Reply #119: You might be surprised [View All]

[zipplewrath]

You said "tending", and that's the key word.

My sister, being the daughter of a PhD, came up with a simple "science fair" experiment along these lines. Basically she rolled dice... ALOT. Actually, we all did since we all got involved before it was all over. Red die, blue die, roll 'em, write down the results. We literally did THOUSANDS of rolls, and the data was constantly moving around, never really wanting to fall into the predicted categories. The were "approaching" them, but one of the numbers, or one of the die would constantly be more outta whack than the others.

Years later, in college, a professor of mine would write a paper addressing this a bit. It was basically an attempt to identify if samples were "truly" random. It wasn't a proof methodology as much as an error detection methodology. i.e. it didn't prove that data sets WERE random, it tried to detect if they potentially WEREN'T random. But what he developed was a set of guidelines (based upon sample size) for finding REGULAR distributions of data, i.e. apparently highly ordered or repetative. In a "truly" random set, there should be "runs" of numbers in order (10, 20, 30, 40..., or 1,2,1,2,1,2). the length and frequency was the point of the paper. Look for them and if you don't find them, somethings probably wrong. Years later I read that forensic accounts do something similar in which the just take all the digits used in accounting data and do a histogram. There should be a roughly "even" distribution of digits in a large report. If there isn't someone's cooked the books.

In your example, it wouldn't be surprising if there was a deviation by a couple hundred dollars either way. It wouldn't be "common" but it wouldn't be out of the relm of reality either.

And again, you're trying to extrapolate a narrow, limited variable, result to a system with large numbers of variables (potentially infinite) and say that because randomness exists in the limited case, it must exist in the larger case as well. Don't mistake me, I know randomness exists, and I'm not much of a believer in Karma. I merely make the case that proving them wrong is difficult.