I started learning cluster analysis (using SPSS) and I need some help in a practical problem.

Given the following variables:
The respondents were asked to indicate the importance of the following factors when buying products and services using a 5-point scale (1=not at all important, 5=very important)

Saving time (x1)
Getting bargains (x2)
Getting products that aren’t on the high street (x3)
Trying new things (x4)
Being aware of what companies have to offer (x5)
Gender (0=Female; 1=Male)


1) Given the levels of measurement, which clustering method would you prefer? Carry out a cluster analysis using this procedure.

As I understand there are both categorical and continuous (income) variables - so the right method is two-step clustering. And we cannot use hierarchical method, correct?

  • $\begingroup$ Likert scales can be considered as scale (interval/ratio) or as ordinal. To treat them as ordinal is more contrite but is less flexible. There is very few options in cluster analysis realm designed specifically for ordinal variables. You might use Gower similarity and then hierarchical clustering. Or think of them as continuous and use Euclidean distance or whatever you want. The idea to do cluster analysis on a mixture of variables of different type is not particularly a good idea. $\endgroup$
    – ttnphns
    May 5, 2015 at 15:41
  • $\begingroup$ Is this a question from a course or textbook? If so, please add the [self-study] tag & read its wiki. $\endgroup$ May 5, 2015 at 15:47
  • $\begingroup$ @ttnphns Thanks, Can I use Two-Step clustering method in this case? $\endgroup$
    – user25963
    May 5, 2015 at 17:03


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy