November 2, 1999

Dear Dr. Stewart:
Following is the referee's report on your revised submission, "Using random
numbers to control events in the outside world."
Regretfully we will not be able to publish this paper.
Regards,
Bernard Haisch
Editor-in-chief, JSE

>>Referee's report:
>>"A New Interpretation of Marshall's Theory of Resonance, With Experimental
>Evidence"
>>and
>>"Using Random Numbers to Control Events in the Outside World"
>>
>>The first manuscript seems virtually unchanged from its previous
>submission, which this reviewer recommended rejecting for publication in
>JSE for a variety of reasons. There is no persuasive evidence in the
>second paper to alter that recommendation or the reasons for it
>>.
>>    1) The method employed for generating random events (hand flipping of
>coins) are vulnerable to any number of potential artifacts;
>>    2) The database is far too small to support any reasonable statistical
>inference;
>>    3) There are numerous variables which could, in principle, bear on the
>claimed results that are not controlled for in the design;
>>    4) Reporting the results of only one experiment of a series of 13,
>>based
>on the fact that it has the most significant results, is tantamount to data
>selection. Referring the reader to a website for a report of the full
>series is not a suitable substitute when submitting scientific findings to
>a peer-reviewed journal.
>>
>>In the opinion of this reviewer this paper is not appropriate
>for publication in JSE.

I did not in fact reply to the JSE because I thought it was better to wait until I had more data. Had I replied my response would have been something like the following.

 
Dear Dr. Haisch,

It seems to me that really new ideas in "paranormal" research are rare. Most articles in the field follow a predictable pattern: (i) a mysterious phenomenon such as telepathy, remote viewing, dowsing etc is investigated. (ii) Data are collected which weakly support the paranormal interpretation. (iii) The researchers conclude that more research needs to be done.

It therefore seems a pity when an article that contains new ideas is rejected (by one referee only) especially when the reasons for rejection are not valid.

Your referee gave four reasons for rejecting my article, three of which are wrong. The fourth is an inherent problem in all research in new scientific fields.

His reasons were as follows:

This is not true - a coin that is flipped by hand to a reasonable height is an almost perfect random number generator (albeit a rather slow one).

Clearly, if the number of turns of the coin could be made nearly constant (e.g. the standard deviation is below say 0.2 turns) the coin would be biased. However, if the standard deviation increases to only 0.5 turns, the bias drops to 0.92% assuming a normal distribution. I have shown this below on a simple Excel spreadsheet which uses the NORMDIST function.

This comment ignores the fact that my statistical analysis takes into account the size of the database. Clearly, one can pick up a weak trend in a large database with great certainty. But a strong trend can be seen clearly in a small database. Imagine an experiment such as mine where every single flip turned out to be a tail. Only 16 flips would be needed to confirm the hypothesis with an uncertainty of one part in 65,536! Clearly there is no necessity for a large database.

The results of the whole series of 13 experiments are used in the statistical analysis. I left out most of the raw data because I felt it was irrelevant and therefore confusing. I would be happy to put it back in if you feel this makes the paper read better!

When I was a student, we used a microelectrode to take readings from a snail’s brain. We knew when we had managed to impale a neuron, because we started to see action potentials. "Fishing" for results in this way is not unscientific, provided that the unsuccessful trials are included in any statistical analysis.

This objection applies to any investigation of phenomena that are not yet fully understood - an infinite number of variables that we have not thought of could be influencing the results. We therefore cannot be sure that we are interpreting the results correctly. This is true, and readers should feel free to come up with new hypotheses that explain the data better.

I believe that the ideas contained in this article have far-reaching implications for all paranormal research.  Your referee agrees that "the author raises an interesting theoretical
>concept with potential for fruitful experimental study".   Surely interesting theoretical concepts are exactly what paranormal research needs right now?

I therefore request that you reconsider my article.

Yours sincerely,