On: * cures (or causes) cancer

[joe_the_user]

Statistics should only be applied to fairly uniform and understandable phenomena.

Applying statistical likelihood in this instance is stretching the bounds of statistics considerably more than The Daily Mail stretched them.

[nevinera → joe_the_user]

No, applying statistical methods to this instance is unreasonable. Observing that selection bias skews the probabilities involved is perfectly reasonable, and indeed almost necessary to rational discussion.