Date:         Mon, 26 Jul 1999 09:38:30 -0500
Reply-To:     "Nichols, David" <nichols@SPSS.COM>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:         "Nichols, David" <nichols@SPSS.COM>
Subject:      Re: Data entry for Hierarchical Clustering

That would seem reasonable. Perhaps the problem is that you're reading the
data into the CLUSTER procedure as raw data, and it's transforming your
dissimilarity matrix to another matrix, which will give gibberish. The
ALSCAL procedure for MDS expects data to be in the form of one or more
similarity or dissimilarity matrices, but the CLUSTER expects it to be "raw"
data in a rectangular format. You'd need to use command syntax with the
MATRIX IN subcommand in order to get CLUSTER to realize that this is already
a proximities matrix and not to further transform it.

David Nichols
Principal Support Statistician and
Manager of Statistical Support
SPSS Inc.

> -----Original Message-----
> From: murrays4 [mailto:murrays4@GATEWAY.NET]
> Sent: Monday, July 26, 1999 10:20 AM
> To: SPSSX-L@LISTSERV.UGA.EDU
> Subject: Data entry for Hierarchical Clustering
>
>
> I have done a cardsort with 40 cards and 65 subjects. For analyses I
> want to perform MDS and hierarchical clustering. Currently my data are
> in the form of the lower half of a 40x40 symetrical
> similarities matrix.
> To create a dissimilarities matrix, can I simply subtract the
> number of
> matches for each pair of cards from the total possible (65)? When I do
> this, the MDS solution seems to make sense, but the hierarchical
> clustering doesn't. Can anyone point me in the right direction?
>
> Bill Murray
>