Dear R users,
Firstly, I would like to congratulate the R core team in bringing out R
1.0.0 and all who have helped in developing it.
I have been having problems with using se.contrasts and would be pleased
if someone help.
I have been doing a repeated measures ANOVA using aov using a split plot
design for a single variable, color. The aov results were as follows:
> summary(aov(CD2~cont + Error(cont:rpl)+times+cont:times, color))
Error: cont:rpl
Df Sum Sq Mean Sq F
value Pr(>F)
cont 6 8.6214 1.4369
6.8651 0.2841
Residuals 1 0.2093 0.2093
Error: Within
Df Sum Sq Mean Sq F
value Pr(>F)
times 7 53.133 7.590
1053.32 4.886e-10 ***
cont:times 42 86.030 2.048
284.24 2.149e-08 ***
Residuals 7 0.050 0.007
---
Signif. codes: 0 `***' 0.001 `**' 0.01 `*' 0.05 `.' 0.1
` ' 1
I then used model.tables to get the effects and standard errors , which
gave me this:
> model.tables(cd.aov2, "effects", se=T)
Standard error information not returned as design is unbalanced.
Standard errors can be obtained through se.contrast.
........... followed by the table of effects
However, whatever I do I cannot calculate se.contrast.
In the above test the factor cont has 7 levels (1,2,3,4,5,6,7,7) , rpl
is the replication, and times has 8 levels.
I want to test if there is a significant difference between the
following:
(1) if there is a difference between cont (1,2,3) between times (6,7,8)
(2) if there is a difference between cont (4,5,6) between times (6,7,8)
essentially these are the main contrast I am interested in, as a
treatment is applied between times 6 and 7 at 3 levels within
cont(1,2,3) and cont(4,5,6).
Interaction effects between cont with times can be seem by plotting
effects, but I suppose this isn't enough to show significant
differences, isn't it?
I would be grateful if someone could suggests how I might calculate the
contrasts using se.contrasts.
As there are also different contrasts.ie. helmert,treatments, etc. This
confuses me even more.
Also I mentioned in the beginning of the text that I did the ANOVA for a
single variable. I infact measured 13 variables and wanted do
originally try a repeated measures MANOVA, but did not know how to do
this. Any suggestions here would also be grateful. Cons and Pros of
univariate and multivariate approach.
I have looked at the growth and rmutil, which the author J Lindsey
suggested could be used, but as I had no experience using them, again I
did not know what to do. Has anyone tried this approach?
Thanks
Peter
-----------------------------
Peter Ho
Escola Superior de Biotecnologia
Universidade Cat?lica Portuguesa
Rua Dr. Ant?nio Bernardino de Almeida
Porto 4200-072
Portugal
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !) To: r-help-request at
stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
