Word-of-Mouth:
Influences on the Choice of Recommendation Sources

by

K Schoefer

1998

A dissertation presented in part consideration for the degree of M.A. in Corporate Strategy and Governance.

Contents

Chapter 1 Introduction 1

Chapter 2 Literature Review 3

1. Consumer Decision-Making 3 1. The Decision-Making Process 4 1. Problem Recognition 5 2. Information Search 6 3. Evaluation of Alternatives 8 4. Purchase 9 5. Post-Purchase Evaluation 10 2. Individual Influences 12 3. Environmental Influences 15 1. Culture 15 2. Social Class 16 3. Reference Groups 16

2. Word-of-Mouth Communication 20 1. Definition 20 2. Scope and Significance 20 3. Characteristics of WOM 22 4. The Nature of WOM 24 1. Types 24 2. Process 24 3. Conditions 28 4. Motives 29 5. WOM and the Consumer Behaviour Literature 29 1. Adoption and Diffusion on Innovations 29 2. Post-Purchase Decision-Making 31 3. Pre-Purchase Decision-Making 33

Chapter 3 Empirical Analysis 34

1. The Model 34 1. Recommendation Sources 34 2. Evaluative Cues 35 3. Task Difficulty 36 4. Prior Knowledge 37 5. Hypotheses 39 2. Research Context 43 3. Method 45 1. Research Instrument 45 2. Procedure 47 3. Sample 48 4. Results 50 5. Discussion 57 6. Limitations 61

Chapter 4 Conclusion 62

Appendix A
Appendix B

References

Chapter 1: Introduction The idea of understanding consumer behaviour as a sequential decision-making process is one that is common in marketing (Engel et al., 1993; Wilkie, 1994; Solomon, 1993; Assael, 1992; Loudon and Della Bitta, 1993; Kotler, 1997). The decision-making process itself is presented as a logical flow of activities, working from problem recognition to purchase to post-purchase evaluation. This decision-making process is affected by a number of other more complex influences. Some of these influences relate to the wider environment in which the decision is being made while others relate to the individual who makes the decision.

In this context, ".. [o]ne of the most widely accepted notions in consumer behavior is that word-of-mouth communication (hereafter WOM) plays an important role in shaping consumers' attitudes and behaviors." (Brown and Reingen, 1987) More specifically, WOM communications between consumers are a topic of interest in both the pre-purchase and post-purchase decision-making literature. Research into the diffusion of innovations has focused on modelling the role of WOM in product adoption at various stages of the diffusion process (Mahajan et al., 1990). WOM has also been studied as a mechanism through which consumers convey both informational and normative influences in the product evaluation (Arndt, 1967; Brown and Reingen, 1987). Finally, WOM has been identified as an important post-purchase complaining option (Day, 1984; Singh, 1990).

Although WOM plays an important role in consumer pre-purchase and post-purchase decision-making, research into this phenomenon has been fragmented. Importantly, relatively little attention has been directed at understanding key issues with respect to WOM recommendation sources and the factors that influence their use (Brown and Reingen, 1987, Duhan et al., 1997). The aim of the present work is to add to this small body of empirical research.

The main part of this paper is divided into two chapters. Chapter 2, that follows an introduction to the work, is a theoretical one. It is a review of the literature on consumer decision-making and the individual and environmental influences on it. Emphasis is being placed on WOM communication and its role in understanding consumer behaviour. Chapter 3 focuses on the choice of WOM recommendation sources. Empirical research is presented which explores the influences on the choice over WOM recommendation sources. Finally, concluding remarks and recommendations for further research can be found in chapter 4.

Chapter 2: Literature Review
2.1 Consumer Decision-Making Engel et al. define consumer behaviour as "... those activities directly involved in obtaining, consuming, and disposing of products and services, including the decision processes that precede and follow these actions." (1993, p. 4). Thus, in the marketing context, the term 'consumer behaviour' refers not only to the act of purchase itself but to any pre- and post-purchase activities (Foxall, 1985, 1997; Ennew, 1993). Pre-purchase activities would include the growing awareness of a want or need, and the search for and evaluation of information about the product and brands that might satisfy it. Post-purchase activities would include the evaluation of the purchased item in use, and any attempt to reduce feelings of anxiety which frequently accompany the purchase of expensive and infrequently bought items like consumer durables. Each of these has implications for purchase and repurchase and they are amenable to marketing communications and the other elements of the marketing mix. Our understanding of both consumer behaviour and the capacity of marketing activities to influence it rest on knowledge of the ways in which consumers form decisions (Foxall, 1985, 1997).

There have been many attempts to create models of consumer decision-making such as those proposed by Howard and Sheth (1969), Nicosia (1966) and Engel et al. (1968). Since a review of these models would be beyond the scope of this chapter, a simplified approach has been adopted to guide the discussion. A diagram of this approach is presented in figure 1.

Figure 1: Consumer decision-making framework [adapted from: Dibb et al., 1997]

Reference to figure 1 reveals that it is made up of three major sections: (1) the consumer's decision-making process, (2) individual determinants of behaviour, and (3) environmental variables influencing behaviour. These major sections will be examined in more detail below.

2.1.1 The Decision-Making Process As shown in figure 1, a major part of consumer behaviour is the decision process used in making purchases. This decision-making process, according to Engel et al. (1993), includes five stages: (1) problem recognition, (2) information search, (3) evaluation of alternatives, (4) purchase, and (5) post-purchase evaluation. An obvious criticism of this conceptualisation, however, would be that not every purchase will involve such an extensive decision-making exercise. The extent to which each of these steps is followed in the precise form and sequence can vary from one situation to the next. Some decisions are rather simple and easy to make, whereas others are complex and difficult. Consumer decisions can thus be classified into one of three broad categories: routine response behaviour, limited decision-making and extensive decision-making (Howard, 1977; Brassington and Pettitt, 1997; Loudon and Della Bitta, 1993; Solomon, 1993; Wilkie, 1994).

Routinised-response behaviour occurs in purchasing situations which the consumer is likely to experience on a regular basis. The items that fall into this category do tend to be low risk, low priced, frequently purchased products such as food and household products. In this situation, the actual identification of a need may not occur explicitly; there may be little or no information search and the consumer may rely heavily on brand loyalty. Over-time, the repeat purchase becomes habitual, with little or no re-evaluation of the decision. Consumers engage in limited decision-making when they buy products occasionally and when they need to obtain information about an unfamiliar brand in a familiar product category. This type of decision-making requires a moderate amount of time for information gathering and deliberation. Typical examples include electrical goods, furniture and holidays. Finally, extensive decision-making comes into play when a purchase involves unfamiliar, expensive or infrequently bought products like cars or houses. Extensive decision-making is usually initiated by a motive that is important to the buyer's self-concept and the eventual decision is perceived to carry a high degree of risk. Moreover, the consumer will engage in extensive information search and evaluation prior to purchase and the purchase itself will be a relatively long process (Brassington and Pettitt, 1997; Loudon and Della Bitta, 1993; Solomon, 1993; Wilkie, 1994; Ennew, 1993).

2.1.1.1 Problem Recognition Problem recognition represents the beginning of a consumer's decision-making process. At this stage the consumer perceives a need and becomes motivated to solve the problem that he/she has just recognised. Once the problem is recognised, the remainder of the consumer decision-making process is invoked to determine exactly how the consumer will go about satisfying the need (Wilkie, 1994). Conceptually, problem recognition occurs when the consumer identifies a discrepancy between his/her actual and desired state. However, the presence of need recognition does not automatically activate some action. This will depend on two factors. First, the recognised need must be of sufficient importance. Second, consumers must believe that a solution to the need is within their means. If need satisfaction is beyond a consumer's economic or temporal resources, for instance, then action is unlikely (Engel et al., 1993; Ennew, 1993).

Need recognition can be triggered by internal or external stimuli. In the former case, one of the consumer's personal needs - hunger, thirst, sex - rises to a threshold level and becomes a drive. In the latter case, a need is aroused by an external stimulus such as advertising. Additionally, changes in one's actual or desired state are likely to create new needs (Kotler, 1997; Wilkie, 1994; Ennew, 1993). For example, the birth of a child may create a need for baby-care products that were not needed before.

2.1.1.2 Information Search After a need has been recognised, the consumer may then engage in a search for potential need satisfiers. Information search, the second stage of the decision-making process, can be defined as the motivated activation of knowledge stored in memory or acquisition of information from the environment (Engel et al., 1993). As this definition indicates, information search can be either internal or external in nature.

In the internal search, the consumers search their memory for information about products that might solve the problem. This information may be based on past experience of the product, information that has been absorbed form past marketing campaigns, or information collected from WOM recommendations. If they cannot retrieve enough information from their memory for a decision, they seek additional information in an external search. The external search may focus on communication with friends and colleagues, comparison of available brands and prices, marketer dominated sources, such as television or press advertisements, and public sources (Engel et al., 1993; Loudon and Della Bitta, 1993; Ennew, 1993; Dibb et al., 1997). According to Bloch et al. (1986), external search can be subdivided into purposeful and ongoing search. The first mode, purposeful search, refers to external search that is driven by an upcoming purchase decision, whereas the second mode, ongoing search, refers to information acquisition that occurs on a relatively regular basis regardless of sporadic purchase needs.

The amount of information that is collected depends on the nature of the decision process. An extensive problem-solving process will usually entail a considerable amount of search. The consumer may consider a number of brands, visit several stores, consult friends, and so on. Information overload, however, may cause problems for the consumer. There is evidence to suggest that consumers cannot cope with too much information and, in fact, tend to make poorer choices when faced with large amounts of information (Keller and Staelin, 1987). At the other extreme is routine-response behaviour. Here the consumer minimises search time and effort by relying on past knowledge of the product or the brand name. Other sources of information are ignored. Information search under a limited problem-solving process falls between these two extremes (Loudon and Della Bitta, 1993; Engel et al., 1993; Ennew, 1993).

A successful information search yields a group of brands that a consumer views as possible alternatives. This group of products is sometimes called the consumer's evoked set (Dibb et al., 1997).

2.1.1.3 Evaluation of Alternatives As the consumer is engaged in search activity, he/she is also actively engaged in information evaluation. At this stage of the decision-making process, the consumer evaluates alternatives to make a choice. Four tasks are involved: the consumer must (1) determine the evaluative criteria to use for judging alternatives, (2) decide which alternatives to consider, (3) assess the performance of considered alternatives, and (4) select and apply a decision rule to make the final choice (Engel et al., 1993).

When evaluating the products in the evoked set, consumers may employ a number of different evaluative criteria in making their decision. These criteria are the characteristics or features that the consumer wants (or does not want). Evaluative criteria will usually vary in their importance or salience. Price, for example, may be a dominant dimension in some decisions and yet rather unimportant in others. The salience of evaluative criteria depends on a host of product, situational and individual factors (Loudon and Della Bitta, 1993; Kotler et al., 1996; Engel et al., 1993).

Consumers must also determine the set of alternatives from which a choice will be made (that is, the evoked set). In some situations, the evoked set will depend on the consumer's ability to recall alternatives from his/her memory. On other occasions, alternatives will be considered if they are recognised at the point of purchase. If consumers lack prior knowledge about choice alternatives , they must then turn to the environment for assistance in forming their evoked set (Engel et al. 1993).

A consumer may also rely on his/her existing knowledge for judging the performance of choice alternatives along salient evaluative criteria. Otherwise, external search will be required to form these judgements. In judging how well an alternative performs, ranges for acceptable values ('cut-offs'), that a consumer imposes for evaluative criteria will strongly determine whether a given alternative is perceived as acceptable. Additionally, judgements about choice alternatives can depend on the presence of certain cues or signals. Such is the case when price is used to infer product quality (Engel et al., 1993).

Finally, the procedures and strategies used for making the final selection from the choice of alternatives are called decision rules. These rules may be stored in memory and retrieved when needed. Alternatively, consumers may built constructive decision rules to fit situational contingencies (Engel et al., 1993). Decision rules vary considerably in their complexity. They can be very simplistic (for example, buy what I bought last time) but they can also be quite complex, such as when the rule resembles a multi-attribute model. Another way to differentiate among decision rules is to divide them into compensatory and non-compensatory ones. Non-compensatory rules, such as conjunctive, lexicographic, and elimination-by-aspects, do not permit product strengths to offset product weaknesses. Compensatory rules, in contrast, do allow product weaknesses to be compensated by product strengths (Solomon, 1993; Loudon and Della Bitta, 1993; Assael, 1992; Engel et al., 1993; Wilkie, 1994).

2.1.1.4 Purchase The outcome of the alternative evaluation stage is an intention to buy (or not to buy). The fourth sequence in the decision-making process involves purchasing the intended product. In general, this will be the product which has the most satisfactory performance in relation to the evaluative criteria (Assael, 1992; Ennew, 1993).

As long as the consumer's circumstances or the circumstances in the marketplace remain stable, the decision to purchase will lead to an actual purchase (Kotler et al., 1996; Ennew, 1993). However, in executing a purchase intention, the consumer may make up to five purchase sub-decisions or instrumental actions, i.e. brand decisions, vendor decisions, quantity decisions, timing decisions and payment-method decisions (Kotler, 1997; Assael, 1992). The extent of instrumental action is likely to vary for different degrees of complexity of the decision-making process (Assael, 1992). For example, buying salt gives little thought to the vendor or payment method.

2.1.1.5 Post-Purchase Evaluation The consumer decision-making process does not end when a purchase has been made. Once the product is purchased, the will evaluate its performance in the process of consumption. The outcome is one of satisfaction or dissatisfaction. Whether the consumer is satisfied or dissatisfied depends on the relationship between the consumer's expectations and the product's perceived performance. If the product exceeds expectations, the consumer is delighted; if it meets expectations, the consumer is satisfied; if it falls short of expectations, the consumer is dissatisfied (Kotler, 1997; Kotler et al., 1996). These feelings determine whether consumers make a complaint, purchase the product again or talk favourably or unfavourably about the product to others (Dibb et al., 1997).

Shortly after the purchase of an expensive product, the post-purchase evaluation may result in cognitive dissonance. This, in very simple terms, can be understood as doubts that occur because the consumer questions whether the right decision was made in purchasing the product. Since such psychological discomfort is not pleasant, the consumer will be motivated to act to reduce the amount of dissonance he/she is experiencing. Thus, a consumer may attempt to return the product or may seek positive information about it to justify the choice. An important role of marketing, therefore, is reminding the consumers that they have made the correct decision (Loudon and Della Bitta, 1993; Assael, 1992; Brassington and Pettitt, 1997; Foxall, 1997; Dibb et al., 1997; Ennew, 1993).

2.1.2 Individual Influences The manner in which the individual consumer influences the decision-making process is central to an understanding of consumer behaviour. Following Kotler (1997), these influences can be broadly categorised into psychological and personal factors.

Psychological factors operating within individuals partly determine people's general behaviour and thus influence their behaviour as consumers. The primary influences on consumer behaviour are (1) personality and self-concept, (2) motivation, (3) learning, (4) perception and (5) the impact of attitudes.

Personality and self-concept provide the consumer with a central theme. That is, they provide a structure for the individual so that a consistent pattern of behaviour can be developed (Kotler et al., 1996; Brassington and Pettitt, 1997; Kotler, 1997; Dibb et al., 1997).

Motives are internal factors that energise behaviour and provide guidance to direct the activated behaviour. They will affect which needs a consumer regards as important and therefore the priority in which they should be satisfied. Maslow's theory of motivation, for example, suggests that needs are arranged in a hierarchy, from the most pressing to the least pressing. According to this theory, a consumer would seek to satisfy lower needs (e.g. physiological needs) before progressing to higher needs such as self-esteem or status (Feldman, 1989).

Most human behaviour is learned. Consequently, what consumers learn, how they learn and what factors govern the retention of learned material in memory are all issues of considerable importance for understanding consumers. Not only do consumers acquire and remember product names and characteristics, but they also learn standards for judging products, places to shop, problem-solving abilities, behaviour patterns and tastes. Such learned material, stored in memory, significantly influences how a consumer reacts to each situation that he/she faces (Engel et al., 1993; Wilkie, 1994).

Perception represents the process of selecting, organising and interpreting information inputs to produce meaning. Information inputs are the sensations received through the senses, i.e. sight, taste, hearing, smell and touch. However, each consumer receives, organises and interprets this sensory information in an individual way. Consequently, three perceptual processes can be distinguished: selective attention, selective distortion and selective retention. Selective attention refers to the selection of inputs that people expose to their awareness. Selective, on the other hand, is the changing and twisting of currently received information. Finally, selective retention is the process of remembering information inputs that support personal feelings and beliefs and of forgetting those that do not (Dibb et el., 1997; Kotler, 1997; Brassington and Pettitt, 1997).

Attitudes guide a consumer's basic orientation toward objects, people, events and his/her activities. As such, attitudes strongly influence how consumers will act and react to products and services, and how well they will respond to communications that marketers develop to convince them to purchase their products (Kotler, 1997; Dibb et al., 1997).

As indicated earlier, there is another category of individual factors influencing consumer decision-making: personal factors. These personal factors include demographic and situational variables.

Demographic variables are individual characteristics such as sex, age, race, ethnic origin, income, family life cycle and occupation. A consumer's income, for example, determines his/her spending power and therefore influences whether it is possible for him/her to satisfy a particular need.

Situational factors, are the external circumstances or conditions that exist when a consumer is making a purchase decision. For example, the amount of time a consumer has available for decision-making is a situational variable that strongly influences consumer decisions. A consumer may therefore quickly decide to buy a readily available brand if there is little time available for selecting and purchasing a product (Dibb et al., 1997; Kotler, 1997).

2.1.3 Environmental Influences Consumers are not isolated units but are members of a society, interacting with others and being influenced by them. These social attachments include culture, social class and reference groups.

2.1.3.1 Culture Culture has the broadest of all environmental influences on consumer behaviour. It is "... the values, norms, and customs that an individual learns form society and that leads to common patterns of behavior within that society." (Assael, 1992, p. 319). As this definition indicates, culture includes both material and abstract elements. In a consumer behaviour context, artifacts of the material culture would include products and services, supermarkets and advertisements. Abstract elements would include values, attitudes and ideas (Engel et al., 1992).

Consumption choices cannot be understood without considering the cultural context in which they are made. Consumer goods, for example, have a significant ability to carry and communicate meaning. This, basically, occurs through a process in which cultural meaning is drawn from a particular cultural world and is transferred to a consumer good through advertising and the fashion system and then from these goods into the life of the individual consumer through certain consumption rituals (Kluckhohn, 1951; quoted by Loudon and Della Bitta, 1993).

Culture also mandates the success or failure of specific products and services. A product that provides benefits consistent with those desired by members of a culture has a much better chance of attaining acceptance in the marketplace (Solomon, 1993). A culture can be divided into sub-cultures based on age, geographic regions or ethnic identity. Within these, there are even greater similarities in people's attitudes, values and actions than within the broader culture (O'Shaughnessy, 1995; Dibb et al., 1997).

Finally, the prevailing culture will also determine how consumers react to certain aspects of the marketing mix (Ennew, 1993; Dibb et al., 1997).

2.1.3.1 Social Class Within every society, people rank others into higher or lower positions of respect. This rankings results in social classes. A social class is a social category, usually defined by its members having roughly equivalent socio-economic status. Typically, occupation and income serve to distinguish social classes but some researchers stress other factors such as education, lifestyle, prestige or values as better descriptive measures (O'Shaugnessy, 1995).

Social classes show distinct product and brand preferences in many areas, including leisure activities, clothing and cars. Some products may even be considered as status symbols which serve to associate a consumer with a particular social class (Kotler, 1997; Ennew, 1993).

2.3.1.2 Reference Groups Consumers do not behave as isolated individuals. They belong to various groups. Traditionally, a group is referred to as "... a set of two or more individuals who are in reciprocal communication or associate with each other for some purpose." (O'Shaughnessy, 1995, p. 128) Two generic types of groups can be identified: primary and secondary groups. Primary groups include the family, friends, or working colleagues and involve an individual in direct and frequent interaction with other members. Secondary groups, on the other hand, are groups which tend to be more formal and require less continuous interaction, e.g. a political party (Chisnall, 1985). In the consumer behaviour literature, the groups of interest are reference groups and the family.

Reference groups. A group becomes a reference group when an individual identifies with it so much that he/she takes on many of the values, attitudes or behaviour of group members (Dibb et al., 1997). Most people have several reference groups, such as friends, families, colleagues, religious and professional organisations.

The consumer need not be a member of the group since some groups are those to which the consumer aspires to join (aspirational group). For example, a young junior manager might aspire to the middle management ranks. A group can also be a negative reference group for an individual (dissociative group). Such a group is one whose values or behaviour an individual rejects.

A reference group may serve as a point of comparison and a source of information for an individual. A consumer's behaviour may change to be more in line with the actions and beliefs of group members. Generally, the more conspicuous a product, the more likely it is that the brand decision will be influenced by reference groups (Dibb et al., 1997; Kotler, 1997). An individual may also seek information from the reference group about other factors regarding a prospective purchase, such as where to buy a particular product. The degree to which a reference group will affect a purchase decision depends on an individual's susceptibility to its influence and the strength of his/her involvement with the group (Dibb et al., 1997, Kotler, 1997). Reference groups can on occasions influence what product is bought but not the brand while, on other occasions, they can influence the brand bought but not the decision to buy the particular product (Assael, 1992; Bearden and Etzel, 1982)). Conformity to group norms is motivated both by social conformity and informational conformity. Social conformity arises from the desire for acceptance. It is expressed by an individual's desire to harmonise his/her relations with others. Social conformity is more common if the purchase is socially visible like a car or if the purchase has relevance for the consumer's reference group such as clothing. Informational conformity, on the other hand, results from the desire to make sense of the world around. In other words, while social (normative) conformity to group norms emanates from the desire to be accepted, informational conformity is a way of seeking a more accurate view of reality. Much conformity, however, involves informational as well as normative influence (O'Shaughnessy, 1995).

In most reference groups, one or more members stand out as opinion leaders. Marketers trying to use reference group influence, therefore, attempt to reach and influence the opinion leaders in the reference group of their target customers. Generally, an opinion leader provides information about a specific sphere that interests reference group members who seek information. Opinion leaders are viewed as being well informed about a particular area. However, they are not necessarily the foremost authorities on all issues (Dibb et al., 1997; Kotler, 1997).

Family. Family members constitute the most influential primary reference group. The needs of the family affect what can be afforded, where the spending priorities lie and how a purchase decision is made. All of this evolves as the family matures and moves through the various stages of its life-cycle. Over time, the structure of a family changes. For example, children grow older and eventually leave home, or 'events' break up families or create new ones (Brassington and Pettitt, 1997).

Regardless of the structure of the family unit, members of a household can participate in each other's purchasing decision-making. In some cases, family members may be making decisions that affect the whole family and thus a family can act as a decision-making unit where individual members play different roles in reaching the final decision (Kotler, 1997; Chisnall, 1985; Dibb et al., 1997). In this context, a study by Davis and Rigaux (1974) found that the roles and relative influence of the husband, wife and children in the purchase varies between certain products categories. Husband dominance was identified for automobiles and liquor whereas wives tended to dominate decisions for food, toiletries and small appliances. Joint decision-making was found to be likely for housing, vacations and furniture.

Clearly, groups of all kind have the potential to act as both facilitators and inhibitors of consumer behaviour. For each purchase, the individual has to decide which group's influence is the strongest or most important and act accordingly. A key issues, in this respect, is WOM communication as it represents the way in which members of reference groups influence each other. (Assael, 1992)

2.2 Word-of-Mouth Communication
2.2.1 Definition Arndt defines WOM as "... oral person-to-person communication between a receiver and a communicator whom the receiver perceives as non-commercial, regarding a brand, product or service." (1967, p. 291) However, it is important to point out that WOM need not necessarily be brand, product or service-focused. It may also be organisation-focused. Neither need WOM be face-to-face, direct, oral or ephemeral. The electronic community, for example, generates virtual WOM which is not face-to-face, not direct, not oral, and not ephemeral (Buttle, 1998).

2.2.2 Scope and Significance One of the most widely accepted notions in consumer behaviour is that WOM plays an important role in shaping consumers' attitudes and behaviours. In an early study, Whyte (1954) investigated the diffusion of air conditioners in a Philadelphia suburb. He concluded, on the basis of anecdotal evidence, that the pattern of ownership could be explained only be the presence of a vast and powerful network consisting of neighbours exchanging product information. In a more formal study, Katz and Lazarsfeld (1955) found that WOM was the most important source of influence in the purchase of household goods and food products. It was twice effective as radio advertising, four times as effective as personal selling, and seven times as effective as newspapers and magazines.

Subsequent investigations of the WOM phenomenon have confirmed the dominance of personal influence in choice decisions. Engel et al. (1969), for example, found that almost 60 percent of consumers cited WOM as the most influential factor regarding their adoption of an automotive diagnostic centre. Similarly, Arndt (1967) showed that respondents who received positive WOM about a new food product were three times more likely to purchase it as those who received negative WOM. More recent research is provided by Herr et al. (1991). They observed that WOM communication had a much stronger impact on brand evaluations than information from neutral sources such as the 'Consumer Reports' magazine.

The power of WOM communication stems from various factors. First, consumer recommendations are usually perceived as being more credible and trustworthy than commercial sources of information (Day, 1971). It is common to assume that another consumer has no commercially motivated reasons for sharing information (Engel et al., 1993). Also the discussions with either friends or family tend to be friendly and can offer support for trying certain behaviours. Second, the WOM channel is immediately bi-directional and interactive which allows for a 'tailored' flow of information to the information seeker (Gilly et al., 1998). The third strength of consumer WOM comes from its 'vicarious trial' attributes. Potential consumers of a product, for example, can gain some of the product experience by asking somebody who has an actual experience with the product.

WOM is of particular importance to the services sector. The typical characteristics of services such as intangibility, simultaneous production and consumption, perishability, heterogeneity and the need for the consumer participation results in the fact that suppliers are not able to present the product in advance of the purchase (Helm and Schlei, 1998; Zeithaml and Bitner, 1996). Services, therefore, are high in experience and credence properties which the consumer can only ascertain after purchase and use (Zeithaml and Bitner, 1996). As a consequence, consumers of services rely to a large extent on personal communication and the exchange of experiences with other customers since their experiences of serve as a 'vicarious trial' (Engel et al., 1993). Empirical support for the importance of WOM when purchasing services is provided by Murray (1991) who found that services consumers prefer to seek information from family, friends and peers rather than sponsored promotional sources.
2.2.3 Characteristics of WOM According to Buttle (1998), WOM can be characterised by valence, focus, timing, solicitation, and intervention.

Valence. From a marketing point of view, WOM can be negative as well as positive. In the case of negative WOM, consumers convey information on poor performance, lack of service, high prices or rude sales personnel. Positive WOM is the mirror image. Assael (1992) notes that dissatisfied consumers complain to approximately three times as many friends and relatives as when they are satisfied. Additionally, Mizerski (1982) indicates that a consumer is more likely to pay attention to negative than to positive information. A study by Heath (1996), however, shows that people do not display a simple preference for bad news. Instead, they pass along information that matches the emotional valence of the conversation topic.

According to File et al. (1994), valence and volume of post-purchase WOM can be influenced by management policy. More specifically, they cite work that provides evidence for the contention that the handling of the complaints process, services recovery programmes and unconditional service guarantees influence the frequency and direction of WOM. Richins (1983), for example, shows that if complaints are encouraged, the retailer has the chance to remedy legitimate complaints and win back a customer who may also make positive reports to others, enhancing goodwill.

Focus. WOM activity is not only limited to consumers. In fact, the extent of WOM activity can be seen as a function of the following: the people with whom the company and its employees come into contact (customers, suppliers, agents, competitors, the general public, and other stakeholders); its communications; and the inherent interest in the company as a result of its actions (Haywood, 1989). Similarly, the S.C.O.P.E. model (suppliers, customers, owners or investors, partners and employees) of relationship marketing indicates that WOM is not only restricted to consumers. WOM, for example, is an important source of information in the recruitment of employees (Buttle, 1998). However, the majority of management writings on WOM is that of the satisfied customer communication with a prospect. The assumption is that positive WOM draws customers on the loyalty ladder (figure 2), thereby converting a prospect into a customer (Christopher et al., 1991 quoted in Buttle, 1998).

Prospect Customer Client Supporter Advocate Partner

Figure 2: The loyalty ladder [adapted from: Buttle, 1998, p. 101]

Timing. WOM may be uttered at different stages of the decision-making process; i.e. before or after a purchase. WOM that operates as an important source of pre-purchase information is referred to as input WOM. Output WOM, on the other hand, is uttered after the purchase or the consumption experience (Buttle, 1998).

Solicitation. Not all WOM communication is customer-initiated. WOM may be offered with or without solicitation; it may be offered even though it is not sought. If authoritative information is thought, however, the consumer may see the input of an influential or opinion leader (Buttle, 1998).

Intervention. The power of WOM has not gone unnoticed. An increasing number of companies are proactively intervening in an effort to stimulate and manage WOM activity. Some even consider customer WOM as the most effective marketing tool and also the one with the lowest cost (Wilson, 1994). Specifically, marketers seek to influence opinion leaders directly, stimulate WOM communication in advertising, simulate WOM communication through advertising and/or portray communications form opinion leaders. Additionally, marketers try to curb, channel and control negative communications (Assael, 1992, Engel et al., 1993, Buttle, 1998; Haywood, 1989).

2.2.4 The Nature of Word-of-Mouth
2.2.4.1 Types Richins and Root-Shaffer (1987), in a study on personal influence in buying cars, identified three basic types of WOM communications: product news, advice giving and personal experience. Product news is information about the product such as features or performance attributes. Advice giving relates to expressions of opinions about a product. Personal experience involves comments about product attributes or reasons for buying the product. These categorisation implies that WOM serves two functions, to inform and to influence. Whereas product news informs consumers, advice and personal experience are likely to influence consumer decisions. This, in turn, suggests that each of these types of communication is probably most important in different stages of the decision-making process. Product news, for example, is important in creating awareness about a product and its features. Hearing about product experiences from a friend or relative support the consumer in the evaluation of the relative merits of one brand or another. Finally, through the opinion of 'relevant others', advice giving is important in making the purchase decision stage.

2.2.4.2 Process Until the 1940s, marketers assumed that communication was a one-way process flowing from the marketer to consumers. This view, however, was challenged by Lazarsfeld et al. (1948). Their study of voting behaviour indicated that mass media messages were intercepted and distributed by so-called opinion leaders. This two-step flow hypothesis suggests that marketer-controlled mediated communication flows to opinion leaders who in turn communicate it through WOM to their peers and thereby exercising influence. In this theory, opinion leaders are distributed in all levels and groupings of society and may be influential on just one of several topics (Buttle, 1998). A study by Katz and Lazarsfeld (1955), for example, found that opinion leadership is product specific. They profiled different attributes for fashion opinion leaders, food opinion leaders, public affairs opinion leaders and movie-going opinion leaders. However, a study by Rogers (1962) claimed to have identified three traits which broadly define an opinion leader: social status, social participation and cosmopolitanism. Robertson, (1971) on the other hand, found that opinion leaders tend to be more gregarious, more innovative and knowledgeable than followers. As far as cosmopolitanism is concerned, Robertson (1971) could not distinguish followers from opinion leaders. A set of demographic characteristics of influentials [Engel et al. (1993) prefer to talk of influentials rather than opinion leaders] is provided by Engel et al. (1993). They note that influentials are active information searchers, more innovative, more gregarious, more socially active, more fashion conscious and independent.

The two-step flow hypothesis and the associated concept of opinion leadership have been criticised on a number of grounds. First, the follower is not passive. He/she may well request information as well as listen to unsolicited opinions of others. Second, those who transmit information are also likely to receive it; that is opinion leaders are also followers and vice versa. Third, the opinion leader is not the only one to receive information from the mass media. Followers are also influenced by advertising (Assael,1992; Engel et al., 1993). Katz and Lazarsfeld (1955) also realised that there may be a 'gatekeeper'. This gatekeeper may be distinct from the opinion leader; he/she may introduce ideas and information to the group but may not influence it. Because of these limitations in the concept of a two-step flow, a multi-step flow model (figure 3) of WOM communication came into broader acceptance.

Gatekeepers
Mass Media Opinion Leaders Followers

Figure 3: Multi-step flow model [Source: Assael, 1992, p. 430]

In this model the mass media can reach the gatekeeper, opinion leader, or follower directly but are less likely to reach the follower (indicated by the dotted line). The gatekeeper represents a source of information to both opinion leaders and followers. However, the dissemination of information to opinion leaders is more likely. Furthermore, WOM communication between opinion leaders and followers is represented as a bi-directional flow, as opinion leaders may seek information from followers, and followers may solicit information form opinion leaders (Assael, 1992; Engel et al., 1993).

The recognition in the multi-step model that opinion leaders and followers both may transmit and receive information leads to four possibilities (Reynolds and Darden, 1971).

Figure 4: A categorisation of consumers by opinion leadership and information seeking for clothing decision [Source: Reynolds and Darden, 1971, p. 451]

Consumers who score high on both opinion leadership and information seeking are classed as socially integrated consumers. Those who score high on influencing others, but low on being influenced themselves, are classed as socially independent consumers. Socially dependent consumers, on the other hand, are those who score low on influencing other but high on being influenced by others. Finally, socially isolated consumer are those who score low on both opinion leadership and information seeking (Reynolds and Darden, 1971).

More recently, the concepts of the 'market maven' emerged. Market mavens are defined as individuals having information about many kinds of products, places to shop, and other facets of markets, who initiate discussions with consumers and respond to their request for market information (Feick and Price, 1987). Other than the finding that market mavens are morel likely to be women, there is no clear demographic or psychological profile identified yet for these influencers (Slama and Williams, 1990; Higie et al., 1987; Bayus et al., 1985). Gelb and Johnson (1995; quoted by Buttle, 1998) note that not only does the market maven prompt WOM, but those with links to such individuals are disproportionately likely to act on what they are told.

2.2.4.3 Conditions Although WOM is an important factor in consumer decision-making, it is not the dominant factor in every situation. A study by Herr et al. (1991) shows that WOM is not as important in the evaluation of a car if the consumer already has a strong impression of the product and/or negative information regarding the product is available. WOM, therefore, is unlikely to change attitudes of consumers who have strong brand attitudes. Herr et al.'s (1991) findings also suggest that WOM is unlikely to change a consumer's attitudes if the consumer has doubts about a product because of credible negative information.

The influence of WOM also varies across product categories. According to Assael (1992), it is most important when reference groups are likely to be sources of information and influence. Two cases are of particular importance: First, the consumer is involved in the purchase decision and second, the purchase of a risky product. Consumers who are involved with a product are more likely to communicate about it and influence others, especially if they are involved on an ongoing basis (enduring involvement). Empirical evidence comes from a study by Richins and Root-Shaffer (1987). They found that individuals with enduring involvement are most likely to be opinion leaders. Individuals who were involved on a situational basis only were unlikely to influence others, although they could inform friends and relatives about new products or product attributes. Consumers are also more likely to initiate product-related conversations and to request information from friends and relatives if they see risk in the purchase (Cunningham, 1966).

2.2.4.3 Motives There are several motives for engaging in WOM communication. Being involved in a decision, as mentioned before, is likely to encourage consumers to transmit information and influence. Katz and Lazarsfeld (1955) provide evidence that those most likely to transmit information are not those with experience but those who are experiencing the product. Situational involvement, or involvement in the product decision is one important ingredient in personal communications. Another motive for WOM communication is the inherent interest in the product category (enduring involvement). Individuals who have an ongoing interest in a product category get enjoyment in talking about it (Assael, 1992). Moreover, WOM communication is sometimes initiated to erase any doubt about product choice. According to cognitive dissonance theory, a consumer may attempt to reduce discomfort by describing the positive qualities of a recently purchased product to friends and relatives. Ideally, a purchase of the same product by a friend or relative confirms the consumer's original judgement. Another reason for WOM communication is involvement with a group. Talking about products may simple be a mean of social interaction as Dichter (1966) suggests. The greater the importance of the group for the individual, the greater the likelihood that the he/she will seek to transmit information to it. WOM communications can also be initiated by a consumers desire to be influential. Talking about the product and thus influencing people may give them personal satisfaction (Dichter, 1966).

2.2.5 WOM and the Consumer Behaviour Literature
2.2.5.1 Adoption and Diffusion of Innovations The diffusion process refers to a group phenomenon, indicating how an innovation spreads among consumers. The diffusion process, of course, necessarily involves the adoption process of many individuals over time. The adoption process of a new product is an individual phenomenon relating to the sequence of stages through which an individual passes from first hearing about a product to finally adopting it.

