Conjoint Surveys to Infer Preferences

In order to better understand economic preferences and econometrically estimate utility functions, it's often useful to create some artificial choice sets that allow revealed preference among choice objects that have characteristic combinations not commonly observed or when choice sets and choices are not easy to observe.

For instance, its really difficult to figure out teacher preferences over school characteristics including the composition and level of pay, student characteristics, and school policy. Not only is it uncommon to observe teachers making choice sets and decisions but its unclear what characteristics teachers are aware of. And there's not a lot of variation in, for instance, pay composition.

Asking teachers to make decisions among hypothetical schools with specified characteristics is often the best economists can do.

What you want for this is a conjoint survey.

The only problem is that you can't specify every element of the state space. The question is what is the subject imagining in the background. For instance, do they associate some unspecified characteristic with one of your specified characteristics? If so, you're getting bias.

Keywords: Andrew Johnston, Andrew, Johnston, Education, Wharton, vita, curriculum vitae, cv, economics, applied economics, economist, microeconomics, empirics, empirical economics, Andrew, Andy, Andrew Johnston, Andy Johnston, Andrew Cowley Johnston, Andy Cowley Johnston, Johnston, Education, Wharton, vita, curriculum vitae, cv, economics, applied economics, economist, microeconomics, empirics, empirical economics