+ All Categories
Home > Documents > Why does Segmentation Matter ... DATASET · JULY 2014

Why does Segmentation Matter ... DATASET · JULY 2014

Date post: 21-Nov-2023
Category:
Upload: lisboa
View: 0 times
Download: 0 times
Share this document with a friend
27
See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/264043821 Why does Segmentation Matter ... DATASET · JULY 2014 READS 104 2 AUTHORS, INCLUDING: Jaime R. S. Fonseca University of Lisbon 44 PUBLICATIONS 158 CITATIONS SEE PROFILE All in-text references underlined in blue are linked to publications on ResearchGate, letting you access and read them immediately. Available from: Jaime R. S. Fonseca Retrieved on: 05 February 2016
Transcript

Seediscussions,stats,andauthorprofilesforthispublicationat:https://www.researchgate.net/publication/264043821

WhydoesSegmentationMatter...

DATASET·JULY2014

READS

104

2AUTHORS,INCLUDING:

JaimeR.S.Fonseca

UniversityofLisbon

44PUBLICATIONS158CITATIONS

SEEPROFILE

Allin-textreferencesunderlinedinbluearelinkedtopublicationsonResearchGate,

lettingyouaccessandreadthemimmediately.

Availablefrom:JaimeR.S.Fonseca

Retrievedon:05February2016

1

Why does Segmentation Matter? Identifying market segments through

a mixed methodology

Jaime R.S. Fonseca

Abstract

The purpose of this chapter has been to develop an overall framework that describes

how market can be segmented, that is the objective of this study is the way of grouping

customers together for the most effective targeting, by means of a new conceptual

model which intends to combine the use of Latent Segment Models with a mixed

research scheme (by merging qualitative and quantitative researching methodologies). A

particular retail market segmentation solution is a function of both the market

segmentation base variables and a specific segmentation procedure, providing a better

understanding of market. The knowledge of segment structure is of high importance for

marketing because of its managerial utility, particularly in what concerns targeting and

positioning. Companies that identify underserved segments can then outperform the

competition by developing uniquely appealing products and services. This research

began with an overview on segmentation aspects and aims, and presents, under mixed

research scheme, an application with Latent Segment Model (LSM) procedure for retail

market segmentation, and information criteria AIC3 and AICu for model selection, in

order to uncover the segment structure underlying to a dataset from a retail chain

customers.

Keywords

Market segmentation; Segmentation base variables; Segmentation methods; latent

segment models; mixed research

Jaime R.S. Fonseca (corresponding author)

Chair for Data Analysis, Higher Institute of Social and Political Sciences (ISCSP),

Centre for Public Administration and Policies (CAPP), Technical University of Lisbon,

Portugal; E-mail: [email protected]

2

1. Introduction and Objectives

Market segmentation is a theoretical marketing concept involving artificial groupings of

consumers constructed to help managers design and target their strategies (Wedel &

Kamakura, 1998). Today, companies recognize that they cannot appeal to all customers

in the market or at least not to all customers in the same way, because each of the

customers is unique, and they come from different backgrounds, live in different areas

and have different interests and goals. As a result, they are too varied in their needs and

buying practices; what’s more, the companies themselves vary widely in their abilities

to serve different segments of the market, and rather than trying to compete in an entire

market, each company must identify the parts of the market that it can serve best and

most profitably, (Sun, 2009). Companies that identify efficiently segments can then

outperform the competition by developing uniquely appealing products and services.

By dividing the market into relatively homogenous subgroups or target markets, both

strategy design and tactical decision making can be more effective and robust for

successfully bridging the gap between segmentation principles and successful

application, which continues to be a major challenge for the marketing community.

Segmentation technique – identifying homogenous sub-populations within larger

heterogeneous populations – has emerged as an important marketing tool over the past

half-century, as a response to the need to effectively communicate with, and motivate to

action, an increasingly diverse population of individuals, families and businesses, who

rely on a rapidly multiplying set of communication channels (Heuvel & Devasagayam,

2004). It is well known that customer segmentation is most effective when a company

tailors offerings to segments that are the most profitable and serves them with distinct

competitive advantages; this prioritization can help companies develop marketing

campaigns and pricing strategies to extract maximum value from both low and high

profit customers. By tailoring the offering product to different groups, companies are

able to more precisely meet the needs of more customers and consequently to gain a

higher overall level of share or profit from a market.

The purpose of this article is to develop an overall framework scheme that describes

how market can be segmented, that is the objective of this research study is the way of

grouping customers together for the most effective targeting, by means of a new

conceptual scheme which intends to combine the use of Latent Segment Models on a

mixed research (by merging qualitative and quantitative researching methodologies),

3

expecting to result in market segments that satisfy the homogeneity within and

heterogeneity across segments; regardless of the employed tool to segment a population,

each segment must contain elements which are homogeneous. The bases of these

similarities should be easily interpretable and should provide useful guidelines for the

promotion of products or services specific to each segment.

From the purpose’s paper we intended to deal more deeply with the third part of market

segmentation scheme (Table 3), that is, best conceptual scheme for effective market

segmentation, and is organized as follows: in section 2, we overview concerning the

subject; in section 3, we present the proposed model for market segmentation, and

corresponding information criteria; in section 4 we report on results from a retailing

data set and finally, in section 5 we present some concluding remarks.

2. Why Segmenting?

Consumer diversity is increasing rapidly and companies have long sought to

differentiate their products relative to competitors, and this is where market

segmentation comes in. Why segmenting? Because of identifying segments where

competitors see an undifferentiated mass market creates several opportunities for new

marketing strategies based on a better knowledge of specific customers’ needs and

preferences. It is consensual that the foundation of strategic marketing is formed by

market segmentation, target marketing and product positioning.

Nowadays, segmentation is a crucial marketing strategy, helping marketers for

identifying consumer needs, preferences and find new marketing opportunities, and

enables marketers to regulate marketing mixes to meet the needs of particular segments.

Several marketing researchers have responded to the needs of management by

conducting market segmentation studies, for instance, (Assael & Roscoe, 1976),

(Calantone & Sawyer, 1978), (Punj & Stewart, 1983), (Beane & Ennis, 1987),

(Kamakura, Kim, & Lee, 1996), (Lockshin, Spawton, & Macintosh, 1997), (S. H.

Cohen & Ramaswamy, 1998), (Dibb, 1999), (Kim, Srinivasan, & Wilcox, 1999), (Bock

& Uncles, 2002), (Palmer & Miller, 2004), (Sun, 2009). Marketing planning process

flows from the selection of target markets to the formulation of specific marketing mix

and positioning, objective for each product of retail chain. Segmentation theory suggests

that groups of customers with similar needs and purchasing behaviors are likely to

demonstrate a more homogeneous response to marketing programs, the constitution of

segments being essential to target marketing (Fonseca & Cardoso, 2007b).

4

Segments are derived from the heterogeneity of customer wants (Smith, 1956), who

defines market segmentation as a process that involves viewing a heterogeneous market

as a number of smaller homogeneous markets, in response to differing preferences,

attributable to the desires of consumers for more precise satisfaction of their varying

wants. The definition of (Kotler, 1972), was conceptually consistent with Smith’s one,

and he defines it as the subdivision of a market into homogeneous subsets of customers,

where any subset may conceivably be selected as a market target to be reached with a

distinct marketing mix. For (Dolnicar, 2008), market segmentation is the strategic tool

to account for heterogeneity among individuals by grouping them into market segments

which include members similar to each other and dissimilar to members of other

segments, and following (Sun, 2009), market segmentation is to divide the whole

market into meaningful, relatively small and identifiable market segments, which are

groups of individuals or organizations with similar product needs.

