Recently Haukur, one of my best and dearest friends, posted something interesting on Facebook:
Since my blog is in English and this Haukur status is in Icelandic on Icelandic subject, I think a rough translation and explanation might be required for the swarm of the non-Icelandic speaking Eikonomics readers.
In his post Haukur refers to a “fun-fact publication” released by the Icelandic Statistics Office. The publication provides us with the three least common birthdays in Iceland. Haukur comments on the major birthday deficits and concludes that he can accept Leap Day but that the Christmas deficit confuses him.
How confused exactly should Haukur be?
While most people love rushing into explaining statistical discrepancies with speculative causal factors, I am a big fan of attributing everything to chance. So I decided to check if I could tell my friend that “it’s just a small variation and can more or less be explained by chance”. Well, let’s see how that turns out.
First, I calculated simple probabilities. In this exercise, I used a parallel universe identical to Iceland in every single way except, the people there make out and give birth at any random day in the year (no seasonal pattern of births). I this world, the following probabilities should hold:
- The chance of being born on any day, including the 24th or 25th of
December, would be around 0.27%.
- The chance of being born on a Leap day would then be around 0.7%.
Based on these numbers and a recent population figure of 329,100 I came up with the following chart:
The chart above – that shows actual birthday persons in purple and the hypothetical birthday in yellow – does indeed confuse. Why is the deficit so large (in relative terms) in December and almost non-existent on Leap Day?
One explanation that could be added and still keep my chance argument intact would be the Icelandic-human-breading-cycle. Could it be that Icelandic people are full of hormones in May, but all have headache in March?
Headache and hormone cycles
If headaches and hormones are to explain the December deficit, then the few births in December might be explained by pure chance within the month.
But the fact of the matter is that historically, people have been about equally interested in baby making in March and May. Also the super shortness of February adds to the problem and the babies that are made in early June are also more likely being pushed into the early days of March anyway and therefore the deficit (in relative terms) should even be larger on Leap Day then on the December days.
After loads of over-complicated statistical analysis including hundreds of different regression runs over multiple time periods as well as many different specifications of averages, I did indeed feel comfortable enough to produce the chart above and expand my calculations away from the parallel random universe into the real world.
The result: Leap Day looks even more sensible but Christmas days remain, for the lack of a better word, totally bunkers.
On Leap day, the model predicts around 216 birthday people. Six missing individual from the actual figures, I am fine with that and conclude that the Leap-Day discrepancy is up to nothing more than chance. Haukur was right.
Unfortunately, the model tells me that around 318 people are missing in the December days and all my statistical tests imply that this result is not up to chance. Furthermore, if this gap was to be explained by chance alone we would expect there to be some sort of drop in fertility at the end of March and beginning of April (see yellow bars in chart above). However the data suggests that in the period headaches are on the run and hormones on the rise.
Since chance does not explain the December deficits I could hypothesis about several factors that then might drive this result. But why would I rob the reader of the joy of speculating about causes. That is the most fun bit! Instead I will tell you an interesting fact that is a by-product of my statistical analysis.
Winter babies are coming
Historically Icelandic babies have had a higher chance of being born in the six „summer“ – or rather non-winter – months. But with increasingly better central heating, that has been changing. The chart below shows the probability (from my model) of being born in the summer months (yellow) and being born in the winter months (purple).
Notice that if you had been in your mom’s womb in the 19th century you would have been 10% more likely to pop out in the nice mild summer months. But in the 21st century we have converged and your unborn unmade kids are only around 3% more likely to pop out in time for wild-berry picking then their 19th century forefathers.
Merry Christmas to you all,