In most markets, the choice that a person will make is contingent on the situation they are in. For example:
- Wine choice made in a restaurant may differ to that in a bottle shop or when ordering online.
- Breakfast cereal choice may differ when buying for one’s self than for a child.
- The price we will be prepared to pay for a hot dog at a sports stadium may be a lot higher than in a park.
One approach to addressing this is to use volumetric and constant-sum questions but, as discussed below in Types of Conjoint Questions, it is extremely difficult to validly analyze such data (difficult to the point where it has perhaps never been done rigorously).
A generally better approach is to explicitly cue the situation, writing questions that fully describe the relevant situation by mentioning the situation's:
The goal is to put the user in a very specific state of mind, where this state of mind is analogous to the state of mind they are in when purchasing. For example, a good question is “Thinking about the last time you purchased chocolate for your husband at a supermarket, if only the alternatives below were available, which would you have purchased?”. A poor question would be “Which of these would you buy?”, as the situation is so vague that the person answering the question will not easily access any of the relevant constraints and considerations required to make a realistic choice.
In addition to ensuring that the questions are appropriately cued, there is a need to ensure that the cueing is representative. For example, if 70% of purchases in a market are made in supermarkets, then ideally you will structure the questionnaire so that 70% of the choice questions are also about supermarkets.
Analysis is greatly simplified by asking a person to provide all their data about a single situation as if you cue for multiple situations for each respondent it is necessary to model interactions between situations and utilities.
Please sign in to leave a comment.