202 The Marketing Book
using, to a larger extent, marketing tools and
techniques, while cluster 1 strategy is more
‘passive’. Low coefficients of hierarchy (or low
dissimilarity) indicate high relationships or
high similarity. For example, variables 1 and 2
are highly correlated, that is, they aggregate
early, at a coefficient of 0.27. In contrast,
variable 12 (use of advertising) is highly dis-
similar to other variables and associates with
the remaining coefficients in cluster 2 only at
the 3.5 level. A possible explanation of this
could be that hotels adopting the marketing
strategy indicated by cluster 2 use an alter-
native communication mix and/or other mar-
keting tools (e.g. variables 13, 14, 15, 4, 5, 10,
etc.).
Multidimensional scaling
Multidimensional scaling (MS) is a measure-
ment technique concerned mainly with the
representation of relationship, differences, dis-
similarities (or similarities), substitutability,
interaction, etc. among behavioural data such as
perceptions, preferences and attitudes. The
input data on various objects (variables) which
are to be analysed are collected from the subjects
(respondents) by a number of direct or indirect
questions. The questions can be either of Likert
type (i.e. a five-point scale questionnaire indicat-
ing the level of agreement or disagreement to
statements) or, alternatively, asking each of the
respondents to rank the variables to be investi-Figure 9.2 Hierarchical clustering of variables associated with a marketing strategy for hotels
Source: Meidan (1983).