Post by
IBCoupe »
https://forums.nicoclub.com/ibcoupe-u134097.html
Mon Jul 18, 2011 5:35 pm
No study; just stuff I've heard on NPR.
Questionable judgment or otherwise, it could also be a function of the statistic-gathering method. Are we sampling drivers or are we sampling accidents? In other words, what is the study asking, and whom?
If it asks male drivers and female drivers to self-report on their accident rates, that's one thing, with its own reliability questions. If it asks insurance companies for the genders involved in accidents, the reliability issue goes out the window, but your sample is very easily skewed. One method would account for the difference in gender ratios on the road, and the other would not. Looking at the slides, those statistics appear to come from a study based around the latter.
If 75% of the cars on the road are driven by men, and 25% by women, what does that do? Imagine that men and women are equally accident-prone, and that 25% of men and women each year will get into an accident.
If there are 100 people on the road, that means roughly 19 men will be in accidents, and 6 women will be in accidents. But what would the insurance companies report? Of 25 accidents this year, 76% were from men, and 24% were from women. Here's where the logical flaw appears: "Thus, men are 52% more likely to be in accidents than women."
Lies, damned lies, and statistics, son.