I have a data set of about 200 individuals who have been rated on 7 behavioral
performance dimensions by subject matter experts. I would like to identify
clusters of individuals that share similar performance profiles across
dimensions.
My initial intention was to approach this via cluster analysis, possibly with
the TwoStep cluster analysis technique in SPSS. However, I recently read an
article (from 1997), that approached a similar research question using inverse
principal components analysis (aka inverse factor analysis, or Q-factor
analysis). They stated that cluster analysis was not as effective as inverse
principal components analysis for grouping profiles when there was measurement
error and/or overlap between groups.
Does anyone have any insights on the pros/cons & comparative appropriateness
of these two techniques? Have recent improvements in clustering techniques
(such as the TwoStep technique in SPSS or fuzzy clustering techniques) made
these techniques more effective than inverse principal components techniques?
Regards,
Taylor
=====================
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
