Segmentation with value(s)

Segmentation with value(s)

An assessment of the Global Value List as basis for global market

segmentation

by means of the simultaneous clustering of countries and consumers

Author: Eline

de Vries

Fist Supervisor: Prof. dr. T.H.A. Bijmolt

Second Supervisor: Dr. J.E.M

Van Nierop

Place/Date:

Segmentation with value(s)

An assessment of the Global Value List as basis for global market

segmentation

by means of the simultaneous clustering of countries and consumers

Author: Eline Louise Elisabeth de Vries

Student number: 1337025

Rijksuniversiteit Groningen

Faculty of Management & Organization and

Economics

1

st

supervisor: Prof. Dr. T.H.A. Bijmolt

2

nd

supervisor:

Dr. J.E.M Van Nierop

_{Unilever Research & Development Vlaardingen}

Supervisor: C.

Boucon

Groningen, July 2006

Foreword

In front of you lies my Master thesis which I have made to finalise my Master Research Based Marketing at the University of Groningen, The Netherlands.

The research, on which this thesis is based, is performed at Unilever R&D Vlaardingen. This multinational gave me the opportunity to assess whether the in-house developed Global Value List, a measure instrument to measure social values, was a useful tool to conduct global market segmentation research. The increasing globalisation and competition demanded for such a tool; Unilever however was not sure whether they had the right one at hand.

The assignment formed a great challenge. Not only because of the huge sample size – over 44 thousand respondents had all answered 63 questions – which laid a heavy burden on computer capacity, but also because of the time limit I imposed myself. I wanted to perform the research between February and July of the year 2006.

However, challenges are where I live for and overall I really enjoyed performing this research.

Yet, the successful finishing of my thesis would not have been possible without the help of certain persons, which I like to acknowledge here.

First of all, I would like to thank Prof. Dr. T.H.A. Bijmolt, who has taken his function as first supervisor very seriously and provided me with excellent comments on the several draft versions.

In addition, I would like to thank my second supervisor here as well, Dr. J.E.M Van Nierop, for giving me feedback on my final draft version.

Furthermore, I would like to thank Unilever and in particular my supervisor Claire Boucon for giving me this opportunity and letting me perform the research with such a high level of independence.

Finally, I thank my friends for giving me a pleasant time during my entire study period, as well as my family for supporting me unconditionally.

Last but not least, I thank my boy-friend Vincent, for letting me see the positive sides of another crashed computer analysis.

Thank you.

Eline de Vries

Management Summary

In order to evaluate whether the Global Value List provides a useful basis for global market segmentation by means of simultaneous clustering of countries and consumers, on the basis of which the viability of rolling out new products across new countries can be assessed, the following analyses were performed.

First the important criteria one has to take into account while performing global market segmentation research were identified. The following criteria turned out to be hereby of importance. First the conceptual issues, which consists of the level of aggregation, evaluation criteria for market segmentation bases and construct equivalence.

In addition, the methodological issues as measure equivalence, sampling equivalence, stimulus commonality and management centrality of great importance as well.

All these conceptual as the methodological issues have to be satisfied in order to assure cross-cultural validity of the research instrument and the results.

Not all these requirements were taken into account however, during the development of the Global Value List and the data collection phase. Unilever is therefore recommended to improve their awareness and usage of these issues, since successful global market segmentation starts obviously with a cross-cultural valid research instrument.

Second, the food-specific relationship between social values was investigated, which turned out to be of a direct nature. Although the social value-consumer behaviour relationship in general tended to be indirect, with regard to food no mediating elements were assumed to exist. This makes relative accurate prediction of food-related consumers behaviour based on social values likely and can be explained by the fact that food is considered a low involvement product. No well defined attitudes are hereby defined in advance and no thoroughly evaluation of the product’s attributes takes place before purchase either. In contrast, the product choice is solely based on one’s social values.

In addition, values were characterized as relative abstract concepts, which transcend specific objects and situations. Moreover, values are highly stable and change only little if at all during one’s life. All of this makes social values more generalizable and cross-cultural applicable than the often used concepts as attitudes, needs and lifestyles as segmentation basis.

Segmentation performed on social values leads to highly stable segments, with good identifiability, good substantiality, moderate responsiveness and moderate actionability. After the social values had proven themselves suitable to use as segmentation basis, the data collected based on the Global Value List was used. A factor analysis performed on row-standardized data, identified 11 latent dimensions, interpreted as the social values: benevolence, family care, conformity, power, esteem, recognition, achievement, reward, social competence, relatedness and traditionalism.

The probabilities to belong to a certain consumer segment while judging a particular value as important or non-important, showed however that differences between consumers based on social values are small.

Finally, the validity of the segmentation model, assessed by split-half validation, proved to be rather good. Although some different and unexpected combinations of countries were made in both models, the segmentation of the consumer segments turned out to be almost identical. All consumer segments from above could be found within the validation models as well.

Concluding could therefore be said that the Global Value List provides a useful basis for global market segmentation by means of simultaneous clustering of countries and consumers, performed by multi-level latent class analysis. Based on the relative importance attached to the social values by the different consumer segments within the country segments, the viability of rolling out new products across new countries can be assessed. Since for food a direct relationship exists between social values and consumer behaviour, it is likely that such an assessment can be done relatively accurate.

However, since no actual buying data was available for all countries, the real predictive ability of analyses based on the Global Value List remains to be tested. This is a nice opportunity for further research.

The last two recommendations for further research are moreover; a more profound study after the course of the CAIC and AWE statistics with large sample sizes and within multi-level latent class analysis, and the adjustment of the Global Value List towards a smaller set of items. This will reduce the demanded computer time and satisfy the requirement of measure instruments that are as short as possible. It kills two birds with one stone.

Table of Contents

Foreword

…….……… . . .

. . .3

Management Summary

... 4

Table of Contents

... 6

1. Introduction

...10

1.1. Preface ...10

1.2. Research design...10

1.2.1. Research goal ...10

1.2.2. Research question ...10

1.2.3. Partial questions ...10

1.2.3. Research question ...11

1.3. Limitiations...11

1.4. Scientific and Managerial Relevance ...11

1.4.1. Scientific relevance ...11

1.4.2. Managerial relevance ...12

1.5. Structure of thesis ...12

_Toc141672599

2. Global Market Segmentation

...13

2.1. Introduction ...13

2.2. Market Segmentation Strategies ...13

2.2.1. Country-as-segment segmentation ...14

2.2.2. International segmentation ...14

2.2.3. Global segmentation...14

2.3. Conceptual issues...15

2.3.1. Level of aggregation ...15

2.3.2. Market Segmentation Bases...16

2.3.2.1. Evaluation criteria

...17

2.3.2.2. Values as segmentation bases

...17

2.3.4. Construct equivalence ...18

2.4. Methodological issues ...18

2.4.1 Measure equivalence ...18

2.4.2. Sampling equivalence...20

2.4.3. Stimulus commonality...20

2.4.4. Management centrality...20

2.5. Conclusion...21

3. Social Values

...22

3.1. Introduction ...22

3.2. Definition of social values ...22

3.3. Difference between attitudes and values...22

3.4. Value-behaviour relationship ...23

3.4.1. Indirect relationship...23

3.4.3. Restrictions ...25

3.5. Values and food ...25

3.6. Cultural influence on values ...26

3.7. Conceptual value framework...27

3.8. Conclusion...28

4. Research design

...30

4.1. Introduction ...30

4.2. Data description ...30

4.2.1. Main characteristics...30

4.2.2. Global Value List...31

4.2.3. Socio-demographic data ...31

4.3. Biases ...31

4.4. Missing values ...33

4.5. Weighing ...33

4.6. Standardization of data ...33

4.7. Factor Analysis...34

4.7.1. Conditions ...34

4.7.2. Factor selection criteria...34

4.7.3. Factor Rotation ...35

4.7.4. Interpretation of loadings ...35

4.7.5. Communalities...35

4.7.6. Labelling of factors...35

4.8. Multi-level Latent Class Analysis ...35

4.8.1. Latent class modelling ...35

4.8.2. Multi-level latent class modelling...36

4.8.3. Formulation of the model...36

4.8.4. Determination Number of Segments ...37

5. Assessment of global value structure

...39

5.1. Introduction ...39

5.2. Principal Axis Factoring over Original data ...39

5.2.1 Conditions ...39

5.2.2. Overall analysis ...40

5.2.2.1. Number of factors...40 5.2.2.2. Rotation method...42 5.2.2.3. Communalities...42 5.2.2.4. Labelling of factors

...42

5.3. Per country analyses ...45

5.4. Subconclusion...47

5.5. Principal Axis Factoring over Standardized Data ...47

5.5.1. Introduction...47

5.5.2. Conditions ...47

5.5.3. Overall analysis ...48

5.5.4. Number of factors ...48

5.5.5. Significance of loadings...48

5.5.6. Communalities...48

5.5.7. Labelling of factors...48

5.7. Conclusion...52

6. Simultaneous country and consumer segmentation

...54

6.1. Introduction ...54

6.2. Estimation of the model...54

6.2.1. Technical characteristics ...54

6.2.2. Original data...54

6.2.3. Determination Number of segments...54

6.3. Description of model: 7 Country and 8 Consumer segments ...57

6.3.1. Country segments ...57

6.3.1.1. Entropy measures...57

6.3.2. Consumer segments...59

6.3.2.1. Entropy measures...59

6.3.2.2. Possession of social values

...59

6.3.3. Combination country and consumer segments...60

6.3.4. Probabilities...62

6.3.4.1. Important values

...62

6.3.4.2. Unimportant values...62

6.4. Conclusion...63

7. Assessment of validity segmentation model

...64

7.1. Introduction ...64

7.2. Model formulation ...64

7.2.1. Number of segments...64

7.2.2. Comparison ...64

7.2.2.1. Segment sizes

...64

7.2.2.2. Classification statistics...65

7.2.3. Composition of country segments ...65

7.2.4. Composition of consumer segments...66

7.3. Conclusion...67

8. Conclusion & Recommendations

...69

8.1. Introduction ...69

8.2. Conclusion...69

8.3. Recommendations ...71

8.3.1. Recommendations for Unilever...71

8.3.2. Recommendations for further scientific research ...71

References

...73

Appendices

...Error! Bookmark not defined.

Appendices belonging to Introduction

...Error! Bookmark not defined.

Appendix A1.Year and agency of data collection

...Error! Bookmark not defined.

Appendix A2

.

Means on value items per country

...Error! Bookmark not defined.

Appendices belonging to 3.Social Values

...Error! Bookmark not defined.

Appendix B1

.

Overview Global Value List

...82

Appendices belonging to

4.Research Design

...83

Appendices belonging to

5. Assesment of global value structure

...85

Appendix D1.FA Unstandardized data: correlation matrix

...86

Appendix D2.FA Unstandardized data: total variance explained

...87

Appendix D3.FA Unstandardized data: rotated factor matrix

...88

Appendix D4.FA Unstandardized data: results with varimax rotation

...89

Appendix D5.FA Unstandardized data: communalities

...93

Appendix D6.FA Standardized data: correlation matrix

...94

Appendix D7.FA Standardized data: total variance explained

...95

Appendix D8.FA Standardized data: scree plot

...96

Appendix D9.FA Standardized data: communalities

...97

Appendices belonging to 6. Simultaneous country and consumer segmenation. . …

98

Appendix E1

.

Technical input in Latent Gold 4.0

...99

Appendix E2. Output based on original data

...100

Appendix E3

.

Overview CAIC values

...102

Appendix E4

.

Overview AWE values based on 100 percent of data

...103

Appendix E5

.

Overview CAIC and AWE values based on 10 percent of data

...104

Appendix E6

.

Overview BIC values

...105

Appendix E7

.

Overview of posterior country probabilities

...106

Appendix E8. Output Latent Gold - Profile consumer segments

...110

Appendix E9

.

Differences between value items

...113

Appendix E10.Relative importance of values by consumer segments

...115

Appendix E11

.

Calculation of probabilities

...118

Appendix E12

.

Calculation of the mean probabilities

...129

Appendices belonging to 7.Validation

...141

Appendix F1

.

Validation: CAIC and AWE values based on70 percent of data

...142

Appendix F2

.

Model 70: Output Latent Gold - Profile consumer segments

...143

Appendix F3

.

Model 30: Output Latent Gold - Profile consumer segments

...146

Appendix F4

.

Peason’s correlation between Model-100 and Model-70 / Model-30

...149

Appendices belonging to

Conclusion & Recomendation

...151

Appendix G.1Validation: CAIC and AWE values based on 1 percent of data

...152

1. Introduction

1.1. Preface

For an organization as Unilever, and actually for any organization, it is of great importance to have a clear picture of the needs of consumers. To gain insight in the needs of consumers and which products and services they value, strategic consumer research is required.

Unilever has performed research in the form of habits and attitudes studies, in which one attempted to understand consumer behaviour and to identify the underlying reasons for their behaviour by measuring respondents’ actual behaviour and their attitudes towards food products and meal preparation (Klerk et al., 2005).

However, these studies were performed on a country by country basis, while Unilever is a global company, operating across the globe. As a result, one realised, Unilever would benefit from economies of scale if it started to expand the scope of the research performed and to roll out similar products across the world. Yet, this policy demanded a global instead of a country based model to perform the segmentation. And this was not the only change required. Using attitudes to predict the suitability of products to roll them out globally did not seem the best way. Attitudes in general after all, but attitudes towards food and meal preparation as well, differ substantially from country to country and depend heavily on a country’s culture (Klerk et al., 2005). Instead, consumers must be grouped together on characteristics that are culturally independent in order to be able to obtain real global and cross-border segments.

This condition of cultural-independency makes many more widely used segmentation bases, as social structures, education systems and living standards, unsuitable for international and global market segmentation. They perform quite poorly once implemented in an international setting (Nachum, 1994).

However, Unilever expected this requirement of universality to be satisfied by social values, since values are of a more abstract level then attitudes and all people across the globe possess certain values, as was the thought. Consequently Unilever developed the Global Value List, a questionnaire to collect the social value data and proceeded to collect the data.

In this report will be investigated whether social values are indeed suitable for use in global market segmentation. If turns out that it is possible to group consumers and countries by means of multi-level latent class analysis (Bijmolt et al., 2004), based on the values measured by the Global Value List, a promising global segmentation method is born.

1.2. Research design

1.2.1. Research goal

Main goal of this study is to provide insight whether the Global Value List is a useful basis for global market segmentation, performed by multi-level latent class analysis.

1.2.2. Research question

The research question of this study is the following:

1.2.3. Partial questions

In order to answer the research question above, the following partial questions are formulated:

Does the Global Value List provide a useful basis for global market segmentation by means of simultaneous clustering of countries and consumers, on the basis of which the viability of

1.2.3. Research question

In order to answer the research question above, the following partial questions are formulated:

• What are important criteria one does have to take into account while performing global market segmentation research?

• What is the relationship between social values and behaviour?

• Are social values suitable to use as global segmentation basis?

• What does the underlying value structure of the Global Value List look like?

• What does a simultaneous country and consumer segmentation performed on Global Value List data look like?

• How valid is the method of simultaneous country and consumer segmentation performed on Global Value List data?

1.3. Limitations

This study takes place in a consumer context as opposed to a Business to Business context. Moreover, the data is of an international nature, however mainly originating from countries within Europe and Africa.

Goal of this study is in addition not to give thé truth with regard to global market segmentation in general, but to find the optimal segmentation given the data of the Global Value List. Whether this is indeed the best method to perform global market segmentation can not be said, since no data is available about actual consumer behaviour within all countries, which made a real evaluation of the segments impossible. Neither is any data available about the performance of alternative segmentation methods.

The product restriction was: the final research report for Unilever must contain an assessment of the usability of the Global Value List as basis for global market segmentation. The process restrictions finally: the study at Unilever must be performed between February and June of the year 2006. After that, the resulting thesis based on this research could be finished externally from Unilever.

1.4. Scientific and Managerial Relevance

1.4.1. Scientific relevance

In this report a framework is given of the value-behaviour relationship and the intermediating concepts. Such a framework could not be found yet in literature.

In addition, in this report is investigated whether the multi-level latent class analysis as segmentation method is applicable with highly abstract, cognitive value data instead of ownership data about financial services as is done by Bijmolt et al. (2004). In other words: with cognitive data is stead of actual behaviour.

Based on these global data are assessed the generalizability of social values and the validity of social values as segmentation basis within a multi-level latent class analysis. With this, the research described in this report meets the wishes of several scholars to examine the validity and generalizability of theories and models in a more international setting. Steenkamp (2005) called recently “to move out the ‘U.S. silo’ and to conduct more research on an international basis.”

This is exactly what is done in this report, which makes the results of substantial scientific relevance.

1.4.2. Managerial relevance

Managers can use the value-behaviour framework to enhance their understanding of the value-behaviour relationship and the intermediating concepts. In this way they can form a clear picture of the factors that can be used in order to influence the behaviour of their consumers.

Moreover, until now only limited attention has been paid to the specific requirements of global marketing segmentation research. In this report an overview of the conceptual and methodological issues of global market segmentation research will be given, which should be taken in consideration while conducting global market segmentation research. Since there are serious pitfalls concerning research performed across national borders which are easily overseen or underestimated, this overview is of great importance for marketing practice in the future.

1.5. Structure of thesis

The structure of this thesis is the following:

First the concept of global market segmentation will be elaborated on in chapter 2. Different market segmentation strategies will be explained, as well as the conceptual and methodological issues which one has to keep in mind while conducting market segmentation across national borders.

After that the content of social values and their relationship with consumer behaviour will be described in the third chapter. Also the cultural influence on values and the development of the Global Value List will be explained here.

Following this literature based part, in chapter 4 the research design will be described. Issues like the threat of missing values, the standardization of the data and the specifications of the different statistical analyses will be described.

Subsequently, the results of the analyses will be reported in chapter 5 and 6. Followed by a validation part in chapter 7.

Finally, a conclusion will be drawn and implications will be given in detail in chapter 8.

Global Market Segmentation Criteria Chapter 2 Global Market Segmentation Analysis Chapter 6 Use of Value data Assessment Value-Behaviour Relationship Chapter 3 Assessment Universality of Social Values Chapter 5 Validation Segmentation method Chapter 7 Conclusion & Recommendation Chapter 8

Figure 1.5. Research model

2. Global Market Segmentation

2.1. Introduction

Yet in 1956 the classic concept of segmentation was originally proposed. But even before that, segmentation had been carried out for many years. After all, the

identification of consumer needs and satisfying them effectively is a central part of the marketing philosophy and market segmentation is a useful method to carry this out. When it comes to the application of segmentation itself, not much has changed since then. What has changed however is the scope of segmentation.

In recent years firms have expanded more and more globally. International or even global marketing research plays an important role in this development. However, only limited attention has been paid to the specific requirements global market

segmentation has to satisfy. Although the process of conducting research for global market segmentation does not differ from domestic marketing research, there actually are substantial pitfalls and problems associated with extending the scope of research beyond a single country’s borders (Douglas and Craig, 2006).

In this chapter, first the different segmentation strategies will be explained in more detail. After that an overview of the conceptual and methodological issues of global market segmentation research will be given, which should be taken into consideration in order to tackle global markets. This is knowledge which will be of use in the following analyses in this report, but is of importance for marketing practice in the future as well. After all, although this chapter is focused on Unilever, the content is generalizable to other companies conducting market research on an international scale, as well.

