How do I figure out mortality statistics for people born in 1890?

April 22, 2008 9:40 PM

I'm trying to figure out what percentage of people born in 1890 were still alive in 1920 and then 1950 and so on, until we can say for certain that every person born in 1890 is 100 percent definitely dead. I've looked at the 1890 and 1920 census, but they don't have mortality tables.

(I'm specifically looking for values for 19th century populations. Not necessarily 1890, but in that ballpark...I figure 1890 is easier than 1891 because there was a census in 1890.)

I'm mainly curious about how WWI and WWI and other major events shaped the mortality curve -- for instance, statistically were __% of men born in 1890 likely to be alive in 1917, but really only __ % of men born in 1890 were alive in 1917 because of the war?

I'm interested in statistics for both world and US populations. I know there are actuarial mortality tables for questions like this, but they don't extrapolate backwards, as far as I'm aware. If they do extrapolate backwards, then I don't know how or where to access these statistics.

posted by melodykramer to grab bag (12 comments total)

This is for something I'm working on with WWI vets specifically. I want to be able to say what kinds of odds the few left have beat to become the last remaining survivors...
posted by melodykramer at 9:46 PM on April 22

The census records ages, right? So if the 2000 census lists nobody older than 109, then everybody born in 1890 is dead.

Of course, it's not really that simple because the census doesn't count every person, but I'd wager it's still a pretty good estimate. You might ask a statistician what kind of distribution lifespans follow; if that's been constant over the years, you can use that combined with census ages to get a probability.

Come to think of it, you could build that distribution yourself from the census by fitting a curve to the series {people born in 1890 still alive in Y, for Y=1890,1900,1910...}
posted by qxntpqbbbqxl at 10:06 PM on April 22

If you're looking for odds of of vets, I think it's just as simple as taking the total that served for a given country and dividing by the total survivors. For the US, if there were 4.7 million and one survivor, the odds would be 4.7 million to 1 of being the last.
posted by Yorrick at 10:12 PM on April 22

On this page you'll find reports of mortality statistics by age. Note: 1917 weighs in at a hefty 72.8MB. All the age related tables (except for the one on page 11*) have this rider on them though: [This table excludes the deaths of soldiers and sailors after the United States entered the war.], so I don't know if it will suit your requirements.
*This table says - Death from all causes in the 25 to 29 years bracket was 41,732. Incidentally the deaths of under 5's was 243,708!
posted by tellurian at 11:06 PM on April 22

I just got the 1905 tables. Death from all causes in the 15 to 19 years bracket was 50,135.
posted by tellurian at 11:21 PM on April 22

You want the publication called 'Vital Statistics Rates in the United States 1900-1940' for the second one. I have the giant PDF on my computer, and I think it came from the NIH website somewhere, definitely not the census bureau.

Mortality rate is relatively simple to calculate if you have the bulk statistics. The 1900 census statistics (the file names are horrible - 13982433v4_TOC.pdf for instance) which are available on the census website has a chart, somewhere in the file that ends v4ch1.pdf, of mortality from 1850-1900.

That being said, you don't just want mortality tables - you want male mortality tables, and specifically you might want age-restricted data. I have to go to work, but I have 1850-1950 male mortality calculated for one state, for instance, for men 20 and over (I needed it for my thesis.) Briefly, though, the CDR for Connecticut males looks like this:

CDR all CT SDR Males 20+
1850 15.59 15.776
1860 11.11 **
1870 12.64 14.073
1880 14.74 15.197
1890 19.39 19.952
1900 16.98 17.946
1910 15.60 16.219
1920 13.60 13.550
1930 10.70 12.686
1940 10.50 14.242
1950 9.60 12.988

The 1860 census can go to hell, by the way. Mefi mail me if you want more info.
posted by cobaltnine at 4:19 AM on April 23 [1 favorite]

Gah, morning. Sorry, you don't want CDR, you want the actual tables, but the Age Specific Death Rate shows sort-of what you need. If one wanted to calc an Age Specific Death Rate for males between 18 and, say, hm, 45 or 50, which is your population, you could use that, too.
posted by cobaltnine at 4:35 AM on April 23

@ Yorrick -- Yup, that'd be the final answer -- but I'd like to say "___ % of men were alive in 1960, 1980 etc.

@ tellurian -- do you think the under 5's would be similar for 5's in 1895?

@Cobaltnine - Thanks. I'd like women too, because the person I'm writing about lives a lot longer than they do, too.
posted by melodykramer at 5:50 AM on April 23

Why not just take several random samples of ages from the census records, average and extrapolate? It can be a valid statistical method.
posted by JJ86 at 6:01 AM on April 23

It might be worth knowing that most of the 1890 census, having been destroyed by fire, is missing, so that might not be the best primary source of data for you.
posted by timetoevolve at 9:35 AM on April 23

(That is, the original schedules - quite a bit of information is still available.)
posted by timetoevolve at 9:37 AM on April 23

do you think the under 5's would be similar for 5's in 1895?
The next set of figures available on that site are for 1900:
Death from all causes in the 5 to 9 years bracket was 36,748
Death from all causes in the 10 to 14 years bracket was 24,500
posted by tellurian at 6:58 PM on April 23