|
Dear Listers
I would be much grateful if you could give me a clear answer as I might use twostep cluster analysis for rating scale variables, inasmush items(variables)are ordinal(ordered)so as not to treat them as continuous nor as categorical(unordered)ones. This is a common asked question often raised in many statistical queries but unfortunately without any definite answer. Ultimately, what about CATPCA as a means of transforming (quantifying)ordinal measures before approaching Twostep or K-means for clustering? Many thanks G. Siardos ===================== 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 |
|
>
> This is a common asked question often raised in many statistical > queries but unfortunately without any definite answer. That is because there is no simple definitive answer. Seldom would one want to rely on single method of clustering. Any clearer answer depends a lot on what data you have, what you are using the data to find out, etc. How were the data obtained? When you say "rating scale variables" what response scale are you using? Where did the questions used to create the variables come from? Are the items from pre-existing scales of values, attitudes, etc.? Is this an exploratory study or are you testing specific hypotheses? How many variables are there? Were subsets of the variables chosen to represent particular constructs (ideas)? Once you have a new nominal level variable from the clustering what do you intend to do with that variable? Art Kendall Social Research Consultants On 2/13/2010 9:07 AM, George Siardos wrote: > Dear Listers > > I would be much grateful if you could give me a clear answer as I > might use twostep cluster analysis for rating scale variables, inasmush > items(variables)are ordinal(ordered)so as not to treat them as continuous > nor as categorical(unordered)ones. > This is a common asked question often raised in many statistical > queries but unfortunately without any definite answer. Ultimately, what > about CATPCA as a means of transforming (quantifying)ordinal measures > before approaching Twostep or K-means for clustering? > > Many thanks > G. Siardos > > ===================== > 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 > > ===================== 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
Art Kendall
Social Research Consultants |
| Free forum by Nabble | Edit this page |
