More Options

Moonshine

Book Review

James Rotton

Volume 21.3, May / June 1997

The cover of How the Moon Affects You advertises it as a “revised and updated version” of The Lunar Effect, which was published in 1978 (Anchor Press). There are two differences between this update and its predecessor. First, Lieber has given the book a title more likely to appeal to those who believe in astrology. This is not entirely bad. With its new title, the book is less likely to be mistaken for a scientific work. It will probably end up in the New Age section of bookstores. Second, its author has added a fourteen-page afterword entitled “New Research.”

The book’s predecessor introduced the world to Lieber’s “biological tides theory.” This theory maintains that since the moon causes ocean tides and the human body is 80 percent water, the moon’s gravitational pull also causes tides in the human body. If one accepts these premises, it’s a short step to believing that some people behave irrationally when their water balance is upset. Astronomers reject this argument as being based on a crude and misleading analogy. In a cogent review that appeared in the Spring 1979 Skeptical Inquirer, astronomer George O. Abell pointed out that the moon’s gravitational pull was less than the weight of a mosquito. Two of my colleagues, Roger Culver and Roger Ianna, subsequently showed that the moon’s “pull” was less than that of a wall of a building six inches away (see Culver and Ianna 1988).

This book contains very little that is new, as Lieber acknowledges: “When invited to revise and update the material in the present book, I was surprised to learn how slight the modifications were. Most of the earlier findings are still relevant” (p. 156).

It would take too much space to identify (let alone correct) all of the errors that precede this conclusion. However, Lieber’s claims have turned out to be wrong so often in the past that I felt compelled to do a line-by-line comparison of the two books. Lieber’s revisions served to save face. For example, Lieber’s first edition describes how he learned that the earth, sun, and moon would line up in January and February of 1974. Because this “cosmic coincidence” would cause high tides, Lieber alerted the Miami police, the local news media, and the emergency room of a public hospital to expect an outbreak of abnormal and criminal behavior. Lieber repeats this story in his revised edition, complete with the line: “Sure enough, all hell broke loose” (p. 61). After two pages of anecdotal evidence, his previous edition concluded by noting that there would be another cosmic coincidence in December 1990 and in January 1992. But of course these dates passed uneventfully. So it is not surprising to find that this is one of Lieber’s few revisions.

Another revision can be found in Lieber’s reference to The Jupiter Effect by John R. Gribbon and Stephen H. Plagemann (Random House, 1974), a book that, according to Lieber, reports “the significance of tides as triggers of earthquakes” (p. 86). He states that Gribbon and Plagemann are concerned that a coincidence of cosmic cycles “may trigger another great California earthquake.” But when? He gives no date. But, if we turn to Lieber’s earlier edition, we find that he gave a date: 1982. Of course, no earthquake of any note occurred in California that year.

Unfortunately, instead of correcting past errors, Lieber has added fourteen additional pages of faulty logic, distorted facts, and ad hominem arguments in the chapter entitled “New Research.” He reports that there have been “at least 43 data-based studies” since The Lunar Effect appeared in 1978. That number sounds about right. However, Lieber errs when he asserts that “critical review of documented findings reveals the following: positive and negative findings are pretty much equally divided” (p. 156). The fact is that most studies have failed to uncover anything resembling support for the lunar hypothesis, as I. W. Kelly, Roger Culver, and I have documented (Kelly, Rotton, and Culver 1996).

Lieber doesn't cite (let alone review) any of the “at least 43 studies” that he mentions, nor does his bibliography contain references to works that appeared since the first edition of his book. Instead of describing what other scientists have found, he devotes seven pages to a failure to replicate his work with C. R. Sherin. In a 1972 article, Lieber and Sherin uncovered what they claimed was “a lunar effect” for homicides in Dade County, where Miami is located. Their data spanned the years between 1956 and 1970. The present book contains a graph (on page 41) that seems to suggest that homicides peaked at the full moon. As Kelly, Culver, and I have documented (Kelly, Rotton, and Culver 1985-86), Lieber and Sherin based their claims on three tests that attained significance. They neglected to tell readers that they performed forty-eight tests of significance in all. Not divulging this information is like a gambler failing to tell us how many times a coin was tossed before three heads came up. Since Lieber and Sherin chose a conventional (p ¢ .05) level of significance, we would expect 2.4 (i.e., .05 3 48) of the tests to attain “significance” by chance alone.

