home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!sun-barr!ames!purdue!yuma!csn!boulder!ucsu!yertle.Colorado.EDU!mcclella
- From: mcclella@yertle.Colorado.EDU (Gary McClelland)
- Newsgroups: sci.math.stat
- Subject: Names for Joint Central Moments?
- Keywords: central moments, joint distributions
- Message-ID: <1992Jul25.180635.25360@ucsu.Colorado.EDU>
- Date: 25 Jul 92 18:06:35 GMT
- Sender: news@ucsu.Colorado.EDU (USENET News System)
- Organization: University of Colorado, Boulder
- Lines: 40
-
- I recently submitted a statistical article to a psychological journal.
- This article made use of various joint central moments of bivariate
- distributions. One of the reviewers expressed the opinion that while
- many readers of that journal would know what a "Kodak moment" was,
- very would have any intuitions about the joint central moments I was
- using. I've looked without success for names or intuitive explanations
- for various joint central moments. Can anyone on the net help?
-
- More specifically, here is the problem. For a bivariate distribution
- of X and Z, we can define joint central moments (using the notation of
- Kendall & Stuart) as
-
- mu(j,k) = E[(X-Xmean)^j (Z-Zmean)^k]
-
- So mu(2,0) and mu(0,2) are the univariate variances and intuitive
- explanations are available. Further, mu(3,0) represents skewness and
- mu(4,0) reflects kurtosis. For the bivariate, mu(1,1) is the
- covariance and it is fairly easy to explain that to the "Kodak moment"
- folks. However, I'm really in need of names and/or explanations for
- mu(2,1), mu(1,2), and mu(2,2). Help please!
-
- mu(2,1) and mu(1,2) are related to skewness because one can show that
- if the conditional univariate distributions are symmetric then mu(2,1)
- = mu(1,2) = 0. But does anyone have a name for mu(2,1) or any other
- insights about what it tells us about the joint distribution?
-
- I've informally in my own notes been calling mu(2,2) the "double
- variance" because when X and Z are stochastically independent, mu(2,2)
- equals the product of the respective univariate variances. Anyone have a
- better name or insights about what mu(2,2) tells us?
-
- Any suggestions or references would be greatly appreciated. I'll
- summarize any direct email replies in a subsequent post.
-
- thanks for any help!
-
- gary mcclelland
- univ of colorado
- mcclella@yertle.colorado.edu
-
-