[R] pvclust: highly varying results for the exactly same analysis, unknown warning message

[paladini at trustindata.de]

Hello,
I have got problems using pvclust() and it would be really nice if  
somebody could help me.

I have got a dataframe called "owner2006" looking like this:


                       company  year share0 share1 share2  share5  
share6  share7 share11 share12
1                   aareal bank 2006  57.800   0.00 42.200   0.000   
0.000   0.000       0       0
3                        adidas 2006  94.730   0.00  5.270   0.000   
0.000   0.000       0       0
5           airbus group (eads) 2006  33.470   0.00  5.020  22.490  
35.530   0.000       0       0
.
.
.
.
(The data frame dimension is 124 rows, 10 colums.)


I use the following code


ownerscale2006=owner2006
ownerscale2006[,c(3:10)]=scale( owner2006[,c(3:10)])

fit=pvclust(t(ownerscale2006[,c(3:10)]), method.hclust="ward",  
method.dist="euclidean")
plot(fit, main="owner,2006,scaled")
pvrect(fit, alpha=.95)


What concerns me is that the outcome varies so much. One time only two  
clusters are proposed next time there are six clusters highlighted.
Shouldn`t the outcome be always nearly the same? Is there something  
wrong with my analysis or can a certain data-structure cause such  
highly varing outcomes?


Moreover I get always  warning messages :
1: In a$p[] <- c(0, bp[r == 1]) :
   number of items to replace is not a multiple of replacement length
2: In a$p[] <- c(0, bp[r == 1]) :
   number of items to replace is not a multiple of replacement length
3: In a$p[] <- c(0, bp[r == 1]) :
   number of items to replace is not a multiple of replacement length
4: In a$p[] <- c(0, bp[r == 1]) :
   number of items to replace is not a multiple of replacement length
5: In a$p[] <- c(1, bp[r == 1]) :
   number of items to replace is not a multiple of replacement length


And I haven't the faintest idea what they mean in this context.


I would be very grateful if somebody could help me.

Best regards

Claudia Paladini