The effect of positive online user message content on consumer’s intention to buy An experimental examination of user message content, construal level and perceived
credibility
Author: Jeffrey Brigitha
Student number: 10001534 Date of Submission: August 2, 2015
Qualification: MSc. in Business Administration – Marketing Track Institution: University of Amsterdam
2 Statement of originality
This document is written by Jeffrey Brigitha who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
3
C
ONTENT
Abstract ... 5
1 Introduction ... 6
2 Literature review and hypotheses ... 10
2.1 Conceptualization of eWOM ... 10
2.1.1 eWOM content ... 11
2.1.2 Arousal and functional eWOM ... 12
2.2 Construal-level ... 15
2.2.1 The level of abstractness ... 16
2.3 Perceived credibility ... 19
2.4 Purchase intention ... 24
2.5 Conceptual model ... 25
3 Data and method ... 27
3.1 Research design ... 27
3.2 Measurement ... 28
3.2.1 Functional and arousal content manipulations ... 28
3.2.2 Level of abstractness ... 29 3.2.3 Perceived credibility ... 29 3.2.4 Purchase intention ... 30 3.3 Analyzing data ... 30 3.3.1 Missing values ... 30 3.3.2 Computing variables ... 31 4 Results ... 32
4.1 Basic data analysis ... 32
4.1.1 Sample ... 32
4.1.2 Manipulation checks ... 32
4.1.3 Validation of key constructs ... 34
4.1.4 Correlation analysis ... 34
4.1.5 Conceptual model analysis ... 35
4.2 Hypotheses test ... 37
4
4.2.2 The overall effect of the model ... 39
4.2.3 The effect of the level of abstractness ... 39
4.2.4 The effect of perceived credibility ... 40
4.2.5 Correlation between the independent variables ... 40
4.2.6 Mediating effect ... 41
5 Discussion, practical implications and future research ... 43
5.1 Discussion ... 43
5.2 Practical implications and future research ... 47
6 Conclusion ... 50
Bibliography ... 51
5
A
BSTRACT
This thesis empirically examined the influence of positive functional and arousal user
message content on purchase intention via the construal level of the mental representation of the message and perceived credibility. Existing literature established the compelling role of informational and emotional content in the effectiveness of eWOM. In this study the focus is on the functional aspect of informational content and the arousal aspect of emotional content. Emotional content in eWOM focus more on the feelings whereas informational content focus more on product attributes. Therefore, the functional content is used to describe the product characteristics where the arousal content is used to express the user’s feelings and
experiences. This research extends previous literature by including two mediating variables, the level of abstractness and perceived credibility. Since the perceived credibility is the second order mediator, the assumption is that the level of abstractness has an impact on the perceived credibility. The influence of the functional and arousal content is investigated through an online experiment. The results partially supports the generated hypotheses which were in line with existing theories. Nevertheless, this study provides researchers and
practitioners with new insights on the role of eWOM content and the mediating role of the level of abstractness and perceived credibility.
6
1
I
NTRODUCTION
The major developments of the internet has had an important influence to the behaviour of consumers, companies and to the field of marketing in general. The internet turned into one of the most essential marketplaces for the transactions of goods and services (Leeflang, Verhoef, Dahlström, and Freundt, 2014). Before the technological breakthrough of the internet,
consumers already networked and informed other consumers about their experiences with products and services or brands and companies. This phenomenon, referred to as word of mouth (WOM), is an important circulation which, both positive as negative, has major
influences on consumer behaviour. If word of mouth is used in a correct way, it is able to built a certain perception about products and services and promote them in a positive way for the long-term. An important process of communication on the internet is electronic word of mouth (eWOM). eWOM is specified as the exchange of negative or positive information about a product or company by existing or future consumers, and, which is made available through the internet to a wide public (Henning-Thurau et al. 2004).
One difference between eWOM and WOM is that consumers and companies can reach more people within shorter time periods with eWOM through social media and web-based opinion platforms. Messaging through the Internet travels faster and therefore reaches
consumers faster. Besides, the recipients of eWOM often want easy and less expensive access to information that can influence the purchase decision (Goldsmith and Horowitz, 2006). The two contexts for eWOM are information-oriented and emotion-oriented context. eWOM in information-oriented context focus primarily on product performance whereas eWOM in emotion-oriented context emphasize more on the subjective product experience (Daughtery and Hoffman, 2014). A well known theory within the field of information transmission is the dual-process theory in which two content sources are described that affect the persuasiveness
7
of the communication. Two content specific influencers are informational and emotional written messages. The emotional influence is described as a message where emotional feelings, either positive or negative, are aroused. The valence is an extensively studied eWOM content element which will not be outlined further in this study. The level of arousal is an interesting content element which is explained as the level of activation that is evoked by the emotional message (Berger and Milkman, 2012). For example, a relax or content feeling is related to a low level of activation while feelings of excitement creates a high level of
activation. On the other hand, eWOM content can be written from an informational perspective. When the informational message provides a functional need, the eWOM will consist of content-specific words that describe a product on its functional characteristics (Chih et al., 2012). The amount and extent of functional words in the message may contribute to the argument quality and effectiveness of the message. The effect of the textual expressions that differ in the level of arousal and functional content is examined more extensively in this study.
The findings in the literature about the interpretation of eWOM state that content-specific factors are processed in different ways due to the level of abstractness and perception of credibility. For example, the construal level theory argue that a higher level of abstractness result into a higher level of construal and, therefore, the content is processes differently (Trope, Liberman, and Wakslak 2007). Functional content in messages often consist of less abstract words than arousal content (Li, 2013). Other studies provide evidence that
informational and emotional content of an online message has a positively influence on the perceived credibility of the message. However, in this study the focus is primarily on the functional and arousal aspects. A higher perceived credibility may influence the level of persuasion of the message which makes the eWOM message more effective (Park, Lee, and
8
Han, 2007). The effectiveness of the eWOM can be reflected by the consumer’s intention to buy, since it is very difficult to measure the actual purchase (Paul and Wang, 2011).
The valence of the online user message is important since it defines the impact on the persuasiveness of the message. However in this study the emphasize will be on the textual eWOM content characteristics rather than the valence of the message. Interesting to study is that content with the same valence, which is positive messaging in this study, but with two different arousal or different functional strength may have different effects on the
persuasiveness (Berger and Milkman, 2012). Within the field of eWOM, many researchers studied the effect on purchase intention. However, not much research is conducted about the eWOM message content with the focus on functional and arousal message content. Interesting therefore will be to examine the difference between functional and arousal content and how they both affect consumer’s purchase intention. Both the functional as the arousal message content is published by a product user. In order to fully understand the relationship between the construal level and perceived credibility, the mediating effect is examined. For example, what is the influence of the positive functional and arousal content on the level of abstractness and, in turn, on the intention to purchase? Besides, will the difference in functional and
arousal eWOM content create a different perception of the credibility and purchase intention? The following research question includes the two independent variables and emphasizes the contribution of this study: How does positive functional and arousal online user message content affect consumers’ purchase intentions via the construal level of the mental representation of the message and its perceived credibility?
The role of positive online user message content and the influence on consumer message processing is thoroughly explained. Furthermore, the main contribution is to include the construal level and perceived credibility as mediators. Existing research conclude that informational message content generates a low level of construal because information is
9
perceived as concrete. This study will emphasize on the functional content and, besides, tests whether arousal content generates a high level of construal due to its abstract perception (Putte, 2009). On the other hand, a high level of perceived credibility by the consumer is generated when the consumer perceive the message with a high level of trustworthiness (Park, Lee, and Han, 2007). The general assumption of this study is that the mediator effect of construal level is more related to the informational user message content and the mediator effect of perceived credibility to the emotional user message content. This assumption is based on the study of Park et al. (2007) where they suggest that consumers prefer personal feelings and experiences in product reviews. The managerial contribution of this research is recognized in the element that eWOM influences other variables before they affect the consumer’s purchase intention. The elemental motivation of eWOM effectiveness contribute to the theoretical and practical field of (online) marketing.
To continue, this thesis starts with the literature review describing contradicting findings within the domain of eWOM, construal level, perceived credibility and consumer’s purchase intentions. Within the theoretical framework the hypotheses which are examined will be outlined, followed by the conceptual framework. Subsequently, the data and method section will clarify more about the research design and process. Thereafter, the major results are explained and the supported and/or rejected hypothesis are outlined. Finally, this thesis ends with a discussion and conclusion to the research question.
10
2
L
ITERATURE REVIEW AND HYPOTHESES
2.1 Conceptualization of eWOM
Gradually, the effect of eWOM on consumers is to receive attention (Paul and Wang, 2011). The difference between eWOM and commercial advertising is that the eWOM message exists of non-commercial content and includes direct experiences from the consumer that posts the message. The traditional WOM is generally considered as more credible than commercial advertisement and the effect of eWOM on sales is also higher (Herr et al., 1991). The reason for this high credibility is that people gain product information from people they know and trust. Companies’ motive to provide information is to persuade customers to buy their
product. With eWOM, the information gained can be from an anonymous person. This makes it more difficult for consumers to decide whether the message is credible and to feel identified with the person that publishes the message. The source credibility has a major effect on the believability of an online user message. This phenomenon can be described as to what extent the reader of the message believes the sender of the message. The credibility of the source positively influences the attitude towards the source and the persuasiveness of the message Gunther, 1992; Hovland and Weiss, 1951). Besides, the persuasiveness of the message
depends on whether the message is rational or emotional oriented. Rational oriented messages focus on more on the functional benefits and product characteristics whereas emotional messages focus more on the feelings in order to impress the reader (Keller, 2007). In order to use eWOM as an effective marketing strategy, it is important to know that the effect of eWOM is influenced by the message appeal and source credibility (Paul and Wang, 2011).
Two types of eWOM contexts are described in the literature, information-oriented eWOM and emotion-oriented eWOM (Wang and Rodgers, 2010). Information-oriented eWOM context is often referred to as websites such as opinion platforms where reviews are
11
given and product feedback is generated. The content of this information is generally related to product performances and consumers’ ‘objective’ experiences with the product/company. Emotion-oriented context includes platforms that are not product focused such as online communities and social media networks. On these platforms, the eWOM content often
includes broader product experiences which are expressed in a subjective way (Daugherty and Hoffman, 2014). The messages are led by emotions which include the feelings of the
consumer towards the product, brand and/or company. The influence of ‘informational’ and ‘emotional’ eWOM content is broadly explained in current literature. A well known theory is the dual-process theory where informational influence and normative influence are described as the sources that affect the persuasiveness of a message. The informational influence is described as information that is determined by the content quality, source of the message and/or the interpretation of the person who reads it. On the other hand, the normative influence is not determined by the information that is received but more on the opinion of other people (Chueng et al. 2009). The dual-process theory highlights that both dimensions, informational and normative, have a significant influence on the persuasiveness and
credibility of the message.
2.1.1 eWOM content
The content of eWOM messaging is a wide concept and can be tested and interpreted in multiple ways. The most often studied field within eWOM content concerns the valence of the message. The effect of the eWOM message on the consumer depends on whether the message is positive or negative. Even though positive eWOM messages influence the reader, negative eWOM messages may have a greater impact on consumer behaviour (Henning-Thureau et al., 2004). Other studies, for example by Berger and Milkman (2012) indicate that positive
content on the internet is more viral and shared more often than negative content.
Contradiction exists in the current field of research about the effectiveness of eWOM valence. Furthermore, the content of the eWOM message can be interpreted as the subject discussed in
12
the message. For example, is the message directed to the company, the brand or the
product/service (Pai, Chu, Wang, and Chen, 2013)? The words that are used are essential to the content of the eWOM and they also influence the message appeal. The message appeal affect the persuasiveness of the eWOM and as a consequence the behaviour of the consumer. Informational appeal is described as a message where the consumer benefit is used as an attribute. The emotional appeal is described as a message where positive or negative
emotional feelings are aroused to influence consumer behaviour (Keller, 2007). In a similar study by Putte (2009) about advertising content strategies, the information strategy is
described as providing instrumental advantages and focusing on the quality and properties of the product in a rational way. On the other hand, the emotional strategy emphasize on the feelings of the user and clarify which feelings the consumer can experience from using the product. Park and Lee (2008) describe the two different message appeals as attribute value reviews and simple-recommendation reviews. The attribute value reviews include
informational content and are written from a rational perspective based on facts. The simple-recommendation reviews consist of non-relative information and provide a subjective
recommendation based on emotions and feelings. The emotional content is partly determined by the level of arousal in the message. The informational content is determined by the amount of functional words in the message. These two concepts, arousal and functional user message content, are described more in the following section.
2.1.2 Arousal and functional eWOM
Different theories exist within the study of emotions. Within an existing study of Jalonen (2014), the psychological researchers divide emotions into two axes which are classified as valence and arousal. As described earlier, valence is explained as the positive or negative state of the emotion. In this study, arousal is described as the extent to which the emotion is
expressed and the activity that induces the emotion (Schweitzer and Garcia, 2010). For example, different people can have positive feelings towards a brand but person A describes
13
himself as being astonished by the brand while person B is satisfied with the brand. Both emotions are positive but person A encourages the brand while person B’s description is discouraging. On the other hand, the research about viral online content of Berger and
Milkman (2012) state that emotions differ on the level of psychological arousal, referred to as the level of physical activation that is evoked. Content that evokes high positive arousal such as awe and amusement is more viral than low positive arousal such as contentment (Berger and Milkman, 2012). These researchers describe that the differences in the level of arousal influence the social transfer of the message and action-related behaviour. Contentment is marked by feelings of satisfaction, surety, safety and relaxation. High positive arousal
emotions are feelings of awe, excitement, amusement and humour. Surprise is one of the main high arousal emotions on social platforms which can be either positive or negative. Different content might have different causal impacts because the emotion is induced and driven by a different level of activation. The level of activation in online user messages depends on the words which are used to express the emotion and which have different emotional driver strength. Especially within the area of social media it is interesting for organizations to have the expertise about the specific emotions that should be evoked to influence consumer
behaviour. Besides, organizations can provide and communicate content themselves which is based on consumer’s preferences and inspires consumer’s emotions. This is a two-way interaction between organisations and consumers but also between consumers and consumers (Jalonen, 2014). The arousal content is part of the emotional and social need that eWOM should provide. A study by Ha and Lennon (2010) examined how online product-related cues influence internal states such as pleasure, arousal and behavioural response. In this study a paradigm demonstrated by researchers found that a high level of arousal, evoked by external stimuli in online settings, improves the level of satisfaction and purchase intention (Eroglu, Machleit and Davis, 2003; Fiore, Jin and Kim, 2005). For example, the external stimuli
14
include the colours, music and textural characteristics that are used in the online message. Wu, Cheng and Yen (2008) also argue that arousal, which is induced by the online
environment of a store, increase the behavioural intention and approach behaviour. Purchase intention and behavioural intention are explained as approach behaviours which reflect the shopping behaviour and may consist of positive outcomes such as the intention to explore, spend money and return to the sites (Ha and Lennon, 2010). Based on this rationale, the following hypothesis is developed.
H1: When the user message includes high arousal content, the purchase intention will be higher than when the user message includes low arousal content.
On the other hand, eWOM also provides a functional need (Chih et al., 2012). The functional need is derived from the social orientation through the information that is provided. A similar rationale is explained by the dual-process theory of Cheung et al. (2009) where the functional need refers to the informational determinants and the social need to the normative
determinants. However, the functional information is obtained from real evidence and its processing is based on the reader’s judgement of the information. The functional eWOM described in this study is related to the content-specific words that are used to describe a brand or product on its functional characteristics. The functional content is part of and contributes to the argument quality. The argument quality is determined by the
comprehensiveness, accuracy, timeliness and relevance of the message. If the online user message content is perceived as qualitative information by the reader, he or she will create a positive feeling towards the brand or product and the message will also be perceived as credible (Teng et al., 2014). The functional message is characterized as utilitarian eWOM where the consumer interprets the message as attribute-value and/or attribute-benefit
15
information. Furthermore, this type of content is discussed as often being more concrete, objective and rational (Wu and Wang, 2011)
In line with existing research, I predict that informational content has more persuasive power than non-informational content which, as a result, positively influences the purchase intention (Putte, 2009). Therefore, high functional content is perceived as more persuasive than low functional content. This will lead to the second hypothesis.
H2: When the user message includes high functional content, the purchase intention will be higher than when the user message includes low functional content.
2.2 Construal-level
The construal level theory (CLT) explains that concrete near events have low levels of construal and distant, abstract events have high levels of construal. Low levels of construal include unstructured representations of the future where high level construal includes a schematic representation. Simultaneously, low level construal events are described richly in detail with mainly primary information where high level events are profoundly abstract and often consist of secondary information (Trope, Liberman, and Wakslak 2007). Primary characteristics of products are the central features and secondary characteristics the peripheral. Central features are for example the engine capacity of a car while a second
feature can be the display on the dashboard. According to Trope et al. (2007), no difference in valuations arise for near future events. For distant future events, the central features are valued more than the peripheral features. For the example, this suggests that for distant future car purchases, more value is given to the engine capacity than to the display. Another
distinction in the CLT can be made between the desirability and feasibility. High level construal emphasize on the desirability features and low level construal assert feasibility features. Feasibility-related information covers the ‘how’ aspects of a situation whereas desirability-related information covers the ‘why’ aspects of a situation (Trope, Liberman, and
16
Wakslak 2007). For example, when guests are motivated to attend a lecture in the near future or distant future. In the near future, more value is given to the timing of the lecture
(feasibility) whereas for the distant future more value is given to the level of interest in the lecture (desirability). Psychological distance is subjective and explains the distance from the self at a specific moment (now) and specific place (here). Psychological distance
encompasses four elements which all have their share in the level of construal.
2.2.1 The level of abstractness
The level of construal in a product review is dependent on the word characteristics that are used in terms of abstractness. A low-level of construal message consists of detailed and precise information. Low-levels of construal imply that there are concrete words used in the messaging and high-levels of construal suggest that there are abstract words used in the content. For example, when a technological product is reviewed, low-level construal words focus on feature characteristics like colour, size in centimetre and capacity characteristics like storage room or power usage based on numbers. On the other hand, high-level construal words do not focus on the unique characteristics and describe the product in general. For example, the product can be described as big and a good price-quality. These characteristics are hard to interpret and contain less detailed information (Trope, Liberman and Wakslak, 2007). Different levels of abstractness can be used to describe a product. When a product is characterized in an abstract way, there is more room for multiple interpretations from the reader’s side. The statement will become more difficult to verify because it offers less specificity about product characteristics. According to Zeithaml (1988), a higher level of abstraction is related to descriptions of convenience or emotional feelings. Information about products is retained in the memory at different levels of abstraction. When product
information becomes more emotional oriented like for example the emotional payoff, the level of abstractness increases. Still, this does not explain the relationship between the previously discussed arousal content and level of abstractness since arousal is an element of
17
the emotional construct. Previous research investigated the relationship between the semantic size of abstract concepts and the emotional arousal. Kousta, Vigliocco, Vinson, Andrews and Del Campo (2011) argue that abstract concepts are more emotionally charged than concrete ones. For abstract words there exists a stronger link between the semantic size and arousal characteristic than for concrete words. Overall, arousal content is stronger related with abstract content than with concrete content but is dependent by the word size (Yao, Vasiljevic, Weick, Sereno, O’Donnell and Sereno, 2013). This research is extended but without the inclusion of the semantic size of text. In this study the relationship between the emotional content, focusing on valence and arousal, and concreteness rating is examined. The results indicate that as the level of arousal increases, the concreteness rating decreases
(Vigliocco, Kousta, Della Rosa, Vinson, Tettamanti, Devlin and Cappa, 2013). This finding is in line with existing research and results in the following hypothesis.
H3: When the user message includes high arousal content, the level of abstractness will be higher than when the user message includes low arousal content.
On the other hand, functional information relates to the relevant features of a product that emphasize on the functional solutions. In line with the construal fit theory in the context of persuasive messaging, laddering up consists of presenting a brand feature, then emphasizing on the functional benefits and ultimately elaborating on the emotional benefits of the product. The functional benefits are discussed as low-level construals and emotional benefits discussed as high-level construals (Lee, Keller and Sternthal, 2010). This theory is in line with the finding discussed above which suggest that low-level construal descriptions focus on functional features. Low level construals use more concrete words in the communication whereas high level construals characterizes a more abstract language. Based on these theories the following hypothesis is advanced.
18
H4: When the user message includes high functional content, the level of abstractness will be lower than when the user message includes low functional content.
When emphasizing more on the construal fit theory, previous research findings indicate that information with promotion focus is processed at different construal levels. The usage of the construal fit in the communication strategy of organisations enhances the effectiveness and persuasiveness of the message. For example, abstract high-level construal are adequately informing about the desirability of an activity or product. Furthermore, a fit is experienced by the reader if the high level message focus on the pursuit and achievement of the goal. On the other hand, a concrete and thus more detailed, low level approach encourages a prevention focus since the descriptive elements of the message enhance the goal of security and safety (liberman et al., 1999). Important from this theory is that a regulatory fit exists when the focus of the people, either prevention or promotion, matches the communication strategy (Higgins, 2006). Therefore, the persuasiveness of the message depends on the focus of the reader. For example, people who have a prevention focus will be more persuaded if a message about sunscreens accentuate the effectiveness rather than the ease of use. However, the opposite communication strategy will result in more persuasive messaging if the reader has a
promotion focus. The results of the study by Lee, Keller and Sternthal (2010) are in line with existing research. Their research indicate that the brand attitude of promotion focused
consumers will be more positive for high level construal product messaging than for low level construal product messaging. For prevention focused consumers the opposite is true and those will have more positive brand attitudes towards low level construal product messages.
Furthermore, According to Li (2013), the helpfulness of a review is indicated by the authorship and the abstractness of the review content. They conclude that the messages with the highest levels of abstractness is paired with the lowest level of helpfulness of the review.
19
The more concrete the message is, the more the review is perceived as helpful in the eyes of the consumer. A study by Borgida and Nisbett (1977) argue that abstract messaging lack cognitive information and does not trigger the reader in a cognitive way. Cognitive messaging is similar to functional message content in a way in which both include a rich amount of details and objective information. Furthermore, when consumers experience a more concrete mental construal they feel a higher level of involvement and this in turn results in higher purchase intentions (Luo, Andrews, Fang, Phang, 2014). However, in contradiction with these findings, another research by Schellekens, Verlegh and Smidts (2010) state that if the product experience description in word of mouth is positive, abstract descriptions result into a higher purchase intention than concrete descriptions. Although contradiction about the influence of language abstraction exists across studies, the expectation in this study is that lower levels of abstractness are more persuasive and result in higher levels of purchase intention. This will lead to the following hypothesis.
H5: When the level of abstractness is low, the purchase intention will be higher than when the level of abstractness is high.
2.3 Perceived credibility
The perceived credibility of online user messages is a broadly studied concept because of its importance in the field of word of mouth. eWOM is the online interaction between
organizations and consumers and often between consumers. Therefore the credibility of the message is at all times questioned (Lis, 2013). eWOM credibility is described in literature as the extent to which the reader of the message perceives the content as believable, true or factual (Simons, 2002). On the other hand, the level of abstraction is lower when product information is more detailed, are easier to interpret and more concrete in the description of product characteristics (Trope et al. 2007). The richness of the text, contextual details and
20
clarity in the memory of the reader are some of the characteristics of the cues of concrete language. When textual content is perceived as more vivid and consists of more detailed information, the reader will interpret the text as more real (Schooler, Gerhard and Loftus, 1986). This finding is supported by DePaulo et al. (2003) in a study about deception who state that shorter and less detailed information is often untrue and people in general are aware of this fact and also believe truthful information is often more detailed (Vrij, 2008). The amount of detail and the usage of concrete feelings in police reports are even used as cues for
credibility (Akehurst et al., 1996). Moreover, the likelihood and representativeness may increase when the amount of detail and concreteness in message contents increases (Tversky and Kahneman, 1982). There are reasons that clarify why concrete information is perceived as more true and believable. Firstly, concrete words are more recognizable than abstract words and thus also easier to understand and process (de Groot, 1989). Secondly, concrete words are faster recalled in the memory than abstract terms and therefore also more easily remembered (ter Doest & Semin, 2005). The relation between memory and concrete content results into familiarity which in turn contribute to the perception of true information. These findings taken together provide evidence that the reality perception, likelihood and truth of the mental
representation is associated to the level of abstractness/concreteness. Therefore, the following hypothesis is advanced.
H6: When the level of abstractness is low, the perceived credibility will be higher than when the level of abstractness is high.
When online user messages are posted, an online platform creates an interaction process between consumers about specific products, services or brands. Consumers that use or has any experience with one of these three share their thoughts and experiences with other (future) consumers. Compared to information distributed by companies, information by other
21
consumers is perceived as more credible. Companies are less willing to provide negative information about their products and the focus of companies is primarily on the superior aspects of the product (Park, Lee, and Han, 2007). Consumers, on the other hand, do not benefit from the sales of the product and are therefore willing to provide information that highlights the positive aspects but more importantly also the downsides of the product. Furthermore, the consumers see each other as peers because they are in the same situation. They face the similar dilemma’s with buying a certain product and can advise each other in buying situations where companies do not have the experiences or just lack the willingness to share information (Moran and Muzellec, 2014). Therefore, the source of the information influences the perceived credibility (Park, Lee, and Han, 2007).
