. Such as: In spite of slowing bike sales, motorcycling is still at its highest popularity ever. Also: while we have to deal with new and metastasizing hazards such as cell phones, Prii, and SUVs, we're crashing less. And, daily commute experience notwithstanding, the Bay Area is the safest urban area in California for motorcyclists.

First up...


Motorcycling is more popular than ever

While the sport has stopped growing--at least for now--and bike sales have cratered, motorcycling itself remains popular in America, in California, and in the Bay Area.

I'm omitted crap VMT estimates from FHWA 2000-2006 (which they acknowledge).

No VMT estimates by state available.

This requires mc registrations by county, which haven't been published since 2015.

A note on sources
I won't be citing sources in each post, but everything that will appear is from a federal or California government publication. Wanna know where something came from? Send me PM and I'll reply with a link or document.
.

 

[budman]

In. As always.

2018 is because that is what is available for those wondering.

 

[ThinkFast]

Vehicles are vehicles. Why single out Prii?? (FWIW I’m not a fan of them either, but I’d call out Apple Car Play regardless of the vehicle before singling out Prii- or SUVs for that matter).

 

[DataDan]

Vehicles are vehicles. Why single out Prii?? (FWIW I’m not a fan of them either, but I’d call out Apple Car Play regardless of the vehicle before singling out Prii- or SUVs for that matter).

It was a jokey reference to a car despised by motorcyclists. Thirty years ago it was the ovloV. Then, legend has it, the torch was passed from one generation of crap drivers to the next--a transition from Swedish safety to Japanese eco-cognition. However, there's no substance in the data to support either stereotype.

SUVs, OTOH, are the real deal. Maybe not in frequency of crashes (no data to support or contradict), but definitely in lethality. A two-vehicle crash between a motorcyclist and a light truck (SUV, pickup, van, etc.) is at least 50% more likely to kill the motorcyclist than a crash with a car. Pedestrians are at a similar disadvantage. I don't expect to cover that in this thread.

 

[beakflawless]

Motorcycles continue to trend and become even more popular, they more convenient and handy to use.

 

[DataDan]

The Declining Motorcycle Crash Rate

As motorcycling grew from the mid-1990s until the recession in 2008, crashes inevitably increased, too. As shown by the gray shaded area in the graph below, US crashes of all severities more than doubled from 55,000 in 1998 to 123,000 in 2007. Crash rate also rose, whether measured per 10 milllion motorcycle miles traveled--from 54 to 58 (black line)--or per 1000 registered motorcycles--from 14 to 17 (red line).

Since 2007, however, annual crashes have been below the 2007 peak (except in 2016, a questionable result from the first year of a new NHTSA surveying system), and the crash rate has been mostly flat, averaging 55 crashes per million VMT and 13 per 1000 registrations 2008-2018.

Motorcycling in the US is now safer than it has ever been. In spite of phenomenal growth over the past 20 years, we are now less likely to crash than we were in the depth of the 1990s US motorcycle depression, and much less likely than in the 1980s boom. While many factors have likely contributed, two stand out in my mind: the aging riding population and the decline of drinking-and-riding crashes. Those will be subjects in later posts.

California motorcycle crashes and the crash rate have generally followed the US pattern, with a noteworthy difference. The drop in crashes from the 1980s to the 1990s was much greater here. With the 1986 launch of the California Motorcyclist Safety Program came the training requirement--riders under age 18 in 1987, under 21 in 1991. By upping the investment of time and effort needed to start out this kept out less serious potential riders, and it better prepared others. In addition, enactment of the all-rider helmet law in 1992 took some off the street.

Crash rate per 1000 registered motorcycles is higher in California than for the US, due undoubtedly in part to our year-round riding weather. I have not shown VMT crash rate because motorcycle VMT by state is not generally available. However, it has been estimated for some recent years and shows that Californians ride, on average, 4400 miles per registered bike per year compared to the national average of 2400. Base on that, California's VMT crash rate is 15% lower than the US average.

I have crashes by county only since 2001, so I cannot break out the Bay Area earlier than that, and I have registrations by county only through 2015, so I cannot calculate later crash rates. However, for 2001-2015, the Bay Area crash rate per 1000 registered motorcycles is slightly lower than the statewide rate.