Awareness Comprehension Attitude Legitimation Trial Adoption

Figure 5: The adoption process [Source: Assael, 1992, p. 272]

When a product is first introduced, communication from the marketer to the consumer is meant to create awareness of the innovation and provide information. Engel et al. (1969), in a study of early users of a new automobile diagnostic centre, found that early adopters of the service relied primarily on magazines and radio for information. Similarly, Sheth (1971) shows that the principal source of information for early adopters of stainless steel blades were mass media. However, once awareness was created, consumers relied more heavily on friends and relatives to help them evaluate new products. Both studies concluded that the greatest influence on a consumer's decision to adopt innovations came from friends and relatives. Consequently, if a new product is to be diffused across groups, it must first be adopted through positive WOM communication within groups (Assael, 1992).

The diffusion of innovations, however, also requires the spread of information across different groups. This diffusion of information happens because consumers spread WOM by interaction with individuals outside their personal network. In this context, Granovetter's (1973) theory on 'the strength of weak-ties' provides an explanation of the process by which WOM behaviour at the micro level (within groups) is linked to the macro level (across groups) phenomena. A consumer's social relations with other relevant actors typically include a spectrum of ties ranging from strong (such as friends) to weak (such as acquaintances). Granovetter (1973) claims that weak-ties play a crucial role in clarifying and explaining the diffusion of innovations. For WOM referrals this 'strengths of weak-ties' arises from their important bridging function that allows information to travel form one densely knit 'clump' of social structure composed of referral actors to another more cohesive segment of the broader referral system through weak-tie WOM communication. "If weak ... [-tie WOM communication] did not exist, a system would consist of disjointed subgroups, inhibiting the widespread diffusion of information." (Brown and Reingen, 1987, p. 352).

2.2.5.2 Post-Purchase Decision-Making Research anchored in the satisfaction/dissatisfaction and complaining literature has focused on the selection of WOM as a post-purchase complaint option. In other words, negative WOM is thought to be one form of customer complaining behaviour. Hirschman (1970) proposed that customers can either voice their dissatisfaction or exit the relationship when faced with unmet expectations. More recently, Richins (1983) differentiates between three reactions to dissatisfaction (1) switching brands or refusing to repatronise the offending store, (b) making a complaint to the seller or to a third party, and (c) telling others about the unsatisfactory product or retailer (negative WOM). This tripartite taxonomy of complaint style was also found by Singh (1988) when applying cluster analysis to complaints data. The potential impact of these responses on a firm can be significant. Data reported by Diener and Greyser (1978) indicated that 34 percent of those dissatisfied with a personal care product told others about their dissatisfaction. If the number of consumers experiencing dissatisfaction is high enough, such responses may have lasting effects in terms of negative image and reduced sales for the firm (Richins, 1983).

A study by Richins (1983) indicates that the responses to dissatisfaction are related to the nature of the dissatisfaction problem. In the case of minor dissatisfaction, consumers' responses are often minimal too. Consumers' do not complain nor do they spread negative WOM. When the dissatisfaction is serious enough, consumers tend to complain, regardless of other factors in the situation. It is at moderate levels of dissatisfaction that management policy may have the most impact. If complaints are encouraged (e.g. toll-free telephone numbers to receive customer comments), the retailer has the chance to remedy legitimate complaints and win back a customer who may also spread positive WOM to others. Research by TARP (1979) showed that even if the complaint is not settled to the consumer's satisfaction, he/she is more likely to repurchase than if no complaint is made. However, if complaints are discouraged, fewer consumers may indeed complain; instead, they may tell others of their unsatisfactory experiences and may not repurchase the product. Richins (1983) also showed that negative communication was more likely when the consumer placed the blame for dissatisfaction directly on the manufacturer or retailer and/or when the consumer believed that complaining directly to the source would not do any good.

Watkins and Liu (1996) provide general support for the contention that consumers dissatisfied with durables will exhibit higher levels of voice and lower levels of exit than for non-durables. Singh (1990) sees this phenomenon explained by the relative investment of the consumer in the product and thus the value of any redress.

As already noted, after purchases are made, consumers often engage in a post-purchase evaluation of the product. If the product's perceived performance is below the consumer's expectation, he/she might sense dissonance. Since such psychological discomfort is not pleasant, the consumer will be motivated to act to reduce the amount of dissonance he/she is experiencing. One strategy for customers who experience discomfort from cognitive dissonance is to seek WOM from sources which can reduce the discomfort (Buttle, 1998; Dibb et al., 1997; Assael, 1992).

2.2.5.3 Pre-Purchase Decision-Making WOM has also been studied as a mechanism through which consumers convey both informational and normative influence in the product evaluation and purchase intention of fellow consumers (Tax et al., 1993). This kind of information can be given by the choice of the recommendation source and also by a number of factor related to the product or the task of choosing the product. An empirical exploration of those issues follows.

Chapter 3: Empirical Analysis
3.1 The Model Duhan et al. (1997) developed a model to describe the way certain factors influence the consumers' use of WOM recommendation sources in the decision-making process. The model involved the choices of an either strong-tie or weak-tie recommendation source. It explains the way that prior knowledge of the product or service, perceived decision task difficulty and the type of evaluative cues of the product or service affect that choice.

3.1.1 Recommendation Sources The study of WOM communications involves mostly, personal sources of information (Duhan et al., 1997). Brown and Reingen (1987) suggested the categorisation of WOM recommendation sources according to closeness of the relationship between them and the consumer who makes the decision. Individuals that know the decision maker personally are strong-tie recommendation sources. Acquaintances or individuals that are unknown to the decision maker, on the other hand, are weak-tie recommendation sources. Brown and Reingen (1987) found that strong-tie sources are activated in the flow of influence, while weak-tie sources play a more crucial role in the flow of information across groups. Belk (1971) reports that the use of strong-tie sources as providers of information is stimulated by environmental or situational cues. The influence might be greater, compared to that by weak-tie sources, because the more frequent contact and communication with the strong-tie source provides with the opportunity for more situational cues. Brown and Reingen (1987) argue that this can not be generalised to weak-tie sources because the decision maker actively approaches them for information. Weak-tie sources tend to be approached by the decision maker because of their expertise in a particular area. Strong-tie sources are approached because of the similarity with the decision maker. "Thus, if consumers feel a need for reassurance regarding some personal aspects of the decision, they are likely to seek out strong-tie sources for that kind of information." (Duhan et al., 1997, p. 284)

Duhan et al.'s (1997) model is based on the assumption that once the two types of recommendation sources differ qualitatively form one another, different factors influence the consumers use of them. The three factors they explored were (1) the types of information (evaluative cues) important to the decision, (2) the difficulty of the decision task as perceived by the consumer and (3) the type and level of knowledge the consumer has about the product or service.

Task Difficulty

Prior Knowledge Source Choice

Evaluative Cues

Figure 6: General Model of Recommendation Source Choice [Source: Duhan et al., 1997, p. 285]

3.1.2 Evaluative Cues Duhan et al. (1997) acknowledge the existence of different taxonomies of evaluative cues. The one they employ for the purposes of their model is the one suggested by Ben-Sira (1976), which differentiates between affective cues and instrumental cues.

"The evaluation of affective cues is generally based on subjective criteria established by and related to the purchaser ..., whereas the evaluation of instrumental cues is generally based on characteristics of the product that can be evaluated independently of the purchaser." (Duhan et al., 1997, p. 284) A strong-tie source knows both the purchaser and the product. This way he/she is able to evaluate the aspects of the product that the purchaser is more likely to like. Thus, strong-tie sources are more reliable when it comes to affective cues. Similarly, a purchaser that is more interested in idiosyncratic aspects of a product is more likely to rely on a strong-tie recommendation source. On the other hand, knowledge of the product (or service) is enough when instrumental cues are important. This knowledge remains the same regardless of the purchaser. Once the weak-tie sources are more varied and more numerous, chances are that a weak-tie source is going to be the most appropriate to provide with evaluation of instrumental cues. In this case, strong-tie recommendation sources are less important and less sought after.

3.1.3 Task difficulty According to Newell and Simon (1972) the task environment determines the behaviour of the problem solver. In other words, the attributes of the problem rather than the problem solver's internal representation and processes may account for variations in problem solving behaviour and may provide a basis for relevant predictions. The attribute of the task environment that Duhan et al. (1997) chose to incorporate in their model is task difficulty. Although they acknowledge the variety of definitions of task difficulty, in their theory "... the construct [is being] defined in terms of how 'overwhelming' the decision task is for the decision maker." (Duhan et al., 1997, p. 285) Research shows (Kim and Khoury, 1987; Paquette and Kida, 1988) that decision task difficulty can be evaluated as to the number of product alternatives from which one has to choose or as to the number of attributes of the products from which one has to choose. A small number of options or a small number of attributes makes a task easier because the information load is small (and vice versa) (Jacoby et al., 1974). Decision tasks that are perceived to be difficult are also perceived to involve more risk, which affects the types of information sources that the decision maker seeks (Locander and Herman, 1979). High levels of perceived difficulty correlate with low levels of self-confidence about making the right judgement. This situation motivates people to seek information form people with high levels of perceived similarity in whom they have more confidence (Brown and Reingen, 1987). Thus, as perceived difficulty increases the likelihood to seek information form strong-tie sources increases (Duhan et al., 1997). Moreover, Duhan et al. (1997) refer to research which indicates that consumers tend to consider cues that they feel confident, they can evaluate and that when task difficulty increases, decision makers tend to minimise the cognitive effort in the decision. They conclude that when a decision maker is faced with a difficult task he/she will search less and he/she will rely on easily processed cues, i.e. affective cues.

3.1.4 Prior Knowledge Prior knowledge can be defined as the extend of experience and familiarity with a product (good or service) (Duhan et al., 1997). It refers to information that is accessible from memory and that is being accessed before external search for information starts.

The higher the levels of prior knowledge about a product, the more developed the schema of the product is (Marks and Olson, 1981). Also, as prior knowledge increases, the ability and the efficiency to process new information increases (Johnson and Russo, 1984). Prior knowledge can be conceptualised in terms of experience or as being subjective or objective (Duhan et al., 1997).

Experience-based knowledge can be defined as familiarity with products. Familiarity may result from search experience, usage experience, or ownership (Park and Lessing, 1981), or it can be defined experientially as product-related experiences of the consumer with the product (Alba and Hutchinson, 1987). Duhan et al., (1997) preferred to restrict their definition of prior knowledge to actual purchase or usage of the product.

Subjective prior knowledge is self-assessed by the consumer. As a measure it is liable to individual biases of the consumer who may not be objective in comparing him/her self to others or he/she might be under- or over-confident of their actual knowledge level (Duhan et al., 1997). Although it is subject to uncontrollable factors, it is a valuable measure because perception of one's knowledge affects the choices of information sources (Duhan et al., 1997).

Objective prior knowledge is based on the content of knowledge and it can be observed and measured more reliably (Spreng and Olsharsky, 1989). The importance of objective knowledge and the process by which it is acquired is interesting for research purposes because it is specific to the knowledge domain in which different theories are being applied (Duhan et al., 1997).

Prior knowledge, either objective or subjective increases the likelihood for seeking strong-tie sources indirectly through perceived task difficulty (Duhan et al., 1997). Individuals with high levels of prior knowledge have more developed schemata, which are facilitative for the decision-making process. Moreover, they engage in the decision-making process with less cognitive effort (Alba and Hutchinson, 1987). According Brucks and Schurr (1990), low levels of prior knowledge account for the decision task's perceived difficulty. This may account for the expectation that the relationship between prior knowledge (objective or subjective) and task difficulty is negative (Duhan et al., 1997).

As far as experience-based prior knowledge is concerned, it "... should be conceptualized as related to, but different from, subjective or objective knowledge in the way it influences consumer decision making." (Duhan et al., 1997, p. 287). Duhan et al. (1997) see experience as the basis for the development of both objective and subjective prior knowledge that influences the decision-making process indirectly through them. Experience is not retrieved as such at the time of the decision-making task. According to Hoch and Deighton (1989), it is being integrated into pre-existing beliefs. This integration results in new modified beliefs that are retrieved for the evaluation of information relevant to the decision-making task. Based on research by Alba and Hutchinson (1987) and on the observation that individuals tend to believe that they get knowledge through their experiences, Duhan et al., (1997) hypothesise that both objective and subjective knowledge must be positively related with experience.

Finally, Marks and Olson (1990) suggest that consumers with high levels of subjective prior knowledge have more faith in their ability to evaluate instrumental cues and can be expected to rely on them more than consumers with low levels of subjective prior knowledge.

3.1.5 Hypotheses Based on the theoretical considerations mentioned above, Duhan et al. (1997) developed a model to describe direct and indirect influences of prior knowledge, task-difficulty and evaluative cues on the choice of recommendation source. They tested their model in the context of the professional services market and more specifically the medical services. Their choice was based on the fact that first of all, this is a context in which WOM recommendations play an important role, second that is familiar to the average consumer and third, that has been the focus of a number of studies that contribute to a considerable body of literature.

Consumer Mediating Variables Recommendation
Prior Knowledge Source Type Level

Task Difficulty

Strong-tie Objective Affective Cues
Experience Weak-tie Subjective Instrumental Cues

Figure 7: Theoretical Model of Recommendation Source Choice [Source: Duhan et al., 1997, p. 286]

From the model that is schematically presented in figure 7, Duhan et al. (1997) explored nine relationships. More specifically, they tested the following nine hypotheses:

H1: The greater the importance of affective evaluative cues, the greater the likelihood that strong-tie sources will be sought for recommendation.

H2: The greater the level of perceived task difficulty, the greater the likelihood that strong-tie sources will be sought for recommendation.

H3: The greater the level of perceived task difficulty, the greater the importance of affective cues in the decision process.

H4: The greater the level of objective prior knowledge, the lower the perceived level of task difficulty.

H5: The greater the level of subjective prior knowledge, the lower the perceived level of task difficulty.

H6: Experience is positively related to objective prior knowledge.

H7: Experience is positively related to subjective prior knowledge.

H8: The greater the importance of instrumental evaluative cues, the greater the likelihood that weak-tie sources will be sought for recommendations.

H9: Subjective prior knowledge is positively related to the use of instrumental cues.

Duhan et al.'s (1997) results, first of all, confirmed previous findings, that there are different influences on the consumer's choices of different types of recommendation sources. The choice of strong-tie sources was influenced by task difficulty and prior knowledge while the choice of weak-tie sources was influenced by the importance of instrumental cues and subjective prior knowledge. Moreover, they found two indirect effects. First, there was an indirect effect of objective and subjective prior knowledge on strong-tie sources, while there was an indirect effect of subjective knowledge and experience on weak-tie sources. Another two hypotheses that were supported by the results was that high levels of experience were positively related to objective and subjective prior knowledge. Finally, three of their hypotheses were not confirmed by the results. Subjective and objective prior knowledge were found to relate to perceived task difficulty, but in both cases the relationships was of the opposite direction. Also the importance of affective cues did not relate to strong-tie recommendation sources.

3.2 Research Context The aim of the present research was to further investigate the model proposed by Duhan et al. (1997). An additional factor was introduced and a different research context was employed.

It has been found that risk avoidance is an important factor in making purchase decisions (Assael, 1992). According to Cunningham (1966), consumers who see risk in the purchase are more likely to initiate product-related conversations and to request information from friends and relatives. In addition to the nine hypotheses proposed by Duhan et al. (1997), a tenth hypothesis is going to be tested. This is:

H10: The greater a consumer's level of risk aversion, the greater the likelihood that strong-tie sources will be sought for recommendation.

As far as the context decision is concerned, tourism services are common enough to ensure a reasonable test of the theory. Choosing a tourism product is a decision task that was both likely to involve recommendations and to be common enough that the typical consumer would see it as realistic and familiar.

Furthermore, research has consistently found that recommendations play a major role in the selection of tourism products (Morrison et al., 1996; Haywood, 1989) In fact, experiences with popular services such as travel and food seem to generate significantly more interpersonal communication than do experiences with more mundane products. Consumer experiences with tourism products are generally considered "hot topics." People are in the habit of telling others about their experiences with travel destinations, new restaurants, and so on (Haywood, 1989). Moreover, resulting from the increasing importance of services in our economy, there is a body of literature on services to use as a guidance for measurement and other methodological issues.

To perform the study, the large array of services within the tourism industry needed to be narrowed down. Package holidays were chosen, since a high proportion of consumers have had experience with choosing a package holiday. In the UK, such holidays are formed by the packaging together by tour operators of flights, accommodations and intangible services. These are represented as a single product in a brochure, and are generally sold via a travel agent to the consumer (Gilbert and Soni, 1991).

3.3 Method
3.3.1 Research Instrument The research instrument used was a self-administered survey questionnaire (see appendix B). It was pre-tested with a number of number of participants to check for wording and clarity. The research context was operationalised through the use of the scenario method. This approach offers the advantage of avoiding situational effects associated with the selection of a service provider. Additionally, scenaria enable the researcher to more consistently portray decision situations. Two scenaria were created. A 'easy' scenario, offering two options of package holiday with one attribute each. The 'difficult' scenario offered a choice between four different package holidays with varying attributes.

The questionnaire had six sections consisting of closed-end questions and open-end questions.

Recommendation Sources The development of items for measuring the likelihood of using various recommendation sources was based on measures reported in the literature (Duhan et al., 1997; Morrison et al., 1996) and informal talks with travel agents and consumers of the product. The sources typically consulted for recommendations were friends, neighbours, relatives, colleagues, the internet, guidebooks, travel agents and tour operators. Sources identified by this process were similar to those found in other research (Duhan et al.,1997; Morrison et al., 1996). The likelihood of use for each source was measured on a 5-point-likert scale ranging from very likely to very unlikely. Friends, neighbours, relatives and colleagues were considered strong-tie sources; the internet, guidebooks, travel agents and tour operators were considered weak-tie sources.

Task difficulty The level of task difficulty perceived by the respondent after reading a scenario was measured on a 5-point-likert scale ranging from very difficult to not difficult at all (higher values indicating lower difficulty).

Cue Characteristics The findings from the exploratory research and previous empirical studies (Duhan et al., 1997; Morrison et al., 1996; Gilbert and Soni, 1991) were incorporated to develop items for the measurement of the relative importance of affective and instrumental cues. Affective cues were measured with items that scaled the importance of safety at the destination, familiarity with the destination, similarity of the destination and suitability of the holiday resort. The importance of instrumental cues was similarly measured, with items including whether the package was flexible, was offering high-quality accommodation, provided extra services and had rules for compensation. Again 5-point-likert scales, ranging from 'extremely important' to 'extremely unimportant' were used.

Prior Knowledge Experience has generally been measured as the number of product purchases as well as product-related experiences. Experience with package holidays was measured on the basis of visits to the Caribbean, purchases of package holidays as well as the number of such purchase during the last three years. Subjective prior knowledge has generally been measured with subjective self-assessment knowledge level. Studies by Duhan et al. (1997); Rao and Monroe (1988) and Rao and Seiben (1992) each measured subjective prior knowledge with a subject-reported assessment of prior knowledge on a 5-point-likert scale that ranged from completely unfamiliar to extremely familiar. In this research, subjective prior knowledge was measured using an adaptation of the same measure for self-assessment of knowledge, with high values indicating lower subjective knowledge. A scale of objective prior knowledge within this research context was developed following empirical studies and procedures similar to those used by Rao and Monroe (1988), Rao and Seiben (1992), Park et al. (1994), and Sujan (1985), Duhan et al. (1997). The set of items included in the scale was designed to follow the major knowledge dimensions discussed by Brucks (1985). A review by travel agents confirmed the content validity of the scale. An item analysis and inspection of comments from the pre-test led to some modifications to the scale (°C temperature was added and corrected). The resulting scale consisted of 6 multiple-choice items. Objective prior knowledge was measured as the number of items answered correctly by the subject.

Risk Aversion Finally, risk aversion was measured by four statements that rate on a 5-point-likert scale ranging from 'strongly agree' to 'strongly disagree'.

All the answers of the respondents where coded in order for the statistical analysis to be feasible.

3.3.2 Procedure The survey questionnaire was self-administered and was distributed personally or through the internal mail at the University of Nottingham. The survey took place in September and October 1998. The respondents were debriefed for the objectives of the research and were informed that the returned questionnaires would be treated confidentially. 200 questionnaires were distributed (100 'easy' scenaria and 100 'difficult' scenaria). 70 of those were returned (23 'easy' and 43 'difficult' scenaria), which makes a 35 % response rate.

3.3.3 Sample A total of 200 participants was approached. This was a convenience sample since representative sampling was not possible provided the limited time available for this research. The demographic characteristics of the 70 participants who returned the completed questionnaires are summarised in table 1:

|Demographics |Frequency |Percent |
| | | |
|Age | | |
|Under 16 |0 |0 |
|16-24 |17 |24.3 |
|25-34 |23 |32.9 |
|35-44 |18 |25.7 |
|45-64 |11 |15.9 |
|65+ |0 |0 |
|Missing |1 |1.4 |
|Sex | | |
|Male |35 |50 |
|Female |35 |50 |
|Missing |0 |0 |
|Marital Status | | |
|Single |36 |51.4 |
|Married |21 |30.0 |
|Living together |8 |11.4 |
|Divorced/Separated |3 |4.3 |
|Missing |2 |2.9 |
|Education | | |
|GSCE/O-level |11 |15.7 |
|A-level |1 |1.4 |
|Further education (e.g. HNO) |6 |8.6 |
|University |51 |72.9 |
|Missing |1 |1.4 |
|Income (£) | | |
|Less than 10,000 |16 |22.9 |
|10,000 - 19,999 |17 |24.3 |
|20,000 - 29,999 |12 |17.1 |
|30,000 - 39,999 |11 |15.7 |
|40,000 - 49,999 |7 |10.0 |
|over 50,000 |6 |8.6 |
|Missing |1 |1.4 |
|Occupation | | |
|Students |28 |40.0 |
|Administrative Staff |18 |25.71 |
|Academic Staff |16 |22.86 |
|Other |8 |11.43 |
|Missing |0 |0 |

Table 1(continued):

3.4 Results Subsequent data analysis was undertaken using the statistical package SPSS. Basically, this analysis was structured into three parts. First, the hypothesised relationships were tested. Second, an exploratory factor analysis was conducted. Third, a test of relationships others than the ones hypothesised was carried out.

Test of the hypothesised Relationships In order to test the hypothesised relationships, proposed by the model, a path analysis approach was employed. The aim of this approach is to provide quantitative estimates of the causal connections between sets of variables. The connections proceed in one direction and are viewed as making up distinct paths. In order to provide estimates of the postulated paths, path coefficients are computed. A path coefficient is a standardised regression coefficient. The path coefficients are computed by setting up structural equations, that is, equations which stipulate the structure of the hypothesised relationships in a model (Bryman and Cramer, 1997). In our case , seven structural equations were required:

(1) Objective knowledge = f(prior experience) (2) Subjective knowledge = f(prior experience) (3) Task difficulty = f(objective knowledge, subjective knowledge) (4) Affective cues = f(task difficulty) (5) Instrumental cues = f(subjective knowledge) (6) Strong-tie = f(task difficulty, affective cues, risk aversion) (7) Weak-tie = f(instrumental cues)

In order to compute the path coefficients, it is necessary to treat the seven equations as multiple-regression equations and the resulting standardised regression coefficients provide the path coefficients. Table 2 and figure 8 provide the results of the path analysis.

|Hypothesis |Direction of |Significance |R² |T |Standardised |
| |Relationship |(Sig T) | | |Regression Coefficient|
| | | | | |(Beta) |
|H1 |+ |.0075 |.41574 |2.770 |.303921 |
|H2 |+ |.0000 |.41574 |5.042 |.523969 |
|H3 |+ |.2425 |.02064 |1.179 |.143665 |
|H4 |- |.2866 |.02139 |-1.075 |-.141405 |
|H5 |- |.9141 |.02139 |-.108 |-.014250 |
|H6 |+ |.057 |.10836 |2.853 |.329180 |
|H7 |+ |.1758 |.2845 |1.369 |.168658 |
|H8 |+ |.2611 |.02032 |1.134 |.142556 |
|H9 |- |.4624 |.00847 |-.739 |-.092029 |
|H10 |- |.9631 |.41574 |-.046 |-.005200 |

Table 2:

Task Difficulty -.141407 .523969*

Objective .143665 ..329180 -.014250 .303921*
Experience Affective Cues Strong-tie

.168658 Subjective .142556 Weak-tie -.092029 Instrumental Cues

-.005200 Risk Aversion

Figure 8: Path diagram with path coefficients (* = p < .05)

As table 1 indicates, the relationship between affective cues and strong-tie sources of recommendation was found to be positive (Beta = ..303921) and significant (p < .05). Regarding the effect of task difficulty on strong-tie sources, a positive (Beta = .5239969) and significant (p < .05) relationship was found. Together with risk aversion, affective cues and task difficulty accounted for 41.5 % of the variance in strong-tie.

