http://www.youtube.com/watch?v=_4UMOsfV1Gc
http://www.youtube.com/watch?v=PHnA8wMwhNg
http://www.youtube.com/watch?v=FHqVfJgDfBs
http://www.youtube.com/watch?v=iVGhrxf-Fgc
http://www.youtube.com/watch?v=WlOq7rG-wIs

These are the links to what I captured as my friends results on his Clairvoyance Test!
Please Write me and tell us what you think
We're SO EXCITED ) <3

thats pretty cool stuff. congrats to your friend!

Then could we have the final tallies, please? Just X hits in Y trials, where Y includes all the trials, whether they were recorded on video or not. Better yet would be the Y results, one by one. Same comment about Y applies.

These results were not achieved at the same time. When he got an exceptionally high score, I recorded it. And he also got the results of 7 out of 12.
Now as I said earlier, you can slice it any way in the world that you like, but getting results like that is not normal, by any means
If you want final tallies, well, if you watch the videos, you will see that he has two 11/25 scores, (and if I upload the still picture, it'll be three)
and a 0/25, all three scores displaying significant psi ability . I can also post his high scores from past recordings if you'd like

6 out of 11 I captured once the significance of receiving such a score. It was 2.85, which means significant psi ability
6 out of 12
10 out of 25
5 out of 10
10 out of 25
5 out of 10
5 out of 10
5 out of 8
3 out of 50
your results indicate psi missing
I.E. your score is significantly LESS than would be expected by chance. paradoxically this is itself evidence of ESP since it may be argued that to consistently MISS the targets, some unconscious knowledge of them must be present.
5 out of 9
10 out of 25

oh, and by he way...we found one he did awhile ago where his scores were tallied like you wanted
8 out 25
5 out 25
7 out of 25
8 out of 25
7 out of 25
35 out of 125
Chances of that happening by chance: 1 in 29
Percentile chance of that happening by luck: 0.05% chance

(edited a bit to reflect that some of my concerns were covered in a PM I saw after posting... and later I noticed that I corrected for something twice, and so fixed that)

If he got two sets with high marks and one with low marks, then overall he got what he got, 22 out of 75.

Round numbers, about one in 20 people would do that well or better against a truly uniform random selection of cards without replacement from an "infinite shoe."

It is boderline, and to know which side of the 5% probability cutoff it falls on (the most meagre criterion for "statistical significance" that anybody would begin to take seriously), I'd need to know a few more details of how the site works. (Definitely not uniform random without replacement from an infinite shoe, so the computed figure from the simple model will be a bit off)

(The "5% cutoff" used here reflects that both high scores and also low scores "win." I have a problem with deciding this after the results have been viewed, since that invalidates one of the assumptions of the calculation.)

The 5 older sets (35 out of 125) are more clearly over the 5% line (not the .05% line, which is 5 in 10,000 not 5 in 100). I'm assuming, of course, that this record of 5 sets does not survive because it is "good" while others were lost because they were less "good."

I cannot undertsand the rest of the post. Why are there "such-and-such out of 11" entries? There are 25 trials in a set. What happened to the other 14 in that set?


You need to record every score. If this is just the times when he won, then there is no point doing an analysis to see if his good showing could plausibly be by chance. It isn't. You only recorded the good scores.

You definitely want to Google "cherry picking" statistics for guidance on some of these issues.

What you report is tantalizingly better than chance, but the methods strongly suggest both selective reporting and also making up the criterion for success after you have seen the outcome.

(EV may have some further testing-related posts, however, and so it is best to stop here and turn things back to him).

Did you not get the email that I sent you eight bits? You wanted tallies, so i sent you the youtube with my friend's scores that did 10 runs of 25 cards.
63 right out of 250. There was a 1 out of 20 that would happen by chance. With significance probability being 0.05. Now, according the psi test calculator which I will send you a link to, the chances of getting 35 out of 125 is 1 in 29 chance, again 0.05. As far as your other comments, I have two rebuttals
If you watch the video, you will see (or if you want me to write them down, as the camera is sort of blurry) that there were times when he got 3 and 4 out of 25 and scores like that. I was by no means hiding the low scores. and about cherry picking, I will say this again. We both seem to be in an agreement that by chance and chance alone, a person is suppose to get 5 out of 25. When a person, who by the time they only get to 12, gets seven correct, or gets 5 out of 7 correct, I had to record that data because by the time a person got to 12, by chance the most they should have gotten correctly was 2 or 3 at best,
and out of 7, the most they should have gotten correct is 1 or 2 at best. that is why i recorded those results. And no matter if he got 11 out of 25 twice in a row, the odds of that happening even once is 1 in 100 or 0.001. They boy did it three times in three recordings. If you want more info, I'll talk to him some more about it.

Now it's your turn to check your PM's .

http://www.psychic-experiences.com/psychic...zener-cards.php
http://www.mdani.demon.co.uk/para/esp2.htm
These are the sites where I toolk my test

You should have taken the test I had put up.

I held a test hear to test clairvoyance. Some did bad and some did good proving my prediction was correct that the results were random.

If I again or another member creates another similar test for clairvoyance, drop in the thread an take it.

which test was that, (and by the way, my friend was the one with the Gift, not me...though I wish it were)
I might look around to see if I could find the old one for my friend, Though he generally prefers to do Zener Card Tests or jotting down his predictions....I'll talk it over with him.

