Hi:
Try this:
Trat <- c(2:30) # number of treatments
gl <- c(2:30, 40, 60, 120)
# Write a one-line 2D function to get the Tukey distribution quantile:
f <- function(x,y) qtukey(0.95, x, y)
outer(Trat, gl, f)
It's slow (takes a few seconds) but it seems to work.
HTH,
Dennis
On Wed, Nov 3, 2010 at 3:52 AM, Silvano <silvano@uel.br> wrote:
> Hi,
>
> I'm building Tukey's table using qtukey function.
>
> It happens that I can't get the values of Tukey's one degree of
freedom and
> also wanted to eliminate the first column.
>
Firstly, one needs at least two treatments to find a studentized range
(which is why you get NaNs across the first row when you set Trat = 1).
Secondly, if you have at least two groups, you need at least two
observations per group to get a variance estimate, which means that the
variance estimate of the difference needs to have at least 2 df. If one
group has only observation in it, the variance of the difference is the
variance of the group with >= 2 observations, which doesn't make
intuitive
sense. This is why you get NaNs along the first column.
HTH,
Dennis
>
> The program is:
>
> Trat <- c(1:30) # number of treatments
> gl <- c(1:30, 40, 60, 120) # degree freedom
>
> tukval <- matrix(0, nr=length(gl), nc=length(Trat))
>
> for(i in 1:length(gl))
> for(j in 1:length(Trat))
> tukval[i,j] <- qtukey(.95, Trat[j], gl[i])
>
> rownames(tukval) <- gl
> colnames(tukval) <- paste(Trat, "", sep="")
> tukval
>
> require(xtable)
> xtable(tukval)
>
>
> Some suggest?
>
> --------------------------------------
> Silvano Cesar da Costa
> Departamento de EstatÃstica
> Universidade Estadual de Londrina
> Fone: 3371-4346
>
> ______________________________________________
> R-help@r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
> http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>
[[alternative HTML version deleted]]