"Rod Speed" wrote in message
...
NY wrote
Hmm. So for statistical reasons which don't have a cause (so I'm wasting
my time looking for one!), that fact that more people are born on one day
of the year than another means the each person is more likely to die on
the anniversary of when they were born than on any other date of the
year?

Nope, the day they die is completely irrelevant. It’s the lumpiness in the
day of the year they were born on that produces the small statistical
quirk.

The date when they die isn't irrelevant: the original question was "Why do
more people die on their birthday than any other day of [the year]". In
other words, if you plot a frequency distribution for a large population of
(birth_date - death_date) [ignoring both years] against number of people who
die on each day you get a peak around zero (birth date = death date
[ignoring both years]). This is the statistical quirk that you have been
describing. Why you suddenly say "the day they die is completely irrelevant"
baffles me and makes me wonder if goalposts have inadvertently got moved
somewhere in the discussion.

I'm curious *why* it should be so, but statistics and frequency distribution
only tell us *what* happens, with a vague suggestion that it is as a result
of the varying birth rate over a year. I can't being to imagine *why* a
varying birth rate should cause a spike when birth date = death date, but
I'll take it on trust that it's a statistical quirk.

Given that I can't understand why it should happen, I certainly would never
have guessed that it could happen, if I hadn't already heard the question
"Why do more people die on their birthday than any other day of [the year]",
but at least now I know it does.

Maybe somewhere I'll find its cause explained in more detail that "it's a
statistical quirk" due to varying birth rate over a year.


You are still mangling the day you die in with the day you were born on.

The day you die on irrelevant to the statistical quirk.

If you say that the date when you die is irrelevant to the statistical
quirk, are you sure you're still talking about original question. We're
looking at the spike when birth date and death date are almost the same
which peaks when they are exactly the same. We can't compare birth and death
dates, to see this quirk, unless we look at death date as well as birth
date.


I wonder what the magnitude of the statistical quirk peak is, compared with
the completely separate issue that the death rate also varies over the year
(we initially assumed it was constant to keep things simple and to avoid
varying two things at once). I wonder if the variation in death date
sometimes masks the peak around birth=death date that is the statistical
quirk resulting from the varying birth rate over the year.


So maybe the annoying **** at university was right for more reasons than
that death rate varies over the year and that there are social/accident
factors which cause more death around the birthday. Evidently the variation
in birth rate is also significant. Maybe somehow he deduced that a varying
birth rate would cause a spike. If such as deduction (in the absence of your
description of the statistical quirk) is beyond me when I'm in my 50s with A
level maths and university maths-for-engineering courses, then the fact that
he worked it out at 10 makes him a very clever (but still annoying!) ****
;-)