Causation

[Dancing David]

(From a course in research methods my wife is taking.)

The Criteria for Causation

1. That the event labled as causal be located prior to the event labeled as effect.

2. That the effect is subject through testing of dependant and independant variables.

3. That all confounding factors be eliminated or controlled for.

[T'ai Chi]

Originally posted by Dancing David

3. That all confounding factors be eliminated or controlled for.

In theory, sure. However, often in practice you don't even know what these are, certainly not what all of them are.

[Jorghnassen]

Originally posted by jzs
In theory, sure. However, often in practice you don't even know what these are, certainly not what all of them are.

And if you include too many in your model, you might end up not discerning an effect that is really there. Model selection is a real pain.