OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.

office-comment message

[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]

Subject: Re: [office-comment] Re: Gaussian Distribution vs Normal Distribution

• To: office-comment@lists.oasis-open.org
• Date: Thu, 14 Feb 2008 14:50:21 +0100

```Hello Brad,

> > FALSE

It was no meant to shout, I am just used to write some things upper-case
(like TRUE/FALSE, and a lot of other things - never understood why
people understand that's shouting) ;-)

> > "normal distribution": 782,000
> > "gaussian distribution": 573,000
> I don't get that either. Without the quotes, I get David Wheeler's result.
> With quotes

Strange, I tried it today a 2nd time on google.co.uk (I'm in the UK now)
and get slightly different results (all results are rather relative,
"gaussian distribution": 584,000
"normal distribution": 885,000

The problem is "normal distribution" may include many other pages
not related to the gaussian distribution

A much more interesting picture arises when one looks at the literature,
though this is biased, too:

IEEE journals use predominantly gaussian (~2.500 vs ~1.500 articles)

while other journals use more often the "normal" term
(especially in biomedical research)

Now, my strongest argument against the use of the "normal" term
was explained in my previous e-mail:
- recent statistical methods allow us to accurately compute
a distribution for any set of data and any statistic
- therefore eliminating the need for antiquated methods and
concepts
(limited to mean, central limit theorem, sampling,
theoretical distribution not consistent to reality)

We may call this computed and accurate distribution the
*normal* distribution for that particular set of data.

And therefore the confusion would arise. Why call something normal,
when it is NOT normal for that particular set of data? And how to name
the actual distribution otherwise than "normal"?

Why resort to the central limit theorem, when the data set NEVER
arose through random sampling and we will never sample a 2nd data set.

That just makes no sense.

Hope this helps explain my position.
(By the way, the late prof Feinstein had similar views.
http://jech.bmj.com/cgi/content/full/56/5/328)

Sincerely,