ESP/Remote Viewing vs. Just Guessing

[Steven Howard]

Here's something I don't get. When parapsychologists conduct experiments on ESP or remote viewing, their positive results are always just slightly better than you'd get by guessing, but better enough that they claim statistical significance. This usually leads to a lot of debate about methodology, meta-analysis and so forth. What I don't get is, why don't they just repeat the experiment with a larger pool of possible targets?

Suppose you did a study with Zener cards and found that your subjects were successful 22% of the time, as opposed to the 20% you'd expect. And further suppose that, given the design of your experiment, a combined score of 22% was right on the edge of being statistically significant. Rather than argue about math for ages, why not just repeat the experiment with a deck of cards that have ten distinct symbols rather than five?

As I see it, one of two things will happen. Either your subjects will still be correct around 22% of the time, which would be much more interesting, or else it will be closer to the expected 10%. In the latter case, even if they score, say, 12%, wouldn't you have to conclude that they're just guessing? I mean, why else would the size of the target pool affect the results?

I haven't read a lot of parapsychological research, so maybe I'm wrong and this question has been addressed. Does anybody know?

[JLam]

Sounds plausible to me. If they're so goddamn good, why not up the target pool to 50 different symbols?

[Nucular]

Absolutely, that is one necessary way of confirming results: replication.

Using larger sample sizes increases the power of statistical tests too, so 'the more the better'.

But the key is methodology and design: a rubbish experimental design will lead to false positives whether there are 10 subjects or 1,000 - that's why the methodological wrangling/criticism stage is a necessary chunk of the process, I think.

[I less than three logic]

They’d still claim victory in your example. In fact, they’ll claim they did twice as well with 10 symbols compared to 5. They would say it went from 10% over chance (20 being chance, so 22 being a 10% increase) to 20% (12 being a 20% increase from 10). :)

[Ersby]

From what I recall, the reason why parapsychologists use small target sets (of four or five elements) is to keep judging easier and less boring.

More recently, in the paper "Managing the Target Pool Bandwidth:
Noise Reduction for Anomalous Cognition Experiments" from SAIC of STARGATE fame put forward the idea that a target set which is too varied actually reduces psi, for reasons not really explained in the paper.

Make of that what you will!

[Nucular]

Oh, I misread the OP in my last post, I thought you meant larger sample size, rather than more target stimuli - soz.

[Jeff Corey]

I see no reason that they should not use just two targets, say a black field and a white field. That would eliminate any ambiguity in scoring and the results could easily assessed using the binomial test.
There is no advantage in adding more targets in terms of the experimental design or statistics,

[Steven Howard]

More recently, in the paper "Managing the Target Pool Bandwidth:Noise Reduction for Anomalous Cognition Experiments" from SAIC of STARGATE fame put forward the idea that a target set which is too varied actually reduces psi, for reasons not really explained in the paper.

See, they're getting there. Now they just need to give it a name (as they did with "the shyness effect") and they can just hand-wave it aside as though they'd actually explained something.

[Tanstaafl]

I don't understand why no one seems to ask:

If remote viewing works, why don't blind people do it?

I mean who has more motivation to get it working? Who has more time without normal vision to complicate things? Wouldn't some blind people just stumble onto this if it were real?

I suppose they probably have a semi-rational answer, but I'm betting I wouldn't buy it.

[Soapy Sam]

Oh they do. They do. But they remote view a quiet street ten miles away and step under the bus in front of them.

It's called "Natural Selection".

God has a truly wicked sense of fun.

[jimtron]

How's this for an RV test. A string of ten characters (numbers and/or letters) is written in bold black on a white card, and placed on an easel. The RV'er correctly identifies the string from the next room, or down the block, or wherever they want to be (as long as they can't see the string with their eyes, of course).

Wouldn't that be simple and eliminate lucky guesses?

[Jeff Corey]

Two targets would be simpler and the lucky guesses could be easily evaluated by the statistical test.

