jzs,
Not trivial, as we are talking about VARIANCE here, stimpy. Morevoer, the incriminating part is, is that even after I pointed out your error, you still didn't believe me.
You did not point out the error. You just said that I was wrong. Why should any of us simply believe each other's unsubstantiated claims? If you want me to believe you when you say that I have made an error, then you need to explain what the error was, and demonstrate that it was, in fact, and error.
You seem to be completely incapable of doing this.
You said
(bold mine)
"I am simply pointing out (1) that if there is any correlation in the bits, then mean of the XORed data may still be biased, and (2) that even if there is not, the initial bias in the mean will result in correlations in the XORed data. "
Your critique, for it to have any serious impact, hinges on their being correlation in the bits to cause the bias, and the initial bias in the mean. Your own words above show that. If these are small, your critique falls flat.
This has been explained to you several times already. My argument is not that if such biases are present, their analysis is invalid. My argument is that unless they demonstrate that such biases are
not present (which they have not done), their analysis is meaningless, since positive results could simply be due to those biases.
This is really basic stuff, which anybody who has ever done any kind of data analysis should be familiar with.
If you want to not present any measure of these, fine, but don't expect your merely raising the possibility of these things existing to be the same as exposing a serious flaw in their analysis.
You've got it backwards.
It is the PEAR people who are "merely raising the possibility" that such biases are
not present. That
is the flaw in their analysis. I am sorry that you do not understand this. Why should we just accept on faith that such biases are not present? If we do not, then we cannot conclude from their results that anything unusual, or even interesting, has happened. Only by
assuming that the REGs will, in the absence of external influences, be sufficiently unbiased, could we possibly conclude anything from their results.
I am not prepared to make such an assumption. Why do you think that we should?
I guess you finally understood that bit about the variance?
You really have no idea who you are talking to, do you?
I am a physicist. I do experimental data analysis for a living. I have several papers published in pear reviewed physics journals on the subject, and have given invited lectures on the subject at major physics conferences.
I "understand about the variance" quite well, thank you very much. But I am sure you would rather believe that I don't, and that my error was not a simple arithmetic error, but instead the result of me not understanding how variance works. You seem to be quite adept at believing whatever you want to believe is true, regardless of any evidence to the contrary.
Back on ignore for stimpee.
Boo hoo. It breaks my heart to know that you will now be passively ignoring my posts by having the board hide them from you, rather than actively ignoring them by simply refusing to respond to my arguments.
Please, do keep me on ignore. I am not posting for your benefit anyway. That would be a complete waste of time, since you have neither the basic understanding of statistical analysis needed to get anything from them, nor any interest in learning.
Dr. Stupid
But he still was trying to 'get' me, and wouldn't listen to me. His mindset makes this an exercise in futility.
I guess you finally understood that bit about the variance?