Log in

No account? Create an account
Thoughts Online Magazine
Collected Articles on Culture & Politics
Bad Statisticians at Oxford 
6th-Mar-2009 08:48 am
In this case, a really pathetic study on drinking and cancer, whose upshot, as the article properly notes, is that you should be cautious about assuming that widely reported studies contain a grain of truth unless your prior intuitions confirm it.

Drink up, ladies.
6th-Mar-2009 05:50 pm (UTC)
Who commissioned the study? Rotten statistics are usually a sign of a study created with a result already determined.

6th-Mar-2009 08:43 pm (UTC)
Cancer Research UK evidently has a thing for meta-analysis, cohort studies, and "collaborative reanalysis" -- in other words, taking the numbers out of where they were generated and pretending that you had designed one common experiment that included them all as data, and drawing conclusions by assuming that the definitions, assumptions and methodological differences don't matter. I've learned to beware of all of these forms of "x is associated with" conclusions, and my teachers would have been chorusing "correlation is not cause" to their authors.

The real cause is some form of asceticism, a determined horror of the thought that someone might be enjoying themselves instead of taking life seriously.
This page was loaded Oct 15th 2018, 10:43 am GMT.