The authors show how one can perform a 'meta-analysis' of pictures in a conjoint based task. The authors used data from the world-wide coffee study, comprising 38 pictures. They identified the components of the pictures, and used the presence/absence of those components as independent variables. They were then able to relate these components to the utilities of the pictures and to the locations of the pictures on a set of semantic scales. They did this analysis by country and by concept response segment.

Click here for complete paper