The Marketing Book 5th Edition

Results of cross-tabulation used as raw data

Row and column scores plotted

on a correspondence map

Cumulative proportion explained by

the number of dimensions calculated

Contribution of row points to the

inertia of each dimension calculated

Contribution of column points to the

inertia of each dimension calculated

Row scores calculated Column scores calculated

204 The Marketing Book

3 In the case of a new product that differs
substantially from its principal competitors, the
same elements cannot be used for aggregating
utilities.
4 The effects of promotion and distribution
effort on competitive reaction are not
considered.
5 Perceptions from a concept statement and
those from the actual product may differ.
6 New products may take several years to reach
the market, during which time customer
preferences and competitive products may
have undergone substantial changes. Conjoint
analysis has been applied widely on consumer
research (Vriens, 1994), in advertising
evaluation (Stanton and Reese, 1983) and
other commercial uses (Cattin and Wittink,
1982).

Correspondence analysis

Correspondence analysis (CA) is a visual or
graphical technique for representing multi-

dimensional tables. It can often be impossible to

identify any relationships in a table and very

difficult to account for what is happening.

Correspondence analysis unravels the table and

presents data in an easy-to-understand chart.

One approach for generating maps uses cross-

classification data (e.g. brands rated as having

or not having a set of attributes) as a basis

(Hoffman and Franke, 1986). In this approach

both brands and attributes are simultaneously

portrayed in a single space. This technique is

particularly useful to identify market segments,

track brand image, position a product against

its competition, and determine who non-res-

pondents in a survey most closely resemble.

Correspondence analysis shows the relation-

ships between rows and columns of a corre-

spondence or a cross-tabulation table. This

method can be used for analysing binary,

discrete or/and continuous data. CA belongs to

the family of multidimensional scaling tech-

niques and could be employed to scale a matrix

of non-negative data to represent points

Figure 9.3 Procedural steps for correspondence analysis