Parapsychology: Science or Pseudoscience?

No. To an experienced data analyst, it's blatantly obvious that he was picking among alternative data analyses to support whatever hypotheses he had, prior or not, and this is supported by Francis's statistical analysis, an analysis that Andrew Gelman criticized as being superfluous in light of what is so obvious to the naked eye.

Sorry, but just saying "it's obvious" won't do. It was "obvious" to several of us that those presentiment averages weren't biased, but we were wrong.

Francis's statistical analysis may indicate something is amiss - thought I'm not sure how it could tell you exactly what was amiss - but I think it would need to be looked at carefully. It seems a bit ironic that it's essentially a matter of rejecting a psi hypothesis on the basis that the p value is less than 0.1! But on a more fundamental level, it rests on a statistical model of how psi might work. Perhaps the analysis is telling us something about that statistical model, rather than something about Daryl Bem?
 
Sorry, but just saying "it's obvious" won't do.

I can't add much to the explanations by Wagenmakers, Rouder, Alcott, and others.

Francis's statistical analysis may indicate something is amiss - thought I'm not sure how it could tell you exactly what was amiss - but I think it would need to be looked at carefully.

Francis's analysis shows that there were too many successful hypothesis tests (ie, rejections of null hypotheses) given the statistical power of each reported study to detect the reported (or combined) effect size, if all studies conducted were fully reported, and the tests performed according to the assumptions of classical hypothesis testing, namely, that the sample sizes and hypotheses were predetermined, and that a single test of each hypothesis was designed in advanced and performed exactly as designed. In other words, that there was no publication bias, p-hacking, undisclosed exploration, multiple testing, or exploiting researcher degrees of freedom. Francis's analysis is strong evidence that at least one of these requirements for valid statistical inference was violated.

It seems a bit ironic that it's essentially a matter of rejecting a psi hypothesis on the basis that the p value is less than 0.1!

Francis's test doesn't reject the psi hypothesis; it indicates that the experimental results aren't valid evidence for the hypothesis. That is, it's the evidence that Francis's results say we should reject, not the hypothesis.

But on a more fundamental level, it rests on a statistical model of how psi might work.

Huh? It has nothing to do with how psi might work. It depends only on the statistical power of the experiments to detect the reported effect sizes.
 
I don't think Linda should be allowed to sockpuppet without full disclosure of the fact.

It must be nice not having to be honest...
And even having the audacity to change gender.
 
I asked wether you believe "creation science" is a science or pseudoscience?

I know what you asked, and I answered in a manner that prevented you from going on a tangent by arguing that "both depend on belief", which was your apparent intention. But, since you went there anyways... I don't believe that anything dependent on pre-existing demarcations is really science.

The resemblance is that Kennedy attributes the unsustainability and evasiveness of psi as being guided by a "higher consciousness" (presumably God) do induce a sense of awe and wonder by deliberately baffling us.

Kennedy has stated that he had several "personal experiences" that support the existence of the paranormal, that is his personal stance and he is free to express it... But he is actually critical of the us vs. them approach, which sets a clear contrast between him and fundamentalism. Nevertheless, you did not answer how parapsychology as a whole could be realistically compared to what is essentially a lobby group.

The main difference between both is that "creation science's" methodology is dependent on a pre-set "truth", were as parapsychology has proven its willingness to adapt its methods and protocols in response to criticism. Furthermore, you seem to be lumping all researchers (and research areas) into a "whole" and asking us to label it one way or another... This generalization is possibly the biggest flaw in your argument. Sure, we have people that may come in with an interest to prove their own bias, but we also have several parapsychologists that came into the field as "debunkers" and lingered after getting surprising results.
 
"Trust" prior odds? What are you talking about? Prior odds is what you use to make rational judgments in light of new data. It's up to you whether you use prior odds or not. However, they are required for rational judgment about experimental results.

It's not rational if the prior is biased. Usually you get some very low prior because it is felt that psi contradicts what we know about science, but that really isn't a criticism of it and cannot be used to justify an extremely low prior to nullify a small to medium effect size. priors don't take into account pessimistic meta-induction, the fact that psi wasn't just an anamoly found in psychology research but a purportedly natural phenomenon with any thousands of years of history, with field research as well. It seems any significant scientific discovery would have a very low prior, so I am extremely skeptical of the validity of subjective priors.
 
Right, that was what Wiseman and Sheldrake were doing. But the idea that the dog was telepathic came from Pam's parents in the first place. If they were mistaken, was there any reason to think the dog was telepathic?

You still haven't addressed my point. All this is irrelevant to the fact that Wiseman chose, prior to the experiment, an invalid criteria for falsification.



Does that matter? Looking at the "no obvious reason" visits was just a way to reduce some of the noise in the experiment. But regardless of whether or not there was still some noise there, there wasn't any sort of alternative pattern which confirmed the parents' impression. All you can conclude is that Pam's parents were mistaking "noise" for telepathy.

Of course it matters. There could be a squirrel rustling in leaves outside that the dog hears and goes to check out and Wiseman with his relatively deaf human ears doesn't hear and then now rejects the entire trial even though after this the dog was in fact at the door much much more often when the owner was coming home.

The part I put in bold clearly shows you have not looked at the data.


fls said:
Wiseman demonstrated that Pam's parents were mistaken about whether the dog was telepathic. How is that "fraudulent"?

Linda

It's obvious that you haven't looked at the data. you clearly don't have a clue about this. I don't mean to be rude, but you are demonstrating an extremely agregious lack of knowledge for a subject you are making pronouncements of. Look at the data. If you don't see it then you are clearly too biased to carry on a rational discourse with.
 
"Trust" prior odds? What are you talking about? Prior odds is what you use to make rational judgments in light of new data. It's up to you whether you use prior odds or not. However, they are required for rational judgment about experimental results.



As opposed to just considering the experiment in isolation, which takes no other relevant information into account at all. That would sure work well.



http://link.springer.com/article/10.3758/s13423-012-0227-9

When I click on the "download pdf" button, I get the pdf, but that might be because I have access to it by an institutional subscription. If you can't download the paper, I can email you a copy if you PM me your email address.

Thank you for the link. I was able to get the pdf.
 
I can't add much to the explanations by Wagenmakers, Rouder, Alcott, and others.

I wasn't intending to discuss Bem's experiments, which is why I was talking about the autoganzfeld, but I find Utt's response to Wagenmakers much more convincing, so I also don't find that to be obvious in the direction you say.
 
You still haven't addressed my point. All this is irrelevant to the fact that Wiseman chose, prior to the experiment, an invalid criteria for falsification.

I don't know how else to address this. Wiseman chose the obvious criteria. What else could he have chosen that would lead you to believe the dog knew when Pam was on her way home?

Of course it matters. There could be a squirrel rustling in leaves outside that the dog hears and goes to check out and Wiseman with his relatively deaf human ears doesn't hear and then now rejects the entire trial even though after this the dog was in fact at the door much much more often when the owner was coming home.

But how would that indicate that the dog was telepathic and that there was something anomalous going on? The dog spent more time at the window the longer Pam was away, regardless of whether or not Pam was on her way home. So noticing that the dog was at the window didn't tell you whether Pam would show up in the next 5 minutes or the next hour. What's anomalous about that?

The part I put in bold clearly shows you have not looked at the data.

Okay, so presuming that you have looked at the data - what criteria could you use that would reliably indicate that Pam was going to be home in the next 10 minutes?

It's obvious that you haven't looked at the data. you clearly don't have a clue about this. I don't mean to be rude, but you are demonstrating an extremely agregious lack of knowledge for a subject you are making pronouncements of. Look at the data.

I've looked at the data in these experiments more than anybody else I know here. Over the years, I've pointed out stuff that nobody else has noticed (including Sheldrake). Just tell me what you see in the data that you think I don't see.

Linda
 
It's not rational if the prior is biased. Usually you get some very low prior because it is felt that psi contradicts what we know about science...

That is exactly why the prior probability of psi is low. To quote physicist Sean Carroll, if any psi phenomenon is true, then "the laws of physics that have been tested by an enormous number of rigorous and high-precision experiments over the course of many years are plain wrong in some tangible macroscopic way, and nobody ever noticed." The probability of those experiments being wrong is thus an upper bound on the probability that psi phenomenona are real.

...but that really isn't a criticism of it and cannot be used to justify an extremely low prior to nullify a small to medium effect size.

The fact that psi contradicts a large body of rigorous experiments in physics is, again, exactly why its prior probability is low. I have no idea what you're talking about with "to nullify a small to medium effect size.

priors don't take into account pessimistic meta-induction, the fact that psi wasn't just an anamoly found in psychology research but a purportedly natural phenomenon with any thousands of years of history, with field research as well.

Okay, you just doubled my prior probability for psi. See, priors can take into account all those unreliable anecdotes.

It seems any significant scientific discovery would have a very low prior, so I am extremely skeptical of the validity of subjective priors.

