1000 participants have viewed 1 of 15 adverts (a categorical, numerical, variable) and rated their responses ( 5 different continuous variables). I want to see if the 15 adverts can be clustered together in smaller groups based on the responses. Hope you can help, my statistics experience is limited to SPSS. Thanks
Yes you can do this but the best methods IMHO are probably not available in SPSS. The best book (and best methods) around is probably Kaufman and Rousseeuw, Finding Groups in Data, Wiley Interscience. Since you probably don't want to read an entire book, I recommend looking at Venables and Ripley, Modern Applied Statistics with S 4th ed. Springer. Sect.11.2. pp315-321. Here you get a brief overview of the methods I'm suggesting and coded examples in R. The methods themselves are available in R package cluster. R is a free download from the CRAN website. If you decide to go this way, and I really would based on my own experience, a great (and low cost) introduction to R is Lander, R for Everyone, Addison Wesley. This book is available on Amazon for almost nothing. It takes you through modern data analysis and Chapt.22 is all about package cluster. It contains code and worked examples. If I can help, please contact me. I really like this stuff. Best wishes.
What are your clustering variables? Do you want to cluster 1000 respondents or cluster the advertisements? The advertisements are already clustered into 15 groups. Is your research problem to cluster the advertisements? If you clarify the above questions , I may be able to assist you more. Best wishes.
Thanks for your help. My clustering variable is the adverts. I want to be able to cluster them into smaller groups than the original 15, based on the responses by the participants.
- Badges
- Science topic
- Similar topics
- Psychology
- Developmental Psychology