Skip to content

Use nparcomp function from nparcomp library to run post hoc

6 messages · Dennis Murphy, Jun Shen, David Winsemius

Dennis Murphy
# 
Hi:

Is this the function nparcomp() in the nparcomp package or the one
from the mutoss package? When using functions from packages, it is
useful to indicate the package name. I'm assuming you're using the
nparcomp package, because your code worked for me when that package
was loaded:
Loading required package: multcomp
Loading required package: mvtnorm
Loading required package: survival
Loading required package: splines
Nonparametric Multiple Comparison Procedure based on relative
contrast effects , Type of Contrast : Dunnett
 NOTE:
 *-------------------Weight Matrix------------------*
 - Weight matrix for choosen contrast based on all-pairs comparisons

 *-----------Analysis of relative effects-----------*
 - Simultaneous Confidence Intervals for relative effects p(i,j)
      with confidence level 0.95
 - Method = Multivariate Delta-Method (Logit)
 - p-Values for  H_0: p(i,j)=1/2

 *----------------Interpretation--------------------*
 p(a,b) > 1/2 : b tends to be larger than a
 *--------------Mult.Distribution-------------------*
 - Equicoordinate Quantile
 - Global p-Value
 *--------------------------------------------------*
$weight.matrix

    < snipped for brevity - all zeros >

$Data.Info
       Sample Size
1     Duoderm   24
2     Fibrase   24
3 Kollagenase   24
4 Non-treated   24
5    Stimulen   24
6     Vehicle   24

$Analysis.of.relative.effects
                  Comparison rel.effect confidence.interval t.value
1     p(Non-treated,Duoderm)        0.5   [ 0.499 ; 0.501 ]       0
2     p(Non-treated,Fibrase)        0.5   [ 0.499 ; 0.501 ]       0
3 p(Non-treated,Kollagenase)        0.5   [ 0.499 ; 0.501 ]       0
4    p(Non-treated,Stimulen)        0.5   [ 0.499 ; 0.501 ]       0
5     p(Non-treated,Vehicle)        0.5   [ 0.499 ; 0.501 ]       0
  p.value.adjusted p.value.unadjusted
1                1                  1
2                1                  1
3                1                  1
4                1                  1
5                1                  1

$Mult.Distribution
  Quantile p.Value.global
1 2.568766              1

$Correlation
[1] NA

A graphic also appears indicating zero effect, which is what one would
expect since Ulceration = 5 for every observation in the data frame.
R version 2.13.0 (2011-04-13)
Platform: x86_64-pc-linux-gnu (64-bit)

locale:
 [1] LC_CTYPE=en_US.UTF-8       LC_NUMERIC=C
 [3] LC_TIME=en_US.UTF-8        LC_COLLATE=en_US.UTF-8
 [5] LC_MONETARY=C              LC_MESSAGES=en_US.UTF-8
 [7] LC_PAPER=en_US.UTF-8       LC_NAME=C
 [9] LC_ADDRESS=C               LC_TELEPHONE=C
[11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C

attached base packages:
[1] splines   stats     graphics  grDevices utils     datasets  methods
[8] base

other attached packages:
[1] nparcomp_1.0-1  multcomp_1.2-5  survival_2.36-9 mvtnorm_0.9-999
[5] sos_1.3-0       brew_1.0-6      plyr_1.5.2

loaded via a namespace (and not attached):
[1] tcltk_2.13.0 tools_2.13.0

Check your version of R and the nparcomp package against this. If you
have an older version of R or nparcomp, perhaps an upgrade is
sufficient to fix the problem.

HTH,
Dennis
On Fri, Apr 29, 2011 at 2:49 PM, Jun Shen <jun.shen.ut at gmail.com> wrote:
David Winsemius
#
On Apr 29, 2011, at 7:17 PM, Jun Shen wrote:

            

      
      
      
    

        
You can see from Dennis' console output that 4 other packages are  
loaded when nparcomp is loaded. You have apparently not installed all  
of the needed packages for the new version of R. There is an argument  
to the install.packages function that will check for dependencies. Or  
perhaps you need to set that option in a GUI interface.