In other words, market segmentation is the science of dividing an overall market into

segments, whose members share similar characteristics and needs (members’

homogeneity).

A market segmentation solution is a function of the market segmentation base variables

and of a specific segmentation (clustering) procedure, and it provides a better market

understanding and, consequently, means to develop more successful business strategies

(Fonseca & Cardoso, 2005), by addressing the specific needs of the selected segments.

Because an organization adopts either a mass-market or a market segmentation

strategies, when a decision is made for market segmentation, two essential questions

must be addressed: (1) which method is to be used to segment the market, and (2) which

segmentation base variables to use.

Concerning methods, since the appearance of Smith’s now-classic article (1956), market

segmentation has become an important tool both academic research and applied

marketing, (Punj & Stewart, 1983), and the primary use of Cluster Analysis in

marketing has been for market segmentation; Cluster Analysis is a very weak analytical

segmentation technique, but it is perhaps the most used technique for segmentation,

traditionally; thus we select several applications of this tool in marketing (Table 1),

from 1967 to 2007.

5

Table 1 Cluster Analysis Applications

Authors Goal

(Green, Frank, & Robinson, 1967) That intended to identify matched cities for test

marketing

(Green & Carmone, 1968) Aiming to identify similar computers in the computer

market

(Bass, Pessemier, & Tigert, 1969) Trying to identify market segments with respect to

media exposure

(Montgomery & Silk, 1971) To identify opinion leadership and consumer interest

segments

(Morrison & Sherman, 1972) To determine how various individuals interpret sex

appeal in advertising

(Greeno, Sommers, & Kernan, 1973) In order to identify market segments with respect of

personality variables and implicit behavior patterns

(Sexton, 1974) To identify homogeneous groups of families using

product and brand usage data

(W. T. Anderson, Jr , E. P Cox, & Fulcher,

1976)

On the identification of the determinant attributes in

bank selection decisions and use them for segmenting

commercial bank customers

(Calantone & Sawyer, 1978) To study the stability of market segments in the retail

banking market

(Schaninger, Lessig, & Panton, 1980) To identify segments of consumers on the base of

product usage attributes

(Kiel & Layton, 1981)) For the development of consumer taxonomies of search

behavior by Australian new car buyers

(Becker, Brewer, Dickerson, & Magee, 1985) Tried to divide consumer markets by looking at a

consumer’s personality

(Jain, 1993)

Analyzed markets through social, economic, and special

segmentation variables such as brand loyalty and

consumer attitude

(Segal & Giacobbe, 1994) The cluster analysis uncovered four basic “natural”

demographic segments

(Wayne S. DeSarbo et al., 1995)

K-means Cluster Analysis conducted for a major

packaged goods a major packaged

goodsffjfjkkkkk11111111111mmmmmmm\\\\\\\

(Kotler, 1997)

Has proposed that consumer markets should be divided

according to geographic, demographic, psychographic

(lifestyle and personality), and behavioral variables

(Dibb, 1998) Cluster analysis to identify segments in 270

pregnant

women, by using demographic and satisfaction variables (Hruschka & Natter, 1999a)

K-means using demographic and attitude

Variables

(Hofstede & Steenkamp, 1999)

Developed an integrated methodology based on

consumer means-end chains to identify segments in

international markets

(Baker & Burnham, 2001)

Market segments are identified based on a cluster

analysis of respondents' preferences for brand,

price

(Lin, 2002) Considering demographic and psychographic variables

(Dibb, Stern, & Wensley, 2002) Cluster Analysis for measuring the impact on

organizational performance

(kau, Tang, & Ghose, 2003) Cluster Analysis for seeking patterns as well as

their

motivations and concerns for online shopping (Lee et al., 2004)

To segment festival market based on

motivation factors

(Jayawardhena, Wright, & Dennis, 2007) Cluster Analysis and K-means for stability

6

Hierarchical Cluster algorithms are among the most commonly used clustering analysis

in marketing research; however, users of these approaches, tend to discard much of the

detail found in the dendogram (Arabie, Carroll, DeSarbo, & Wind, 1981). Moreover, as

it is well known, the dendogram did not constitute a unique solution, which is a

disadvantage of Hierarchical Cluster Analysis.

Quantitative segmentation tools can range from simple categorization analysis, such as

CART and CHAID/Regression Trees analyses (McCarty & Hastak, 2007), (Thomas &

Sullivan, 2005), (Chen, 2003), (Levin & Zahavi, 2001), to more sophisticated clustering

techniques, such as Hierarchical Cluster Analysis, TwoStep Cluster Analysis, K-means,

(Lee, Lee, & Wicks, 2004), (Hruschka & Natter, 1999), (Jedidi, Jagpal, & DeSarbo,

1997), Conjoint Analysis (Wayne S. DeSarbo, Ramaswamy, & Cohen, 1995), (Green &

Krieger, 1991), (Green & Srinivasan, 1990), Multidimensional Scaling (Carroll &

Green, 1997), (Biggadike, 1981), (Wind, Douglas, & Perlmutter, 1973), Discriminant

Analysis (Tsai & Chiu, 2004), (Harvey, 1990), (Moore, 1980), or Latent Segment

Models (Fonseca, 2010), (Steven H. Cohen & Ramaswamy, 1994).

In this issue of analyzing data, concerning the purposed Latent Segment Models

approach to clustering, part of our conceptual scheme, it offers some advantages when

compared with other most traditional techniques, such as: (1) it identifies market

segments (Dillon & Kumar, 1994); (2) it provides means to select the number of

segments (McLachlan & Peel, 2000); (3) it is able to deal with diverse types of data

(different measurement levels) (Vermunt & Magidson, 2002); (4) it outperforms more

traditional approaches (Vriens, 2001), and (5) it is appropriate to deal with covariates, in

order to a better understanding of customers (Fonseca & Cardoso, 2007a).

Basically they enable us to simultaneously optimize a research function (LSM and

information criteria) and efficiently find segments of cases within that framework thus

being useful for better understanding market structures.

In order to be valuable to marketers, the plan of market segmentation needs to be able to

identify different segments of customers having uniform and stable responses to a

particular set of marketing variables, the segmentation base variables (See Table 2).

This is the second question we have to address, and several authors conducted

researches about it, for instance, (Sharma & Lambert, 1994), (Wedel & Kamakura,

1998), (Kim et al., 1999), (González-Benito, Greatorex, & Muñoz-Gallego, 2000), (W.

S. DeSarbo, Degeratu, Wedel, & Saxton, 2001), (Vriens, 2001), (Heilman & Bowman,

2002), (Fennell, Allenby, Yang, & Edwards, 2003).

7

The greatest opportunity for creating competitive advantage often comes from new

ways of segmenting, because a firm can meet buyer needs better than competitors or

improve its relative cost position (Porter, 1985), thus the identification of segmentation

variables is among the most creative parts of the segmentation process.

Table 2 Some segmentation base variables

Segmentation bases Brief description

Demographics

Consumers can be grouped on the basis of

characteristics such as age or household

composition

Socioeconomic

Consumers can be grouped on the basis of

characteristics such as income, occupation and

education

Product Usage

Potential to use the firm’s product is a

behaviourally based segmentation basis, with

attributes such as awareness, used in the past,

would consider using

Psychographics Consumers can be grouped on the basis of

personality, attitudes, opinions, and life styles

Generation

Generation, or cohort, refers to people born in the

same period of time: similar age, similar

economic, cultural, and political influences in

formative years

Generally, a combination of psychographics (in order to understanding) and

demographics (for targeting) will give good results. For instance, concerning

demographic variables, (Sharp, Romaniuk, & Cierpicki, 1998), and (Lin, 2002) have

suggested that they are useful in segmenting markets, the majority of evidence does not

support this assertion (Fennell et al., 2003) and (Uncles & Lee, 2006). Some studies

showed insignificant or no effect of demographics on consumer price responsiveness,

such as (Kim et al., 1999) and (Scriven & Ehrenberg, 2004). (Granzin, 1981) suggests a

simple solution to the problem that links in with (Simcock, Sudbury, & Wright, 2006)

calling for more sophisticated segmentation: choosing other variables to work alongside

demographics. We argue that demographic variables are very important, in order to a

better understanding of segments, and can be used as covariates when estimating latent

segment models, not as being part of segmentation base variables.

8

3. Methodology

The process of identifying segments necessitates a thorough analysis of the entire

market, not only focusing on the customer’s needs and shopping habits but also

providing knowledge of changing market conditions and competitive actions (Segal &

Giacobbe, 1994). From the traditional market segmentation studies, including now the

mixed research methods, we can enumerate six steps in the market segmentation

process, and we summarize them into Table 3.

Table 3 Segmentation Steps

Segmentation Steps Description

Step 1

Determine the

boundaries of the market

Selecting a market or product category for study

Step 2

Segmentation base variables

Marketers must use their knowledge of the market to select a few

relevant variables in advance.

Step 3

Selecting market research tools

(Mixed research process)

Select tools for collecting and analyzing data. From the stages of the

social research, we notice that qualitative and quantitative can coexist in

each researching process: (1) in a first phase we have the research

preparation, with the establishment of the subject of study (specification

of the problem, papers overview, research theory) and the research

structure (structure of the test, measurement, sampling, ethics); (2) it is

followed by research formation (direct observation, indirect interview,

life history, discussion group, content analysis, survey, secondary data,

simulation); (3) finally, information analysis (data treatment, data

analysis). It would be very difficult to exclude one of the two

methodologies in each one of these three phases, but social scientists

frequently do not manage the available information in its statistical

results, thus losing chances to present statistics that could result in a

bigger clarification of its research questions (King, Tomz, & Wittenberg,

2000).

Step 4

Profiling of each market segment

Involves selecting those variables that are most closely related to

consumers' actual buying behavior

Step 5

Segment targeting

A marketer should look for opportunities that provide a good In step 3,

selecting tools for collecting and analyzing data, we introduce a mixed

methodology, in order to test the solution, by using all the information

obtained from the qualitative data collection tools, such as interviews,

focus groups, participant observation, for exploring new topics, assisting

theory building, providing context for quantitative data, and to help on

explain /clarify quantitative findings (segments). We think that we are

becoming with a better knowledge about the needs and preferences of

customers, by merging knowledge, by using qualitative (quantitative)

conclusions to update quantitative (qualitative) conclusions.

Concerning step 2, one of the most important step in segmentation

schemes, there is a large array of possible segmentation bases - set of

variables or attributes used to assign potential customers to homogeneous

segments - and for a review we can see (Wilkie, 1990) and (Wedel &

Kamakura, 1998), for instance. Following the last one, “The

identification of market segments is highly dependent on the variables

and methods used to define them.” This sentence stresses the great

importance of segmentation base variables, and methods for analyzing

data. Table 2 summarizes some examples of possible segmentation bases.

match for the organization and its resources

Step 6

Product positioning

Involves developing a product and marketing plan that will appeal to the

selected market segment

9

As for step 3, selecting market research tools, we can use data collecting tools - varying

from qualitative to quantitative tools. Market research design concept, staged design,

can be sustained by a mixed methodology or pragmatism methodology which can be

defined as research using both qualitative and quantitative methods and mixing the two

methods when beneficial (Onwuegbuzie & Leech, 2005) and (Leech, Dellinger,

Brannagan, & Tanaka, 2010); in this methodology both quantitative and qualitative

approaches are about taking observations of the world (data) and presenting them within

a framework (a model) (White, 2002).

In order to reach a market research questionnaire design, often we start the step 3

process with qualitative research to define ways customers view the product category or

service, and the differences in those views; we conduct preliminary focus groups or

other qualitative methods - such as depth interviews - in order to uncover insight as to

how several consumers and business audiences see and feel about the product category

and competitive brands, for instance, uncovering and refining our learning about

customers, thus obtaining a fuller picture and deeper understanding of the segments.

Owing to its use of situation and context appropriate designs and methods, mixed

methods research seems particularly suited for action research (Vitale, Armenakis, &

Feild, 2008). In questionnaire design, we can use for instance market segmentation

dimensions such as behavioural, attitudinal, or a combination of these which may form

psychographic segments, and another dimension, demographic for instance, as

covariates, in order to a better characterization of the segments. By a developmental

survey, we also use a questionnaire for collecting data, and then a quantitative method

for analyzing the obtained dataset.

In this study we are more focused on steps 3 and 4, namely on tools for analyzing data

and profiling reached segments. But the market segments identified should mostly

satisfy three criteria which we display in Table 4.

These criteria are all reached by using Latent Segment Models (LSM), by means of the

aforementioned advantages. It is a probabilistic/statistic clustering approach which

assumes that the variables’ observations in a sample arise from different segments of

unknown proportions. They are very good models for modelling of complex

phenomena, then synthesizing and extracting knowledge. The proposed segmentation

conceptual scheme (1) provides internal homogeneity/external heterogeneity, (2) it

enables marketers to reach segments separately using observable characteristics of the

segments, and (3) because of the using theoretical information criteria for model

10

selection balances parsimony (fitting a model with a large number of components

requires the estimation of a very large number of parameters and a potential loss of

precision in these estimates (Leroux & Puterman, 1992), and model complexity (which

tends to improve the model fit to the data), the selected latent class model does evidence

a good trade-off between good description of the data and the model number of

parameters.

The segmentation process is used to distinguish between customers and non-customers,

where "customers" here are extended to include buyers, payers, loyal customers, etc.,

and to understand their composition and characteristics - Who they are? What do they

look like? What are their attributes? Where do they reside? This analysis supports a

whole array of decisions, ranging from targeting decisions to determining efficient and

cost effective marketing strategies, even evaluating market competition, (Levin &

Zahavi, 2001). The three most relevant criteria for segments (Table 4) are always

reached by this conceptual scheme, when segments’ structure really exists, which

doesn’t happen with other tools, such as Cluster Analysis models.

Table 4 Segments’ criteria

Criteria Meaning

Internal Homogeneity/ External

Heterogeneity

Customers within a segment should have similar

responses to the marketing mix variable of interest

but a different response to members of other

segments

Parsimony

The degree to which the segmentation makes every

customer a unique target. That is, the

segmentation should identify a small set of

groupings of substantial size

Accessibility

The degree to which marketers can reach segments

separately using observable characteristics of the

segments

Let )( ipyi

y denote the vector representing the scores of the ith case for the pth

segmentation base variable (i = 1,…,n ; p = 1,…,P).

We consider that the cases on which the attributes are measured arise from a population

which we assume to be a mixture of S segments, in proportions s (mixing proportions

or relative segment sizes), s = 1,…,S. The statistical probability density function of the

vectori

y , given that i

y comes from segment s, is represented by )|( siysf , with s

representing the vector of unknown parameters associated with the specific chosen

11

probability density function. Then the population density can be represented as a finite

mixture of the densities )|( siysf of S distinct segments, i.e.

)|(

11

)|( siy

P

psf

S

ssi

yf

(1)

where i = 1,…,n, },,1{ , }1,,1{ with }, ,{ ,1

1

,0 ss

S

sss

.

is the vector of all unknown parameters.

The LSM estimation problem, simultaneously addresses the estimation of distributional

parameters and classification of cases into segments, yielding mixing probabilities. The

estimation process is typically directed to maximum likelihood using the Expectation-

maximization (EM) algorithm (Dempster, Laird, & Rubin, 1977), (McLachlan & Peel,

2000).

LSM naturally provides means for constituting a partition by means of assigning each

case to the segment with the highest segment-membership probability, that is with

isSs

Max ̂,...,1

, where

S

j if

sf

k

j

k

j

k

si

k

sis

j1

)()(

)()(

ˆ

)ˆ|y(ˆ

)ˆ|y(ˆ

(2)

In order to derive meaningful results from clustering, the mixture model must be

identifiable, that is, a unique maximum likelihood solution should exist (Bozdogan,

1994). A goal of traditional LSM estimation is to determine the smallest number of

latent segments S that is sufficient to explain the relationships observed among the

variables of segmentation base variables. If the baseline model (S = 1) provides a good

fit to the data, no LSM is needed since there is no relationship among the variables to be

explained; otherwise, a model with S = 2 segments is then fitted to the data. This

process continues by fitting successive LSM to the data, each time adding another

dimension by incrementing the number of segments by 1, until a parsimonious model is

found that provides an adequate fit. They are very good models for modelling of

complex phenomena, then synthesizing and extracting knowledge.

Concerning methodologies for the selection of the appropriate latent class model, we

propose the use of the traditional information criteria. Especially, because all the

12

observed variables have similar measure, all of them categorical, we will use the

information criterion AIC3, the more advised for this situation (Fonseca, 2010).

Table 5 Variables and covariates of the used dataset

Variables used for a retail chain customers’ segmentation Segmentation base

USAGE FREQUENCY

Psychographic

VISIT PATTERN

ERN

COMING FROM

TRAVEL TIME

WHY SHOPPING

MONTHLY SPENDING ON PURCHASES FOR THE HOME

MONTHLY EXPENSES IN STORE

QUALITY OF FRESH PRODUCE

STORE TREATMENT

EFFICIENCY OF STORE’S STAFF

STORE VARIETY OF PRODUCTS

STORE PRODUCT PRESENTATION

STORE ENVIRONMENT

CLEANING SHOP

PRICES

PRODUCT QUALITY PRIVATE LABEL

INTERNET USAGE

Covariates

SEX

Demographic AGE

FAMILY SIZE

LIVE CYCLE

INCOME

Socioeconomic EDUCATION

OCCUPATION

CLASS

We can now answer the questions on page 5, concerning segmentation tool and

segmentation base variables. Thus, considering the best segmentation tool we purpose

latent segment models, and as for segmentation base variables we consider, with

marketers, some attributes of stores and some attributes of customers which interact

between them.

These variables are considered as manifest variables or indicators (USAGE FREQUENCY, …,

INTERNET USAGE), from which model parameters are estimated, and used some covariates

(SEX, ... , CLASS), which only are used for a better understanding of segments and their

members (Table 5). Results from these models estimation are valid for all cases,

products, branches, countries, services, and for all kind of variables (categorical,

continuous, or mixed), because of they are probabilistic/statistic models.

4. Results from a retailing data set and discussion

13

As we already assume, we use two types of data collecting in the research: qualitative

and quantitative. Here, we only report quantitative data analysis, based on a dataset

obtained from a questionnaire administered to customers of a retail chain.

Table 6 AIC3 for model selection

Model LL AIC3

1-Cluster -45613,3 91484,666 2-Cluster -44022,5 88681,023

3-Cluster -43082,7 87179,382

After eliminating the questionnaires with several non responses, we retain a dataset with

1449 customers characterized by the segmentation base variables presented in table 5.

Table 7 Parameters’s estimates of two-class latent model

Cluster Size Cluster1(61 percent) Cluster 2 (39 percent)

Indicators USAGE FREQUENCY Everiday 0,2280 0,4265

Two or three times a week 0,3754 0,2670 Once a week 0,1842 0,1523 Two times a month 0,0410 0,0278 Once a month 0,0700 0,0586 Occasionaly 0,1013 0,0679 VISIT PATTERN

During a week 0,3511 0,1898 During the weekend 0,1838 0,1246 Both situations 0,4652 0,6857 COMING FROM

From home 0,6468 0,8012 From Job 0,2643 0,1133 Passing by 0,0598 0,0622 Other 0,0291 0,0234 TRAVEL TIME

Two minutes walking 0,1587 0,2085 Two to five minutes walking 0,1963 0,2524 Five to ten minutes walking 0,1584 0,1521 More than ten minutes walking 0,0673 0,0770 Less or equal five minutes by car 0,1467 0,1582 Five to ten minutes by car 0,0858 0,0674 Ten to fifteen minutes by car 0,0982 0,0425 More than fifteen minutes by car 0,0886 0,0417 WHY SHOPPING

Near form home 0,6015 0,6912 Near from Job 0,0851 0,0436 Passing by 0,1291 0,0419 Low prices 0,0430 0,0424 Variety of brands 0,0066 0,0216 Variety of products in general 0,0196 0,0278 Habit 0,0346 0,0432 Quality products 0,0250 0,0353 Quality of fresh produce 0,0060 0,0047 Cleaning / hygiene shop 0,0068 0,0070 Fast service 0,0119 0,0061 Sympathy in attendance 0,0072 0,0243 Promotions 0,0034 0 Opening hours 0,0077 0,0023 Other 0,0011 0,0018

MONTHLY SPENDING ON PURCHASES FOR THE HOME Mean 266,1543 380,0798

MONTHLY EXPENSES IN STORE Mean 89,3863 200,9802

By estimating these LSM, from the baseline model (homogeneity model or non-

structure segments) to a three-class latent model, we select a two-class latent model by

14

using AIC3 and AICu (Fonseca, 2010a), for model selection, because we have a mixed-

mode dataset (Monthly spending on purchases for the home and Monthly expenses in

store are continuous, the others categorical); these models select automatically the

number of segments, 2-segment in this case, because of graph for AIC3 presents an

elbow (Table 6), by using an information criterion, which is an advantage when

compared with Cluster Analysis.

Because we aim to proceed with a segmentation scheme, as we have explained, we are

going to estimate latent class models, in order to select effective segments and then

being able to target marketing and product positioning.

By estimating parameters of model (1) from the used segmentation base variables we

reach model parameters’ estimates, which we display in Table 7.

Table 7 Parameters estimates of two-class latent model (cont.)

Cluster Size Cluster1

(61 percent) Cluster 2 (39 percent)

QUALITY OF FRESH PRODUCE

Very good 0,0236 0,2793

Good 0,5042 0,4427

Fair 0,3837 0,1650

Bad 0,0498 0,0474

Very bad 0,0011 0

Dn/Da 0,0400 0,0633

TREATMENT

Very good 0,0506 0,6155

Good 0,7112 0,2995

Fair 0,2190 0,0476

Bad 0,0180 0,0053

Very bad 0,0011 0

Dn/Da 0 0,0320

EFFICIENCY OF STAFF

Very good 0,0207 0,4654

Good 0,6971 0,3982

Fair 0,2572 0,0692

Bad 0,0169 0,0142

Very bad 0,0023 0

Dn/Da 0,0058 0,0531

VARIETY OF PRODUCTS

Very good 0,0087 0,1641

Good 0,5063 0,5212

Fair 0,4460 0,2357

Bad 0,0364 0,0322

Very bad 0,0045 0

Dn/Da 0,0023 0,0427

Table 8 summarizes parameters estimates for this model, from the used covariates, used

as inactive, that is, they are not used for parameters’ estimates; thus we use covariates in

order to a better understanding of segments, namely segments’ customers.

