W
whit3rd
Guest
On Wednesday, September 30, 2020 at 9:15:00 PM UTC-7, Bill Sloman wrote:
> On Thursday, October 1, 2020 at 2:52:50 AM UTC+10, whit3rd wrote:
[about older folk having better mortality in the 1918 flu]
After a century, there have been lots of hypotheses; this one survived testing. You and I should be so lucky.
> On Thursday, October 1, 2020 at 2:52:50 AM UTC+10, whit3rd wrote:
[about older folk having better mortality in the 1918 flu]
It was the statistics on older folk having a better survival probability than the young.
So it\'s an inference from the statistics long after the event. In order to test that kind of hypothesis it\'s nice to work on only part of the data, so that you can check whether the hypothesis that looks good on the data you\'ve been over in detail, still looks good on data that you haven\'t had a chance to cherry-pick in some way.
You don\'t have to be consciously trying to cheat to do this kind of cherry-picking - your sub-conscious will do it for you.
There\'s no need to pick small data sets for the 1918 flu, the large data sets support the hypothesis.
Statistical analysis is not cherry-picking in the sense of using a small sample, in a pandemic
that generated millions of death records.
It depends what you are looking for. If you keep on going through the data twenty times, you are likely to find one 95% correlation and it will probably be spurious.
Statistical analysis isn\'t done by the sub-conscious.
Picking out what to analyse for - what might correlate with what - does depend on choosing what to look at .
When or if another explanation for the age distribution arises, it can compete with the previous-similar-infection
hypothesis.
Ho hum. And the other hypotheses that you should have though of, but didn\'t.
After a century, there have been lots of hypotheses; this one survived testing. You and I should be so lucky.