" Each candy bar's average rating across judges is used as a point in a scatterplot with nine points. The scatterplot looks like the constellation Orion. It also looks like a pattern described by two dimensions, sweet and salty. The stress statistic for two dimensions has an acceptable value, but not as good as when a third dimension is added. The third dimension appears to one of texture: creamy vs. crunchy. As Borgatti (1997) observed, there are two things to look for in interpreting an MDS map: clusters and dimensions. An inspection of the clusters of points reveals that highest average rating seems distant from the other eight and falls in the region describable as somewhat sweet, not very salty, and partly crunchy (Snickers). The lowest rating, which has several near neighbors, falls in the very sweet, very salty, and fairly creamy region (Reese). Based on these results,...

He realizes that a good competitor will have to have a name as appealing as Snickers, so he commissions MRO for another job -- helping him decide on a name for the new candy bar.
References

Borgatti, S. (1997). Multidimensional scaling. Retrieved 4/25/2010 from http://www.analytictech.com/borgatti/mds.htm.

Cooper, L. (1983). A review of multidimensional scaling in marketing research. Applied Psychological measurement, 7, 427-450.

Garson, G.D. (2009). "Multidimensional scaling" from Statnotes: Topics in Multivariate Analysis. Retrieved 4/25/2010 from http://faculty.chass.ncsu.edu/garson/PA765/statnote.

Young, F, (1983). Multidimensional scaling. Retrieved April 24, 2010, from http://forrest.psych.unc.edu/teaching/p208a/mds/mds.html.