15

In these tables’ parameters there are two kinds of probabilities, (1) simple probabilities,

or mixing probabilities (relative segments size), from which we can see that we have 61

percent of customers at segment 1 and 39 percent of customers at segment 2, and (2)

conditional probabilities: probabilities of customer selecting some category for

answering a question, knowing that s(he) belongs to a certain segment; for instance,

0.7033 and 0.7743, from Table 8, are the probabilities of respondents’ answers being

female, given that they belong to segment 1 and segment 2, respectively. Thus it allows

us for concluding that in segment 2 we have majority female respondents.

Table 7 Parameters estimates of two-class latent model (cont.)

Cluster Size Cluster1 Cluster2

0,6119 0,3881

PRODUCT PRESENTATION

Very good 0,0050 0,266

Good 0,6154 0,5839

Fair 0,3593 0,1039

Bad 0,0203 0

Dn/Da 0 0,0462

STORE ENVIRONMENT

Very good 0,0101 0,3700

Good 0,7391 0,5151

Fair 0,2388 0,0627

Bad 0,0120 0,0042

Dn/Da 0 0,0480

CLEANING SHOP

Very good 0,0414 0,4647

Good 0,7527 0,4083

Fair 0,1919 0,0690

Bad 0,0128 0,0082

Very bad 0,0011 0

Dn/Da 0 0,0498

PRICES 0,0414 0,4647

Very good 0,0082 0,0635

Good 0,2689 0,2482

Fair 0,5361 0,4582

Bad 0,1614 0,1741

Very bad 0,0110 0,0112

Dn/Da 0,0144 0,0449

PRODUCT QUALITY PRIVATE LABEL

Very good 0,0144 0,1676

Good 0,4232 0,4726

Fair 0,3942 0,1733

Bad 0,0231 0,0169

Very bad 0,0028 0,0028

Dn/Da 0,1423 0,1668

INTERNET USAGE

At home 0,1851 0,1544

At work 0,0998 0,0471

Both 0,1723 0,1320

No Access 0,5428 0,6664

16

The use of probabilities as parameters is another advantage concerning Cluster Analysis

using, which uses distance measures (there are several), and/or different clustering

methods, then resulting on different solutions.

Table 8 Parameters estimates of two-class latent model by covariates

Cluster Size Cluster1 (61 percent) Cluster 2 (39 percent)

Covariates

SEX

Female 0,7033 0,7743

Male 0,2967 0,2257

AGE

Less than 25 years 0,2274 0,1127

From 25 to 34 years 0,2204 0,1717

From 35 to 44 years 0,2119 0,1993

From 45 to 54 years 0,1646 0,2189

More than 55 years 0,1757 0,2974

OCCUPATION

Independent 0,1814 0,1763

Dependent 0,7885 0,7929

Both 0,0241 0,0261

Dn/Da 0,006 0,0048

EDUCATION

Incomplete 1th cycle 0,0189 0,0502

Complete1th cycle 0,1664 0,1964

2th cycle 0,0984 0,0903

3th cycle 0,1721 0,1519

12º year 0,1624 0,1636

Graduate frequency 0,0991 0,0553

Bachelor 0,0499 0,0404

Graduate 0,2198 0,2349

Dn/Da 0,0129 0,017

FAMILY SIZE

1 person 0,1701 0,1568

2 persons 0,2568 0,2240

3 persons 0,3141 0,2694

4 persons 0,2248 0,2128

5 persons 0,049 0,0703

6 or more persons 0,0179 0,0339

LIVE CYCLE

Single pre family 0,2857 0,1648

Couple pre family 0,0634 0,0477

New Family 0,0861 0,0509

Maturing Family 0,0936 0,0765

Established Family 0,2468 0,3507

Single Post Family 0,0485 0,0764

Couple Post familly 0,0793 0,0741

Older Single 0,0463 0,0711

Older Couple 0,0504 0,0877

INCOME

Les than 400 € 0,0512 0,0562

de 401 € a 798 € 0,1475 0,1551

de 799 € a 1197€ 0,1885 0,1996

de 1198 € a 1596 € 0,1284 0,1461

de 1597 € a 1995 € 0,0741 0,0841

More than 1996 € 0,115 0,105

Dn/Da 0,2841 0,2651

CLASS

Class A 0,0838 0,0766

Class B 0,1515 0,1311

Class C1 0,2689 0,3087

Class C2 0,3384 0,3938

Class D 0,1094 0,137

17

The estimated probabilities allow us for naming segments and also for knowing the

segments profile, based on both segmentation base variables and covariates, to a better

understanding of clusters (Table 9).

Table 9 Profile of retail chain customers

Variables Occasional Customers (61%) Loyal Customers (39%)

USAGE FREQUENCY Occasionaly to Two or three times a

week Everiday

VISIT PATTERN During a week; During the weekend Both situations

COMING FROM From Job; Other From home; Passing by

TRAVEL TIME

Five to ten minutes walking; Five or more minutes by car

Two to five minutes walking;

More than ten minutes walking; Less or equal five minutes by car

WHY SHOPPING

Near from Job; Passing by; Low prices;

Quality of fresh produce; Fast service;

Promotions; Opening hours

Near form home; Variety of

brands; Variety of products in

general; Habit; Quality products; Cleaning / hygiene shop;

Sympathy in attendance; Other

MONTHLY SPENDING ON

PURCHASES FOR THE HOME

266,2 380,1

MONTHLY EXPENSES

IN STORE 89,4 201,0

QUALITY OF FRESH PRODUCE Good; Fair; Bad; Very bad Very good

TREATMENT Good; Fair; Bad; Very bad Very good

EFFICIENCY OF STAFF Good; Fair; Bad; Very bad Very good

VARIETY OF PRODUCTS Fair; Bad; Very bad Very good; Good

PRODUCT PRESENTATION Good; Fair; Bad; Very bad Very good

STORE ENVIRONMENT Good; Fair; Bad; Very bad Very good

CLEANING SHOP Good; Fair; Bad; Very bad Very good

PRICES Good; Fair; Bad; Very bad Very good

PRODUCT QUALITY

PRIVATE LABEL

Fair; Bad; Very bad Very good; Good

INTERNET ACCESS At home; At work; Both No Access

Covariates

SEX Male Female

AGE Up to 44 years old More than 44 years old

OCCUPATION Independent Dependent; Both

EDUCATION 2th and 3th cycle; Graduate frequency;

Bachelor Incomplete and Complete1th

cycle; 12º year; Graduate

FAMILY SIZE Up to 4 persons 5 or more persons

LIVE CYCLE Single pre family; Couple pre family;

New Family; Maturing Family

Established Family; Single Post Family; Older Single; Older

Couple

INCOME More than 1996 € Up to 1995 €

CLASS Class A; Class B Class C1; Class C2 e Class D

As a result, we can name segments as Occasional Customers (Segment 1), with 61

percent of customers, and Loyal Customers (Segment 2), with 39 percent of customers.

Concerning Occasional Customers, we can see that they live near from store and they

are more concerned with opening hours, fast service, and promotions; about store

products they think that all is up to good for quality, cleanness, environment, efficiency,

18

variety, prices; they spend monthly 266€ on purchases for home, 89€ at store (only 33

percent), and they access Internet both at home and at work.

In contrast with Occasional Customers, Cluster 2 or Loyal Customers live far away

from store and they are more concerned with variety of brands, variety of products in

general, quality products, cleaning / hygiene shop and sympathy in attendance; about

store products they think that almost all is very good for quality, cleanness,

environment, efficiency, variety, prices. Next, after identifying the two segments, we

evaluated their socio demographic profiles, regarding the socio economic and

demographic variables, here used as covariates; as we can see, concerning sex, they are

majority male, aged up to 44 years old, independent, with lower level of education,

family size up to 4 persons, as for live cycle, they are majority single pre family or

couple pre family, new family or maturing family, with more than 1996€ for income,

and they are in class A or B.

