How does the type of product influence the motivation to read
more OCR’s after reading one review, when searching for
product information online?
2
How does the type of product influence the motivation to read
more OCR’s after reading one review, when searching for
product information online?
Master Thesis
June 18, 2018
Simon Pieter Smits
University of Groningen Faculty of Economics and Business
MSc Marketing
Specialization Marketing Management
3
Abstract:
This study investigates the motivation to read more online consumer reviews (OCR’s) after reading one review, when searching for product information online. There is a lot of academic literature regarding the motivation to read an OCR, but there is a lack of research on why consumers are motivated to continue reading OCR’s after already reading one review. This study investigates the effect of different types of products (search products and experience products) and the effect of valence of the review read by consumers (positive, neutral and negative). Different consumer characteristics where measured in order to see if they had effect on the motivation to read more reviews after reading one review. The following consumer characteristics were measured: risk attitude, information processing style (analytical and intuitive), attitude towards OCR’s, gender and age. An online experiment has been conducted, with 191 respondents. Based on the data, consumers searching for search products were significantly more motivated to read more OCR’s after reading one review, compared to consumers searching for experience products. Valence of the review does not have a main effect, but consumers searching for experience products are more motivated when reading a negative OCR compared to a neutral and a positive review. Furthermore, women are more motivated to read more OCR’s after reading one review, compared to men. Also, consumers with a positive attitude towards OCR’s are more motivated to read more OCR’s after reading one review.
4
Table of content
1. Introduction ... 6
1.1 The importance of online reviews ... 6
1.2 The rise of eWOM ... 6
1.3 Motivation of reading OCR’s ... 7
1.4 Types of products... 8
1.5 Valence of OCR’s... 9
1.6 Relevance of topic ... 10
2. Theoretical framework... 12
2.1 Type of product affect motivation to read more OCR’s ... 12
2.2 The effect of valence on motivation to read more OCR’s after reading one review .. 13
2.3 Consumer characteristics ... 15
2.3.1 The influence of risk attitude ... 15
2.3.2 The influence of the information processing style ... 17
2.3.3 Covariates ... 18
2.4 Theoretical framework ... 18
3. Research Design ... 20
3.1 Type of research... 20
3.2 Design of the surveys ... 20
3.2.1 3x2 between subject experimental design ... 20
3.2.2 Preliminary experiment ... 21
3.2.3 Main experiment ... 22
3.3 Respondents ... 24
3.4 Operationalization ... 25
3.4.1 Factor analysis and Cronbach Alpha ... 27
3.4.2 Motivation to read more OCR’s ... 27
3.4.3 Risk attitude ... 28
3.4.4 Information processing style: Analytical ... 28
3.4.5 Information processing style: Intuitive ... 28
3.4.6 Attitude towards OCR’s ... 29
3.5 Manipulation check ... 29
3.6 Plan of Analysis ... 30
4. Results ... 32
4.1 Mean conditions ... 32
5
4.3 Linear regression models ... 34
4.4 Multicollinearity ... 38
4.5 Hypothesis testing ... 38
5. Discussion and recommendations ... 42
5.1 Conclusion ... 42
5.2 Discussion ... 42
5.3 Managerial implications... 44
5.4 Limitations and further research ... 45
6. References ... 47
7. Appendix ... 54
Appendix A: Survey ... 54
Appendix B: Chi – Square Test ... 66
Appendix C: ANOVA for Age ... 67
Appendix D: Factor analysis ... 68
Appendix E: Reliability Analysis ... 72
Appendix G: ANOVA Type of Product and Valence ... 78
Appendix H: Regressions ... 80
6
1. Introduction
1.1 The importance of online reviews
8th of March, 2018: eight out of ten millennials never buys anything without reading an online review firsthand (Independent, 2018). A study of 2000 adults shows that millennials are making more purchases online than generation X or baby boomers, but that they also do more research before spending their money (Independent, 2018). Matt Moog, CEO of
PowerReviews, conducted a study in combination with Northwestern University and found that the more reviews there are of a product, the more likely it is that a customer will purchase that product (New York Post, 2017). But are consumers really reading all these reviews? BrightLocal’s Local Consumer Review Survey showed that in 2017, just 3% of the
respondents trust a business after reading just 1 review. 29% of the respondents trusted a local business after reading 2 to 3 reviews and 34% had trust in the business after reading 4 to 6 reviews (Brightlocal, 2018). Although there are hundreds of thousands of online reviews available, consumers apparently just read a few of them. This thesis researches why consumers are motivated to read more reviews after they have read one review, when searching for product information online. The effect of different type of products and the positivity of the review will be examined.
1.2 The rise of eWOM
The rise of the internet has created multiple options for consumers to gather product information and has provided opportunities for consumers to engage in electronic word-of-mouth (Hennig-Thurau, 2004). eWOM is a source of brand and product information for consumers in today’s digital marketing environment (Chen, 2008). Online consumer reviews (a form of eWOM) can be defined as any positive, neutral, or negative evaluation of a
product, a service, a person, or a brand presumably posted by former customers on websites that host consumer reviews (Filieri, 2018). Many consumers search on the Internet before they make purchases (Smith, 2005), where they do not only search for product information
7
uncertainties in the buying process. The consumer buying process is commonly described as a five-stage linear process (Blackwell, 2003; Hawkins, 2003): need recognition, information search, alternatives evaluation, purchase decision, and post-purchase behavior. During the information search and evaluation of alternatives, consumers can reduce their perceived risk of a potential purchase (Lawrence, 2005). Theory of risk taking suggests that consumers decide to buy a product under a degree of uncertainty about the given brand (Sheth, 1968). The conceptualization of perceived risk starts with Bauer (1960), who recognizes that consumer behavior involves risk-taking. The perceived risk dimensions identified in the literature include financial, performance, social, psychological, physical and
time/convenience risks (Girard, 2010). It is theorized that when perceived risk falls below an individual’s acceptance value, it has little effect on intended behavior and is essentially ignored (Greatorex, 1993). On the other hand, an extremely high level of perceived risk can cause a consumer to postpone or avoid a purchase entirely (Lawrence, 2005). In presence of uncomfortable levels of perceived risk, consumers apply risk reduction strategies during the second and third stage of their buying process, by relying on personal recommendations and additional information seeking (Lawrence, 2005), thus consumers are often motivated to read multiple OCR’s during their buying process to reduce their perceived risk on a purchase.
1.3 Motivation of reading OCR’s
8
of the perceived quality and uniqueness of information about products, to help them become more informed in order to make better buying decisions. Consumers gain empowerment by attaining information through reading realistic ‘non-expert product opinions’ written by ordinary people like themselves (Burton, 2010). Consumers whose use of digital media is disproportionately reflected in information retrieval are more strongly motivated than others to reap the benefits of enhanced information access (Labrecque, 2013). Reading non-expert opinions in the ‘consumer empowerment’ theme echoes with users’ opinions being more trusted sources than expert reviews. (Bickart, 2001). The ‘consumer empowerment’ theme highly overlaps with trust also leading to ‘risk reduction’ (Bickart, 2001). Consumers are motivated to read multiple reviews in order to reduce risk of being misled by individual sources (Burton, 2010). They also attempted to mitigate for misleading sources by
interpreting reviews with care, trying to select reviews by reviewers with similar views, and using their own experience and judgment to appraise review accuracy (Burton, 2010). Hence, consumers are motivated to read OCR’s to empower themselves by attaining information and by reducing perceived risk. Interestingly, there is a little research regarding the motivation to read more OCR’s after already reading one review. This research makes a distinction between motivation to read a single OCR and motivation to read more OCR’s after reading one
review. According to Kang (2018) it is important to get more insights in the online information attainment of consumers and not only the motivation to read an OCR. This research examines the motivation to read more OCR’s after reading just one OCR, when consumers are searching for information online.
1.4 Types of products
Consumers are motivated to reduce risk relating to products that they are interested in buying (Burton, 2010). According to him, the motivation to reduce risk could be different for
9
and erases differences between search and experience products. According to Huang (2009), the classification of search and experience products is still widely used among current research. Because the classification of Nelson (1974) might have been outdated, Weathers (2006) re-examined again how well consumers could estimate quality of products before and after purchase and concluded that this classification is still useful. An alternative perspective on different type of products is the information search per type of product, how consumers acquire and process information to make decisions. Literature shows that different types of information are associated with different cognitive processes that affect the way information is acquired (Johnson, 2003), the amount of information acquired (Johnson 1984), and the time spent processing each piece of information (Payne, 1988). If consumers seek different information for search and experience goods, this perspective implies that online search and purchase behavior may be different for these two types of goods. Huang et al. (2009) revealed that online consumer behavior for experience and search goods is distinct. Park (2009) backed this up with his findings that the effect of eWOM differs for search and experience goods on websites, whereas the effect of eWOM is greater for experience products. So, the information search process is different for search and experience products. This research will investigate if the different types of products, search versus experience, have an effect on the motivation to read more OCR’s after reading one review.
1.5 Valence of OCR’s
10
perceived risk, both positive and negative, therefore it is suggested that the valence of OCR’s has an effect on consumers motivation to read more OCR’s after reading one review.
1.6 Relevance of topic
Due to increasing competition between retailers, both online and offline, understanding of information attainment and specifically the motivation to read more OCR’s after reading one, could be very valuable. Kang et al (2018) show that information attainment, social interaction, and assortment seeking were found to be antecedents of webrooming: researching products online and buying them in offline stores. That view is in line with the concept of ‘research shopping behavior’ by Verhoef et al. (2007). They defined research shopping behavior as the tendency of consumers to research the product in one channel and then purchase it through another channel, where reading OCR’s is an important part of product research. At this moment, there is little empirical evidence of the information attainment in webrooming behavior (Kang, 2018) and therefore the motivation to read more OCR’s after reading one could be interesting. According to Verhoef (2007), webrooming is the most commonly pursued form of research shopping behavior and webrooming is also the most extensive cross-channel behavior that consumers show according to Flavian (2016). Pure numbers underline these views, according to Forrester Research (2014), webrooming sales are currently five times larger than online sales, and by 2018, 44% of all in-store purchases will be influenced by the Web. Marketing literature reveals a number of reasons why consumers webroom. According to Chou et al. (2016) state that webrooming behavior helps consumers to mitigate the risks involved in directly buying online, and as we have seen, consumer are motivated to read OCR’s to reduce perceive risk.
11 1.7 Outline of thesis
This research will consist of 5 chapters. The study examines the following problem statement: How does the type of product and the valence of the review read by the consumer, influence the motivation to read more OCR’s after reading one review when searching for product information online? The objective of this research will be to determine in what situations consumers are more motivated to read more reviews after reading one review. This will be done by answering the following research questions:
1. How does the type of product influence the motivation to read more OCR’s after reading one review, when searching for product information online?
2. How does the valence of the online review read, influences the motivation to read more OCR’s after reading one review, when searching for product information online?
3. Is there an interaction effect between the type of product and the valence of the review read, on the motivation to read more OCR’s after reading one review when searching for product information online?
4. Which personal characteristics might play a role in the motivation to read more OCR’s after reading one review, when searching for product information online?
These questions will be answered in the following four chapters. In the next chapter, the academic literature will be analyzed and the conceptual model will be shown. The third chapter will explain the methodology of this research. In the fourth chapter, the results will be shown. The fifth chapter will contain the discussion of the results, the implications for
12
2. Theoretical framework
This section will explain the different concepts, according to current academic literature. Relations between the concepts will be examined and the hypothesis will be made. In chapter 2.1 the type of product will be explained and the effect on the motivation to read more OCR’s after reading one will be explained. In chapter 2.2 The effect of the valence of the OCR’s will be explained. In chapter 2.3 the consumer characteristics risk attitude and information
processing style will be analyzed. Also other covariates in this study will be elaborated. This chapter will finish with chapter 2.4 where the conceptual model of this research will be showed.
2.1 Type of product affect motivation to read more OCR’s
The main effect of the type of product
13
When looking at browsing behavior, Huang (2009) also reports that experience products involved a greater depth and lower breadth of search compared to search products. In other words, the amount of time spent per page was greater for experience products, but the number of pages searched was greater for search goods consumers. They found that communication mechanisms, such as consumer feedback and experience simulation (e.g., consumer reviews, multimedia), increase the time spent in a domain but only for buyers of experience goods. So consumers that are browsing for search products spend less time on a website and see more different pages, while consumers browsing for experience products spend more time on one page (Huang, 2009). This could imply that the information, including the OCR’s, is more important when browsing for experience products, compared to browsing for search products. So, these findings imply that consumers browsing for experience products are spending more time reading information and reviews, thus are more motivated to read OCR’s after reading one review, compared to consumers searching online for search products. The differences in browsing behavior could also be because of the different levels of perceived risk per type of product. Girald et al. (2010) state that there are significant differences in consumer risk perceptions for search and experience products. According to them, the levels of the six types of risk were perceived significantly lower for search products than for experience products. Consumers are motivated to read OCR’s to reduce risk, thus consumers buying experience products are more motivated to read OCR’s after reading one review because they want to reduce the higher perceived risk, compared to browsing for a search product.
Based on all the above reasoning, the following hypothesis is made:
H1: Consumers searching online for experience products are more motivated to read more OCR’s after reading one review, compared to consumers searching online for search products.
2.2 The effect of valence on motivation to read more OCR’s after reading one review
The main effect of valence
This section explains why the valence of the review has an effect on the motivation to read more OCR’s after reading one. According to Chiou, the appearance of negative eWOM in online communication can exert a greater impact on corporate image than it would in traditional face-to-face communication (Chiou, 2003). Although positive reviews are
14
negative information has a greater weight compared to equally strong positive information in creating judgements (Ahluwalia, 2002; Wu, 2013). Comparing to positive information, negative information tends to influence evaluation more strongly (Ito, 1998). Positive
reviews can increases consumers’ willingness to trust the product and affects the consumers’ perception of buying the right product (Zhang, 2007). While positive reviews tend to
increase prospective customers’ anticipation of benefits, negative reviews induce expectation of risks (Lee et al. 2008). Pee (2016) collaborates this view with his findings that negative reviews increase the perceived risk and potential loss associated with a purchase. So,
negative reviews in general have more influence on the evaluation and increase the perceived risk. Following this logic, a negative review could have a positive effect on the motivation of consumers to read more OCR’s after reading one review. Thus, this leads to the following hypothesis:
H2: The more negative (positive) the valence of the OCR read, the higher (lower) the motivation to read more OCR’s after reading one review.
H3: The difference between a negative and a neutral OCR has a more positive effect on the motivation to read more OCR’s compared to the difference between a neutral and positive OCR.
The interaction effect of valence and the type of product.
An interaction effect between valence of an OCR and the type of product has been examined by multiple researchers and can be divided in two main views: the usefulness of the OCR and the effect of the valence of the OCR. According to Willemsen et al. (2011), the perceived risk and uncertainty increase along the search–experience product continuum. Therefore, the effect of review valence on review usefulness is limited for search products, as consumers can assess product characteristics objectively. On the contrary, the assessment of an
15
interaction between product type and valence of the OCR and leads to the following hypothesis:
H4: A positive (negative) OCR negatively (positively) influences the effect of the type of product towards the motivation to read more OCR’s after reading one review.
2.3 Consumer characteristics
The motivation to read more OCR’s after reading one review may be influenced by personal characteristics. This study will examine the individual risk attitude and the information processing style. Risk attitude has been conceived mostly as related to decision-making behavior (Pennings, 2000; Weber, 1997). Risk attitude can provide the explanation for the individual differences in the way people resolve decisions involving risk (Wu, 2007). Next to the risk attitude, the information processing style could influence the motivation to read more OCR’s after reading one. Soane et al. (2015) state that information processing styles, typically characterized as tendencies to use analytical or intuitive approaches to choice influence decision processes and outcomes. Analytical processes are required for novel, complex problems whereas intuitive or heuristic processes are applied to numerous daily choices. In this paragraph both characteristics will be analyzed based on academic literature
2.3.1 The influence of risk attitude
16
Risk attitude affects motivation to read more OCR’s after reading one review.
Consumers with higher risk attitude tend to be more confident about expected outcomes (Wu, 2007). He also showed that risk attitude positively is associated with both evaluation-based and emotion- evaluation-based satisfaction (Wu, 2007). This implies that consumers with higher risk preference may not only be overconfident, but also more likely to be pleased. These findings could imply that a higher risk attitude will have a negative effect on the motivation to read more OCR’s after reading one review. These results are also consistent with Cosbey’s (2001) argument that the relationship between self-perception and satisfaction implied that risk attitude plays a critical role in online shopping. The easier consumers make a transaction, the less research they will do, and will have less motivation to continue reading OCR’s. Tan (1999) proposed that consumers with higher risk aversion often perceive online shopping to be a risky activity. This could imply that consumers with low risk attitude are more
motivated to read more OCR’s to reduce the perceived risk. Based on the above reasoning, the following hypothesis is made:
H5: Higher (lower) risk attitude will have a negative (positive) effect to the motivation of consumers to read more OCR’s after reading one review.
Risk attitude and different type of products:
Different risk attitudes can illustrate why some people prefer transactions that are riskier than others (Wu, 2007). Consumers’ risk aversion varies depending on the product type and consumption contexts (Hennigs, 2015). Experience products are perceived with higher risk compared to search products. According to several researchers, risk-neutral consumers are more likely than risk-averse consumers to make a purchase when faced with buying a product with uncertain outcomes or possible loss (Kahneman, 1982; Taylor, 1974). This implies that risk attitude will have a moderating effect on the type of product towards the motivation to read more OCR’s after reading one review. Based on the above, the following hypotheses is made:
17
2.3.2 The influence of the information processing style
The way that consumer process information showed in an OCR, could influence the
motivation to read more OCR’s after reading one review. Research in information processing styles came up with several theories: the risk and information seeking and processing theory (Griffin, 1999), dual process theory (Epstein, 1990; Epstein, 1996), and broaden-and-build theory (Fredrickson, 2001). Among other, Dane (2007) made a distinguish between an analytical and intuitive information processing style. Theories of analytical and intuitive thinking rest on the dual-process concept which proposes two parallel, interactive systems of thinking (Epstein, 1990). System 1 is intuitive, affect-laden and rapid, system 2 is cognitive, resource intense and requires time (Soane, 2015). Analytical thinking is associated with effective decision making due to logical reasoning and fewer decision biases (Stanovich, 2002), and ability to focus on important aspects of information relevant to decisions rather than non-relevant contextual information (McElroy, 2003). Intuitive thinking is associated with expertise (Dreyfus, 2005) and effectiveness in solving everyday problems (Todd, 2007). While the dual-process model has universal application, the extent to which System 1 and System 2 are applied, and the situational contingencies that influence their use, are subject to individual differences (Epstein, 1996).
Information processing styles affect motivation to read more OCR’s after reading one review
The information processing style of consumers can change per information seeking situation (Epstein, 1996). Consumers can have an analytical information processing style in situation A but can have an intuitive information processing style in situation A. Soane (2015) proposed that an analytical information processing styles would be associated positively with
information seeking. She showed that there was a direct effect of analytical information processing style on information seeking. Hence, when consumers have a preferences for an analytical information processing style, information seeking is likely to form part of their strategy for finding and evaluating information systematically prior to making a choice (Soane, 2015). They also showed that the intuitive processing style is negatively associated with information seeking. Based on the above reasoning the following hypothesis is made:
H7: Consumers showing a more analytical (intuitive) information processing style are more (less) motivated to read more OCR’s after reading one review.
18
Information processing style interacts with product
Analytical processes are required for novel, complex problems whereas intuitive or heuristic processes are applied to numerous daily choices (Soane, 2015). It could be argued that the information processing style depends on the type of product. Where the quality of search products is easily assessed, assessing product quality of experience goods is more difficult. Hence, the information processing style could be analytical for experience goods and could be intuitive for search products. Following this reasoning and hypothesis 7, the following
hypothesis is developed:
H8: The motivation to read more OCR’s after reading one review will be higher (lower) when consumers showing a more analytical (intuitive) information processing style when searching online for experience products compared to searching online for search products.
2.3.3 Covariates
The motivation to read more OCR’s after reading one review will be controlled for two demographic variables: age and gender. This research will also control for the covariate attitude towards OCR’s. Attitude towards OCR’s might influence the motivation to read more OCR’s. Attitude towards OCR’s is based on Hennig-Thurau et al. (2003), who did research towards the motivation to read OCR’s.
2.4 Theoretical framework
The variables and hypotheses derived from the theoretical framework are presented in the conceptual framework of this study. This framework provides an overview of the
relationships between the different variables that are tested by conducting an experiment. The variables review valence and type of product are the manipulated variables in this study. Motivation to read more OCR’s after reading one review, risk attitude, the information
19 Figure 1: Theoretical framework
Type of product
- Search - Experience
Valence of online review
- Positive - Neutral - Negative Motivation to read more OCR’s Risk attitude
Information processing style
20
3. Research Design
This research consist of two experiments. First a preliminary experiment is conducted to compare consumers’ perceived ability to assess product quality of different products before the purchase and after a purchase. The second experiment conducted was a 2x3 factorial design to measure motivation to read more OCR’s. In this chapter the type of research will first be explained. The preliminary experiment will be explained and afterwards the design of the main survey will be elaborated. The operationalization and respondents will also be explained in this chapter.
3.1 Type of research
This chapter will explain the design of this research. This research aimed to provide
conclusive answers on the motivation to continue reading OCR’s. Conclusive studies, on the contrary, aim to provide final and conclusive answers to research questions (Saunders, 2012). This research is conclusive research.
Conclusive research can be divided into two categories: descriptive and causal. This research aimed to find a cause-and-effect relationship. This research searches for effects of the type of product and the valence of the review on the motivation to continue reading OCR’s. Both effects are cause-and-effect relationships and therefore is this research is causal research. Causal research design is conducted to study cause-and-effect relationships (Saunders, 2012). Concluding, this research is a causal research design
3.2 Design of the surveys
3.2.1 3x2 between subject experimental design
In this study, a 3 x 2 between subject experimental design is used to test the hypotheses. All participants were allocated to 1 of the 6 conditions. Table 1 shows the 6 different conditions and the different levels of the concepts.
Type of product Search
product
Experience product
Valence of the review
Positive Condition 1 Condition 4
Neutral Condition 2 Condition 5
Negative Condition 3 Condition 6
21
Respondents of condition 1, 2 and 3 saw a search product and afterwards correspondingly a positive, neutral or negative review about this product. Respondents of condition 4, 5 and 6 saw an experience product and afterwards correspondingly a positive, neutral or negative review about this experience product.
Nelson (1974) created a category classification between search and experience goods. This original classification is widely used and Nelson Validated this classification by using a variety of secondary sets (Huang, 2009). While this classification is widely used, multiple researchers studied this classification in the past years (Nakayama, 2010). While the classification of search and experience products may be widely used among research, the classification dates from decades ago. That fact, combined with the assumption that this research will be conducted mostly on students, this research bases its type of product classification on the research of Weathers et al. (2006). Weathers asked 108 students to classify 35 goods into either predominantly experience goods, predominantly search goods or that the respondents do not know. According to the research of Weathers et al. (2006) luggage products, such as a backpack, were rated with 63,6% as search products. Clothes were rated with 95.3% as experience products according to his research.
3.2.2 Preliminary experiment
22 Table 2: Results of preliminary experiment
The results of the preliminary experiment show that when asked to rate a backpack before purchase, participants rated with a mean of 5,08. When asked to rate a backpack after the purchase, participants rated with a mean of 5,94. This shows that a backpack is a good
example of a search product. When asked to rate a pair of jeans before purchase, participants rated with a mean of 3,82. When asked to rate a pair of jeans after purchase, participants rated with a mean of 6,27. A pair of jeans is therefore a good example of an experience product. 3.2.3 Main experiment
Based on the research of Weathers (2006) and the preliminary experiment there is chosen for a backpack as search product and a pair of jeans as an experience product. Both products can normally be bought in a similar price range, from 50 to 150 euro. The respondents were framed with a story where they went traveling for a few months. There was chosen for traveling for a few months so that respondents thought that the purchase was an important purchase because they needed to use the product a lot in the coming months. The scenario said that the respondents were behind a computer and were browsing in different websites. Familiar websites for clothing and luggage were chosen so that the respondents were framed that they were really searching online. To frame the respondents in a way that the respondents were thinking that they were browsing and searching information about the product, a figure
Before purchase After purchase
Product Mean SD Variance Mean SD Variance
23
was showed of a real online web shop. Figure 2 shows the scenario that respondents of condition 1, 2 and 3 received.
Figure 2: Backpack scenario of the survey
24
Figure 3: Valence manipulation in the survey
3.3 Respondents
The survey was posted on Facebook and shared via Whatsapp to friends, colleagues and fellow students. A minimum of 30 participants per condition was chosen, which means that the questionnaire should be answered by at least 180 participants (Malhotra, 2009). In total 234 respondents accessed the survey and 28 respondents did not fill in the survey. The experiment had a response rate of 88%. During the survey there was one statement which said: “For this statement answer: Strongly disagree.” Right after publishing the survey, respondents asked questions about this statement because they didn’t understand what to do with this statement. Several respondents did not fill in “Strongly disagree”. The statement was then immediately changed to: “To make sure you still pay attention please answer for this statement: Strongly disagree”. 206 respondents answered the survey but 15 did not answer “strongly disagree” and the answers of those respondents were excluded. From the 191 respondents who did fill in the survey and correctly filled in the question to make sure the paid attention, 14 did answer the questions about the dependent variable but stopped
25
between the 6 conditions. The SPSS output can be found in appendix B. Secondly, an ANOVA showed that age is not significantly different between the 6 conditions with p = 0,075, but age is almost significantly different between the 6 conditions. The SPSS output of the ANOVA can be found in Appendix C.
Gender Mean age Number of
respondents
Condition Male Female
1 41% 59% 24,7 27 2 50% 50% 28,9 34 3 56% 44% 23,2 30 4 29% 71% 23,5 32 5 41% 59% 28,5 34 6 31% 69% 26,6 34 Total 41% 59% 26,0 191
Table 3: Gender and age per condition
3.4 Operationalization
26
Table 4: Operationalization
Variable Items Source Measurement Eigenvalue & Cronbach’s Alpha
Motivation to read more OCR’s
- I would like to read more reviews about this backpack/pair of jeans - I would like to spend more time to find extra reviews about this backpack/pair of jeans
- I would like to read more reviews about this backpack/pair of jeans to make sure I make the right decision - I would like to read more reviews about this backpack/pair of jeans to get more negative and positive information about the product.
