Demography

Life Expectancy Reported Down, with multiple reasons

I’ve seen another release about the US life expectancy dropping a year during 2020 – but this one didn’t credit Covid exclusively.  It pointed out that the US Life expectancy has been dropping for several years due to an increase in drug overdoses and suicides.  Please remember – causality is inferred, not statistically proven.

Covid, with most fatalities occurring among the the oldest, has a hard time reducing the life expectancy by a year. (Social Security has its work on life expectancy, going back to 1940, another table, for life expectancy at specific ages, is available at here)

The article reminded me of the drop in life expectancy that followed the end of the Soviet Union.  That was credited to alcohol overdoses, violent death, and suicides.  The chart shows that it happened there, so it can happen here.  The thing about the calculated life expectancy is that one 21-year-old male death takes 55.91 years from the life expectancy chart, while a 75-year-old male death takes only 11.14 years from the collective pool.

The Soviet figures suggest that a major economic or governmental change can have some immediate changes – though today’s Russians, who made it through the collapse of the Soviet Union were back on track in 2019.  CDC has released data showing excess US deaths in 2020, but they are by state and weekly.  Hopefully they will condense the data – 50 states and 52 weeks make a spreadsheet that takes a lot of effort to get through.  Summing up the data to one nation and one year will make it a lot easier to comprehend,  The data that is currently available is at this link.  It is interesting to look at – and I expect that they will have it compiled at a national level soon.

A Science for Everyone, Demography

Death Rates by Country

One of the more useful publications to compare nations is the CIA World Factbook.  While we tend to think of the CIA as secret agents, a lot of them are data geeks crunching numbers.  The data they develop about each country is impressive, and like the US Census, the CIA sets the standard for the most accessible and reliable information.  When I started using it, I needed a land-grant college library.  Now, I click World Factbook.

National death rates in 2018 ranged from 19.3 per 1000 in South Sudan down to 1.6 per 1000 in Quatar.   The reasons vary – a higher median age (Japan is 48.36) combined with healthy living and good health care can still have relatively low death rates (Japan was 9.9 in 2018).  The explanation is Demographic Transition theory – in the old days we had high birth rates and high infant/youth mortality.  The second stage occurred with health care improvements – birth rates remained high, but death rates dropped.  Stage 3 showed lower birth rates and death rates continuing to drop, but more slowly.  The fourth stage maintains the lower birth rates, but in an aging population the diseases change – in the US, the big killers are heart disease and cancer.

Lesotho, in Southern Africa, has the second highest death rate – high infant mortality (44.6 deaths per 1000 births), the world’s second highest HIV rate.  A dozen years ago, I first encountered https://www.worldlifeexpectancy.com/ and the website gets increasingly useful.  It isn’t that the covid is so insignificant in Lesotho, it’s that Diarrhea is so much more prevalent.  Click the link – and check out the demographic factors for your own country.  In the US, it shows life expectancy changes since 1960:

US life expectancy from World Life Expectancy

The personal computer has taken demography from being a science that need a major university’s library facilities in my undergraduate days into being a science with the data available to a Fortine resident who has insomnia at 3:00 am. 

Community, Demography

The Useable data on Covid

Figuring out the data to use is important.  On May 9, 1864, Union General John Sedgwick said “They couldn’t hit an elephant at this distance” three separate times.  After the third statement, he became the highest ranking Union officer to die in battle.  He was missing a couple relevant pieces of information – first, the Confederates had snipers with Whitworth rifles, and second, at 800 yards, the Whitworths were making 12 inch groups – far smaller than an elephant.

It’s similar with Covid – each of us is like General Sedgwick, not knowing which piece of data is relevant, or even exists.  Some data ties in blood types.  I’m not certain it is data we can use.  Death and age combine to give us the most solid data that we have – and it may be of some use.  Statista provides this information, as of January 9. 2021, for the US.

AgeNumber of Deaths
Under 134
1-421
5-1455
15-24510
25-342,196
35-445,742
45-5415,558
55-6438,830
65-7470,230
75-8490,744
85+105,673

As you look at these numbers, it’s probably worth remembering that over 50 million Americans are over 65 years old – where most of the casualties are.  Soon we’ll have the 2020 census, but until then the 2010 census data is usable – a bit low, but usable.  The tragic loss of 55 kids under 4 comes from over 20 million population in the 0 – 4 age cohort.  The 105,673 deaths of people 85 and over comes from a cohort of 5½ million.

If you want to calculate the years of life lost to covid, the social security life tables are available at: https://www.ssa.gov/oact/STATS/table4c6.html  Still, we’re dealing with big data – and, as I realized when I was being treated for colon cancer, the most important thing about your life expectancy isn’t the number for your age, it’s which side of the median you’re on. 

