200 The Marketing Book
company’s major objective is to maximize
total profits by establishing optimal sales
target volumes and marketing mixes for the
two channels (or customer segments) subject
to the existing limiting constraints.
8 Causal modelconsists of two main analytical
models for testingcausal hypotheses (path
analysis and LISREL). Path analysisis used
on those occasions when some of the
variables are unobservable or have modest
reliabilities. (This tool should not be confused
with the critical path method (CPM), which is
one of the networking programming models
discussed below.) LISREL is of paramount
importance in marketing situations, when we
want to investigate both measurement and
cause, i.e. structural components, of a system
(e.g. in a consumer behaviour study).
9 Hybrid modelsare methods that combine
deterministic and probabilistic (stochastic)
properties (e.g. dynamic programming,
heuristic programming and stock control).
These models are particularly useful in
handling distribution problems, as explained
below.
10 Networking programming modelsare generally
used for planning, scheduling and controlling
complex projects. There are two fundamental
analytical techniques: the critical path method
(CPM), and the performance evaluation and
review technique (PERT). The differences
between the two are, first, that the PERT
acknowledges uncertainty in the times to
complete the activities, while the CPM does
not. Second, the PERT restricts its attention
to the time variable while the CPM includes
time–cost trade-offs. These two together are
also called critical path analysis (CPA)
techniques.
The ten sets of methods above in no way
exhaust the quantitative methods in marketing.
The selection of techniques presented in this
chapter is based either on their particular
current relevance of handling many marketing
problems or/and because of their potential in
marketing research and analysis.
Multivariate methods
The multivariate methods in marketing are
probably the predominant techniques of the
last two decades, not only because of the wide
variety of flexible techniques available in this
category, but mainly because they answer the
most pressing need of marketing research,
which is to obtain the ability to analyse com-
plex, often interrelated and interdependent
data. There are six main multivariate sets of
methods: factor analysis; latent analysis; cluster
analysis; multidimensional scaling; conjoint
analysis; and correspondence analysis.Factor analysis
Factor analysis (FA) is primarily a tool to
reduce a large number of variables to a few
interpretable constructs. The method is used for
exploration and detection of patterns in the
data with the view to obtaining data reduction,
or summarization, which could be more ame-
nable for reaching decisions and taking market-
ing management actions. The software for FA is
readily available and is standard in any SPSS
(Statistical Package for Social Science) package.
The input data are collected from respondents
and the main limitations are how many factors
to extract and the labelling of the emerging
factors. Factor analysis could be used for
analysing consumer behaviour, market seg-
mentation, product/service attributes, com-
pany images, etc.Latent analysis
Latent structure analysis (LA) is a statistical
technique somewhat related to factor analysis,
which can be used as a framework for inves-
tigating causal systems involving both manifest
variables and latent factors having discrete
components. Latent structure analysis shares
the objective of factor analysis, i.e. first, to
extract important factors and express relation-
ships of variables with these factors and,