Hacker News new | ask | show | jobs
by Retric 1114 days ago
Yes and no,

Statistics have an inherent weakness when you don’t have consistent buckets to place everything into. If individual specialists classify cases differently they really can see statistically valid numbers from relatively few cases where double blind studies using different classification criteria see improvements below the noise floor.

1 comments

Also, largely, there is no scientific evidence for the vast majority of problems.

The only evidential basis for causal analysis are experiments where (1) you intervene to bring about the cause deterministically; and (2) where you control all confounding causes.

Those conditions are impossible to meet for the vast majority of interactions in complex systems, such as people & medicine.

If you run associative statistics on counfounded data collected in observational studies, you may as well correlate star signs with outcomes -- no doubt people being treated under winter signs will do worse than those under gemini. Astrology QED.

Brings to mind an interesting tidbit I ran into--long ago your sign probably did correlate with meaningful things in life. Not that the stars mattered, but the seasons did. Early life nutrition would vary. Even today you'll find a correlation between signs and ADHD diagnoses--because of the school year. There is almost a year of difference between the oldest students entering school and the youngest--and those youngest are appreciably more likely to get an ADHD diagnosis.