According to the framework of Moran and Muzellec (2014), the perceived eWOM credibility is dependent on the source credibility and the message credibility. The source credibility is, besides the argument quality, one of the determinants of informational content. The four C’s described by Moran and Muzellec (2014) are community, competence, content and consensus. The first two C’s influence the source credibility and the latter two influence the message credibility. In this study the focus will be on the message credibility. In order for eWOM to effectively influence the reader, it should be interpreted as credible in the eyes of the reader. The content of eWOM is described as the clarity of the message and valence of the message which, as a consequence, influence the persuasiveness of the communication and establish the effectiveness of the marketing strategy (Park and Lee, 2009). The clarity of the message focus on the way the reader interprets the message based upon its ambiguity and abstraction. When the message is perceived as more informational, the attitude towards the brand or product will be more positive. An informational message influence the reader’s level of knowledge or cognition about the product (Wu and Wang, 2011). Besides, the information that is present in the message, such as functional text-based information about the product, is
22
used by the reader to make an evaluation. For example, within the functional food industry, researchers studied the concept of functional food and its impact on the credibility. Functional foods are described as foods that contain specific physiological benefits to health in
combination with the sensory characteristics such as colour, texture and flavour (Astawan, 2005). This clarification of functional content is in line with the functional content
characteristics described before in this study. Furthermore, the researchers suggest that people intent to buy food products due to the credence of the functional contribution of the food. High source credibility advertising for functional foods leads to a higher confidence level amongst the consumers than low source credibility. The findings clarify that functional textual aspects provide more assurance on the quality and reduces the perception of risks (Biswas et al., 2006). The study about functional food advertisement support the statement that
functional product features are linked to the perceived credibility. Still, important to note, the level of involvement of the consumers play an important role. Consumers who are high
involved in the product are more sensitive to functional content attributes than consumers who are low involved in the product (Soliha, Dharmmesta, Purwanto, Syahlani, 2012). Based on the research findings discussed above, functional content increases the level of assurance, trustworthiness and the perceived credibility. High positive functional content would therefore increase the level of perceived credibility more than low functional content. So, the following hypothesis is formulated.
H7: When the user message includes high functional content, the perceived credibility will be higher than when the user message includes low functional content.
In line with the dual-process theory, the perceived credibility of online product messages is, besides the functional information content, influenced by the normative factors. The
23
product users (Cheung et al., 2009). From an emotional oriented perspective, a reason why consumers like consumer reviews more than company reviews is that the consumer reviews include personal feelings and do not describe the product purely by its objective specifications (Park, Lee, and Han, 2007). Furthermore, the research by Park et al. (2007) also suggest that consumers value eWOM by other consumers more than by organizations. The personal opinion and emotional experience of the product’s user is appreciated and taken into consideration. As a consequence, content that includes functional and arousal words may therefore contribute to the information credibility of the online user message. If the user message is perceived as credible, he/she will receive the message positively and use the review for future decision making (Cheung et al., 2009). The persuasiveness of the message is the outcome which is influenced by the amount of arousal content. Accordingly, message senders influence and can target the receiver of the message by manipulating the message. The effectiveness and credibility level of the message depends, besides the content of the message, also on the receivers’ personal characteristics, the product category and the context in which the message is provided (Wu and Wang, 2011).
Arousal content in user messages is valued by consumers due to its honesty compared to information provided by the organization (Park et al. 2007). The perceived credibility of user generated messages is therefore higher than organization generated messaging. High arousal content would therefore result in a higher perception of the credibility than low arousal content. This will result in the following hypotheses.
H8: When the user message includes high arousal content, the perceived credibility will be higher than when the user message includes low arousal content.
The level of information in the message determine the actual weight in the final purchase decision. Both the image as the text in the user message determine the credibility and its
24
weight in purchase decision. Besides, the credibility can explained as the brand credibility, source credibility and message credibility. In this study the focus will be only on the textual characteristics of the message credibility. But, the reader uses the quality of the text to determine credibility of the source. Therefore the message credibility and source credibility are correlated. The quality of the text, especially the information within the message,
influence the trustworthiness of the message and the source of the message (Wu, 2013). The underlying thought essentially is that people use the information from the online message to determine how much they can rely on the information and to make an evaluation of the source. A high level of perceived credibility in eWOM is positively related with a high level of trustworthiness and has more persuasive power which positively influence the intention to purchase (Li and Zhan, 2011; Prendergast, Yin and Yuen, 2010). The perceived text
credibility is positively associated with source trust. As a consequence, source trust positively influences purchase intention. This leads to the final hypothesis (Li and Zhan, 2011; Hamilton and Mineo, 1998).
H9: When the perceived credibility is high, the purchase intention will be higher than when the perceived credibility is low.
2.4 Purchase Intention
Within the literature of eWOM communication, purchase intention is one of the most popular outcome variables (Bataineh, 2015). The purchase intention is used as an outcome variable if researchers want to test the influence on the purchases. Since actual purchase is, in most studies, very difficult or even impossible to test, researchers use the purchase intention as measurement. The purchase intention is the mediating variable precedent to purchase
behaviour. Within the field of eWOM, the intention to purchase depends on the favourability or unfavourability of the online user message and its concept can be defined as: the
25
product (Wu and Wang, 2011). Positive online user messages are positively related to purchase intention and can directly influence the customer’s product choices (Chang and Chin, 2010). Both the online quality and online quantity of the message is positively associated with the intention to purchase (Do-Hyung et al., 2007). Other influential factors which are positively related to purchase intention and often referred to in the literature are the persuasiveness of the message and product attitude (Park and Lee, 2009).
2.5 Conceptual model
The focus of this study is on the influence of eWOM content on consumer behaviour within the field of online marketing. The message content refers to the online message that is published by a user of a specific product. The two distinct content characteristics of the
independent variable are functional user message content and arousal user message content. In this study the functional content and arousal content can both be expressed as either high or low. Four different combinations of the independent variables, showed by figure 1, are manipulated: high functional/high arousal, low
functional/high arousal, high
functional/low arousal and low functional/low arousal. The effectiveness is determined by the effect of the independent variables on the dependent variable: purchase intention.
Purchase intention can be described as the intention to buy a product, which is different from the actual behaviour (Park, Lee, and Han, 2007). This effect will be the main effect in this experiment.