2.2. Market Segmentation Strategies

Segmentation analysis in general and – of more importance, since Unilever is a globally operating company – globally applied market segmentation as well, attempts to identify people within the market who are more likely to be influenced by

marketing effort with respect to a particular product or service, than the rest of the market population (Boote, 1981).

segmentation, especially if applied on a global scale, increases marketing efficiency by directing effort more specifically toward the designated segment in a manner

consistent with that segment’s characteristics (Boote, 1981).

Market segmentation can be implemented within one single country. This

segmentation strategy is of no relevance for this research however, because Unilever operates in more than one single country. Therefore this strategy will not be

considered any further.

In addition, since no consistency exists in the segmentation literature about

terminology of the segmentation methods considering multiple countries, in this report the following three strategies are distinguished, based on several authors.

2.2.1. Country-as-segment segmentation

A form of international segmentation which is seen often is to adopt a multi-domestic strategy, where each country represents a separate segment. With such a ‘countries-as-segments’ (Ter Hofstede et al., 2002) segmentation, countries are selected on the basis of their local advantages and national brands are tailored to the needs shared by groups of consumers in the same country (Steenkamp & Ter Hofstede, 2002). In each country consumers are targeted by distinct advertising, distribution and pricing

strategies and competition is managed at a national level. Segments are accessible and entry strategies cost efficient (Ter Hofstede et al., 2002). Actually, organisations

operating according to a multi-domestic approach simply apply the standard

segmentation techniques that are developed for domestic markets in an ‘international’ setting; namely using multiple single countries.

However, since an accelerating trend toward global market convergence exists in addition with within-country fragmentation of consumer needs, such a strategy may no longer be valid today (Ter Hofstede et al., 2002)

2.2.2. International segmentation

With an international segmentation strategy, the scope of segmentation is wider than with the country-as-segment strategy, since the segmentation is not only based on a single country, but on a few national markets. Countries which are similar on certain characteristics are combined as a whole, in order to be able to treat them with the same mix of marketing instruments. As a result one can see international segmentation as country-as-segment segmentation applied over combinations of similar countries.

2.2.3. Global segmentation

A global strategy on the other hand is a segmentation strategy integrated across national borders and therefore meets the requirements of the latest developments as regional unifications, shifts to open economies, rapid increases in education and literacy levels, as well as global investments and advances in information- and communication technology. All these developments have made national borders less and less important and multi-domestic strategies no longer appropriate. Strategies must now be integrated and globalized. Use of only country as segmentation basis assumes after all that a country is an isolated or independent unit. In today’s world this is obviously not the case; people travel and communicate and national borders form by no means a restriction anymore.

Advantages of global strategies are economies of scale, caused by the use of a similar marketing strategy in multiple countries, which leads to a reduction in the average production, advertising and distribution costs. Improved quality of products and increased bargaining and competitive power are also named (Steenkamp & Ter Hofstede, 2002). Besides, because groups of consumers in different countries often have more in common with one another than with other consumers in the same country, organisations can best serve segments that transcend national borders (Ter Hofstede, Steenkamp & Wedel, 2002). A major challenge for organizations and

products and marketing programs that meet the common needs of these consumers in global market segments (Ter Hofstede et al., 2002).

In this way international segmentation combines the benefits of standardization, e.g. lower costs, with the benefits of adaptation; close to the needs of consumers for example (Steenkamp & Hofstede, 2002).

In addition, research has shown that products especially designed for the global market achieve market shares almost twice the size of products with domestic design that are aimed at the same overseas markets (Ter Hofstede et al., 2002; Kleinschmidt & Cooper, 1988). Understanding consumers needs and their behaviour across the globe, in order to perform global segmentation, is therefore of vital importance in today’s economy.

Concluding can be said that the differences between country-as-segments, international and global market segmentation depends on the scope of segmentation held by the firm. A country-as-segments strategy concerns several separated countries, where an international firm posses a scope of segmentation which is based on the combination of few national markets. And a global firm at last views the globe and their market more broadly, goes across borders to combine pieces of several countries, and develops segmentation strategies independent of the countries’ physical positions in the world (Hassan et al., 2003).

Today Unilever can be considered as a company with an international segmentation strategy. However, it strives for the implementation of a global market segmentation strategy. In the following chapters will be examined whether the Global Value List as data gatherings method, in combination with the multi-level latent class analysis as analysing method, provide a useful method to conduct global market segmentation research as foundation for such a strategy.

2.3. Conceptual issues

Below, the conceptual issues of international market segmentation will be discussed. This means the level of aggregation of market segmentation, the market segmentation bases and their evaluation criteria and at last the importance of construct equivalence.

2.3.1. Level of aggregation

The level of aggregation of market segmentation can range from no aggregation to country-level aggregation (Steenkamp & Ter Hofstede, 2002). With no aggregation, the data from individual consumers from different counties constitute the basic information for segmentation, which results in segments of consumers from different countries that are similar on the segmentation basis used. This are cross-national consumer segments and is the kind is the kind of aggregation level Unilever had in mind when they asked the thousands individual respondents from different countries to answer the questions of the Global Value List. Moreover, this is the level of aggregation that is necessary for global segmentation strategy.

Country-level aggregation on the other hand, pools the information across consumers within each country and forms groupings of countries based on their similarities on the segmentation basis, as is be done with international segmentation strategies. Country-level aggregation results in cross-country segments. Advantage of this kind of

Market Segmentation Strategy Level of Aggregation Segments

International Country level Cross-country

segments Global

Individual level Cross-national consumer segments

Table 2.3.1. Market segmentation strategies and level of aggregation with accompanying segments

2.3.2. Market Segmentation Bases

Market segmentation bases are the criteria employed for grouping countries or consumers.

It is important that these bases are not too general, in such a way that they do not discriminate enough among people. Dhalla & Mahatoo (1976) for example found that cognitive needs and styles often are far too broad to discriminate among users of different brands within a product category. Since social values are cognitive concepts as well, this is an important finding.

In addition, the bases may neither be too specific since very specific segmentation bases ignore the different nuances of consumer behaviour and are for that reason not very helpful in developing marketing strategies (Dhalla & Mahatoo, 1976).

Moreover, segmentation bases can be devided in general and domain-specific segmentation bases (Steenkamp & Ter Hofstede, 2002). General segmentation bases on the one hand are independent of the domain in question, and can further be divided into observable bases (like geographic regions, economic characteristics and demographics), and unobservable bases (like values and life-styles).

Domain-specific bases on the other hand depend on the particular domain or product. Examples of these are brand penetration rates or attitudes or.

It is important to emphasize here that, although many organisations use segmentation strategies with segmentation bases based on characteristics of countries in stead of consumers, it are the consumers which are of importance, not simply the concept of countries. Countries are often grouped (international segmentation with cross-country segments) founded on differences in macro-level geographic, political, economic or cultural data, whereas only a few organisations use responses from individual consumers, or so called cross-national consumer segmentation (Steenkamp & Ter Hofstede, 2002). It is unrealistic however, to assume that each country is totally homogeneous. In addition, although information on countries often can be acquired easily through secondary data sources, the results of such a country segmentation do not always provide relevant information for managerial decisions, since it only provides insights into which groups of countries can be targeted and nothing about how the consumers within these countries will respond to marketing strategies or new product developments. The so important within-country heterogeneity is often ignored.

Reason for this is the lack of available cross-national consumer data and the costs to collect it yourself, which limits the use of cross-national consumer segments. Whereas the increased globalization has made the homogeneity of behaviour of consumers within countries smaller and increased the similarities among consumers across countries (Ter Hofstede et al., 2002; Hassan et al., 2003; Lewitt, 1983).

One way to apply this is by using multi-level latent class analysis as is done in chapter 6.

2.3.2.1. Evaluation criteria

The strategic attractiveness of the segments resulting from the use of a segmentation basis, is determined by the stability, reachability, measurability, sustainability, receptivity and profitability of these segments.. As a result, these criteria can be used to evaluate segmentation bases (Wedel & Kamakura, 1998:207). See table 2.3.2. for a short explanation of the exact meaning of these criteria (Steenkamp & Ter Hofstede, 2002).

Evaluation criteria

Definition

Identifiability Extent to which distinct segments can be identified Substantiality Sufficiently large segment size

Accessibility Degree to which segments can be researched with promotional and distributional efforts

Stability Temporal dynamics of segments

Actionability Extent to which the segments provide basis for formulation of effective marketing strategies

Responsiveness Whether segments respond uniquely to marketing efforts targeted at them. Responsive international segments make it possible to target consumers from different countries with standardized marketing mixes.

Table 2.3.2. Evaluation criteria segmentation bases

2.3.2.2. Values as segmentation bases

Much research has supported the use of human values as segmentation basis (Schopphoven, 1991), since they indicate underlying consumer motivations and, at the same time, define market segments desiring similar product benefits (Wedel & Kamakura, 1998:56).

Besides, a global market segmentation strategy based on social values can be the basis of much standardization. Hassan et al. (2003) gave the following example about the universal desire for beauty. He stated that women in Tokyo, New York, Istanbul and Berlin are not only ‘sisters under the skin’, but also on their skin, lips, fingernails, and even in their hairstyles. Consequently, they are likely to value similar cosmetics with similar appeals. This makes that using human values by rolling out similar products across the world can lead to enormous economies of scale and other advantages for worldwide operating organisations like Unilever.

Evaluated with the above mentioned criteria, social values used as market segmentation bases are considered to deliver segments which have good identifiability, good substantiality, poor accessibility, very good stability, moderate responsiveness (Steenkamp & Ter Hofstede, 2002) and moderate actionability. The good identifiability and substantiality are due to the fact that social values can be measured well and enough people exist who share the share the same values.

The highly positive score on stability is caused by the fact value priorities are central to people’s self-concept. As a result people’s values change only little and segments based on these values as well.

However, the relationship between socio-demographic variables and human values is considered weak (Wedel & Kamakura, 1998:202) which makes people within certain value segments difficult to identify and to reach. As a result value segments have only low accessibility.

2.3.4. Construct equivalence

Besides the six criteria mentioned above, any global segmentation basis must exhibit construct equivalence. Construct equivalence refers to whether the basis used for segmentation is equivalent across countries (Steenkamp & Ter Hofstede, 2002), which means that no indistinctness about the definition of the basis may exist.

Construct equivalence is that important because if cross-national differences exist with respect to the meaning of the segmentation basis, or if in the most extreme case, the segmentation basis has no meaning at all in some of the countries, segments are likely to be based on these differences instead of similarities in the segmentation basis (Steenkamp & Ter Hofstede, 2002).

However, since every culture has its own language related cognitive categories and dimensions, segmentation bases will not automatically be interpreted the same across countries. To assure construct equivalence often some small emic; country-specific adjustments, by local experts are desired. The concepts used as segmentation bases are often not universal after all. Straightforward is however, that the more similar the socio-cultural environment is among countries, the more equivalent the segmentation bases will be.

Scholars who consider constructs and theories to be emic, argue that constructs and theories are specific to a clearly defined context and as a result must be developed in these contexts as well. Segmentation bases used in different settings therefore must be adjusted to that specific context.

However, the more cross-culturally applicable a research instrument must be, the more aspects idiosyncratic to the culture will be left out. As a result, an important trade-off exists between cross-cultural applicability on the one side and depth and comprehension of the area of interest on the other side (Brunsø & Grunert, 1998). The school which lies at the other extreme of this trade-off and which leaves out all the idiosyncratic aspects of culture is the etic school. They see constructs and theories as universal and applicable in all contexts (Douglas & Craig, 2006). According to them segmentation bases can be universally applied without any adjustments at all.

While developing the Global Value List, Unilever followed a combination of the emic and etic approach. Although in most countries the questions were assured to have translation equivalence by back-translation (which can be considered as an emic element), everywhere the same value items of the Global Value List were used to measure the values (an etic approach).

However, although social values can differ across countries in the way they must be measured (view section 2.4.1. about measure equivalence), chapter 3 makes clear that social values themselves exist in every culture. Therefore, even though a partly etic approach is used, no reason exists to cast serious doubt on the construct equivalence within the data gatherings process of Unilever.

2.4. Methodological issues

After the conceptual requirements of global market segmentation research, now the methodological issues will be elaborated, which further research must satisfy in order to ensure cross-cultural validity.

2.4.1 Measure equivalence

Calibration equivalence is equivalence in monetary units and in measures of weight, distance and values. Examples of variables with which it is often difficult to reach calibration equivalence, are GNP/capita, income or education. Since education systems for instance differ substantially across the world, people of different countries will interpret the content of relevant terms as ‘high school’ differently, which would make education unsuitable to use as segmentation base, unless extremely broad categories as higher versus lower education are used. However, such broad categories lead to a great loss of information, which negatively affects the stability of the segments (Steenkamp & Ter Hofstede, 2002). A possible solution in this case would be to take the years of education of a person after a certain age as segmentation base, which can be measured relative objectively.

