May 06 2013

The Lunar Effect and Confirmation Bias

I gave a seminar recently to science teachers and the topic of whether or not there is a lunar effect came up. I was not surprised to find that 80% of them believed that emergency rooms and police stations are more busy during a full moon. I was also not surprised, but only because I have been there before, that they were highly resistant to my claim that the scientific evidence shows that there is no such effect.

Several questions emerge from the notion that the phases of the moon affect human behavior: what is the plausibility of such a claim, is there actually such an effect, and if not why do so many people believe that there is?

Plausibility

One of two justifications are commonly given for how the moon might influence human behavior. The moon basically has two physical effects on our environment – gravity and light. Astrological influences are not worth further discussion in this article, and I rarely hear that as a justification from the general public in any case.

Gravity is the far less plausible explanation of the two physical effects of the moon. The reasoning often goes something like this: the moon causes tides, which are powerful gravitational effects on water. Our bodies are mostly water, and in fact our brain are floating in water, therefore the moon’s tidal effect might affect our brain function.

This reasoning fails on many levels. First, the moon has a tidal effect regardless of phase. The only difference with lunar phase is the relationship between the lunar tide and the solar tide (yes, the sun has a tidal effect on the Earth as well). During a full moon the lunar and solar tides are lined up, and therefore the combined effect is additive (which is called a spring tide).

However, the same is true of the new moon, therefore if the full moon effect were due to tidal forces there should be an equal new moon effect.

The bigger problem with the gravity explanation is that tidal forces are dependent upon the difference in the distance of the near and far side of an object from another large object. So the ocean tides are caused by the different gravitational pull of the moon on the near side vs far side of the Earth. The tidal force of the moon between the near side and far side of your head is negligible. This is simply not a viable source of an effect on human behavior.

The more plausible mechanism for an alleged lunar effect is the light from the moon. It is reasonable to hypothesize that people are more willing to be outside and active during a full moon because there is more light. This effect, however, should therefore not be present when the sky is overcast or in large cities where artificial light trumps moonlight.

Is There a Lunar Effect?

This is one question that is very amenable to standard observational studies – is there a correlation between some kind of event and the lunar cycle? There have been hundreds of such studies over the last few decades, involving emergency room visits, births, accidents, crime, crisis center calls – just about any marker of human behavior you can think of. Systematic reviews of this research consistently demonstrate that there is simply no evidence for any such effect.

Many of these studies cover years or decades of data, and thousands or tens of thousands of data points. A recent German study, for example, found no lunar effect for births between 1920 and 1989. A review of studies looking at crisis center calls found:

12 studies are reviewed that have examined the relationships among crisis calls to police stations, poison centers, and crisis intervention centers and the synodic lunar cycle. On the basis of the studies considered it is concluded that no good foundation exists for the belief that lunar phase is related to the frequency of crisis calls. In addition, there is no evidence whatsoever for the contention that calls of a more emotional or “out-of-control” nature occur more often at the full moon.

An often cited 1985 review by Rotton, Kelly and Culver found no lunar effect for a long list of events, including suicide, homicide, crime, sleep walking, alcoholism, and many others. Several later studies of psychiatric hospital admission also found no effect.

The bottom line is this – this question has been asked and answered, there is no lunar effect. A massive amount of data simply shows no connection between lunar phases and human behavior.

Belief in the Lunar Effect

Why, then, does belief in this effect remain high? Surveys show that belief in this effect remains at about 40-45%, even among those highly educated. In fact, mental health workers have a greater belief in the lunar effect. The short answer as to why this is – confirmation bias.

Confirmation bias is a bias in human thinking that causes us to notice, accept, and remember information that confirms beliefs we already have, while ignoring, forgetting, or explaining away contradictory data. This is a powerful effect that can lead to the very compelling illusion that an effect is real when it isn’t.

Belief in a lunar effect, therefore, feeds on itself. When we encounter human behavior that seems out of the ordinary, or we are simply noticing the right side of the Bell curve of variation of crazy events, and we believe in the lunar effect, then we are likely to notice if there is a correlation.

For example, one night when I was working in the ER, and it was a busier than average night, a nurse commented, “Wow, it’s really crazy tonight. Is there a full moon?” The answer was no, the moon was in a completely different phase that night. When I informed her of this fact she simply shrugged and promptly forgot the whole thing. One can imagine that on other busy ER nights that happen to fall on or near the full moon that would resonate with her and confirm her belief in a lunar effect.

Confirmation bias can make it seem like there is strong evidence (anecdotal) for something that does not exist outside of our belief.

When I informed the teachers that the evidence shows there is no lunar effect, they simply did not believe it. Their personal experience spoke much more powerfully to them then abstract data.

Of course, the essence of scientific and critical thinking is setting aside personal anecdotes, experience, and belief, and embracing more rigorous data and analysis.

Conclusion

The data is clear, there is no lunar effect for any measured human behavior or medical events. This question has been sufficiently studied that this conclusion is very firm. We can never prove with finite data that an effect size is zero, but we can say there is no evidence for the effect, and if there is such an effect it must be extremely tiny – too small to have been detected even by hundreds of studies with tens of thousands of data points.

The real phenomenon here is one of human belief that persists in the absence of a real phenomenon. The real lunar effect is confirmation bias and understanding how the human brain is a belief machine.

27 responses so far