Heninig-Thurau et al. (2003), Goldsmit (2006)
7-point Likert scale: from 1 “strongly disagree” to 7”strongly agree.” EV: 3,077 α = ,899 Risk attitude
- I would like to try any new product - I would like to spend time to investigate a new brand
- I would like to try the most unusual item
Wu et al. (2007)
7-point Likert scale: from 1 “strongly disagree” to 7”strongly agree.” EV: 1,377 α = ,409 (,374 if item 2 is deleted and ,356 if item 3 is deleted) Information processing style: Analytical
- I don’t like to have to do a lot of thinking (R)
- I try to avoid situations that require thinking in depth about something (R) - I prefer to do something that
challenges my thinking abilities rather than something that requires little thought
- I prefer complex to simple problems - Thinking hard and for a long time about something gives me little satisfaction
Pacini et al. (1999)
7-point Likert scale: from 1 “strongly disagree” to 7”strongly agree.” EV: 2.180 α = ,654 (,367 if item 5 is included) Information processing style: Intuitive
- I trust my initial feelings about people - I believe in trusting my hunches - My initial impressions of people are almost right
- When it comes to trusting people, I can usually rely on my “gut feelings” - I can usually feel when a person is right or wrong even if I can’t explain how
Pacini et al. (1999)
7-point Likert scale: from 1 “strongly disagree” to 7”strongly agree.” EV: 3.154 α = ,851 Attitude towards OCR’s
- I like to read reviews when I’m searching online for a product - Online reviews help me to make a good product choice
- I search multiple reviews to make sure I make a good purchase
Heninig-Thurau et al. (2003),
7-point Likert scale: from 1 “strongly disagree” to 7”strongly agree.” EV: 2,497 α = ,897 Manipulatio n check
- How positive would your rate the reviews you just read on a scale from 1 to 7
- 7-point Likert scale: from 1 “very negative” to 7”very positive .”
27 3.4.1 Factor analysis and Cronbach Alpha
A factor analysis was performed to reduce the large set of variables into a smaller set. A factor analysis was performed for every variable. According to Malhotra (2012) a factor analyses can be done if the Bartlett’s test of sphericity can reject the null hypothesis that the variables are uncorrelated in the population. The Kaiser-Meyer-Olkin statistic should have a value greater than 0.5. The number of factors can be based on the eigenvalues, were only factors higher than 1.0 are used Malhotra (2012). All variables comply with both these statistics. The SPSS output of the factor analysis can be found in Appendix D.
To measure the internet consistency reliability, the split-half reliability test was done. According to Malhotra (2012) the value of the Cronbach’s Alpha should be higher than 0.6. All the variables had a Cronbach’s Alpha above 0,6 except for the variable risk attitude. In the next section will be explained why and how the analysis has been continued. The SPSS output of the reliability analysis can be found in Appendix E
A final factor analysis was conducted for the variables: risk attitude, information processing style and the covariate attitude towards OCR’s. The factor analysis was rotated with the Verimax rotation. Table 5 shows the rotated component matrix of the different items.
Factor Item 1 2 3 4 Risk Attitude Q1 0,079 0,075 0,024 0,870 IPS Analytical Q1 0,039 0,083 0,816 -0,192 IPS Analytical Q2 0,028 0,124 0,823 0,012 IPS Analytical Q3 0,077 0,072 0,509 0,280 IPS Analytical Q4 0,080 -0,149 0,548 0,453 IPS Intuitive Q1 0,751 0,053 0,130 0,012 IPS Intuitive Q2 0,787 0,030 0,022 -0,021 IPS Intuitive Q3 0,814 0,048 0,014 0,064 IPS Intuitive Q4 0,850 -0,031 -0,007 -0,022 IPS Intuitive Q5 0,761 0,096 0,075 0,204 Tendency OCR Q1 0,033 0,920 -0,026 0,059 Tendency OCR Q1 0,065 0,913 0,107 -0,051 Tendency OCR Q1 0,062 0,891 0,129 0,050
Table 5: Full factor analysis
3.4.2 Motivation to read more OCR’s
28
Responses on motivation were made on a seven-point Likert scale, ranging from “strongly disagree” to “strongly agree”. The 4 statements are based on previous research from Heninig-Thurau et al. (2003) and Goldsmit (2006). An example of one statement is: “I would like to read more reviews in order to make sure that I make the right decision”.
3.4.3 Risk attitude
Risk attitude is operationalized with three statements reflecting the risk attitude of
respondents during online shopping. Responses on risk attitude were made on a seven-point Likert scale, ranging from “strongly disagree” to “strongly agree” . The three statements are based on the research Wu (2007) did in online shopping. An example of one statement is: “I am willing to try the most unusual items“. Risk attitude has an Cronbach Alpha of ,409 Although the variable risk attitude is based on Wu (2007), the Alpha is lower than the 0.6 and according to Malholtra (2012), there is not sufficient internal consistency reliability in this variable. The reason this variable does not have this reliability could be because the translation from English to Dutch is not 100% correct and respondents therefore answered differently. Another reason could be that within the 3 items of risk attitude, a statement was asked, if respondents were still paying attention. Because the variable does not meet the qualification of having a value above 0.6, only the statement: “I would like to try any new product” is used for further analysis. This statement is chosen because it correspondents mostly with the definition of the concept risk attitude.
3.4.4 Information processing style: Analytical
The analytical information processing style is operationalized on 5 statements. Responses on the information processing style were made on a seven-point Likert scale, ranging from “strongly disagree” to “strongly agree”. The 4 statements are based on previous research of Pacini et al. (1999). An example of one statement is: “I try to avoid situations that require thinking in depth about something”.
3.4.5 Information processing style: Intuitive
29 3.4.6 Attitude towards OCR’s
The attitude towards OCR’s is operationalized on 3 statements. Responses on the information processing style were made on a seven-point Likert scale, ranging from “strongly disagree” to “strongly agree”. The 3 statements are based on previous research of Heninig-Thurau (2003). An example of one statement is: “Online reviews help me to make a good product choice”. 3.5 Manipulation check
The manipulation check is done by asking the respondents to rate the valence of the review on a seven-point Likert scale, ranging from “Very negative” to “very positive”. With an oneway-ANOVA it was tested if the manipulation of the valence of the review resulted in different scenario’s, according to the respondents. Table 6 shows the descriptives of the oneway-ANOVA manipulation check. On a 7-point Likert scale: from 1 “very negative” to 7”very positive”, respondents who received a positive review scored the review with a mean of 6.03. Respondents who received the neutral review scored the review with a mean of 3.66,
respondents who received the negative review scored the review with a mean of 1.89. Table 6 also shows the 95% confidence interval for the means, for all three levels of valence, the lower and upper bound corresponds with the level of valence.
