New telepathy test, the sequel.

Michel H said:
For the first of these tests, the hit rate for the participants who had complied with the protocol was 2/3=67% (only 3 valid answers were given
Click to expand...


What is the statistical significance of 2 out of 3 correct guesses on a 25% chance quiz? With what degree of confidence (as the term is used in statistics) can we say that 2 out of 3 correct guesses shows a definite non-chance cause?

I don't think you can answer these questions, as your statistics knowledge seems to stop at punching numbers into the first online calculator that shows up in a google search. Still, I hope somebody who knows number stuff will answer.
 
Loss Leader said:
What is the statistical significance of 2 out of 3 correct guesses on a 25% chance quiz?
Click to expand...
One can calculate a p-value, using (for example) this online binomial calculator: https://stattrek.com/online-calculator/binomial.aspx.

Input:
Probability of success on a single trial=0.25
Number of trials=3
Number of successes=2
Output: p=0.15625
This is the probability that, if random chance only is responsible for the results, the hit rate will be larger or equal to 2/3. One often admits that, if p is equal to, or smaller than 5%, the experiment is statistically significant.
For typical analysis, using the standard α = 0.05 cutoff, the null hypothesis is rejected when p < .05 and not rejected when p > .05.
Click to expand...
(https://en.wikipedia.org/wiki/P-value).

Because p=16% here, and 16% is larger than 5%, statistical significance has not been achieved in this small test alone, the sample is too small. One needs to combine with other tests or experiments to achieve statistical significance.

Note that, even if statistical significance has not been achieved, this does not mean that a hit rate of 67% is "worthless".

Note: error correction for post #1699:
(only 3 valid answers were given, this can be compared with 22 valid numerical tests in the previous test)
should be replaced by:
(only 3 valid answers were given, this may be compared with 22 valid numerical answers in the previous test)
 
Last edited:
Michel H said:
...in this small test alone, the sample is too small.
Click to expand...

Correct, and that has consequences that you don't get to just sweep under the rug.

One needs to combine with other tests or experiments to achieve statistical significance.
Click to expand...

Correct, and one of the questions you were challenged to answer was how many trials it would take to justify a hit-rate of 50% to within standard confidence. But since that's a screw-question and not a nail-question, you can't answer it with your online hammer. That means you have to demonstrate actual knowledge, not just bluster around.

Further, I discussed at length how your proposed method of combining trials and tests was invalid. You didn't bother to respond, so I conclude you really don't know how proper experiment design works for these kinds of experiments.
 
JayUtah said:
Nope, wrong answer. Clearly you have no idea what Loss Leader asked for. I do, and I computed the value he asked for. (Normally I don't do math on weekends.) Try again.
Click to expand...


Well, would you PM me? I really want to know.
 
Michel H said:
I am not interested in trying to do one of these tests again.
Click to expand...

I appreciate your honest answer. As I'm sure you understand, this means no one here will ever believe you are telepathic. It also means no one with training in statistics will ever believe you are telepathic, although I don't know whether you understand that as well.

If you were telepathic, this would be a great loss to the world. Nonetheless, I wish you the best.
 
RoboTimbo said:
How many more blinded tests like that do you think you should do to have statistical significance?
Click to expand...
I believe the first blinded test achieved some degree of success (a hit rate of 67%), (only) because of some sympathy from members of this forum who participated, who knew I did this test, bowing to their demands.

However, I got only 3 valid answers in the two "blinded" tests (and no valid answer in the last one), while I got 35 in the first two tests on this forum.

In addition, in the first "blinded" test, the credibility method, developed and validated for "unblinded" tests (not the "blinded" ones) was also (almost) useless. For example, I had given a credibility rating of -5 for Ladewig, for his (correct) answer:
I concentrated on the assigned task and an image of a brightly lit piece of paper appeared, the circled number on it was 2.
Click to expand...

In retrospect, I could perhaps have given him a better credibility, but I had not seen good answers with such comments in the "unblinded" tests.

So, I don't think the blinded method is good, in spite of seeming to be a darling of the members of this forum.