26
Besides, the effectiveness of the positive user message content on the purchase intention is dependent on the mediating effect of the construal level of the mental
representation of the message and the perceived credibility. In this study, the construal level of the mental representation of the message refers to the level of abstractness of the user message content. When the user message content includes more abstract messaging, the construal level is higher (Li, 2013). The perceived credibility is a result of the trustworthiness of the message content where higher level of trustworthiness is in line with a higher
perception of message credibility (Park, Lee, and Han, 2007). Figure 2 illustrates the (direction of the) variables in this study.
27
3
D
ATA AND METHOD
3.1 Research design
A fictitious brand is chosen in the research of this study to avoid an effect of brand image on the measured relationships. The product category is the smart phone industry which is deliberately chosen since (almost) everyone is familiar with the smart phone industry.
Particular to this study, there is a high chance that the respondents are familiar with the smart phone industry and specifications of smart phones since they are also participating in this study on the internet. Besides, the smart phone industry represent shopping and convenience goods which, in this study, demonstrates external validity as to this product category both rationale and emotional values are important (Keller, 2007). The fictitious brand is called Symphony and its imaginary nature causes no association to arise with an existing brand.
The experiment was distributed online via mailing lists of the University of Amsterdam, personal Facebook account and through forwarding e-mails and other online messaging. After the participants received the link to the experiment, they could choose between a Dutch and English version. Then, they were introduced to the conductor of the experiment, the goal of the experiment and the buying situation. The participants were also informed about the non-existence of the brand. Furthermore, they were asked to make decisions in the experiment as if they were in a real-life situation. After continuing the experiment, the participants were randomly assigned to one of the four manipulations. Hereafter, they answered questions that measured the following sequence of variables:
- The purchase intention
- The perceived credibility
28
Next, they were asked to state to what extent they find the online user message content functional and to what extent arousal. After completing these questions the participant was asked to fill out demographic questions and is thanked for his/her participation.
3.2 Measurement
3.2.1 Functional and arousal content manipulations
The independent variable consists of positive functional user message content and positive arousal user message content. Functional user message content consists of an online post by a person who used a product and gives information about functional advantages, quality and/or properties of the product. The information given is primarily rational. Arousal user message content consists of an online post by a similar person, but who focuses on the arousal he/she feels by using the product or the feelings that can arise when using the product. The effect on the purchase intention of these two treatments can be measured on a net promoter score ranking from one to ten. The main effect is illustrated by four situations where the participants either read a positive user message with high functional/high arousal content, low
functional/high arousal content, high functional/low arousal content or low functional/low arousal content. The variable functional content is manipulated by using highly functional product characteristics like ‘8mm thin smart phone’ and ‘5,1 inch large screen’ whereas for the low functional manipulation product descriptions were used like ‘thin smart phone’ and ‘large screen’. Arousal content is manipulated by making distinction between high positive arousal and low positive arousal feelings. Feelings that represent high positive arousal are feelings such as amusement, awe and excitement. On the other hand, low positive arousal are feelings like contentment, satisfaction, surety and safety (Berger and Milkman, 2012). The arousal content variable doesn’t focus on the product attributes but more on the feelings of the user and the experience of using the phone. The high arousal content characterizes
descriptions such as ‘the camera on the back of the phone works impressive and makes
29
the back of the phone works pleasant and makes fulfilling real-life pictures’. In the high arousal user message words such as ‘impressive’, ‘amazing’, ‘astonishing’ and ‘surprising’ are used to describe the feelings of the user. On the other hand, the low arousal user message includes words like ‘easygoing’, ‘carefree’, ‘fulfilling’ and ‘makes me feel at ease’. The manipulations are broadly similar in content but deviate for approximately ten per cent.
3.2.2 Level of abstractness
The effect of the user message content on purchase intention is mediated by the level of construal and perceived credibility. The level of construal refers to the abstractness of the message content and is measured on two items for each manipulation. To test the effect of the user message content on the level of construal, two situations are created. The items that measure the level of abstractness are (1) the review did not help me to evaluate the product, and (2) the review helped me to understand the performance of the product (Li, 2013). The results were measured on a 7-point likert scale anchored by ‘1= strongly disagree and 7= strongly agree’. The first item is phrased in a negative sentence to prevent that the results are influenced by a ‘yes-saying’ bias.
3.2.3 Perceived credibility
The other mediator is the credibility of the user message perceived by the reader. This
mediating variable can be perceived as relatively high or low. A low perceived credibility is a result of a perceived untrustworthy message content and a high perceived credibility is a result of a perceived trustworthy message content. Two items are questioned to measure the perceived credibility (1) I perceive the review is not good, and (2) I perceive that the review is trustworthy (Li, 2013). The results were measured on a 7-point likert scale anchored by ‘1= strongly disagree and 7= strongly agree’. Again the first item is phrased in a negative sentence to prevent the ‘yes-saying’ bias.
30 3.2.4 Purchase intention
Purchase intention is defined as the probability and possibility that indicates the willingness of the consumer to purchase a product (Wu and Wang, 2011). The purchase intention is the outcome variable in this study which indicates the attitude of the consumer towards the product and its actual behaviour. Two items are designed to measure the purchase intention of the main effect and indirect effect through the level of abstractness and perceived credibility. The items are (1) how likely would you choose the Symphony smart phone when you want to purchase a phone that is similar in nature? And (2) would you recommend the Symphony smart phone to your friend? Both items are measured on a 7-point likert scale hold by ‘1= strongly disagree and 7= strongly agree’.
3.3 Analyzing data
3.3.1 Missing values
The experiment is, besides in English, also executed in Dutch since many of the participants can only read and understand the user message properly in Dutch. Despite the difference in language the two versions did not differ in content. While analyzing the data, the results of the two different versions were converted into the same variables. With the multiple choice questions (for example, with measuring the purchase intention) the respondents had to answer before they could proceed the survey. Therefore, no missing values were noted on these questions for the people who finished the experiment. Some participants did not finish the survey and stopped already in an early stadium. A discrete missing value is entered in the ‘value’ field of the variables that did not contain data. The two independent variables hold the values ‘0’ and ‘1’, the value of ‘999’ is used as discrete missing value to exclude them in analysis. To the variables that involved the 7 likert scale values, the value ‘0’ is filled in as discrete missing value. In total, the sample incorporated 15 participants with incomplete data. However, with the open questions some participants either filled in a space or a ‘-’, meaning
31
they did not know or perhaps did not want to answer the question. These characters were also reported as missing values.
