#11
|
|||
|
|||
Re: Site bias test
You hope bad players keeping getting in as dogs and winning? Yeah, that makes perfect sense.
|
#12
|
|||
|
|||
Re: Site bias test
[ QUOTE ]
You hope bad players keeping getting in as dogs and winning? Yeah, that makes perfect sense. [/ QUOTE ] Ha! Uh, dude, wake up...I hope bad players keep getting in as dogs! We both know that how much they win is irrelvant since statistically they'll win and lose how much they should. But, when they win as a dog, they remember , and continue getting in as a dog feeding the better players more chips. I'm sure you're the only person on here that didn't get that. [img]/images/graemlins/blush.gif[/img] Class dismissed. |
#13
|
|||
|
|||
Re: Site bias test
[ QUOTE ]
We both know that how much they win is irrelvant since statistically they'll win and lose how much they should. [/ QUOTE ] So easy to say, so damn difficult to prove. |
#14
|
|||
|
|||
Re: Site bias test
It cannot be proven. We can't "prove" the status quo, we can only try to find enough evidence to disprove it. (As your "experiment" is attempting to do.) If we can't find enough evidence, that doesn't mean that the status quo is true, it only means that we didn't have enough evidence to say that it wasn't!
|
#15
|
|||
|
|||
Re: Site bias test
I think you're the only idiot here, "dude". Nothing is wrong with people getting in as dogs and LOSING, but that is not what your post said. If they keep getting in as dogs AND WINNING, you lose all your money. Class is same place, same time tomorrow.
|
#16
|
|||
|
|||
Re: Site bias test
[ QUOTE ]
I have only just begun to gather data. Please comment on the experimental design and possibilities for data analysis. [/ QUOTE ] There's nothing wrong with the design, except that it's based on a very specific theory. If you're going to gather the data, it makes sense to collect information everytime you are all-in heads up with one other player. You can compute the probability of your winning, and do a statistical test for whether you win less or more than expected. You can do subtests for things like fresh faces or other factors. Starting off by looking for one specific effect can lead to tunnel vision. Say for example that your probability of winning is uniformly distributed between 40% and 65% (I assume you're better than average). After 100 hands, the standard deviation of your winning percentage will be about 5%; so you could detect gross cheating, say you won only 40% or less, but not subtle cheating. You also couldn't tell the difference among subsets, like fresh faces. After 1,000 hands, you could detect subtle divergences from random, and also differences among subpopulations. |
|
|