[jimtron]

Bottom line: if anyone really has ESP and/or RV capabilities, I don't see why they couldn't prove those abilities repeatedly under test conditions (conditions that would rule out guessing, deception, etc.).

[blutoski]

Here's something I don't get. When parapsychologists conduct experiments on ESP or remote viewing, their positive results are always just slightly better than you'd get by guessing, but better enough that they claim statistical significance. This usually leads to a lot of debate about methodology, meta-analysis and so forth. What I don't get is, why don't they just repeat the experiment with a larger pool of possible targets?

Suppose you did a study with Zener cards and found that your subjects were successful 22% of the time, as opposed to the 20% you'd expect. And further suppose that, given the design of your experiment, a combined score of 22% was right on the edge of being statistically significant. Rather than argue about math for ages, why not just repeat the experiment with a deck of cards that have ten distinct symbols rather than five?

As I see it, one of two things will happen. Either your subjects will still be correct around 22% of the time, which would be much more interesting, or else it will be closer to the expected 10%. In the latter case, even if they score, say, 12%, wouldn't you have to conclude that they're just guessing? I mean, why else would the size of the target pool affect the results?

I haven't read a lot of parapsychological research, so maybe I'm wrong and this question has been addressed. Does anybody know?

I think part of the reason for keeping the same protocol in repeat experiments is to use the data in meta-analysis down the line. The second benefit is that you don't have to undergo the expense of establishing new thresholds for acceptance. The big part of the expense is time and effort arguing with skeptics and other psi experimentors. The Zener / Ganzfeld experiments have enough debate around them already: why invite a whole avalanche of this to strike out on another testing protocol?

Hyman has done a good job reviewing the Ganzfelds and I'd make an effort to review his findings to understand that there's no real problem with the protocol itself. It's just the dishonesty or credulity of the testers that causes problems.

[Zygar]

Two targets would be simpler and the lucky guesses could be easily evaluated by the statistical test.

I agree whole-heartedly. Two possible targets would make the statistical anomalies much easier to track, and would introduce a lot more consistency to the results. There would be a lot less room to argue about the results which would allow the parapsychologists to focus 100% on improving their methodology.

[Steven Howard]

I guess part of my problem is that I don't really have the background or training in statistics to appreciate why fewer random targets would be better than more. I'll take your word for it that it's the case, it's just counter-intuitive to me.

At a more fundamental level, I wonder why we're comparing the results to random chance in the first place. Why are we measuring how much better the "psychics" are than somebody's who's guessing randomly, rather than how much worse they are than somebody who can actually see the target?

[Steven Howard]

Okay, I think I see part of where my thinking was wrong on this. I was incorrectly assuming that a success rate of, e.g, 25% with Zener cards would -- if it were actually an indicator of real ESP -- mean that the person's ESP was reliable 25% of the time and unreliable 75% of the time. So repeating the experiment with ten or twenty possible cards instead of five would still show a 25% success rate. Now that I think about it, I realize I wasn't accounting for the times when the ESP didn't work, but the subject guessed right anyway. Which would mean -- again, assuming that ESP was real -- that this person had ESP that was reliable about 5% of the time, and that therefore you'd always expect them to be about 5% better than random chance, regardless of what the random chance was.

All of which makes the supposedly "successful" ESP and remote viewing results even less impressive.

[CFLarsen]

I see no reason that they should not use just two targets, say a black field and a white field. That would eliminate any ambiguity in scoring and the results could easily assessed using the binomial test.
There is no advantage in adding more targets in terms of the experimental design or statistics,

It makes a lot of sense to the parapsychologists, because it obscures the fact that their simple experiments yield no positive result.

So, instead of accepting that there is no phenomenon, they move on by designing a new experiment with increased complexibility, which ensures that the result is even more suitable for interpretation. And, then, on to a new, even more complex design.

The only reason they can do this, is because their colleagues don't keep them in check. Real scientists do this, but not pseudoscientists. Oh, no.

Deceit is an inherent trait in paranormal beliefs.