Quality service is a very important construct; through building on and extending earlier

research ((Aaker, 1991); (E. W. Anderson, Fornell, & Lehmann, 1994); (Oliver, 1997)),

forwards a framework of service that presents quality which leads to satisfaction which

in turn affects loyalty (Harris & Goode, 2004); (Fonseca, 2009) from an empirical

research about customers’ satisfaction, also concludes that quality service leads to

customers’ satisfaction. Loyal Customers spend monthly higher levels of mone (Harris

& Goode, 2004), in our case 380€ on purchases for home, 200€ at store (53 percent!),

and they did not access Internet. Thus, loyal customers buy more, are willing to spend

more, are easier to reach, and, more than that, they act as enthusiastic advocates for

firms.

Again, through socio economic and demographic profile, we can understand better this

segment, by become knowing that they are majority female, oldest, dependent, with

highest level of education, family size with more than 4 persons, established family,

single post family, and older single for live cycle, with less than 1995€ for income, and

they are in class C1, C2 or D.

From the Table 8, it is evident that there is a great difference in the socio economic and

demographic profiles of occasional and loyal customers. The knowledge of the

segments’ structure is very important because of it managerial utility, particularly in

what concerns targeting and positioning. Because of customers in each segment must

respond differently to variations in the marketing mix compared with those in other

19

segments, it implies that this classification into two segments is a true market

segmentation scheme, as the segments exhibit behavioral response differences.

All of this is reinforced by the interviews we made at the beginning of this plan of

marketing segmentation. They confirm the existence of two segments, because all of

them fall into the uncovered segment structure. This is one important aspect of mixed

methodology, merging knowledge, in the case, by using knowledge from the qualitative

research to complement the quantitative research findings. It is of great importance this

mixed methodology, in order to be sure that we reached effective segments, because a

bad segmented market is often worse than making the mass-market assumption.

Finally, in order to know which factors mostly influenced the Internet usage (after

recoding Internet access: 1, use; 2, non-use), we use binary logistic regression. The

model estimation allows us to conclude that education (p value = 0), age (p value = 0),

income (p value = 0), sex (p value = 0.005), and usage frequency (p value = 0.022), by

this order, are the variables which contribute more for explaining Internet usage. The

model also states that concerning education and age the influence is, inverse; as for

income, the influence is direct. This is in accordance with the uncovered segment

structure: occasional customers access internet at home and at job, they are majority

male, youngest, with lower education, but with higher income. Thus, we have good

reasons to believe on the veracity of the reached segment structure, firstly reinforced by

the interviews (qualitative validating quantitative – mixing), and now by quantitative.

As future research, concerning Internet usage, this retail chain organization would like

to get enough information about it, in order to consider future electronic commerce.

5. Conclusions

Firstly, we overview the great importance of market (retail market in particular)

segmentation, because of through the growing knowledge about the market segments,

by tailoring the offering product to different groups, companies are able to more

precisely meet the needs of more customers, and consequently to gain a higher overall

level of share or profit from a market.

Accurate measurement of preferences allows the marketer to gain a deeper

understanding of consumers’ wishes, desires, likes, and dislikes, and thus permits a

better implementation of the tools of the marketer (Steven H. Cohen & Neira, 2003), by

20

the concentration of marketing energy and force on the segments to gain a competitive

advantage within the segment.

Secondly, we apply Latent Segment Modelling to market segmentation, for a dataset

with mixed-mode data, and AIC3 and AICu for model selection. A two-class latent

model fitted well the data, and the two segments are internal homogeneous/external

heterogeneous, they constitute a very parsimonious solution, and marketers can reach

segments separately using the observable characteristics of the segments. This is a good

solution, and the advantages of efficient clusters is that marketers can easily understand

them, being able for developing different and more successful business strategies, and

this allows these managers to focus limited resources on meeting or exceeding the needs

of particular customers (Beynon, Goode, Moutinho, & Snee, 2005).

We tested this solution by using all the information obtained from the qualitative

collection methods, such as interviews, focus groups, participant observation, in

accordance with retail chain marketers at the beginning of the segmentation scheme, in

order to demonstrate that the derived segments will respond differently to variations in

the marketing mix; with the information obtained from the qualitative treatment, we

became knowing that customers are likely to react to an offer, a price or a promotion in

accordance with occasional customers and loyal customers. Thus, throughout a mixed

scheme of market segmentation and AIC3 and AICu information criteria, we can

support the idea that latent segments models are accurate on representing efficiently the

heterogeneous customers, by means of identifying two homogeneous segments which

accommodates the needs and preferences of customers. The great strength of this

pragmatic approach to social science research methodology in general, and market

segmentation in particular, is its emphasis on the connection between epistemological

concerns about the nature of the knowledge that we produce and technical concerns

about the methods that we use to generate that knowledge.

This moves beyond technical questions about mixing or combining methods and puts us

in a position to argue for a properly integrated methodology for the social sciences

(Morgan, 2007).

Our empirical findings indicate that customers’ perceptions are quite different for all the

used variables, including Internet usage; loyal customers indicate that quality is very

important to them, varying from quality of fresh produce to prices. Concerning used

covariates, we find that all of them are important in order to differentiate the retail chain

customers.

21

At last, we conclude that education, age, income, sex, and usage frequency, by this

order, are the variables which contribute more for explaining Internet usage (use/not

use, after recoding), by using a binary logistic regression; is quite evident, from these

results – testing the solution – the effectiveness of the uncovered structure.

To sum up, we must conclude that the used scheme of latent segment models for

segmentation combined with mixed methodology is an advantageous scheme research

in order to uncover the underlying market typology, when compared with most

traditional methods. In future study, this retail chain organization must carefully study

the situation about electronic commerce, given the importance of Internet access, in

order to find out what benefits the customer seeks, and what risks s(he) fears.

Acknowledgements

Thanks go to the anonymous referees for suggesting a number of improvements.

References

Aaker, D. (1991). Managing Brand Equity: Ontario: The Free Press.

Anderson, E. W., Fornell, C., & Lehmann, D. R. (1994). Customer satisfaction, market

share, and profitability: Findings from Sweden. Journal of Marketing, 58(3), 53

- 66.

Anderson, W. T., Jr , E. P Cox, I., & Fulcher, D. G. (1976). Bank Selection Decisions

and Market Segmenta. Journal of Marketing, 40, 40-45.

Arabie, P., Carroll, J. D., DeSarbo, W., & Wind, J. (1981). Overlapping Clustering: A

New Method for Product Positioning. Journal of Marketing Research,

XVIII(August), 310-317.

Assael, H., & Roscoe, A. M. (1976). Approaches to Market Segmentation Analysis.

Jounal of Marketing, 40, 67-76.

Baker, G. A., & Burnham, T. A. (2001). Consumer Response to Genetically Modified

Foods: Market Segment Analysis and Implications for Producers and Policy

Makers. Journal ofAgricultural and Resource Economics, 26, 387-403.

Bass, F. M., Pessemier, E. A., & Tigert, D. J. (1969). A Taxonomy of Magazine

Readership Applied to Problems in Marketing Strategy and Media Selection.

Journal of Business, 42(337-363).

Beane, T. P., & Ennis, D. M. (1987). Market-Segmentation - a Review. European

Journal of Marketing, 21(5), 20-42.

