How customers are influenced by
product reviews: research on the persuasive
effects of source credibility in an online review
context
Exploring the relative effects of source credibility, cognitive responses, online
trust and eWOM adoption in an online review context
by
Thije Boksebeld
How customers are influenced by
product reviews: research on the persuasive
effects of source credibility in an online review
context
Exploring the relative effects of source credibility, cognitive responses, online
trust and eWOM adoption in an online review context
by
Thije Boksebeld
University of Groningen
Faculty of Economics and Business
Master’s thesis MSc Marketing Management (Final version)
Completion date: June 15, 2018
Lemelerweg 4 8105 AD Luttenberg
+ 316 154 239 40 thijeboksebeld@gmail.com
Student number: 3159299
ABSTRACT
Since not much is known yet about the differential effects of source credibility in the online review context, the purpose of this study was to explore the underlying psycho-logical mechanisms of source credibility (i.e., the indirect effects of the two dimen-sions), and to which extent source credibility is applicable in the online review context. I therefore examined the relative effects of the two dimensions of source credibility, cognitive responses, affective- and cognitive trust and eWOM adoption in influencing the customer’s attitude towards a product. Using survey data from 156 participants, the results showed that perceived source credibility positively affects attitude, where a trustworthy communicator is more persuasive than an expert communicator. Contrary to the source credibility literature, unfavorable cognitive responses negatively moder-ate the relationship between source credibility and attitude. Seemingly, source credi-bility does not show the dampening effect on counterarguing in an online review con-text than it does in advertising concon-text. Furthermore, a theoretical contribution to the literature of eWOM communication and the role of source credibility in it, is the ap-plicability of source credibility in an online review context. Source credibility plays an important role in reducing uncertainty of eWOM message credibility. An increase in source credibility namely resulted in an increase in cognitive trust, resulting in turn in the use and higher adoption of the information from the online review, which even-tually results in more positive attitudes towards the product.
Keywords: Source credibility; Trustworthiness; Expertise; Consumer behavior; eWOM;
PREFACE
This master’s thesis is my final project of the study Marketing Management at the Uni-versity of Groningen. Writing this master’s thesis — including the high amount of independ-ency — was definitely one of the highlights of this study. Debit on this was the underlying topic of consumer behavior. After researching social behavior during my Pre-MSc (i.e., investigating materialism), I found out that I have a strong preference to investigate social consumer psy-chology. So it was fun to read more about certain topics concerning consumer behavior.
Further, for helping me in this period, I am thankful to my supervisor dr. Martijn Keizer for providing me very useful feedback and for challenging me to get the most out of this study. Secondly, I will also want to thank my fellow group members for the feedback and for providing a pleasant working environment during group meetings.
Next to that, the two years of studying in this lively town not only improved my mar-keting knowledge and my analytical skills, but it also helped me in growing as a person. I am really happy that I chose to continue my student life by applying to the Pre-MSc Marketing course and subsequently following the MSc Marketing. Lastly, referring back to this thesis, I hope that you enjoy reading this piece, and hopefully it gets you interested in this topic.
MANAGEMENT SUMMARY
Nowadays before purchasing new products or services, individuals increasingly consume online product reviews that are attached to the product description, in order to gather and ex-change product information. Online product reviews are even considered one of the most influ-ential types of electronic word-of-mouth in shaping consumer attitudes and facilitating purchase decisions. However, the anonymous character and the reduced social presence in the online world makes it difficult for consumers to assess whether trust is warranted or not. Trust in an online review context therefore seems a very crucial construct. People namely tend to look for online safety cues in order to determine the credibility of the message. A reviewer’s perceived credibility could potentially be such a safety cue. However, despite the fact that online reviews are increasingly important for consumers (and marketing managers), research on source credi-bility in the online review context is underexposed.
Research on source credibility nevertheless is researched thoroughly, be it, with the ma-jority of it in an advertising context. In this context, the use of source credibility turned out to be an effective persuasion tactic for marketing managers. Highly credible sources namely cause increased acceptance of a message because they are associated with favorable outcomes. Fur-ther, research shows that source credibility is especially effective when people hold unfavorable thoughts towards a communication. Source credibility in this light have a dampened effect on counterarguing, induced by those unfavorable thoughts towards an advertising (i.e., unfavora-ble cognitive responses). Furthermore, source credibility appeared to be effective in both, low-effort and systematic processing.
Where much is already known about the persuasive effects of source credibility, it is still quite uncertain how the two dimensions of source credibility differ from each other, and even if they show differential effects, which dimension is better applicable in which situation. The main objective of this research therefore is to contribute to the source credibility literature by exploring to which extent the antecedents differ in effectiveness regarding attitude change. Furthermore, as stated before, it is still uncertain to which extent source credibility is applicable in an online review context. This paper therefore tries to provide empirical support for how both dimensions function, and to which extent source credibility is applicable in the online review context
ppurchase information search for a specific product. Participants were shown an online re-view from either a trustworthy, expert, combined, or no specific credible source. After being presented to the product page with the corresponding online review, participants were asked to answer questions that measured the variables of interest (i.e., attitude, eWOM adoption, affec-tive- and cognitive trust and cognitive response favorability).
Various statistical tests (i.e., t-tests, One-way ANOVA and mediation models by Hayes) were conducted to analyze the results. The major results of this experiment showed that the dimensions of source credibility (i.e., trustworthiness and expertise) both accounted for a sig-nificant proportion of variance in attitude change. Both dimensions showed a sigsig-nificant higher positive attitude than respondents exposed to a source who was perceived low in credibility. Nevertheless, perceived trustworthiness showed higher attitude change than respondents with a perceived expert did. Thus, in an online review context, perceived trustworthiness is likely to be more effective than perceived expertise.
Furthermore, contrary to the source credibility literature regarding advertising, unfavor-able cognitive responses in an online review context are negatively moderating the effects of source credibility on attitude change. This means that the effects of source credibility and its antecedents in an online review context are less strong when cognitive responses are unfavora-ble. Apparently, the resistance to the persuasive effect of advertising — that comes from unfa-vorable cognitive responses — increased counterarguing, that in turn resulted in a tendency to disbelieve online reviews. Lastly, and consistent with the prediction of the applicability of source credibility in an online review context, results show that individuals can be positively persuaded by source credibility in an online review context. The mechanism behind this com-prises an increase in cognitive trust, resulting in higher adoption and use of eWOM messages, which subsequently positively affects attitudes. It can be stated that source credibility in this light functions as a so called online safety cue for reducing uncertainty.
TABLE OF CONTENTS
1. Introduction 1
2. Literature review 3
2.1 Source Credibility 3
2.1.1 Cognitive response theory 4
2.1.2 Dual process theory 5
2.1.3 Persuasion under low effort processing conditions 6
2.1.4 Persuasion under high effort processing conditions 6
2.1.5 Multiple role assumption 6
2.1.6 Source credibility affecting message-relevant thinking 7
2.2. The antecedents of source credibility 8
2.2.1 Differential effects expertise vs. trustworthiness 8
2.2.2 Mechanism expertise dimension 9
2.2.3 Mechanism trustworthiness dimension 10
2.3 Source Credibility in an online review context 11
2.3.1 Persuasion in the online sphere 12
2.3.2 Source credibility in an eWOM context 12 2.4 The moderating effects of cognitive response favorability 14
2.5 Conceptual Model 16
3. Method 17
3.1 Participants and design 17
3.2 Procedure and stimuli 17
3.3 Analysis plan 22
4. Results 22
4.1 Sample demographics 22
4.2 Results of hypotheses testing 23
4.3 Post-hoc analyses for indirect effects 29
5. Discussion 32
5.1 Summary of findings 32
5.2 Theoretical contributions 33
5.3 Practical implications 35
5.4 Limitations and further research directions 36
References 38
Appendices 44
Appendix A 44
While it may be true that the best advertising is word-of-mouth,
1
1. INTRODUCTION
Marketing managers frequently hire people of ostensibly high integrity and credibility to sponsor, enhance and endorse their products and services. By doing so, managers can use different types of credible individuals in their commercials. Consider a typical advertising sce-nario in which a consumer is exposed to an advertising commercial promoting nutrition prod-ucts. Very often, these kinds of advertisements contain athletes promoting the use of that par-ticular brand, in order to strengthen associations regarding the healthiness of their products by using sportsmen. An example of this is when Friesland Campina used Dafne Schippers, a Dutch athlete, to promote their products (Friesland Campina, 2016).
Another example with reference to this phenomenon is the use of actors wearing white dental lab coats as a cue for expertise regarding toothbrush commercials. Not to mention — maybe one of the most recognizable examples — George Clooney in the well-known Nespresso coffee commercials (Nestlé Nespresso, 2015) to increase the likeability of their products. All these examples are based on the belief that highly credible sources are more persuasive than those of lower credibility. Highly credible sources cause increased acceptance of a message because they are associated with favorable outcomes (Heesacker, Petty & Cacioppo, 1983). Source credibility, which refers to a message communicator’s perceived expertise and trust-worthiness (Kelman & Hovland, 1953), in this light results in significant attitude change. In fact, when the emphasis is on expertise or trustworthiness, highly credible sources have been found to be more persuasive than those of lower credibility (Pornpitakpan, 2004; Weitzl, Wolfsteiner, Einwiller & Wagner, 2016).
2
Coney, 1982). Highly credible sources also appeared to be effective when people engage in low-effort processing. In these situations, people use heuristic cues (e.g., experts can be trusted) to make their judgments (Petty, Cacioppo & Goldman, 1981; Weitzl et al., 2016). Those heu-ristics help individuals forming attitudes towards products and services.
Research on source credibility such as described above, mainly focuses on the overall effects of source credibility, despite the fact that the two dimensions of source credibility might have differential weights (McGinnies & Ward, 1980; Pornpitakpan, 2004; Xu 2014; Shan, 2016). Very little research is performed on the separate underlying dimensions of source cred-ibility (i.e., trustworthiness and expertise) on attitude change.
Both dimensions seem to function via different types of trust. Trustworthiness appears to function as an affective trust cue (i.e., the extent of intrinsic motivation perception; Johnson & Grayson, 2005), whereas expertise is likely to function as a cognitive trust cue (i.e., the con-fidence in one’s competence, responsibility and knowledge; Xu, 2014). The main objective of this research therefore is to contribute to the source credibility literature by exploring to which extent the antecedents differ in effectiveness regarding attitude change.
Furthermore, where the vast majority of research on source credibility covers source credibility in an advertising context, little is known about the effects of source credibility in the digital environment. More specifically, not much is known about the relevance of source cred-ibility in an online review context, a relatively recent and growing phenomenon in the digital world. This new form of word-of-mouth communication consists of consumer opinions and recommendations on products or services. This online consumer sharing activity is widely known as electronic word-of-mouth (i.e., eWOM; Hennig-Thurau & Walsh, 2004). Even though the importance of reviews as a criterion in judging the reliability of online information, the communicator’s source has not been fully investigated in studies regarding online product reviews (Dou, Walden & Lee, 2012). The importance of source credibility in judging and adopt-ing online reviews is still not quite clear.
3
Thadani (2012) found that source credibility has a positive effect on eWOM adoption, and that eWOM adoption in turn plays an important role in shaping one’s attitude and purchase inten-tion. The effects of source credibility on consumers’ attitude in this light seem to be mediated by eWOM adoption in this online review context.
Summarizing, although many research papers addressing the effects of source ity in an advertising environment, not much is known yet about the effects of source credibil-ity’s dimensions in the online review context. The objective of this article therefore is to explore the underlying psychological mechanisms of source credibility (i.e., the indirect effects of the two dimensions) and to present and provide empirical support to which extent source credibility is applicable in the online review context. Lastly, this paper examines the prediction of cogni-tive response theory, that messages with high source credibility lead to more persuasion when thoughts in response to a communication are unfavorable (Bock & Saine, 1975). In what fol-lows, I will provide an overview of the concepts: source credibility and its antecedents, source credibility in an online review context, and the role of cognitive response favorability in more detail. Subsequently, the methods, stimuli and procedures used in this study are presented. I then review the results, implications, limitations, and recommendations for further research on source credibility.
2. LITERATURE REVIEW
2.1 Source Credibility
The concept source credibility has been widely identified to consists of two antecedents, namely expertise and trustworthiness. Expertise refers to the extent to which a communicator is perceived to be capable of making correct statements, and trustworthiness refers to the attrib-ution of the source’s motivation to share information (Hovland, Janis & Kelley, 1953). A source with high credibility, compared to one low in credibility, appears to result in a positive attitude toward the endorser and the advertisement. Regarding advertising, commercials associated with highly credible sources were rated as more believable and truthful than were those associated with the low credibility sources (Kelman & Hovland, 1953; Braunsberger, 1996; Pornpitakpan, 2004).