I could perhaps invite the readers and posters of this thread to read again the analysis of a large unblinded test: http://www.internationalskeptics.com/forums/showthread.php?postid=9516155#post9516155
and to reflect a little about the "universality" of the credibility assessments, which is the fact that all reasonable persons can agree that some answers are credible, and some are not (and agree on the list of the credible ones).

This is the method which seems to have a real future, and that you should (in my opinion) study if you have a real interest in telepathy.
 
Last edited:
Michel H said:
However, I got only 3 valid answers in the two "blinded" tests (and no valid answer in the last one), while I got 35 in the first two tests on this forum.
...
So, I don't think the blinded method is good, in spite of seeming to be a darling of the members of this forum.
Click to expand...

It's the "darling" of all experimental science. Looking at the answers and then deciding based on whimsical ad hoc criteria which answers are "credible" is a blatant show-stopper in real science. So comically so that this is why you get no serious answers. Everyone can plainly see that you're cheating.

The number of answers you get is not connected methodologically in any way with your subjective culling of data. It's your responsibility to proceed only when you get enough subjects. No one who is even remotely serious about science is going to accept your subjective non-blind data selection.

I could perhaps invite the readers and posters of this thread to read again...
Click to expand...

No. Simply repeating your beggardly claims to accept your method as valid will not work. Can I perhaps invite you to respond to the lengthy analysis I did previously of your method? Or are you going to continue to pretend it doesn't exist?
 
JayUtah said:
It is thorough. And somehow you missed it. How could you miss such a long post?
Click to expand...


In all fairness, most of the time I see one of your long posts, I tend to skip it. I know most of it is going to go way over my head. I still have nightmares about the Moon Landing threads.
 
JayUtah said:
It's the "darling" of all experimental science.
Click to expand...
It beggars belief that anyone who claims to have a PhD in physics does not know that, and understand exactly why it is the case. Cognitive biases were the reason why the scientific method had to be invented, and blinding is the most useful tool in its toolbox.
 
Pixel42 said:
It beggars belief that anyone who claims to have a PhD in physics does not know that, and understand exactly why it is the case.
Click to expand...

Exactly. This is Intro to Science 101.

Cognitive biases were the reason why the scientific method had to be invented, and blinding is the most useful tool in its toolbox.
Click to expand...

More so in the social sciences than in the more "dispassionate" sciences that don't involve human subjects.
 
Loss Leader said:
In all fairness, most of the time I see one of your long posts, I tend to skip it. I know most of it is going to go way over my head.
Click to expand...

That's fair. But you don't have a PhD in physics, so I don't expect it to resonate with you. Just so you know, there was no actual math in the post. That I recall.

Conversely, I just grabbed a Constitutional Law textbook, turned to a random page, and read it. I got maybe a quarter of the content. The rest is stuff that I presume gets learned prior to that page, or perhaps in other classes. In your field of expertise, I'm content reading the briefs and the summaries. I don't read the whole docket and depositions. But then again, you don't expect me to be a competent lawyer.
 
JayUtah said:
More so in the social sciences than in the more "dispassionate" sciences that don't involve human subjects.
Click to expand...
True, but all experimental science involves at least one human being - the experimenter, who needs to ensure he's not seeing patterns that aren't really there.
 
JayUtah said:
I don't read the whole docket and depositions. But then again, you don't expect me to be a competent lawyer.
Click to expand...


Depositions are exceedingly rarely published, even in the age of computerized court filings. Lawyers can usually only do research based on decisions, orders, and judgments. Strangely connected to this topic, they do demonstrate confirmation bias. The judge (or justice) only puts in the decision the facts of the case that uphold the legal point being made - even though, ideally, the legal point arises from the facts of the case. One of the most famous cases in tort law had markedly different facts from those recited by the court.
 
Loss Leader said:
One of the most famous cases in tort law had markedly different facts from those recited by the court.
Click to expand...

I know this case, which I guess follows from it being famous, and slightly absurd. And also because a surprising number of landmark tort cases seem to involve railroads. Isn't there also one where a guy gets hit by a train and his flying torso takes out some old lady? I mean, this thread is about the probability of silly events, right?
 
Last edited:
You must log in or register to reply here.

Back
Top Bottom