It's merging two different statistical blocks that are not related.
The younger one has a decreasing "percent chance of dying each year" because infants are more likely to die of any number of conditions (including bad genetic lottery: being not capable of maintaining life outside the womb) - pre-adolescents have the lowest mortality rate.
The older group has an increasing chance of dying each year due to such causes as teenage low-impulse control, giving birth, progressively weakening immune system, cancer, war, etc.
In other words: They die for different reasons, and taking a total average instead of separating out the populations is like saying "The average American has one ovary and one testicle" - which is statistically very close to true, but doesn't describe the experience or reality of any real person. Or like saying, "The average person has fewer than 2 but more than 1 hands" - also true, but meaningless - where a better analysis would be "X% of Americans have fewer than 2 hands, and of those, Y% have no hands at all."
Similarly, the standard "life-expectancy at birth" is useful only for whole-population averages, as in "Estimate number of children born live per woman in France in 1752". And that still doesn't take into consideration such factors as nuns or old maids who never give birth, or the birth-count disparity between peasants and nobility.
If you want to trot out examples of how long people lived, you have to focus on a single distinct population - only count population after the inflection point in the "percent chance of dying this year" curve, or better yet, separate out men from women, and separate out women who died in childbirth and men who died in warfare, separate out instances of widespread plague or famine, etc. The more you generalize, the less you know.
![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png)
zaluzianskya and ![[community profile]](https://www.dreamwidth.org/img/silk/identity/community.png)
skylark_photos
Bad statistics
Date: 2009-11-06 05:40 pm (UTC)The younger one has a decreasing "percent chance of dying each year" because infants are more likely to die of any number of conditions (including bad genetic lottery: being not capable of maintaining life outside the womb) - pre-adolescents have the lowest mortality rate.
The older group has an increasing chance of dying each year due to such causes as teenage low-impulse control, giving birth, progressively weakening immune system, cancer, war, etc.
In other words: They die for different reasons, and taking a total average instead of separating out the populations is like saying "The average American has one ovary and one testicle" - which is statistically very close to true, but doesn't describe the experience or reality of any real person. Or like saying, "The average person has fewer than 2 but more than 1 hands" - also true, but meaningless - where a better analysis would be "X% of Americans have fewer than 2 hands, and of those, Y% have no hands at all."
Similarly, the standard "life-expectancy at birth" is useful only for whole-population averages, as in "Estimate number of children born live per woman in France in 1752". And that still doesn't take into consideration such factors as nuns or old maids who never give birth, or the birth-count disparity between peasants and nobility.
If you want to trot out examples of how long people lived, you have to focus on a single distinct population - only count population after the inflection point in the "percent chance of dying this year" curve, or better yet, separate out men from women, and separate out women who died in childbirth and men who died in warfare, separate out instances of widespread plague or famine, etc. The more you generalize, the less you know.