Hacker News Reader

Comment by veidr | Hacker News Reader

Sure, whole classes of problems will disappear as old people die.

But we may not need to wait for grannie and her bridge club to die for this one; TFA notes that support for full-blown legal weed is up 14% in people over 65 since 2011.

View on HN

Note that the average life expectancy in the US is 78.6 years old. 79 minus 65 is 14 years. So in two years, we can expect that 2/14ths of the population over 65 will have died, or ~14% of that population. A 14% increase in the population over 65 in favor of legalization in two years does not conflict with the notion that this is a demographic issue.

Life expectancy for someone who is 65 years old is actually 82.5 or 85.2 years (depends on gender) [1]. You are making a very common mistake by using life expectancy for someone born today -- that's a very different thing.

[1] http://www.ssa.gov/oact/STATS/table4c6.html

I was hoping someone would correct me on that. So 84-65 = 19 years, and 2/19 = 10.5% predicted population churn in two years. 14% is 3.5%/10.5%, or 33% higher than the demographic factor alone would predict.

Your calculations assume nobody lives past the age of 84. You may want to rethink that.

It assumes that the average person aged 65 will live to 84, meaning as many will die before as after.

No. It is a prediction that someone who is 65 will live on average another 19 years to age 84. It does not assume "as many will die before as after." The wikipedia article on life expectancy is actually quite good, if you are interested in learning more.

If the definition of average used by life expectancy is the median, then yes, by definition as many will die before as die after.

I hope nobody is actually redefining the word "average" to mean median, because that would confuse all I thought I knew about statistics.

Is there a less allegedly ambiguous term for the "traditional" definition of average, that is, the sum of many measurements of the same characteristic, divided by the number of elements measured?

Mean, surely?

Very true. Thank you.

Average can refer to any of the common measures of central tendency, as Wikipedia calls them: https://en.wikipedia.org/wiki/Average#Harmonic_mean

Average intelligence, for example, refers to median intelligence (equivalent to an IQ of 100). Mean intelligence, however, is likely the same as median intelligence due to its normal distribution.

I'm way too stoned to follow this conversation...

Not quite: http://www.ssa.gov/oact/STATS/table4c6.html

That can still be a demographic effect. The people over 65 today are not the same people who were over 65 two years ago.

The correct way to control these statistics is by birth-year cohort, not age.

Reply on news.ycombinator.com