UNBELIEVABLE - Stacy Horn - Nonfiction

In 1930, backed by prevailing cultural interest, J.B. Rhine started a laboratory at Duke University to study the possibility of parapsychology. By now, all that remains are letters and records in the Duke archives, and a few splinter groups, one run by the descendants of a purported medium, one a controversial member of the AAAS, and none university-affiliated, despite Princeton's dabbling in PEAR, the Engineering Anomalies Research lab. Though the phenomnea investigated by Rhine and his colleagues ranged from poltergeist activity that inspired many films to mediums, psychics, and unexplained 'hauntings' and spooky lights, Rhine's attitude was always that of a strict scientist and experimentalist, and the vast majority of his experiments dealt with ESP cards--cards printed with a series of five symbols drawn from a deck and not shown to a subject, who was supposed to guess which symbol was on the card. The point of the experiment is that, if the subject is truly guessing, then probabalistically he should only get one out of every five cards right. If he gets many more than that correct, then something else is at play. According to Horn, who regrettably does not reproduce any of the data or statistical analyses, Rhine had a number of subjects perform significantly above chance, a conclusion which he took as evidence of latent ESP ability in much of the population.

Rhine is frequently described as a charismatic, driven figure, and presumably because of this, the book is much more taken with descriptions of the people and relationships that formed the parapsychology research community, rather than the actual experiments or data. Experiments and fieldwork are described, and some tantalizing evidence presented--a mother who envisioned her son in danger on the battlefield and yelled "Duck!" later to find in a letter that her son heard her, followed her direction, and saved his life in doing so; a deathbed vision of a recently deceased relative who the ill woman didn't know had died; a house where objects slid across counters and fell to the floor of their own volition, in the presence of reporters and policemen. Of course, the skeptic's reaction to all of these is that it must have been trickery, and the viewers and scientists present simply must not have investigated thoroughly enough. No evidence presented on the page is going to be great enough to convince a reader who isn't able to be there for herself. But of course, Rhine knew that, which is why he shied away from many of the unexplained occurrences brought to his attention by members of the public, and persisted with his ESP card experiments for his entire scientific career.

Horn reports that she has seen the data from these experiments, and believes it, at least in the case of Rhine's most spectacular cases. I have no reason to doubt her, but where her report falls flat is that she doesn't print even a single data table, for us to run the numbers on ourselves. She mentions that after years of debate centered around Rhine's statistical analysis, an American society of statisticians published an official report validating Rhine's methods, and insisting that any objections to his results have to come from methods, not from statistics. However, this was much before Bayesian statistics caught on, and as we now know, frequentist p-values and confidence intervals can be highly overstated without taking into account prior knowledge. If I were to run a statistical analysis on Rhine's data, I would use a very skeptical prior, which might show that even runs of card-guessing much higher than chance would fall within normal probability. Pointedly, Horn mentions that Rhine was accused of stopping experiments whenever he wanted, probably because such a card-flipping experiment could go on as long as the experimenter was willing to reshuffle the deck. However, in any run of coin-flipping, there are bound to be periods of greater-than-average tails frequency, as well as periods of greater-than-average heads frequency. It is easy to imagine Rhine stopping when he was ahead, or going only one round with a subject who was performing so far below chance that no ability was suspected at all. Horn takes a very sympathetic viewpoint toward Rhine, and dresses down skeptics for being willing to ignore evidence, but she also doesn't indicate that she has done her own statistical analyses, or investigated any of these phenomena herself. Her references to pertinent literature are spot-on, and recognizable to any student of the field, but even accepting that Rhine had the best of intentions, it's easy to see what one wants to see in perfectly objective data. In a number of anecdotes, Rhine is unwilling to exploit the grief of mourning families to test or recommend mediums and psychics, but he is willing to take the money of philanthropic donors interested in the question of life after death, or a persistence of spirit, even though he never directly addresses these questions. It may have been his conscience that drove him to produce results to satisfy his benefactors in some way, even though he was unwilling to venture into questions less scientifically examinable. Even the best scientists feel pressure to produce, and when one's salary is in some way dependent on one's continued progress, it would be no surprise to find that, even subconsciously, one starts to skew the data.