Suppose That Answering "True" to the Question "I Drink a Lot

Multiple Choice

Suppose that answering "true" to the question "I drink a lot of water" happens to help predict depression. Would it be wise to include this item on a test used to identify depression?

A) yes, because the essential characteristic of a good test is that it predicts relevant behaviors
B) no, because there is no theoretical link between drinking water and being depressed
C) yes, because there is a theoretical link between drinking water and being depressed
D) maybe, depending upon the theoretical orientation of the test developer