home *** CD-ROM | disk | FTP | other *** search
- Comments: Gated by NETNEWS@AUVM.AMERICAN.EDU
- Path: sparky!uunet!paladin.american.edu!auvm!!SERVICES"
- X-Organization: Concordia University, Montreal, Quebec
- X-Envelope-to: stat-l@MCGILL1.BITNET
- X-VMS-To: IN%"stat-l@mcgill1"
- X-VMS-Cc: ILFC594
- Message-ID: <2F37649540010E42@Vax2.Concordia.CA>
- Newsgroups: bit.listserv.stat-l
- Date: Wed, 2 Sep 1992 10:02:00 EDT
- Sender: "STATISTICAL CONSULTING" <STAT-L@MCGILL1.BITNET>
- From: "GEOFF SELIG - CONCORDIA U. COMPUTING SERVICES"
- <ILFC594@VAX2.CONCORDIA.CA>
- Subject: Re: Equal sample sizes - unequal variances?
- Lines: 26
-
- Deanna Wild writes:
-
- > Why does having equal sample sizes mitigate the requirement of equal
- > variances in an ANOVA/t-test?
-
- The SSE term in ANOVA estimates the cell variances. It is generated, for
- all intents and purposes, by _weighting_ the observed cell variability by
- the cell size. With equal cell size, cells with overly large or small
- variances contribute an equal but "opposite" share of variability to the
- overall SSE value. However, with unequal cell sizes, the cell with the
- largest _n_ will contribute more than its share to SSE while the cell with
- the smallest _n_ will contribute less. Should the largest _n_ fall in a
- cell with an overly large or overly small variability, the SSE value will
- over- or under-estimate the _true_ value for the (overall) cell
- variability. That is to say, SSE will be overly large if the largest _n_
- falls in the cell with the largest variability or SSE will be
- overly small in the largest _n_ falls in the cell with the smallest _n_.
-
- I hope this helps.
-
- Geoff Selig Phone (514)-848-7666 | _, /| The Cat's Meow
- Academic Services FAX (514)-848-7622 | \ o.O'
- Department of Computing Services | =(___)=
- Concordia University (H928) | U
- Montreal, Quebec H3G 1M8 | ilfc594@vax2.concordia.ca
- | ilfc594@conu2.BITNET
-