Wednesday, December 2, 2009

Milo left this comment...

Depends on your point of view. From the tree's fixed position point of view, both are going round the tree; neither is going round the other. From the squirrel's point of view -- always facing toward the center of the tree as he moves around the tree-- the man is always straight staight-ahead: straight through the tree and is not moving around the squirrel. From the squirrel's point of view -- always facing in same direction as he moves around the tree, the man is going round the squirrel.

PS. Your definition of confounding seems a bit off. Unless a confounder is associated with at least two elements in an association, it can't influence the association.

Here is my response to Milo's comments...

Milo - I am amazed someone is actually reads this blog besides myself. Your answer to my challenge is definitely well thought out and makes it seem like I thought of a very clever questions. The last part of your answer concluding that the man is going around the squirrel because the squirrel is always facing the same direction - well, I must confess that answer confounds even my mind, and that is hard to do. So thus I must give you credit for answing this correcly because I am unable to devise a logical response to counteract it.

Now about my defition of confounding - I think you are just trying to give me a hard time with that one. I see no reason why there would need to be more than one other factor besides the confounder in an association in order for the counfounder to influence the association.

For exammple, take the relationship between shoe size and mathematic ability in students Grades K-8 in a school where all the students are given the same test. We would find that is a very high correlation between shoe size and test scores. But in fact the shoe size has no effect on the test scores, it is confounded by age. We would find that if we took only kids the same age and then correlated their shoe sizes with their test scores, no correlation would be found - identifying the confound of shoe size with age in influencing test scores. So here all you need is the one other factor, age, the true factor associated with test scores, to exist in order for there to be a confounding factor shoe size, a false association with test scores.

1 comment: