home *** CD-ROM | disk | FTP | other *** search
- Comments: Gated by NETNEWS@AUVM.AMERICAN.EDU
- Path: sparky!uunet!paladin.american.edu!auvm!HFNET.BT.CO.UK!SMFURNER
- Message-ID: <3288.9212211847@hfs1.hfnet.bt.co.uk>
- Newsgroups: bit.listserv.stat-l
- Date: Mon, 21 Dec 1992 18:47:39 GMT
- Sender: STATISTICAL CONSULTING <STAT-L@MCGILL1.BITNET>
- From: Steve Furner <smfurner@HFNET.BT.CO.UK>
- Subject: SuperANOVA
- Lines: 44
-
- I have a colleague who is experiencing a problem with
- the ABACUS product SuperANOVA. He is using a Mac Quadra 700
- with 8 MBytes of RAM.
-
- The analysis has 5 factors, 3 factors having 3 levels
- each, and one factor with two levels. The 12 subjects are
- included as a factor. Two of the main effects are
- confounded. There are 9 missing cells in the design, and
- there are 648 data points
-
- The manual states that unbalanced and missing cell analyses
- can be carried out, but it does not give much information
- about running them.
-
- With only 12MB of virtual memory SuperANOVA runs out of
- memory when 3rd order cross-terms are included in the model.
- We would like to include at least all the 3rd order terms and
- some of the 4th order terms. However, when we increased the
- virtual memory to 48MB and attempted a complete analysis, the
- machine took all night to finish just the sums of squares
- calculations. It seemed to be paging excessively, which is of
- course not surprising with so much memory!
-
- We were wondering if any anybody has any experience
- of this type of analysis with SuperANOVA. What we
- would like to know is:
-
- a) Is the time and memory being taken reasonable or
- is there a problem with the analysis
-
- b) Is there a way of determining what the machine
- resource requirements are for an analysis
-
- c) Is there a way of determining how long the analysis
- should take
-
- d) Any comments, suggestions or hints on carrying out
- this type of analysis.
-
- Please send replies to:
-
- William Thorpe
-
- thorpe@saltfarm.bt.co.uk
-