Exam 21: Multidimensional Scaling and Conjoint Analysis

In MDS, objects cannot be projected onto two, three, four or even higher dimensions.

Respondents who are overwhelmed by a ranking task will tend to ignore variations in the less important attributes.

If all of the possible levels of an attribute have the same utility, the attribute is not important in influencing overall attitude.

In theory, before respondents can knowledgeably make trade-offs, they must compare different attributes and evaluate the desirability of the various levels of each attribute.

When there is a need to reduce the number of attributes in a study, factor analysis has an advantage over discriminant analysis in that a significance test can be done of the output of factor analysis.

Factor analysis groups attributes that are similar

The greater the difference between the highest and lowest valued levels of an attribute, the less important the attribute.

Consider the following MDS solution, in which A, B, C and D represent brands of a product class. The solution implies that B is sweeter than D but less creamy than A.

Before introducing a new product, Innovators Incorporated conducted a preference study.The data were analyzed by using an ideal-point MDS solution.The company then designed and introduced a product which was extremely "close" to the target market's ideal point. However, since a new product has been introduced into the space, another study should be done since the ideal point may have been relocated.

The pair-wise trade-off approach produces concepts that are more realistic than those of the full profile approach.

Number of dimensions in an MDS cannot be decided using

Multidimensional scaling addresses the general problem of positioning objects in a perceptual space.