3.3.2 Computing variables
The experiment data is converted from the online platform to SPSS. Therefore, the variables were not named and coded clearly and/or properly. For example, the variable ‘A_0’ was renamed in to the variable highF_highA which implicated the manipulation of high
functional/high arousal. In this way the variables are easily recognized while analyzing them. The data of the four distinct manipulations are used to compute two new variables by taking the mean of the four manipulations and integrate them into the variables ‘FunctionalMessage’ and ‘ArousalMessage’. The functional message represents the first independent variable where a distinction is made between high functional and low functional content. High
functional content is reported with the value of ‘0’ and low functional content is reported with the value of ‘1’. The arousal message represents the second independent variable whereby high arousal content is announced with the value of ‘0’ and low arousal content is announced with the value of ‘1’. Then, the negatively asked questions, for the first items of the perceived credibility and level of abstractness (PerceivedCredbility_1 and LevelAbstractness_1), were re-coded into new variables with reverse scoring and are called ‘PC_1’ and ‘LA_1’. For example, a score of 1 was reversely coded into a 7 and a score of 7 reversely coded into a 1. A new variable is computed called ‘PC_TOT’ for which the mean value of the two items ‘PC_1’ and ‘PerceivedCredibility_2’ is calculated. The same process is executed to compute the variable ‘LA_TOT’ for which the mean value of the two items ‘LA_1’ and
‘LevelAbstractness_2’ is calculated. The user message is perceived as high in credibility when the score is relatively high. For the level of abstractness, the variables measures whether the user message is helpful in understanding the product and its performances. Therefore, a high score indicates a concrete message while a relatively lower score indicates an abstract message.
32
4
R
ESULTS
4.1 Basic data analysis 4.1.1 Sample
A total of 198 experiments were conducted, including 170 Dutch speaking and 28 English speaking participants. This matches a distribution within the sample of 85,9 percent Dutch and 14.1 percent English. The sample consists of a higher proportion of female with 54,5 percent (108) than male with 36.4 percent (72). 9.1 percent (18) of the participants did not fill in their sex and were reported as missing. The age group within the sample varies between sixteen and 85. Although the age within the sample is widely spread, the largest percentage of consumers are of the age 21-30 with 52 percent of the sample. The average age, also
illustrated in table 2, is 30.24. This makes the sample skewed towards female and younger aged consumers. A large proportion of 88.3 percent of the sample announced that they were living in the Netherlands. No other countries besides European countries were included in the participant’s country of residence. The level of education was quite equally spread as 20.2 percent finished MBO degree, 27.3 percent finished HBO and 31.3 percent finished university. 192 participants were randomly distributed to the four different user message groups, six were missing. 22.2 percent (44) were high functional/high arousal, 22.7 percent (45) were low functional/high arousal, 28.8 percent (57) were high functional/low arousal and 23.2 percent (46) were low functional/low arousal.
4.1.2 Manipulation checks
The manipulation checks were executed to test whether the different levels of functional and arousal content for the user message is recognized by the participants. The results of the one-way Anova measure the mean values for the functional content and arousal content per user message group. For the functional content check, the results show different means for high
33
functional/high arousal (M=6.64), low functional/high arousal (M=6.29), high functional/low arousal (M=6.70), low functional/low arousal (M=6.61). However, the different mean values are not significant F(3, 177) = 0.526, p > 0.05. The arousal content check showed different scores for high functional/high arousal (M=6.05), low functional/high arousal (M=7.11), high functional/low arousal (M=6.20), low functional/low arousal (M=6.30). Though, the mean differences are not significantly supported F(3, 177) = 2.484, p > 0.062. The arousal content check for the first manipulation is notably low (6.05) where for the second manipulation the same arousal content is rewarded with a substantial higher mean value (7.11). Most of the participants, who were linked to the first manipulation and evaluated very low arousal scores, were very negative about the message in general or valued the functional aspects in particular. Some of the participants announced they only liked the functional aspects and/or would not buy this phone anyway. Based on this results, there is no support that in this sample the present experiment adequately manipulated the user message content of eWOM. The failed manipulation check does not intend to verify that the manipulated independent variables cause variation in the dependent variable. But, eventually, the manipulation in this study failed to influence the dependent variable. The results of the manipulation check is illustrated in table. The manipulations, illustrated in table 1, are numbered for which 1 is high functional/high arousal, 2 is low functional/high arousal, 3 is high functional/low arousal and 4 is low functional/low arousal.
34 4.1.3 Validation of key constructs
A measurement of the Cronbach’s Alpha is conducted to illustrate the reliability of the constructs measuring the level of abstractness, perceived credibility and purchase intention. All the three variables consist of two items which are measured on a sample of 183, excluding 15 participants. Firstly, the Cronbach’s Alpha for purchase intention is 0.800. This
Cronbach’s Alpha reflects a good internal consistency between the two items of purchase intention. Secondly, for perceived credibility the Cronbach’s Alpha is 0.673. Finally, the Cronbach’s Alpha for the level of abstractness is 0.600. The Cronbach’s Alpha’s for perceived credibility and the level of abstractness are below the acceptable cut-off value of 0.70. The values of the Alpha between 0.6 and 0.7 represent a low internal consistency of the items. The items of both variables have relatively low inter correlation which means that the two items do not hang strongly together in what they measure. This low Cronbach’s Alpha is also related to the number of items that measure the scale, which is two per variable. When more items are used in measurement, the Cronbach’s Alpha may go up. However, the distinctness of the two items does not mean that the two items do not measure the construct properly. Still, the poor reliability illustrated by the two low Cronbach’s Alpha, could mean that the variable constructs are insufficiently reflecting the perceived credibility and level of abstractness. In turn, this may influence the mediating effect, reliability measurement of the variables in this framework and eventually the results of this study. The Cronbach’s Alpha’s are showed in table 2 which demonstrates the correlation analysis of the variables age, sex, user message, employment status, school degree, purchase intention, perceived credibility and level of abstractness.
4.1.4 Correlation analysis
The relationship between the variables is examined through a correlation analysis. The results of the correlation analysis, illustrated in table 2, clarify the intensity and meaning between two or more variables and exhibit the means and standard deviations. The results suggest
35
there is a significant positive relationship between age and sex of 0.25. Furthermore, age and level of abstractness are significantly correlated with -0.206. The variables sex has significant correlations with the employment status (0.273), educational level (0.306), purchase intention (0.218) and perceived credibility (0.237). The variable sex has a mean of 1.54 and represent that the majority of the sample are female. The employment status has a positive and
significant relation with school degree, embodied by the correlation of 0.473. Moreover, significant evidence prove purchase intention has a positive relation with perceived credibility and level of abstractness. Finally, the two mediators in this study, perceived credibility and level of abstractness have a significant positive correlation of 0.688. The correlations clarify that the variables move in the same direction. However, this specific number does not explain the causation of one variable in the other variable.