Table 6: Manipulation check
The results of the manipulation check with an one-way-ANOVA showed that there is
significant difference between the three levels of valence. The difference between groups got an F=169,506 and p =0,000. A multiple comparison between the three levels had been conducted. A post-hoc test with Tuckey showed that the all the three levels of valence also differed significantly from each other. The SPSS output of the model can be found in Appendix F.
95% Confidence interval for Mean
30 3.6 Plan of Analysis
This section describes the plan for further analyzing the data. An analysis of variance will be conducted on the set of data in order to examine the different levels of the experimental variables, and to see if there are significant main and interaction effects. The means of the different conditions will be showed. Next, multiple regression will be performed with different models. Starting with model 1, a multiple regression of only the main effects of the two experimental variables will be tested. In model 2 the interaction effect between the two experimental variables will be added. Model 3 will test the main and interaction effects of the two experimental variable combined with the moderator risk attitude. Model 4 will test the main and interaction effects of the two experimental variable combined with the moderator information processing style. Model 5 will test the main and interaction effects of the two experimental variable combined with the moderators risk attitude and information processing style. Model 6 will test the full model, including the covariate attitude towards OCR’s.
The experimental variable valence consists of three levels. With coding, two dummies are created. The neutral condition will occur when both dummies are 0. A positive condition will occur when dummy 1 has got a value of 1 and a negative review will occur when dummy 2 has got a value of 1
The models will be tested with the following bivariate linear regressions:
Model 1: MOT = β0 + β1 * TOP + β2 * Dummy1 + β3 * Dummy2 + ε
Model 2: MOT = β0 + β1 * TOP + β2 * Dummy1 + β3 * Dummy2 + β4 * TOP * Dummy1 + β5 * TOP * Dummy2 + ε
Model 3: MOT = β0 + β1 * TOP + β2 * Dummy1 + β3 * Dummy2 + β4 * TOP * Dummy1 + β5 * TOP * Dummy2 + β6 * RIS + B7 * RIS * TOP + ε
Model 4: MOT = β0 + β1 * TOP + β2 * Dummy1 + β3 * Dummy2 + β4 * TOP * Dummy1 + β5 * TOP * Dummy2 + β8 * ANA + β9 * ANA * TOP + β10 * INT + β11 * INT * TOP + ε Model 5: MOT = β0 + β1 * TOP + β2 * Dummy1 + β3 * Dummy2 + β4 * TOP * Dummy1 + β5 * TOP * Dummy2 + β6 * RIS + B7 * RIS * TOP β8 * ANA + β9 * ANA * TOP + β10 * INT + β11 * INT * TOP + ε
31
Where:
MOT = Motivation to read more OCR’s TOP = Type of Product
Dummy 1 = 0 when neutral valence condition, 1 when positive valence condition Dummy 2 = 0 when neutral valence condition, 1 when negative valence condition RIS = Risk attitude
ANA = Analytical information processing style INT = Intuitive information processing style ATO = Attitude towards OCR’s
32
4. Results
In the following chapter the results of the analysis are presented. First the means of different conditions will be showed, then the main effects and the interaction effects of the analysis of variance will be presented. Next, the linear regression models will be showed. Due to
multicollinearity, an extra analysis of variance was conducted. Finally, the hypothesis will be tested according to the linear regressions and the analysis of variance.
4.1 Mean conditions
Table 7, on the next page, shows the means and standard deviation of the 6 different
33
Motivation to read more OCR’s after reading one review.
Type of product
Search product Experience product
Mean (SD) Mean Total
Valence of the review Positive 5,6019 (1,19) 4,2578 (1,60) 4,8729 (1,57) Neutral 5,5294 (1,50) 4,4412 (1,40) 4,9853 (1,54) Negative 5,0917 (1,73) 4,8309 (1,59) 4,9531 (1,65) Total 5,4066 (1,50) 4,5150 (1,53)
Table 7: Mean scores per condition
4.2 Analysis of variance
A univariate ANOVA was conducted to see the main effects of the different levels of valence and the different type of products on the motivation to read more OCR’s. A
two-way-ANOVA was conducted. The interaction between the two variables was also tested. Table 8 shows the results of the two-way-ANOVA.
Source Type III
Sum of Squares df Mean Square F Sig. Corrected Model 48,086a 5 9,617 4,173 ,001 Intercept 4662,643 1 4662,643 2023,212 ,000 Type of Product 38,200 1 38,200 16,576 ,000 Valence ,097 2 ,048 ,021 ,979 Type of Product * Valence 9,997 2 4,998 2,169 ,117 Error 426,346 185 2,305 Total 5135,125 191 Corrected Total 474,433 190
Table 8: Results of the two-way ANOVA
34
the three different levels of valence, there was nog significant difference between the different levels. The ANOVA also tested for an interaction effect between the different levels of
valence and the type of product. There is no significant interaction effect between those two variables but with a p = ,117, it is almost significant for 90% of the population. This can also be seen in Figure 3, which shows the marginal means. Interestingly, the figure shows that the means for the search product are declining and the means for experience are increasing. The figure also an indication of a small interaction effect with the negative valence condition. For both the search and experience conditions, they differ from the neutral and positive condition. The figure and the significance level of p = ,117, show that there is some indication of
interaction. The next section will analyze these effects and the possible interaction effect with regression analysis. The SPSS output of the ANOVA can be found in Appendix G.
Figure 3: Plot of estimated means of the motivation to read more OCR’s after reading one
4.3 Linear regression models
35
Table 9: Summary of regression models. * p < ,10, ** p < ,05, *** p < ,01
Effect Model Main effect 1 2 3 4 5 6 Constant 5,430*** 5,474*** 5,574*** 5,606*** 5,596*** 5,080*** Type of product (0 = search product, 1 = experience product) - ,890*** -,977*** -1,078*** -1,102*** -1,108*** -,823** Positive Dummy (0 = neutral, 1 = positive) - ,075 ,128 ,012 ,028 ,022 -,04 Negative Dummy (0 = neutral, 1 = negative) - ,004 -,319 -,403 -,388 -,370 -,178 Interaction effect Type of product * Positive Dummy -,367 -,248 -,181 -,154 -,053 Type of product * Negative Dummy ,635 ,838** ,811 (p = ,119) ,802 (p = ,124) ,548 Moderators Risk Attitude ,109 ,090 ,077
36
Model 1 tested the main effects of type of product and the valence of the reviews on the motivation to read more OCR’s after reading one review. The overall model is significant with p = 0.001 and explains 8% of the variance of the motivation to read more OCR’s with R2
= 0,080. The type of product has got significant influence on the motivation to read with p = 0.000 and β = -,890. Consumers are more motivated to read more OCR’s after reading one
review while searching for search products compared to searching for experience products. The valence of the review does not have a significant influence on the motivation to read more OCR’s after reading one review with p = 0,735 for the positive condition and p = 0,881 for the negative condition.