The data shows that Covid’s deadliness increases with the age of the person it infects.  The data isn’t adequate to show either the probability of being infected or missing the virus.  It’s easy to describe the statistics of the Spanish Flu – but most of the data was compiled and available by 1920.  This article describes how Gunnison, Colorado isolated their way out of the Spanish Flu: Isolation worked well for Gunnison. 

The folks at APM Research Lab have calculated out death rates by race – and their study is worth a look, despite the problem that race is more a social construct than genetically identifiable.  Here’s some of their data (as of January 5).

“These are the documented, nationwide actual mortality impacts from COVID-19 data (aggregated from all available U.S. states and the District of Columbia) for all race groups since the start of the pandemic.

  • 1 in 595 Indigenous Americans has died (or 168.4 deaths per 100,000)
  • 1 in 735 Black Americans has died (or 136.5 deaths per 100,000)
  • 1 in 895 Pacific Islander Americans has died (or 112.0 deaths per 100,000)
  • 1 in 1,000 Latino Americans has died (or 99.7 deaths per 100,000)
  • 1 in 1,030 White Americans has died (or 97.2 deaths per 100,000)
  • 1 in 1,670 Asian Americans has died (or 59.9 deaths per 100,000)
  • Indigenous Americans have the highest actual COVID-19 mortality rates nationwide—about 2.8 times as high as the rate for Asians, who have the lowest actual rates.

Data is data.  There’s not a lot of difference between Latino and White rates. 

Community, Demography

Easy Math but Fake News

Yesterday, I read that US Life Expectancy had dropped by a full year due to Covid.  I didn’t really think about it – I had taught about the drop in life expectancy accompanying the Spanish Flu, and had invented hypothetical plagues for student exercises in demography class.  But when I had the full-year drop in life expectancy cited to me a second time, I realized that large numbers keep us from checking the math, even when the data is readily available.  Here’s the basic math for checking the assertion, worked as we would have in the slide rule era.

The US population is just a little under 330 million.  At present there are approximately 400,000 Covid deaths.  Using the Social Security life expectancy tables was a good decision – the data is readily available to check your work . . . but we don’t need complex math to check the claim that US life expectancy will drop 1 year due to covid.  It’s probably worth mentioning that life expectancy is a statistical thing, accurate for a large group but not particularly accurate for an individual.  I’ve known people who lived past 100 and others who died at 14.  At age 12, they had similar chances to live to old age.

To reduce US life expectancy by one year, Covid would have to take away 330 million years of life (remember, there are 330 million people. If each loses one year…)

This is possible, but to make the math easy, lets state the problem in millions to get away from the tyranny of large numbers. .  We’re left with 330 for population, and 0.40 for deaths.  To reduce US life expectancy by one year, we have to have 330 (million years of life) lost by 0.40 (million people).

US PopulationCovid Deaths
330,000,000400,000
330 million.4 million

Checking the math is nothing more than setting up a word problem: How many years of life are lost for each covid victim? Can there be 330 (million) total years of life lost with 0.40 (million) deaths due to covid?

Well, 330 years of life lost divided by .40 is: 825 years lost per covid death. That implies the average Covid death deprives its hypothetical victim of 825 years of life.  Since average life expectancy is now about 80 years, it looks like several orders of magnitude were lost in someone’s calculations.  The old slide rule techniques still have value in checking one’s work.

The same day, another stats guy ran numbers showing that the average Covid death was 13 years early.  That seems to have a bit more face validity – we can go to the charts that show death rates by age, develop percentages, and check his data against the tables – but I’m still making the math easy:

400,000 Covid deaths X 13 years = 5,200,000 lost years of life, or 5.2 million
5.2 million (lost years) divided by 330 million (population) = 0.0158 years of life expectancy per individual. 
0.0158 X 365 (days in a year) = 6 day drop in life expectancy.

The availability of data makes it possible for demography to be a science for everyone, and not confined to university campuses.

Demography

Musical Life Expectancy

Over 20 years ago, a study was published about the life expectancy of saxophone players. It found that playing saxophone correlated positively with a significantly shorter life expectancy, and suggested that it might be caused by circular breathing and posture – but the data just showed correlation.  Of course, regardless of the quality of statistical data, causation is inferred.

A 2015 article from the Conversation showed differing life expectancies for different styles of music.  The graph was impressive:

It looks like jazz and blues are the healthiest genres, while Rap musicians tend to die young.  Still, the graph misses an important element – time.  Rap and Hip Hop started in the seventies, while Blues started a century earlier, and Jazz wasn’t far behind.   The data are skewed, and it isn’t surprising that Blues musicians have lived longer than Rap musicians.  It’s hard to be an old dead musician in the newest genre.

Music and Longevity (2014) looks at the mean age at death of nearly 9,000 musicians, and concludes that harpists live longer (average 80.9) and guitarists have the shortest life expectancies (54.4).  Still, looking at the data, Zharinov and Anisimov included 32 harpists and only 9 guitarists. 

I know a lot more guitar players than harpists, so either the sample is skewed or Keith Richards longevity is normal for guitarists.