4.1.5 Conceptual model analysis.
In order to test the hypotheses in this study, the different relationships between the variables are examined. First, to examine how the difference in functional user message content and arousal user message content affects the intention to purchase, a 2 (functional message) x 2 (arousal message) ANOVA is conducted. As figure 3 (A) demonstrates, the main effect is tested to clarify whether significant differences exist between the high and low functional user message and between the high and low arousal user message (path c’). Likewise, an ANOVA is processed to clarify the association of the independent variables with the level of
36
other variables are analysed. If both independent variables show significant effects on the dependent variable, two separate process analysis are executed which test the framework (A and B) per independent variable. If either one of the independent variables show significant variances in the dependent variable, then this independent variable is used in a process
analysis to test the framework in figure 3 (A and B). The process analysis tests the direct (path c’), indirect (path a1b1, path a2b2, path a1a3b2) and total effect (path c) between one of the independent variables and the other variables. But, if the ANOVA results do not show any significant differences between functional user message and arousal user message, the paths of the independent variables are not further examined and the sequential mediation model and process analysis are redundant. The focus will be on the remaining part of the framework and a regression analysis is performed to test the effect of the two mediating variables on the dependent variable.
37 4.2 Hypotheses test
4.2.1 The effect of user message content
The sample in this study is divided in four different user message content groups. The four groups have in common that they include positive evaluations towards a fictitious brand but they are distinct in the level of functional and arousal content. In the present study the focus is on the influence of the independent variable on the outcome variable through two mediating variables. Therefore, an ANOVA is carried out to test the main effect between the
independent variables and purchase intention, level of abstractness and perceived credibility. Since there are two independent variables, a factorial ANOVA is applied to a sample of 183. The results suggest that there is no significant main effect of functional user message content on the purchase intention, F(1, 183) = .072, p > .05, n² = .000. Moreover, there is no
significant main effect of arousal user message content on the purchase intention, F(1, 183) = 1.115, p > .05, n² = .006. And, no significant interaction effect between functional user message content and arousal user message content on the purchase intention is tested, F(1, 183), = .011, p > .05, n² = .000.
An ANOVA is processed to examine the effect of the independent variables on the level of abstractness and perceived credibility. The results state there is no significant
variance in the level of abstractness for the functional user message content, F(1, 183), = .084, p > .05, n² = .000. Furthermore, the effect of arousal user message content on level of
abstractness is non-significant, F(1, 183), = .029, p > .05, n² = .000. There is a non-significant interaction effect between functional user message content and arousal user message content on the intention to purchase, F(1, 183), = .396, p > .05, n² = .002. Finally, the results of the independent variables on the perceived credibility, illustrated in table 3, shows there is no significant effect of the functional user message content on the perceived credibility, F (1, 183), = .397, p > .05, n² = .002. There is no significant effect of the arousal user message content on the perceived credibility, F (1, 183), = 2.013, p > .05, n² = .011. Besides, there is
38
no significant interaction effect between functional user message content and arousal user message content on the perceived credibility, F (1, 183), = .092, p > .05, n² = .001
These results taken together indicate that there is no significant prove that there are differences in variance of functional and arousal user message content on perceived
credibility, level of abstractness and purchase intention. This finding suggests that the paths of the independent variables to the other variables are not further examined. At this point the current research is focused on the association between level of abstractness, perceived credibility and purchase intention.
39 4.2.2 The overall effect of the model
From this point, the focus of the study will be on the association between the level of
abstractness, perceived credibility and purchase intention. The R-square, illustrated in table 4, imply that this model with two independent variables (level of abstractness and perceived credibility) explain 36,3% in the outcome variable (purchase intention). The adjusted R-square may be a better indicator of the explanatory power of the regression model since this number has been adjusted for the number of predictors in the model. Where the R-square increases as the amount of predictors increase, the adjusted R-square only increases when the predictors improve the model. The number of 0.356 shows that, though slightly lower than the R-square, the predictors explain 35.6% in the outcome variable. The F-ratio from the
ANOVA table suggests that the overall regression of the independent variables predicts the independent variable with statistical significance, F (2, 180) = 52,235, p < 0.000.
4.2.3 The effect of the level of abstractness
A multiple regression analysis is performed to examine the influence and association between the level of abstractness and intention to purchase. The level of abstractness is measured on a 7-point likert scale based on two items. A high score suggests that the online user message content is evaluated as more concrete while a low score represents a more abstract level of user message content. The assumption is that a higher score, and thus a more concrete
evaluation of the user message content, is positively associated with the intention to purchase. The unstandardized coefficient (B) in table 4 indicate the variation of the independent variable when all other variables are held constant. The unstandardized coefficient for the level of abstractness is 0,338. This number implies that when the evaluation of the level of abstractness increases with one, and thus being more concrete, the purchase intention
increases with 0,338. The standardized coefficient (Beta) is evaluated to analyse and compare the individual impact on the dependent variable. The Beta of 0.336 state that an increase of one standard deviation in the level of abstractness would result in an increase of 0.336
40
standard deviation in the intention to purchase. Furthermore, the coefficients are significantly proved, p < 0.005. These results indicate that there is a positive relationship between the two variables, proving that if an online user message is evaluated with high values, and thus being more concrete, this will have a positive impact on the intention to purchase.
4.2.4 The effect of perceived credibility
The perceived credibility is measured on a point likert scale based on two items. The 7-point likert score, for this specific variable, suggests that a higher evaluation is in line with a higher perception of the credibility. The effect of perceived credibility on purchase intention is illustrated by an unstandardized coefficient of 0.293 and is significantly proved, p < 0.005. This result means that if the perceived credibility increases with one, the purchase intention increases with 0.293. The Beta of 0.319 point out that an increase of one standard deviation in the perceived credibility reacts to an increase of 0.319 standard deviation in the purchase intention. Thus, the results clarify that there is a positive and significant relation between the two variables.
4.2.5 Correlation between the independent variables
The multicollinearity in a regression refers to the correlation between multiple independent variables. Multicollinearity occurs when the independent variables are to an extent more correlated to each other than to the independent variable, making them redundant. The Pearson correlation in the regression analysis demonstrate that the level of abstractness and perceived credibility have a correlation of 0.688. Besides, the correlation of the level of
