msb@hal.UUCP (Monty S. Baveja) (04/18/89)
Thanks to all the input I have received concerning uni/multi variate analysis
of repeated measures. I have decided against using SPSS/PC, which is what I
normally use. Instead I am going to use BMDP/PC initially and if the data needs
more massaging I am going to use the GLM procedure in SAS. For all those
interested here are some of the responses and evaluations I received.
Thanks again,
Monty Baveja (msb@hal.cwru.edu)
____________________________________________________________________
SAS does virtually any kind of design you want. You need to know what
you are doing however with repeated measures (specify SUBJECTS nested
within the between factors in the Model statement).
SPSS can do repeated measures but has more limitations.
BMDPC has some routines that do repeated measures.
PowerStat does both balanced and unbalanced repeated measures, but
univariate only.
There are various other packages that purport to do repeated measures
but are often quite limited.
There are also packages like GLIM that require more expert knowledge to
use, but are very powerful.
Avoid SPSS which does not handle repeated measures well. SAS, Systat
or PowerStat are the way to go. SAS is best but most expensive, Systat
has an awful interface and is overpriced. PowerStat is limited (does
not does MANOVA) but does SAS-type repeated measures and sells for
less than $300 US.
BMDP PC is the premier package for repeated measures manovas, they
have several types. They also have a good monograph on repeated measures
statistics. Don't have their address handy, but see back of PC Mag
[Dfor phone, address. I have not used directly but have seen their
main-frame packages used by psychophysiologists for years.
mvolo@ecsvax.UUCP
There is a Stats package from called FIRST
Serious Statistcal Software
Lynwood
Bentyheath Lane
Willaston
South Wirral
L64 1SD
United Kingdom Tel 051-327 4268
I have seen the Acorn Archimedes version but there is a PC version. The
Author will supply a demo version. I do not use stats myself so I cannot
comment on the package other than to say that it seems to work and
is very complete and easy to use.
From jch@osupyr.mast.ohio-state.edu Wed Apr 12 13:14:48 1989
PROC GLM in SAS does both univariate and multivariate repeated measures.
was just reading the mail and noticed your query about repeated measures.
SAS does repeated measures analysis very well and provides a number of
options with it as well. I personally use SYSTAT and it also does
repeated measures analysis but with somewhat fewer options. I find this
method of analysis becoming quite important in my field (soil and plant
nutrition).
There is a BMDP PC package available. It is expensive, however, and I believe
that it has all the features of the mainframe versions --- namely, very
good ANOVA/MANOVA programs. SAS/PC can also handle these things --- PROC
GLM (I think).
Stanley Wasserman
stanwass@uxh.cso.uiuc.edu
years. I think its hard to beat SAS - it's a power-statistician's package par
excellance. It does have a stiff learning curve, but there are a number of
good shortcuts (e.g. good books, training seminars, etc). I don't think PC
Magazine did SAS justice - but it's not surprising. It is an imposing package
to the novice. The returns on investment in SAS are very high, however.
_______________________________________________________________________________
Jim Goes | InterNet GOES@OREGON.UOREGON.EDU
Graduate School of Management | UUCP: {fav backbone}!uoregon!oregon!goes
University of Oregon | MaBell: (503) 686-3309
________________________________________________________________________________
Grad School. Its not just a job, it's an indenture...
SAS certainly works fine for this. You need to have EMS if your model is very
complex (lots of degress of freedom) but it works well on PC.
-phil
> -*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
> J. Philip Miller - Div of Biostat - Washington Univ Medical School
> phil@wubios.WUstl.edu - Internet phil@wubios.wustl - bitnet
> (314) 362-3617 c90562jm@wuvmd - alternate bitnet
>