Date:         Wed, 24 Nov 1999 00:48:47 -0600
Reply-To:     "Nichols, David" <nichols@SPSS.COM>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:         "Nichols, David" <nichols@SPSS.COM>
Subject:      Re: Cluster Analysis

The only way that the order of cases can affect the results from CLUSTER is
when you have tied distances. When the smallest distances between existing
clusters (which are just cases at the beginning) are equal for two or more
pairs, then the order can affect the results in that an arbitrary decision
is made in terms of which pair to join. If I remember correctly, the search
and replace algorithm that goes through the matrix replaces the marked
minimum only if the new one is smaller, so in the case of ties, the first
pair will be the one that gets joined.

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

> -----Original Message-----
> From: Felix Brosius [SMTP:FelixBrosius@CSI.COM]
> Sent: Sunday, November 23, 1997 3:48 AM
> To:   SPSSX-L@LISTSERV.UGA.EDU
> Subject:      [SPSSX-L] Cluster Analysis
>
> Dear Listmembers,
>
> under which circumstances/options does the order of the cases in the data
> file have an influence on the results of an hierarchical cluster analysis,
> especially in the context of the Ward-Method?
>
> Many thanks in advance
> felix brosius