[Sigia-l] RE: Cluster analysis is dead - was Jakob wrong? (TryMDS)

[Boniface Lau]

> From: Donna Maurer
>  
> Boniface Lau wrote:
> 
> >To determine whether multidimensional scaling (MDS) or cluster
> >analysis is more appropriate, people should look into the
> >underlying information structure.
> >
> >  
> >
> That's funny - that's why people conduct card sorting exercises in
> the first place ;)

ISTM you are confusing the information structure in the card
sort-produced data with the one recommended by the subsequent
analysis. Don't forget that the actual card sorting itself is only
step 1 of a 2-step process. The second step is to analyse the data
produced in step 1.

After completing step 1 (card sort), you know whether the underlying
structure of the produced data is hierarchical. Thus, you have the
necessary information to decide whether cluster analysis or
multidimensional scaling should be used in step 2 to analyse the data.


Boniface