Lefty
New member
- Joined:
- Apr 19, 2010
- Posts:
- 2,241
- Liked Posts:
- 780
Was thinking about it, and here's a quick example to help you (all of you, really) understand why correlation does not necessarily indicate causation:
Many government buildings are closed on days beginning with the letter "S". If you were to graph the number of government buildings closed as a function of days beginning with the letter "S", you would find a very definite correlation (the correlation coefficient would be close to 1).
Now, is the closing of government buildings caused by certain days beginning with the letter "S"? No, the buildings are closed because it's the weekend, not because of the way the days of the week are spelled.
See? Correlation does not mean causation. It can imply causation, but you need to look deeper at the problem (usually by way of a regression analysis, which I was at one point able to perform, but not any longer).
Many government buildings are closed on days beginning with the letter "S". If you were to graph the number of government buildings closed as a function of days beginning with the letter "S", you would find a very definite correlation (the correlation coefficient would be close to 1).
Now, is the closing of government buildings caused by certain days beginning with the letter "S"? No, the buildings are closed because it's the weekend, not because of the way the days of the week are spelled.
See? Correlation does not mean causation. It can imply causation, but you need to look deeper at the problem (usually by way of a regression analysis, which I was at one point able to perform, but not any longer).