Skip to content
Prev 2236 / 7420 Next

R-sig-ecology Digest, Vol 39, Issue 16

hi,

sorry - i'm afraid this was only due to a trivial syntax error:  
"...,dist = "euclidean",..." should say "...,method = "euclidean",..."  
with the wrong argument adonis defaults to "bray" - and with  
bray-curtis dissim. the difference between 1 and 2 is really not the  
same as the one between 2 and 3 or 4 and 5, isn't it?

so the message to you, regarding your suspicious p-values, might be to  
check for the dissimilarity-index you use, what its characteristics  
are and if it's appropriate for the questions you ask..

yours,
kay

following up my example:

# distance matrix with bray-curtis dissimilarity:
# dissimilarities *are the same*

dm <- as.matrix(vegdist(spdf, method = "bray"))

# assign names
dimnames(dm) <- list(paste(treat, imp), paste(treat, imp))

# see average off-diagonal dissimilarities (within treatments, between  
impact levels)
mean(dm[11:20,1:10])
mean(dm[31:40,21:30])
mean(dm[51:60,41:50])

# distance matrix with euclidean dissimilarity:
dm <- as.matrix(vegdist(spdf, method = "euclidean"))

# assign names
dimnames(dm) <- list(paste(treat, imp), paste(treat, imp))

# see average off-diagonal dissimilarities (within treatments, between  
impact levels)
# dissimilarities *are not the same*
mean(dm[11:20,1:10])
mean(dm[31:40,21:30])
mean(dm[51:60,41:50])




Zitat von Kari Lintulaakso <kari.lintulaakso at gmail.com>: