Master Thesis
THE EFFECT OF INFORMATION TYPE ON
CONSUMER ATTITUDE AND BEHAVIORAL
INTENTIONS: THE MODERATING ROLE OF
INVOLVEMENT
August 2012, University of Groningen
ABSTRACT
This study investigates one type of electronic word-of-mouth (eWOM), the online consumer review. It is examined to what extent information type does have an influence on consumers’ attitudes and behavioral involvement and how involvement moderates this relationship. The hypotheses were tested using a 3 (type of information) x 2 (involvement) research design. The results show that both attribute- and benefit-centric information have more influence on attitudes of low-involvement consumers rather than high-involvement consumers. Also, low-involvement consumers were more influenced through both types of information than attribute-centric information. Theoretical and managerial implications of these findings are provided.
PREFACE
This thesis is the final step in completing my Master of Marketing Management of the University of Groningen. Since it has been a journey writing this report, I would like to use this preface to thank several persons whom have made this thesis possible. First of all I would like to thank my supervisor Debra Trampe who has been very helpful in providing constructive feedback and assistance during the whole process. She has been very supportive and responded always very quickly when necessary. In this respect, I would also like to thank my second supervisor Eline de Vries, for reviewing and providing helpful suggestions during the last part of this thesis. In addition, a word of thanks to Martijn van der Veen and Marcella Vermeij for their helpful tips during the statistical part of this study. Moreover, I would like to thank all my friends for listening to all my complaints, especially Leon Scholte Albers. Finally, many thanks go out to my parents who have always supported and encouraged me during my study.
TABLE OF CONTENT
1 INTRODUCTION ...6
1.1 Background and Problem Statement ... 6
1.2 Theoretical and Managerial Relevance ... 7
1.3 Research Question ... 8
1.4 Structure of the Thesis ... 9
2 THEORETICAL BACKGROUND ... 10
2.1 Online consumer reviews... 10
2.2 Motivations to use OCRs ... 10
2.3 Website type and type of recommendation source ... 11
2.4 Search goods versus experience goods ... 12
2.5 Effects of messages on attitude and behavior ... 13
2.6 Type of information ... 15
2.7 The moderating role of involvement ... 17
2.8 Involvement and message content ... 18
2.9 Hypotheses... 20
3 RESEARCH DESIGN AND METHODOLOGY ... 23
3.1 Sample and design ... 23
3.2 Procedure ... 23 3.3 Manipulations ... 24 3.4 Dependent variables ... 26 3.5 Control variable ... 27 3.6 Scales ... 27 4 RESULTS ... 29 4.1 Manipulation checks ... 29 4.2 Attitude ... 30 4.3 Behavioral Intentions ... 31 4.4 Hypotheses... 32
5 CONCLUSION AND DISCUSSION ... 34
5.2 Limitations & Further Research ... 37
APPENDIX I Reviews ... 39
APPENDIX II Questionnaire ... 40
1 INTRODUCTION
1.1
Background and Problem Statement
Dining out goes beyond just satisfying hunger and has become a popular form of entertainment. People go out to eat, and they expect to derive pleasure and satisfaction from it (Warde and Martens, 2002). In 2011, the total expenditure for food away from home was approximately € 18 million, almost one-third of the amount of the total food expenditure (Foodservicemarkt Nederland, 2012). According to Gregory and Kim (2004), eating away from home becomes increasingly ingrained in our culture. Researchers have classified restaurants as experience goods, which mean that full information on certain attributes cannot be known without the direct experience (Xiang et al., 2007). Because of their intangibility, consumers rely often on word of mouth for this kind of products. Word of mouth (WOM) is defined as “all informal communications directed at other consumers about the ownership, usage, or characteristics of particular goods and services or their sellers” (Westbrook, 1987).
According to Chen and Xie (2005) consumers make use of friends, relatives, salespeople and publications such as consumer reports as recommendation sources. With the rapid growth of internet, consumers could benefit from the availability electronic word of mouth (eWOM) when making their purchase decision. The options for using eWOM are much larger than the traditional scope WOM. Whereas traditional WOM is often limited to friends and relatives, eWOM occurs through various channels such as discussion forums, blogs, social networking sites, virtual communities, and consumer review sites (Hennig et al., 2004; Chu and Kim, 2011; Senecal and Nantel, 2004). Another difference between traditional WOM and eWOM is the unlimited availability to other consumers for an indefinite period of time. Moreover, eWOM can be illustrated with pictures and supporting comments of other consumers (Bickart and Schindler, 2001).
prior to paying for a service delivered offline (comScore, 2008). Moreover, OCRs are an important factor in product sales. Chevalier and Mayzlin (2006) investigated review statistics and found that the number of reviews is positively related to product sales. For example, Amazon began to offer consumers the possibility to evaluate its products on the website. Nowadays, the company has over 10 million reviews, which are one of its most popular and successful features (Harmon, 2004). Accordingly, many e-tailers have adopted the same strategy and are taking advantage of OCRs as a new marketing tool (Dellarocas, 2003). OCRs could be controllable because marketers can decide whether to allow consumer reviews to be shown or not, and if they are shown marketers can offer a specific review format in order to guide consumers to post their opinions in the way they want. Some firms even strategically manipulate online reviews in an effort to influence consumers’ purchase decisions (Dellarocas 2006). An underlying belief behind such strategies is that OCRs can significantly influence consumers’ purchasing decisions, especially for experience goods (Godes and Mayzlin, 2004; Liu, 2006; Dellarocas et al., 2007).
When looking at the written OCRs, it is striking that their structure and content can vary to a great extent. Hence the question arises what kind of reviews influence consumers most? Which messages fits best in order to influence consumers’ attitudes and behavioral intentions? For example, the review could vary in number and quality of arguments, and could consist of either subjective or objective information. Moreover, the credibility and expertise of the reviewer could vary. Within literature, different researchers have shed their light of the effects of these different types of message cues on consumers’ attitude and behavioral intentions. It has been showed that the varied message strategies obviously could have differential persuasive impact on consumers' attitudes.
1.2
Theoretical and Managerial Relevance
the type of information they seek out. Therefore, consumers’ involvement will be regarded as the moderating factor. To date, no study has specifically investigated the effects of review content characteristics and the moderating role of consumer involvement on consumers’ attitudes and behavioral intentions. Combining these gaps, the goal of this study is to examine whether different types of information could influence consumers’ attitude and behavioral intentions and how these effects are moderated by consumer involvement. In addition, most studies have focused on products, instead of services. This lack of studies that focus on the side of the service providers provides an exiting research opportunity, as this could give some direction how firms may potentially use OCRs as a strategic tool.
Because OCRs are a very important communication and marketing channel nowadays, it is important for managers to know which type of information fits best to their consumers. Although both types of consumers increase their search when an important purchase is imminent, low involved consumers still search less and will be more difficult to reach with advertising than high involved consumers (Richins et al., 1992). When the type of information in the OCRs is represented in a way that fits to consumers’ involvement level, those reviews can positively affect attitude and behavioral intentions. Also, content factors, in contrast to context factors, appear to produce a persisting change in attitudes (Petty and Cacioppo, 1984). Taking this into account, it is important for managers to know which information type fits consumers with a high- or low involvement level. Equipped with this knowledge, managers will be able to use OCRs as a strategic communication channel.
1.3
Research Question
This study extends prior research on OCRs by examining whether different types of information could influence consumers’ attitude and behavioral intentions and how these effects are moderated by consumer involvement. Therefore, the research question would be:
What is the influence of information type on consumers’ attitude and behavior and how does involvement moderate this relationship?
1.4
Structure of the Thesis
2 THEORETICAL BACKGROUND
2.1
Online consumer reviews
In this section, the main literature for OCRs will be outlined. OCRs are considered to be a major informational source for consumer judgments and a critical factor for purchase decisions (Chatterjee, 2001). Not only is the amount of individuals posting their evaluations online growing, but also the amount of consumers relying on information created by other consumers is increasing, especially for services (Moe and Trusov, 2011).
Review platforms differ in minor ways but have similar basic functions (Senecal and Nantel, 2004). They enable consumers to read the opinions and experiences of other consumers relating to a wide range of product and service categories. Contributions on opinion platforms usually include both a verbal account of a consumer’s experience with a product and a formalized rating of the product. Readers have the opportunity to assess the quality and trustworthiness of individual contributions, and their ratings are visible to other readers. Some platforms rely on business models, which are often similar. Revenues are earned from banner advertising and from offering market-research services. Some income may also derive from sales commissions which could be a possible source of conflict with the platform’s trustworthiness.
2.2
Motivations to use OCRs
To examine the impact of OCRs on consumers’ attitude and behavioral intentions, it is necessary to identify the motives that induce consumers to seek information from these sources. Motives are defined as the “general drivers that direct a consumer’s behavior toward attaining his or her needs.”
Henning-Thurau et al. (2004) found that social benefits, concern for others, economic incentives, and self enhancement are the most important reasons for posting reviews. Schindler and Bickart (2004) distinguish three motives for using OCRs; information motives, entertainment motives, and support and community motives.
risk, which is greater when buying a service than when buying a physical product. The motivation to determine the social position is derived from the finding that consumers read reviews in order to evaluate the product and its associated social prestige (i.e. determine their social status). This social orientation through information has an influence on change in buying behavior. Thus, it could be said that online review platforms sometimes functions as “social positioners”, meaning that they serve as the infrastructure of a virtual community that offers social and information utility by helping consumers to compare and process their product experiences. The dissonance reduction motive is derived from the cognitive dissonance theory (Sweeney et al., 2000) which explains that when consumers have to make their choice, they often experience cognitive incongruence related to the information about the alternatives they have rejected. Moreover, cognitive incongruence may be caused by conflicting information from other sources. This could be reduced by unbiased or neutral information. Since OCRs platforms offer neutral information, cognitive incongruence could be reduced by using such platforms. The remuneration motive refers to the fact that a lot of online review platforms reward their consumers for reading contributions. It has been proved that rewards, either directly or indirectly, are a motive for consumers to read other-consumers’ reviews.
2.3
Website type and type of recommendation source
Review platforms can be used and promoted by different types of websites. Senecal and Nantel (2004) distinguished three different types: sellers (e.g., retailer or manufacturer websites such as Bol.com), commercially linked third parties (e.g., comparison shopping websites such as Kieskeurig.nl), and non-commercially linked third parties (e.g., product or merchant assessment websites such as Consumentenbond.nl). Research has indicated that third party websites are preferred by consumers and have a significant influence on consumers’ purchase decisions (Chen and Xie 2005; Thurau and Walsh 2003). They are viewed as more credible and they are decreasing consumers’ search costs.
influence than sources that have no personal knowledge about the consumers and providing non-personalized information (e.g. other consumers).
Park and Kim (2008) distinguished two information sources; information provided by consumers and information created by sellers. They found that information created by consumers was followed in a greater extent than information created by sellers because it is perceived as more credible and trustworthy. Moreover, consumer-created content will tend to be more consumer-oriented than seller-created content, since it describes the product- or service experience from the consumers’ perspective. Because it represents their personal feelings and satisfaction about the product, the information is more understandable and familiar to the consumers.
The fact that researchers found that both type of website and type of source are having an impact on the likelihood that the review has influence, could be explained through Kelman’s (1961) research to source credibility. According to Kelman (1961), source credibility consists of senders’ expertise and trustworthiness. Expertise is here defined as “the extent to which the source is perceived as being capable of providing correct information” (Bristor 1990, pp. 73) and trustworthiness as the perceived information source’s motivation to communicate this expertise without bias (McGuire, 1969). Both have been found to be positively correlated with consumers’ attitude and behavioral intentions (Senecal and Nantel, 2004). For example, Bansal and Voyer (2000) found that the greater the perceived expertise of the source, the greater the influence of this message on the consumers’ purchase decision and Smith et al. (2005) found a positive relationship between consumers’ trust in a recommender and the influence of the source on the choice decision.
2.4
Search goods versus experience goods
preferences are shared (Park and Kim, 2008). This may overcome the problems of low comparability and few search qualities associated with services. Since they discuss their purchase intentions with referrals, consumers presumably are influenced in their decision-making because they interact and communicate with others. In line with this, Huang et al. (2009) find that OCRs and multimedia that enable consumers to interact before purchase have a greater effect on consumer search and purchase behavior for experience goods than for search goods. Park and Lee (2009) agree on this stating that eWOM related to experience goods have a greater effect on consumer decisions. Therefore, especially marketers from experience goods should make every effort to strengthen the perceived usefulness of OCRs (Park and Lee, 2009).
According to the investigated literature, it has been shown that consumers rely more on WOM for experience goods than for search goods. Furthermore, research regarding type of source found that consumer-created information is more influencing than seller created information because of their trustworthiness and credibility. Contradictory results exist regarding the influence of website type. Taking this into account, the OCRs used in this study will relate to experience products, and will consist of consumer-created information. Because of the contrary results for website type, this factor will not be taken into account.
In the next section, the main constructs of this research, information type and involvement, will be widely discussed.
2.5
Effects of messages on attitude and behavior
(Chaiken, 1987). In this mode of attitude change, which also has been labeled the ‘peripheral route’ to persuasion (Petty and Cacioppo, 1981), attitude change has been shown to be largely independent of message content. A related model is the heuristic-systematic model from Chaiken (1980, 1987). Here, systematic processing and heuristic processing are distinguished. Applied to persuasion, systematic processing implies that people form their attitudes by actively attending to and cognitively elaborating persuasive messages. In heuristic processing, on the other hand, peoples’ attitudes are formed by invoking heuristics such as “experts can be trusted”, “more arguments are better”, and “long messages are valid messages”
Which of the two ‘routes’ or processing modes is used depends on consumers’ motivation and ability to process the persuasive message. The ability to process information refers to consumers’ expertise (Celsi and Olson, 1988), and motivation to process refers to consumers’ involvement (Cohen 1983; Houston and Rothschild 1978; Petty and Cacioppo 1981; Wright, 1974; Zaichkowsky 1985). If both motivation and ability are high, the probability of systematic processing is increased. If either motivation or ability are low, heuristic processing may still provide an economic way of forming an attitude judgment. For instance, a person lacking the ability to carefully process message content because of external distraction, they may still be able to form an attitude judgment on the basis of simple heuristic cues such as the reputation of the source, the number of arguments presented, or the length of an argument, without careful consideration of the content (Park and Kim, 2008).
be more salient in memory than attitudes based on peripheral cues, and thus people may be more able to act like them.
As shown above, a lot of research has already been conducted on the different persuasion effects of context- and content related cues. This study focuses on message content rather than message context. The next section will discuss literature related to message content.
2.6
Type of information
The way in which information is presented in a message is an important choice that can greatly affect cognitive activity among receivers and, subsequently, the message's persuasiveness. Therefore, a lot of research on reviews has focused on the message content. Message content can vary in quality and sort of information. This is also mentioned as type of information.
One categorization of information type is characterized as either objective or subjective. According to Cohen (1983), objective messages use explicit information (e.g., technical specifications, numerical cues, visual presentations) directly indicating an attribute or benefit. Subjective claims use surrogate indicators to indirectly suggest existence of attitudes or benefits a brand provides. Within subjective messages, on the other hand, claims are not explicitly made but implied by colors, endorsements (e.g., by experts), symbols (e.g., phrases, brand names) or magnitudes (created by camera technique).
Objective information provides potent, high-quality arguments (Gill et al., 1988). Objective messages are less ambiguous, therefore generating more support arguments and fewer counterarguments than subjective claims (Edell and Staelin 1983). Moreover, objective messages are seen as more credible (Holbrook 1978). In contrast, subjective messages more typically serve as peripheral cues designed to produce positive affective responses (Petty et al., 1983), which may be induced through associations (i.e., classical conditioning), examples and visual descriptions, and with positive cues (e.g., attractive pictures or endorsers) (Kelman 1961; Mitchell and Olson 1977; Mowen 1980). They also may occur as receivers draw simple inferences about merits of a position implied by subjective cues (e.g., symbols or magnitudes suggest drawing certain inferences) (Petty et al., 1983).
Strong messages are understandable and objective (manipulated via statistical evidence and credible sources), whereas weak messages are emotional and subjective (e.g., an opinion in a testimonial). An example of a strong message is: “This product is twice as fast as other comparable goods, and it is even cheaper”, because it specific, clear, and supported by arguments. An example of a weak review is: “I can’t believe I got this; I’m proud of it”, or “It’s so good that I will never buy another one”. These kind of reviews are subjective, emotional, and do not make any reasoned arguments. Park et al. (2007) found that high quality messages have more influence on consumers’ purchase intentions than weak messages. In line with this, Wood et al. (1985) demonstrated that strong messages (manipulated via statistical evidence and credible sources) were more persuasive than weaker claims (e.g., an opinion in a testimonial) when participants processed and evaluated the validity of the informational content.
than when only benefits were presented. The absence of attribute information was believed to have prevented experts from assessing the veracity of benefits claimed, leading them to discount the claim.
Research related to product information shows that experts prefer technical attribute information, while novices prefer literally expressed benefit information (Maheswaran and Sternthal, 1990). For example, experts make decisions about food on the basis of technical attributes like nutritional information, whereas novices rely on benefit information about the items (e.g. good for you).
Although it is not known if the effects reported in the above literature were attributable to differences in participants' information processing because no cognitive measures were attained, it would seem that the presentation of statistical evidence in support of product claims logically requires more cognitive elaboration than do anecdotal or associative messages. Such logic is implicit in the definition of informational advertising, which provides "consumers with factual... data in a clear and logical manner such that they have greater confidence in their ability to assess the merits of buying the brand" (Puto and Wells, 1984, p. 638).
The essential criterion for persuasion effectiveness, and thus influencing consumers’ attitudes and behavioral intentions, is the acceptance of the content. Therefore, a cognitive fit is essential. According to the explained models of persuasion, consumers’ involvement plays a very important role. Therefore, this construct will be discussed in the next section.
2.7
The moderating role of involvement
involvement is low and contributes little to involvement responses. For example, few consumers are generally interested in products such as vacuum cleaners, ketchup, or paint (Richins et al., 1992). According to Gill et al. (1988), the common meaning underlying these terms is personal relevance. This refers to how personally important or connected a product is to consumers’ values.
In this study, situational involvement will be used as the moderating variable because of four reasons. First, dining out is a high-risk product, because services are intangible and might not be fully understand before its consumption (Bristor, 1990). Second, consumers’ involvement can be different for the same product, depending on their personal characteristics (Park et al., 2004). Third, Park et al. (2004) suggest that the situational importance of a purchase decision is likely to be the most representative of the variance in the consumers’ involvement (Mittal, 1995). They argue that some products are inherently involving because of the nature of the purchase, but various situations can elicit individuals concern for their behavior in a situation. Finally, the experiment ensures that respondents generate situational involvement, even when they may not have general involvement with the subject.
Furthermore, many researchers have distinguished between high and low involvement situations (Chaiken, 1980; Petty et al., 1981). Although there are many specific definitions of "involvement", there is considerable agreement that in high involvement situations, the persuasive message under consideration has a high degree of personal relevance to the recipient, whereas in low involvement situations, the personal relevance of the message is rather trivial. As such, involvement is related to the motivation for processing messages (Mitchell, 1979). The next section will explain this in more detail.
2.8
Involvement and message content
Involvement is related to message-processing motivation which may be derived from the self-concept, needs and value-relatedness of the product, and is associated with a readiness to acquire and process information (Gill et al., 1988). OCRs may have more general relevance for those perceiving the product as personally important. As a result of this ongoing relevance, they should be more selectively attentive and willing to process product-related information (Mitchell, 1979). Moreover, because they regard the product class as important, high-involved consumers’ motivation and concern should enhance the likelihood they will evaluate the review content in a reasoned and critical manner.
message content is the most important determinant of the amount of persuasion that occurs. On the other hand, under low involvement conditions, the focus of thought is more on context cues of a communication which allows a person to evaluate a message or decide what attitudinal position to adopt without engaging in any extensive cognitive work relevant to the issue or product under consideration. This assumes that attitude change is determined by different elements under high and low involvement conditions. For example, when people are thinking carefully about information, they should be affected by the quality of the arguments a message contains (Petty & Cacioppo, 1986). In line with this, Park et al. (2007) found a strong relationship between involvement and information processing. When involvement increases, people have greater motivation to comprehend the salient information. Highly involved consumers are found to be more influenced by issue-relevant arguments and product-relevant attributes. On the other hand, when involvement is low, individuals are more influenced by peripheral cues such as the characteristics of information sources (Chaiken, 1980), number of arguments (Petty and Cacioppo, 1984), famous endorsers, or high expertise of the source of the message. Focusing on each of the latter aspects of a communication allows a person to evaluate a message or decide what attitudinal position to adopt without engaging in any extensive cognitive work relevant to the issue or product under consideration. Thus, when consumers are not highly involved with a persuasive message they rely on a short-cut means of evaluation.
In addition to this, Wood et al. (1985) demonstrated that strong messages (manipulated via statistical evidence and credible sources) were more persuasive than weaker claims (e.g., an opinion in a testimonial) when participants processed and evaluated the validity of the informational content. In contrast, participants who did not actively scrutinize the message were not persuaded by the information contained in the message but rather by peripheral, heuristic message-based cues (e.g., length, structural features, and tone of voice). Thus, when consumers are highly involved, stronger messages had a greater impact on persuasion than weaker messages.
Within this study, two types of information will be distinguished; attribute-centric information and benefit centric-information. Attribute-centric information consists of arguments based on attributes such as numbers representing attribute levels. Thus, the information is objective, which means that the message could be typed as strong and that the argument quality could be typed as high.
Benefit centric information consists of subjective interpretations about the attributes. Thus, this information is more subjective, which means that the message could be typed as weak, and the arguments as less strong.
2.9
Hypotheses
Combining the findings of the persuasion models ELM and HSM and extended research to persuasive cues and information types, the information processing strategy of highly involved consumers fits attribute-centric reviews, because these consumers are influenced by strong, high quality, and objective messages. On the other hand, lower involved consumers fit reviews typed as benefit-centric.
When individuals are able to process online consumer reviews represented in a way that is cognitively fit, they can efficiently process given reviews, thereby those reviews positively affect attitude and behavioral intentions. Therefore, holding that the reviews deliver information about the same elements of the restaurant, the following hypotheses are proposed:
H1a: product-attribute information has more influence on attitude for consumers with high involvement than for consumers with low involvement.
H2a: product-benefit information has more influence on attitude for consumers with low involvement than for consumers with high involvement.
revisit, recommendation, and positive word-of-mouth. These findings all support the significant link between customer attitude and behavioral intentions. Thus, more favorable attitudes lead to more favorable behavioral intentions. Therefore, the effects of attitude and behavior are expected to behave in the same way, which results in the following hypotheses:
H1b: product-attribute information has more influence on behavioral intentions for consumers with high involvement than for consumers with low involvement.
H2b: product-benefit information has more influence on behavioral intentions for consumers with low involvement than for consumers with high involvement.
Besides the attribute-centric and benefit-centric condition, a combination of these conditions will be added. This review will consist of both attribute- and benefit-centric information, which means that the information consist of either strong/objective or weak/subjective messages.
unwilling to engage in the effortful cognitive work necessary to evaluate the quality of the arguments and are not willing to give counter argumentation. Holding that the number of arguments is the same in the attribute- and benefit-centric condition as in the attributes-only condition, but that the number of high-quality arguments will be reduced, and the number of low-quality arguments will be increased, high-involved consumers will engage in producing more counterarguments, which leads to less persuasion, and thus are their attitudes and behavioral intentions less influenced. Low-involved consumers, on the other hand, will not engage in counter argumentation, and are thus more influenced by both types of information. This results in the following hypotheses:
H3a: both product attribute- and benefit information have more influence on attitude for consumers with low involvement than for consumers with high involvement.
3 RESEARCH DESIGN AND METHODOLOGY
In this chapter, the research design and methodology will be outlined. In the first paragraph, the research method will be discussed, including a description of the data collection and the procedures involved. Finally, an overview of the measures will be presented. The questionnaire can be found in Appendix II.
3.1
Sample and design
The target group consisted of persons aged 20 to 29 years old. This group was chosen because these people are visiting restaurants most frequently; 90% of them are dining out at least four times a year (CBS, 2009). Moreover, research has shown that students tend to go out more frequently than other groups (Auty, 2006). A convenience sample was used in order to guarantee a fast data collection in a period of two weeks. The online survey was distributed with the help of a snowball sampling (Blumberg et al. 2005).
A total of 169 respondents participated in the survey (104 female, 65 male; Mage = 31, SD=13.98, range 19-72). Most of them were students (57.4%), or work either fulltime (21.9%) or part-time (16.0%). Their average frequency of visiting a restaurant for dinner was 18 times per year (SD=12.63).
They were randomly assigned to one of the six conditions. Table 1 shows the number of participants in each scenario. They participated voluntarily in a 3 (type of information) x 2 (involvement) between subjects factorial design. All constructs were measured on a seven-point Likert scale. Although pre-existing multi-items scales were used where appropriate, some items were adapted or added to suit the context of the study. The survey instrument also included manipulation checks to ensure that the attribute levels were perceived as intended.
Attribute-centric Benefit-centric Attribute- and benefit-centric
Low involvement N=28 N=30 N=26
High involvement N=29 N=29 N=27 Table 1: Number of participants per scenario
3.2
Procedure
high-involvement condition or low-high-involvement condition. Next, the participants were introduced to one type of review; the attribute-centric information review, the benefit-centric information review, or the review with both attribute- and benefit information. After reading, participants were asked to answer a set of questions related to their attitudes and purchase intentions towards the restaurant. When participants were ready, they answered a set of questions relating to their perceptions of the task goal, and their involvement. Also, they answered some questions related to their attitudes towards online reviews. Besides this, they answered some measures of individual differences like age, gender, and working life. Finally, the participants were thanked for their participation. It took approximately 10 minutes to complete the process. The whole questionnaire can be found in Appendix II.
3.3
Manipulations
3.3.1 Information typeType of information was manipulated by three different types of reviews; one with just attribute-centric information, one with just benefit-centric information, and one with both attribute- and benefit information. Previous studies within the restaurant industry came up with the following elements which are most important to consumers when selecting a restaurant (Almanza et al., 1994; June and Smith, 1987; Morgan, 1993): quality of food and method of preparation, type of food, speed of service, friendly servers, price, location, decor, atmosphere, and menu. Therefore, these elements have been incorporated in the three types of reviews. Besides the similar content, the length of the reviews was constant across conditions because it can affect quality and quantity (Chevalier and Mayzlin, 2006). Moreover, for all three types of reviews, the amount of positiveness is controlled as the same strength. Also, the amount of arguments was similar across conditions. The reviews can be found in Appendix I.
3.3.2 Involvement
Involvement was manipulated by two different types of dinner; a social occasion with friends and a formal business occasion. Within literature, a distinction has been made across four types of dinner (Auty, 1992): a celebration (birthday or anniversary), a social occasion, a convenience/ quick meal, and a business meal. It has been investigated that type of dinner is one of the main important variables that has influence on the restaurant choice. According to Quester and Smart (1998), situational involvement and personal relevance play a very important role here. Business versus leisure as a purchase occasion has been shown to effectively differentiate between high and low involvement in prior studies (Ostrom & Iacobucci, 1995; Sundaram et al., 1997). Consumers are highly involved for occasions with a high level of risk such as business meals and lower involved for a leisure dinner (Mattila, 1993). Since students show a marked tendency to eat out for social rather than for specifically celebratory occasions or convenience motivations (Auty, 1992), the focus in this study is on social dinners for the low-involvement condition. The high low-involvement condition will be manipulated as business meals. Figure 1 shows both manipulations.
Figure 1: First manipulations involvement
perceived involvement was 5.00 in the high-involvement condition, and 4.97 in the low-involvement condition. This was not a significant difference as indicated by Independent Samples T-Test (F=.24,
p<.94). Therefore, new manipulations for the high- and low involvement condition were developed.
Figure 2 shows the new manipulations.
Figure 2: Manipulations involvement
Measures Cronbach’s alpha Questions Measurement Involvement (Smith et al., 2005)
0.88 1. Als ik voor bovengenoemde situatie een restaurant moet uitkiezen dan is deze beslissing erg belangrijk voor mij.
2. Als ik voor bovengenoemde situatie een restaurant moet uitkiezen dan ben ik erg betrokken bij het maken van deze beslissing.
7-point Likert-scale (1=strongly disagree, 7=strongly agree)
Table 2: Cronbach’s alpha involvement
3.4
Dependent variables
3.4.1 Attitude3.4.2 Behavioral intention
Behavioral intention was measured on a 7 point Likert scale, derived from previous studies published in the information technology and marketing literature (Boulding et al., 1993; Cronin and Taylor, 1994). The questions were “How likely is it that you will recommend this restaurant to your friends?” and “How likely is it that you will visit this restaurant next time when you go out for dinner?”
3.5
Control variable
It is very important to take one’s attitude towards reviews into account, since this is an important individual difference variable for studies towards consumer behavior (Bearden et al, 1989; Valck et al. 2009), which indicates how receptive one is to opinions and experiences of others.
Attitude towards reviews was measured on a 7 point Likert scale in terms of four items, adapted from Park et al. (2007). Participants were asked to indicate to what extent they agreed with statements relating to online consumer reviews. (e.g. When I buy a product online, I always read reviews that are presented on the website, When I buy a product online, the reviews presented on the website are helpful for my decision making). The full questionnaire can be found in Appendix II.
3.6
Scales
The Likert scales used for attitude and behavioral intentions were tested on their internal consistency. Therefore, a Cronbach’s alpha was calculated, which is the most widely accepted formulation of reliability (Malhotra, 2007). Besides this, a factor analysis was performed, in order to check dimensionality.
3.6.1 Cronbach’s Alpha
Measures Cronbach’s alpha
Questions Measurement
Attitude (Petty et
al., 1986; Petty and Cacioppo, 1984; Mitchell, 1981)
0.93 Mij lijkt het beschreven restaurant: 1. Ongunstig – gunstig 2. Onbevredigend – bevredigend 3. Onaangenaam – aangenaam 4. Ontoepasselijk – toepasselijk 5. Onaantrekkelijk – aantrekkelijk 7-point Likert-scale (1=strongly disagree, 7=strongly agree) Behavioral intentions (Boulding et al. 1993, Cronin and Taylor, 1994)
0.78 1. Uitgaande van de informatie die ik heb gelezen is de kans groot dat ik dit restaurant bezoek als ik de volgende keer uiteten ga.
2. Uitgaande van de informatie die ik heb gelezen is de kans groot dat ik dit restaurant aanbeveel aan andere mensen. 7-point Likert-scale (1=strongly disagree, 7=strongly agree) Attitude towards online reviews (Park et al., 2007) 0.85 for all 4 questions 0.87 when dropping question 4
1. Als ik een product koop, lees ik altijd reviews over dit product 2. Reviews helpen mij bij het maken
van een aankoopbeslissing 3. Het lezen van reviews geeft mij
vertrouwen bij het doen van een aankoop
4. Als ik geen reviews lees voorafgaand aan een aankoop, maak ik mij zorgen over mijn beslissing
7-point Likert-scale (1=strongly
disagree, 7=strongly agree)
Table 3: Cronbach alpha Attitude, Behavioral intentions & Attitude towards online reviews 3.6.2 Factor analysis
A factor analysis is conducted in order to see whether the factors load on different components. As can be seen from table 4, for all variables, the KMO value, which examines the appropriateness of the factor analysis, is sufficient. Moreover, the eigenvalues, which represent the total variance explained by each factor, are greater than 1.0. Finally, the total variance explained also had a satisfactory level above the recommended 60% for all variables. Since all items loaded on one component, the items were averaged to compose an attitude-score, a behavioral intention-score and an attitude towards online reviews-score.
Variables KMO > 0.5 Eigenvalue > 1.0 % of variance explained
Attitude 0.85 3.94 78.84%
Behavioral intentions 0.51 1.64 81.91%
4 RESULTS
To examine if randomization between conditions was performed well, an Independent-sample Kruskal-Wallis test was used. The test showed that there was no significance difference in gender between conditions (p=.70). Moreover, there was no significant difference between groups in the number of times participants were visiting a restaurant per year, as was tested by means of an ANOVA test (F<1, p = NS). Thus, the randomization was successful.
4.1
Manipulation checks
Checks were carried out to ensure that the manipulations were successful. A one-way ANOVA test indicated a significant difference in means between the three information types. The test showed that the three conditions significantly differ (F=12, 29, p < .05), with a lower mean rating for attribute centric information (M=4.30, SD=2.26), a higher mean rating for benefit centric information (M = 8.80, SD=1.87), and a moderate rating for both attribute- and benefit-centric information (M= 6.77, SD=1.96). In conclusion, this manipulation was successful.
Involvement was measured in terms of two items. On a 1-7 scale, with higher numbers indicated higher involvement, the mean rating of perceived involvement was 6.31 in the high-involvement condition, and 4.22 in the low-involvement condition. This was a significant difference as indicated by Independent Samples T-Test (F=4, 67, p < .05). Thus, this manipulation also was successful.
The control variable general attitude towards online reviews was analyzed to see if there were significant differences among groups. No significant difference was shown (F (1,161) = .67, p < .42) Thus, this control variable was excluded in the following analysis.
4.2
Attitude
As 2 x 3 ANOVA indicated, there was no significant main effect for involvement and type information. There was a marginally significant interaction between the effects of involvement and type of information (F (2, 16) = 2.73, p = .07).
A Tukey’s HSD test was conducted to evaluate pairwise differences among means. For the low involvement condition, a marginally significant effect was found between attribute-centric information (M=5.35, SD=1.11) and both attribute- and benefit information (M=5.94, SD=1.00; LSD=.59, p=0.6) on attitude. Thus, for lower involved consumers, both types of information had a significantly greater effect on attitude and behavioral intentions than attribute-centric information. This was in line with the expectation that for low involvement, both types of information had more effect than attribute-centric information. No significant effects were found between attribute-centric (M=5.35, SD=1.11) and benefit-centric information (M=5.55, SD=1.28; LSD=.20, p=.50) and between benefit-benefit-centric information (M=5.55, SD=1.28) and both types of information (M=5.94, SD=1.00; LSD=.39, p=.21) for the low-involvement condition.
Moreover, a significant effect was found for both types of information between high involvement (M=5.38, SD= 1.29) and low involvement (M=5.94, SD=1.00; LSD=.56, p=0.7). Thus, lower involved consumers where more influenced by both types of information than higher involved consumers. This is in line with H3a; thus, H3a is accepted. No significant effects were found for attribute-centric information between high-involvement (M=5.72, SD=.82) and low-involvement (M=5.35, SD=1.11; LSD=.37, p=.21) and for benefit-centric information between high-involvement (M=5.80, SD=1.23) and low-involvement (M=5.55, SD=1.23; LSD=.25, p=.40).
Type of information Attribute-centric Benefit-centric Attribute + Benefit Main effect Involvement Low involvement M = 5.35 SD = 1.11 (N=28) M = 5.55 SD = 1.28 (N=30) M = 5.94 SD = 1.00 (N=26) 5.60 High involvement M = 5.72 SD = .82 (N=29) M = 5.80 SD = 1.23 (N=29) M = 5.38 SD = 1.28 (N=27) 5.64 Main effect 5.54 5.68 5.66
Figure 3: Graph visually illustrating differences in Attitude between scenarios
4.3
Behavioral Intentions
As 2 x 3 ANOVA indicated, there was a significant main effect for involvement ((F (1, 16) = .02, p = .03). No significant main effect was found for type information ((F (2, 163) = .02, p = .98). Also, no interaction effect was found ((F (2, 16) = 1.25, p = .29).
Although there was no significant interaction effect, an inspection of pattern of means suggested that the means for high- and low- involvement seemed to behave as predicted. Therefore, an explorative Tukey’s HSD test was conducted. This test showed that there was a significant difference between high involvement (M = 5.26) and low involvement (M = 4.54) for the attribute-centric condition (F (1, 16)5.08,
p = .03) on behavior. Thus, for consumers reading attribute-centric information, behavioral intentions
were more influenced when they were high-involved than when they were low-involved, which is in line with hypothesis H1b. No significant effects were found for benefit-centric information between high-involvement (M=5.16, SD=1.35) and low-high-involvement (M=4.67, SD=1.39; LSD=.49, p=.12) and for both types of information between high-involvement (M=4.87, SD=1.22) and low-involvement (M=4.87, SD=1.98; LSD=.01, p=.99). 5,35 5,55 5,94 5,72 5,8 5,38
Attribute Benefit Both
Type of information Attribute-centric Benefit-centric Attribute + Benefit Main effect Involvement Low involvement M = 4.54 SD = 1.23 (N=28) M = 4.67 SD = 1.39 (N=30) M = 4.87 SD = .98 (N=26) 4.69 High involvement M = 5.26 SD = 1.01 (N=29) M = 5.16 SD = 1.35 (N=29) M = 4.87 SD = 1.22 (N=27) 5.10 Main effect 4.90 4.92 4.87
Table 6: Mean scores on Behavioral intentions
Figure 4: Graph visually illustrating differences in Behavioral intentions between scenarios
4.4
Hypotheses
When looking at the hypotheses, hypothesis H3a, stating that both product attribute- and benefit-centric information have more influence on attitude for consumers with low involvement than for consumers with high involvement, was accepted. However, contrary to what was expected, this effect was not found for behavioral intentions. This means that H3b was rejected.
For H1b, stating that attribute information has more influence on behavioral intentions for consumers with high involvement than for consumers with low involvement, ambiguous support was found. Although there was no significant interaction effect, a specific contrast analysis showed a significant
4,54 4,67 4,87 5,26 5,16 4,87
Attribute Benefit Both
difference between high-involvement and low involvement consumers for attribute-centric information. For hypothesis H1a, stating that attribute information has more influence on attitudes for consumers with high involvement than for consumers with low involvement, no significantly differences were found between high and low involvement. Thus, H1a was rejected.
Also, the results for benefit-framed reviews were different than expected. For this condition, no significant differences were found between high and low involvement. Thus, hypotheses H2a and H2b, stating that benefit-information has more influence on attitude and behavioral intention for consumers with low involvement than for consumers with high involvement, were rejected.
H1a: attribute-centric information has more influence on attitude for consumers with high
involvement than for consumers with low involvement Rejected
H1b: attribute-centric information has more influence on behavioral intentions for consumers with
high involvement than for consumers with low involvement Ambiguous
H2a: benefit-centric information has more influence on attitude for consumers with low involvement
than for consumers with high involvement Rejected
H2b: benefit-centric information has more influence on behavioral intentions for consumers with low
involvement than for consumers with high involvement Rejected
H3a: both product attribute- and benefit-centric information has more influence on attitude for
consumers with low involvement than for consumers with high involvement Supported
H3b: both product attribute- and benefit-centric information has more influence on behavioral
intentions for consumers with low involvement than for consumers with high involvement Rejected
5 CONCLUSION AND DISCUSSION
The aim of this study was to investigate whether different types of information could influence consumers’ attitude and behavioral intentions and how these effects are moderated by consumer involvement. The study shows a significant type information x involvement interaction effect for attitude. Low involvement consumers were more affected by both types of information rather than attribute-centric information. This is in line with previous research (Maheswaran and Sternthal, 1990; Park et al., 2007), which states that low-involvement consumers are less persuaded through high-quality and objective data. Moreover, low-involvement consumers were more affected through both-types of information than high-involvement consumers. This is also in line with earlier research, stating that greater involvement heightens recipients’ tendencies to scrutinize the message content, depending on the argument quality (Chaiken, 1980; Petty and Cacioppo, 1984). Since the review consisted of both attribute- as benefit centric information, it could be assumed that persuasion was inhibited. Low-involvement consumers will process the information less critically, resulting in a greater effect on their attitudes.
Although, no support is found for the expectations that attribute-centric information has more influence on high-involvement consumers and benefit-centric information has more influence on low-involvement consumers. This lack of support could be explained by several reasons.
showed their participants eight different reviews. To avoid a high rate of dependency of one review, it is suggested to include more reviews during further research.
Second, the limited evidence for the proposed hypotheses could be explained through the assumption that elaboration occurs as a function of separable antecedents (Andrews, 1988). Recently, researchers have explored relationships among combinations of possible antecedent conditions to message elaboration. Besides involvement, they also controlled for consumers’ expertise (Petty and Cacioppo, 1981; Petty and Cacioppo, 1984; Petty, et al., 1983; Maheswaran and Sternthal, 1990; Park et al., 2007). The explanation for this can be found in ELM models. Consumers need both the motivation (referring to involvement) and ability (referring to expertise) to process. The separability of these concepts is based upon the following reasoning: "If a person is highly able to process a message but lacks the prerequisite motivation, little processing will occur" (Petty and Cacioppo 1986, p. 81). Conversely, if a message processor is motivated to process message content (due to personal relevance of an advertised product) but lacks the required ability (due to limited knowledge) and/or opportunity (due to distracting stimuli), little message elaboration will occur. Therefore, each individual antecedent condition serves only as a necessary, but not sufficient condition for message elaboration.
However, others have argued for conceptual distinctiveness between ability (due to expertise) and motivation (due to involvement). Zaichkowsky (1985, p.296) indicates that “expertise may not be necessarily related to involvement because involvement is a motivational construct whereas expertise is a sustaining construct representing knowledge structure. One does not necessarily have to be an expert in order to be involved with the product. However, involvement may motivate one to gather information and in time become increasingly knowledgeable about the product.” In line with this, Wright (1974, p. 194) argues that “involvement in actively processing information is largely a function of a person's recognition that the information has goal-satisfaction value to him. Involvement is thus grounded in the perceived meaning of the specific content. A message that is highly arousing to a receiver by virtue of its content may be transmitted in a form that restricts his opportunity for response, or vice-versa. Arousal to process and opportunity to process should, therefore, be treated as theoretically separate variables.”
however this variable could not be included as a moderator within this research. Analyses indicated a large dispersion among results which led to a skewed distribution of the six conditions.
Although it was anticipated that the same effects hold for attitudes as for behavioral intentions, no support was found for these effects. This discrepancy between attitudes and behavioral intentions could be explained through the role of prior knowledge and experience (Feldman and Lynch, 1988). Feldman and Lynch (1988) note that prior knowledge increases the likelihood that either the actual response measured or its immediate cognitive antecedents will be retrieved for use in guiding subsequent judgment or behavior pertaining to the same object. They state that if memories of beliefs, attitudes, intentions, or past behaviors exist, cues directing activation of any one of these can cause it to be the direct determinant of a judgment or behavior. Also Schwartz and Tessler (1972) noted that regular experience leads to a tendency for participants to infer their own attitudes toward acts, their personal and social normative beliefs, as well as their intentions for future behavior from past choices. For participants who have prior experience with the dinner occasion as described in the survey, might refer to their past experiences instead of using the given review. This could lead to contradictory behavioral intentions. For example, the participants could form favorable attitudes towards the restaurant mentioned in the review. However, when thinking about past experiences in similar occasions, they may still choose a restaurant they have visited in the past, although one’s attitude towards the reviewed restaurant is positive. This could result in discrepancies between attitude and behavioral intentions. On the other hand, it could be that some of the participants did not have experience with the manipulated dinner occasion. In that case, they form attitudes according values they perceived as important during their past experiences. Since they have never participated in an occasion like described within the survey, they have more difficulty with indicating their behavioral intentions, which also could lead to discrepancies.
5.1
Managerial Implications
The findings of this study have several managerial implications.
consumers and high-involvement consumers are important to marketers making a strategic plan (Kassarjian, 1981). Low-involvement consumers may not be very interested in buying products at the moment, but they should not be treated as “‘non-profitable consumers.” Instead, they should be treated as “potential consumers” for the long-term profit of companies. Therefore, marketers should adopt a variety of tactics to attract these potential consumers and stimulate their interest in a product. This means that online sellers need to deliver product information framed in a cognitively fitted way for consumers with different levels of involvement.
However, it is unrealistic for online sellers to provide a standardized review format for previous buyers because ‘‘word-of-mouth” messages are supposed to be informal and, therefore, format-free. OCRs are presented on the Internet without any standard format (Chatterjee, 2001), which means that consumers freely write about their experience with the product. Therefore, a different strategy is suggested; marketers should sort reviews depending on the type of reviews (attribute-focused reviews and benefit-focused reviews). Then, depending on the degree of involvement of the product, they should first show the reviews that match the level involvement of potential consumers. Park et al. (2007) suggest that involvement can be detected through click-stream data because on-line shopping tasks differ with the level of involvement. For example, the searching task is close to high-involvement action, while the browsing task is related to low-involvement action (Hong et al., 2004). Using click-stream data, online sellers can categorize consumers by level of involvement and show reviews that fit their involvement up front. Once sellers develop a system for recommending appropriate reviews to consumers involved at both high and low levels, the effects of online consumer reviews will be more convincing.
5.2
Limitations & Further Research
This study has some limitations that should be addressed.
First, this study incorporated just one review per condition on which participants had to base their attitude and behavioral intentions. In a normal setting, consumers will base their decisions on more than just one review (Lee et al., 2007) and will search on different websites. Moreover, the reviews that were presented were shown on 1 page without the possibility of browsing through the website. Therefore, it will be relevant to investigate whether comparable effects will occur under normal conditions
general, the number of positive reviews occupies a larger portion of total reviews. According to Resnick and Zeckhauser (2002), 99.1% of customer feedback on eBay in the late 1990s was positive. Also, Mulpuru (2007) found that more than 80% of the reviews were positive after evaluating 4000 reviews in the Electronics and Home & Garden categories on the Amazon.com site. Although the number of positive reviews largely exceeds the number of negative reviews, negative reviews are influential to consumers. Chevalier and Mayzlin (2006) noted that negative reviews have a greater effect than positive reviews on consumer decision making. Other researchers (Pavlou and Dimoka, 2006; Ba and Pavlou, 2002) found evidence on the stronger effect of negative comments compared to the positive ones. Besides this, it has been found that consumers will have a more positive attitude toward the product when they processed positive reviews that fit them, and a more negative attitude when processing negative reviews that fit them (Park and Kim, 2007; Petty and Cacioppo, 1983). Therefore, it will be interesting to investigate information type in the context of negative reviews. Furthermore, a combination of both positive and negative reviews could be another research area of interest, as this is a more realistic representation of reality.
Third, six elements were included in the reviews (e.g. price, service, menu, preparation, location, and atmosphere). Service was operationalized as speed of service, and preparation as durability of the ingredients. However, one can imagine that consumers may attach different values to these attributes, or even find other elements more important. Also, they might assign different values to different dinner occasions. For example, atmosphere could be a more decisive element for a business dinner, whereas price could be more decisive for a dinner with friends. Thus, although the manipulation checks were successful, it cannot be said with certainty that people have valuated the variables exactly the same as they would have when deciding to select a restaurant in reality.
Fourth, behavioral intentions were measured with only two items. Although the construct was extended from previous literature (Park et al., 2007), and demonstrated adequate reliability and construct validity, no results were found for this variable. Therefore, future research could be enhanced by using more items to measure the construct.
APPENDIX I
Reviews
1. Attribute-centric review2. Benefit-centric review
REFERENCES
Almanza, A., Jaffe, W. and L. Lingchun (1994) “Use of the Service Attribute Matrix to Measure Consumer Satisfaction,” Journal of Hospitality & Tourism Research, 17 (2), 63-75.
Andrews, J. C. (1988) “Motivation, Ability, and Opportunity to Process Information: Conceptual and Experimental Manipulation Issues,” Advances in Consumer Research, 15, 219-225.
Auty, S. (1992) “Consumer choice and segmentation in the restaurant industry,” The Service Industries
Journal, (3), 324-339.
Ba, S. and P.A. Pavlou (2002) “Evidence of the effect of trust building technology in electronic markets: price premium and buyer behavior,” MIS Quarterly, 26 (3), 243–268.
Bansal, H.S. and P.A. Voyer (2000) “Word-of-Mouth Processes within a Services Purchase Decision Context,” Journal of Service Research, 3 (2), 166-177.
Bearden, W.O., Netemeyer, R.G. and J.E. Teel (1989) “Measurement of Consumer Susceptibility to Interpersonal Influence,” Journal of Consumer Research, 15, 473-481.
Bickart, B. and R.M. Schindler (2001) “Internet forums as influential sources of consumer information,”
Journal of Interactive Marketing 15 (3), 179-195.
Blumberg, B., Cooper, D.R. and P.S. Schindler (2005) Business Research Methods. McGraw-Hill Education, New York.
Boulding, W., Kalra, A., Staelin, R. and V.A. Zeithaml (1993) “A dynamic process model of service quality: from expectations to behavioral intentions,” Journal of Marketing, 30, 7-27.
Bristor, J.M. (1990) “Enhanced Explanations of Word of Mouth Communications: the Power of Relationships,” Research in Consumer Behavior, 4, 51-83.
CBS (2009) Toerisme en recreatie in cijfers 2009. Available from:
http://www.cbs.nl/NR/rdonlyres/CF6D75E3-B5CC-4310-A782-0CAAAE88C9B1/0/2009g82pub.pdf [Accessed June 2012].
Celsi, R.L., and Olson, J.C. (1988) “The role of involvement in attention and comprehension processes,”
Journal of Consumer Research, 15 (2), 210–224.
Chaiken, S. (1980) “Heuristic versus systematic information processing and the use of source versus message cues in persuasion,” Journal of Personality and Social Psychology, 39 (5), 752–766.
Chaiken, S. (1987) “The Heuristic Model of Persuasion”. In M.P. Zhana, J.M Olson and C.P. Herman (Eds.)
Social Influence: The Ontario Symposium, 5, 3-39. Hilssdale, NJ: Lawrence Erlbaum.
Chen, Y., and Xie, J. (2005) “Third-party product review and firm marketing strategy,” Marketing Science, 24 (2), 218–240.
Chevalier, D.A. and D. Mayzlin (2006) “The effect of word of mouth on sales: online book reviews”,
Journal of Marketing Research 43 (3), 345–354.
Chu, S.C. and Kim, Y. (2011) “Determinants of consumer engagement in electronic word-of-mouth (eWOM) in social networking sites,” International Journal of Advertising, 30(1), 47-75.
Cohen, J. B. (1983), "Involvement and You: 1000 Great Ideas," Advances in Consumer Research, 10, Richard P. Bagozzi and Alice M. Tybout, eds., Ann Arbor, MI: Association for Consumer Research, 325-328.
ComScore (2007) Online consumer-generated reviews have significant impact on offline purchase behavior, Press release, 2007. Available from:
http://www.comscore.com/press/release.asp?press=1928 [Accessed August 2012].
Cronin, J.J. and S.A. Taylor (1994) “SERVPERF versus SERVQUAL: reconciling performance-based and perception-minus-expectations measurement of service quality,” Journal of Marketing, 58, 125-131. Dellarocas, C. (2003) “The digitization of word of mouth: Promise and challenges of online feedback mechanisms,” Management Science 49(10), 1401–1424.
Dellarocas, C. (2006). “Strategic manipulation of internet opinion forums: Implications for consumers and firms,” Management Science, 52(10), 1577–1593.
Dellarocas, C., Xiaoquan M. Zhang, Neveen Awad. (2007) “Exploring the value of online product reviews in forecasting sales: The case of motion pictures,” Journal of Interactive Marketing 21(4), 23–45.
Edell, J. A. and R. Staelin (1983), "The Information Processing of Pictures in Print Advertisements,"
Journal of Consumer Research, 10, 45-61.
Feldman, J.M. and J.G. Lynch (1988) “Self-Generated Validity and Other Effects of Measurement on Belief, Attitude, Intention, and Behavior,” Journal of Applied Psychology, 73 (3), 421-435.
Fishbein, M., & Ajzen, I. (1974) “Attitudes toward objects as predictors of single and multiple behavioral criteria,” Psychological Review, 81, 59-74.
Fishbein, M., and Ajzen, I. (1975) “Belief, attitude, intention, and behavior: An introduction to theory and research”. Reading, Mass.: Addison-Wesley.
Foodservicemarkt Nederland (2012) Available from:
http://www.foodservicexs.nl/index.php?option=com_docman&task=doc_view&gid=39&Itemid=34 [Accessed August, 2012].