Model 2 tested the main effects of the type of product and the valence of the reviews and the interaction effect of the type of product and the valence of the reviews on the motivation to read more OCR’s after reading one review. The overall model is significant with p = 0,002 and explains 9,7% of the variance of the motivation to read more OCR’s after reading one review with R2 = 0,97 The main effect of type of product is significant with p = 0,006 and β =
-,977 . The interaction effect between a positive review and the type of product is not significant with p = ,489. The interaction effect between a negative review and the type of product in not significant with p = 0,208.
Model 3 tested the main effects and interaction effects of the two experimental variables and effect of the moderator risk attitude on the motivation to read more OCR’s. The overall model is significant with p = 0.002 and explains 11,6% of the variance of the motivation to read more OCR’s with R2 = 0,116. The main effect of type of product is significant with p = 0,002
and β = -1,078 . The moderating effect of risk attitude on the effect of type of product on the motivation to read more OCR’s is not significant, with p = 0,274. There is no significant direct effect of risk attitude on the motivation to read more OCR’s with p = 0,266. The interaction effect between the negative review and the type of product is almost significant with p = 0,095 and β = ,838 for 90% of the population. Consumers searching for experience products are more motivated to read more OCR’s after reading a negative review compared to consumers searching for search products after reading a negative review.
37
of the variance of the motivation to read more OCR’s with R2 = 0,112. The main effect of type
of product is significant with p = 0,002 and β = -1,102 . The main effects of an analytical information processing style and an intuitive processing style are not significant with p = ,411 and p = ,386. The moderating effect of the information processing style on the effect of type of product on the motivation to read more OCR’s is not significant with p = ,763 and p= ,636. The interaction effect between the negative review and the type of product is almost
significant with p = 0,119 and β = -,811 for almost 90% of thepopulation. Two separate regression analysis, with either just the analytical information processing style or just the intuitive information processing style have been conducted. The results of model 4 correspond with the extra regressions, the significance levels were not significantly different.
Model 5 tested the main effects and interaction effects of the two experimental variables and effects of the moderators risk attitude and information processing style on the motivation to read more OCR’s after reading one review. The overall model is significant with p = 0.024 and explains 11,9% of the variance of the motivation to read more OCR’s after reading one review with R2 = 0,119. The main effect of type of product is significant with p = 0,002 and β
= -1,108 . The direct effects of the moderators risk attitude and information processing style on the motivation to read are not significant with p = ,359, p = ,425 and p = ,482. The
moderating effect of the risk attitude and the information processing style on the effect of type of product on the motivation to read more OCR’s after reading one review is not significant with p = ,359, p = ,873 and p= ,745. The interaction effect between the negative review and the type of product is almost significant with p = ,124 and β = -,802 .
Model 6 tested the main effects and interaction effects of the two experimental variables, the effects of the moderators risk attitude and information processing style and the control variables on the motivation to read more OCR’s after reading one review. The overall model is significant with p = ,000 and explains 30,7 % of the variance of the motivation to read more OCR’s after reading one review with R2 = ,307. The main effect of type of product is
38
,000 and β = ,423. Consumers who are have a a more positive attitude towards OCR’s are more motivated to read more OCR’s after reading one review while searching for online product information. Age is not significant on the motivation to read more OCR’s after reading one review. Gender does have a significant effect on the motivation to read more OCR’s after reading one review with p = ,008 and β = ,568. Women are more motivated to read more OCR’s after reading one review, compared to men.
4.4 Multicollinearity
The regression model 5 without mean-centering resulted in VIF-scores higher than 10. The interaction between the 2 factors of the moderator information processing style and the type of product resulted in VIF-scores for the type of product of 40,714, and for the two factors of information processing style: 23,096 and 26,689. One solution to solve multicollinearity is to create dummy variables based on the median split. The median split of the analytical
information processing style factor is: 4.75. A new dummy is created, which is 0 for all scores below 4.75 and a score of 1 for all scores above 4.75. The median split of the intuitive
information processing style factor is: 4.80. A new dummy is created, which is 0 for all scores below 4.80 and a score of 1 for all scores above 4.80.
A new ANOVA has been conducted and the SPSS output can be found in Appendix H. The main effect of type of product is significant with p = ,000. The main effect of valence is not significant with p = ,972. The interaction effect between type of product and valence is almost significant with p = ,110 for 90% of the population. The main effect of analytical information processing style on the motivation to read more OCR’s is significant with p = ,086.
Consumers with a more analytical information processing style are more motivated to read more OCR’s compared to consumers with a more intuitive information processing style. The main effect of the intuitive information processing style on the motivation to read more OCR’s is not significant with p = ,810. The moderating effects of both factors of the information processing style are not significant with p = ,388 and p = ,178.
4.5 Hypothesis testing
This section will test the hypothesis made in chapter two. This is based on the bivariate regressions, the analysis of variance and the extra analysis of variance due to
39
overview, Table 11, shows whether the hypothesis has been accepted, partly accepted or have been rejected.
Hypothesis Result
H1: Consumers searching online for experience products are more motivated to read more OCR’s after reading one review, compared to consumers searching online for search products.
Rejected
H2: The more negative (positive) the valence of the OCR read, the higher (lower) the motivation to read more OCR’s after reading one review.
Rejected
H3: The difference between a negative and a neutral OCR has a more positive effect on the motivation to read more OCR’s compared to the difference between a neutral and positive OCR.
Rejected
H4: A positive (negative) OCR negatively (positively) influences the effect of the type of product towards the motivation to read more OCR’s after reading one review.
Partly supported
H5: Higher (lower) risk attitude will have a negative (positive) effect to the motivation of consumers to read more OCR’s after reading one review.
Rejected
H6: The motivation to read more OCR’s after reading one review will be lower (higher) when consumers have a higher (lower) risk attitude when searching online for experience products compared to searching online for search products.
Rejected
H7: Consumers showing a more analytical (intuitive) information processing style are more (less) motivated to read more OCR’s after reading one review.
Partly supported
H8: The motivation to read more OCR’s after reading one review will be higher (lower) when consumers showing a more analytical (intuitive) information processing style when searching online for experience products compared to searching online for search products.
