Mar 1, 2013

Liars, Damned Liars, and Statistics

One of the things that I'm enjoying the most about The Geezer with a Grudge: Average Mileage spreadsheet is the arguments about how invalid the data is. NHTSA, the states, the AMA, and insurance companies simply make up the data they use for motorcycle fatalities per mile traveled because they are too lazy to get real data. If they wanted real data, all they have to do is use the information we're all forced to provide every time a motorcycle is sold: the manufacturer, VIN number (which can easily be linked to model, engine size, year of manufacturer, etc.), and mileage at the time of the sale.

The fact is, what the database is showing so far is about what I expected to see; the average motorcyclist could get by with a 100cc scooter. With the big mileage riders skewing the data to the high-side, average miles/year is moving towards 1500/year. Still, it's hard to justify more than 250cc if that's all you're going to do with a motorcycle. This is what statisticians call a "fat-tail curve," though. There are going to be a few bikes on the right side of the curve and a crap-load of bikes on the low side. I'm not even a little surprised.

All of my life, I've bought my motorcycles by cherry-picking the 80% of motorcycle buyers (Note: I did not call them "motorcyclists."). Using good old Pareto Principles, 80% of the people who buy a motorcycle are unlikely to use it for much. Apparently, based on our current data, the more the bike costs the less likely it is to be ridden. More importantly, for me, the trendier the bike the more likely it will be found super-cheap a year after the trendiness wears off.

For example, my 1999 Suzuki SV650. I bought that bike in mid-2000, after the press had raved this bike into unobtainium for two years. My local dealers had no SV's in stock and were asking a slight premium for next year's 2001 production. However, I kept an eye on Craigslist around the Midwest and found about a half-dozen for sale for about 1/2 list price with less than 100 miles on the odometer. A young "motorcycle buyer" fell for the hype, bought a new SV for full list-plus, put the recommended farkles on the bike, dropped it in his driveway, and decided motorcycles were too dangerous. Five years and about 30k miles later, I sold the SV for slightly more than I'd paid for it and bought a year-old 2004 V-Strom (the "next big thing") for the same amount with less than 1,400 miles of local use (owned by an old guy who dropped the bike once and gave up). Well-used V-Stroms were going for serious money in 2005, but I waited for the usual suspects to show up and was not disappointed. I expect to unload my 2004 V-Strom for close to what I paid for it 50-some-k-miles later.

As our data is demonstrating, most motorcycle owners are not motorcyclists.


Trobairitz said...

Me thinks I need to get out and ride more. I average 3500-4000 miles per year on the Gladius. To me that is low but non-moto related things seem to get in the way on the weekends.

Your post just makes me want to get out on the bike more.

I do find it funny how many bikes we see for sale that have very low miles. 3-4 year old biks with less than 1,000 miles, etc.

T.W. Day said...

The Gladius is a great bike. I had a chance to ride one the first year it was available here. There is no good reason why you shouldn't expect it to replace a lot of your cage miles during the year. It's 31F out today and the roads are mostly clear. I'm digging my gear out for a late Friday afternoon ride.

Flyboy said...

Is the "average" miles an accurate data point or would the "median" be more accurate?
As for most efficient cc size, I am sure a motorized bicycle would trump pretty much anything. But then, that wouldn't be cool or fun would it?
I am sick of hearing how sports bike or large tourer riders are saving the planet and cutting down on traffic congestion by riding their bikes. (rolls eyes)

T.W. Day said...

There are 3 basic kinds of "averages": mean, median, and mode. This is the Excel "average" stat, or mean. I suspect most of the cruiser crowd would be confused by listing that as "mean." I'd like to include a distribution chart in the spreadsheet, but haven't done the work for that yet. It will be a very fat tail curve, though.

Unknown said...

Actually, I added a cumulative distribution chart to the spreadsheet a few days ago. Just click on the "Cumulative distribution" tab at the bottom of the screen. Let me know if it's not showing up and I can see what the issue is.

T.W. Day said...


Brilliant. The "histogram" page is illustrating the fattest "fat tail" I've seen in years. ;-)