MAD in my vocabulary (that is, as used in the books and so forth that I usually meet) is 'mean absolute deviation' - and 'deviation' is always interpreted as 'deviation from the mean'. AD has no meaning, since the word 'average' is ambiguous. MD is 'mean deviation', and hence 'mean deviation from the mean'. (And hence zero!)
If terms are used in full, the suggested 'MAD' is 'median absolute deviation from the median' (and this is a very useful measure of variability in some circumstances). But calling it MAD is confusing......
Drives you mad, doesn't it?
Alan
On Tuesday, March 4, 2003, at 12:50 AM, dennis roberts wrote:
At 11:23 PM 2/28/2003, John Poole wrote:Alan McLeanThanks Jason. Here is a related reply, which I am forwarding to the list. -- John
From: [EMAIL PROTECTED] (Jim Jaccard) Newsgroups: comp.soft-sys.stat.spss
John -
One of the more common indices of variability used that is reasonably outlier resistant and that often is reported in conjunction with the median is called the Median Absolute Deviation statistic or MAD.
of course, the more symmetrical distributions are .. the more MAD is like AD ... average deviation ... finding average of absolute deviations around the MEAN
even if the median is a bit different than the mean ... MAD and AD are essentially the same quantities
..
..
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
.. http://jse.stat.ncsu.edu/ .
=================================================================
. . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
