Hi folks
Can anyone provide me with some information on the following:
In Conjoint analysis for the aggregate model, papers often refer to
carrying out a cluster analysis on the part worth functions for each
individual. Say this resulted in 3 clusters then an aggregate analysis
would be performed on each of the 3 groups.
Am I interpreting this correctly....say if we have n individuals and
there are 4 attributes each with 2 levels, we may end up with the
following values for the part worths (hypothetical)
ATTRIBUTES
Person A B C D
level1 level2 level1 level2 level1 level2 level1 level2
1 0.756 -0.756 0.64 -0.64 0.987 -0.987 0.86 -0.86
2 0.726 -0.726 0.69 -0.69 0.889 -0.889 0.82 -0.82
3 0.736 -0.736 0.74 -0.74 0.798 -0.798 0.94 -0.94
4 0.432 -0.432 0.31 -0.31 0.437 -0.437 0.2 -0.2
5 0.354 -0.354 0.40 -0.40 0.678 -0.678 0.41 -0.41
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
n . . . . . . . .
If we decided to group these individuals as discussed above by cluster
analysis before generating aggregate models, would we enter all of the
data above...i.e. for each individual, the values of the part worth for
*both* levels of each attribute?
If anyone knows of an appropriate text which gives a worked example
then it would be much appreciated too. The papers and books I have read
merely say that "cluster analysis was performed using individual part
worth functions" and do not give much detail. Some papers also refer to
the standardising of each individual's part worths prior to conducting
cluster analysis. Has anyone any views on this?
Many thanks for your help on this.
All the Best,
Kim.
|