Discrepancy and Social Pressure Effects on People’s Online Consumer
Review Intention
University of Groningen Faculty of Economics and Business
Department of Marketing Nettelbosje 2 9747 AE Groningen Author Robin Pot Student number: 1611917 Ceintuurbaan 47-5 1072 ET Amsterdam +31 6 28 57 28 08 robinpot@gmail.com Supervisors
1st Supervisor: Dr. J.A. Voerman Co-assessor: Dr. H. Risselada
2
I
Management Summary
The purchase process consists of multiple phases of which the post-purchase phase is investigated here by specifically focusing on people’s intention to post an online consumer review this was done by focusing on two antecedents that were likely to have an effect: the degree of discrepancy between pre-purchase expectations and post-purchase performance; and social pressure. Controlling for the effects that people’s age and experience with posting online reviews might have, an ANCOVA is performed in order to gain more insights into people’s online review intention. The final phase of the purchase process is important as post-purchase
evaluations and recommendations can have a profound effect on sales and future profits of businesses.
This study was performed by using an online survey mechanism where people were randomly assigned to eight possible scenarios based on different combinations of the degree of
discrepancy and social pressure.
The results from this study indicate that there is non-significant evidence that people differ in their online consumer review intention based on the degree of discrepancy between pre-purchase expectations and post-pre-purchase performance. The results signal however that people who have their expectations confirmed (small or no degree of discrepancy) show higher intentions to post online reviews than people who have their expectations disconfirmed (high degree of discrepancy). This effect was strongest when people had high expectations and had their expectations confirmed by high performance.
Social pressure in addition was found not to be significant as well. This is because this study defined social pressure as either positive or negative reviews from other people being present. The results did signal however, that people who see other people’s positive reviews have a slightly higher intention to write a review.
What was found to be a significant variable that influences online consumer review intention is people’s experience with giving such reviews. Age in this sense proved to be of no significant influence on any of the relationships in the model.
3
II
Preface
So it finally happened, finishing my master thesis at the end of a chapter. Sounds cliché, right? Well it is! Starting a new chapter, life as a working man, the serious stuff. People say, it is just like marriage, your life is over. My time in Groningen was indeed great and I have learned a lot. I have learned what I can, what I cannot and especially what I like and do not like. Starting my Groningen chapter with a Bachelor degree in International Business, I have now completed this chapter with a Master degree in Marketing.
The chapter knew some interesting paragraphs full of stories and experiences which I will never forget. Towards the end of the chapter I have made a decision which some people would
probably would not make. In January, I decided to step into a new startup business and immediately I was working overtime. Early mornings and late evenings as you might expect from a startup. Logically, finishing the thesis became a challenge.
Luckily, I had a lot of support from the people surrounding me. My girlfriend, my parents, my colleague, of which especially my girlfriend gave me the time and space to combine both a full time job and writing a thesis. Long days and short nights were the result of my decision, but hard work have structured the business and thus my live. Special thanks go to my supervisors who have guided my throughout this thesis process.
4
Contents
I Management Summary ... 2
II Preface ... 3
1. INTRODUCTION ... 6
1.1. The Purchase Process ... 6
1.2. Online Consumer Reviews ... 8
1.2.1. Discrepancy Effect on Online Consumer Review Intention ... 8
1.2.2. Social Pressure Effect on Online Review Intention ... 9
1.2.3. Effects of Gender, Age and Experience ... 10
1.3. Problem Statement and Research Question ... 11
1.4. Research Relevance and Structure ... 12
2. Literature Review ... 13 2.1. Degree of Discrepancy ... 13 2.1.1. Pre-Purchase Expectations ... 13 2.1.2. Post-Purchase Performance ... 14 2.1.3. Discrepancy ... 14 2.1.4. Asymmetric Effect ... 16 2.2. Social Pressure ... 16 2.3. Background Characteristics ... 18 2.3.1. Gender... 19 2.3.2. Age ... 20 2.3.3. Experience... 20 2.4. Conceptual Model ... 21 3. Methodology ... 22 3.1. Experimental Design ... 22 3.1.1. Scenario Manipulation ... 23 3.2. Data Collection ... 23 3.3. Operationalization ... 24 3.3.1. Degree of Discrepancy ... 24 3.3.2. Social Pressure ... 25
3.3.3. Online Consumer Review Intention ... 25
3.3.4. Age ... 26
3.3.5. Gender... 26
3.3.6. Experience... 26
3.4. Pre-testing ... 27
3.5. Population and Sample ... 27
3.6. Randomization ... 27
3.7. Reliability ... 28
5 3.8.1. Manipulation Checks ... 29 3.8.2. Main Analysis ... 33 4. Results ... 34 4.1. General results ... 34 4.2. Homogeneity of Slopes ... 35 4.3. ANCOVA ... 37 4.4. Results Summary ... 41
5. Discussion and Conclusions ... 43
5.1. Discussion ... 43 5.2. Conclusions ... 43 5.2.1. Degree of Discrepancy ... 43 5.2.2. Social Pressure ... 44 5.2.3. Experience... 44 5.2.4. Age ... 45 5.3. Managerial Implications ... 45
5.4. Limitations and Further Research ... 46
6. References ... 48
7. Appendix ... 54
A: Example Survey (Dutch) ... 54
B1: Homogeneity of Slopes... 61
6
1. INTRODUCTION
The Internet has evolved into a complex and dynamic channel used by consumers every day. Online shopping has increased enormously over the last decade. In 2004 for example, only 46% of consumers was shopping online whereas in 2009 it was estimated that nearly 64% would be shopping online in 2010 (Delafrooz, 2009). Important to recognize is that the Internet has given the consumer a voice. Consumers can now post information on the Internet the same way as companies do. In other words, the consumer has the possibility to contribute to the body of online information with their own user generated content (UGC). UGC is defined as content
“created or produced by the general public rather than by paid professionals and primarily distributed on the Internet” (Daugherty et al., 2008). The fact that consumers can do this is
important when looking at online shopping. Besides having the opportunity to post their own content, consumers have the chance to read the content provided by other consumers and use this for purchase decision making. This development has an effect on the purchase process consumers go through when they are buying a new product (Ansari, Mela and Neslin, 2008). Therefore, more and more retailers have added the Internet as an extra retail channel. More specifically, retailers have done so because they are trying to satisfy the growing needs of customers to buy through multiple channels (Van Bruggen et al, 2010).