.
.
.

 

[ThinkFast]

Motorcycling in the US is now safer than it has ever been. In spite of phenomenal growth over the past 20 years, we are now less likely to crash than we were in the depth of the 1990s US motorcycle depression, and much less likely than in the 1980s boom. While many factors have likely contributed, two stand out in my mind: the aging riding population and the decline of drinking-and-riding crashes. Those will be subjects in later posts.

I would've guessed increasing prevalence of ABS has something to do with it, too. No?

 

[Maddevill]

It was a jokey reference to a car despised by motorcyclists. Thirty years ago it was the ovloV. Then, legend has it, the torch was passed from one generation of crap drivers to the next--a transition from Swedish safety to Japanese eco-cognition. However, there's no substance in the data to support either stereotype.

.

I agree 100% It used to be white Volvo station wagons leading every car parade on just about every twisty road. Volvo emphasized their safety, which Volvo buyers needed since they sucked as drivers. Now, it seems a lot of former Volvo owners now drive Prius or Honda CRV. With the same results. I know a Prius can actually go pretty fast,( It's true!) but the operators are still mostly incompetent.

Mad

 

[DataDan]

Rider age, part 1

One reason the motorcycle crash rate has declined over the past 30 years, in my opinion, is the aging rider population. Old folks tend to be more risk averse in physical activities (though possibly not in poker and commodity futures trading). The graphs below compare the motorcycle owner age distribution from Motorcycle Industry Council data to the crash-involved rider age distribution from NHTSA data:

When US motorcycling collapsed between the mid-1980s and mid-1990s, it was riders under age 30 who left in the greatest numbers. From 1985 to 1998, that group dropped from 3 million motorcycle owners to 1 millon. At the same time, the 50+ group grew by 50%, from less than 500,000 to 750,000, while those in between, 30 to 49, were approximately unchanged. Because the <30 group is more likely to crash than the 50+, this demographic change helped drive down the crash rate. Statistically, a higher risk group was being replaced by a lower risk group.

But there's more to the story. Meanwhile, the high-risk young riders improved, their crash rate falling 35% from nearly 40 crashes per 1000 registrations in 1990 to 25 in 1998. Risk fell in other age groups, too, but the drop in the <30s had the biggest impact on overall crash rate.

A downside of the older riding population has been increased crash lethality--the chance of death in the event of a crash. Older riders are more likely to die when they crash than younger riders, and, as a result, the decline in fatality rate has been less than the decline in crash rate. That will be the subject of a later post.

Next I will compare the age distribution of crash-involved riders in the US, California, and the Bay Area.


My US crash data goes from 1988 to 2018, owner age distribution from 1985 to 2016. For consistency in the graphs, I've plotted only the overlapping years, and I have smoothed the sometimes jumpy NHTSA crash estimates with a 3-year moving average. Age groups were dictated by available ownership data.
.

 

[budman]

That is an interesting theory Dan.

Rings true to me. Well thought out :thumbup

 

[DataDan]

Rider age, part 2

The following graphs compare age distributions for crash-involved riders in the US, California, and the Bay Area 2001-2018 (the extent of my California age data). All police-reported crashes are included (estimated by NHTSA for the US, counted for California). About 4% are fatal, 21% major injury, the rest non-injury or minor injury.

One slight difference is that fewer older riders--55+--and more younger riders--under 35--appear in California and the Bay Area crashes. This most likely reflects the proportions of riders in those age groups.

Another way to look at the same data is by rider generation. The Millennials have evidently taken up motorcycling in greater numbers here in California than in other states. Note that the Millennial numbers in 2018 are greater than Generation X in 2002, when they were the same age.

I have used what seems to be a common definition of generations:
Baby Boom: birth year 1946-1964
Generation X: 1965-1980
Millennial: 1981-1996
Generation Z: 1997-
Birth year estimated as crash year - rider age, so not perfectly accurate
.

 

[DataDan]

Drinking and riding and crashing

To recap, the US motorcycle crash rate--both per mile traveled and annually per registered bike--has fallen substantially over the past 30 years. Riding is now as safe as it has ever been in the US. In part, this is due to a demographic shift from higher-risk young riders to lower-risk old folks. But declining crash rate within age groups has also contributed--particularly among young riders. Why is that?

A big contributor is a reduction in drinking-and-riding crashes. The graph below shows drinking rider crashes from NHTSA's sampling system since 1988. Over the past 30 years, they have dropped by more than 30%. This is an accomplishment the motorcycling community can be proud of.

The next three graphs show how the US, California, and the Bay Area compare since 2001. In the past five years, alcohol involvement has been around 6% for all, down from 8% in the early 2000s.

Fatal crashes

A few years ago some mistaken information was circulating about alcohol involvement in fatal motorcycle crashes, and I would like to correct that. You may have heard that in California it was over 50%. In fact, it has been unchanged at around 30% for the past 10 years. Before that, though, it had fallen from nearly 50% in the early 1990s. The next set of graphs compares alcohol use in fatal motorcycle crashes in the US, California, and the Bay Area over 30 years. Data is from NHTSA's fatal crash database for all.

.
.
.

 

[DataDan]

Fatalities

As we've seen in previous posts, motorcycling in the US grew tremendously over the past 20 years. At the same time, motorcycle crashes increased as well. Inevitably, deaths followed:

However, increasing fatalities generally followed the growth of the sport, as seen in the fatality rate per registered motorcycle (gray shaded area). The exception is the disproportional rise during the period of high growth from the late 1990s until the recession. This may be due to the decrease in average experience as new riders took up the sport. When growth stopped in 2008, the fatality rate fell back to the low pre-boom level and flattened, and is now near the all-time low.


California has been similar, with a notable exception. After a sharp drop in deaths and fatality rate at the onset of the recession, the pre-recession increase resumed. Fatality count has exceeded the pre-recession high, the fatality rate nearly so.

The recent history of Bay Area fatalities is similar to the statewide record, though my lack of registration data results in an incomplete picture of fatality rate. Note that the Bay Area graph, unlike those for the US and California, is not a 3-year moving average. I didn't want to smooth out the sharp peak of 2007 and 2008. This was real and had a devastating effect on BARF--sometimes two or three RIP threads a week.

Coming up next, I'll take a closer look at fatality rate by breaking it down into crash rate and crash lethality.

 

[motomania2007]

Dan, does your alcohol involvement data mean .08 BAC (legally drunk) and higher or just measurable BAC (impaired and drunk rather than just legally drunk)?

 

[budman]

Thanks DD.
The flattening mentioned is good.. more training? More..:dunno

Just good to see.

I remember 07-08 well. Got the start up of 1Rider in my link.
First radio appearance as we launched. Enchanter smoked me on the MIC. :laughing

Really appreciate your effort then and now. :thumbup

 

[DataDan]

Dan, does your alcohol involvement data mean .08 BAC (legally drunk) and higher or just measurable BAC (impaired and drunk rather than just legally drunk)?

It's either measured BAC > 0, or assessed by the investigating officer as had been drinking (but not necessarily impaired).

If the rider is killed, there is often BAC from an autopsy. In California, more than 80% of deceased riders have BAC reported (even if 0); in other states it can be less.

In other cases, the investigator usually makes a judgment call on whether the rider had been drinking.

Data is from three source: FARS, the USDOT fatality database; CRSS, the USDOT crash sampling system; and SWITRS, the CHP database of all California crashes. FARS and CRSS use BAC if reported, officer assessment if not. SWITRS uses only officer assessment.

 

[DataDan]

Thanks DD.
The flattening mentioned is good.. more training? More..:dunno

My opinion is that the flattening of the US fatality rate curve post-recession resulted from the end of the motorcycling boom. I'll get into this a bit more tomorrow, but my theory hypothesis conjecture wild-ass guess is that crash and fatality rates move in the same direction as growth in the sport--up when growing, down when declining.

When the popularity of motorcycling is increasing, the noob percentage is high and, consequently, average experience falls. On the other side of the boom--the bust, such as in the early 1990s--fewer noobs means higher average experience. Because greater experience decreases crash risk (seen both in Hurt and MAIDS), the crash rate curve will follow the sales curve.

