Hi All,
*Second Request*
I created a proximity matrix of dissimilarity distances (measure = SEUCLID; 83 X 83). I performed a hierarchical cluster analysis (method = Ward) on that matrix, which
showed a dendrogram of 1 to 5 clusters. I would like to determine the “jump” or increase in the dissimilarity of joined clusters by evaluating the variance between cases
within clusters. So I will need to calculate the variance for the pairwise distances of cases within clusters (grouping variable). For example, let us say that the initial solution
is 5 clusters. Are the within cluster variances (four clusters now) significantly different at stage one when I combine the cases that makeup cluster 4 and cluster 5? I know
that this is not a clear-cut syntax procedure because I am only interested in calculating the dissimilarity distances between cases within clusters. So variance across the
rows is not useful, but variance for pairwise cells by the group I believe is. I GREATLY APPRECIATE any thoughts and syntax to accomplish this task that has baffled me. I
would be happy to explain more if my explanation here is insufficient. Thank you.
Best,
Frank
=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD
Locked
