Re: Percentiles

On Friday 26 March 2004 06:42, John Darrington wrote:
> By this algorithm, the 0'th percentile always equals the smallest
> datum.  Is that what you'd expect.

Yes, that's right. The 100th percentile is the largest.

> In my research these last couple of days, I've only found one
> statement about percentiles with which I wholly concur:
>
>  "There is no universally accepted definition of a percentile"

I did a quick survey of all of my undergraduate research design books.  Just 
less than half avoid the discussion altogether.  The rest use a definition 
consistent with mine.  There was one book that used a completely different 
definition.

So....

At least my definition is consistent with SPSS.  I checked R and MATLAB and 
they both do it the same way.

- M.

>
> This comes from http://cnx.rice.edu/content/m10805/latest/ which has
> some interesting discussions.
>
> Anyway, I'm running with Michael's definition and I'll have something
> checked in soon.
>
> J'

-- 
Michael Kiefte, Ph.D.
Assistant Professor
School of Human Communication Disorders
Dalhousie University