Latent Class/ Cluster analysis with uncompleted ranking data ...

#1
Dear all,

For my thesis I am looking for more ideas to do the best with uncompeted data sets and using the best software solution for a market segmentation problem.

I have ca. 20 variables ranked in an acending order of importance.
ca. 60 respondents gave this stimuli and some of them gave me not a full ranking - real world case with 5 ranks, 11 ranks, 20 ranks etc. of the 20 variables.
Can I analyze this with SPSS??? does anyone has experience for that?

I also would like to find a theoretical discussion for clusteranalysis results with ranking, rating data. I have found this discussion regarding benefit segmentation several times but I think they discuss about evaluating the conjont profiles/ choice sets with rating, ranking.

Does anyone knows Latent Class modells? www.statisticalinnovation.com provides this software for choice data. I would like to find someone with experience in a similar case like mine for uncompleted ranking data and cluster problems.

thank you very much for help,
Matt:)
 
#3
percentile ranking

hey guyz.. hey matt..
just want to post this question, i think it's pretty much easy but i really am still confused.. "what does percentile rank mean in terms of passing and/or failing?" (e.g. 80th percentile of those who failed)
help..
thanks :confused:
 

JohnM

TS Contributor
#4
maddie,

Percentile rank has nothing to do with passing and/or failing - it just indicates what percentage of the scores are lower.

So, a statement such as "80th percentile of those who failed" means that they've broken the scores into two major groups, pass and fail, arbitrarily, and if you're in the "fail" group at the 80th percentile, it means that 80% of the scores that failed are lower.

John