Becker, B. W., Brewer, B., Dickerson, B., & Magee, R. (1985). The influence of

personal values on movie preferences. In Austin, B. A. (Ed.), Current Research

in Film: Audiences, Economics, and the Law (pp. 37-50): Ablex Publishing

Company, Norwood, NJ.

Beynon, M., Goode, M. M. H., Moutinho, L., & Snee, H. (2005). Modelling

Satisfaction with Automated Banking Channels, Using Variable Precision

Rough Set Theory. Services Marketing Quarterly, 26(4), 77-94.

22

Biggadike, E. R. (1981). The Contributions of Marketing to Strategic Management. The

Academy of Management Review, 6(4), 621-632.

Bock, T., & Uncles, M. (2002). A taxonomy of differences between consumers for

market segmentation. Intern. J. of Reasearch in Marketing, 19, 215-224.

Bozdogan, H. (1994). Mixture-Model Cluster Analysis using Model Selection criteria

and a new Informational Measure of Complexity. In Bozdogan, H. (Ed.),

Proceedings of the First US/Japan Conference on the Frontiers of Statistical

Modeling: An Approach, 69-113 (Vol. 2, pp. 69-113): Kluwer Academic

Publishers.

Calantone, R. J., & Sawyer, A. G. (1978). The Stability of Benefit Segments. Journal of

Marketing Research, XV, 395-404.

Carroll, J. D., & Green, P. E. (1997). Psychometric Methods in Marketing Research:

Part I, Conjoint Analysis. Journal of Marketing Research, XXXII 385-391.

Chen, J. S. (2003). MARKET SEGMENTATION BY TOURISTS’ SENTIMENTS.

Annals of Tourism Research, 30(1), 178-193.

Cohen, S. H., & Neira, L. (2003). Measuring preference for product benefits across

countries: Overcoming scale usage bias with Maximum Difference Scaling.

Paper presented at the ESOMAR 2003 Latin America Conference Proceedings.

ESOMAR: Amsterdam, The Netherlands.

Cohen, S. H., & Ramaswamy, V. (1994). Latent Segmentation Models. New tools assist

researchers in market segmentation Marketing Research, 10 (2), 14-21.

Cohen, S. H., & Ramaswamy, V. (1998). New Tools for Market Segmentation: An

Introduction to Latent Class Models. Marketing Research, 10(2), 14-21.

Dempster, A. P., Laird, N. M., & Rubin, D. B. (1977). Maximum Likelihood from

incomplete Data via EM algorithm. Journal of the Royal Statistics Society, B,

39, 1-38.

DeSarbo, W. S., Degeratu, A. M., Wedel, M., & Saxton, M. K. (2001). The spatial

representation of market information. Marketing Science, 20(4), 426-441.

DeSarbo, W. S., Ramaswamy, V., & Cohen, S. H. (1995). Market Segmentation with

Choice-Based Conjoint Analysis. Marketing Leners, 137-147.

Dibb, S. (1998). Market segmentation: strategies for success. Marketing Intelligence &

Planning, 16, 394–406.

Dibb, S. (1999). Criteria guiding segmentation implementation: reviewing the evidence.

Journal of Strategic Marketing, 7, 107-129.

Dibb, S., Stern, P., & Wensley, R. (2002). Marketing knowledge and the value of

Segmentation. Marketing Intelligence & Planning, 20, 113-119.

Dillon, W. R., & Kumar, A. (1994). Latent structure and other mixture models in

marketing: An integrative survey and overview, chapter 9 in R.P. Bagozi (ed.),

Advanced methods of Marketing Research, 352-388, Cambridge: Blackwell

Publishers. In.

Dolnicar, S. (2008). Marketing Segmentation in Tourism,

http://ro.uow.edu.au/commpapers/556

Fennell, G., Allenby, G. M., Yang, S., & Edwards, Y. (2003). The effectiveness of

Demographic and Psychographic Variables for Explaining Brand and Product

Category Use. Quantitative Marketing and Economics, 1, 233-244.

Fonseca, J. R. S. (2009). Customer satisfaction study via a latent segment model.

Journal of Retailing and Consumer Services, 16, 352-359.

Fonseca, J. R. S. (2010). On the Performance of Information Criteria in Latent Segment

Models. Paper presented at the Proceedings of World Academy of Science,

23

Engineering and Technology, Volume 63, March 2010, WASET 2010, Rio de

Janeiro, March 29-31, 2010, p. 45-72. ISSN: 2070-3724. .

Fonseca, J. R. S., & Cardoso, M. G. M. S. (2005). Retail Clients Latent Segments In

Bento, C., Cardoso, A. & Dias, G. (Eds.), Progress in Artificial Intelligence

(Vol. Lecture Notes in Computer Science, pp. 348-358): Springer

Fonseca, J. R. S., & Cardoso, M. G. M. S. (2007a). Mixture-Model Cluster Analysis

using Information Theoretical Criteria. Intelligent Data Analysis, 11(2), 155-

173.

Fonseca, J. R. S., & Cardoso, M. G. M. S. (2007b). Supermarket Customers Segments

Stability. Journal of Targeting, Measurement and Analysis for Marketing, 15(4),

210-221.

González-Benito, Ó., Greatorex, M., & Muñoz-Gallego, P. A. (2000). Assessment of

potential retail segmentation variables. An approach based on a subjective MCI

resource allocation. Journal of Retailing and Consumer Services, 7, 171-179.

Granzin, K. L. (1981). An Investigation of the Market for Generic Products. Journal of

Retailing, 57, 39-55.

Green, P. E., & Carmone, F. J. (1968). The Performance Structure of the Computer

Market: A Multivaiate Appproach. Economic and Business Bulletin, 20(1-11).

Green, P. E., Frank, R. E., & Robinson, P. J. (1967). Cluster Analysis in Test Market

Selection. Management Science, 13(B), 387-400.

Green, P. E., & Krieger, A. M. (1991). Segmenting Markets With Conjoint Analysis

Journal of Marketing, 55 20-31.

Green, P. E., & Srinivasan, V. (1990). Conjoint Analysis in Marketing: New

Developments With Implications for Research and Practice. Journal of

Marketing, 54(2), 3-19.

Greeno, D. W., Sommers, M. S., & Kernan, J. B. (1973). Personality and Implicit

Behavior Patterns. Journal of Marketing Research, 10, 63-69.

Harris, L. C., & Goode, M. M. H. (2004). The four levels of loyalty and the pivotal role

of trust: a study of online service dynamics. Journal of Retailing, 80, 139-158.

Harvey, J. W. (1990). Benefit Segmentation for Fund Raisers. Journal of the Academy

of Marketing Science, 18 77-86.

Heilman, C. M., & Bowman, D. (2002). Segmenting consumers using multiple-category

purchase data. Intern. J. of Reasearch in Marketing, 19, 225.252.

Heuvel, D. A., & Devasagayam, P. R. (2004). Multivariate cluster analysis model of

benefit-based market segmentation: a case study from the recreation and leisure

industry. http://www.danavan.net/pdf/cluster.analysis.publication.pdf

Hofstede, F. T., & Steenkamp, J.-B. E. M. (1999). International market segmentation

based on consumer-product relations. Journal of Marketing Research, 36, 1-17.

Hruschka, H., & Natter, M. (1999). Comparing performance of feedforward neural nets

and K-means for cluster-based market segmentation. European Journal of

Operational Research, 114, 346-353.

Jain, S. C. (1993). Marketing Planning & Strategy, 4th ed., : South-Western Publishing

Co., Cinciannati, OH.

Jayawardhena, C., Wright, L. T., & Dennis, C. (2007). CONSUMERS ONLINE:

INTENTIONS, ORIENTATIONS and SEGMENTATION. International

Journal of Retail and Distribution Management, 35(6), 515-599.