Be that as it may, let’s take a closer look at Lieber’s new evidence. It consists of a plot of homicides in Dade County (2,714 new cases) between 1971 to 1980. Although Lieber does not give a value for his statistical test, he reports that it resulted in a “chi-square with a high significance (p ¢ .001).” But there’s a problem: Homicides did not peak when the moon was full! Rather, as Lieber reports, “the new-moon and full-moon peaks vanished. We found a peak midway between new moon and first quarter” (p. 160). This is obviously embarrassing, and Lieber suggests that “A. D. Pokorny and colleagues, Rotton and colleagues, and G. O. Abell would say ‘Aha, failure to replicate: previous results were Type I error. I told you so!'” (p. 161)

I can't speak for Pokorny, Abell, and others, but my guess is that Lieber’s result stemmed from his failing to control for weekend and other secular trends. Of particular importance is the fact that the 29.54 days in the moon’s synodic cycle is very close to 28-31 days in the months of the year. There are ways to avoid the biases that these trends introduce, as my colleagues and I describe in some of the articles that Lieber cites. However, instead of using these methods, which are pretty standard in econometric analyses, Lieber devotes a long paragraph to recounting all of the problems that have afflicted Miami during the 1970s: the Mariel boatlift, population increases, Cocaine Cowboys, et cetera. None of these would surprise viewers of Miami Vice, but one has to wonder why Lieber did not apply the same reasoning to his earlier study when Miami had to contend with Castro’s revolution, an influx of refugees, Watergate burglaries, and the five thousand deaths chronicled in Edna Buchanan’s Pulitzer Prize-winning The Corpse Had a Familiar Face.

After his long digression, Lieber uses a three-day moving average to smooth out the graph for 1971-80 homicides, and he applies the same operation to the data for 1956-70. Now, however, Lieber does not blame a full moon. He instead concludes that the curves reveal “an underlying semi-monthly lunar rhythm” (p. 165). Lieber has stumbled on an old truth that would be obvious to anybody who has taken a first-year course in econometrics: The easiest way to generate a curve is to average adjacent observations. The curve will be even more impressive if one superimposes a dotted sinewave over it, as Lieber does; doing so distracts attention from the data’s irregularities and differences between the graphs for homicides during the sixties and seventies. Those who teach statistics might want to use the graphs in this chapter to show students how figures can be made to lie.

Not only does Lieber distort his own work, he also misrepresents what others have written. As an example of sloppy scholarship, he refers to an article by Culver (1988), which was actually written by Culver, Kelly, and myself. He states that Culver committed a Type II error “when he [sic] reported on G. O. Gilbert’s findings” (p. 165).1 As Lieber reports, Gilbert found that fewer assaults occurred when the moon was full. He takes my coauthor, Culver, to task for concluding that this is contrary to the biological tides theory: “This is a clear example of a Type II error that results when a researcher is either blind to the theoretical underpinnings of his subject or is blinded by unconscious bias” (p. 166). The fact is that it was Gilbert who concluded, in his thesis, that “the number of significant full moon correlations was only slightly greater than expected by chance. . . . Since the sign of the correlation [for assaults] was negative, the relationship was a new moon relationship rather than a full moon relationship” (p. 166). Unlike Lieber, who cites a page from Dissertation Abstracts as his source, my colleagues went to the trouble of obtaining a copy of Gilbert’s thesis. Ironically, Lieber lists Gilbert as one of two persons who are “finding lunar connections.” The other person he lists is a scholar (Michael Persinger) who has done interesting work on geomagnetic activity but who has never published an empirical study on lunar cycles.

By now it should be clear that Lieber interprets any and all departures from chance as a “lunar effect.” He began by claiming that he had uncovered a lunar effect when three out of forty-eight tests of statistical significance uncovered a greater number of homicides during the full moon in Miami between 1956 and 1970. He extended his theory to account for an apparent increase in homicides that occurred after full moon periods in Cleveland (see Sanduleak 1985). Now he interprets peaks during the new moon and first quarter as lunar effects.

Rather than deal with these inconsistencies, Lieber tries to distract attention away from flaws in his theory by attacking those who have raised questions about his theory and the data on which it is based. Referring to reviews written by Kelly, Culver, and myself, he charges:

While they intellectually can accept that much of the evidence (including their own) supports the existence of a small lunar effect on behavior, they were unable to comprehend that such small effects, under certain circumstances, can have dramatic impacts upon behavior. Their viewpoint suffers from the absence on their team of a researcher trained in neurochemical effects on behavior. . . . One marvels at how adept these authors were at waving a mathematical wand and causing to vanish the results of seventy years of field and laboratory research. (P. 159)

First, we have never accepted that “the evidence supports the existence of a small lunar effect.” We would not, and have not, used a term (namely, lunar effect) that implies causality when discussing alleged correlations. What we have stated, in more than one article, is that the effect size for lunar associations in this area is so small that it is best chalked up to chance. In Kelly, Rotton, and Culver 1996, we stated that meta-analyses of previously published findings indicated that phases of the moon “accounted for no more than 3/100 of 1 percent of the variability in activities usually termed lunacy” (18). Such a small percentage is too close to zero to be of any theoretical, practical, or statistical interest or significance.

