Bungee Jumper

There is another thread he started about statistics (it may be on page 2 of the board already). Maybe it's there. But he did say that according to a researcher named Weiss, the cause of regression toward the mean was error...

And myself. Physics Review E for one...that would be the volume concerned with statistical physics...

But you can use it to determine the error of a single die roll!

The statistics discussion ("Err America") on PPP is a riot. Holcomb's Arm is actually insisting that a single die has a "true value" of three and a half. Until that, though...seeing BF argue with a chef about the proper way of cooking refried vomit in the retatta thread was classic...

I honestly think BF would have figured all this out by now. Seriously, have I really been that unclear in explaining regression toward the mean? I thought my two dice example was pretty !@#$ing clear, myself...

True enough...but like you said, it's a nitpick. I'm 76 inches tall, Nervous Guy's 76.003124 inches tall. But in the context of this discussion...we're both 76 inches tall, simply because that's how the data's collected and portrayed by anyone trying to determine the distribution of height among the population (if you'll pardon that phrasing, I know it's awkward). That's also why I used the two dice example before...it is a definitively discrete phenomenon that displays no inaccuracy, no error, and the binomial distribution of possibile values is a reasonable approximation of a normal distribution for the purpose of explaining regression toward the mean. Population height can be treated as discrete...but like you said, it isn't...and now you've just introduced the fact that height measurement is in fact only possible to a certain degree of precision, which is going to make Mr. Potato Head say "See, there's error! I was right!"

THE MEAN OF THE NORMALLY DISTRIBUTED ERROR, YOU !@#$ING MORON!!!!

The problem is, even though the average value of a multitude of die rolls (or even the faces of a single die) is 3.5...it doesn't really mean anything. It's an average of a discretely valued system that doesn't actually exist in the set of values of the system. So who cares? It's not a meaningful characterization of any property of the system - it is a characterization, mind you, just not a meaningful one. It sure as hell doesn't mean the die has a "true value" of 3.5, like HA is trying to say.

Isn't this a !@#$ing riot?

Or a die...which you've already said exhibits regression toward the mean because of error. This all just proves you can't define "measurement" or "error", either. But to take your height example...there is no regression toward the mean when you measure the height of the same person twice, because it's a discrete, exact value. Not probabilistic. However, when two very tall (or very short) people have kids, it is likely (i.e. "it is probable", "there is a probability greater than 50%") that the kids will be closer to average height than the parents. And it doesn't mean the parents or children are the wrong height. It's strictly because of the frequency distribution of people's height in the population, which is directly related to probability. Got it now? Is your warped little idiot mind starting to twig to the difference between "error" and "probability" yet?

831830[/snapback] AND IT'S NOT BECAUSE OF ERROR, IT'S BECAUSE OF THE PROBABILITY DISTRIUBTION, YOU IDIOT!!!! Jesus Christ...

Hey, women don't have rights. Fetal collections of cellular material do...at least until they grow up to be women who want to have babies. Live with it.

No, the fact that I'm "afraid" (sic) to answer the question suggests that I'm aware you need a basic education in statistics before you can comprehend the answer. The answer, by the way, is neither.

Given the way local politics is run and the cultural history, I think any one of these would be appropriate.

We're just all wrong. It's that simple.

That marvelous die analysis he posted last night actually caused me physical injury. Literally. I laughed so hard, I fell out of my chair and sprained my wrist when I landed.

It's not my birthday.

I'll tell you what, then...send your example to the writers of the article. See what they think.

No, a die roll is NEVER consistent with a normally distributed data set...because a die roll is not normally distributed. A pair of dice, however, is a reasonable approximation...and is the example I used. And a pair of dice does not have error either. As I said many posts ago: you roll an 11, your next roll will tend to regress toward the mean because of probability: the distribution of possible values is such that there's three chances to roll an 11 or better, but 33 to roll less than 11. You have an 11 in 12 chance of getting a roll closer to the mean value of 7 than otherwise. So let's say you roll a...oh, let's say 9, for the sake of argument. That does not mean your error reduced from 4 to 2...it means you rolled a more likely value than 11 or 12. That is the difference between probability and error. That is what causes regression toward the mean. Then you shouldn't have introduced the blithering example of a single die. I think, before I answer such a complicated question, it would be better if you understood the basic statistical concepts embodied in a more simple system like a pair of dice.

No, I didn't. I challenged you to explain dice. You couldn't even do that. Seriously...what part of "error and chance are two completely different things" are you having trouble with? What? The kids score closer to the mean because their parents' scores were wrong? Does that actually make sense to anyone?

And if they don't roll double 3.5's, we'll know the game's rigged, because the dice were wrong! Brilliant!

Trust me, we understand it. That's why it's so !@#$ing funny.

How the !@#$ can you expect to get a value that doesn't exist in the system?????? Let me make this as clear as I can: A die has six faces. They are numbered: 1,2,3,4,5,6. No face is numbered three and a half. You can never EXPECT to roll three and a half. You cannot derive an "expectation" of some event that doesn't exist from a "conglomerate" of expectations that do exist. It's akin to saying the "expected value" of a coin flip is "hails" or "teads", because it's a "conglomeration" of heads and tails. And since when is rolling a die an attempt to measure an average value of the die? Where the !@#$ do you get this sh-- from? What, exactly, is your !@#$ing problem with this concept?

Can't do it. I already measured the bet as $20, so your bet is wrong by $10.50. But maybe if you bet again, it'll regress toward $20.

But if it's more reliable than average, the reliability's wrong anyway...