home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!wupost!waikato.ac.nz!maj
- From: maj@waikato.ac.nz
- Newsgroups: sci.math.stat
- Subject: Re: Standard Deviation.
- Message-ID: <1992Aug18.151741.10172@waikato.ac.nz>
- Date: 18 Aug 92 15:17:41 +1200
- References: <1992Aug14.172833.11844@cbfsb.cb.att.com> <c48nbgtf@csv.warwick.ac.uk> <thompson.714070323@daphne.socsci.umn.edu> <44620@ucbvax.BERKELEY.EDU>
- Organization: University of Waikato, Hamilton, New Zealand
- Lines: 29
-
- I'd just like to add a couple of points to the many that have
- been made. (More in the direction of people who are mathematically
- sophisticated, but with little exposure to statistics. Unfortunately
- this class remains sizeable.)
-
- Firstly the ' n - 1 ' is really the dimension of a vector space:
- the vector space of contrasts in the observations. In more complex
- situations other numbers will appear in place of n -1. For example
- if the observations come from several groups we are often
- interested in the within-group contrasts. Here we head towards the
- idea of 'degrees of freedom'.
-
- Secondly the ' n ' divisor is the one which makes the standard
- deviation a functional on the space of empirical distribution
- functions: if you take each observation twice the e.d.f. does not
- change, and neither does the n-definition standard deviation.
-
- For a lot of work in the theory of robust statistics and diagnostics
- it is convenient to have statistics as functionals.
-
- Both definitions are here to stay, as each leads to convenient
- theory in different contexts.
-
- --
- Murray A. Jorgensen [ maj@waikato.ac.nz ] University of Waikato
- Department of Mathematics and Statistics Hamilton, New Zealand
- __________________________________________________________________
- 'Tis the song of the Jubjub! the proof is complete,
- if only I've stated it thrice.'
-