Friday, May 22, 2009

Testing Mrs Patricia Putts

Introduction

Dr Richard Wiseman and Professor Chris French recently tested Mrs. Patricia Putts, a proported medium in the U.K. This was a prepatory test in an attempt to win the JREF million dollar prize. Mrs. Putts sat with 10 sitter subjects. She wrote her impressions for each; no verbal communication was allowed. The medium was unable to see details about the subjects.

After she wrote all 10 readings consisting of some number of statements each, all readings were shown to all sitters in the blind. No sitter was successful in picking their respective actual reading.The experiment was declared a failure by Wiseman and French.

Analytical Technique

But this was more of a test of the sitters' ability to guess their readings than a test of Mrs. Putts' mediumship. It is unclear why Wiseman and French decided to pass on the more common statement analysis technique used in modern mediumship research. Given the subtle nature of the phenomena, it would seem that the statement analysis approach would be more meaningful. There are too many rater bias opportunities to consider the protocol used as fair.

As an alternative, it would have made more sense to ask the sitters to review a list of 100 statements (presuming 10 statements per reading) and have the sitters mark those statements which applied to them. Those would count as 'hits' and the remainder as 'misses'. Scores for each reading would then be rank-ordered for each sitter.

If the highest scoring reading for each sitter were in fact the reading associated with that sitter, the test would be a success, otherwise the test would be a failure. Given ten sitters and ten readings, there is a 1 in 10 chance that any single sitter would be associated with their test. This is below acceptable levels of chance outcomes for small effect sizes according to conventional statistics. Associating 2 out of 10 readings with the correct sitters would occur at the 1 in 100 probability level. Three out of 10 would be at the 1 in 1000 probability level and so on. In fact the chances of all ten readings associating with their correct sitters is 1 in 1 billion. So this technique is capable of measuring incredibly small probabilities.

The original protocol measured a large effect size phenomenon, namely the ability of 10 sitters to each select their respective readings. No special equipment, abilities or knowledge required except that each sitter supposedly had the personal knowledge required to complete the task. In contrast, the method described above supports a far smaller effect size, which is consistent with anomalous psi phenomena (which includes mediumship). Therefore the described procedure would seem to be more correct for the purpose as well as being a far more objective a measure. The chosen protocol for example does not provide a way to know why the sitters selected the readings as they did. So there is no way to know if any influences occurred that might explain the outcomes.

Experimenter Effect

There is also the issue with respect to Experimenter Effect. This was conclusively demonstrated when the same Dr. Richard Wiseman failed to replicate a staring experiment by Dr. Marilyn Schlitz in the United States. When his objections were published, Schlitz exchanged places with Dr. Wiseman in an attempt at replication. Surprisingly replication occurred consistently with Dr. Schlitz and failed consistently with Dr. Wiseman, even when they exchanged test subjects. From this it was mutually concluded that the failure to replicate was something to do with Wiseman and might well have to do with his skeptical belief system. Subsequently, the Experimenter Effect has been noted in parapsychology experiments and in more mainstream research involving small effect sizes and subtle phenomena including medical research.

There was no attempt to counter this bias by using a skeptical and a non-skeptical investigator for example. This despite the fact that both Wiseman and French know and understand the implications of experimenter effect. If the test were truly unbiased they should have objected and at minimum recommended appropriate controls.

Summary

Wiseman and French seem to have followed a protocol that is contrary to modern best practice in mediumship research. It certainly appears they both are well aware of the shortcomings and pitfalls involved based on their published papers. One wonders why they would endorse such a protocol.

It was unwise to have paired Dr. Wiseman, of Schlitz/Wiseman staring experiment fame, with another skeptic like Professor French. The lack of balance not only calls into question the objectivity of the design, apparently supported by the protocol executed, but also suggests that ad hoc there was little opportunity for success. Perhaps Mrs. Putts' surprise at her complete failure was in fact warrented.

The acceptance by Wiseman and French of biased analytical techniques and their lack of attempt to control for well-known Experimenter Effect calls the basis of the entire test and their findings into question. It would be most interesting to reanalyze the data from a statement analysis perspective to see if the score changed.