home *** CD-ROM | disk | FTP | other *** search
- Comments: Gated by NETNEWS@AUVM.AMERICAN.EDU
- Path: sparky!uunet!uvaarpa!darwin.sura.net!paladin.american.edu!auvm!NIOSHE2.EM.CDC.GOV!DKW4
- Return-Receipt-To: <dkw4@NIOSHE2.EM.CDC.GOV>
- Encoding: 27 TEXT
- X-Mailer: Microsoft Mail V3.0 (beta-2)
- Message-ID: <2A939339@router.em.cdc.gov>
- Newsgroups: bit.listserv.stat-l
- Date: Thu, 20 Aug 1992 16:40:00 EST
- Sender: "STATISTICAL CONSULTING" <STAT-L@MCGILL1.BITNET>
- From: dkw4@NIOSHE2.EM.CDC.GOV
- Subject: Correction for Shapiro-Wilk test
- X-To: STAT-L%MCGILL1.BITNET@CUNYVM.CUNY.EDU
- Lines: 24
-
- Question? Has anyone heard of a correction factor for the Shapiro-Wilk test
- (for normality) produced by SAS's PROC UNIVARIATE? If you have heard of
- such a correction factor, do you think it really helps?
-
- That is, I seem to find the S-W test to lack power for small sample sizes and
- to almost always reject the null hypothesis (of normality) for large sample
- sizes. Translation: I find it to be a worthless test most of the time.
-
- However, a colleague (who was quoting me....gulp) was challenged by another
- person who said Shapiro-Wilk worked just fine with the correction factor.
-
- So, does anyone know anything about this? Thanks much for any help.
-
- Deanna
-
- *-----------------------------------------------------------------*
- Deanna Wild
- Statistician
- National Institute for Occupational Safety and Health
- 4676 Columbia Parkway MS: R-4
- Cincinnati, Ohio 45226
-
- INTERNET: DKW4@NIOSHE2.EM.CDC.GOV
- *-----------------------------------------------------------------*
-