cooper@pbsvax.DEC (Topher Cooper HLO2-3/M08 DTN225-5819) (07/24/85)
Sorry to keep this going against such hostility, but I really do think that this is relevant to physics. Normally, I feel that posters who try to "explain" what they said, are wasting their time, since they already said it. But, since I have frequently admired the insight of Ethan Vishniac's postings (and agreed with them more often then not) I must assume that I did not express myself clearly. I will post separately, if at all (I really am not trying to annoy people), my responses to points which did not seem to be simply my failure to communicate clearly. > The reason why >all this is not particularly interesting to physicists as an experimental >endeavor is that the problems here are unique to the field. The point I was trying to make is that I can see no reason to assume that the problems _are_ unique to the field (of parapsychology). In particular much of the modern (last 15 years) work in "PK" on "unselected" subjects, seems as far as the mechanics of the experiments, very similar to many experiments in modern physics. An "operator" observes a device which measures some quantity subject to a great deal of random fluctuation. The operator has a preference for some particular result (this factor is ignored in descriptions of physics experiment, but is normally there). The output from the measuring device is analyzed statistically and any bias is noted (this applies also to an underlying "noisy quantity" passing some threshold value, i.e., in a grain of photographic film). In parapsychology, when a bias is found, the experimenter says, in essence: "we have found an example of an experimental anomaly, of the particular type we call PK". In physics, with essentially the same set up, when a bias is found, the experimenter says, in essence: "my hypothesis is confirmed". I should point out that in PK experiments, the operator (agent, subject) doesn't seem to have to "do" anything, just to "want" a particular result. As a matter of fact, trying to do something seems to reduce the size of the anomaly. The strongest anomalies seem to happen quite consistently in the time period immediately _after_ the subjects stops trying. This is called the "release of effort effect." In what way is the Fabry-Perot optical interferometer used in a PK experiment different from that used in a physics experiment? In what way are the parapsychologist's matched thermistors distinguishable from the physicist's? Is there any real difference between the apparent sub-atomic events which trigger Schmidt's Geiger counter (already discussed here) and those found in deep mines by scintillation detectors or those found by photon counters in an X-ray telescope? It doesn't really matter whether or not the cause of the anomalies is poor controls, the incorrectness of elementary statistical theory, experimenter fraud, operator fraud (e.g., a lab tech), or something mysterious that people are able to do sometimes. Why do we assume that physics experiments are more immune to the source of the anomaly (whatever it is) than parapsychology experiments? How far can we trust the latest results about some new particle, proton decay, fractional charges, magnetic monopoles or heavy neutrinos? >Attempts to replicate experiments in ESP by independent, and skeptical observers >seem to have a much higher chance of rejection than 1 in 9. To date it would >appear to be closer to 8 (or 9) in 9. That's what I was trying to say. Conventional statistics says that (with an alpha of .05), one out of 20 carefully conducted experiments will produce false positives, and that one out of 400 carefully conducted experiments will produce false positive results in both the experiment and its replication (though pure replication is very rarely attempted). The results of parapsychology experiments, which can be viewed as control runs in a test on general experimental methods, indicate that when the effect size is small the actual values are closer to one out of three and (therefore) one out of nine respectively. Topher Cooper USENET: ...{allegra,decvax,ihnp4,ucbvax}!decwrl!dec-rhea!dec-pbsvax!cooper ARPA/CSNET: cooper%pbsvax.DEC@decwrl Disclaimer: This contains my own opinions, and I am solely responsible for them.