Joel Aelwyn wrote:
On Thu, Feb 17, 2005 at 11:53:47AM +1000, Anthony Towns wrote:martin f krafft wrote:What would be the point? That would tell us nothing about causation, which is the question at issue.Maybe we should tabulate most commonly bashed roles and see if there is a correlation with inavailability of information?The assertion that causation and correlation are never tied is false.
Good grief.Correlation tells you nothing about causation. It tells you that two things often occur together, so that when you find one, you'll often find the other.
Or, in more homily terms: If there's smoke, there may not be fire, but you'd still be well advised to grab a fire extinguisher and check it out just in case.
In particular, a correlation does not tell you which of the two events was the cause -- or even that there isn't a third event that was the cause. In Brisbane, high electricity use and high ice cream sales are correlated, but the underlying cause is hot weather, so manipulating either one won't assist with the other.
In the smoke versus fire case, you need to be careful that in looking at the correlation instead of the causation, you don't find yourself recommending people install exhaust fans instead of fire extinguishers.
Cheers, aj