STUMP » Articles » Mortality Monday: When Do People Die? » 7 August 2017, 14:11

Where Stu & MP spout off about everything.

Mortality Monday: When Do People Die?   


7 August 2017, 14:11

No, I don’t mean what age do they die (though, that is somewhat involved).

What time of year do people die? Is there something special about the particular day or month? (I don’t have time of day records, but that would be iffy at any rate. Next week, I’ll look at where people die, and you’ll see you may not get a good time on death given where most people die.

I’m going back to my favorite death records — 2014 U.S. deaths!

And I’m tired, so I’m just gonna smack ya with a bunch of graphs!


So let me take my first cut of the data, which is plain total number of deaths per month. Those are the columns in this graph:

Oh wow! Look how few in February! Of course, February has only 28 days, so I put in a line graph where I did a deaths per day calculation for each month.

Still, WOW! Look at that trend! Summers are awesome! Winters suck!

Okay, that graph is a bit distorted. Notice where the scales begin for each — waaaay above zero.

Here is a more truthful graph:

Not nearly so exciting. Yes, there are more deaths in winter. The rate in December was 17% higher than the rate in August. Not huge, but definitely a real effect.


Thing is, it’s going to be pretty clear the time effect comes from old folks — because they die in greater numbers than young folk.

But let’s see if we can tease this apart.

First, a stacked graph:

Um, that’s not terribly helpful. Yes, old folks die in greater numbers.

Let’s turn it into a line graph:

Ah, so we can now see a clear seasonal effect for old folks in this graph. Indeed, the seasonal effect gets worse the older the population — the biggest difference for those over age 85 is 25% in death count… for those 55-69, that gap is only 18%.

It’s really difficult to see the curves for younger folk, as few die, so let’s break those out into their own graph:

Those patterns are rather odd. It’s difficult to do anything at all with them. But note the numbers — fewer than 100 people per day die in each of these age groupings. Very few young people die compared to the thousands of old people per day.


Now we can do the raw numbers and rates again — luckily, for days of the week, in non-leap years only one day has an extra (and 2014 wasn’t a leap year).

I’m going to do the misleading and the more truthful graphs, one after the other:

Yeah, not much going on there. But the reason not much is going on there is that most deaths are natural (and of old folks), and those don’t show much variation in day of week — only a 2% disparity. Even if I could show a “statistically significant” difference, it wouldn’t be an important difference.

So let me pick out three death types where it does make a significant difference, in the usual, normal way people consider significant.

Far more accidents on weekends, which is no surprise; more suicides on Mondays… which I have no expectation either way; and more homicides on weekends.

You can’t see it well on the graph for homicides, so let me do one of the homicides line by itself.

There ya go. Thursdays are the lowest day for homicides. Sundays the highest — indeed, they’re 37% higher than Thursdays. Indeed, homicides have the strongest day-of-week pattern of all the types. Suicides are next at 26% difference, and then accidents at 23% difference.

Liven up your next bar conversation with these handy facts.

Maybe now you see why I always bring a book with me to the bar.

Underlying spreadsheet (yeah, it’s a mess)

Related Posts
Literal Suicide and Media
Video: U.S. Mortality Trends 2020-2022 part 8: External causes of death
Mortality Monday: Suicide -- Rates