Rejected
Table 10: Hypothesis testing
40
Hypothesis 2 states: “The more negative (positive) the valence of the OCR read, the higher (lower) the motivation to read more OCR’s after reading one review”. The analysis of variance showed that the different levels of valence do not have a significant effect on the motivation to read more OCR’s. With p =,979 we can conclude that the valence does not have a main effect and that there are no differences between a positive, a neutral and a negative review. Regression analysis in all models also showed no significant effect of valence on the motivation to read more OCR’s after reading one review. Concluding, hypothesis 2 is rejected
Hypothesis 3 states: “The difference between a negative and a neutral OCR has a more positive effect on the motivation to read more OCR’s compared to the difference between a neutral and positive OCR.” No significant difference between positive, neutral and negative has been found in the model 2. Based on the reasoning above, and on the reasoning for hypothesis 2, hypothesis 3 is rejected.
Hypothesis 4 states: “A positive (negative) OCR negatively (positively) influences the effect of the type of product towards the motivation to read more OCR’s after reading one review”. The regression analysis shows that there is no significant effect between the type of product and a positive review. Model 3 shows that there is a significant effect between the type of product and a negative review for 90% of the population. With p = 0,095 and β = ,838 we can conclude that there is an interaction effect and that consumers searching for experience
products are more motivated to read more OCR’s after reading a negative review compared to consumers searching for search products after reading a negative review. Model 4 and 5 show also that this interaction effect is also almost significant with p = ,119 and p = ,124. The ANOVA showed an indication of an interaction effect with the negative valence. Hypothesis 3 is partly supported since there is no interaction effect between the type of product and a positive or neutral review, but there is an interaction effect between a negative review and the type of product for 90% of the population.
Hypothesis 5 states: “Higher (lower) risk attitude will have a negative (positive) effect to the motivation of consumers to read more OCR’s after reading one review”. Model 3 shows that there is no significant effect of the personality trait risk attitude towards the motivation to read more OCR’s. With p = 0,266 hypothesis 5 is rejected.
41
experience products compared to searching online for search products“. Model 3 shows that the moderating effect of risk attitude on the effect of type of product on the motivation to read more OCR’s after reading one review is not significant, with p = 0,274. Concluding,
hypothesis 6 is rejected.
Hypothesis 7 states: “Consumers showing a more analytical (intuitive) information processing style are more (less) motivated to read more OCR’s after reading one review”. With a
regression analysis, model 4 showed that there is no significant main effect towards the motivation to read more OCR’s of analytical information processing style with p = ,411 and that there is no direct main effect of the intuitive information processing style with: p = ,386. The betas were positive with β = ,125 for analytical information processing style and β = ,140 for intuitive information processing style. Due to multicollinearity an analysis of variance is conducted where both factors and two dummies were created, by splitting the factors with the median in 0 and 1. This new analysis of variance showed that the main effect of analytical information processing style on the motivation to read more OCR’s after reading one review is significant with p = ,086. Consumers with a more analytical information processing style are more motivated to read more OCR’s after reading one review, compared to consumers with a more intuitive information processing style. The main effect of the intuitive
information processing style on the motivation to read more OCR’s after reading one review is not significant with p = ,810. Concluding on both the analysis, there is a small main effect of an analytical information processing style and this is a positive effect on the motivation towards reading more OCR’s after reading one review. Therefore, hypothesis 7 is partly supported.
Hypothesis 8 states: “The motivation to read more OCR’s after reading one review will be higher (lower) when consumers showing a more analytical (intuitive) information processing style when searching online for experience products compared to searching online for search products”. The moderating effect of the information processing style on the effect of type of product on the motivation to read more OCR’s after reading one review is not significant with p = ,763 and p= ,636. In the extra analysis of variance, both interacting effects are not
42
5. Discussion and recommendations
5.1 Conclusion
This study delivers contributions to current research literature regarding the motivation of consumers to read more OCR’s after reading one review. Current research literature focused on the main motivation to read a OCR, but there is a lack of research in the area why and when consumers are motivated to read more OCR’s after reading one review. The aim of this study was to find out what variables influence the motivation to read more OCR’s after consumers read one OCR. This study investigated the effect of different type of products, the valence of the review, the risk attitude of consumers and their information processing style.
Based on chapter 4, there is a significant difference in the motivation to read more OCR’s after reading one review, depending on the type of product. Consumers who are searching for search products have a higher motivation to read more OCR’s compared to consumers who are searching for experience products, where consumers searching for experience products are more motivated to read more OCR’s after reading a negative review compared to consumers searching for search products after reading a negative review.
The valence of the review itself does not have a significant effect on the motivation to read more OCR’s, but there is an interaction effect between the type of product and a negative review for 90% of the population.
The attitude towards OCR’s does have a significant effect on the motivation to read more OCR’s after reading one review. Consumers with a positive attitude towards OCR’s are more motivated to read more OCR’s after reading one review. A significant difference in gender is also found, women are more motivated to read more OCR’s after reading one review,
compared to men.
5.2 Discussion
43
Weathers (2006) and a preliminary experiment, this study chose a backpack as a search product and a pair of jeans as experience product. One explanation why consumers were more motivated to read more reviews could lie in the perceived risk of both products. Cunningham (2005) found perceived risk to be a significant factor affecting internet consumer behavior. The frequency of the purchase lowers the perceived risk of the purchase (Heijden et al., 2000). Respondents could have perceived the backpack with higher risk compared to the purchase of a pair of jeans and therefore were more motivated to read more OCR’s for the search product compared to the experience product. That could be because consumers are purchasing a pair of jeans more frequently than a backpack. According to Huang et al. (2009) the Internet also blurs the distinctions between experience and search goods by providing mechanisms that enable online shoppers to gather information on experience and search attributes. It might be argued that other approaches to product classifications, such as Peterson et al. (1997) classification based on purchase frequency, tangibility, and differentiation or Weathers, Sharma, and Wood’s (2007) classification based on the perceived need to see, touch, or hear a product versus read manufacturer-provided information about product attributes, are more relevant in the online context (Huang, 2009).
Valence of the reviews did not have a significant effect on the motivation to read more OCR’s. Based on the negativity effect, it was expected that negative online reviews have a bigger impact, based on an assumption in psychology which states that negative information has a greater weight compared to equally strong positive information in creating judgements. The manipulation check showed that the respondents were assigned to different levels of valence, but it did not have a significant effect. Although the valence of review has been found to have a significant influence on consumer behavior, , other studies have found no effects for review valence (Ketelaar, 2015). For instance, in a study conducted by Cheung et al. (2009) across a variety of products, no significant effects of review valence were found. It could be possible that valence of a review just does not have an effect on the motivation to read more OCR’s after reading one review while searching for products online.
