1

I have 2 groups, each containing 20 individuals. For each individual I have a dissimilarity matrix of a number of judged items. I'd like to evaluate whether there is an overall difference between the groups in the semantic networks underlying their judgements. As a first step, I computed 2D MDS spaces for each group separately using the smacofIndDiff() function from the smacof package in R, which I understand permits one to characterize the overall space for a group of individuals who may vary slightly from one to another. This yields two 2D spaces (really, two 2xK matricies, where K is the number of items evaluated for dissimilarity), but I'm not sure how to proceed with regards to evaluating the degree to which the resulting spaces differ and whether this difference might reasonably be expected by chance.

Any suggestions? Note that I'm not wed to the 2D MDS idea, so feel free to suggest alternative ways to represent the judgement data as a semantic space.

Mike Lawrence
  • 13,793
  • You seem do have done 2 independent simple MDSs. Instead, consider one weighted MDS (INDSCAL model) with two matrices (groups). Even better you could do INDSCAL with all 40 matrices and analyse there appear systematic group differences afterwards. – ttnphns Oct 08 '12 at 16:43

0 Answers0