Originally Posted by
Adam Strange
The inaccuracy of this test just illustrates the problems with doing predictive analysis without accurate feedback from past experiences.
I'm sure that "the men in the boats" meant something to the person who devised the test, but what that might have been is left to speculation.
Sort of like me referring to my cousin Beth and, of course, what she specifically means to me should be intrinsically divined by everyone else.
The most impressively accurate test I ever took was one which was intended to identify radicals in the incoming classes to the University of Michigan. It was billed as a "personality" test, but what it really was looking for was anyone who would cause trouble for the University.
The way it worked was it asked you about 500 random questions, none of which were specific to personality or politics. Questions like, "Do you like cooked carrots, or raw?" "Do you take stairs one at a time, or two at a time?"
Individually, the questions told you nothing important about a person. It was the pattern by which you answered them that the testers could correlate with other, known people, some of whom were agitators. They could say that "this person answered 87% of the questions exactly like these other troublemakers."
And it worked.
A friend of mine found my files one day and told me that they predicted that I wouldn't finish school, and I nearly didn't. It took five years, but I got a degree.
It said other things about me, but I've forgotten most of them. The testers gave us results for five or so psychological categories, but the test actually had about forty results.
I've always admired this approach, although it requires continuous refinement and updating of the questions as the culture changes.