Market Segmentation

Market segmentation is a crucial marketing strategy. Its aim is to identify and delineate market segments or “sets of buyers” which would then become targets for the company’s marketing plans. The advantage to marketing management is that Ais technique divides total demand into relatively homogeneous segments which are identified by some common characteristics. These characteristics are relevant in explaining and in predicting the response of consumers, in a given segment, to marketing stimuli. The market can be subdivided by geographic, demographic, psychological, psychographic or behavioural variables. The advantages and disadvarUages of each of these types of segmentation variables are discussed in detail in this paper. Kotler {1984) has identified four requirements that a marketer can use in evaluating the desirability of potential market segments, namely measureability, accessibility, substantiality and actionability. Once a segment has been identified which meets these requirements, it is possible to develop a product or service which meets the unfulfilled needs of this segment. A marketing mix can then be devised to reach the segment identified economically and efficiently. A strategy of market segmentation attempts to regain some of the benefits of the closer association with customers which was the strength of traditional business operations.

INTRODUCTION
This paper presents a review of the literature concerning the concept and practice of market segmentation. This key strategy is essential to the development of a strategic plan for a brand. It is a decision-making tool
for the marketing manager in the crucial tasks of selecting a target market for a given product and designing an appropriate marketing mix.

The uses of this technique are discussed, together with the procedures for segmenting markets. Possible bases for segmenting consumer markets are reviewed in detail. The more straightforward objective bases have been briefly outlined whereas the more complex subjective behavioural bases are discussed in more depth. The requirements for segmentation to be effective are noted and some criticisms of the technique presented.

Market segmentation has long been “considered one of the most fundamental concepts of modern marketing” {Wind 1978, p. 317). Sheth (1967, p. 728) has described it as “essential to marketing”. According to the definition found in the Oxford English Dictionary “to segment” is to “divide into parts”. In marketing terms these parts can either refer to groups of consumers with similar requirements or to groups of goods or services with similar attributes. The term and concept of “market segmentation” have been attributed to Wendell R. Smith, in a paper first published in 1956. He (Smith 1956, p. 6) commented “Segmentation is based upon developments on the demand side of the market and represents a rational and more precise adjustment of product and marketing effort to consumer or user requirements. In the language of the economist, segmentation is disaggregative in its effects and tends to bring about recognition of several demand schedules where only one was recognized before”. In a brief “Retrospective Note on Market Segmentation” published in the introduction to the special edition of the Journal of Marketing Research edited by Wind, Smith (1978, p. 316) asserts that “the roots of early market segmentation research, carried on almost a quarter of a century ago, can be found in the writings of a group of marketing practitioners and scholars whose undisputed leader was the late
Wroe Alderson”.

Baker (1984, p. 123) considers the “concept of market segmentation rests upon recognition of a differentiated demand for a product, while its use as a marketing tool depends uf)on identification of the most appropriate variable or variables with which to subdivide total demand into economically viable segments. Economically viable segment may be understood as being of sufficient size to enable a marketer to earn an adequate profit by catering to the specific needs of its members. In fact Haley (1968, p. 30) considers that “the idea that all markets can profitably be segmented has now received almost as widespread acceptance as the marketing concept itself”. Howard and Sheth (1969, p. 70) have noted market segmentation depends on the idea that “the company should segment or divide the market in such a way as to achieve sets of buyers”. These sets of buyers, or subsegments of the market, would then become targets for the company’s marketing plans. The potential methods of subdividing total markets must be validated by research. It is then the responsibility of management to devise marketing mixes which are effective in the market segments. Thus market seg[mentation has as its aim the identification and delineation of market segments with a view to providing more efficient and satisfactory marketing service. “The strategy of market segmentation recognises that people differ in their tastes, needs, attitudes, motivations, life-styles, family size and composition etc.” (Chisnall 1985, p. 264).

The concept of market segmentation hsis only been recognised comparatively recently. Historically sellers engaged in mass marketing. That is they mzss produced, mass distributed, and mass promote! one product to all consumers in an attempt to obtain economies of scale. In the face of the competition inevitably generated by this approach producers sought to

MARKET SEGMENTATION

303

obtain a differential advantage through making their products or services
different from those of competitors. This product differentiation strategy is “designed to offer variety to buyers rather than to appeal to different segments” (Kotler 1984, p. 251). According to Staudt et al. (1976 p. 6) product differentiation has followed the approaches shown below: (a)

(b)
(c)
(d)
(e)
(f)

physical differentiation of product
psychological differentiation of product
differences in purchasing environment
difference by virtue of physical distribution capability
differences in after-purchase assurance of satisfaction in use differences in prices and terms of sale.

Product differentiation proved moderately successful but because it does not centre upon the needs and requirements of the consumer it has failed to yield maximum benefits to the producer and consumer (Ogwo 1980, p. 24). Companies are increasingly embracing market segmentation strategies as a result of the dissatisfaction they have experienced with product differentiation.

At a superficial level the theory of market segmentation appears to conflict with basic economic theory. The tailoring of a product to meet the needs and wants of a market segment militates against long production runs and the resulting economies of scale. The development of the segmentation approach to market planning was associated with the end of rationing after the war, the acceleration of technological progress, increased social mobility and growth in the variety of wants felt by consumers and the revival of competition as a market force (Crimp 1985). Manufacturers were thus able to identify marketing opportunities, design and launch a product to fulfil the requirements of that segment and concentrate marketing effort on that
segment. This yielded a two-fold benefit. The manufacturer could adjust prices, distribution channels, promotions and advertising to reach the target market efficiently. “Instead of scattering their marketing effort (‘Shotgun’ approach), they can focus it on the buyers who have the greatest purchase interest (‘rifle’ approach)” (Kotler 1984, p. 251). In addition the manufacturer could develop sufficient loyalty for his product to withstand the price appeal of retailers own label products.

USES OF THE MARKET SEGMENTATION APPROACH
The importance of the market segmentation approach has already been stressed in the introduction to this paf)er. It can help to set the basic objectives for the whole marketing operation, and to indicate appropriate strategies by which these objectives can be realised. The type of objective and strategy will affect the type of segmentation problem posed, which in turn will determine the kind of research necessary. Lunn (1978 p. 346) has

304
A. CAROLINE TYNAN AND JENNIFER DRAYTON
identifled four characteristic types of marketing problem which would normally be solved through a market segmentation study.
(a) Defining the market
It is important to view a market from a consumer’s viewpoint rather than from that of the manufacturer. Products which are viewed as substitutes by the customer may come from several distinct product fields when the manufacturers perspective is considered. On one hand the consumers product field concept may range quite widely, and on the other she may not necessarily consider all the brands from any one field as being suitable for a particular need or set of needs.

(b) To rationalise policies for existing brands and products The company is constantly seeking to devise optimum strategies for its products. The objective may be to improve market share, weaken the pMDsition of a key competitor, or protect a brand from competitive activity. In the light of the market segmentation research, attempts may be made to increase the purchase rate of current buyers, to convert buyers from competing brands or
to attract a new group of customers to the product field.

(c) To position ranges of brands and of product varieties
In a market with several different segments of consumers who have different needs a company is well advised to cater for several of the more important segments if it has sufficient resources. At the same time compietition between the company’s brands in any segment should be minimised.

(d) To idmtify gaps in tbe market wbicb o£Fer new product
opportunities
The aim here is to identify customer segments whose needs are not being met by any existing brand. These needs may be met by launching a new product or by altering an existing product.
PROCEDURES FOR SEGMENTING MARKETS
Yoram Wind (1978) has identified four basic methods for segmenting markets, the traditional a priori and cluster based designs and the newer flexible and componential procedures. The classification of segmentation studies into the first two of these categories, that is a priori and cluster based or post hoc, was suggested by Green (1977). He maintains that a priori segmentation models have had as the dejjendent variable (the basis for segmentation) either product specific variables like product usage or loyalty, or general customer characteristics in demographic terms. Survey results show the segments’ estimated size and their demographic, socio-economic,

MARKET SEGMENTATION

305

psychc^raphic and other relevant characteristics. A typical research design for an a priori segmentation model involves the following seven stages: (a) Selection of the a priori basis for segmentation.

(b) Selection of a set of segment descriptors, including hypwtheses on the jMJSsible link between those descriptors and the basis for segmentation. (c)
Sample design—mostly stratified and occasionally a quota sample according to the various classes of the dejjendent variable. (d) Data collection.

(e) Formation of the segments based on a sorting of respondents into categories.
(f) Establishment of the (conditional) profile of the segments using multiple discriminator analysis, multiple regression analysis or some other appropriate analytical procedure.
(g) Translation of the findings about the segments estimated size and profile into specific marketing strategies, including the selection of target segments and the design or modification of sf>ecific marketing strategy.

Cluster based or post hoc segementation models differ from a priori models in that the basis for segmentation is selected after the data has been collected. Most commonly the variable used in this type of mode! are needs, attitudes, lifestyle and other psychographic characteristics, or benefits sought from the product or service. Frequendy the clustering procedure is preceeded by a factor analysis to reduce the original set of variables. The variables are grouped according to their correlation with each other and the amount of variance they can explain in the dependent variable. As in a priori designs the size and demographics, socio-economic, purchase and other relevant characteristics are estimated.

Flexible segmentation offers a dynamic approach to the segmentation problem. It allows management to develop and examine a large number of alternative segments, each composed of a group of consumers who exhibit a similar response to new “test” products. The approach is based on the integration of the results of a conjoint analysis and a computer simulation of consumer choice behaviour. Conjoint analysis studies usually consist of three parts:

(a) Preference ranking or rating of a set of hypothetical products. (b) Perceptual ranking or rating of current brands on the same set of attributes used in ranking the hypothetical products above.

(c) A set of demographic and other background characteristics. The simulation
uses these three data bases as inputs, and requires management to assist in the development of “new product offerings”. The consumer response to these offerings is then simulated. When a segment has

A. CAROLINE TYNAN AND JENNIFER DRAYTON

been selected, information on its estimated size and discriminating characteristics is available. The compMDnential segmentation procedure, which was promulgated by Green ei al. (1977) shifts the emphasis of the segmentation model from the partitioning of a market to a prediction of which type of person will be most respKjnsive to what type of product feature. The type of person is described in terms of demographic and psychographic attribute levels. The procedure used is an extension of conjoint analysis.

The componential segmentation model offers a new conceptualisation for market segmentation because it offers both an analysis of the market segment for a particular product offering and an evaluation of the most desirable product offering or positioning.

BASES FOR SEGMENTING MARKETS
Realising the potential benefits of market segmentation requires both management acceptance of the concept and an empirical segmentation study before implementation can begin. Wind (1978, p. 318) states that “most segmentation studies have been conducted for consumer goods”. However both the concept of segmentation and the majority of segmentation approaches are equally applicable to consumer and to industrial markets (Webster and Wind 1972, Nicosia and Wind 1977).

The segmentation model requires the selection of a basis for segmentation, (the dependent variable), and descriptors, (the independent variables), of the various segments. There is a very wide selection of variables mentioned in the consumer literature as possible bases for and descriptors of
segments. The segmentation base chosen to subdivide a market will depend on “the type of product, the nature of demand, the method of distribution, the media available for market communication, and the motivation of buyers” (Chisnall 1985, p. 266). These segmentation bases are rarely used alone, a combination of two or more of them is more usual. The various bases of segmentation analysis are discussed under the following headings: 1. Geographic bases in which markets are divided into geographic units. 2. Demographic bases include segmentation studies based on age, sex, socio-economic group, family size, life cycle, income, occupation, education, etc.

3. Psychological bases in which personality factors, attitudes, risk, motivation, etc. are used to divide the market. 4. Psychographic bases include lifestyle, activities, interests, opinions, needs, values and the like as market delineators.

