Reply
Sat 15 Sep, 2012 10:31 pm
Years ago I saw an 'interesting' set of data. The data presented an analysis of stock market (NYSE) performance versus an unnamed variable, x. The data showed an extremely high statistical correlation between NYSE performance and x. It showed that a barrage of statistical tests indicated (presumably) that x 'caused' NYSE. The article concluded with the revelation that x was, in fact, the batting average of the 1912 New York Yankees over some time period.
How can this be? What does that mean? I'm sure you've noticed that the authors are very careful in the way they phrase the results of statistical analysis.....e.g. when you reject an hypothesis you don't conclude that something caused something else. You conclude that "there is or is not enough statistical evidence to reject or accept Ho".
@terps,
Correlation does not mean causation. All sorts of things correlate. If you are using the scientific method, you start with a hypothesis based on your understanding of the underlying principles and then design experiments to try and disprove your hypothesis. Correlation by itself is only useful in giving you clues to look for understanding.
Stumble It!
Tweet This
Bookmark on Delicious
Share on Facebook
Share on MySpace