Test Statistics

Skeptics just like to say that I failed to get 100% on a test and therefore the claim is falsified. My experience with the claim says there's more to it than that, and my training in the science of chemistry says that physical phenomena are not just 100% or 0%.

Since I lack any training in statistics (and find the subject a bit too boring or intimidating to pick up a book and teach myself) I decided to play it with a die and see what kind of statistical behavior shows up. I went to this online die generator and simply did a whole lot of trials. I do know from statistics, that random fluctuations do occur, and that a larger number of trials reduces the random error and approaches a truer value. So I tried to do as many as I could.

The Seed Test

I love the Seed Test I had, because it proves that statistics does work. The Seed Test tested me on something I could not do. Statistics predicted that with a 1 in 2 chance of guessing on the correct answer, and by pure random guess alone, I would get half of my trials correct. And I did. I got 7 out of 15 correct, and 8 out of 15 incorrect. So we can trust statistics, statistics is not broken in any way.

IIG Test

Case 1: two persons out of three The first thing we can look at concerning the IIG test is that I had two people out of three correct. Each trial has six people and one of them is the correct person. I have no idea how to do the statistics. I use a six-sided die (one in six people), one die, and three times at once, pretending that "1" is a correct answer, and count the number of "1"'s:

Set 1
Total number of trials: 137
3 correct: 0 -- 0%
2 correct: 9 -- 6.6%
1 correct: 57 -- 41.6%
0 correct: 71 -- 51.8%

Set 2
Total number of trials: 120
3 correct: 0 -- 0%
2 correct: 8 -- 6.7%
1 correct: 47 -- 39.2%
0 correct: 65 -- 54.2%

Combined
Total number of trials: 257
3 correct: 0 -- 0%
2 correct: 17 -- 6.6%
1 correct: 104 -- 40.5%
0 correct: 136 -- 52.9%

The results indicate that it should be a 6.6% chance to achieve two out of three persons correctly just by guessing, and that it is a 93.4% chance to get only one correct or none correct. With these numbers we can neglect 3 correct as if it were unachievable, as it did not appear once during these 257 trials with the die.

Statistics would want me to get less than 2 correct. Of course, about 6 or 7 people out of 100 people guessing would get 2 correct. But still, I *should* have been one of those who gets only one or none right, because statistics says so.

Case 2: omitting the first trial The Seed Test was conducted with and planned by a real scientist. A scientist agrees that individual trials can be discarded if they do not produce the experience of the phenomenon that is under study. We are analysing for the accuracy, not frequency, of perceptions. I am asked to be accurate, but whether I have a perception in everyone or always is beside the point - who cares.

In the IIG test I knew that I had not made a confident perception in trial 1. It should therefore be omitted. I complained about trial 1 during the entire 10 minute break after that trial and my complaints were by no means ad hoc (ie. after the fact). Meanwhile I was utterly confident in my choice of person in trials 2 and 3. If we do a six-sided die, the results are the following;

Set 1
Total number of trials: 95
2 correct: 3 -- 3.2%
1 correct: 31 -- 32.6%
0 correct: 61 -- 64.2%

Set 2
Total number of trials: 81
2 correct: 4 -- 4.9%
1 correct: 22 -- 27.2%
0 correct: 55 -- 67.9%

Set 3
Total number of trials: 145
2 correct: 5 -- 3.4%
1 correct: 45 -- 31.0%
0 correct: 95 -- 65.5%

Set 4
Total number of trials: 99
2 correct: 2 -- 2.0%
1 correct: 25 -- 25.3%
0 correct: 72 -- 72.7%

Combined
Total number of trials: 420
2 correct: 14 -- 3.3%
1 correct: 123 -- 29.3%
0 correct: 283 -- 67.4%

Statistics says that only about 3 out of 100 people would get two persons out of two trials correct by guessing. Statistics says that I am 96.7% likely to get one or none correct. My results, although 3% likely to occur, are not doing exactly what statistics would want them to do.

For those overly skeptical persons (who are not scientists) who think that giving the claimant the right to omit certain trials, would somehow skew the results in their favor, you are simply mistaken. In the seed test, I omitted a whopping 50% of the trials. But I still achieved a total score on the graded trials which aligned perfectly with statistical prediction. And since my choice to omit trial 1 in the IIG test was not an ad hoc decision, it is a valid decision to make. It is not cheating. Skeptics might think that it's cheating, because in this case it does make my overall score better. But that's because I think there might be something to my claim. Omitting trials does not give better score, unless there is an ability behind which trials were omitted and which ones were kept.

I should have gotten no correct persons. But I got two. Statistics doesn't predict that. Ok there is a small chance, about 3 in 100. That is why repeat trials are needed. If my score was due to random error then additional trials will reveal it as such, or reveal that I have an ability which is as good as this.

Other tests On the TAM test, I saw a kidney in all but two out of ten kidney spaces, and one of those two was missing a kidney. In a reading with Dr. Carlson I saw that he was missing a left kidney. In another reading with a lady skeptic I saw that she was missing her uterus. In Michael Shermer I saw and accurately described that he has the Hepatitis C virus and described other aspects of him that he says were perfectly accurate and only a close friend would have known. Plus case after case, without confirmation bias, without ad hoc. It all adds up to something which statistics is saying should not be happening.

/