Sorry, but your analysis with respect to Skepdic is completely off-point. Obviously, if there was a bias in the PEAR tests, the results would be inaccurate. But that's not what Robert Carroll of Skepdic says; rather, he says that "statistical significance does not imply IMPORTANCE." (emphasis added.) Try as you might, you can't get around the fact that Carroll thinks that, even if PEAR's tests were done perfectly, the statistically significant results aren't important!There are a couple of things to consider here. First of all (I realize that I am repeating myself), "statistical significance" is not a measure of the truth of an idea. So dismissing a statistically significant result does not mean that you are dismissing an idea that might be true. I have already explained to you numerous times why this is the case.
Second (I realize that this is also a repeat), bias (a tendency to create measurements in a particular direction that is not representative of the population under study) is very difficult to eliminate completely. What we can do reasonably successfully is eliminate most sources of bias, so that we can say at the end "any source of bias, if present, would have to be miniscule and therefore cannot account for an effect of this size". However, the effect that we are talking about for PEAR is miniscule. Which means that it would be very easy for any residual bias in the experiment to completely account for the effect.
You have to remember that this data was based on a machine that was supposedly able to generate random numbers. Over the years, various ways have been developed to simulate the process of generating random numbers and we test whether or not they are successful using a Goodness of Fit test. However, these tests are fairly crude and will miss moderate deviations from true randomness. What the researchers did was make a machine, do a rough test to see if it was grossly different from true randomness (no), generate millions of outputs, and then apply an especially sensitive test to see if those outputs were different from true randomness (yes). All they really proved is that their machine was not a perfect random number generator. But what they (and you) wanted to conclude was that it was not perfect because of the existence of psi. What skeptics point out is that maybe it just wasn't perfect because machines are rarely (if ever) perfect.
So I agree with Skepdic that one should favour highly probable events as a explanation over highly improbable events.
And this is not the first time Carroll has shown his ignorance; skepdic.com also maintains that Edgar Cayce "was the first to recommend laetrile as a cancer cure." See http://www.skepdic.com/cayce.html (For those who don't know, laetrile was not synthesized until 1952 -- seven years after Cayce's death. See http://www.thecrimson.com/article.aspx?ref=346523)
Now, with respect to your altogether different contention that PEAR's tests WERE biased, what is your evidence that PEAR's random number generator produced results that favored the operators' intentions?
wouldn't it be hard to reach the keyboard? 