4
2.1.1 Cognitive response theory
Many researchers have considered Cognitive response theory — developed by Green-wald (1968) — to predict the persuasive effects of source credibility (Dean, Austin, & Watts, 1971; Bock and Saine 1975; Stemthal, Dholakia, & Leavitt, 1978; Harmon and Coney, 1982). This cognitive response view of persuasion had showed superiority over other approaches (e.g., associative and message learning approaches) in the explanation and prediction of the persua-sive effects of source credibility (Insko, Lind, & LaTour, 1976; Love & Greenwald, 1978; Cacioppo & Petty, 1979; Petty, Cacioppo, & Heesacker, 1981). According to this theory, social influence depends for a great part on the favorability of thoughts, or object-attribute associa-tions available in memory at the time of judgment (Harmon & Coney, 1982). Two types of thoughts are important when judging information: message thoughts and own thoughts (i.e., cognitive responses; Greenwald, 1968). Message thoughts represent thoughts regarding the pre-sented information, whereas cognitive responses are associations an individual has stored in memory, that are relevant to the communication (e.g., the communicator or product), but not directly to the message arguments.
According to the cognitive response theory, individuals are active thinkers, who engage in silent personal discussions with the presented advertisements (Greenwald, 1968). By doing so, individuals elaborate on the message arguments and consider additional information that is not contained by the communication. People in this stage generate thoughts for or against the arguments presented by the communicator. It is these cognitive responses to the advertisement (rather than the message arguments themselves) which determine the impact of the persuasive communication on attitudes (Fennis & Stroebe, 2016). Research on information processing has considered these cognitive responses to explain the production of counterarguments, which in turn increases resistance (Festinger & Maccoby, 1964). More specifically, an individual can have relatively positive thoughts towards the communication beforehand, but eventually hold negative feelings towards the communication because of the counterarguments induced by un-favorable cognitive responses.
5
resistance to influence in cases when cognitive responses are unfavorable. Since the positive associations of the credibility of the source become tied to the advertisement and the product or service, the use of highly credible sources are believed to increase persuasion when cognitive responses are unfavorable (Heesacker, et al., 1983). However, — providing that thoughts to the communication are favorable — a highly credible source also inhibits the activation of support arguments and thus is not as persuasive relative to a less credible source who facilitates support argument activation (Harmon & Coney, 1982; Tomala, Briñol & Petty, 2006).
Thus, the findings as stated before show that source credibility seems especially persua-sive if cognitive responses are unfavorable. Since people in general hold unfavorable thoughts towards advertisements (Obermiller & Sprangenberg,1998; Trusov et al., 2009), highly credible sources often are a very effective and widely used as persuasion tactic for marketeers.
2.1.2 Dual process theory
Another explanation for the mechanisms of source credibility can be found in the dual process theory of persuasion, which is an extension to the cognitive response theory as de-scribed above. The elaboration likelihood model (i.e., ELM) from the study of Petty et al., (1981) is such a dual process theory of persuasion. This model assumes that persuasive com-munication can induce attitude change trough two types of processing. The way people engage in either type of processing in turn, predicts the likelihood of accepting message arguments (Fennis & Stroebe, 2016). The two modes of information processing according to the ELM are systematic and non-systematic processing (i.e., heuristic processing). People engaging in sys-tematic information processing use notable mental effort to scrutinize the content and argumen-tation of a message. Contrarily, people engaging in heuristic processing, rely on heuristics and use simple decision rules (e.g., experts can be trusted) to make judgments about the provided information (Petty, Wheeler & Tormala, 2003).
6
can namely be seen as a continuum, suggesting only the likelihood of the extent of systematic
processing (Petty et al., 1981).
2.1.3 Persuasion under low effort processing conditions
Prior research on source credibility with respect to the ELM, uncovers a number of mechanisms for source credibility effects. Depending on the consumers’ extent of processing, source credibility can have different functions in inducing persuasion. Under low effort pro-cessing, source credibility appears to evoke — amongst other things — the “experts can be trusted’’ heuristic, indicating that people perceive expertise as a cue for making judgments (Cialdini, 2001). This low processing mode can be seen as the default option and the corner-stone of persuasion via heuristics (Cialdini, 2009; Fennis & Stroebe, 2016). According to the least-effort principle that underlies this behavior, people behave in this mindless manner when there is no sufficient reason to invest effort in mindful behavior. People by default namely tend to save their energy resources and fall back on simple decision heuristics as described above.
2.1.4 Persuasion under high effort processing conditions
As stated before, people can also engage in high effort processing when there is suffi-cient reason to invest effort in a mindful behavior (e.g., high motivation or ability). Under such high effort processing, source credibility can also be persuasive. Source credibility namely in-fluences persuasion by biasing the kind of thoughts the information provides (Chaiken & Ma-heswaran, 1994) and affects people’s confidence in their thoughts by affecting the perceived validity of the information in a message (Fragale & Heath, 2004). When people then find the information to be valid, people namely tend to be more confident in their thoughts (Tormala, Briñol & Petty, 2006; Briñol, Petty & Tormala, 2004) and become more confident in making judgements, even though this confidence is sometimes based on false inferences. High source credibility can namely persuade people to attribute expertise to a source, regardless of the qual-ity of the information the communicator offers (Willemsen, 2012).
2.1.5 Multiple role assumption
7
characteristic of elaboration via the heuristic route, also processing via the systematic route is mediated by the “if–then’’ reasoning leading from evidence to a conclusion according to Kruglanski and Thompson. Kruglanski and Thompson (1999) namely suggest that the two routes are functionally identical in forming judgements (Fennis & Stroebe, 2016). According to the multiple-role assumption, persuasion can also be characterized as a singular process of drawing conclusions from available evidence (Kruglanski & Stroebe, 2005). It does not matter whether this evidence is contained by an argument or comes from a heuristic cue.
2.1.6 Source credibility affecting message-relevant thinking
In situations where processing is not forced to be high or low on beforehand (because the extent of processing can namely be seen as a continuum, suggesting only the likelihood of the extent of systematic processing; Petty et al., 1981), source credibility can influence the amount of processing that occurs (Heesacker, et al., 1983). When the needs (e.g., collecting product information) of the individual namely can be satisfied by the communicator’s source, source variables (e.g., source credibility), because of the increased involvement, may motivate indi-viduals to process the information more systematically. In this way, indiindi-viduals can become motivated to analyze and scrutinize the content of the persuasive message, based on the match between the source variables (DeBono & Harnish, 1988).
In support for this reasoning, Heesacker et al. (1983) showed that respondents with a low tendency to extract the meaning from persuasive messages, show significant differential attitude change (i.e., agreed with strong arguments, disagreed with weak arguments) to strong and weak arguments only when they were presented by a highly credible source. When pre-sented a low credible source, participants with this personality trait showed no differential agreement with the strong and weak messages, meaning that a highly credible source affected message-relevant thinking. Those individuals namely responded as increasing source credibil-ity enhanced their motivation to think critically about the content of the message that was pre-sented. With respondents with an inhibit tendency to extract the meaning from persuasive mes-sages, strong arguments were persuasive for both high, and low credible sources and weak ar-guments were not effective for any source (Heesacker et al., 1983).
8 more systematic information processing.
However, even when individuals are motivated to process information systematically, there are situations in which source credibility functions as a heuristic cue. Source credibility can namely have a persuasive effect under systematic processing when the information is quite indistinctive (Chaiken and Maheswaran (1994). People engaging in systematic information pro-cessing in this light are unable to come to an unambiguous conclusion. People’s ability in mak-ing valid judgements in this way are reduced, and they will fall back on the persuasive impact of the heuristic cues (Weitzl et al., 2016).
Given the fact that source credibility serves as a cue in both types of processing, it seems highly plausible that high source credibility will result in positive attitude change.
H1. High source credibility, regardless of the extent of processing one engages in, will
result in a significant positive attitude change towards the product or service.
2.2. The antecedents of source credibility
With the underlying mechanisms and boundaries of source credibility now discussed, the question of how the two antecedents of source credibility in isolation relate in changing the consumer’s attitude remains. To get a clearer picture, this section therefore tries to examine the antecedents of source credibility. Are there any differential effects? How distinctive are those two dimensions? The following paragraphs will dive into detail regarding those research ques-tions.
2.2.1 Differential effects expertise vs. trustworthiness
Research on the different effects of source credibility by Xu (2014) points out that there are significant differences between the two dimensions, however, the two antecedents of source credibility nevertheless show significant correlation with each other. This is consistent with research of Cialdini (2001) that suggests that the principles he proposes, that are closely related (e.g., authority, social proof) can be discussed separately, but could also be applied in combi-nation due its overlap, and ability to compound their impact. Although the two dimensions are highly correlated, they nevertheless are believed to be empirically distinguishable (Xu, 2014; Johnson & Grayson, 2005). In some occasions, the two antecedents function quite similarly, in other occasions they are very distinguishable in their effectiveness as a persuasion tactic.
9
revealed that a perceived trustworthy communicator was more persuasive than an untrustworthy one, whether or not it was expert, suggesting that in some cases, the use of a trustworthy source can be more effective than the use of an expert. Elaborating on this, results of the Shan (2016) study suggest that perceived similarity has a positive effect on trustworthiness and a negative influence on expertise. Consumers namely tend to trust people more when they perceive them to be more similar. People who are like themselves, do not necessarily have expert knowledge of the product or service.
In turn, expertise — the other antecedent of source credibility — does not only provide the customer with a cue for presumed experts qualities under low processing motivation such as stated earlier, it also implies expertness by a professional title, which also can influence attitude under high processing motivation (Fennis & Stroebe, 2016) by functioning as a strong message argument. An explanation for this effect lies in the multiple role assumption. When we refer back to the multiple role assumption, the cue for expertise can function as a heuristic (e.g., experts can be trusted) for people under low processing motivation, but can also work for people engaging in high information processing. Highly motivated individuals namely also agree with the expert source, regardless of the quality of the message (DeBono & Harnish, 1988).
Given the brief overview of the differential effects of source credibility such as stated in this paragraph, suggesting differential effects of the two dimensions, I also suggest that both dimensions have distinguishable effects. Albeit still quite uncertain to which extent, given the high correlation between the two antecedents of source credibility.
H2. The dimensions of the communicator’s source credibility (i.e.,
trustworthi-ness and expertise) both have different, be it positive, effects in shaping con-sumers’ attitude towards a product or service.
With the argumentation of the distinction being made, it is still not clear how the two dimensions of source credibility relate in changing the consumer’s attitude. More specifically, it is still not very clear how and to what extent the two dimensions differ. The following para-graphs tries to explain the apparent differential effects by exploring each dimension in detail.
2.2.2 Mechanism expertise dimension
10
perception of a expertise contains the identification of competencies associated with the com-municator (Johnson & Grayson, 2005). The status of expert serves as a cue signaling that the source provides validity. This cue can persuade people to attribute expertise to a source, regard-less of the quality of the information the communicator offers (Willemsen, 2012). This finding is consistent with Cialdini’s (2001) Principle of Authority, that states that individuals rely on the heuristic that well-selected experts offer a valuable and efficient shortcut to good decisions. Using well-selected experts as a communicator turned out to be a highly effective persuasion strategy in advertising (Cialdini, 2001).
Based on the characteristics of the expertise-dimension of source credibility, expertise seems to function as a cue for cognitive trust. Cognitive trust has been defined in the social psychology literature as a customer’s confidence or willingness to rely on a message provider’s competence and reliability that arises from an accumulated knowledge that allows one to make predictions, with some level of confidence (Johnson & Grayson, 2005; Lewis & Weigert, 1985). When the reputation effects are strong, interaction with the stimuli may be an oppor-tunity to confirm or disconfirm prior perceptions (Johnson & Grayson, 2005). Cognitive trust originates from the confidence in the competence and responsibility of the communicator, and is mostly driven by its perceived knowledge (Xu, 2014). This is in line with the definition of the expertise concept. Cognitive trust, also seen as the ability-dimension of trust, affects the extent to which consumers are able to be persuaded (McKnight, Choudhury & Kacmar 2002; Fennis & Stroebe, 2016). Concluding, since cognitive trust is the willingness to rely on a source based on specific reliability instances, it seems highly plausible that this type of trust functions as an indirect effect of the dimension expertise.
H3a. Cognitive trust mediates the positive effects of the source’s perceived
ex-pertise in shaping consumers’ attitude towards a product or service. 2.2.3 Mechanism trustworthiness dimension
11
Peer-power in this case functions as an important indicator of trustworthiness. This cor-responds with the Principle of Social Proof in the study from Cialdini (2001). This principle emphasizes on the role of peer-power and similarity. Individuals heavily rely on the people around them for cues on how to think, feel and act. Study shows that persuasion can be ex-tremely effective when it comes from peers (Cialdini, 2001). We namely see something as cor-rect to the extent that we see others believing it or doing it (Fennis & Stroebe, 2016).
The cue of trustworthiness is highly driven by emotion. This seamlessly corresponds with the concept affective trust. Affective trust namely is defined as the confidence one places in a communicator based on feelings generated by the level of care and concern the communi-cator demonstrates. The essence of affective trust is the reliance on a communicommuni-cator based on emotions. Affective trust is closely related to the perception that a communicator’s actions are intrinsically motivated. This is in line with the definition of trustworthiness that states that in-dividuals who judge a source on trustworthiness, base their judgment on inferences regarding the reviewer’s motivation to provide information about a product or service (McCracken, 1989), and also to which extent the source’s actions are intrinsically motivated (Johnson & Grayson, 2005). Concluding, the positive effect of trustworthiness on attitude in this light is likely to be mediated by affective trust. For the antecedent trustworthiness to be effective as decision-mak-ing aid, consumers namely have to first trust the reviewer (Xu, 2014).
H3b. Affective trust mediates the positive effects of the source’s perceived
trust-worthiness in shaping consumers’ attitude towards a product or service.
Summarizing, expertise turned out to be an antecedent of cognitive trust, but not affec-tive trust, whereas trustworthiness seems an antecedent of affecaffec-tive trust (Johnson & Grayson, 2005). This paper proposes that the antecedents of source credibility are very likely to vary in their contributions to persuade the consumer, albeit still quite uncertain to what extent each dimension accounts for source credibility. Thus, both antecedents are likely to influence differ-ent types of trust. This research therefore tries to contribute to closure of this literature gap. I will address this gap by trying to provide evidence that both antecedents have differential indi-rect effects on one’s attitude change.
2.3 Source Credibility in an online review context
12
context. Jensen, Averbeck, Zhang, and Wright (2013) also indicate that much of the past re-search regarding online product reviews has ignored the effects of the reviewer credibility. Fo-cusing instead on the quality, validity, and dispersion of online product reviews, and their direct effects on product sales. To see what kind of role source credibility plays in an eWOM context, it is important to fist see how persuasion in the online sphere is different from persuasion in the offline sphere.
2.3.1 Persuasion in the online sphere
As with information processing in the offline sphere, persuasion in the online sphere also features both, systematic as well as heuristic processing. Nevertheless, Fennis and Stroebe (2016) suggest that heuristic processing probably accounts for the vast portion in the online environment. In this environment there is a substantial dominance of automatic, non-conscious processing (Guadagna, Okdie & Muscanell, 2013). Because consumers are increasingly online, consumers engage in multiple tasks simultaneously (e.g., chatting, buying, recreating). When people engage in multiple tasks simultaneously, interference between the tasks become inevi-table (Salvucci & Taatgen, 2008). These inferences are expected to produce cognitive load (Fennis & Stroebe, 2016). High cognitive load, in turn, promotes heuristic processing rather than systematic processing (Eagly & Chaiken, 1993; Petty et al., 1981).
For heuristics to be effective, people have to feel that their actions are freely chosen (Cialdini, 2009; Fennis & Stroebe, 2016). Since exposure to eWOM information is totally under the consumer’s control (Chatterjee, 2001), the use of heuristics is even more likely to be appli-cable in the online review context rather than in an physical environment. More specifically, since heuristic processing covers such a great part in making online purchases, heuristic cues like perceived expertise and trustworthiness seem to be highly effective in this online context.
2.3.2 Source credibility in an eWOM context
Nowadays more and more people consult product information from discussion forums, consumer review sites, weblogs, and social networking sites, to gather and exchange product information in order to make purchase decisions. In fact, a study from ChannelAdvisor (2010) revealed that ninety-one percent of their respondents mentioned that they consult online reviews and other user-generated content before purchasing a new product/service. Almost fifty percent of them indicated that they were influenced by online reviews in the way they purchase.