Jedidi, K., Jagpal, H. S., & DeSarbo, W. S. (1997). Finite-Mixture Structural Equation

Models for Response-Based Segmentation and Unobserved Heterogeneity

Marketing Science, 16, 39-59.

24

Kamakura, W. A., Kim, B. D., & Lee, J. (1996). Modeling preference and structural

heterogeneity in consumer choice. Marketing Science, 15(2), 152-172.

Kau, A. K., Tang, Y. E., & Ghose, S. (2003). Typology of online shoppers. Journal of

Consumer Marketing, 20(2), 139-156.

Kiel, G. C., & Layton, R. A. (1981). Dimensions of Consumer Information Seeking

Behavior. Journal of Marketing Research, 18, 233-239.

Kim, B.-D., Srinivasan, K., & Wilcox, R. T. (1999). Identifying Price Sensitive

Consumers: The Relative Merits of Demographic vs. Purchase Pattern

Information. Journal of Retailing, 75(2), 173-193.

King, G., Tomz, M., & Wittenberg, J. (2000). Making the Most of Statistical Analyses:

Improving Interpretation and Presentation. American Journal of Political

Science, 44(2), 341-355.

Kotler, P. (1972). Marketing management; analysis, planning, and control (2nd edition

ed.): Prentice-Hall (Englewood Cliffs, N.J)

Kotler, P. (1997). Marketing Management Analysis, Planning, Implementation, and

Control, 9th ed., : Prentice-Hall International, Englewood Cliffs, NJ.

Lee, C.-K., Lee, Y.-K., & Wicks, B. E. (2004). Segmentation of festival motivation by

nationality and satisfaction. Tourism management, 25, 61–70.

Leech, N. L., Dellinger, A. B., Brannagan, K. B., & Tanaka, H. (2010). Evaluating

Mixed Research Studies: A Mixed Methods Approach. Journal of Mixed

Methods Research, 4(1), 17-31.

Leroux, B. G., & Puterman, M. L. (1992). Maximum-Penalized-Likelihood Estimation

for Independent and Markov-Dependent Mixture Models. Biometrics, 48(2),

545-558.

Levin, N., & Zahavi, J. (2001). Predictive Modeling using Segmentation. JOURNAL

OF INTERACTIVE MARKETING, 15 (2), 2-22.

Lin, C.-F. (2002). Segmenting customer brand preference: demographic or

psychographic. Journal of Product & Brand Management, 11(4), 249-268.

Lockshin, L. S., Spawton, A. L., & Macintosh, G. (1997). Using product, brand and

purchasing involvement for retail segmentation. Journal of Retailing and

Consumer Services, 4(3), 171-183.

McCarty, & Hastak. (2007). Segmentation approaches in data-mining: A comparison of

RFM, CHAID, and logistic regression. Journal of Business Research, 60, 656–

662.

McLachlan, G. F., & Peel, D. (2000). Finite Mixture Models (first ed.): John Wiley &

Sons.

Montgomery, D. B., & Silk, A. J. (1971). Clusters of Consumer Interests and Opinion

Leaders' Sphere of Inluence. Journal of Marketing Research, 8(317-321).

Moore, W. L. (1980). Levels of Aggregation in Conjoint Analysis: An Empirical

Comparison. Journal of Marketing Research, XVII 516-523.

Morgan, D. L. (2007). Paradigms Lost and Pragmatism Regained. Methodological

Implications of Combining Qualitative and Quantitative Methods. Journal of

Mixed Methods Research, 1(1), 48-76.

Morrison, B. J., & Sherman, R. C. (1972). Who Responds to Sex in Advertising.

Journal of Advertising Research, 12, Journal of Advertising Research.

Oliver, R. L. (1997). Satisfaction a behavioural perspective on the consumer. New

York: New York. McGraw-Hill.

Onwuegbuzie, A. J., & Leech, N. L. (2005). Taking the "Q" Out of Research: Teaching

Research Methodology Courses Without the Divisive Between Quantitative and

Qualitative Paradigms. Quality & Quantity, 39, 267-296.

25

Palmer, R. A., & Miller, P. (2004). Segmentation: Identification, intuition, and

implementation. Industrial Marketing Management, 33, 779-785.

Porter, M. E. (1985). Competitive Advantage Creating and sustaining superior

performance: New York Free Press.

Punj, G., & Stewart, D. W. (1983). Cluster Analysis in Marketing Research: Review

and Suggestions for Application. Journal of Marketing Research, XX (May

1983), 134-148.

Schaninger, C. M., Lessig, V. P., & Panton, D. B. (1980). The Complementary Use of

Multivanate Procedures to Investigate Nonlinear and Interactive Relationships

Between Personality and Product Usage. Journal of Marketing Research, 17,

119-124.

Scriven, J., & Ehrenberg, A. (2004). Executive Summaries - Consistent Consumer

Responses To Price Changes. Australasian Marketing Journal, 12(3), 19-20.

Segal, M. N., & Giacobbe, R. W. (1994). Market Segmentation and Competitive

Analysis for Supermarket Retailing, International Journal of Retail &

Distribution Management, 22, 38-48.

Sexton, D. E., Jr (1974). A Cluster Analytic Approach to Market Response Functions.

Journal of Marketing Research, 11, 109-114.

Sharma, A., & Lambert, D. M. (1994). Segmentation of Markets Based on Customer

Service. International Journal of Physical Distribution & Logistics

Management, 24(4), 50-58.

Sharp, A., Romaniuk, J., & Cierpicki, S. (1998). The performance of segmentation

variables: A comparative study: Unpublished manuscript, Marketing Science

Centre, Adelaide, Australia.

Simcock, P., Sudbury, L., & Wright, G. (2006). Age, perceived risk, and satisfaction in

consumer decision making: A review and extension. Journal of Marketing

Management, 22, 355–377.

Smith, W. R. (1956). Product differentiation and market segmentation as alternative

marketing strategies. Journal of Marketing, 21, 3-8.

Sun, S. (2009). An Analysis on the Conditions and Methods of Market Segmentation.

International Journal of Business and Management, 4(2), 63-70.

Thomas, J. S., & Sullivan, U. Y. (2005). Managing Marketing Communications with

Multichannel Customers Journal of Marketing, 69 239–251.

Tsai, C.-Y., & Chiu, C.-C. (2004). A purchase-based market segmentation

methodology. Expert Systems with Applications, 27 265–276.

Uncles, M., & Lee, D. (2006). Brand purchasing by older consumers: An investigation

using the Juster scale and the Dirichlet model. Market Letters, 17, 17-29.

Vermunt, J. K., & Magidson, J. (2002). Latent class cluster analysis.Unpublished

manuscript, Cambridge University Press.

Vitale, D. C., Armenakis, A. A., & Feild, H. S. (2008). Integrating Qualitative and

Quantitative Methods for Organizational Diagnosis. Possible Priming Effects?

Journal os Mixed Methods Research, 2(1), 87-105.

Vriens, M. (2001). Market Segmentation. Analytical Developments and Applications

Guidlines: Millward Brown IntelliQuest.

Wedel, M., & Kamakura, W. A. (1998). Market Segmentation: Concepts and

methodological foundations (Second ed.). Boston: Kluwer Academic Publishers.

White, H. (2002). Combining Quantitative and Qualitative Approaches in Poverty

Analysis. World Development, 30(3), 511–522.

Wilkie, W. L. (1990). Consumer Behavior, 2nd edition: John Wiley & Sons, Inc.

26

Wind, Y., Douglas, S. P., & Perlmutter, H. V. (1973). Guidelines for Developing

International Marketing Strategies Journal of Marketing, 37 14-23.


Recommended