The Speculist: Life Expectancy and Longevity


Live to see it.

« FastForward Radio -- The Technological Singularity | Main | Friday Videos -- It's Hot »

Life Expectancy and Longevity

Headline from AP today
"Life expectancy in US up"

I see headlines like this all the time and inevitably the article will then make a complete hash of the data.
In this case -

The new U.S. data is a preliminary report based on about 90 percent of the death certificates collected in 2007. It comes from the National Center for Health Statistics, part of the Centers for Disease Control and Prevention.
Life expectancy is the period a child born in 2007 is expected to live, assuming mortality trends stay constant. U.S. life expectancy has grown nearly one and a half years in the past decade, and is now at an all-time-high.

So if I am reading this right- people who died and were issued a death certificate in the US in 2007 are the data base. But the analysis claims to have anything useful to say about the life expectancy of someone born in 2007.

Even one more simple level of analysis would be useful
Most of the people who died at age 78 in the US in 2007 were born in 1929. Not all of them were born in the US. Not all of them lived in the US. And some were born in 1928 - but died in 2007 before their birthday.

But according to the Centers for Disease Control in 1929 the life expectancy in the US was approx 58 years, a little longer for those who made it out of infancy. So what that means is that the kids born in 1928 who actually lived to 2007 increased the predicted expectancy dramatically.

But here's where it gets really dumb. Just to choose one, well documented example- the polio vaccine wasn't successful until 1952, just when are dat set who survived WWII were becoming young adults and getting ready to make the baby boom. There are other more significant life-saving improvements but the point is that none of them were known in 1929.

Likewise- to look at death certificates in the US in 2007 and use the data to imply that you know anything useful about the life expectancy for anyone born in the US in 2007 is a really strained analysis.


I think it all hinges on these words:

"...assuming mortality trends stay constant. "

So I take it that 2007 is one of many data points they've collected over the years. Of course, all previous data points depended on data from people born BEFORE 1929, so that data is even more suspect for the reasons you cited.

By the time the demographers start to suspect that their growth curve might be too linear (and that it's 80 years out of date) we may have already hit actuarial escap velocity.

It's not fair to say that "all previous data points depended on data from people born BEFORE 1929." In a given dataset, some data points are above the average and some are below. In other words, some of the people who died in 2007 were born before 1929 and some were born after 1929.

I don't presume to know their algorithm, but I assume the life expectancy figure is computed by comparing the percentage of people who died at the age of 10, 20, 30, 40, etc.. in 2007 compared to what those percentages were a year ago, five years ago, ten years ago, etc. and then extrapolating what the percentages will be for babies born today.

Obviously, major medical breakthroughs (like the polio vaccine) would cause huge shifts in these numbers, making the prediction impossible to do. That's why, I presume, they say "assuming mortality trends stay constant."

So, to restate in a (perhaps) slightly better way: without any significant medical breakthroughs, the average baby born today will live to be 78 years old.

As with most statistics, explaining it is just as hard (if not harder) than calculating it...

"actuarial escape"

Well- here it gets interesting in both directions.
A lot of Americans died in WW2. Predominantly healthy, white men. Actuarials have the statistical chops to smooth that impact- but in 1982 they actually smoothed the data down. Meaning, they factored in the probability if WW3 occurred, NBC weapons would be far more lethal and the resulting death toll much higher.

That said- life insurance premiums are too high because they underestimate longevity based on being backward looking (and they exclude war, and other non-normal deaths)

I started looking at some statistics and most have us behind places like Spain, Greece, Jordan, Israel, Netherlands, and the UK, in life expectancy. Maybe waiting in line for health care gives you something to look forward to?

Estimated average lifespans for babies born in 2007 are just that, estimates. And these estimates are based on past trends, including the lifespans of those who died in 2007, and also including projected trends in the art of medicine.

Such estimates can't account for abrupt future changes including terrorism, epidemics, a world war, or extreme life extension technologies.

Obviously, no one really knows how long those babies born in 2007 will really live on average.

So, the reporter was dealing with the data at hand.

Yeah- Sally, dealing with the data at hand weakly.

Even making one simple step to a next level would have been useful. At the least, a comment pointing toward better analysis would be welcome .

Harv --

The CIA posts the most complete list I could find:

I wonder how much of this is interventionist healthcare and how much is preventative / lifestyle stuff? Also, I wonder if our ethnic diversity tends to push the US towards the middle of the list? The bottom of the list is just heartbreaking. Note how Africa just takes over as you move down, and how short life gets by the time you reach the end.

I think you may be extrapolating too much from the (undoubtedly) inadequate AP explanation. I'm sure that the National Center for Health Statistics uses a far more complex formula than the AP spelled out, almost certainly taking into account the factors you mentioned, and more.

The fact is, all predictions of the future are based on past trends, and subject to error as circumstances change. For instance, a massive nuclear war or pandemic, or conversely a major medical breakthrough, would throw all predictions out of whack. But I think the more reasonable among us wouldn't fault the predictors for missing out on those. Predictions are by their nature speculative, and as the financial disclaimers always say, "Past performance is no guarantee of future results."

As for U.S. life expectancy, a relatively high level of violent deaths among lower-income young males is a big factor driving down the overall numbers. Wider availability of government-provided health care would do little to alter that. The same goes for relatively poor lifestyle choices among many Americans. And don't try arguing that more preventative health care would make a significant difference -- studies reveal that this is nothing but a myth.

Post a comment

(Comments are moderated, and sometimes they take a while to appear. Thanks for waiting.)

Be a Speculist

Share your thoughts on the future with more than


Speculist readers. Write to us at:

(More details here.)



Powered by
Movable Type 3.2