Business Analytics - Group D

Group D -Ankur Goel

The Choice Between Conjoint Analysis and Perceptual Mapping:

Choosing between conjoint  analysis and perceptual mapping may appear to many to be trivial—a classic no-brainer.  It is like the choice in a world championship between the LA Rams and the New York Mets.  The outcome trivially depends on your choice of turf and rules of play.  Perceptual mapping is played on a turf of image products, such as cigarettes and bourbon, and its rules specify that the competitive structure can be reduced to two dimensional competitive maps, something that is only possible if the perceptions on products on attributes are strongly correlated with one another.  By contrast, conjoint analysis plays on a conceptually different field.  The soft turf of image products is replaced by the hard surface of functional products such as computers or forklift trucks.  Further, the rules of conjoint keep the attribute sharply distinct, so that the impact of a change in any one of them is clearly discernible.  Finally, the outcomes of the two systems are quite different.  Perceptual mapping forms elegant spaces, which locate consumers’ perception of the brand, while leaving obscure the relationship between attribute levels and preferences.  Compare those maps with the partworth functions of conjoint analysis, which move effortlessly from attribute levels to preferences, apparently finessing the issue of perceptions altogether. While perceptual mapping and conjoint analysis techniques have been traditionally quite different, the Adaptive Perceptual Mapping (APM) program of Sawtooth Software makes them much more similar.  What is novel about the APM approach is that it forms maps at the individual level, and then uses these to predict preferences in a choice simulator.  When its individual-level model is compared with the individual level model in conjoint, the differences between the two become much less pronounced. Our plan today is to examine the similarities and differences between an individual-level perceptual map and a conjoint analysis.  We will then describe a study in which both techniques are used to predict straw votes in the current presidential race.  While the winner is the one that predicts the most votes for each individual, the main insights from this study will involve distinguishing when one system will be more appropriate than the other, and why.