Quote:
Originally Posted by russotto
To reiterate what UT said earlier, the actual figure, taking into account only sampling error, was 101,000 plus or minus 93,000. When your error bars are of the same magnitude as your data points, you don't have data; you have junk.
|
You don't understand statistical sampling. Here's a rough graph of what the results show. The probability (0 - 1) of any given number being the correct one is a point on the y-axis, the number of casualties (1 - 300,000) is a point on the x-axis. When graphed, you get the following bell shaped curve (sorry it ain't prettier, but I've got work to do):
__________________
Macavity, Macavity, there's no on like Macavity,
He's broken every human law, he breaks the law of gravity. - T.S. Eliot, Old Possum's Book of Practical Cats
|