Quote:
|
Originally Posted by towny
I think you're mincing words a little here ... surely calling this evidence an 'interesting statistical correlation' is not dissimilar to a newspaper printing a 'clarification' because it doesn't like to use the word 'correction'?
|
Perhaps your comment about "mincing words" does have some truth to it, but the article is from New Scientist so I am therefore looking at the article from the viewpoint of scientific rigour. I can use the term 'interesting statistical correlation', but that only describes the numerical outcome of the results, not the results when taken in context. You can demonstrate many correlations, that even with a strong statistical result do not have any cause and effect.
In this instance, they have merely demonstrated that the twin who smoked cannabis was more likely to go on to taking harder drugs. Without knowing the complete social background for each twin as an individual, nor the incidence of hard drug use in twins as a whole, the results have no meaning.
Consider this example, which I appreciate may be a tad extreme, but serves for illustration. Imagine a situation where 100% of cannabis users then went on to take heroin. Even here you could not argue that it was a gateway drug as you would have no data on how many of those people would have taken heroin even if they hadn`t taken cannabis.