Second, if I were a psychiatrist, I would point out that Lieber evidences a considerable amount of ambivalence toward statistics. On the one hand, he makes much of the fact that his results achieved statistical significance; on the other, he ends one chapter by repeating Carl Jung’s tired saw about “the ruinous influence of statistics,” and he complains about critics who utilize a “statistical sleight of hand” (p. 158). The fact of the matter is that my colleagues and I used the same statistical procedures that the medical community uses to evaluate empirical findings. I find it disturbing that a physician who lists the Miami Heart Institute as his affiliation shows such a disdain for statistical evidence.

Finally, Lieber’s remark about our research team’s needing “a researcher trained in neurochemical effects” comes close to being an ad hominem argument. However, rather than take umbrage, I will use this forum to invite Lieber to name a published neurochemist who would be interested in joining us in evaluating research on lunar periodicities. I suspect that Lieber is going to have trouble responding to my invitation, because I failed to uncover anything dealing with lunar periodicities and neurochemical changes when I did a computer search of the literature. My search included all of the works listed in PsychLit, the GENMED option on LEXIS-NEXIS, and the computer library for Dissertation Abstracts. I located fifty-three works, including Lieber’s original publications. To put this small number into perspective, I might note that LEXIS stopped searching after one thousand articles when I did a search on “circadian rhythms.”

Most of the lunar studies I found dealt with psychological variables (e.g., homicide, psychiatric disturbances). A good number were by scholars (e.g., Kelly, Saklosfke, Martin, Culver) whom Lieber dismisses as “skeptics.” Only a few considered substances that might interest a neurochemist, and none of these studies has been replicated. My search failed to uncover any studies linking lunar cycles to substances that have been implicated as possible correlates of stress and aggression (e.g., serotonin, melatonin, epinephrine, norepinephrine, testosterone, cortisol, vasopressin [directly relevant to fluid content], growth hormone, pH, 17-OHCS, adrenocrotropic hormone).

It is hard to believe that all of the researchers who have documented circadian rhythms in stress hormones didn't collect enough data to check for lunar cycles. It is even harder to imagine that nobody has looked for a biological tide in biochemical processes when one considers the number of times that physicians send blood and urine samples out to be assayed.

However, instead of presenting data that would support his claims, Lieber resorts to repeating what others have told him in undocumented conversations. To take one of many examples, he states that a scholar working at Stanford “told me that this remarkable finding was replicated by other researchers” (p. 168). It doesn't take a lawyer to recognize this as hearsay.

Lieber ends his book with his response to a question posed to him in October 1978. Two things make Lieber’s response worth quoting. First, it gives a feel for the kind of nonsense that Lieber offers as evidence in this book. Second, I am astounded that anybody could remember such a long soliloquy. When asked to elaborate on how “[m]an’s responsiveness to subtle and persuasive effects provides us with an option for survival of our species,” he responded:

By virtue of our innate responsiveness to these imperceptible influences, we have within us a mechanism for survival under the most adverse of terrestrial circumstances. Suppose that at a time in the future a nuclear conflagration occurs. The only survivors are those lucky enough to have descended into underground shelters or caves located far below the Earth’s surface. Solar and societal timing cues soon would be eliminated in such a hostile environment. The only forces that penetrate to Earth’s core are gravity and geomagnetism; because our systems are capable of resonating with these two forces, we are able to synchronize our body’s rhythms with them. . . . Our biological clocks would not run down, as neurospora aboard Spacelab I did. Man becomes ill and nonfunctional when his body rhythms are disorganized; having the capacity to use gravity and geomagnetism to synchronize rhythms provides us with a crucial survival option.

Maybe. However, I suspect that having a biological clock out of sync is the last thing most of us would be worrying about after a nuclear conflagration.

Notes

I thank W. H. Laverty and P. Quincey for sharing their thoughts about Lieber’s book and I. W. Kelly for commenting on an earlier draft of this review.

  1. A Type II error is a statistical concept described in introductory textbooks. Investigators commit a Type II error when they fail to reject a null hypothesis that is, in fact, false. However, it is more common to discuss the probability of a Type II error in terms of its complement (namely, power). There are several ways to increase the power of statistical tests (see Rotton and Schönemann 1978). The most common way in research on lunacy is to base one’s conclusions on larger samples and data sets that cover a long period of time. As Kelly, Rotton, and Culver (1985-86) report, neither is associated with the probability of significant results in this area.

References

James Rotton

James Rotton is professor of psychology at Florida International University and a member of CSICOP’s Astrology Subcommittee.