However, if you except the principles of causality, and also except that correlation can be reliably measured, then the demonstration of a correlation does indicate the presence of some underlying causal relationship. It is just that the relationship needn't be as simple as A causes B.
In my fresher stats lectures a million years ago we were told about the near-perfect correlation, over many decades, between the salaries of ministers of religion in New England and the price of cigar tobacco in Cuba. A good anecdote sticks around - as that fine cartoon may.
Correlation does not imply causation.
ReplyDeleteHowever, if you except the principles of causality, and also except that correlation can be reliably measured, then the demonstration of a correlation does indicate the presence of some underlying causal relationship. It is just that the relationship needn't be as simple as A causes B.
ReplyDeleteNeil: "if you except the principles of causality.."
ReplyDeleteExcept? Accept?
Should be 'accept', in both cases. Brainfart on my part.
ReplyDeleteIn my fresher stats lectures a million years ago we were told about the near-perfect correlation, over many decades, between the salaries of ministers of religion in New England and the price of cigar tobacco in Cuba. A good anecdote sticks around - as that fine cartoon may.
ReplyDelete