Any hypothesis that violated the known laws of physics would have a very, very low prior. Arguably any such hypothesis would be paranormal.
 
I wasn't intending to discuss Bem's experiments, which is why I was talking about the autoganzfeld, but I find Utt's response to Wagenmakers much more convincing, so I also don't find that to be obvious in the direction you say.

I didn't expect you to. But it is really glaring. If I gave that paper to a statistics class to analyze, I would fail anyone who couldn't point to several evidences of exploratory analyses or hypotheses fine tuned to fit the data.
 
You know things have really taken a turn for the worse when people start quoting Sean Carroll. Resorting to the words of shit-brain terrarium-enclosed academia-indoctrinated indoctrinators is a good indicator that it's time to put an end to the discussion. There's no making your way back up the hill after one makes that kind of move.
 
I don't know how else to address this. Wiseman chose the obvious criteria. What else could he have chosen that would lead you to believe the dog knew when Pam was on her way home?



But how would that indicate that the dog was telepathic and that there was something anomalous going on? The dog spent more time at the window the longer Pam was away, regardless of whether or not Pam was on her way home. So noticing that the dog was at the window didn't tell you whether Pam would show up in the next 5 minutes or the next hour. What's anomalous about that?



Okay, so presuming that you have looked at the data - what criteria could you use that would reliably indicate that Pam was going to be home in the next 10 minutes?



I've looked at the data in these experiments more than anybody else I know here. Over the years, I've pointed out stuff that nobody else has noticed (including Sheldrake). Just tell me what you see in the data that you think I don't see.

Linda


What other criteria? I don't know, maybe what Sheldrake used, which I already said, and which Wiseman replicated, which showed a significant effect. Wiseman replicated this and then lied to promote himself. That's fraud.
 
  • Like
Reactions: K9!
What other criteria? I don't know, maybe what Sheldrake used, which I already said, and which Wiseman replicated, which showed a significant effect. Wiseman replicated this and then lied to promote himself. That's fraud.

Linda and Jay have done this before on Sheldrake, had every point they were trying to make disproven and now they're at it again with you.

Neil, you're being played.
 
That is exactly why the prior probability of psi is low. To quote physicist Sean Carroll, if any psi phenomenon is true, then "the laws of physics that have been tested by an enormous number of rigorous and high-precision experiments over the course of many years are plain wrong in some tangible macroscopic way, and nobody ever noticed." The probability of those experiments being wrong is thus an upper bound on the probability that psi phenomenona are real.

What laws? Other physicists like Henry Stapp have worked on how PK could be described by slight biases in the QM equations which does not make QM "plain wrong." Carroll is also a person that believes in the many worlds interpretation because of mathematical parsimony, evidence be damned. I have also listened to a lecture of his about why God is not an explanatory hypothesis, and if he has a clue of what is actually science and what is metaphysics then he didn't demonstrate it there. Did Carroll specify what laws?

The experiments we have would NOT be wrong. Psi is indicating a new domain of discovery. What we learn would make current laws of physics as wrong as quantum theory made netwonian theory. Yes, there is a level of fundamental wrongness to Newtonian theory, and yes the metaphysics is completely wrong, but we didn't have to throw out Newtonian physics because of the discovery of quantum theory.


jay said:
The fact that psi contradicts a large body of rigorous experiments in physics is, again, exactly why its prior probability is low. I have no idea what you're talking about with "to nullify a small to medium effect size.

What experiments?

Regarding nullifying a small to medium effect size based on biased priors, I am referring to the Lindley-Jeffreys paradox:

https://en.wikipedia.org/wiki/Lindley's_paradox


jay said:
Okay, you just doubled my prior probability for psi. See, priors can take into account all those unreliable anecdotes.

When you have collections of many hundreds of personal accounts that are investigated and analyzed and published, they are, by definition, not anecdotal. It's called field research. Granted it is not as reliable as lab research, but current lab research was started to try to learn more about a phenomenon that seems to be indicated by field research, which was started because there seemed to be the possibility of a natural phenomenon that should be studied. The field research is not so easily dismissed, either.

And what, then, about pessimistic meta-induction? Historically all our theories have proven to be fundamentally incorrect. There is nothing to indicate that our current theories will not share the same fate. How about our incomplete knowledge of the physical world? How about our almost complete cluelessness about our own consciousness? How can one be so certain about such low priors? Especially when these subjective priors are used to bias against new discoveries by nullifying small to medium effect sizes via the Lindley-Jeffreys paradox.

And again, discoveries have, almost by definition, low priors. So my point is, so what if it has a low prior? That can't be used as a legitimate reason against it existing. What was the prior probability of quantum theory?