1.1. The Purchase Process
The Internet as a channel can be used by consumers to search for information about the product they want to buy (Verhoef et al, 2007). Consumers subsequently purchase the selected product and even evaluate their purchase decisions. This can also be done through the Internet. Where one customer might begin searching for information on a website and purchases something in a store, another customer might go to a store and finally order a product on a website. Both customers can subsequently meet each other on a social media channel evaluating the product they both bought by posting a review online. This scenario indicates that customers walk a specific path when purchasing a product. This purchase process is defined by Shankar et al. (2011), as comprising “different stages such as motivation to shop, search, evaluation,
7 focus of marketing is to obtain insights into the total purchase process of consumers. For consumers, the Internet as such has altered this process.
Generally, the literature about the purchase process identifies three different phases in the shopping process: the pre-purchase phase, the purchase itself, and the post-purchase phase (Van Bruggen et al., 2010; Shankar et al., 2011). In the first phase, the consumer devotes time and space to the search for information about products, brands and retailers. Subsequently, the consumer selects the product and related brands he or she wants to purchase. When this is done, the consumer makes the purchase and becomes a customer. Most of the literature on multi-channel retailing for instance, discusses issues for only the first two phases, the pre-purchase and the pre-purchase phase. In other words, their research concerns a path-to-pre-purchase, but fails to go beyond the point-of-purchase. Verhoef et al. (2007) for instance, investigate the so called ‘research-shopper phenomenon’ where they concluded that some consumers use a different channel for searching for information than they do for actually purchasing a product or service. A clear distinction is made between shopping activities. What customers do after their purchase does not come forth. The same goes for Konuş et al. (2008), who distinguish between a pre-purchase and a purchase phase in their study about multi-channel retailing. In both these studies, the focus is on multichannel shopping where consumers might start searching for information about products via for instance an Internet website, but end up making the purchase in a store. However, what the above studies and many other scholars fail to incorporate is the third phase of the purchase process, namely the post-purchase phase. Shankar et al. (2011) state that the post-purchase phase is important as it includes consuming the product, evaluating and recommending it and eventually repurchasing the product.
8
1.2. Online Consumer Reviews
Online consumer reviews are an important part of user-generated content. The amount of online consumer reviews of products has increased significantly over the past couple of years. The impact it has had on commercial activities in an online context has been profound (Forman et al., 2008). Online consumer reviews in this study are defined as “peer-generated product evaluations
posted on company or third party websites” (Mudambi and Schuff, 2010). Examples of company
websites are Amazon.com or the Dutch Bol.com, whereas third party websites refer to any social media platform, such as Facebook and Twitter and evaluation and recommendation websites. Online consumer reviews about products are mostly posted in the form of a numerical rating (ranging from 1 to 5 stars). Additionally, customers have the possibility to write a personal piece of text with the customer’s comments (Mudambi and Schuff, 2010). Earlier research that focused on online consumer reviews identifies the different consequences of such online reviews. Studies like the research done by Forman et al. (2008) for example, focus on the impact they have on future company sales. They state that it has been generally assumed that one primary reason for such influences on sales is that online consumer reviews provide specific knowledge about the retailer and her products. It is therefore valuable for customers that are searching for information about something that they want to purchase (Forman et al., 2008). That, for instance, is the main reason for why consumers use rating websites during their search for information (Dabholkar 2006). Other customers’ evaluation of products serve as a basis for consumers’ decision-making process.
Other research into online consumer reviews covers the effects of specific characteristics of the online reviewer (Forman et al. 2008; Smith et al. 2005). It is argued that such message source characteristics influence the decisions that searching consumers make. Consequently, online consumer reviews that have been written by people with similar characteristics have an effect on the searching consumers. Here, self-selection bias is possible, when customers go looking for reviews written by people with similar characteristics (Hu et al. 2008; Li and Hitt 2008). Forman et al. (2008) consider the effect of what people say about themselves in an online review on peer recognition or future sales. Literature about information processing suggests that attributes of an information source have powerful effects on the way people respond to messages (Kelman 1961; Chaiken 1980; Mackie et al. 1990). Thus, online reviews can hold both product information and information about the reviewer where both types have an effect on future sales.
1.2.1. Discrepancy Effect on Online Consumer Review Intention
9 sales, it seems worthwhile to investigate what makes people post product reviews on a website. As said earlier, people evaluate their purchase of a product and are either satisfied or unsatisfied with the product’s performance (Gilly and Gelb, 1980). After a purchase, the product’s performance is contrasted against the expectations consumers had about a product’s performance prior to purchasing the product. Generally, the expectations of a product and the performance of meeting these expectations, result in satisfaction or dissatisfaction with the product. According to Engel et al. (1978) this is referred to as the degree of discrepancy between pre-purchase expectations and post-purchase performance. When the discrepancy is negative, consumers may decide to complain about this (Kraft, 1977; Landon, 1977). When the degree of discrepancy is positive, consumers may feel delighted. According to Oliver (1977), consumer delight is the emotional reaction that comes from being surprised about the positive disconfirmation of expectations. As other authors say, it is the typical emotional response in a situation where unexpected value, or performance, is perceived (Zeithaml, et al, 1993). In this case, this refers to deciding to write a negative online review about the purchase. It is therefore interesting to look at the effect of this discrepancy on online consumer review intention, which is the dependent variable in this study.1
1.2.2. Social Pressure Effect on Online Review Intention
The discrepancy that can exist between a consumer’s expectations about a product’s performance and the eventual product performance could be a basis for a consumer to write a review online. However, from literature about behavior, intention is said to be determined by other factors as well. One of them is the social factor as explained in the theory of planned behavior (Ajzen, 1991). From this theory in general, it becomes apparent that the intention to perform a certain behavior is a strong predictor of people’s actual behavior. Ajzen’s theory was built with the purpose to be able to predict behavior in specific situations, looking at more than only the aggregation principle. The aggregation principle holds that people’s general attitudes and their personality traits predict general behavior (aggregated across different situations). In the theory of planned behavior, the author wanted to focus on explaining specific behavior, incorporating several scenario-specific factors that help explain intention that consequently
1
10 predicts behavior. One of these antecedents of intention is called subjective norm and seen as a behavior-predicting variable in specific situations. Subjective norm is conceptualized by the author as perceived social pressure on performing a certain behavior. In the light of this study, this indicates that the intention to post an online review is considered to be directly influenced by the social pressure that people experience when forming an intention to post an online review. Subsequently, for this research, the subjective norm from the theory of planned behavior is important to incorporate as the variance in the online consumer review intention will likely be explained by social pressures.
Currently, the Internet and specifically social media has been at the focal point of attention. In the post-purchase phase of the purchase process, people can post reviews online where and whenever they want. Specific product evaluation and recommendation sites are available and an increasing amount of social media channels offer the possibility to post such user generated content. As a lot of people interact with each other through these media, the social element is present here. As said in the introductory part about online consumer reviews, consumers use rating websites during their search for information (Dabholkar, 2006).This information could influences people’s thoughts about the product or service as other people’s opinions are a form of social pressure. Zhu and Xiaoquan (2010) state that people’s online purchase decisions are influenced by the online consumer reviews posted by other consumers. Research from Moe and Trusov (2011) discusses this and the authors claim that the reviews people write are affected by other people’s reviews.
When people thus have an intention to post an online review, they can come across several prior written reviews about the same product or service, which could affect their intention to post an online review. Therefore, it is interesting to investigate if social pressure indeed has an effect on the intention to post an online consumer review. In this study, the term ‘social pressure’ is used to identify the second independent variable in our model.
Reviews can be negative or positive about a certain product or service. Therefore, in this study social pressure refers to the direction of social pressure, either positive or negative.
1.2.3. Effects of Gender, Age and Experience
11 intention and the effect that social pressure has on online review intention. Venkatesh et al (2003) looked at other theory and models about user and technology acceptance and the intention to start using information systems. They say that gender makes a difference when forming an intention (Venkatesh et al. 2000). Males are said to have stronger intentions in certain situations than do females. In other situations it could be the other way around. Furthermore, they say that experience is a strong factor as people learn over time and find it increasingly easy to use a system or technology. Prior experience makes repeated behavior easier. People’s age is finally another variable that needs to be accounted for as difficulty in processing stimuli and focusing on a technology becomes more difficult over the years (Plude and Hoyer 1985). Using the Internet to post an online review might be more difficult for older people.
For this research, it is assumed that there is a relationship between the degree of discrepancy between pre-purchase expectations an post-purchase performance and online consumer review intention and between social pressure and online consumer review intention, taking into account that people differ in age, gender and experience with giving online consumer reviews. The next two paragraphs will focus on the problem statement and the relevant research question as well as the outline for the rest of this study.
1.3. Problem Statement and Research Question
12 In order to investigate the relationships described, a research question is formulated:
“What is the impact of the degree of discrepancy between pre-purchase expectations and post-purchase performance, and the impact of social pressure, on online consumer review intention, taking into account that people differ in age, gender and their experience with giving online consumer reviews?”
1.4. Research Relevance and Structure
This study is relevant for online marketing practitioners as it will contribute to the body of research about post-purchase online consumer reviews. The post-purchase phase of the purchase process has been an understudied topic in the marketing literature. This phase is especially important as post-purchase behavior affects future re-purchase decisions of consumers. The post-purchase phase is among other characterized by evaluating products and services and recommending them to others (Shankar et al., 2011). Consumers have the possibility to post online reviews about their product or service experiences which are a basis for other consumers to base their purchase decisions on (Zhu and Xiaoquan, 2010). For marketing managers, this is important to realize as this post-purchase phase behavior thus has an effect on future sales (Forman et al., 2008). Negative word-of-mouth negatively affects other consumers and thus negatively affects the profits of firms.
In order to predict online consumer review behavior, marketing managers should look at the intention that people have for posting an online review. This study focuses on this online consumer review intention and provides valuable insights into the differences in this intentions and where they come from. Consequently, as intention is a good predictor of behavior, this research will help to investigate online consumer review behavior by looking at the antecedents of online consumer review intention. Subsequently, readers of this research report may use the results to contribute to the creation of new marketing tactics that can help improving business or products and services.
13
2. Literature Review
In order to build a solid and relevant foundation for this research, the independent variables and the background characteristics that influence online consumer review intention are conceptualized here. The first paragraph focuses on the explanation of the degree of discrepancy as an independent variable. Subsequently, the second paragraph explains social pressure as the second independent variable. Furthermore, the third paragraph conceptualizes the three moderating background variables that are used, namely age, gender and experience with online consumer reviews. After each paragraph, hypotheses are formulated that will support the research design in chapter three. Finally, the variables are depicted in a conceptual model which will graphically display the relationships between the variables.
2.1. Degree of Discrepancy
Before people purchase a specific product, they have expectations about what the product will be or do for them (Woodruff and Gardial, 1996). Moreover, the consumer goes through the post-purchase phase by comparing prior expectations to perceived reality. It might happen that the performance of a product or a service does not meet a consumer’s prior expectations. Consequently, a difference, or discrepancy will occur between pre-purchase expectations and post-purchase performance. This resulting gap is referred to as the disconfirmation of expectations (Rust and Oliver, 1994). This refers to a famous paradigm, called the Expectancy-Disconfirmation Paradigm, which implies that the pre-purchase expectations become a standard to which post-purchase performance is contrasted (Reisinger and Waryszak, 1996; Tribe and Snaith, 1998). Furthermore, it suggests that pre-purchase expectations and post-purchase performance determine consumer satisfaction (Oliver, 1980) which in turn influences people’s post-purchase behavior such as word of mouth or recommendation behavior.
In order to conceptually structure the degree of discrepancy properly, pre-purchase expectations and post-purchase performance are defined.
2.1.1. Pre-Purchase Expectations
14 (LaTour and Peat, 1979). As Woodruff et al. (1983) indicate, expectations are the desired standards that consumers would like to see from a product’s or service’s performance. Expectations thus are the standards on which consumers build their performance evaluation. In sum, prior research shows that expectations in the pre-purchase phase come from people’s perceptions of what a certain brand will do, how a certain category of products will perform and what their ideal product is. Ultimately, pre-purchase expectations have an effect on satisfaction and post-purchase behavior but the actual post-purchase performance evaluations are important to realize as well (Gupta and stewart, 1996).
2.1.2. Post-Purchase Performance
Post-purchase performance can be defined as a composition of relevant benefits or attributes that a consumer perceives (Cadotte et al., 1987). In order to be able to better understand satisfaction and ultimately consumers’ post-purchase behavior, post-purchase performance needs to be evaluated. Churchill and Suprenant (1992) for instance, say that post-purchase performance directly affects satisfaction. They suggested that pre-purchase expectations may be less important in comparison with post-purchase performance evaluations. However, this was in the light of consumers being highly familiar with the product and the probability of having expectations highly similar to performance (Johnson and Fornell, 1991). Even though some researchers argue for more importance on pre-purchase expectations and others argue for the higher relevance of post-purchase expectations, the role of the size of the gap between both expectations and performance is what is investigated in this study as research shows its role in determining the post-purchase behavior (Oliver, 1980). Here, it is investigated what the effect is of the degree of this gap on people’s post-purchase behavior, in this case, their behavior to review their product or service experience. This is done by focusing on people’s online review intention.
2.1.3. Discrepancy
Comparing pre-purchase expectations with post-purchase performance may result in a degree of discrepancy, a difference in what is reality and what was expected. This discrepancy can be positive or negative. Positive discrepancy occurs when people have low pre-purchase expectations and experience high post-purchase performance whereas negative discrepancy occurs when people have high expectations and low performance. When no discrepancy exists, people have their expectations confirmed whether they were low or high.
15 disconfirmation of expectations on product ratings. His results indicated that people who have their expectations confirmed rate a product significantly higher than those who have their expectations disconfirmed. In the latter situation, a discrepancy between expectations and performance occurs. He, however, tested only two different scenarios. He did not make a distinction between confirmation when expectations are low and when expectations are high. Furthermore, he did not investigate what the effects are on product ratings when people have their expectations positively disconfirmed. In short, negative disconfirmation was found to result in lower product ratings than when confirmation occurred. These results were confirmed by Cohen and Goldberg (1970). Later, Olshavsky and Miller (1972) saw the flaw in Cardozo’s research and tested the effects on product ratings when positive disconfirmation (low
expectations versus high performance) occurred. Their research design was a 2 x 2 matrix (high or low expectations versus high or low performance). Their results indicated that people base their product ratings more on what they expected than what they experience afterwards. This is the case when positive or negative disconfirmation occurs.
The resulting satisfaction or dissatisfaction that comes from a certain degree of discrepancy will influence people’s post-purchase behavior and thus the way they act (Andreasen 1976; Gupta and Stewart, 1996). The theory of planned behavior (Ajzen, 1991) states that intention is a good predictor of behavior which indicates that satisfaction or dissatisfaction based on the degree of discrepancy links to intention to perform behavior. From Nyer (2000) for instance, it is found that people who have experienced dissatisfaction with a product or service, could feel less dissatisfied when they show their anger or frustrations which might trigger them to share their opinion with others. Hence, online reviews could be their vent for showing others how they feel. This indicates a link between satisfaction or dissatisfaction and word-of-mouth (WOM). This link is studied intensively by Anderson (1998) and others, who state that satisfied customers will engage in favorable recommendation behavior (Bitner, et al., 1990; Oliver, 1980; and Reichheld and Sasser, 1990).
16
H1: A high degree of discrepancy between pre-purchase expectations and post-purchase performance has a positive effect on people’s intention to write an online consumer review.
2.1.4. Asymmetric Effect
However, there is evidence to be found that some asymmetry exists regarding the effect of discrepancy on online customer review intention. In an article by Schifferstein et al. (1999) the asymmetry was found that positive disconfirmations is regarded as ‘‘gains’’ and negative disconfirmations as ‘‘losses’’. People generally respond differently to a loss than to a gain. From Kahneman and Tversky (1979) it is found that the effect of losses have a larger effect on people than the effect of winning. Thus, if a feeling of experiencing discrepancy as a loss occurs, the effect on buying intention was found to be stronger than when a feeling of experiencing discrepancy as a gain. Therefore, one can assume that a negative disconfirmation of expectations (HE-LP) will have a stronger effect on online consumer review intention than when positive disconfirmation of expectations (LE-HP) will occur. Consequently, the following hypothesis is added:
H1a: The effect of the degree of discrepancy between pre-purchase expectations and post-purchase performance on people’s intention to write an online consumer review will be stronger when expectations are high and performance is low (HE-LP) than when expectations are low and performance is high (LE-HP).
The following paragraph discusses the second independent variable used in this study, namely social pressure. After its conceptualization, a hypothesis is formulated for later testing.
2.2. Social Pressure
Here, social pressure is discussed as the second independent variable in this study that is assumed to influence people’s intention to post a review online. As consumers are influenced by other people’s reviews when making purchase decisions, it is expected that the social element plays its part when people want to evaluate or recommend their product or service experience online. Social pressure has increasingly been studied, but the literature about social pressures in online reviews is limited and with mixed results (Sridhar and Srinivasan, 2012). However, as comes forth in their article, the insights that come from the effects that social influences have on online ratings, have high managerial relevance.
17 online product reviews. These reviews can additionally be used as information sources by other consumers (Hoyer & MacInnes, 2008; Lee et al., 2008). Social media platforms and third-party recommendation websites hence are increasingly being used to search for information about products. Schlosser (2011) indicates this by pointing out that almost all respondents that they have studied said they look at other people’s reviews before they purchase something. Moe and Trusov (2011) support this by saying that consumers increasingly rely on what other people have written about the product or service.
Regarding social pressure, consumers who are about to post an online review about their experience with a product or service, could come across other people’s reviews. Consequently, these prior reviews can influence these consumers (Moe and Trusov, 2011). When a consumer posts an online review, by default he or she becomes part of the social group of online reviewers as stated by Hennig-Thurau (2004). This consumer is an opinion leader whose product or service review is used by other consumers for more information. Still, and equally important, this opinion leader consumer could come across other reviews which will likely influence his own review. Basically, consumers make purchase decisions based on what ideas they themselves have and the opinions from other people’s reviews (Rashotte, 2009).
When looking at for instance the uses and gratification theory, the social effects are implied to explain what motivates consumers to choose certain media channels (Diddi and LaRose, 2006). One of these media channels is the Internet, which can be used for sharing user-generated content with other consumers, especially through social media. Hence, studies on why people contribute their personalized content on social media platforms show that social factors are the main motivations for content contribution.
This social element is also present in the theory of planned behavior, from which it can be seen that one of the factors that can explain intentions is called subjective norm. This term refers to the social pressure that people perceive to act in a certain way or perform a specific behavior (Ajzen, 1991). As a general rule, the author indicates that the more favorable the subjective norm, the stronger the intention to perform the behavior. This refers to the direct link between social pressure and intentions. From social influence theory, a similar concept as subjective norm can be found which is conformity (Neuberg et al., 2009). This can be confirmed as it is stated that the actions of other consumers have a profound effect on behavior of other people. These other people are said to be in the same social group. In this case, this group is referred to as the group who reviews online.
18 included two types of interactions between social pressure through other people’s product ratings and the product rating itself. First, they focus their model on the moderating effects of social pressure on the effects that several independent factors have on online reviews. These factors are the features of the product experience and possible product failure or product recovery. Second, they focus on the direct link between other consumers’ ratings, thus social pressure, and the eventual product rating of the consumer in question. From earlier literature this direct link is supported by research from Fromkin (1970) who said that people create an opinion based on the consensus from within a group. This direct link is of interest for this research as it does not focus on specific functions or attributes of a product or service experience.
Based on the direct link between subjective norm and intention (Ajzen, 1991) and the direct link between social pressure from other consumers’ reviews on the actual online consumer ratings (Sridhar and Srinivasan, 2012), this study assumes that a direct link exists between social pressure and online consumer review intention. The theory of planned behavior indicates that the stronger the subjective norm, the stronger the intention to perform a behavior. Social pressure in this study are other people’s previously posted online consumer reviews visible consumers write their own review. In that light, the presence of these reviews should have a direct positive effect on online consumer review intention. The hypothesis accordingly, becomes:
H2: Social pressure has a positive effect on people’s intention to write an online consumer review.
2.3. Background Characteristics
Several background characteristics are found that could have an effect on the relationships between the degree of discrepancy and online consumer review intention. The same holds for the effect they might have on the relationship between the degree of discrepancy and online consumer review intention.
19 It thus becomes interesting to find out whether different people respond differently to a certain degree of discrepancy.
2.3.1. Gender
Obviously, men and women differ from in each other in numerous ways. For this study, it is assumed that the effects that the degree of discrepancy and social pressure have on online consumer review intention are different between men and women. Regarding social influences for instance, it is found that women react more strongly to opinions of others (Miller, 1976; Venkatesh et al., 2000). In other words, it is expected that the effect of social pressure on online consumer review intention is stronger when the consumer is a woman.
The above is based on technology acceptance. The technology acceptance model (TAM) designed by Davis (1989), tries to predict adoption of technological systems and ultimate usage on a person’s job by focusing on intention. Here, the perceived ease of using an information system is indicated to predict usage intention together with the perceived usefulness. From Venkatesh et al. (2003) it becomes apparent that perceived ease of using such a system differs between men and women. Women are said to be more capable of guessing the amount of effort that using or adopting a system takes. Men on the other hand are more task-oriented meaning the perceived usefulness of a technology system will be clearer for them than for women (Venkatesh and Morris, 2000).
A revised version (TAM2) of this model (Venkatesh and Davis, 2000) later included subjective norm as a form of social pressure that can predict intention to use. Since using the Internet as an information system in order to be able to post online reviews, looking at technology acceptance models seems appropriate. In the TAM2 model, the authors included subjective norm in order to better predict usage intention. Regarding subjective norm, or social pressure, it is found that the effect on intention is stronger for women than for men.
Based on the above, it becomes clear that gender has an effect on the relationships between factors that affect intention to use. In the light of this study it is interesting to see what the effects are of gender on the relationships between the independent variables and the dependent variable. Accordingly, the hypotheses are as follows:
H3: The effect of the degree of discrepancy between pre-purchase expectations and post-purchase performance on people’s intention to write an online consumer review will be moderated by gender.
20 2.3.2. Age
Levy (1988) indicated that studies that refer to differences between men and women, often have misleading results because they do not account for differences in age. As for gender, age might help to explain differences in online review intention. Regarding age and social pressure for instance, Rhodes (1983) concluded that the older people are, the more importance they place on social effects. In other words, they are more likely to affiliate with others (Venkatesh and Morris (2000). Furthermore, it is found that age has an effect on the perceived ease of posting a review. In addition, Plude and Hoyer (1985) found that processing difficult stimuli and allocation attention to the right aspects becomes harder when people age. The perceived usefulness of information systems that allow posting online reviews becomes less clear. Therefore, it is hypothesized that age negatively moderates the effect of social pressure and the degree of discrepancy on online review intention. The supporting hypotheses then become:
H5: The effect of the degree of discrepancy between pre-purchase expectations and social post-purchase performance on people’s intention to write an online consumer review will be negatively moderated by age.
H6: The effect of social pressure on people’s intention to write an online consumer review will be negatively moderated by age.
2.3.3. Experience
From the literature, it is found that experience is an important variable as well that needs to be incorporated in this research. This is because of a ‘learning-effect’ that becomes apparent when time passes. In other words, the experience that people build up with posting online consumer reviews might have an effect on intention to post an online review. With repeated usage for instance, media usage behavior is less subject to active self-reflection in order to conserve mental resources (Diddi and LaRose, 2006). In other words, less cognitive action (thinking processes) occurs when experience is higher. Therefore, the third and final co-varying variable in the model will be tested with the use of the following hypotheses:
H7: The effect of the degree of discrepancy between pre-purchase expectations and social post-purchase performance on people’s intention to write an online consumer review will be negatively moderated by experience.
21 In order to have a complete overview of all the interactions between the variables, the analyses in this study will check if there is any interaction between the two factors as well. The hypothesis becomes:
H9: There is no interaction between social pressure and the degree of discrepancy
The next part will graphically display the relationships as assumed in this chapter.
2.4. Conceptual Model
The previous paragraph presented a literature overview and resulted in the identification of several hypotheses that are shown in figure 1 which depicts the conceptual model used in this study. ∆Discrepancy
OCRI
Social
Pressure
ExperienceGender
Age
H3 H5 H7 H2 H1 H4 H6 H8Figure 1: Conceptual Model
22
3. Methodology
In this chapter, the methodology used is explained by outlining the research that has been done and describing the data that resulted from a distributed survey. In order to ensure the appropriateness of the measures and scales used, the validity and the reliability are checked. Subsequently, the scales are checked regarding their internal consistency. In order to do this, a factor analysis is performed referring to the Cronbach’s alpha. The first paragraph describes the experimental design, followed by the explanation of the methods used to collect the necessary data for this study. The third paragraph covers the operationalization of the variables where paragraph four explains pre-testing the survey. Following, the subsequent paragraph describes the population and sample used for analysis. Furthermore, reliability of the scales that are used to measure the hypothesis are discussed and finally a plan of analysis is presented.
3.1. Experimental Design
The research done in this study is characterized as a descriptive conclusive study. According to Malhotra (2010), a research of this kind is structured by using samples that are large enough and representative for the population under study. Quantitative analyses are subsequently performed on the data that results from the sample. These analyses are consequently the basis for the conclusions and recommendations made at the end of the study. The descriptive study that is performed here is preceded by formulated hypotheses which, according to Malhotra (2010), refers to having defined a research problem properly before starting the study.
The research question that is studied here is “What is the impact of the degree of discrepancy
between pre-purchase expectations and post-purchase performance, and the impact of social pressure, on online consumer review intention, taking into account that people differ in age, gender and their experience with giving online consumer reviews?” In order to test this research question,
23
Low expectations vs. high performance under positive social pressure
Low expectations vs. high performance under negative social pressure
high expectations vs. low performance under positive social pressure
high expectations vs. low performance under negative social pressure Low expectations vs. low
performance under positive social pressure
Low expectations vs. low performance under negative social pressure high expectations vs. high
performance under positive social pressure
high expectations vs. high performance under negative social pressure
Table 1: All possible scenarios
3.1.1. Scenario Manipulation
In order to successfully study the effects of the degree of discrepancy on online customer review intention with social pressure being present, both the degree of discrepancy and social pressure are manipulated. Each of the eight scenarios thus differs in either the degree of discrepancy or social pressure. Consequently, the differences in online customer review intention can be measured in each of the eight scenarios previously described. In order to make sure that the respondents have actually experienced the manipulation, several manipulation checks are inserted in the survey.
3.2. Data Collection
In order to collect the necessary data from the samples, this study makes use of an online survey, or questionnaire. A survey is an appropriate data collection technique as it makes sure the data is comparable. Furthermore, it is fast and records data accurately (Malhotra, 2010). The survey used in this research is a personal survey which is distributed through the Internet in order to obtain an acceptable response rate (Malhotra, 2010).
24
3.3. Operationalization
In this section, the operationalization of the variables is presented. This is a necessary process that allows for appropriate testing of the hypotheses presented in chapter two. The variables are constructs that exist of one of more items. In this research, the constructs used are the degree of discrepancy, social pressure and online consumer review intention. Additionally, the three background variables incorporated in this research will be operationalized (age, gender and experience).
Several different scales are found in the literature that are used to measure the constructs in this research. The constructs, together with the corresponding items that will be tested, can be found in table 2.
Construct Authors Items
Degree of Discrepancy Oliver, 1980; Venkatesh & Goyal, 2010;
McKinney et al., 2002
What are your expectations about Eetcafé Sjonnie?; What is your perceived performance of Eetcafé Sjonnie?
Social Pressure I judge the above reviews to be positive
Age What is your age in years?
Gender Are you male or female? (0 = female; 1 = male)
Experience Fuchs, et al., 2010 I feel competent enough to post an online consumer review;
I feel I have the relevant knowledge and expertise to post a sound online consumer review;
I have difficulties posting an online consumer review; I have experience with posting an online consumer review.
Online Consumer
Review Intention Warshaw, 1980; Laczniak et al. 2001 How likely is it that you will post an online consumer review about your experience?
Table 2: Items per variable
3.3.1. Degree of Discrepancy
25 disconfirmation or discrepancy. The scales for measurement is copied from research from Oliver (1980) where respondents had to answer on a 7-point three-item Likert scale. The items the author tested on were usage experience, performance and benefits of an information system. In this study however, the subjects under study have to deal with expectations and performance regarding a dinner at a restaurant. It is therefore decided to test on discrepancy by simply asking a question about pre-purchase expectations and later after post-purchase performance. The possible range from very low to very high.
The degree of discrepancy is manipulated here by having the respondent think of a situation in which he is about to dine in an unknown restaurant called Eetcafé Sjonnie. The respondent has to imagine that he or she does some exploring in order to know what the restaurant will be like. Subsequently, either a negative or a positive news article about the restaurant is shown. In addition, some photos are shown of the kitchen of the restaurant. This kitchen is either very filthy or very clean and tidy. To make sure that the manipulation succeeded, a manipulation check is performed based on the research from Oliver (1980). His method of measuring disconfirmation measured scores on a "greater-than-expected, worse-than-expected" scale. Consequently, respondents are asked to indicate how the perceived performance was. They can range their answers on an 11-point Likert scale ranging from ‘worse than expected’ to ‘better than expected’ with ‘as expected’ in the middle.
3.3.2. Social Pressure
Social pressure in this study is seen to be either positive or negative. In order to manipulate the social pressure that people experience in each scenario, fictive reviews about the restaurant are shown. These reviews are either extremely positive or negative with corresponding numerical ratings. They are graphically displayed in the layout from iens.nl in order to give a sense of reality. On iens.nl people can find a lot of restaurants in Holland on which online consumer reviews can be posted together with a rating on food, service and décor.
In order to check if the manipulation succeeded, people are given the following statement: “I judge the above reviews to be positive“
Respondents have to answer on a 7-point Likert scale whether they agree with the statement or not where a scoring of 1 is strongly disagree and 7 is strongly agree.
3.3.3. Online Consumer Review Intention
26 as they simply ask respondents about the likeliness of behavior. Their research about online purchase intention tested this construct by asking how likely it was that a consumer would perform a certain behavior given a certain situation. For this study, one question only is incorporated in the survey referring to the intention to post an online consumer review. Respondents are thus asked to answer the following question on a 7 point Likert scale (1= very unlikely and 7 = very likely):
“How likely is it that you will post an online consumer review about your experience at Eetcafé
Sjonnie?”
The three background variables from this study need to be operationalized as well in order to be tested appropriately. In the conceptual model, age, gender and experience are identified as moderating background variables.
3.3.4. Age
Age is operationalized as a nominal variable, being the age of the respondent at the time of filling out the survey. Age is measured in the number of years.
3.3.5. Gender
Gender however, is a nominal variable and is treated as a dummy variable having only two options, male and female, where female is 0 and male is 1.
3.3.6. Experience
27
3.4. Pre-testing
Malhotra (2010) recommends pre-testing to be a reliable method in order to test a survey’s validity and reliability. Subsequently, errors and possible other problems can be eliminated. Hence, this survey was tested beforehand using ten respondents. The feedback coming from these respondents was used for improvements in order to arrive at the final version of the survey.
3.5. Population and Sample
For this study a survey was distributed using convenience sampling (Malhotra, 2010) where respondents were in the right place at the right time in order to fill out the survey questions. The target population for the distribution of the survey consisted of every person who has a computer with access to the Internet through email and social media. In order to ensure that enough people filled out the survey, convenience sampling is used in because it is likely that a snowball effect occurs. This is because the survey was posted online and sent through email and receivers of the messages where asked to forward the survey to as much people they know. In total, 335 people started the survey for which they received a message containing a hyperlink. From the respondents that started the survey, 264 people (total sample size) filled out the survey correctly.The response rate cannot be checked as due to the snowball effect using social media, a lot of people might have received the link but decided not to start the survey. From the respondents 65,9% is male and 34,1% is female. The average age of the sample of respondents is 38 (37,88 rounded) with an age range of 15 to 74.
3.6. Randomization
As said before, eight different scenarios can be identified to which the respondents have been randomly assigned using Qualtrics. For the degree of discrepancy the respondent can fall into four identified scenarios (LE-HP; HE-LP; LE-LP and HE-HP). Regarding social pressure, respondents can either be in a scenario with positive social pressure (PSP) or negative social pressure (NSP). For each of the four degrees of discrepancy either positive or negative social pressure is exercised, creating a 4 x 2. The results can be found in table 3.
Scenario # of Respondents Scenario # of Respondents
LE-HP + PSP 33 LE-HP + NSP 34 HE-LP + PSP 31 HE-LP + NSP 28 LE-LP + PSP 36 LE-LP + NSP 32 HE-HP + PSP 31 HE-HP + NSP 39
28
3.7. Reliability
One of the variables used in the conceptual model consist of more than one item. This variable needs to be checked for its internal consistency. In other words, all items used in the survey must measure the same underlying construct. This is done by looking at the Cronbach’s Alpha (Cortina, 1993). If the result is that items measure the same construct, it is appropriate to create a new variable based on the combined items. The Cronbach’s Alpha is the average of the split-half coefficients. This is a result of splitting the scale’s items in different ways. A Cronbach’s Alpha will be somewhere between 0 and 1 and when the resulting alpha is .7 or higher it is considered to be internally consistent and thus reliable for further analyses (Nunally, 1978). In the survey, the construct Experience was measured using four different items of which one item has been used in reverse as a negative worded question, namely difficulty. In order to prevent response bias, the scores on the item have been reversed before checking the items’ combined Cronbach’s Alpha. When including all four items, the Cronbach’s Alpha of Experience is .700 which indicates internal consistency. However, the alpha becomes .726 when experience as an item is deleted. Running the test again with three items, the results indicate that the alpha can still be improved. Deleting difficulty (reversed), results in a Cronbach’s Alpha of .726 (results in table 4). More items could not be deleted.
Construct Cronbach’s Alpha Item
Experience .756 Competence Knowledge
.700* Experience
.726* Difficulty (reversed)
*Cronbach’s Alpha if item included
Table 4: Internal consistency of experience construct
From the reliability analysis, it was found that if two items, experience and difficulty, were to be deleted from the construct, a higher Cronbach’s Alpha would be obtained. As a result, the two items are deleted and the final Cronbach’s Alpha for Experience as a construct is .756 based on the two items competence and knowledge.
29 component, or factor, could be extracted from the construct. Only one Eigenvalue was found to be above 1 (2.227).
As a result, Experience is a combined construct based on competence and knowledge. A new variable is subsequently created by adding up the scores on both items and dividing them by 2. During the course of this research, it was decided that gender as a covariate will be excluded from the model as the spread between the respondents is not in a good balance. From the descriptive data analyses it was found that only 34,1% of the respondents is female. Some scenarios even show a more displaced balance concerning gender, having even less females. A T-test was performed to support this argument. The results show that the mean on online consumer review intention is equal (4.52) for both males and females. As there is no difference between the groups, the effects do not show any significance (.978).
3.8. Analysis Plan
3.8.1. Manipulation Checks
Prior to performing the main analysis on the data from the survey, manipulation checks will be done. This is because this study manipulated the end results by having created different scenarios in which respondents of the survey could fall. The purpose for doing this is based on the interest in the effect of discrepancy between pre-purchase expectations and post-purchase performance on the intention people have to post a review online. Furthermore, it is interesting to see what the effects of social pressure are on people’s online review intention.
In order to see whether or not the manipulations have succeeded, the manipulations must be checked using statistical tests. These analyses have to be performed prior to running the main analysis as the results would be unreliable if manipulations have not succeeded.
To see if the degree of discrepancy manipulation succeeded, the scenarios are split up based on pre-purchase expectations and post-purchase performance. Without accounting for social pressure, four different groups can be identified based on the degree of discrepancy (see table 5). This division, called ‘Factor_DD,’ is the variable used for the main analyses.
Scenario Factor_DD
LE-HP 1
HE-LP 2
LE-LP 3
HE-HP 4
30 Basically, for the manipulation to be a success, the means on low and high pre-purchase expectations should be significantly different from each other. The same holds for low and high post-purchase performance. Table 6 gives the descriptive results from an ANOVA test where the different means are shown for each of the Factor_DD groups.
Factor_DD N Mean Standard
Deviation
Pre-Purchase Expectations 1 (LE_HP) 67 1,48 .959
2 (HE_LP) 59 6,03 .809 3 (LE-LP) 68 1,46 .818 4 (HE-HP) 70 6,07 .937
Post-Purchase Performance 1 (LE_HP) 67 5,64 .900
2 (HE_LP) 59 1,31 .595 3 (LE-LP) 68 1,26 .857 4 (HE-HP) 70 6,53 .583
Table 6: ANOVA descriptives results; α of .05
Using T-tests to check for manipulation success, first it is tested whether people in a scenario where they should experience high pre-purchase expectations (HE-HP and HE-LP or Factor_DD 4 and 2), score significantly higher on the pre-purchase expectations scale than people in the LE-LP and LE-HP scenario (or Factor_DD 3 and 1). The low expectations scenarios are placed in one group and the high expectations scenarios are placed in another group. The means on the high expectations should be significantly higher with an alpha of .05. The results are shown in table 7. Pre-Purchase Expectations N Mean Standard Deviation Significance level (α = .05) Low (LE) 135 1,47 .888 .000 High (HE) 129 6,05 .878
Table 7: Manipulation check Degree of Discrepancy; scenario grouping based on pre-purchase expectations
The same is done for post-purchase performance where high performance scenarios should score significantly higher on the post-purchase performance scale than low performance scenarios. The results from the corresponding T-test are shown in table 8.
Post-Purchase Performance N Mean Standard Deviation Significance level (α = .05) Low (LP) 127 1,28 .744 .000 High (HP) 137 6,09 .873
31 The results from both T-tests show that the manipulation for having people experience a certain degree of discrepancy succeeded. People experience low pre-purchase expectations (Mean=1,47) in low expectations scenarios and they experience high pre-purchase expectations (Mean=6,05) in high expectations scenarios. Furthermore, the second T-test results prove that low post-purchase performance (Mean=1,28) is experienced in low performance scenarios and high post-purchase performance (Mean=6,09) in high performance scenarios. Both p-values (.000) are significant as they are below the alpha of .05.
Table 6 shows that people rank post-purchase performance even higher when they had high pre-purchase expectations (Factor_DD group 4) than when they experienced low pre-purchase expectations. (Factor_Group 1). When contrasting the different scenarios against each other, this is confirmed and shows significantly higher mean in Factor_DD group 4. These results come from a Least Significant Difference (LSD) test. (Table 9).
Factor_DD (I) Factor_DD (J) Significance
level Correct Pre-Purchase Expectations 1 (LE) 2 (HE) 3 (LE) 4 (HE) .000 .887 .000 Yes Yes Yes 2 (HE) 1 (LE) 3 (LE) 4 (HE) .000 .000 .811 Yes Yes Yes 3 (LE) 1 (LE) 2 (HE) 4 (HE) .887 .000 .000 Yes Yes Yes 4 (HE) 1 (LE) 2 (HE) 3 (LE) .000 .811 .000 Yes Yes Yes Post-Purchase Performance 1 (HP) 2 (LP) 3 (LP) 4 (HP) .000 .000 .000 Yes Yes No 2 (LP) 1 (HP) 3 (LP) 4 (HP) .000 .763 .000 Yes Yes Yes 3 (LP) 1 (HP) 2 (LP) 4 (HP) .000 .763 .000 Yes Yes Yes 4 (HP) 1 (HP) 2 (LP) 3 (LP) .000 .000 .000 No Yes yes
Table 9: Least Significant Difference test (LSD); Multiple Comparisons
32 rank their performance significantly different than people in Factor_DD group 4 even though they should both experience high post-purchase performance without a significant difference. The results from the T-tests however show that the manipulation succeeded as people correctly experienced what they supposed to experience.
For the social pressure manipulation check, another T-test is performed as respondents could only fall into one of two social pressure scenarios. In the scenarios people were either shown negative or positive reviews. Here, a new factor is created based on the social pressure scenarios called Factor_SP. This factor is subsequently contrasted against the scores on the social pressure scale, the MC_DD variable. In this case, the means for both groups should be significantly different. The results of the T-test are shown in table 10.
Social Pressure N Mean Standard
Deviation
Significance level (α = .05)
Negative 133 2,64 2,151 .001
Positive 131 5,15 1,181
Table 10: T-test results Social Pressure
33 3.8.2. Main Analysis
In the next chapter, the hypotheses that have been formulated in chapter two will be put to the test. This is a necessary process in order to answer the research question. This will be done through a one-way analysis of covariance (ANCOVA). The ANCOVA is an analysis which combines regression and one-way analysis of variance (ANOVA) features. An ANCOVA is an adapted version of the ANOVA where several covariates are included. An ANCOVA is performed in order to see what the effects are of several independent variables on the dependent variable while controlling for certain confounding variables. According to Field (2000), the ANCOVA is a parametrical statistical way of testing a linear model’s overall fit by looking at a statistic called the F-ratio. In this study the ANCOVA controls for the possible effects that Experience and Age (covariates) might have on the relationship between Degree of Discrepancy, Social Pressure (the factors) and Online Consumer Review Intention (the dependent variable).
Three steps have to be taken when performing an ANCOVA. First, an ANCOVA assumes homogeneity of slopes (Poremba & Rowell, 1997). That is, the ANCOVA assumes that the covariates do not have a significant effect on the factors. Consequently, this needs to be tested before running the actual ANCOVA. The following four relationships will be tested for homogeneity of slopes:
1. Degree of Discrepancy versus Age (Factor_DD * Age) 2. Social Pressure versus Age (Factor_SP * Age)
3. Degree of Discrepancy versus Experience (Factor_DD * Experience) 4. Social Pressure versus Experience (Factor_SP * Experience)
5. The Degree of Discrepancy versus Social Pressure versus Age (Factor_DD * Factor_SP * Age)
6. The Degree of Discrepancy versus Social Pressure versus Experience (Factor_DD * Factor_SP * Experience)
34
4. Results
In the following chapter, the main results of the analyses are presented. General statistics about the total sample and the different scenarios are presented first, followed by the test for homogeneity of slopes and the analysis of covariance. Finally, the analyses will lead to the conclusions based on the hypotheses that have been discussed in chapter two. As discussed in the previous part, it is decided to delete gender from the conceptual model. Consequently, the model to be tested looks like this:
4.1. General results
The sample for this study consists of 264 respondents that have been randomly assigned to the eight different scenarios.. The outcome of the randomization is shown in table 11.
Scenario #Respondents #Males #Females Mean Age Age range Mean OCRI score
LE-HP+NSP 34 20 14 39,21 20-71 4,09 HE-LP+NSP 28 19 9 38,39 21-60 3,96 LE-LP+NSP 32 21 11 37,53 21-61 4,78 HE-HP+NSP 39 26 13 41,41 22-60 4,92 LE-HP+PSP 33 20 13 35,61 15-74 4,03 HE-LP+PSP 31 21 10 35,19 19-62 4,81 LE-LP+PSP 36 24 12 36,58 22-61 4,47 HE-HP+PSP 31 23 9 38,48 17-61 5,00
Table 11: Sample descriptives per scenario
∆Discrepancy
OCRI
ExperienceAge
H5 H7 H2 H1 H6 H8Figure 2: Revised Conceptual Model
35 A minimum was set at 25 respondents for each scenario. This condition is met as the smallest scenario sample has 28 respondents.
The highest mean score on the OCRI scale was found in the group where respondents had high pre-purchase expectations and high post-purchase expectations and were shown positive reviews of other people. The lowest mean score on the OCRI scale was found in the group where respondents faced a scenario with high pre-purchase expectations and low post-purchase performance and were shown negative reviews of other people.
4.2. Homogeneity of Slopes
Before running the actual ANCOVA test, it is necessary to test whether the covariates interact with the factors or not. This is done by testing for homogeneity of slopes (SPSS results in Appendix B1). Being able to run the ANCOVA depends on the result of this test. The results determine whether it is possible to perform an ANCOVA where factors and covariates do not interfere with each other. When one does not account for the output from this test and simply runs the ANCOVA, errors might occur (Poremba & Rowell, 1997).
As said prior to this chapter, there are six interaction effects that need to be tested: 1. Degree of Discrepancy versus Age (Factor_DD * Age)
2. Social Pressure versus Age (Factor_SP * Age)
3. Degree of Discrepancy versus Experience (Factor_DD * Experience) 4. Social Pressure versus Experience (Factor_SP * Experience)
5. The Degree of Discrepancy versus Social Pressure versus Age (Factor_DD * Factor_SP * Age)
6. The Degree of Discrepancy versus Social Pressure versus Experience (Factor_DD * Factor_SP * Experience)
36
Variables (IV*Cov) F-value Significance level (α= .05)
Factor_DD*Age 2.200 .089 Factor_SP*Age .638 .425 Factor_SP*Experience 2.081 .150 Factor_DD*Experience .450 .717 Factor_DD*Factor_SP*Age .578 .630 Factor_DD*Factor_SP*Experience .646 .586
*Dependent variable: Online Consumer Review Intention (OCRI)
Table 12: Homogeneity of slopes results