 

[motomania2007]

It's either measured BAC > 0, or assessed by the investigating officer as had been drinking (but not necessarily impaired).

If the rider is killed, there is often BAC from an autopsy. In California, more than 80% of deceased riders have BAC reported (even if 0); in other states it can be less.

In other cases, the investigator usually makes a judgment call on whether the rider had been drinking.

Data is from three source: FARS, the USDOT fatality database; CRSS, the USDOT crash sampling system; and SWITRS, the CHP database of all California crashes. FARS and CRSS use BAC if reported, officer assessment if not. SWITRS uses only officer assessment.

When you say "80% of deceased riders have BAC reported" does that mean 80% of deceased riders have a positive BAC <0 or that only 80% have a BAC data filled in and some (most?) of them are reported at 0%?

 

[DataDan]

When you say "80% of deceased riders have BAC reported" does that mean 80% of deceased riders have a positive BAC <0 or that only 80% have a BAC data filled in and some (most?) of them are reported at 0%?

California Motorcycle Riders Killed by BAC

[table=head] | 0.000 | .001-.079 | .080-.149 | .150+ | not reported | % reported
.......... | ................ | ................ | ................ | ................ | .................... | ....................
1990 | 239 | 54 | 61 | 129 | 45 | 91%
1991 | 218 | 40 | 54 | 90 | 45 | 90%
1992 | 173 | 19 | 28 | 62 | 24 | 92%
1993 | 164 | 15 | 36 | 42 | 38 | 87%
1994 | 137 | 24 | 46 | 36 | 28 | 90%
1995 | 129 | 22 | 23 | 47 | 25 | 90%
1996 | 115 | 15 | 24 | 40 | 27 | 88%
1997 | 120 | 14 | 26 | 37 | 24 | 89%
1998 | 106 | 11 | 18 | 39 | 18 | 91%
1999 | 132 | 10 | 23 | 30 | 25 | 89%
2000 | 148 | 17 | 24 | 34 | 35 | 86%
2001 | 177 | 21 | 22 | 38 | 18 | 93%
2002 | 190 | 23 | 27 | 37 | 30 | 90%
2003 | 223 | 18 | 29 | 52 | 35 | 90%
2004 | 257 | 20 | 27 | 59 | 42 | 90%
2005 | 276 | 22 | 28 | 55 | 73 | 84%
2006 | 293 | 30 | 42 | 62 | 49 | 90%
2007 | 319 | 28 | 30 | 78 | 41 | 92%
2008 | 336 | 20 | 50 | 83 | 48 | 91%
2009 | 255 | 16 | 34 | 49 | 31 | 92%
2010 | 207 | 24 | 27 | 57 | 25 | 92%
2011 | 231 | 28 | 29 | 49 | 50 | 87%
2012 | 266 | 22 | 26 | 66 | 46 | 89%
2013 | 274 | 30 | 32 | 67 | 42 | 91%
2014 | 289 | 27 | 41 | 80 | 66 | 87%
2015 | 274 | 68 | 22 | 53 | 60 | 87%
2016 | 336 | 30 | 47 | 75 | 64 | 88%
2017 | 329 | 37 | 35 | 90 | 71 | 87%
2018 | 208 | 19 | 28 | 44 | 173 | 63%
[/table]

"Not reported" includes: not tested, tested but results unknown, unknown if tested, and not reported.

The reported % for 2018 is low because it's from the first database release. The final usually comes out in October.

Data for this table is from the US DOT fatality database, which includes all riders involved in fatal crashes. Around 10% of riders in fatal crashes survive (passenger, another rider, or non-motorcyclist killed), and only about 20% of those have BAC reported.

 

[DataDan]

Crash Lethality

In yesterday's post I tried to show that the motorcycle fatality rate per registered bike increased as the sport grew quickly from the late 1990s until the recession hit in 2008. Maybe this was caused by a rising crash rate--more crashes, more deaths? No. A graph back in post #6 showed that crash rate had not climbed that way. In fact, it declined in the early 1990s then flattened through the early 2000s. What gives?

