Dear R People: I have a data set with trt and resp: trt resp A 23 A 28 A 37 A 30 B 37 B 44 B 31 B 35 C 42 C 47 C 52 C 38 And I did read.table, constructed an aov object and ran the summary on the aov object. I also ran the TukeyHSD function. All is well so far. My question is: is there a way to put together a confidence interval from the aov object on a particular mean, please? It seems like there should be. Thanks, Have a great night/day, Sincerely, Erin -- Erin Hodgess Associate Professor Department of Computer and Mathematical Sciences University of Houston - Downtown mailto: erinm.hodgess at gmail.com
Hi Erin,
how about this:
dta<-read.table(textConnection("trt resp
A 23
A 28
A 37
A 30
B 37
B 44
B 31
B 35
C 42
C 47
C 52
C 38"),header=T)
summary(ano<-aov(resp~trt,dta))
ano0<-update(ano,.~.+0)
confint(ano0)
# and the corresponding means are
coef(ano0)
# or use the effects package
library(effects)
summary(allEffects(ano))
hth.
Am 20.01.2013 03:11, schrieb Erin Hodgess:> Dear R People:
> I have a data set with trt and resp:
> trt resp
> A 23
> A 28
> A 37
> A 30
> B 37
> B 44
> B 31
> B 35
> C 42
> C 47
> C 52
> C 38
>
> And I did read.table, constructed an aov object and ran the summary on
> the aov object. I also ran the TukeyHSD function. All is well so
> far.
> My question is: is there a way to put together a confidence interval
> from the aov object on a particular mean, please? It seems like there
> should be.
>
> Thanks,
> Have a great night/day,
> Sincerely,
> Erin
>
>
--
Eik Vettorazzi
Department of Medical Biometry and Epidemiology
University Medical Center Hamburg-Eppendorf
Martinistr. 52
20246 Hamburg
T ++49/40/7410-58243
F ++49/40/7410-57790
--
Pflichtangaben gem?? Gesetz ?ber elektronische Handelsregister und
Genossenschaftsregister sowie das Unternehmensregister (EHUG):
Universit?tsklinikum Hamburg-Eppendorf; K?rperschaft des ?ffentlichen Rechts;
Gerichtsstand: Hamburg
Vorstandsmitglieder: Prof. Dr. Martin Zeitz (Vorsitzender), Dr. Alexander
Kirstein, Joachim Pr?l?, Prof. Dr. Dr. Uwe Koch-Gromus
Hello Erin, You might want to compute *credibility* intervals instead. They have the true meaning of intervals within which the *unknown parameter* (not the sample statistic) has a fixed probability to lie (confidence intervals do not). The AtelieR package, especially in its "Bayesian inference on several means" section, lets you input the data summary (means, standard deviation and counts) and automatically provides these intervals on the unknown means. The approach implemented is the one described in: Neath, A. & Cavanauigh, J. (2006). A bayesian approach to the multiple comparisons problem. Journal of Data Science, 4, 131-146. Besides, it will give you the best constrained model (in terms of equal means) for your data, and the probability that this model is the true one. It is an elegant solution to the multiple comparisons problem. HTH, Yvonnick Noel University of Brittany, Rennes France