What the Unilever study concerns; in those countries were respondents were asked about their income, the multiple choice answer possibilities were adapted to the local currency and income level. Income could then relatively be compared between countries. An absolute comparison would lead to biases.

Translation equivalence is a form of measure equivalence which indicates whether the measurement instrument is interpreted similarly by respondents in different countries. Respondents differ for example in terms of language, education, level of literacy and rules of social interaction, which all have influence on respondent comprehension, suitability of different response formats and response styles (Douglas and Craig, 2006). To assure this equivalence, careful back-translation and extensive pretesting of translations in different countries is required. Back-translation consists of a procedure in which a questionnaire is translated by a bilingual translator, who is a native speaker of the target language. After that, a translator who is a native speaker of the initial language translates the questionnaire back. After that the two questionnaires in the source language are compared and errors corrected (Douglas and Craig, 2006). The importance of such a translation is often underestimated, since it turned out that international and global segmentation studies are not always careful in their linguistic translation and only few studies provided any evidence on international pretesting at all (Steenkamp & Ter Hofstede, 2002).

In addition, achieving translation equivalence remains a very complicated process, especially since equivalent words may not be available in other countries. Steenkamp & Ter Hofstede (2002) gave an example of the absence of an equivalent for the Dutch word ‘gezelligheid’; defined as sitting together with a number of people and feeling good because of a purported unity in feelings and emotions…

However, conducting research across countries using one single language is not the solution,

This would help to overcome potential language-based response biases, but ignores differences in language fluency and interpretation of English, which can create even more serious biases (Dawar & Parker, 1994).

In the data gathering process of Unilever translation equivalence could have been better as well. Although most of them were, not all the questionnaires were translated in the native language of the respondents using back-translation, but were conducted in English. This could have led to biases in the responses.

The third form of measure equivalence is score equivalence. Score equivalence refers to the equivalence of the observed scores on the measures. Observed scores must be comparable across countries, otherwise international segmentation solutions will be biased. This holds not only for primary, but for secondary data as well. Score equivalence is a more often occurring problem than may be thought on first sight. For example, even the wealth of countries, measured by GNP/capita, is difficult to compare due to different tax morals.

supposed to measure. Some best-known response styles are extreme responding and answering with social desirable answers (Paulhus & Reid, 1991).

As can be read in section 4.3 also in the data used within this study some response styles did exist.

Finally, measure equivalence in general is hard to obtain for psychological measurements, like social values, across cultures. Unilever only assumed measure equivalence of the Global Value List while collecting the data and did not actually test it (Klerk et al., 2005). As a result, the existence of bias in the data due to a lack of measure equivalence is a possibility that can not be ruled out.

2.4.2. Sampling equivalence

Another form of equivalence is sampling equivalence. This kind of equivalence does not mean the sample must consist of people with the same socio-demographic characteristics across different countries. Nor does it mean that the method used to sample must be the same. It does mean however, that the information collected from the sample must be equivalent and that the sample must be representative of the relevant target population. (Steenkamp & Ter Hofstede, 2002).

Especially in developing countries where the use of questionnaires, which requires a particular amount of education, excludes a part of the population, sampling equivalence is often hard to reach (Steenkamp & Ter Hofstede, 2002).

Moreover, here belongs also the possibility to make sample sizes per country proportional to the population sizes, as is recommended by Steenkamp & Ter Hofstede (2002) in case of global research. If sample sizes are not proportional, the pooled sample is not representative of the pooled population, which could bias the results since the weight of a subject in the sample is the same, although the chances of a subject being included in the sample is not equal across countries (Steenkamp & Ter Hofstede, 2002).

In order to neutralize this effect, sample sizes can be weighed by population size or samples can be drawn for each country proportional to their population. However, with this method a risk exists that the sample sizes of smaller countries would become too small for reliable estimation. This exactly is the reason why this option is given up within the Unilever study (view section 4.5 as well).

2.4.3. Stimulus commonality

Use of a common research instrument requires equal familiarity with visual and verbal stimuli, like Likert Scales or paired comparisons and photographs, as well. Moreover, colours and combinations of colours have different associations in different countries (Douglas & Graig, 2006).

In addition, most developing countries are characterized by lower levels of formal education, while many measurement instruments require a fairly high degree of respondent sophistication (Steenkamp, 2005). In order to conduct valid market segmentation research in developing countries, measurement instruments has to be as simple and short as possible. Since the Global Value List contains no less than 63 value items, it does not satisfy this criterion very well (see section 8.3. recommendations as well).

2.4.4. Management centrality

However, next to this decentrality, one must be cautious that some degree of centrality is maintained while performing the research.

Within the Unilever research there was some lack of centrality, which resulted in inconsistently collected socio-demographic data.

2.5. Conclusion

Concluding can be said that Unilever and worldwide operating companies in general would benefit substantially from global market segmentation. However, before conducting global market research one has to be sure the research design satisfies the above mentioned conceptual and methodological issues, in order to assure cross-cultural validity. In the case of Unilever, this is something which can still be improved. The economies of scale which then can be obtained by rolling out similar products across the world can build an important competitive advantage, which is of great value in today’s world of increasing competition. After all, the globalisation forces continuously more companies to target their products at markets across national boundaries.

Nevertheless, this can only be done successfully if organisations thoroughly understand and adequately respond to the core values and needs of their consumers (Ter Hofstede et al., 2002; Hassan & Kynak, 1994). A specific global segmentation method would allow international companies to do this, and at the same time at a much larger scale than just cross-country. In chapter 6 will be examined whether a multilevel latent class analysis can provide such a global segmentation method, by segmenting simultaneously on countries and consumers.

3. Social Values

3.1. Introduction

In order to be able to determine whether segmentation performed using the Global Value List is suitable to assess the viability of rolling out new products across new countries, insight has to be gained in the existing relationship between social values and behaviour. This chapter serves mainly to explain the nature of this relationship. After all, if there is no relationship between these two concepts, consumer behaviour can not be predicted from social values and any usefulness of the Global Value List is per definition out of question.

Moreover, assuming that a relationship between both concepts exists, the accurateness of behaviour prediction based on values highly depends on the nature of that relationship; whether this is a direct one without noise, or an indirect one, with disturbing elements in-between.

