home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: sci.math.stat
- Path: sparky!uunet!zaphod.mps.ohio-state.edu!news.acns.nwu.edu!nucsrl!ddsw1!pfleury
- From: pfleury@ddsw1.mcs.com (Patrick J. Fleury)
- Subject: Re: Quartiles in S-plus
- Message-ID: <Bxy40J.HI6@ddsw1.mcs.com>
- Sender: Patrick Fleury pfleury@ddsw1.mcs.com
- Date: Thu, 19 Nov 1992 04:19:29 GMT
- References: <1992Nov17.170005@masg1.epfl.ch>
- Organization: ddsw1.MCS.COM Contributor, Chicago, IL
- Lines: 20
-
- In article <1992Nov17.170005@masg1.epfl.ch> zoubeidi@masg1.epfl.ch (Toufik Zoubeidi) writes:
- >Hello,
- >
- >In preparing a class example we found out that in computing quartiles
- >S-plus uses different methods under different commands. The method
- >used in the commands "summary" or "quantile" is different from the
- >one used to compute quartiles for a boxplot in the command "boxplot".
- >We know which method is used in each case. Does anyone know the reason
- >for this difference.
- >
- There was an informative article a couple of years ago in _The Americna
- _Statistician_ which might shed some light on this. I think it was called
- _Some implementations of the boxplot_ and two of its authors were Iglewicz
- and Hoaglin. I think the year was 1989 but I am not absolutely sure.
-
- As I remember it, the authors point out that there are eight different
- definitions of the sample quartile which are used in different circumstances.
-
- It is worth looking at.
-
-