Hypotheses 3, 6, 7, 8, 9 showed a positive direction of the relationship (Beta = .143665; .3229180; .168658; .142556) but were not significant at the .05 level. Moreover, the R² values (R² = .02064; .10836; .2845; .02032) for these hypotheses were rather disappointing. At most, 28.455 % of the variance in the dependent variable were explained by the independent variable(s).

Hypotheses 4, 5, 9, 10 were found to be negative in their direction (Beta = -.141407; -.014250; -.092029; -.005200) but not significant (p > .05). As far as the R² values (R² = .2866; .9141; .4624; .9231) are concerned, at least 28.66 % and at most 96.31 % of the variance in the dependent variable could be explained by the independent variable(s).
Exploratory Factor Analysis In addition to a test of the hypothesised relationships, an exploratory factor analysis was conducted in order to assess the factorial validity of the questions related to affective cues, instrumental cues, risk aversion, strong-tie and weak-tie sources of recommendation. In other words, it was examined whether the items which made up these five constructs where not measuring something else.

First, an exploratory factor analysis was conducted with the variables AATM, AFAM, ASIM, ASAF, IQUA, IFLE, IRUL and IEXT suggesting that these variables form three factors: factor 1 (IEXT, IFLE, IQUA, IRUL), factor 2 (AFAM, ASAF) and factor 3 (ASIM). Based on these findings, three aggregate variables were computed: EVAL1 (mean of IEXT, IFLE, IQUA, IRUL), EVAL2 (mean of AFAM and ASAF) and EVAL3 (ASIM).

The second factor analysis focused on the variables RMIS, RNEW, RSTI and RNER. The items have similar contribution to the one extracted factor (.65048; .82414; .68086; .57138), suggesting that they represent a single construct (RISK).

The third factor analysis was run with a data set containing AFRI, AREL, ACOLL, ANEI, AGUI, AINT, ATAG and ATOP. Three factors were extracted: factor 1 (AFRI, AREL, ACOLL, ANEI), factor 2 (ATAG, ATOP) and factor 3 (AGUI, ANEI). Based on these findings, three aggregate variables were computed: SOU1 (mean of AFRI, AREL, ACOLL and ANEI), SOU2 (mean of ATAG and ATOP) and SOU3 (mean of AGUI and AINT).

Taking these new factors into consideration, the hypothesised relationships were redefined in term of the factors the have emerged . A new set of linear regression was then used to test these relationships (table 3).

|Relationship |Direction of |Significance (Sig T) |R² |Beta |T |
| |relationship | | | | |
|EVAL1 - SOU1 |+ |.1095 |.04078 |.201951 |1.624 |
|EVAL1 - SOU2 |+ |.2944 |.01772 |.0133105 |1.057 |
|EVAL1 - SOU3 |+ |.4687 |.00836 |.091455 |.729 |
|EVAL2 - SOU2 |+ |.019 |.14457 |.380221 |3.237 |
|EVAL2 - SOU3 |- |.9663 |.00003 |-.005342 |-.042 |
|EVAL2 - SOU1 |+ |.3343 |.01504 |.122648 |.0973 |
|EVAL3 - SOU1 |+ |.0090 |.010497 |.323996 |2.697 |
|EVAL3 - SOU3 |+ |.8544 |.00054 |.023214 |.0184 |
|EVAL3 - SOU2 |+ |.9750 |.00002 |.003997 |.031 |
|Task difficulty - EVAL2 |+ |.3531 |.01308 |.114370 |.935 |
|Task difficulty - EVAL2 |+ |.8762 |.00037 |.019255 |.156 |
|RISK - SOU2 |+ |.0040 |.12778 |.357460 |2.989 |
|RISK - SOU3 |- |.6680 |.00299 |-.054652 |-.431 |
|RISK - SOU1 |+ |.0770 |.05036 |.224402 |1.799 |

Table 3:

Reference to table 3 reveals three significant (p < .05) relationships. First, EVAL2 is positively (Beta = .380221) related with SOU2. According to the R² value, EVAL2 accounts for 14.457 % of the variance in SOU2.

Second, EVAL3 has a positive (Beta = .323996) relationship with SOU1. The R² value for this linear regression is .10497, implicating that EVAL explains 10.497 % of the variance in SOU1.

Third, RISK is positively related with SOU2 (Beta = .357460). RISK accounts for 12,778 % of the variance in SOU2. The remaining relationships were not significant at the .05 level of significance.

Exploratory Research The final step of the analysis focused on relationships which had not been initially described by the model. Table 4 provides a summary of this step:

|Relationship |Direction of |Significance (Sig |R² |Beta |T |
| |relationship |T) | | | |
|Instrumental Cues - |+ |.1095 |.04078 |.201952 |1.624 |
|Strong-tie | | | | | |
|Affective Cues - Weak-tie |+ |.0625 |.05486 |.234214 |1.897 |
|Task-difficulty - Weak-tie |+ |.2646 |.02003 |.141526 |1.126 |
|Task difficulty - Instrumental Cues |+ |.7618 |.00140 |.037436 |.304 |
|Risk aversion -Task difficulty |+ |.0499 |.05877 |.242416 |1.999 |
|Objective - Affective Cues |- |.8064 |.00090 |-.030048 |-.246 |
|Objective - Instrumental Cues |- |.2245 |.02914 |-.148125 |-1.226 |
|Subjective - Affective Cues |- |.3528 |.01350 |-.116209 |-.936 |

Table 4: (continued)

As table 4 indicates, risk aversion had a positive (Beta = .242416) and significant (p < .05) relationship with task difficulty. However, risk aversion accounts for only 5.877 % of the variance in task difficulty. The other relationships did not reach significance at the .05 level.

3.5 Discussion
Hypothesised Relationships The test of the model reveals a number of interesting relationships. Contrary to Duhan et al. (1997), the study found support for the hypothesised relationship between affective evaluative cues and strong-tie sources. Consumers who place greater importance on affective evaluative cues in choosing a holiday are more likely to seek advice from strong-tie sources of recommendations. Support was also found for the hypothesised relationship between task difficulty and strong-tie sources, which is consistent with Duhan et al.'s (1997) findings. Consumers who perceive greater difficulty in the decision task are likely to look toward strong-tie sources of recommendations.

Experience was found to have a positive with both subjective and objective knowledge although this relationships were not found to be significant. This is very surprising, given that previous research findings and common sense all point to the fact that knowledge is a function of prior experience (Duhan et al, 1997; Alba and Hutchinson, 1987; Park et al., 1994). A possible explanation of this might be the under-representation of people with very high levels of experience. Only 2.9 % of the participants in the study scored "very high" on the experience scale while 48.6 % scored "very low". This might be due to either of two factors. First of all, the Caribbean as a destination and/or package holidays may not be as popular as they were initially considered. Second, the destination and/or package holidays may not be popular amongst the participants in this study.

The results also show that the relationships between objective and subjective prior knowledge and perceived task difficulty were negative, as hypothesised, but they were not found to be significant. The reasons for that might lie in the content of the 'difficult' scenario. Only 21.28 percent of the participants, who were given the difficult scenario, rated the choice task as difficult (score 1 or 2 on the 5-point-likert scale). Two factors might account for this outcome. First, it may be that the number of product choices and attributes of the products were not as many as it would take to account for high levels of information load. Second, the absence of any financial risk in the choice may also have contributed to the generally low levels of perceived task difficulty. A further consideration may be that participants were not employing ht same knowledge in their evaluation of the task difficulty as the objective and subjective knowledge measured by the questionnaire. One could safely assume that any individual has experience of holiday of some sort and consequently more or less developed schemata of holidays. Those schemata can be applied to the choice involved in the two scenaria. On the other hand, objective and subjective knowledge in the questionnaire was strictly measured in relation to package holidays and the Caribbean. Hence, a person which general knowledge about holidays may find the choice easy although he/she is not familiar with the specific nature of the product or the specific destination.

The relationship between task difficulty and affective cues was found to be positive, as expected, but the small number of low scores (1 or 2) on the task difficulty 5-point-likert scale may have made these finding insignificant.

According to the results of this study, the hypothesised relationship between instrumental cues and weak-tie sources of recommendations was positive in direction but not significant.

Contrary to what was expected, a negative relationship between subjective knowledge and instrumental cues was found. Given that this relationship was not significant, further research should be conducted to establish its validity.

As far as risk aversion is concerned, no significant results have been found to support the hypothesised relationship with strong-tie sources. On the contrary, the insignificant relationship that emerged was negative. Surprisingly, though, risk aversion has a significant positive relationship with task difficulty. Once the argument that the perception of a decision-making task as difficult makes people risk averse is irrational, one may assume that it is the wish to avoid risk in decision-making that makes people perceive decision-making tasks as difficult. This suggests that risk aversion has an indirect positive effect on strong-tie sources of recommendation through perceived task difficulty.

Exploratory Factor Analysis An exploratory factor analysis showed some interesting results. First, it was found that the strong-tie/weak-tie classification of recommendations sources was not precisely reflecting the underlying construct. In fact, three factors were extracted: Factor 1 comprises of all four strong-tie recommendation sources, i.e. friends, relatives, colleagues and neighbours. Factor 2 comprises of travel agents and tour operators. Factor 3 comprises of guidebooks and the internet. It can, therefore, be suggested that weak-tie recommendation sources that involve personal contact are qualitatively different from non-personal weak-tie sources.

The second factor analysis focused on the construct underlying evaluative cues. The evaluative cues that were labelled "instrumental cues" were indeed grouped under one factor. However, the evaluative cues which have originally labelled "affective cues" were grouped under two separate factors. In other words, factor 1 includes extra services, flexibility, quality and compensation. Factor 2 includes familiarity and safety and factor 3 includes only similarity. A tentative interpretation might be that the underlying construct in factor 2 is safety provided that familiarity is easily confounded with safety or that familiar places are often perceived as safer than unfamiliar ones. A re-examination of the relationships, previously tested, that takes into account the grouping of the variables as it emerged from the factor analysis provides with different interesting results. Although affective cues were found to have a significant and positive relationship with strong-tie sources of recommendations, only factor 3 (similarity) had a significant positive relationship with strong-tie sources while factor 2 (familiarity, safety) did not. It can be argued that individuals who value being in groups with people that are similar to them are more likely to approach people similar to them (friends, relatives, colleagues, neighbours) for advice. On the other hand, people who value familiarity and safety are more likely to contact travel agents and tour operators for advice. This relationship was not proposed by Duhan et al. (1997) and it is contradictory to the finding that affective evaluative cues related with strong-tie recommendations sources. If one considers, though, that safety can also be evaluated objectively an expert that has the relevant factual knowledge can be considered more appropriate as a recommendation source than a person who knows the decision-maker personally (Brown and Reingen, 1987; Duhan et al., 1997). This is compatible with another finding of the present study. A significant positive relationship was found between risk aversion and a factor that grouped travel agents and tour operators together. People who wish to avoid risk rely more readily on 'experts' with whom they have personal contact.

3.6 Limitations The findings of this study are subject to several limitations. First, a convenience sample was used for the data collection which makes the results not readily generalisable. Moreover, the general limitations of the scenario approach have to be taken into consideration. A scenario may appear realistic but is not necessary compatible with the reality of individual participants. Being the winner of a holiday to the Caribbean is well beyond ordinary every-day experience. A participant's responses may be a function of the extraordinary context rather than a true representation of the same person's decision-making behaviour. Furthermore, WOM has not been extensively studied in the context of tourism services. As a result, the design of the questionnaire was mainly based on limited exploratory research and empirical studies in other services markets.

Chapter 4: Conclusion The present paper has examined the role of WOM in relation to consumer decision-making. The results indicated that persons who perceive a decision-making as being difficult were more likely to sought advice from others closely related to them. Moreover, the consumes who value idiosyncratic attributes of a product or service tend to rely on recommendations of persons closely related to them. Finally, an exploratory analysis showed that risk avers individual perceive decision-making tasks as being more difficult. Apart from those relationships, the findings of the research were inconclusive due to the fact that the results were not significant.

The exploratory factor analysis pointed to the fact, that the initial conceptualisation of the constructs involved in the model was not an accurate reflection of the underlying categorisation. This was the case for weak-tie recommendation source and affective evaluative cues.

As the results of the present study challenge the simple weak-tie/strong-tie classification of recommendation sources, further research is needed to explore the issue of a more detailed classification. Cultural Influences may account for the choice of recommendation source as well as the impact that different factors have on recommendation choices. In order for the effect of cultural influences to be identified and measured, a cross-cultural study of the model is needed.

To sum up, we may say that this study provided limited evidence for the Duhan's et al. (1997) model but it raised conceptual and methodological questions appropriate for future research.

Appendix A

VARIABLE VARIABLE DESCRIPTION

AATM Atmosphere
ACOLL Colleague
AFAM Familiarity
AFRI Friend
AGE Age
AGUI Guidebook
AINT Internet
ANEI Neighbour
AREL Relative
ASAF Safety
ASIM Similarity
ATAG Travel Agent
ATOP Tour Operator
EDU Education
EXPE Experience
IEXT Extra Services
IFLE Flexibility
INC Income
IQUA Quality
OCC Occupation
MSTA Marital Status
IRUL Compensation
OKNO Objective Knowledge
RMIS Mistake
RNER Unknown Area
RNEW New Places
RSTI Known Area
SEX Sex
SKNO Subjective Knowledge
TDIFF Task Difficulty

Appendix B

WORD-OF-MOUTH QUESTIONNAIRE

I am undertaking research at the University of Nottingham on the role of word-of-mouth in consumer decision-making. I am particularly interested in sources of advice and factors influencing choice of holiday. I would be grateful if you could spare five minutes to complete this questionnaire and return it to me. All information will be treated in confidence.

Please read the following scenario and, as best as you can, place yourself into the role described.

About three months ago you took part in a prize-competition at your local supermarket. Surprise! Last week you received a letter saying that you are the winner of the first prize; a two-week holiday for two in the Caribbean. Things couldn't be any better since you have been working for almost eight months without any major breaks. Today you have been contacted by the travel agency in charge. They have been authorised to offer you two package holidays in Jamaica. Package A is an all-inclusive (i.e. all food and drink) holiday in a resort which is known for being very lively whereas package B, also all-inclusive, offers a resort which has a reputation for being rather relaxed and quiet.

To answer the following questions, please tick the appropriate number.

1. How difficult do you feel this scenario is in terms of the choice of a package holiday?

Very difficult 1 2 3 4 5 Not at all difficult

2. A common way people go about selecting a package holiday is to ask someone else for their advice. Given the scenario you have just read, please indicate how likely you are to use the following sources of advice.

Very likely Very unlikely

a) Ask a friend for advice 1 2 3 4 5 b) Ask a relative for advice 1 2 3 4 5 c) Ask a colleague for advice 1 2 3 4 5 d) Ask a neighbour for advice 1 2 3 4 5 e) Consult a guidebook for advice 1 2 3 4 5 f) Search the Internet for advice 1 2 3 4 5 g) Ask a travel agent for advice 1 2 3 4 5 h) Ask a tour operator for advice 1 2 3 4 5

3. Regarding yourself, to what extent do you agree or disagree with the following statements?

Strongly Strongly agree disagree a) It is easy to make a mistake when choosing a holiday. 1 2 3 4 5 b) I am nervous about choosing a holiday in an area that 1 2 3 4 5 I don't know. c) I like to visit new places even if I know little about them. 1 2 3 4 5 d) I prefer to stick with a holiday destination that I know. 1 2 3 4 5

4. How important would each of the following factors be when choosing between these package holidays.

Extremely Extremely important unimportant a) The country is familiar (language, food, etc.). 1 2 3 4 5 b) You will be with a group of similar people. 1 2 3 4 5 c) The holiday destination is safe (safe to walk about, health 1 2 3 4 5 risks, etc.). d) The atmosphere in the resort suits you. 1 2 3 4 5 e) Extra services are provided (e.g. free transfers to airport, 1 2 3 4 5 free insurance). f) The package is flexible and can be tailored to meet your 1 2 3 4 5 personal needs. g) The accommodation is of a high quality (e.g. star-rating). 1 2 3 4 5 h) There are rules for compensation in the case of cancellation, 1 2 3 4 5 alteration and over-booking. i) Other. Please specify ....................................................................

5. Have you been to the Caribbean before?

Yes 1 No 2

6. Have you previously booked a package holiday?

Yes 1 Go to question 7 No 2 Go to question 8

7. How many times have you been on such trips during the last three years?

............................ times

8. Regarding package holidays to areas such as the Caribbean, would you consider yourself

Extremely familiar 1 2 3 4 5 Completely unfamiliar

Please answer the following questions as best as you can. It is important that only you answer these questions. Please do not consult any persons or books. Remember, your responses are completely private.

9. Your best chance of avoiding delays on charter flights is to be on the first outbound flight in the morning.

True 1 False 2

10. If you are a late booker, it is probably sensible to avoid going to this year's fashionable destination in case you are caught up in an accommodation shortage.

True 1 False 2

11. Can you name three major tour operators in the UK package holiday industry.

......................................... ......................................... .........................................

12. On average the temperature in the Caribbean is almost constantly around

87 °F (31 °C) 1 58 °F (14 °C) 2 71 °F (22 °C) 3

13. What is the average flying time to the Caribbean?

15 hours 1 6 hours 2 11 hours 3

14. Which of the following is the correct time zone for the Caribbean?

UK minus 5 hours 1 UK minus 7 hours 2 UK minus 12 hours 3

For purposes of classification, I would be grateful if you could complete the following questions about yourself.

15. Are you

Male 1 Female 2

16. What is your current occupation?

............................................................................................

17. Can you tell me which of these age groups you fall into?

Under 16 1 16-24 2 25-34 3 35-44 4 45-64 5 65+ 6

18. Please indicate your marital status

Single 1 Married 2 Living Together 3 Divorced/Separated 4 Widowed 5

19. Can you tell me something about your education?

GSCE / O-level 1 A-level 2 Further Education (e.g. HNO, NVQ, BTEC) 3 University 4 Other. Please specify........................................................

20. What is your approximate family income each year?

less than £10,000 1 £10,000 - £19,999 2 £20,000 - £29,999 3 £30,000 - £39,999 4 £40,000 - £49,999 5 over £50,000 6

THANK YOU VERY MUCH FOR YOUR HELP.

Please return to either
Klaus Schoefer or Christine Ennew at the University of Nottingham Business School

WORD-OF-MOUTH QUESTIONNAIRE

I am undertaking research at the University of Nottingham on the role of word-of-mouth in consumer decision-making. I am particularly interested in sources of advice and factors influencing choice of holiday. I would be grateful if you could spare five minutes to complete this questionnaire and return it to me. All information will be treated in confidence.

Please read the following scenario and, as best as you can, place yourself into the role described.

About three months ago you took part in a prize-competition at your local supermarket. Surprise! Last week you received a letter saying that you are the winner of the first prize; a two-week holiday for two in the Caribbean. Things couldn't be any better since you have been working for almost eight months without any major breaks. Today you have been contacted by the travel agency in charge. They have been authorised to offer you four package holidays. Package A is an all-inclusive holiday in a three-star hotel in Jamaica. Package B includes half board supplement in a four-star accommodation on the Cayman Islands. Package C offers Bed and Breakfast in a five-star hotel in Jamaica. Finally, package D represents a two-centre holiday; one week in Jamaica and one week on the Cayman Islands.

To answer the following questions, please tick the appropriate number.

1. How difficult do you feel this scenario is in terms of the choice of a package holiday?

Very difficult 1 2 3 4 5 Not at all difficult

2. A common way people go about selecting a package holiday is to ask someone else for their advice. Given the scenario you have just read, please indicate how likely you are to use the following sources of advice.

Very likely Very unlikely

a) Ask a friend for advice 1 2 3 4 5 b) Ask a relative for advice 1 2 3 4 5 c) Ask a colleague for advice 1 2 3 4 5 d) Ask a neighbour for advice 1 2 3 4 5 e) Consult a guidebook for advice 1 2 3 4 5 f) Search the Internet for advice 1 2 3 4 5 g) Ask a travel agent for advice 1 2 3 4 5 h) Ask a tour operator for advice 1 2 3 4 5

3. Regarding yourself, to what extent do you agree or disagree with the following statements?

Strongly Strongly agree disagree a) It is easy to make a mistake when choosing a holiday. 1 2 3 4 5 b) I am nervous about choosing a holiday in an area that 1 2 3 4 5 I don't know. c) I like to visit new places even if I know little about them. 1 2 3 4 5 d) I prefer to stick with a holiday destination that I know. 1 2 3 4 5

4. How important would each of the following factors be when choosing between these package holidays.

Extremely Extremely important unimportant a) The country is familiar (language, food, etc.). 1 2 3 4 5 b) You will be with a group of similar people. 1 2 3 4 5 c) The holiday destination is safe (safe to walk about, health 1 2 3 4 5 risks, etc.). d) The atmosphere in the resort suits you. 1 2 3 4 5 e) Extra services are provided (e.g. free transfers to airport, 1 2 3 4 5 free insurance). f) The package is flexible and can be tailored to meet your 1 2 3 4 5 personal needs. g) The accommodation is of a high quality (e.g. star-rating). 1 2 3 4 5 h) There are rules for compensation in the case of cancellation, 1 2 3 4 5 alteration and over-booking. i) Other. Please specify ....................................................................

5. Have you been to the Caribbean before?

Yes 1 No 2

6. Have you previously booked a package holiday?

Yes 1 Go to question 7 No 2 Go to question 8

7. How many times have you been on such trips during the last three years?

............................ times

8. Regarding package holidays to areas such as the Caribbean, would you consider yourself

Extremely familiar 1 2 3 4 5 Completely unfamiliar

Please answer the following questions as best as you can. It is important that only you answer these questions. Please do not consult any persons or books. Remember, your responses are completely private.

9. Your best chance of avoiding delays on charter flights is to be on the first outbound flight in the morning.

True 1 False 2

10. If you are a late booker, it is probably sensible to avoid going to this year's fashionable destination in case you are caught up in an accommodation shortage.

True 1 False 2

11. Can you name three major tour operators in the UK package holiday industry.

......................................... ......................................... .........................................

12. On average the temperature in the Caribbean is almost constantly around

87 °F (31 °C) 1 58 °F (14 °C) 2 71 °F (22 °C) 3

13. What is the average flying time to the Caribbean?

15 hours 1 6 hours 2 11 hours 3

14. Which of the following is the correct time zone for the Caribbean?

UK minus 5 hours 1 UK minus 7 hours 2 UK minus 12 hours 3

For purposes of classification, I would be grateful if you could complete the following questions about yourself.

15. Are you

Male 1 Female 2

16. What is your current occupation?

............................................................................................

17. Can you tell me which of these age groups you fall into?

Under 16 1 16-24 2 25-34 3 35-44 4 45-64 5 65+ 6

18. Please indicate your marital status

Single 1 Married 2 Living Together 3 Divorced/Separated 4 Widowed 5

19. Can you tell me something about your education?

GSCE / O-level 1 A-level 2 Further Education (e.g. HNO, NVQ, BTEC) 3 University 4 Other. Please specify........................................................

20. What is your approximate family income each year?

less than £10,000 1 £10,000 - £19,999 2 £20,000 - £29,999 3 £30,000 - £39,999 4 £40,000 - £49,999 5 over £50,000 6

THANK YOU VERY MUCH FOR YOUR HELP.

Please return to either
Klaus Schoefer or Christine Ennew at the University of Nottingham Business School
References

Alba, J. W. and Hutchinson, J. W. (1987) 'Dimensions of Consumer Expertise', Journal of Consumer Research, 13(March), 411-454.

Arndt, J. (1967) 'Role of Product-Related Conversations in the Diffusion of a New Product', Journal of Marketing Research, 4(August), 291-295.

Assael, H. (1992) Consumer Behavior and Marketing Action, Boston: PWS-Kent Publishing Company.

Bearden, W. O. and Etzel, M. J. 'Reference Group Influence on Product and Brand Purchase Decisions', Journal of Consumer Research, 9(September), 183-194.

Belk, R. W. (1971) 'Occurrence of Word-of-Mouth Buyer Behavior as a Function of Situation and Advertising Stimuli', in F. C. Alvine (ed.), Proceedings of the American Marketing Association's Educators Conference, Chicago: American Marketing Association.

Ben-Sira, Z. (1976) 'The Function of the Professional's Affective Behavior in Client Satisfaction: A Revised Approach to Social Interaction Theory', Journal of Health and Social Behavior, 17(March), 3-11.

Bloch, P. H., Sherrell, D. L. and Ridgway, N. M. (1986) 'Consumer Search: An Extended Framework', Journal of Consumer Research, 13(June), 119-126.

Brassington, F. and Pettitt, S. (1997) Principles of Marketing, London: Pitman Publishing

Brown, J. J. and Reingen, P. H. (1987) 'Social Ties and Word-of-Mouth Referrral Behavior', Journal of Consumer Research, 14(December), 350-362.

Brucks, M. (1985) 'The Effects of Product Class Knowledge on Information Search Behavior', Journal of Consumer Research, 12(June), 1-16.

Brucks, M. and Schurr, P. H. (1990) 'The Effects of Bargainable Attributes and Attribute Range Knowledge on Consumer Choice Processes', Journal of Consumer Research, 16(March), 409-419.

Bryman, A. and Cramer, D. (1997) Quantitative data analysis with SPSS for Windows: A guide for social scientists, London: Routledge.

Buttle, F. A. (1998) 'Word-of-mouth: understanding and managing referral marketing', The Academy of Marketing Annual Conference, 100-106.

Chisnall, P. M. (1985) Marketing: A Behavioural Analysis, London: McGraw-Hill.

Cunningham, S. M. (1966) 'Perceived Risk as a Factor in the Diffusion of New Product Information', in R. M. Haas (ed.), Proceedings of the American Marketing Association Fall Conference, 28, 698-721.

Davis H. L. and Rigaux, B. P. (1974) 'Perceptions of Marital Roles in Decision Processes', Journal of Consumer Research, 1(June), 51-62.

Day, G. S. (1971) 'Attitude Change, Media and Word of Mouth', Journal of Advertising Research, 11(6), 31-40.

Day, R. L. (1984) 'Modeling Choices Among Alternative Responses to Dissatisfaction', in T. C. Kinnear and A. Arbor, Advances in Consumer Research, vol. 11, MI: Associsation for Consumer Research.

Dibb S., Simkin, L., Pride, W. M. and Ferrell, O. C. (1997) Marketing: Concepts and Strategies, Boston: Houghton Mifflin.

Dichter, E. (1966) 'How Word-of-Mouth Advertising Works', Harvard Business Review, 44 (November-December), 148-152.

Diener, B. J. and Greyser, S. A. (1978) 'Consumer Views of Redress Needs', Journal of Marketing, 42(October), 21-27.

Duhan, D. F., Johnson, S. D., Wilcox, J. B. and Harrell, G. D. (1997) 'Influences on Consumer Use of Word-of-Mouth Recommendation Sources', Journal of the Academy of Marketing Science, 25(4), 283-295.

Engel, J. E., Blackwell, R. D. and Kegerreis, R. J. (1969) 'How Information is Used to Adopt an Innovation', Journal of Marketing Research, 9(December), 3-8.

Engel, J. F., Blackwell, R. D. and Miniard, P. W. (1993) Consumer Behavior, New York: Dreyden Press.

Engel, J. F., Kollat, D. J. and Blackwell, R. D. (1968) Consumer Behavior, New York: Holt, Rinehart and Winston.

Ennew, C. T. (1993) The Marketing Blueprint, Oxford: Blackwell.

Feick, L. F. and Price, L. L. (1987) 'The Market Maven: A Diffuser of Marketplace Information', Journal of Marketing, 51 (January), 83-87.

Feldman, R. S. (1989) Adjustment: Applying Psychology in a Complex World, New York: McGraw-Hill.

File, K. M., Cermak, D. S. P. and Prince, R. A. (1994) 'Word-of-mouth effects in professional services buyer behaviour', The Services Industries Journal, 14(3), 301-314.

Foxall, G. R. (1985) 'Consumer behaviour', in M. J. Backer (ed.), The Marketing Book, Oxford: Butterworth-Heinemann.

Foxall, G. R. (1997) 'Consumer decision making' in M. J. Backer (ed.), The Marketing Book, Oxford: Butterworth-Heinemann.

Gilbert, D. C. and Soni, S. (1991) 'UK Tour Operators and Consumer Responsiveness', The Services Industries Journals, 11(4), 413-424.

Gilly, M. C., Graham, J. L., Wolfinbarger, M. F. and Yale, L. J. (1998) 'A Dyadic Study of Interpersonal Information Search', Journal of the Academy of Marketing Science, 26(2), 83-100.

Granovetter, M. S. (1973) 'The Strength of Weak Ties', American Journal of Sociology, 78(May), 1360-1380.

Haywood, K. M. (1989) 'Managing Word of Mouth Communications', The Journal of Services Marketing, 3(2), 55-67.

Heath, C. (1996) 'Do People Prefer to Pass Along Good or Bad News? Valence and Relevance of News as Predictors of Transmission Propensity', Organizational Behavior and Human Decision Processes, 68(2), 79-94.

Helm, S. and Schlei, J. (1998) 'Referral Potential - Potential Referrals: An Investigation into Customers' Communication in Services Markets', in P. Andersson, Track 1 Market Relationships, Proceedings of the 27th EMAC Conference Marketing: Research and Practise, 41-56.

Herr, P. M., Kardes, F. R. and Kim, J. (1991) 'Effects of Word-of-Mouth and Product Attribute Information on Persuasion: An Accessibility-Diagnosticity perspective', Journal of Consumer Research, 17(March), 454-462.

Higie, R. A., Feick, F. L. and Price, L. L. (1987) 'Types and Amount of Word-of-Mouth Communication about Retailers', Journal of Retailing, 63(Fall), 260-278.

Hirschman (1970) Exit, Voice and Loyalty: responses to decline in firms, organizations and States', Cambridge, MA: Harvard University Press.

Hoch, S. and Deighton, J. (1989) 'Managing What Consumers Learn from Experience', Journal of Marketing, 53(April), 1-20.

Howard, J. A. (1977) Consumer Behavior: Application of Theory, New York: McGraw-Hill.

Howard, J. A. and Sheth, J. N. (1969) The Theory of Buyer Behaviour, New York: John Wiley and Sons.

Jacoby, J., Speller, D. E. and Kohn, C. A. (1974) 'Brand Choice Behavior as a Function of Information Load', Journal of Marketing Research, 11(February), 63-69.

Johnson, E. J. and Russo, J. E. 'Product Familiarity and Learning New Information', Journal of Consumer Research, 11(June), 542-550.

Katz, E. and Lazarsfeld, P. (1955) Personal Influence, New York: Free Press.

Keller, K. L. and Staelin, R. (1987) 'Effects of Quality and Quantity of Information on Decision Effectiveness', Journal of Consumer Research, 14(September), 200-213.

Kim, C. and Khoury, M. (1987) 'Task Complexity and Contingent Information Processing in the Case of Couple's Decision Making', Journal of the Academy of Marketing Science, 25(Fall), 32-43.

Kotler, P. (1997) Marketing Management: Analysis, Planning, Implementation, and Control, Englewood Cliffs: Prentice Hall.

Kotler, P., Armstrong, G., Saunders, J. and Wong, V. (1996) Principles of Marketing: The European Edition, Hemel Hemstead: Prentice Hall Europe.

Lazarsfeld, P., Berelson, B. and Gaudet, H. (1948) The People's Choice, New York: Columbia University Press.

Locander, W. B. and Hermann, P. W. (1979) 'The Effect of Self-Confidence and Anxiety on Information Seeking in Consumer Risk Reduction', Journal of Marketing Research, 16(May), 268-279.

Loudon, D. L. and Della Bitta, A. J. (1993) Consumer Behavior: Concepts and Applications, London: McGraw-Hill.

Marks, L. J. and Olson, J. C. (1981) 'Toward a Cognitive Structure Conceptualization of Product Familiarity', in K. Monroe and Arbor, A. (eds.), Advances in Consumer Research 8.

Mahajan, V., Muller, E. and Bass, F. M. (1990) 'New Product Diffusion Models in Marketing: A Review and Directions for Research', Journal of Marketing, 54(January), 1-26.

Mizerski, R. W. (1982) ' An Attribution Explanation of the Disproportionate Influence of Unfavorable Information', Journal of Consumer Research, 9(December), 301-310.

Morrison, A. M., Yang, C.-H., O'Leary, J. T. and Nadkarni, N. (1996) 'Comparative Profiles of Travellers on Cruises and Land-based Resort Vacations', The Journal of Tourism Studies, 7(2), 15-27.

Murray, K. B. (1991) 'A Test of Services Marketing Theory: Consumer Information Acquisition Activities', Journal of Marketing, 55(January), 10-25.

Newell, A. and Simon, H. A. (1972) Human Problem Solving, Englewood Cliffs: Prentice Hall.

Nicosia, F. M. (1966) Consumer Decision Processes, Englewood Cliffs: Prentice-Hall.

O'Shaughnessy, J. (1995) Competitive Marketing: A Strategic Approach, London: Routledge.

Paquette, L. and Kida, T. (1988) 'The Effect of Decision Strategy and Task Complexity on Decision Performance', Organizational Behavior and Human Decision Processes, 41, 128-142.

Park, C. W. and Lessig, V. P. (1981) 'Familiarity and Its Impact on Consumer Decision Biases and Heuristic', Journal of Consumer Research, 8(September), 223-230.

Park, C., Mothersbaugh, D. and Feick, L. (1994) 'Consumer Knowledge Assessment', Journal of Consumer Research, 21(June), 71-82.

Rao, A. R. and Monroe, K. B. (1988) 'The Moderating Effect of Prior Knowledge on Cue Utilization in Product Evaluations', Journal of Consumer Research, 15(September), 253-264.

Rao, A. R. and Seiben, W. A. (1992) 'The Effect of Prior Knowledge on Price Acceptability and the Type of Information Examined', Journal of Consumer Research, 19(September), 256-271.

Reynolds, F. D. and Darden, W. R. (1971) 'Mutually Adaptive Effects of Interpersonal Communication', Journal of Marketing Research, 8(November), 449-454.

Richins, M. L. (1983) 'Negative Word-of-Mouth by Dissatisfied Consumers: A Pilot study', Journal of Marketing, 47(Winter), 68-78.

Richins, M. L. and Root-Shaffer, T. (1987) 'The Role of Involvement and Opinion Leadership in Consumer Word-of-Mouth: An Implicit Model Made Explicit', in M. J. Houston (ed.), Advances in Consumer Research, 15, 32-36.

Robertson, T. S. (1971) Innovative Behaviour and Communication, New York: Holt, Rinehart and Winston.

Rogers, E. M. (1962) Diffusion of Innovations, New York: Free Press.

Sheth, J. N. (1971) 'Word-of-Mouth in Low-Risk Innovations', Journal of Advertising, 11(June), 15-18.

Singh, J. (1988) 'Consumer complaints intentions and behavior: definitional and taxonomical issues', Journal of Marketing, 52(January), 93-107.

Singh, J. (1990) 'Voice, exit and negative word-of-mouth behaviors across three service categories', Journal of the Academy of Marketing Science, 18(Winter), 1-16.

Slama, M. W. and Williams, T. G. (1990) 'Generalization of the Market Maven's Information Provision Tendency Across Product Categories', in M. E. Goldberg, G. Gorn and R. W. Pollay (eds.), Advances in Consumer Research, 17, Association for Consumer Research, Provo: UT.

Solomon, M. R. (1993) Consumer Behavior, Englewood Cliffs: Prentice Hall.

Sujan, M. (1985) 'Consumer Knowledge: Effects on Evaluation Strategies Mediating Consumer Judgments', Journal of Consumer Research, 12(June), 31-46.

TARP (Technical Assistance Research Programs) (1979) Consumer Complaint Handling in America: Summary of Findings and Recommendations, Washington, DC: Office of Consumer Affairs.

Tax, S. S., Chandrashekaran, M. and Christiansen, T. (1993) 'Word-of-Mouth in Consumer Decision-Making: An Agenda for Research', Journal of Consumer Satisfaction, Dissatisfaction and Complaining Behavior, 6, 75-80.

Watkins, H. S. and Liu, R. (1996) 'Collectivism, individualism and in-group membership: implications for complaining behaviors in multi-cultural contexts', Journal of International Consumer Marketing, 8(3/4), 69-96.

Wilkie, W. L. (1994) Consumer Behavior, New York: John Wiley and Sons.

Wilson, J. R. (1994) Word-of-Mouth Marketing, New York: John Wiley and Sons.

Whyte, W. H. Jr. (1954) 'The Web of Word of Mouth', Fortune, 50(November), 140-143.

Zeithaml, V. A. and Bitner, M. J. (1996) Services Marketing, New York: McGraw-Hill.

-----------------------
Individual influences

Environmental influences

Problem recognition

Information Search

Evaluation of alternatives

Purchase

Post-purchase evaluation

Decision-Making Process

Feedback

Socially Integrated (32%)

Socially Dependent (18%)

Socially Independent (18%)

Socially Isolated (32%)

OPINION
LEADEERSHIP High

Low

INFORMATION SEEKING

High Low