This stuff's always confusing. I think this is the way it works...[hazarded the philosophy major :blink: ]:
Let's say you have 100 M.C. riders in your study group and you want to study accident results. Your default presumption going in is that all the riders are exactly the same, meaning every rider has an equal probability of being in an accident.
Okay, so after the study period closes you find that, of those 100, 40 have had an accident.
Okay, so that's a 40% accident rate, which translates into the default probability that every rider had a 40% probability of being in an accident.
You begin to look at all the variables attendant to the accident and non-accident participants: type of M.C., rider age, rider gender, day or night, and of course, rider apparel--color, reflectivity, bad taste, Hawaiian shorts, tassels, Harley logos, t-shirts stating "If you can read this the bitch fell off, whatever....
If 20% of the participants wore "bright-colored, reflective" gear, then your default expectation is that 20% (on average, over time, blah blah blah statistical caveats) of the accident victims, or 8 victims (.20 x 40), would be wearing yellow Aerostitch's (or equivalent but of course there is no equivalent to an Aerostitch which is why I forked out $800 not including the back and hip protectors though I don't care how visible it is I just couldn't go with the yellow and besides I was riding a Harley at the time and anyway my wife said it wasn't cool).
Where was I...
But what you REALLY find is that only, say, 3 (or 3%) of the brightly-appareled riders had accidents. So you conclude that riders who wear bright apparel have only a 3% probability--or a 37%
lower probability (40% - 3%)--of being in an accident. The "strength" of this correlation can be measured, and indeed they do point in the article to the R value or
coefficient of determination or whatever the hell that thing is that I forgot from my statistics class.
Any statistics majors out there? Did I get that right? :blink:
As long as we're on the subject of accident research, here's a link to a couple articles I published in
Friction Zone, analyzing 100,000 California motorcycle accidents that occurred over 10 years:
Rush-Hour M.C. Accident Statistics
Jb
:graduated: