Oh, what the hell. Let's break out the whole drum kit, namely the Berkeley report on "lane splitting" (filtering in the UK)...
Safety implications of lane-splitting among California motorcyclists involved in collisions
we found significant variation in the manner in which lane-splitting was done. Lane-splitting was done in traffic flowing at a range of speeds. The motorcycle speed almost always exceeded the traffic speed by a small margin but, in many cases, exceeded it greatly. We compared the proportion of collision-involved, lane-splitting motorcyclists with injury across several body regions by whether the lane-splitting was done only in traffic flowing at 30 MPH or less and that the motorcycle speed should exceed the traffic speed by no more than 10 MPH. We found that the proportion with each injury type was high when the lane-splitting was consistent with neither speed component, was lower when it was consistent with one speed component, and was lower still when it was consistent with both speed components.
From table 9 of
http://www.sacbee.com/news/local/tr.../California Lane Splitting Crash Analysis.pdf
Fatal injury collisions while lane-splitting: 16
Non-fatal collisions while lane-splitting: 1147
Probability of death given that you are lane-splitting *and given that* you crash = 16/(16+1147) =
1.38%
Fatal injury collisions while _not_ lane-splitting: 204
Non-fatal collisions while _not_ lane-splitting: 6388
Probability of death given that you are _not_ lane-splitting *and given that* you crash = 204/(204+6388) =
3.09%
1.38% is less than 3.09% right? And 16 is
much less than 204! So doesn't that
prove that lane-splitting is safer?
NO! It does not take account of the probability of crashing during each of the two options (lane-splitting or not.)
This is a bit like reading that fewer people die each year from shark attacks than from dog attacks and concluding that
a shark is less deadly than
a dog.
For demonstration's sake, what if there were a total of 200 million miles travelled by motorcycle, and 5% (10 million) of those were spent lane-splitting? We get:
Fatalities per million miles of lane-splitting = 16/10 = 1.6
Fatalities per million miles of not lane-splitting = 204/190 = 1.07
This would show that lane-splitting is about 50%
more likely to lead to death!
(Note that the total mileage drops out of the equation when looking at the relative risk - we can assume any amount and we'd still get that 50%.)
To repeat, that 5% value is made up - but it illustrates the point. The Berkeley report does not
prove that lane-splitting has less chance of death.
And here is the bottom line (not speculation). The magic figure is around the 7.2% mark.
If more than 7.2% of motorcycle miles are spent lane-splitting, then lane-splitting has been shown to have
less chance of fatality.
If less than 7.2% of motorcycle miles are spent lane-splitting, then lane-splitting has been shown to have
more chance of fatality.
We don't know which it is BECAUSE THE PERCENTAGE OF MILES SPENT LANE-SPLITTING HAS NOT BEEN ESTABLISHED!