5. Behavioural bases include brand loyalty, usage rate, benefits sought, use occasions.
6-7. Industrial segmentation and product segmentation are briefly discussed for the sake of completeness.

I. Gcogra^dc segmmtatioD
Geographic segmentation was jserhaps the first type of segmentation to exist, historically speaking (Lunn 1978). This is because many companies operate along geographic lines. “Small manufacturers who wished to limit their investments, or whose distribution channels were not large enough to cover the entire country, segmented the US market in effect by selling their products only in certain areas” (Haley J968, p. 30). These comments apply to many countries other than the United States of America.

Markets can be analysed nationally, regionally or locally. When assessing
market oppiortunities in different countries a useful categorisation is based upon gross national product per capita. This generates three main segments; those of the industrialized countries, the developing countries and the less developed countries. Weber (1974) has noted the piotential ofthe developing countries which account for only 19% of the world’s population but 32% of the world’s income. Areas can be studied for differences in buying behaviour attributed to locale. Food habits, for instance, tend to have regional variations. In Scotland, the consumption of both vegetables and beverages recorded by the National Food Survey is considerably lower than that of England and Wales (Household Food Consumption and Expenditure 1981 (1983), pp. 45-52}. One major development in the field of geographic based segmentation is Richard Webber’s ACORN (A Classification Of Residential Neighbourhoods). The system was developed from sociological research into urban deprivation in Liverpool and it classifies people and households according to the typie of neighbourhoods in which they live. After joining Consolidated Analysis Centre Inc. (CACI), Webber extended his original classification. Presently ACORN recognises 38 neighbourhood types, identified by a combination of 40 variables from census data. These 40 variables include age and household comjjosition, housing type, social and employment status. This division into 38 different types of household offered more detail than was necessary and so the 11 Family Group Classification shown below was devised (Chisnall 1986).

ACORN is a powerful segmentation tool which offers a detailed profile of particular segments together with their precise location. “It clearly has particular relevance to direct marketing, leaOet distribution and local media selection” (Chisnall 1986, p. 280).

In 1983 the Consumer L,ocation System (CLS) was launched. This system has combined BMRB’s Target Group Index (TGI) and ACORN. TGI examines the purchasing habits of approximately 24,000 consumers across more than 500 product fields. The TGI results are correlated with ACORN neighbourhood types to identify the concentrations of potential purchasers for a specific product. Actual names and addresses of people in relevant target groups can be generated as the electoral register has been computerised by CNN systems. This is clearly a much tighter specification of a mass consumer market than is currently available through any other medium (Rines 1983).

2. Demographic segmentatioai
This consists of dividing the market into groups on the basis of demographic variables such as age, sex, socio-economic group, family size, life cycle, income, occupation and education. Kotler (1984, p. 255) states that “demographic variables are the most piopular bases for distinguishing
customer groups”, pxjssibly because of the ease with which this kind of data can be collected. These characteristics have become the basic terms in which many marketers consider the consumer. This is reasonable in as much as demographic variables describe impiortant aspects of the consumer which give rise to purchasing requirements. Additionally demographic data has been collected over such a long period of time that relationships with other marketing variables e.g. media use have become well-known (Lunn 1978, p. 349). Thus many marketers collect demographic data on the characteristics of their consumers routinely, even when they intend to use some other bjise for segmentation.

In recent years demographic segmentation has been subject to considerable criticism. Stan ton (1978) has commented that “Looking at the demographic variables… rarely is a useful market segment identified by a single market factor”. McCarthy (1978) takes a similar stance commenting that product choice is only weakly related to demographics. A number of studies have revealed that demographic variables such as age, sex, income and occupation are pKXjr predictors of behaviour, and as such are of limited value in the formulation of market segmentation studies. Haley (1968, p. 31) has noted that they rely on descriptive rather than causal factors and as such are “not efficient predictors of future buying behaviour, and it is future buying behaviour that is of central interest to marketers”. Perhaps the most

MARKET SEGMENTATION

309

widely cited study quoted in this context is the study of 57 grocery products by Frank et al. (1967, p. 189). These authors concluded “socioeconomic and demographic characteristics are poor predictors of consumption for a wide range of specific grocery products”. Support for their conclusions have come from other researchers (Koponen 1960, HUdegaard and Krueger 1964; Frank 1967, Massy et al. 1968}. The only researchers to take exception to this conclusion are Bass et al. (1968) and more recently Wheatley et al. (1980). Both these research teams have taken the approach of examining group rather than individual behaviour which may account for their success in using demographic segmentation variables where approaches based on the individual consumer failed. However demographic bases for segmentation of both individual and group behaviour are still avidly defended by many marketing research practitioners (e.g. Ckjrnish 1981).

Age has frequently been used as a base for segmentation on the basis that consumer wants and capacities change with age. ‘A notable state of the art review of this variable was conducted by Phillips and Stemthal (1977). They examined the differences made by age in information processing with a view to designing more appropriate advertising communications for different age segments. The study addressed two issues:

(a) Whether elderly individuals show a differential sensitivity in processing information in relation to younger people.
(b) At what age these differences are manifested.
They concluded that age differences result in a complex set of changes in individuals’ sources of information, ability to learn and susceptibility to social influence. These changes do not necessarily occur at 65 yeare but they are related to the social, psychological and physical changes that accompany ageing. In many product areas the sex of a consumer determines the product he or she will buy. For instance, sex segmentation has long been applied to clothing, magazines, cosmetics and toiletries (Kotler 1984). More recent work has indicated sex differences in a product area not normally considered to have a sex link, that of food (Dickens and Chappell 1977). In some cases the life cycle concept has proved a more useful segmentation variable than age (Lansing and Kish 1957). These authors made a comparison between life cycle and age on six aspects of the family’s consumption pattern and concluded that life cycle discriminated better than age in all six cases, and that life cycle analysis provided some useful information that analysis by age tended to conceal. life cycle is a composite variable, made up of factors which include age, number of years married, age of children and working status. The concept was postulat«i in the 1930s but only developed in a marketing sense in the 1950s and 1960s. It is applicable to the conventional nuclear family. Berkman and Gilson (1981, p. 179) consider that it may “not reflect current trends such as the two income family”. Derrick and Lehfeld (1980, p. 214) give a detailed analysis of the litnitations of the traditional life cycle which they consider “fall under

two main headings—operationalising the stages and interpreting the results”.
An important conference, entitled “The Life Cycle and Consumer Behaviour” was held in Michigan in 1954 (Clark 1955) where several key papers were presented. Other important contributions came from Lansing and Kish (1957), Wells and Gubar (1966) and more recently StampQ (1978). A detailed and extensive analysis of the relationship betwen life cycle and consumption behaviour is presented by Reynolds and Wells in their book Consumer Behaviour (1977). Their model is presented below in Figure 2.

The basic assumption underlying the family life cycle approach is that most households pass through an orderly progression of stages each with its own characteristic purchasing pattems. In spite of the difficulties of classifying some respondents who do not fit neatly into any of the usual stages, e.g. older single pieople, or the widow who has young children, life cycle remains a useful segmentation base at a general level. Income is another important segmentation base, and one which has been reasonably well researched. There are difficulties in establishing household

income from all sources after deductions, some of which can be attributed to peoples’ reluctance to divulge this type of information, and some to the complications of having more than one wage earner in the family. Several authors have concluded that income is the best of the demographic bases for segmenting markets (Allt 1975, Stanton and Haug 1971, Slocum and Mathews 1970).

Socio-economic classification has the advantages of being both widely understood and used as the basis for media classification for many years (Chisnall 1985). To some extent it also subsumes the other segmentation variables, income, occupation and education. In Britain the classification system most popular with market researchers for socio-economic group is the A to E grading used in readership surveys and advocated by the Market Research Society (Wolfe 1973). This classification is shown below in Figure 3.

Source: Monk, D. (1970, “Social grading on the national readership survey”, London, Research Services, Joint Industry Committee for National Readership Surveys. FIGURE 3 Socio-economic classification

Social class, by whatever definition, is popular as a segmentation variable in spite of its doubtful ability as a predictor of consumer behaviour. It has been cited as a useful segmentation base by several authors. Martineau (1958) found a close relationship between store choice, patterns of spending and social class. Packard (1969) also noted the “pride, pleasure and prestige” which many women feel in patronising a high-class store. According to Rich and Jain (1968) the higher the social class of a consumer the more quickly she wished to complete her shopping. However since this early research, the market place and the consumer have both changed considerably with the result that many authors consider social class is now a poor discriminator in many product fields, particularly for fast moving consumer goods. The social mobility of many individuals in our affluent society has undoubtedly contributed to this change. The sharp divisions between different social classes are blurring as consumption and earning habits change. It is unlikely that social class will ever regain its prominence as a segmentation variable.

In an attempt to improve the j>oor predictive ability of segmentation based on single demographic variables Research Services Limited developied SAGACITY, a classification based upon a combination of life cycle, income and socio-economic group. SAGACITY is founded on the premise that consumers have different aspirations and behaviour at different stages of the life cycle. Four life cycle stages are subdivided by income, either better or
woKe off and by occupation, either manual (blue-collar) or non-manual (white-collar). The SAGACITY model is shown below with a brief description of the twelve groups and an indication of their size when compared to the total adult population (Market Research Society 1984).

Combining severat demographic variables offers an improvement in discrimination over using them alone but this method is still liable to the criticisms levelled at income and socio-economic classifications mentioned above.

3. Psychological segmentatioii
General dissatisfaction with geographic and demographic characteristics as segmentation bases led to the use of psychological variables as a basis for predicting consumer behaviour. It was anticipated that procedures developed and tested by anthropologists, sociologists and psychologists could be applied to the segmentation of consumer markets. The work has centred on the use of variables such as personality, risk, reference groups and attitudes. The use of personality as a segmentation variable has not met with any significant success. Kassarjian (1971) calls the results “equivocal”, Van Vetdhoven (1973) calls them “disappointing”, although a great deal of work has been conducted in this area. The idea of classifying people by personality tyjje is not new, one famiUar personality segfmentation is that of Hippocrates who used the four humours, choleric, melancholic, sanguine and phlegmatic. Another example is provided in Spranger’s book. Types of Men., published in 1928 and reported much later by E)^enck (1954). He identified six types:

Theoretical —dominant interest in discovering the truth, a cognitive attitude towards life.
Economic —dominant interest in what is useful, a practical approach to life.
Aesthetic —dominant interest in form and harmony, an artistic approach to life.
Social

—dominant interest is in people—love is his main approach to life.
Political
—dominant interest is in social relations but with a basic life approach of power.
Religious —dominant interest in doctrines and philosophy, with a mystical approach to life.
According to Kassarjian (1971) there is no accepted defmition of the term “personality”. However Jahoda and Warren (1969, p. 9) have defined it as “the total organisation of internal psychological functioning”. One underlying basic assumption of personality theories is that “personality reflects enduring needs of the individual; that is needs that are ‘common denominator?’ of the person’s behaviour regardless of the nature of the problem situation with which he is faced” (Mostyn 1977, p. 27). This implies that an individual has an enduring set of tendencies to behave in a given way to given classes of stimuli.

One of the earliest papers linking psychological factors with consumer behaviour was that of Mason Haire (1950). Resp)ondents were asked to describe the personality and character of the women whose shopping list they examined. They were given two lists which only differed in respect of the coffee listed, one was Nescafe Instant, the other Maxwell House Drip grind (a type of finely ground coffee bean). The respondents were able to characterise these two women with the result that they saw the Nescafe shopf>er as lazy and a poor household planner and the Maxwell House shopper as thrifty and a good wife. Since this work psychological segmentation has proliferated. A useful examination of the topic has been published by Mostyn (1977). Firstly, some of the important pajiers which have indicated a significant correlation {in the author’s opinion) between personality and consumer behaviour will be examined. This will be followed by those studies which found a weak correlation.

Koponen (1960) used the Edwards Personal Preference Schedule in a study of smoking and found that sex dominance, aggression and achievement needs were positively related to cigarette smoking in men. A significant correlation between extroversion, as measured by Eysenck’s scales, and smoking was
established by Eysenck et al. (1960). Tucker and Painter (1961) found significant correlations between pereonality traits and the use of headache remedies, vitamins, mouthwash, alcoholic drinks, motor cars and chewing gum using the Gordon Personal Profile. Westfall (1%2) established significant differences between convertible, compact and standard car owners using Thurstone’s Temperament Schedule. Claychamp (1965) used the Edwards Personal Preference Schedule and found that personality variables predicted better than demographic variables whether a resfKindent was a customer of a bank or savings and loan association. A study in 1970 by Lehmann, found that mothere who were both anxiotis and high on self-esteem were less likely to be persuaded to try new ideas for their babies, whereas mothers who were low on these traits were easier to persuade.

Fry (1971) used a variety of personality tests, including a special self-confidence scale, and established a significant correlation between sex, class and brand choice for cigarettes. Using the Rokeach dogmatism scale in a study conceming fashion and cosmetic products, Jacoby (1971) found the low-dogmatic women signiGcantly more likely to make innovative choices. Coney (1972) replicated and extended this study by including men and additional product categories. He confirmed that low-dogmatics were significantly more likely to be innovators. Blake et al. (1973) established that dogmatism was significantly related to the acceptance of new products, but not the acceptance of novel products.

The foiiowing pajiers all found poor correlation between personality and consumer behaviour. In a study reported by Britt (1966, p. 182) and sponsored by the Advertising Research Foundation, it was found that “in predicting toilet tissue purchase behaviour, information on the demographic and personality traits was little better than no information at ali”. Myers (1967) reported little correlation between pereonality and attitudes towards private brands. He used the Cattell Personality Factor Inventory and could explain only 5% of the variance in the purchase data. In a similar study of own label (store brands) versus branded goods, Massy et ai. (1968) used the Edwards Personal Preference Schedule. They concluded that in only 26% of cases did personality variables add a significant increment to the prediction that could be made with socio-economic data alone.

Robertson and Myers (1969) used the California Personality Inventory plus measures for innovativeness and opinion leadership in a study of new appliances, food and clothing products. They found only a minimal relationship between piersonality variables and behaviour towards new products. Levonian (1969) stated that of eight studies claiming to associate personality and opinion change foiiowing a mass communication, 40% actually yielded results in the opposite direction. Personality was not a useful discriminator variable in differentiating between “new season’s” and “last year’s model” car buyers (Wiseman 1971). Pizam (1972) found that only 16 of the 37 personality traits he tested had a significant association with innovativeness. Finally, Villani (1975) found that demographic variables “outperformed” personality variables in his study of television programme viewing. In the research cited above both significant and not significant results have been obtained using the same personality tests and frequently with reference to the same subject, innovativeness for example.

Clearly some of the studies suffer from problems of reliability and validity. An additional problem has best been summed up by Kassarjian (1965, p, 146) “The consumer researcher too often expects more from an instrument than it was originally intended to furnish”. As Kotler (1984) has jrointed out, even where evidence has been found of the influence of personality on consumer behaviour, the implications for marketing strategy have remained unclear. Another of the psychological areas that has been used by marketen to predict consumer behaviour is risk, Bauer first proposed the concept of “perceived risk” in buying decisions in 1960. He intimated that risk is assessed differently according to the perceptions of individual buyere. Since then many empirical studies have been conducted on various aspects of risk in consumer decision making, Taylor (1974) has hypothesised that the element of choice in consumer behaviour involves risk because the outcome of the choice is uncertain, and that this risk is seen in terms of a possible financial, social, performance, psychological, physical or convenience loss.

The individuals reaction to these “piotential losses would depend on that individual’s amount of perceived risk”, Peter and Ryan (1976) have defined perceived risk as the expectation of losses associated with a purchase which acts as an inhibitor to purchase behaviour. On the assumption that individuals would behave in such a way as to reduce the amount of risk in a purchase situation, Cunningham (1967) has shown that brand loyalty is determined to a great extent by perceived risk, Arndt (1968) showed a similar relationship between the use of word of mouth communication about a product and perceived risk, as did Bearden and Mason (1978) in prescribed drug purchases. Another risk related theory is Festinger’s (1957) Theory of Cognitive Dissonance, In essence this theory states that an individual wil! seek to try to reduce dissonance or disharmony within his cognitive structure and attempt to reach a state of harmony. This is achieved by resolving the conflict between the various factors which are not psychologically consistent with one another. The individual is motivated to change his opinion, attitude or behaviour in order to reach a state of harmony. In two studies of automobile purchase, Ehrlich et al. (1957) indicated that as automobile buyers seek information on model and make already bought they are dissonant consumers, Engel (1963) however suggested that as automobiles are no longer a status symbol there should be little dissonance after purchase.

The psychological concept of reference groujjs has also been used to explain consumer behaviour and segment markets. The concept of reference group was first propiosed by Hyman (1942) to describe the kind of group used by an individual as a pwint of reference for his own judgement, beliefs and behaviour, Venkatesan (1966), Friedman and Fireworker (1977) have confirmed that much consumer decision-making is influenced by the pressure to conform to group norms. Bourne (1957) indicated that the more conspicuous a product, the more likely its purchase is susceptible to reference group behaviour.
Reference group influence is more likely to be effective in the case of products which reflect {>ersonal taste (Chisnall 1985),

There is also the unproved implication in many of these studies that an individual’s susceptibility to reference group influence is related in some way to his personality.
The final psychological variable to be considered separately in this study is attitude. This concept is frequently considered for use as both a segmentation base and a possible predictor of consumer behaviour. In segmenting a market the users of a product are frequently identified by means of their attitude towards that brand or product. This assumes that there is some causal link between the attitude and the purchase behaviour, a link which while frequently hypothesised has not been proved. McGuinness et al. (1977), Crespi (1977), Howitt and McCabe (1978) all consider that attitude does predict behaviour. Fishbein (1967), Ajzen and Fishbein (1973), Pinson and Roberto (1973) however deny the causai link between attitude and behaviour. Work still continues to determine the true link between attitudes and behaviour.

In conclusion it should be noted that much of the published work on psychological bases of segmentation is in conflict. Although these variables do influence buying behaviour there is no reason to believe there exists a generalised pattern of influence. According to Chisnall (1985, p. 268) “individual products should be carefully analysed for the potential or actual personality factors influencing their sales”.

4. Psychographic segmentation
During the 1960s, a blend of personality and motivation research began to take shape (Wells 1975). It has been referred to “as the marriage between the richness of motivational research with its emphasis on qualitative methods and projective techniques and the statistical sophistication of the factor and trait theorists who made psychological segmentation studies possible” (Mostyn 1977, p. 31). This new area has been variously cjdled lifestyle (Plummer 1971, 1971-2), psychographics (Nelson 1969, 1971a, 1971b, Demby !971, Pernica 1974), activity and attitude research (Hustad and Pessemier 1971, 1974, Wells and Tigert 1971) and “activities, interest and opinions” research (Engel et al. 1978). Therefore psychographic researchers have moved beyond demographics and considered activities, interests, opinions, needs, values, attitudes and personality traits.

There is as yet no generally agreed definition. In his excellent review article. Wells (1975, p. 197) has proposed an operational defmition of psychographic research as “quantitative research designed to place consumers on psychological as distinguished from demographic dimensions”. The technique divides the market into segments on the bauis of interest, values, opinion, personality characteristics, attitudes and demographic variables using techniques of factor analysis, cluster analysis and canonical correlation (Kassarjian 1971). It is assumed that products which fit into the lifestyle will have value for the consumer. Wells (1968) named this approach “backward segmentation” because it groups people by their behavioural characteristics before seeking correlates. In this approach the analysis of buyer behaviour starts with the behaviour itself. Complex statistical techniques such as factor analysis and cluster analysis are applied to purchasing data across a wide variety of products to seek for pattems of complementary and substitutable products (Bass et al. 1969). The three main applications of backwards segmentation are to

(a) Stimulate ideas and guide future research.
(b) Simplify marketing strategies.
(c) Increase understanding by stimulating researchers to question why sets of products group together as they do.

Nelson (1971b} has noted some of the reasons for the growing ptopularity and importance of psychographics.
(a) The general acceptance of the need for the application of behavioural science information to advertising and marketing problem solving. (b) The availability of computer programmes that can perform multivariate analysis on large numbers of pieople. (c) Acceptance of the concept of market segmentation.

(d) Decreasing relevancy of certain demographic characteristics. (e) Change taking place in our social structure.
Psychographics emphasises the importance of general environmental, cultural and social factors, e.g. socialisation and group pressure and as such takes up where psychological segmentation leaves off (Mostyn 1977). Psychographic surveys usually employ Likert Scales and self-administered questionnaires which are largely precoded to facilitate analysis. The samples tend to be large and the analysis ranges from the use of simple crosstabulations to the more complex factor analysis, cluster analysis or canonical correlation.

Wells (1974) has discussed the uses of psychographics in detail. He has summarised the techniques used in market segmentation the following way. “Life style and psychographic research can assist market segmentation in a variety of ways. It can provide useful descriptions of existing segments of present markets. It can help the analyst understand the results of multidimensional scaling or prciduct benefit segmentation. It can contribute new and useful dimensions along which consumers may be segmented. It can create new segments based upon product and/or brand related interests, needs and values. And it can create new segments based ufjon more general aspects of life style”.

Tipton (1972) has noted the general uses to which psychographic segmentation has been put, apart from market segmentation they are; (a) New product development in which a “gap” of unfulfilled needs or wants is identified in the market and a new product designed to fill that gap.

(b) Media selection—the knowledge of the psychographic profile enables the selection of the most effective and economical media mix to reach the segment,
(c) Creative application—knowledge of how consumers live and think is helpful to a researcher in designing an advertising campaign or making a new product appealing.
Wells (1975) divided segmentation studies into those where the segmentation was based on general life style dimensions and those in which the psychographic items were product si>ecific. An example of the “general” approach has been drawn from Wells (1975, p, 201), In this example approximately 4000 respondents, answered psychographic, product use and media exposure questions. Using factor analysis the eight homogeneous groups described in Figure 5 were derived.

The product and media use of these eight psychographic groups were then calculated.
In Britain, Attwood Statistics characterise the housewives on their consumer panel by their psychographic groups and behaviour into the following types (Crimp 1985, p. 115),
1, Conscientiousness related to housework,
2, Economy consciousness,
3, Conservatism in brand (tending to better known brands rather than experimenting with a new brand or product),
4, Traditionalism in housework (related to the use of labour-saving or convenience products),
5, Willingness to experiment in shopping.
Many general life style profiles have appeared in the literature. They include the following papers. Ziff (!971, 1973) used psychographics to show that both products and product classes can be differentiated by attitudes, needs and values. Both Plummer (1971-2) and Tigert (1974) researched psychographic profiles for magazine readers, Tigert also investigated television viewers. Scales representing fashion interest, fashion venturesomeness, cognitive style, information seeking, relative popularity and relative confidence have been used to predict fkshion opinion leadership by Darden and Reynolds (1972), Other general profiles have investigated carry-out foods (Tigert el al. !97I), beer (Tigert 1971), and bank charge cards (Plummer 1971). An interesting study of shopper types, recently published by Lesser and Hughes (1986), examined psychographic segmentation solutions from different locations. The authors concluded that psychographic segments developed for markets in one geographic location are generalizable to markets in other geographic locations.

The second group of studies involves those in which product specific variables have been employed. Pernica (1974) has reported on a psychographic segmentation for stomach remedies. He focused on product related

He is a self-sufficient man who wants to be left alone and is basically shy. Tends to be as little involved with community life as possible. His life revolves around the family, simple work and television viewing. Has a marked fantasy life. As a shopper he is practical, less drawn to consumer goods and pleasures than other men.

Low education and low economic status, he tends to be older than average. Group II.

“The Tradiuonalist” (16% of total maies).

The man who feels secure, has self-esteem, follows conventiotial rules. He is proper and respectable, regards himself as altruistic and interested in the welfare of others. As a shopper he is conservative, Ukes popular brands and well known manufacturers. Low education and low or middle socio-economic
status; the oldest age group. Group III.

“The Discontented Man” (13% of total males).

He is a man who is likely to be dissatisfied with his work. He feels by passed by life, dreams of better jobs, more money and more security. He tends to be distrustful and socially aloof. As a buyer, be is quite price conscious.

Lowest education and lowest socio-economic group, mostly older than average. Group IV.

“The Ethical Highbrow” (14% of total maies).

This is a very concerned man, sensitive to people’s needs. Basically a puritan, content with family life, friends, and work. Interest in culture, religion and social reform. As a consumer he is interested in quality, which may at times justify greater expenditure. Well educated, middle or upper socio-economic status, mainly middle aged or older. Group V.

“The Pleasure Oriented Man” (9% of total males).

He tends to emphasise his masculinity and rejects whatever appears to be soft or feminine. He views himself a leader among men. Self-centred, dislikes his work or job. Seeks immediate gratification for his needs. He is an impulsive buyer, Ukely to buy products with a masculine image.

Low education, lower socio-economic class, middle aged or younger. Group VI.

“The Achiever” (11 % of total males).

This is likely to be a hardworking man, dedicated to success and all that it implies, social prestige, power and money. Is in favour of diversity, is adventurous about leisure time pursuits. Is stylish, likes good food, music, etc. As a consumer he is status conscious, a thoughtful and discritninating
buyer.

Good education, high socio-economic status, young.
Group VII.

“The He-Man” (19% of total males).

He is gregarious, likes action, seeks an exciting and dramatic life. Thinks of himself as capable and dominant. Tends to be more of a bachelor than a family man, even after marriage. Products he buys and brands preferred are likely to have “self-expressive value”, especially a “Man of Action” dimension.

Well educated, mainly middle socio-economic status, the youngest of the male groups. Group VIIl.

“The Sophisticated Man” (10% of total males).

He is hkely to be an intellectual, concerned abou! social usues, admires men with artistic and intellectual achievements. Socially costnopolitan, broad interests. Wants to be dominant, and a leader. As a consumer he is attracted to the unique and fashionable. Best educated and highest status of all groups, younger than average.

The Severe Sufferere are the extreme group on the potency side of the market. They tend to he young, have children, and he well educated. They are irritable and anxious people, and believe that they suffer more severely

than others. They take the ailment seriously, fuss about it, pamper themselves, and keep trying new and different products in search of greater potency. A mer given time span should be

used in preference to purchase frequency as a measure of user rate. They found that either purchase frequency, or size purchased together with number of units purchased at one time can be deceptive when used alone. An interesting paper by Dickson (1982) makes an excellent case for the adoption of the usage situation in market segmentation. He redefines markets and demand in terms of people and usage situations offering the following “Sjjeculative Person Situation Segmentation Matrix for Suntan Lotion” as an example of his segmentation framework. (See Figure 8) Alternatively, many markets can be segmented into non-users, potential users, first time users, regular usere, intermittent users or ex-users. A different marketing approach will be required for each of these groups. Little research has been conducted on this topic but at a pragmatic level many companies with a high market share will attempt to convert potential users into actual users, whereas smaller firms would be more likely to encourage users to switch to their brand.

6. Industrial segmentatioii
As mentioned in the introduction to this section on Bases for Segmenting Markets, the majority of segmentation approaches are equally applicable to consumer and industrial markets (Webster and Wind 1972, Nicosia and Wind 1977). Clearly some situational variables are specific to either the consumer market or to the industrial market, but it should be noted that the industrial market can be segmented by geograpfiic, demographic, psychological, psychographic and behavioural variables in just the same way as the consumer market,

A two stage approach to industrial market segmentation was developed by Wind and Cardozo in 1974, The first sUge is macro-segmentation where the market is segmented by bases including industry demographics, size, indtistrial sector (i.e, commercial firm or public sector), SIC (Standard Industrial ClassiOcation) code and product usage. The second stage called micro-segmentation is based on the demographic and behavioural characteristics of decision making units or buying centres. Subsequently ChofTray and Lillien (1978) attempted to operationalize this second stage of Wind and Cardozo’s model.

Since this early work much of the academic research has “unfortunately attempted to transpose literature on the industrial buying process to explain industrial market segmentation” (Hlavacek and Reddy 1986, p, 13), Segmentation based on industrial end use, buying power and industrial concentration are recommended (Chisnall 1986), with insufficient emphasis being placed on the development of an understanding of buyers needs.

6. Product segmentation
It should be mentioned for the sake of completeness that it is possible to cluster products rather than consumers by using product segmentation, Bamett (1969), an enthusiastic proponent of tfiis approach, has argued that researchers should abandon consumer segmentation and concentrate instead on deriving product field speciGc criteria by which consumers themselves distinguish between brands and products. However “it is just as valuable to cluster consumers in terms of their requirements from product field specific variables as it is to cluster products in terms of the extent to which they are perceived to satisfy these requirements” (Lunn 1978, p, 366-7), A full description and critical appraisal of the different methodological approaches to product segmentation can be found in Beazley (1973). Product segmentation has proved especially useful in identifying gajK for new product oppwrtunities, and in checking which brands or products are competing with each other.

REQUIREMENTS FOR EFFECTIVE SEGMENTATION
There is general agreement in the literature on factors which affect the feasibility of market segmentation, Kotler (1984) originally cited three factors namely, measurability, accessibility and substantiality. In later revisions he has added a fourth factor, i,e, actionability, A brief^ description of these factors follows:

Measurability This refers to the effective size and purchasing power of a submarket. It is dependent on the availability of suitable market research data concerning the segmentation variable chosen.

Accessibility This is the degree to which a segment can be effectively reached and served. It largely rests upon the ability of a firm to direct its marketing effort at a particular segment. Media coverage, distribution and the influence of behavioural factors, all need to be evaluated. It is important to choose a media mix which will reach the target segment both economically and efficiently. Likewise the distribution network chosen must be effective in reaching the sub-segment. Chisnall {1985, p. 265) also recommends that “reference should be made to group behaviour, opinion leadership, family life styles” for the sub-segment under consideration. Substantialitj The segment must be sufficiently large and profitable to be economically viable for the firm. Kotler (1984, p. 265) maintains “a segment should be the largest jKJSsible homogeneous group worth going after with a tailored program”.

Actionability The degree to which a firm can develop and manage effective programmes for attracting and serving segments. This factor relates to the requirements and capabilities of particular firms.

To these four factors Thomas (1980) adds another, that of “stability” i.e. can you predict the segment’s behaviour in the future.
Clearly there are many ways to segment a market, although they are not all equally effective. In order to select an appropriate method of segmenting a particular market, the segment (s) identified must exhibit the four characteristics explained above. If the segment chosen is measurable, accessible, substantial and actionable then in Kotler’s words it should be “maximally useful”.

CRinCSSMS OF THE THEORY OF MARKET SEGMENTATION
The concept of market segmentation, and most market segmentation studies, is “based on the premise that the given market is heterogeneous and can be segmented” (Wind 1978, p. 327). Chisnall {1985, p. 262) goes further, he maintains that “practically every market is capable of refinement into significant sub-markets”. Young et al. (1978) disagree, they have suggested three occasions in which market segmentation is not useful. (a) where the market is so small that marketing to a small pwrtion of it may not be profitable.

(b) where heavy users constitute such a high proportion of the sales volume that they comprise the only relevant targets.

(c) where a brand is the dominant brand in the market. (However, even here, segmentation provides information about the consumers who constitute the target market which will enable the marketing manager to “taiior” tbe marketing mix to their needs). Collins (1971) has argued that the traditional assumption of all marketing studies that a product can be placed in a market segment is an oversimplification. He considers not all consumers can be placed in a segment and that most brands of products do not operate within a definable segment. His hypothesis is that a brand’s sales depends on its penetration and buying rate rather than on the segment size. Collins accepts the concept of market segmentation in the case of what he calls the “segment brands”, which he defines as appealing to some relatively restricted section of the market and having more heavy than normal buyers. Resruk et al. (1979) have drawn attention to their belief that segmentation has gone too far.

They have identified markets which are hyper-segmented, a condition they define in cases where increasingly smaller market segments are identified and targeted, thus causing corresponding increases in production and marketing costs. These authors have postulated the strategy of “counter segmentation” as an appropriate response to hyper-segmentation. In counter segmentation market segments are aggregated or clustered so that a simpler product is offered, with lower production and marketing costs, and some of the savings are passed on to the consumer through lower prices. In spite of the interest this paper initially generated the strategy of counter segmentation does not appear to have been widely adopted.

While all these criticisms are valid in a small number of cases they are not universally applicable and thus do not seriously limit the theory or practice of market segmentation.

CONCLUSION
Market segmentation is clearly a crucial marketing strategy. It enables the
marketing manager to divide total demand into relatively homogeneous segments identified by geographic, demographic, psychological or behavioural variables. These characteristics are relevant in explaining and in predicting the response of consumers, in a given segment, to marketing stimuli. Once a segment has been identified which fulfils the requirements of measurability, accessibility, substantiality and actionability it is possible to develop a product or service to meet the needs of the segment. A marketing mix can then be devised to reach the segment identified efficiently and economically.

References
Ajzen, I and Fishbein, M. (1973), “Atdtudina! and normative variables as predictors of specific behaviour”, Journal oJPersonality and Social Psychology, 27(1), pp. 41-57.

Allt, B. (1975), “Money or class; New light on household spending”, Advertising Qaarttrly, 44, Summer, pp. 6-9
Amdt, J. (1968), “Word of mouth advertising and perceived risk”. In: Perspective in Consumer Behaviour. (Eds.) Kassarjian, H. H. and Robertson, T. S., Glenview, Illinois, Seott, Foresman and Co., pp. 330-336.

Baker, M. J. (Ed) (1984), Macmiltan Diciianary of Marketing and Advertising, London, Macmillan.
Barker, S. M. and Trost, J. F. (1973), “Cultivate the high-volume consunier”. Harvard Business Review , 51(2), pp. 118-122.
Bamett, N. L. (1969), “Beyond market segmentation”, Harvard Business Review, 47(1), pp. 152-166.
Bass, F. M., King, C. W. and Pessemier, E. A. (1968), AppticaHms af the Sdmces ia Marktting Management, New York, John Wiley and Sons.
Bass, F. M., Pessemier, E. A. and Tigert, D. J. (1969), “Complementary and substitute patterns of purchasing and use”, Joumat of Advertising Research,
9(2), pp. 19-28. Bass, F. M., Tigert, D. J. and Lonsdale, R. T. (1968), “Market segmentation: Group versus individual behavior”. Journal of Marketing Research, 5(3), pp. 264—270. Bauer, R. A. (1960), “Consumer behaviour as risk taking”. In: Dynamic Marketing for a Changing Wortd. (Ed.) Hancock, R. S., Chicago, Illinois, American Advertising Association, pp. 389-398. Bearden, W. O. and Mason, J. B. (1978), “Consumer perceived risk and attitudes toward generically prescribed drugs”. Journal of Applied Psychology, 63(6), pp. 741-746. Beazley, D. (1973), “Alternative approaches to brand piositioning research: How do we choose?”. In: Brand Positioning. (Eds.) Green, P. A. and Christopher, M., Cranfield Institute of Technology, UK.