Five years or more ago, I came across an article – reference 4 – by an astronomer about how the moon did not intervene very much in human affairs – say not in the patters of hospital admissions and not in the pattern of births. Passing notice at references 5 and 6. The fact that the moon was the cause of tides notwithstanding. The fact that the moon has figured large in man’s imagination – one result being lots of moon goddesses – notwithstanding.
Much more recently, I revisited an article about p-values, as noticed at reference 2.
Then a few days ago I came across an article – reference 1 – about the effect on the moon on sleep in Uppsala, in Sweden, a place which is at about the same latitude as Lerwick in the Shetlands. So a place with long winter nights and long summer days. And the abstract was littered with p-values. So it seemed right and proper to take a look.
Very roughly speaking, on the basis of a sample of about 800 sleeps, people seemed to sleep better when the moon was waning than when it was waxing and the effect was more pronounced with men than with women. That said, the effects did not seem to be very large. From where I associate to the tricky business of trying to make significant statements when confronted with a great heap of rather messy data from brain scanners.
I was introduced to the world of sleep science, so plenty of new terms to learn about. There does seem to be a lot of it about, with plenty of sleep institutes, plenty of sleep scientists – and a sleep institute figures in at least one episode of ‘Lewis’ on ITV3. And reference 7 reports lots of problems with sleep. Certainly, speaking for myself, sleeping is not so good as I get older, so maybe sleep is a big deal for older people, and trying to do something about it is a big deal. Whisky is not the best way forward!
The sleeps used here were taken between 2001 and 2018, in three very roughly equal chunks. The first chunk was data recycled from a previous study. I didn’t find out where the other two chunks came from. So I have no idea how good the sample was for the purpose.
All the sleeps were monitored by something called polysomnography, which appears to mean that the subject is wired up. Which all sounds rather intrusive, and not conducive to sleep, but lots of it goes on and it can be done at home. Read all about some suitable equipment at references 8 and 9. With reference 9 containing lots of fierce looking health warning about its use – as well as the helpful diagram included above. While I thought the picture at the top of the flier was just so much eye-candy. Nothing much to do with people with problems being wired up at all.
It seemed odd that sleep should be affected by the direction of travel of the moon, whether it was waxing or waning. I would have thought that the amount of moon – say no moon, thin moon, fat moon, full moon – would have been a more natural place to start, but I did not find anything about that.
Nor did I did any reference to the lengths of either days or nights, which one might have thought would have an effect on sleep patterns in the far north. And given that the authors knew about the state of the moon for each sleep, they could have known about the state of the sun, that is to say the length of the day.
Or to the amount of ambient light – solar, lunar or otherwise – during the sleep sessions, probably not known as the sleeps were recycled from previous studies. With one advantage of sleep laboratories with sleep cubicles being that all this sort of thing is under control and easily logged.
All that said, the idea here was to test the relation between quality of sleep (say statistic Q) and the dichotomy between waxing and waning. So we will be interested in whether the mean value of Q for when the moon waxing is significantly different from that when it is waning. On the usual assumptions about the distribution of the statistic Q, I would have thought that getting p-values and making proper use of them ought to be a relatively straightforward application of something like the Student’s t test. A test which I now know is available in Microsoft’s Excel.
It comes to mind that a long time ago, when I was trying to learn something about statistics, one had books of statistical tables, rather like books of logarithms, and you could look p-values up in them. But Microsoft notwithstanding, it might take me a little time now to work out what to do in their absence.
Instead, I had a poke around using Bing and lighted upon the helpful freebie about statistical inference at reference 10, described as course notes, but actually rather a lot better than the course notes I remember. And pitched at about the right level for me. Maybe it will drag me to a point where I can make more progress with the use and abuse of p-values.
Further consideration of the effect on the moon put aside for the time being. Maybe it will resurface in the weeks to come.
PS: reference 12 suggests that any connection between the menstrual cycle and the lunar cycle remains a lively subject of debate. They might be near enough the same in length, but synchronisation of the onset of the menstrual cycle with any particular phase of the moon is another matter.
References
Reference 1: Sex-specific association of the lunar cycle with sleep - Christian Benedict and others – 2022.
Reference 4: No Evidence of Purported Lunar Effect on Hospital Admission Rates or Birth Rates - Jean-Luc Margot – 2015.
Reference 5: https://psmv2.blogspot.com/2015/06/statistics.html.
Reference 6: http://psmv2.blogspot.com/2015/06/part-2-of-2.html.
Reference 9: Natus Brain Monitor & Embla Dx Series: User and Service Manual – Natus – 2018.
Reference 10: Fundamental Theory of Statistical Inference – G. A. Young – 2021. The course notes.
Reference 11: Essentials of Statistical Inference – Young and Smith – 2005. The source book. Available for £30, including postage, from Abebooks, compared with the £10 I paid Ai Printers on our High Street to print the course notes version.
No comments:
Post a Comment