[davidsmith73]

It makes a lot of sense to the parapsychologists, because it obscures the fact that their simple experiments yield no positive result.

So, instead of accepting that there is no phenomenon, they move on by designing a new experiment with increased complexibility, which ensures that the result is even more suitable for interpretation. And, then, on to a new, even more complex design.

The only reason they can do this, is because their colleagues don't keep them in check. Real scientists do this, but not pseudoscientists. Oh, no.

Deceit is an inherent trait in paranormal beliefs.


What do you base all these accusations on?

[Beady]

What do you base all these accusations on?

You have to get used to Claus. His thing is demanding evidence, not giving it.

[delphi_ote]

What do you base all these accusations on?
I don't know about the rest of it, but the last sentence is definitely right. Whether it's self deception or the deliberate deception of others, somewhere along the line there's a disconnect between the parapsychologists' theories/experiments and reality.

[CFLarsen]

What do you base all these accusations on?

I based them on the verifiable facts.

Take a look at the Global Consciousness people (http://www.skepticreport.com/psychics/radin2002.htm).

Take a look at the PEAR people (http://www.skepticreport.com/psychics/shapesintheclouds.htm).

Take a look at Gary Schwartz (http://www.skepticreport.com/psychics/schwartzkeen.htm).

On and on it goes. None of them repeats an experiment and get the same positive result. None of them has ever had an experiment replicated by independent researchers. Their research is not only sloppy, they ignore criticism and refuse to change anything that will improve the quality of their experiments.

None of these people are slobbering idiots, mind you. They are highly educated, highly intelligent people. They know what they are doing. They are just not interested in doing it right.

[blutoski]

None of these people are slobbering idiots, mind you. They are highly educated, highly intelligent people. They know what they are doing. They are just not interested in doing it right.

I think that was the epiphany of Dr. Susan Blackmore: "Holy crap, after all this time, I finally realized they're not investigating, they're confirmation-seeking." She abandoned the field as beyond rescue, because there is no interest among the major players to conduct an actual investigation. They need to ensure positive results. Motives vary.

This is over and above evidence for fraud, of course, which taints the efforts of those who are merely self-deceived.

[Ashles]

I think the most interesting question would be - if I had an ability that was so miniscule I could only detect it by use of large scale studies and statistical analysis... what would lead me to assume I had this ability in the first place?

[CFLarsen]

I think that was the epiphany of Dr. Susan Blackmore: "Holy crap, after all this time, I finally realized they're not investigating, they're confirmation-seeking." She abandoned the field as beyond rescue, because there is no interest among the major players to conduct an actual investigation. They need to ensure positive results. Motives vary.

This is over and above evidence for fraud, of course, which taints the efforts of those who are merely self-deceived.
Indeed.

I find it increasingly difficult to come up with excuses for the professionals in this area. The higher it goes, the crookeder it becomes.

[blutoski]

I think the most interesting question would be - if I had an ability that was so miniscule I could only detect it by use of large scale studies and statistical analysis... what would lead me to assume I had this ability in the first place?

In their defense, most of the experiments are designed to detect an overall effect, rather than identify a 'star'. Their hypothesis is that there is 'something' that can't be explained by random chance, not so much that there are individuals with outstanding talent. Although they do sometimes claim to find the latter in their studies. These people have been either exposed as hoaxers, or have been unable to repeat their results, which just goes back to a chance explanation.

[Ashles]

In their defense, most of the experiments are designed to detect an overall effect, rather than identify a 'star'. Their hypothesis is that there is 'something' that can't be explained by random chance, not so much that there are individuals with outstanding talent.
Exactly. So why would this hypothesis be formed in the first place?

I could form the hypothesis that I could mentally make leaves move a tiny amount on trees 5km away. But why would I spend ages testing such a hypothesis without any clear evidence to start with?

It seems to be simply the desirability of such abilities that is the impetus for the creation and continuous testing, rather than any actual evidence towards their existence in the first place.