The phrase “life expectancy” is used all over the place but I have never seen it defined. What does it mean to say that currently life expectancy in some place is 80? I can think offhand of several possibilities:
of all people dying today, the average age is 80
we guess that the average life of a person born today will be 80
some other average
or
any of the above with “average” replaced by “median”
So someone must know what it means. Wiki suggests the second choice above, but that seems just like a guess. Who knows what medical advances or civilization collapse will happen to someone born today?
If I were doing it, it would be the first above, but using median not average.
This definition seems to imply that you measure at birth, that is, life expectancy is the average age a person just born can expect to achieve.
However, the definition on the wiki page is
You see this is very different. If the life expectancy of a population, as measured at birth, is 75, say, what is it for someone aged 80? So you see context matters.
The first definition is useful when measuring the overall health of a population. It is totally useless in estimating how long someone has to live.
Wikipedia is correct in that sense. The life expectancy for an individual goes up every day that they survive but the effect is much more pronounced for passing through certain ages. Making it past childhood alone gives a boost to total life expectancy for an individual because childhood mortality brings down the average as a whole. Once an individual get past that period, the mortality during that period doesn’t apply to them anymore they are expected to live longer on average than a newborn.
A friend of mine who has written a popular statistics text (around reliability) uses this exact example. It is very important to distinguish average lifetime of a product in the field from average expected life given that it has been in the field a certain amount of time. When we had a big infant mortality problem, I saved my company a lot of money (and made my exec VP very happy) by demonstrating that replacing the components in the field would actually hurt the overall reliability of our installed base. Tracking failure rates proved I was right.
It’s not “just a guess”. It’s calculated based on mortality rates (numbers of people dying) in various age groups, and assuming that those mortality rates continue in the future.
However, it’s not the same thing as the mean or median age of people dying today, because the age distribution of the population is distorted due to factors like immigration (or emigration), fertility rates being higher (or less) that that for zero population growth, and mortality rates being different in the past.
I think it’s similar to that, but it can’t be exactly the same, because that’s being used for medical research, where the populations are those who have received various treatments (with possibly a control group that receives no treatment). Life expectancy in the sense used by the OP is for the population of a country/state/etc. as a whole, and is based on mortality statistics for the whole population. A medical research project will have “censored data”, i.e., people who drop out of touch with the project, so you don’t know if they are still alive or dead. That should not happen with general population statistics, or it happens so rarely with mortality statistics that you don’t need to allow for it…
This isn’t correct. Kaplan-Meier is applicable to any non-negative data, and it doesn’t require modification when there’s no censoring. Per Wikipedia on calculating life expectancies, it looks like that’s not what they use, but it’s a reasonable first guess.
You really need to go back and re-read the *Foundation *Trilogy, Hari. Your theory of Psychohistory was a natural outgrowth of 20th-Century actuarial tables. It is similar to radioactive decay—you can never tell when any specific nucleon will decay, but, given a sufficiently large sample, you can say, with some precision, how long it will take for half of them to do so. Life insurance companies make their money by calculating, given the present age of the insured, their health, activities and many other factors, how long they may be expected to live, on average. To say that Mr. Seldon, at 45 years old today, will live X number of years is wrong. It is better to say that Mr. Seldon, and the hundreds of thousands of people of similar age and health can expect to live X number of years.
With hundreds of millions of people in the US and Europe, and collecting data for at least two centuries, they can get pretty darn close to the expected age of decease of any individual. When you get quoted a rate for insurance, the actuaries at that company have cross-referenced and tabulated mountains of data to come up with a figure whereby, on average, they stand to make money.