3.2. Definition of social values

Rokeach (Ter Hofstede et al., 2002) defined social values as an enduring prescriptive or proscriptive belief that a specific mode of behaviour or end-state of existence is preferred to an opposite mode of behaviour or end-state. He views values as a standard that guides and determines action, attitudes toward objects and situations, evaluations, judgements, comparisons of self with others and attempts to influence others.

This definition consists of two parts, which corresponds to the distinction made by the same author, between two types of values:

- Instrumental values; enduring beliefs that a specific mode of conduct is personally or socially preferable

- Terminal values; enduring beliefs that a specific end-stage of existence is personally or socially preferable

These terminal values are desired modes of existence and represent enduring sets of values. As they are the preferred end-states of existence, people strive for them and they give meaning to their life. Instrumental values at the other hand are desired modes of behaviour or ‘instruments’ used in order to achieve the terminal values. Some examples of terminal values are equality, family security and salvation. Helpful in the achievement of terminal values and therefore called instrumental values, are ambition, independence and honesty (Wedel & Kamakura, 1998:16).

Terminal values then are motivational constructs, since living up to a value by instrumental values and concrete behaviour, fulfils a particular, highly abstract goal. The prospect of the achievement of such a goal or terminal value motivates to perform certain behaviour or instrumental value.

Obviously, there are only a limited number of fundamental values. These values are held by all human beings across the world; they all posses the same values. The difference is only the different degrees to which people possess these values. People namely do differ with regard to their values, but only in the relative importance of specific values (Verplanken en Holland, 2002). One value may be very important to one person, but at the same time quite unimportant to someone else.

3.3. Difference between attitudes and values

The difference between attitudes and values lies in the fact that an attitude refers to the organization of several beliefs around a specific object or situation. A value on the other hand, refers to a single belief of a very specific kind, for example a proscriptive belief (Ter Hofstede et al., 2002). Moreover, values transcends specific objects or products, something which attitudes do not (Ter Hofstede et al., 2002).

Attitudes are, in contrast to values, very concrete with respect to certain desirable behaviour. Values are considered to be universal (Grunert & Juhl, 1995; Schwartz and Blisky, 1987; 1990; 1992), are more abstract and are often seen as the central beliefs of an individual. As a consequence individuals may have thousands of attitudes, but - as said previously- only a few dozen values.

This makes values also more stable than attitudes. Attitudes are dynamic; subjects asked about their attitudes for example, often change their minds over time. People’s most central values at the other hand often remain a whole life as well as unchanged (Bardi & Sachwartz, 2003). As a consequence values may be more useful than attidudes in prediciting human behaviour.

However, instrumental values are indeed more stable than attitudes, but less stable than terminal values. With this, instrumental values are comparable with attitudes, since they are reflected in human behaviour by influencing decisions, but are not that stable.

Terminal values at last, strongly affected by cultural factors, are more stable because they are acquired early in life. As a result instrumental values are more susceptible to changes in socialization processes (Wedel & Kamakura, 1998:145).

Finally, also a more technical distinction can be made between values and attitudes, which probably contributes to a clear understanding of the differences. It concerns the way both concepts are measured; attitudes varies on an evaluative (favourable- unfavourable) dimension, while variation in values lies, as mentioned earlier, in its importance (Verplanken & Holland, 2002).

3.4. Value-behaviour relationship

No agreement exists about the exact nature of the relationship between values and consumer behaviour, although one does agree that some kind of relationship exists and that values guide behaviour (e.g. Pitts & Woodside, 1984; Kim et al., 2002). Main point of discussion however is whether this relationship is of a direct or an indirect nature.

3.4.1. Indirect relationship

Some authors view the relationship between values and consumer behaviour as indirect, which means that no direct relation between the two concepts exists, but that they are mediated by one or more other concepts. Influence of values on behaviour is present in such a case, but goes through the mediating concept(s). Which concepts precisely act as the mediating concept is still unclear.

First of all, Homer & Kahle (1988; Shim & Eastlick, 1998) state that the influence of abstract values flows via less abstract attitudes to specific behaviour. Two other studies (Shim & Eastlick 1998; Pitts & Woodside, 1983) have confirmed that the relationship between values and behaviour is mediated by attitudes and that only a weak direct relationship exists.

Besides, the Means-end Theory of Gutman (1982) links perceived product attributes to values. ‘Means’ here are products or activities in which people engage and ‘ends’ are desired states of being, for example: happiness or accomplishments.

psychological) accruing directly or indirectly to the consumer, sooner or later, from his behaviour (Gutman, 1982). These consequences contribute to the achievement and reinforcement of a person’s values.

Whether a product produces specific consequences depends on the attributes the product possesses. It are these attributes, defined as the physical and psychological constructs by which respondents describe and/or differentiate between products (Manyiwa & Crawford, 2001), which provide the consequences.

As a result the attributes are not valued for their own sake, but because they make the achievement of desired end states i.e. values possible. Consumers therefore desire that attributes, which provide favourable consequences or which can avoid unfavourable consequences, that are consistent with their values (Gutman, 1982; Manyiwa and Crawford, 2001). Consequently, people buy products with attributes leading to consequences which are consistent with and which reinforce their values.

Obviously, like with attitudes and values, there are only a couple of ends (values) and far more means (consequences).

Besides, these consequences can occur directly or indirectly. Direct consequences occur if they come direct from the product consumed. Indirect consequences at the other hand come from other people reacting favourably or unfavourably to us and our purchase- and consumption behaviour

(Gutman, 1982).

Finally, since a person’s values are ordered in importance, a person’s consequences are ordered in importance as well. Because one can not and does not want to live completely according to all existing values, one strives for the satisfaction of the most important consequences by the purchase and consumption of the matching products (according to its attributes) in order to fulfil their most important values (Sim & Eastlick, 1998; Kahle & Kennedy, 1989; Baker et al., 2002).

With these most important values, that values are meant, which are most central to one’s self-concept. In other words: the values which a person wants to identify himself with.

A slightly similar view on this value behaviour relationship is the view that this relationship is mediated by the needs of consumers (Kim et al., 2002). In this way values drive consumers to buy the products that fulfil their specific needs. Instead of valued end states as Gutman (1982) does, Kim et al. (2002) speak about needs which have to be satisfied. They distinguish three basic types of consumers’ needs: functional, social and experiential needs.

A person’s functional needs are low-level motivators which encourage consumers to focus on intrinsic advantages of the product. Functional needs can therefore be satisfied by the functional characteristics of a brand (Kim et al., 2002).

Social needs, as social approval for example, can be satisfied by the degree of social visibility of the product or the prestige it provides (Kim et al., 2002; Keller, 1993). Experiential needs at last, are consumers’ need for novelty, variety and pleasure and can especially be satisfied by new products (Kim et al., 2002).

