Conjoint Analysis – Typical Problems

Typical Problem Studied with Conjoint Analysis: Conjoint analysis is typically used to identify the most desirable combination of features to be offered in a new product or services (e.g. what features should be offered in a new public transportation system?). In such studies, respondents are told about the various combinations of features under consideration and are asked to indicate the combination they most prefer, to indicate the combination that is their third preference, and so on. Conjoint analysis uses such preference data to identify the most desirable combination of features to be included in the new product or service.

What Conjoint Analysis Does: A conjoint analysis applies a complex form of analysis of variance to the preference data obtained from each respondent. This analysis calculates a value (or utility) for each feature. Features with the highest values are judged the most important to respondents.

Types of Variables used in Conjoint Analysis: Conjoint analysis is applied to categorical variables, which reflect different features or characteristics of the product or service under consideration. For example some new product characteristics of interest to researchers could include color (red or blue); size (large. Medium, or small); shape (square or cylindrical); price ($1.00, $1.50, or $2.00); so on. Because it is applied only to categorical variables, conjoint analysis is different from both factor analysis and cluster analysis.

Conjoint Analysis Identifies Interdependencies among variables: Conjoint analysis differs from cross tabulation, regression, LDA, and AID in that it is not concerned primarily with a single dependent variable. Rather, conjoint analysis is like cluster and factor analysis in the sense that these methods try to identify the interdependencies which exist between number of variables. In the example involving a new public transportation system, the variables are the features and characteristics that can be designed into the new system and conjoint analysis tries to measure the relative importance of various combinations of those features and characteristics.

An example of a Conjoint Analysis Application:

A medium sized southern city was planning to improve its public transportation system, and city officials wanted to identify the characteristics of the system that potential users would find attractive. City officials were concerned with the relative desirability of the three main system attributes of fare; frequency of service and comfort (identified as the presence or absence of air conditioning and recorded music). Each of these attributes could be offered in three or four different ways or levels. For example, city officials were considering three fare levels (75¢, $1.00, $1.25); three levels of frequency of service (every 10 minutes, every 15 minutes, every 20 minutes); and four different comfort features both air conditioning and recorded music, air conditioning only, recorded music only, and neither air conditioning nor recorded music.

City researchers selected a representative sample of 500 adults interested in using the improved public transportation if it were available. Each of these adults was shown the 12 combinations that resulted from the three frequency service levels and four comfort features under consideration. They were then asked to identify which combination of service and comfort was their first preference, which combination of service and comfort was their second preference, and so on until they rank ordered all 12 combinations. These adults were asked to repeat this procedure two more times once for all possible combinations of fares and comfort, and once for all possible combinations of fares and frequency of service. In other words, respondents were asked to rank order their preferences for all possible pairs of attributes and for each level or feature being considered for each as attribute. These preference rankings were used in a conjoint analysis to calculate values (called “utilities”) that could help managers decide on the best combination of features to offer.

To obtain a good understanding of how conjoint analysis can be applied to the public transportation system project, readers must have an understanding of (1) respondent preference rankings, (2) how the utilities correspond to preferences, and (3) how the utilities can be used. It should be noted that the following discussions of these three items pertain to a single respondent, and that what applies to a single respondent applies also to every other respondent.–