C5 - simpson's paradox
- two potential treatments for kidney stones: A and B
- A more effective than B for small and large stones
- ignoring patient classifications, B is found to be more effective
- the fact that the 'winner' changes when classification is removed is called simpson's paradox
| A | B | |
|---|---|---|
| small | ||
| large | ||
| aggregate mix |
-
clearly, A is best for both small and large stones
-
large cases are more difficult, hence A is reserved for them, therefore there are more studies
-
B is reserved for small cases, which are easier, hence more studies
-
this paradox is caused by a combination of:
- the biggest contributors to the mix are quadrants -- [large, A] and [small, B] -- which involve tests with 1000 patients
- but [small, B] has a higher effective percentage than [large, A] because, although A is better, since it is dwarfed by the fact that small stones are easier