"Liz Tuddenham" wrote in message
id.invalid...
Rod Speed wrote:

NY wrote
[...]
OK. So this all hinges on the fact that the birth rate (and maybe death
rate) varies throughout the year.

Yes, and all you need is the birth day to vary, the
death day doesn't need to to get the statistical quirk.

That makes sense. Birth rate is under human control (to some extent)
and shows annual peaks, but while death rate only shows general seasonal
trends.

If the birth rate is peaky by calendar date and the death rate
is flat, more of the deaths will be on days which coincide with
birthdays - not because there are more deaths on those days but because
there is a greater chance that each dead person had been born on that
day.

That's the bit I don't understand. *Why* will the chance that *I* die on a
given date (eg my birthday) be affected in any way whatsoever by how many
people were born on that date? Why is it more likely that I will die on a
date when many rather than few other people were born?

but... if the death rate were also peaky and the peaks didn't coincide
with the peaks of the birthdays, the effect could be reduced or even
reversed. This could happen if voluntary euthanasia becomes more
widespread.

I may be starting to understand why I'm having problems with this. (Hooray,
says Rod!)

There are two ways in which the original assertion could have been worded:

- Why do more people die on their birthday than any other day of the year?

- Why is there a greater chance that Person A (as an isolated person) will
die on his birthday than any other day of the year?

These two may or may not be identical. One considers the population as a
whole, and the other treats each person as an independent individual.

I'm really not sure what the exact wording of the question was: I've just
given the gist of it, as I remember it several decades later.

I think I'm looking at the problem from the point of view of the second
assertion (each person is an isolated, independent individual), and I don't
see how the chance of Person A dying on any day is affected in any way by
how many other people happened to have been born on that day (in one year or
another). Before all this analysis, I would have expected that the chance of
a person dying on any given date was affected solely by environmental
factors (seasonal variation in diseases, climatic variation in immune
system) etc, the person's own gradually increasing chance of dying as they
get older); and was otherwise the *same* probability of dying on any day of
the year, without a spike on the anniversary of the person's birth.