Conjoint Analysis

analysis of features considered jointly

Forms of conjoint analysis

• Choice-based
• Adaptive: Each customer is asked a different set of questions which are decided dynamically based on their responses
• Full-profile: a ****full suite of options is presented to the consumer
• Menu-based: consumer is shown a list of attributes with associated prices and they choose their ideal product

Process

• Define products as a collection of attributes
• Consumers react to a number of alternatives
• Infer attributes’
  • importance
  • most desired level of each consumer

Optimization Method (Linear Programming)

<aside> 💡 used when consumer choice data is pairwise data and attribute values are continuous

</aside>

• Set of options on which the preference judgement is made: $j = \{1,2,..,n\}$
• The $n$ options are described in terms of $t$ dimensions: $P = \{1,2,..,t\}$
• The pre-specified location of the $j^{th}$ option in the $t$-dimensional space is denoted by $Y_j$ i.e. $Y_j = \{Y_{j,p}\} p \in P$
• The ideal point of the subject is $X = \{x_p\}p \in P$ i.e. the product location most preferred by the individual.
• Unweighted distance $d_j^u = [\sum_{p \in P} (y_{j,p} - x_p)^2]^{1/2}, \forall j \in J$
• Weighted distance $d_j^w = [\sum_{p \in P} w_p(y_{j,p} - x_p)^2]^{1/2}, \forall j \in J$