Another way to look at fatality rate is as a combination of crash rate and crash lethality--the percentage of motorcycles involved in crashes multiplied by the percentage of crashes that kill the motorcyclist. If 2% of registered motorcycles crashed over a year's time, and 3% of those crashes resulted in death, then 0.06% of motorcycles were involved in a fatal crash. Crash lethality helps explain the apparent contradiction between a falling crash rate and a rising fatality rate.

This graph shows three plots for the US since 1990. Fatality rate (gray shaded area) is the same as in yesterday's post. But instead of showing it along with registrations and deaths, I've combined it with crash rate (gold) and crash lethality (blue).

At first, crash rate and fatality rate declined in parallel as the motorcycle depression of the 1990s came to an end. A lot of factors surely contributed--the spread of training, more capable motorcycles, greater safety awareness among riders. Another factor I suspect is related directly to the decline of motorcycling. As the sport collapsed in the late 1980s, sales dropped by nearly two-thirds, so there were fewer new riders and proportionally more veterans on the road. Because of this, average motorcyclist experience was up, and we know from the Hurt and MAIDS crash studies that experience substantially reduces crash risk.

However, when the motorcycling resurgence began in the late 1990s, crash rate flattened out but the fatality rate reversed and began to climb. This was because motorcycle crashes were becoming more deadly. In the early 1990s, before the boom kicked off, the lethality rate was already climbing. From a low of 30 deaths per 1000 crashes in 1992, it rose by more than one-third in less than 10 years and continued to increase until the recession hit. By then, crash rate was near an all-time low, but, driven by crash lethality, the fatality rate was higher than it had been in more than 15 years.

When the boom came to an end in 2008, the fatality rate dropped by 25% in four years, drawn down by a similar decline in crash rate, and that is how it has remained through 2018. My hunch is that the recent crash rate drop was, again, the noob/vet effect--fewer new riders, more veterans, lower average risk. Unfortunately, our all-time low crash rate--20% less than post-depression low of 1999-2000--has not been accompanied by an all-time low fatality rate. High crash lethality has kept that achievement out of reach.

Tomorrow I will identify factors that have driven crash lethality to its current high level and will keep it there for the near future.


California and the Bay Area

The California graph of fatality rate, crash rate, and crash lethality is similar to the US, so my previous analysis applies, at least generally. However, I do want to point out some differences.

My confidence in California crash counts for the 1990s isn't real high, so neither is my confidence in crash rate and lethality. However, I feel pretty good about crash counts from 2000 forward. I am much more confident about registrations and fatalities, so I trust fatality rate over the entire range.

Notice that the California crash rate (gold line) is higher than the US--most likely due to our year-round riding--while lethality (blue line) is lower--probably due to the helmet law. Yet the fatality rate (gray shaded area) for both is in the neighborhood of 60 deaths per 100,000 registrations. End result is the same, but contributing factors differ.

The precipitous drop in fatality rate from 1990 to 1994 is due to two big factors: First, the training requirement for riders < age 21 went into effect in 1991 (the < 18 requirement began in 1987). Whether training made them safer or dissuaded them from taking up the sport in the first place is more of a philosophical question at this point. Whatever. Young riders were a big demographic in the sport in 1990, so the training requirement very likely prevented crashes. Second, the all-rider helmet law went into effect January 1, 1992. Again the question is dissuasion or prevention, but it certainly saved lives.

Compared to the US, California's fatality rate dropped more sharply at the onset of the recession in 2008, and while the US rate has been flat since, California's increased, though not revisiting the pre-recession high. Notice that crash rate and crash lethality have both driven California's post-recession fatality rate higher since 2011.

Again my lack of registration data limits conclusions about the Bay Area. Worth noting, however, is that the crash rate, crash lethality, and fatality rate are all quite a bit lower than statewide. I have no explanation for the fact that the 2007-2008 fatality peak was a product of crash lethality rather than crash rate. I have piles of data on those years but haven't dug into it to find an answer. Post-recession, the crash rate mostly paralleled the rest of the state, but lethality remained much lower. One guess about the lower lethality is that many Bay crashes are low-speed freeway and city incidents that usually don't seriously injure the rider.