Enneagram test, and thoughts on Gladwell and the MBTI

Another one of these personality tests: the Enneagram test, more about motivations than the MBTI, which is about behaviour. Kottke mentions that there was a Malcolm Gladwell article on the MBTI in a recent New Yorker, which I must have missed - scouring my archives now.

Edit: found it, read it. I love reading the redoubtable Gladwell - often his articles are a tour de force of ideas. One part intrigued me:

"Myers-Briggs has a large problem with consistency: according to some studies, more than half of those who take the test a second time end up with a different score than when they took it the first time."

Clearly this is, as Gladwell suggests, a function of the dichotomous scale (E vs I, N vs S, etc.). The question is: does it just mean that some people are on the borderline between, say, E vs I, and slip across fairly often, or does it mean people are really very different in their expressions of their personalities at different points in time? I'm inclined to think it's the former, which means if instead of grouping you as an E or I, they give you a score on the E-I continuum (as some of the tests do), it would probably be much more revealing. Which is to say, I think people's personalities are variable, but within certain ranges. Gladwell makes a case for both the former and the latter, arguing that the tests are not very good at making predictions of behaviour. However, this seems to be an assertion of fact (based on himself and Sandy Nininger, the arts-loving lieutenant who was the first American soldier to be awarded the Medal of Honor in the Second World War) rather than empirically substantiated - would have loved to see more proof beyond his citing of Walter Mischel's critique of personality tests.