in_cavediver
Contributor
Charlie59:Statistics, statistics? What the heck is wrong with common sense. Why do you guys try to change the real world to meet your warped idea of reasonable.
The numbers don't have feelings and they don't care what they show. The numbers are the numbers and no one has tried to change them for personal gain in this instance. Accept reality, it is quite liberating.
I have summed my opinions up in a prior message but seeing this made me want to make a few more comments.
Statistics - Statistics are nothing more than mathmatical operations done on a set of numbers to attempt to infer meaning. Statistics in there raw form are factual, accurate and totally meaningless. Humans must look at the results and infer meaning.
For instance, lets say 45% of divers in accidents have a BMI over 30. That means 55% have a BMI under 30. This can be looked at in many ways but none have any bearing. There simply isn't enough information.
Lets add some more information (made up since I have no sourceand no one might). Lets say 65% of the diving population has a BMI over 30. Now, looking at the math, the 'obese' divers are doing much better than the 'skinny' divers. (all things equal, 65% should have been dead, not 45%)
Lets add one more factor. Lets say the total deaths were 100 for the year. Do we yet have enough information - not really. We still need the number of divers and average dives per year with respect to both classes. Only then can we determine if the sample was large enough to yield statistically valid information. To small a sample size, nothing but meaningless noise.
This is also neglecting one whole significant area - independence. Can we show that the conclusion is independent of other factors such as training level, number of prior dives, temperature of the water, depth of dive, dive time prior to accident, Barometric pressure on the surface and the average land speed of an African swallow? - Nope, BMI is not independent of the other factors.
Only once we have the full picture of the data with all dependencies addressed will we have enough to make intelligent inferences. From what I have seen in this thread, that is lacking. Without this data, it's just people looking to advance an opinion based on dubious 'facts'.