Sunday, August 29, 2004

On the merits of normality

In statistics, it is often quite convenient if a population is normally distributed. In multivariate statistics, special reverence is reserved for those joint distributions that are "IID normal"--independently and identically normally distributed.

When it comes to humans, however, normality is much overrated. Have you ever thought about the reality that "normal" is, in truth, merely a synonym for "mediocre?"

As I am fond of saying, "Being normal is not a badge of distinction." My "Words of Wisdom" page contains additional pontifications.

Back when I was in graduate school, a bunch of us were discussing eccentricity in academia. I said something to the effect that there are few, if any, normal people in academia. "True," said one of my classmates, "but some are stranger than others."

"Well," I said, "if you do a log transformation, I am not that much of an outlier."

The response: "I just proved my point!"

No comments: