Quote:
Originally Posted by DEpointfive0
I didn't see where you got that, BUT I don't think they sold 100k of them
Skew 1:
It's a total of cars registered between 2006 and 2009... That are over 100k...
Skew 2: It says the model year for the 370Z was 2005-2008... I see reported sales for those years at around 80k.
Statistics is the study of making and analyzing how numbers lie. I have an inverted statistic:
The sold 700-800 Porsche Carrera GTs in the US (Let's call it 1000)
With the death of our beloved Roger and that other guy in the car, the death rate is 2000 million!!!
So honestly they need to specify deaths per model year, OR registration or miles driven per 100000 cars sold or something, but all that being said, the numbers can still lie
|
I don't think what you cite as skewing the results actually do that. It looks like the authors were trying to allow for reasonable bases for comparison. They had a minimum required population size and used a standard time period for that population. What they seem do have done actually makes the results easier to use.
I have no idea how you calculated 2000 million. What I will suggest is the reason they don't include cars with such low sales figures is, in part, because the likelihood of drawing conclusions from rare events (e.g. fatal accidents) is impossible with relatively small (1000 cars) sample sizes.
If you are expressing frustration about the way statistics are used against people when you remarked about "how numbers lie" I am with you. People in politics, in the media, people selling stuff, or people in legal circles often misuse or abuse statistics to knowingly make false (or unsupported claims), or to bait the untrained into drawing those conclusions on their own. Strictly speaking, however, I do strongly disagree with the notion that statistics themselves (done properly) lie - people, however, lie with statistics. Not exactly the same thing.
Back on topic, I think there is some usefulness to the study as long as we don't draw inappropriate conclusions from the data.