People choose to buy that specific product which possesses these specific characteristics that satisfy their specific functional, social and/or experiential needs. Purchase behaviour can therefore be explained as a result of the different kind of needs people have.

Allen (2001) at last found both an indirect and a direct relationship between values on purchase behaviour. An indirect influence of values he explained appears when consumers evaluate a product’s utilitarian meaning by examining the product

attribute-by-attribute. Values hereby influence the importance given to a product’s

attributes, which in turn influence product preference and with that consumer behaviour. The relationship between values and behaviour is thus mediated by the importance of the attributes, like in Gutman’s Means-end Theory.

The direct influence of values on the other hand appears when consumers evaluate a product’s symbolic meaning in an intuitive way. No evaluation of the product’s attributes takes place in that case and purchase behaviour is directly influenced by values.

3.4.2. Direct relationship

Next to the direct influence of values named by Allen above, and the weak, but direct relationship identified by Homer & Kahle (1988), Shim & Eastlick (1998) and Pitts & Woodside (1983) within the previous section, Goldsmith et al. (1995) gave a hypothesis which is not been tested yet, but is similar to the theory of Allen. It states that with low-involvement products, consumers purchase without forming well defined attitudes in advance of the purchase. Instead, their product choice is solely based on their values, assuming a direct value-behaviour relationship.

More evidence of the existence of a direct value-behaviour relationship can not be found in the literature.

3.4.3. Restrictions

Finally, McGregor (2000) pointed out that although values seem to influence consumer behaviour, consumer’s actual behaviour is often in direct conflict with their values. He gives the example of Canadian consumers who are supposed to value the environment, yet are one of the largest polluters of the world…

This assumption is supported by Munson (1984; Brunsø et al., 2004) who found only weak general relations between values and behaviour.

Verplanken and Holland (2002) pointed at the degree of centrality of values in their relationship with behaviour. With centrality they mean the degree of importance of a value to an individual. Their research showed that values influence behaviour only if they are sufficiently central to a person’s self-concept.

Rokeach (1973; Gutman, 1982) agrees that not all behaviour is guided by values. According to him, values only influence behaviour that is related to maintaining and enhancing your self-esteem.

3.5. Values and food

Even less than about the relationship in genereal, is known about the specific relationship between values and food purchase. However, research of Homer and Kahle (1998) proved the causal relationship underlying the value-attitude-behaviour hierarchy in the specific situation of natural food purchase. They showed that people with stronger terminal and instrumental values had more favourable attitudes toward natural food purchase than those with weaker forms of values. In this way they found that social values, via attitudes, are systematically associated with the use of natural foods in general.

people buy food solely based on their values without forming well defined attitudes in advance, indicating a direct value-behaviour relationship.

Grunert & Juhl (1995) did find a relationship between values and the purchase of organic food. Whether this relationship was one of a direct or indirect nature, was not specified.

Brunsø et al. (2004) developed a cognitive structure model for the food-related lifestyle, based on assumptions about the relationship of food products and values in the consumers’ cognitive structure. He found that concepts like the way of shopping, cooking methods, quality, consumption situations and purchasing motives (determining ones lifestyle) intermediate this relationship. So according to Bruno et al. ones values determine ones food-related lifestyle, which at its turn determined ones consumer behaviour.

Finally, Barnhill found a positive association between values for three generations of Japanese-Americans and food preferences, indicating cultural influence involved in the value-behaviour relationship. (Wedel & Kamakura, 1998: 220). This cultural influence will be further elaborated on in the following section.

3.6. Cultural influence on values

Since a culture can also be described as the value system of a society and values are considered to be the consequence of culture, ethnicity, personality, and society with its institutions (Shim & Eastlick, 1998; Rokeach 1973; Baker et al., 2002), values can be assumed to be shared to some extent within a culture.

Culture has an important influence on consumer behaviour (Gutman, 1982; Belk, 1975). Culture creates standardised behaviour, since specific behaviour is seen as normal and appropriate within a specific culture.

Kim et al. (2002) have shown in their research that at the one hand the products and brands people buy and at the other hand the benefits that they desire from their purchases, are all culturally based. This leads to the conclusion that consumer needs are also affected by culture and with that indirectly the value-behaviour relationship as well. Markets with low individualism for example would value products to fulfil social needs to reinforce group membership. On the other hand, consumers in markets with high individualism would value products that appeal to their experiential needs more (Kim et al, 2002). As a consequence consumer behaviour would be different, due to differences in cultural influenced needs.

Applied to the specific domain of food, one can see food habits as a component of culture and with that see culture indirectly influences food purchase decisions that consumers make. Mead (White & Kokotsaki, 2004) described this relationship between culture and food habits. Food habits are seen as the culturally standardised set of behaviours in regard to food, manifested by individuals who have been reared within a given cultural tradition (White and Kokotsaki, 2004). As a consequence, according to these authors not only the general value-behaviour, but also the specific value-food purchase relationship is influenced by culture.

Finally, culture determines which consequences are desired in that specific environment and thus which values are held as important. So, culture determines the relative importance of values.

family life. Europe on the other hand is more individualistic and personal achievement is an important value.

According to Grunert & Juhl (1995) these cross-cultural differences in values can then be used in predicting concrete behaviours. Cultural differences in behaviours, he said, can the easiest be explained by the most abstract level of consumers’ cognitive hierarchy. With other words: by their values. However, in this research the main aim is not to use values to differentiate between cultures but to find similarities between respondents across countries and cultures based on these values. Following the same logic, that too is possible. After all, cultures do not differ that much in that they hold completely different sets of values; they only find certain values more important than does another group. But basically they all possess the same set of values, which makes individual deviations likely because next to the cultural overall value system of a country, all individuals possesses a personal value system existing of their own individual hierarchical value system. Therefore it should be possible by basing the segmentation on values, to identify similar groups of consumers across different countries and cultures. It is likely after all, that the relative importance assigned to values by individuals, is not restricted by country borders. Individuals across countries and cultures can posses the same composition of values.

In this way values used as a segmentation basis make global segmentation possible. Based on, in this way identified segments, products and marketing strategies that address similar values in different countries can be developed.

3.7. Conceptual value framework

The above mentioned relationships proposed by the different authors, can be combined and displayed in a clearly set out model as is done in figure 3.7.

Figure3.7. Value-behaviour relationship framework

In this way it is clearly visible that social values are influenced by culture, personality, ethnicity and the society with its institutions. Values themselves have a direct relationship with consumer behaviour and indirect relationship via attitudes, product