Statistical Gastronomy

In recent years, molecular gastronomy has gone from being an obscure style of cooking practiced by a few, zealous chefs to an accepted artform embraced by the greater foodie community. The proponents of this craft seek to infuse chemistry into cooking, often times resulting in the deconstruction of classic dishes. For instance, consider this deconstructed glass of white wine created by Chef Jose Andres. By tasting each of the individual components that are present in the wine, diners come to appreciate the complexity of the product and respect the notion that the whole is in fact greater than the sum of its parts.

Believe it or not, there is a statistical corrolary to molecular gastronomy - conjoint analysis. Conjoint analysis seeks to understand the relative importance of certain attributes in a product by asking respondents to rate the utility of the product as a whole. Using regression analyses, researchers can then determine what truly matters to consumers, and which levels of particular attributes are ideal. Like Chef Andres deconstructed glass of white wine, conjoint analysis demonstrates to us that consumers may not be aware of which components are truly driving their perceptions of a product.

How might conjoint analysis be used in the restaurant industry? Let's say that one is designing a new restaurant and would like to understand the importance of a few key variables: theme (French v. Italian), average check ($20, $30, $40), and presence of celebrity chef (Yes/No). Individuals can then be asked to rate the utility of combinations of the above attributes and those ratings could be analyzed using regressions. After a little bit of statistical gastronomy, the restaurateur will not only know which of the above attributes is most important, but also which level of each attribute is preferred.

Conjoint analysis has been used quite a bit in the hospitality industry; in fact, Courtyard by Marriott was developed through the process. Nevertheless, conjoint has yet to make its way to the restaurant space, perhaps because it is difficult to determine all of the attributes that impact a diner's experience. Going forward, I expect to see statistical analyses such as conjoint becomming more prevalent in foodservice, particularly for major brands. Just as molecular gastronomy has raised haute cuisine to a whole new level, statistical gastronomy can do the same for consumer insights.