The test was made for Krillanisavillan, the test used however can no longer be used in any reputable test because the answers have been made public. I would be more then happy to formulate another for your friend with the same rules.

P.S. The test is in the metaphysics board here.

I've talked it over with my friend...and he's decided to take you up on your offer! He's not claiming how much he's going to get right, as there are many times when he got as average as 20 out of 100, and those times when he has gotten as good as 18 out of 50. He told me that if he succeeds, he succeeds, and if he doesn't, then he doesn't. the only thing he asks for is what is the statistics in terms of probability correct.
Thanks and we nervously await your response.

Hi.

If you wish to do the 30 question, 4 option test, I wonder if you would be willing to proceed on the basis that only a high number of hits is a win. (Rather than either a high number or a low number is a win).

Either way, the goal would need to be declared in advance for any of the usual, single-valued significance tests to be applied.

And BTW, if your friend does not score a significant success in this test, with whichever goal, that would not be validly interpreted as evidence against his abilities, since another explanation of failure to achieve significant results might be that there are too few trials.

If the number of trials is "large" then that would be one thing, but 30 is small. This was plenty good enough to test an extravagant claim like 80% right, but wouldn't "detect" an ability that achieved 35% right as opposed to the 25% chance level (based on quick, approximate "back of envelope" figuring).

Finally, I trust everyone is in agreement that "80%" as a criterion was based only on what another poster claimed, and has no statistical generality whatsoever, and so has nothing to do with the current proposed test.

no problem, my friend has no problem about saying that only a high score counts.
has AtheistGod finished with making the test?

Crap, youtube's not working for me right now. I'll try to watch the vids again later.

I did a telepathy test, where my friends would think of a number 1-10..And I had to guess it..I got 6/10...And the odds of that happenin jus by chance are 1 in 7257 or soemtin like that...

The odds of getting 0/10 or 10/10 is no less likely then the other.

I have proven somewhat with my test that the process is completely random.

3 tests 10 questions each and 4 possible right answers. There was one person who got 7/10 correct and one person who scored 6/10 on 2 without trying. As I recall you took my test and didn't beat the odds.

The odds of guessing a correct number between 1-10 is 10/1. So where you got 1-7257 is beyond me...

The odds of getting the numbers correct for each set of 10 does not change.

The results of your said test were still random.

Each set carries 10-1 odds. Your chances for each set is 10% the only point where the odds would have been 1 in 7257 would have been if you in fact had to guess the number out of 7257.



I'll get around to it soon the new test will be posted within a couple of days.

The difficulty with H's test was his obvious lack of control for ordinary exchange of information, rather than anything statistical.

To get 6 (or more, the usual test statistic) right with n=10 and p= 0.10 is unlikely to happen by chance (p < .001). Thus, his results were in all likelihood not due to chance, which is what a statistical test informs you about.

What the results were due to instead of chance is not a statistical question, but a matter of what mechanisms could be operating, based on the experimental design. The design was inappropriate for evidentiary use in H's case, fine for social bonding with his buddies, but worthless for establishing any paranormal quality to his abilities.

Turning to the concerns of this thread, I was unable to find out much about the pseudorandom number generators used on the two Zener-card test sites posted. There may be some possibility of contacting the operator of

http://www.mdani.demon.co.uk/para/esp2.htm

by email. He seems to have some sophistication, and I think it is a good sign that a named individual stands prominently behind the software. I am less sure about the other site, but for all I know, they are peachy nifty.

I cranked out the cumulative binomial for n=30, p=0.25, the distribution which describes "chance" performance for AG's test. I can post that whenever it's timely, and it's surely on the web somewhere, too.

if you could post statistics, that'd be great. The website that my friend tested his abilities on that you said had some sophistication also has a psi test calculator, being able to calculate the odds of getting such and such correct....the only problem is that you need at least 50 trials to calculate the odds. But thanks for trying to find the answers. thanks so much for everything

Significance tests are based on the probability of doing at least as well as the observed result if guessing uniformly at random.

Ordinarily, satisfying a "5%" significance level requires you to beat 5%. However, because of the large gap between the probabilities of successive possible outcomes, I believe it is reasonable to consider 12 or more hits "significant" even though the probability of that is slightly higher than 5%.

Five percent is a conventional figure anyway, and as it is, this test has only about a 1 in 3 chance of succeeding if the subject has the ability consistently to achieve a 35% hit rate on average. If the bar were set at 13 hits, then the odds would be almost 4:1 against such a subject.

If guessing uniformly at random,
the probability is . . . of getting at least

(percent . . . p = 0.25, n = 30)

5.066 . . . 12
2.159 . . . 13
0.818 . . . 14
0.275 . . . 15
0.082 . . . 16
0.022 . . . 17
0.005 . . . 18
0.001 . . . 19

and vanishing beyond that (as if 1/1000 of 1% wasn't)

I checked my table built on Excel against lower (more usual) precision tables at

http://www.cardiff.ac.uk/maths/teaching/pr...l%20Tables.html

for reasonableness. Note that the probability of getting k or more correct is
one minus the value tabulated there for k-1.