For a multi-dimensional analysis, the single nominal scaling level finds one set of quantifications for all dimensions, with different weights or loadings for each dimension. With multiple nominal scaling level no weights/loadings are obtaind, but multiple sets of quantifications, one set for each dimension.
In recent versions of SPSS the terms "single nominal" and "multiple norminal quantification have been replaced with "vector quantification" and "centroid quantification". The multiple nominal or centroid quantification for a category is the centroid of the component scores (in CATPCA) of all subjects that have scored in that category. The vector quantifications are restricted centroid quantifications. In graphical terms: centroid category points for a variable are scattered and vector catgory points for a variable are the centroid points restricted to lie on a line (that goes through the origin, and the loadings of the variable for each dimension define the direction of this line).
Centroid quantification is used when you are interested in relations between the categories of a variable with other (categories of) variables. For example, if you have customer satisfaction data and you also have a variable containing country of origin of the customers, you can use CATPCA with "single" quantifications ((spline) nominal, (spline) ordinal, or numeric) for the satisfaction variables, and include the country variable as multiple nominal. Then you can look at the biplot of loadings and centroids (loading vectors for the satisfaction variables and centroid (or group) points for the countries) to see the relation between aspects of satisfaction and countries.
Hope this helps,
Anita van der Kooij
Data Theory Group
Leiden University
________________________________
From: SPSSX(r) Discussion on behalf of Schwarz,Paul
Sent: Fri 18/05/2007 19:37
To:
[hidden email]
Subject: "single nominal" vs. "multiple nominal"
List readers,
Can someone please explain to me the difference between "single nominal"
and "mulitple nominal" variables? The terms appear in some of the
optimal scaling routines, but SPSS documentation I've seen so far,
hasn't helped me understand the difference.
Thanks for your time and patience.
-Paul
**********************************************************************
This email and any files transmitted with it are confidential and
intended solely for the use of the individual or entity to whom they
are addressed. If you have received this email in error please notify
the system manager.
**********************************************************************
Locked