Jay said:
Any hypothesis that violated the known laws of physics would have a very, very low prior. Arguably any such hypothesis would be paranormal.

What laws?
 
Last edited:
I didn't expect you to. But it is really glaring. If I gave that paper to a statistics class to analyze, I would fail anyone who couldn't point to several evidences of exploratory analyses or hypotheses fine tuned to fit the data.

I guess the recently elected president of the American Statistical Association and chair of the department of statistics for University of California, Irvine doesn't understand Bayesian analysis.
 
What laws?

Quantum field theory.

The experiments we have would NOT be wrong. Psi is indicating a new domain of discovery.

"Domain of discovery." Word salad.

What experiments?

You're joking, I hope.

Regarding nullifying a small to medium effect size based on biased priors, I am referring to the Lindley-Jeffreys paradox:

https://en.wikipedia.org/wiki/Lindley's_paradox

The Jeffrey's–Lindley paradox has nothing to do with "nullifying a small to medium effect size," whatever that even means.

When you have collections of many hundreds of personal accounts that are investigated and analyzed and published, they are, by definition, not anecdotal. It's called field research.

The one thing I've noticed that consistently divides proponents and skeptics is that proponents give great weight to anecdotes field research, whereas skeptics consider such testimony research to be too prone to bias to take very seriously.

And what, then, about pessimistic meta-induction? Historically all our theories have proven to be fundamentally incorrect.

I disagree. Mostly, our prior theories have been found to be approximations, limiting, or special cases of more general theories.

How about our incomplete knowledge of the physical world?

According to Carroll, the physics of everything that could effect us on the macrospcopic level of everyday life is completely understood. There's no room for psi. Although he presents summaries of experiments to support this assertion, I have to admit I don't have the necessary background to assess his claims. The remaining gaps in our knowledge of physics relate to the very, very small and the very, very, very large scales

How can one be so certain about such low priors?

Questions about the probability of a prior probability are nonsensical.

And again, discoveries have, almost by definition, low priors.

Sorry, but that's just not true.

So my point is, so what if it has a low prior? That can't be used as a legitimate reason against it existing.

If something has a low probability, it probably doesn't exist. That's what having a low probability means.

What was the prior probability of quantum theory?

The question isn't really answerable as posed. But what was the prior probability that something resembling quantum theory would emerge as the answer to then-unsolved problems in physics, pretty high, I would think.

I guess the recently elected president of the American Statistical Association and chair of the department of statistics for University of California, Irvine doesn't understand Bayesian analysis.

Utts' Bayesian analysis of Bem's paper did nothing to address the question of p-hacking. Neither did Wagenmakers'. Wagenmakers first pointed out the evidence of p-hacking, but then he put that aside, and assumed for the sake of the Bayes analysis to show the evidential value of the results in support (or lack thereof) for psi assuming the results were legit. Utts likewise assumed that the data results were legit in her Bayes analysis.

In other words, she understands Bayesian inference. You don't
 
Last edited:
Quantum field theory.

"Domain of discovery." Word salad.

You're joking, I hope.

The Jeffrey's–Lindley paradox has nothing to do with "nullifying a small to medium effect size," whatever that even means.

The one thing I've noticed that consistently divides proponents and skeptics is that proponents give great weight to anecdotes field research, whereas skeptics consider such testimony research to be too prone to bias to take very seriously.

I disagree. Mostly, our prior theories have been found to be approximations, limiting, or special cases of more general theories.

According to Carroll, the physics of everything that could effect us on the macrospcopic level of everyday life is completely understood. There's no room for psi. Although he presents summaries of experiments to support this assertion, I have to admit I don't have the necessary background to assess his claims. The remaining gaps in our knowledge of physics relate to the very, very small and the very, very, very large scales

Questions about the probability of a prior probability are nonsensical.

Sorry, but that's just not true.

If something has a low probability, it probably doesn't exist. That's what having a low probability means.

The question isn't really answerable as posed. But what was the prior probability that something resembling quantum theory would emerge as the answer to then-unsolved problems in physics, pretty high, I would think.



She understands Bayesian inference. You don't.

Good job on not naming a SINGLE law of physics that psi violates. I guess you can't.
 
She understands Bayesian inference. You don't.

I didn't claim to. I said that her analysis disagrees with Wagenmakers, which you said was "obvious." If she understands Bayesian analysis and disagrees with Wagenmakers, then how can Wagenmaker's response be "obvious"?
 
  • Like
Reactions: K9!
Back
Top