Contextual Inference in Markets: On the Informational Content of Product Lines

Numerous studies demonstrate that seemingly irrelevant factors influence people’s decisions. Perhaps the best known examples of such influence are context effects. A consumer exhibits a context effect if her choice between two alternatives systematically depends on the presence of other options. An extensive literature demonstrates context effects in laboratory settings. One of the most widely studied context effects is the compromise effect (Itamar Simonson 1989), which refers to the finding that people tend to choose the middle option. More precisely, when three alternatives are available, the middle alternative is chosen more often than when it is paired with only one other option. Figure 1 shows the compromise effect obtained by Simonson (1989). This tendency to avoid extreme options has been credited with affecting decisions ranging from the demand for wine (Daniel L. McFadden 1999) to voting (Kaisa Herne 1997) and investing (Shlomo Benartzi and Richard H. Thaler 2002). Even more telling of the importance bestowed on the compromise effect is its didactic use in books such as 101 Ways to Increase Sales (Dirk van Eetvelde, Carl Geens, and Peer Harrington 2002) and Information Rules (Carl Shapiro and Hal R. Varian 1999). Another behavior that is seemingly inconsistent with a well-defined preference ordering is choice overload. In choice overload experiments (e.g., Sheena S. Iyengar and Mark Lepper 2000), customers are less likely to make a purchase if more products are added to the choice set. More broadly, marketing research reveals that minor changes in the decision-making environment can have a substantial impact on what people buy. This malleability of behavior is usually