[ClearCase_guy]

I agree with the statistics here that show a huge increase in these deaths of athletes.

And I find it disturbing that there are “experts” and “fact checkers” out there who can whip out their own set of statistics and just wave this all away “Everything is normal”.

The system is fundamentally broken when the statistics can prove two opposite things.

[Red Badger]

[Alberta's Child]

I work in a STEM field, and I had more than 30 credits of undergraduate science and engineering courses with laboratory components. I routinely work with statistics, and I can tell you that most of what passes for “statistical analysis” in public matters is highly error-prone at best and fraudulent at worst.

One common problem is “sample bias” or “selective sampling” when using samples to measure things over large groups.

Let’s consider average height, for example. If you measured all 330+ million Americans, you’d find that they have an average height of about 5’6”. If you took a truly random sample of 10,000 Americans, you’d probably come up with the same average height.

But if your “random” sample of 10,000 Americans only included men, you’d come up with an average height of about 5’9”.

And if your “random” sample only included men who have played basketball at least at the collegiate level, you’d come up with an average height of 6’3” or taller.

And if you are looking for incidences of health problems or other characteristics, you have to remember that the more you look for them, the more cases you will find. That’s why hospital admissions, for example, are usually a highly inaccurate method for computing the proportions of the population dealing with various health issues … because anyone who shows up at a hospital is already dealing with a problem to begin with.

[ByteMercenary]

Not sure if one can say that the system is broken because statistics blah blah ...

Statistical results are utterly dependent on the amount and quality of input. Since this is a well known property of statistics, the real science is setting up the input to a statistical study so the results favor the opinion of the guy paying the bill.

I had this argument with my stat professor several times. My opinion was that it is ridiculous to say that if I solicit 213 opinions on a subject, get a result, then say that the majority opinion is the opinion of everyone in the USA. So I called Bullshiite which is probably why I got a ‘C’.