One assumption inherent in adaptive and tradition conjoint
analysis is pair-wise independence of attributes — that the importance of a product feature
is not dependent on any other feature. If true, then the part-worth of the attributes is additive.
While the pair-wise independence assumption is valid most of the time, it is unlikely to
hold true for brand and price. Some brands are more sensitive to changes in price than others. The importance of
price, therefore, varies with brand.
When two attributes are pairwise dependent, we say that there is statistical
interaction between the attributes. This is handled by adding the interaction term,
price × brand, to the utility model. Pricing models that account for this interaction
yield better estimates of consumers’ sensitivity to price.
