repository.upenn.edu/cgi/viewcontent.cgi?article=1022&context...
by P Choi - 2005 - Related articles
Apr 1, 2005 - We have discussed the basic concepts and mathematics of conjoint analysis. In this section, we focus on the implementation of conjoint ...Market simulators
1. Conjoint Model After obtaining partworth estimates for conjoint approach, researchers incorporate some kind of simulator that could take each individual’s idiosyncratic partworths and compute its implied choices in various ways to develop forecasts of how the market might respond to managerial changes in product design. The values (partworths) of the attributes could be varied in ways so as to forecast changes in market shares and returns to the sponsor firm and to competitors in the same marketplace. Other aspects, such as market segments, could also be isolated for specific marketing strategy decisions.
Suppose the primary data input to the model consists of a matrix of K individuals’ partworths. Assuming there is no interactions involved. Let denote the utility on the pth attribute for the jth level for individual k, and be the intercept term for individual k. k pjy ka The utility of a profile to individual k is the sum of the individual’s intercept term and the partworths corresponding to the attribute levels that compose the profile. () ∑ = =+ P p kk pjPk ayjjjjU p 1 231 ,,..., The market share estimates for the profile for individual k can then be calculated using the following function: Market Share k = ( ) () Pk Pk jjjjU jjjjU e e ,,..., ,,..., 231 231 1+
We can then aggregate the shares for all individuals (say, a total of K individuals) to obtain ∑
=
=
K
k kaggregate 1 k Market Share w Market Share where wk is the weight of individual k. The weight could be set as 1/K, but it could also be set to reflect the volume of the individual’s purchases in the product category.
No comments:
Post a Comment