13
specific sense, favorable review attitude can lead to higher purchase intentions because con-sumer’s trust, or lack of trust, to a large extent determines whether we buy online or refrain from doing so (Jiménez & Mendoza, 2013). However, despite the fact that source credibility in online reviews is an underexposed field of research, the increasing consumption of online re-views seemingly contributes to a great extent in shaping online behavior, and subsequently forming attitudes (Fennis & Stroebe, 2016).
Thus, eWOM adoption (i.e., the adoption and use of eWOM communication) plays a significant role in making online purchase decisions (Cheung et al., 2009). But the anonymous character and the reduced social presence online makes it difficult for consumers to assess whether or not that trust is warranted (Fennis & Stroebe, 2016). Trust in an online review con-text therefore seems a very crucial construct. Online transactions namely proceed fundamen-tally different from face-to-face transactions. A physical product or service in exchange for money namely cannot be monitored physically by the consumer in such a way it can with bricks and mortar shopping (Fennis & Stroebe, 2016). Not to mention, in the context of eWOM com-munication, information is usually shared by unknown individuals, which also brings over un-certainty.
Consistent with this statement, Cheung and Thadani (2012) suggest that consumers have difficulty in determining the communicator’s credibility of eWOM messages. Results from the study of Chatterjee (2001) suggest that consumers — because of the lack of certainty — tend to look for so called online safety cues (Fennis & Stroebe, 2016) when determining the credi-bility of eWOM messages in order to reduce uncertainty. Cues like reviewer perception and store familiarity according to Chatterjee (2001) serve as such online safety cues.
Research regarding the believability of anonymous online product reviews has shown that the credibility of the reviewer is one of the most important antecedents of eWOM adoption (McKnight & Kacmar, 2006; Cheung & Thadani, 2012). In fact, Cheung, Luo, Sia and Chen (2009) showed that reviewer’s credibility has a direct, positive effect on eWOM adoption. Tra-ditional communication theories also show that informational factors such as source, message and receiver are the main elements that affect an individual’s information evaluation, showing that credibility perception has a significant impact on the adoption of the online review message (Shan, 2016). The extent to which the reviewer is perceived as a credible source, and can be trusted to give an objective opinion on the product or service (Goldsmith, Lafferty & Newell, 2013), therefore is quite important for consumers (and marketing managers).
14
communicator in this light leads to increased adoption of eWOM communication (i.e., the adop-tion and use of eWOM communicaadop-tion for making a purchase decision; Cheung & Thadani, 2012). Adopting these kinds of information from external sources, enhances consumer purchase decisions by affecting one’s attitude (Cheung & Thadani, 2012). Most online shoppers namely heavily rely on online product reviews to make purchase decisions (Freedman, 2008).
Building on this line of literature, it seems plausible that the use of source credibility is highly applicable in an online review context. In a more specific sense, trust plays an important role in adopting eWOM communication, and trustworthiness and expertise seem to function as such cues for trust (Xu, 2014; Johnson & Grayson, 2005). Given the fact that eWOM adoption subsequently plays a significant role in making online purchase decisions (Cheung et al., 2009), I suggest that eWOM adoption mediates the effects of source credibility on attitude change in the online review context.
H4. In an online review context, eWOM adoption mediates the relationship between
source credibility and attitude towards a product or service.
With the predictions about the applicability of source credibility in an online review con-text now being made, there are still occasions in which source credibility is more effective than in other occasions. The following section digs deeper in those occasions of effectiveness. 2.4 The moderating effects of cognitive response favorability
Due its non-commercial character, eWOM is seemingly considered as relatively credi-ble and trustworthy (Mayzlin, 2006). Consumers namely tend to be more susceptive to word-of-mouth than to product information provided by companies. Information provided by com-panies is perceived as less credible due to their self-interest in providing information (Trusov et al., 2009). Since consumers tend to be more susceptive to eWOM than to product information provided by companies, information provided by companies are generally perceived as less favorable.
15 or service (Rifon, Choi, Trimble & Li, 2004).
In general, negative feelings inform people that there must be something wrong, thereby increasing their motivation to scrutinize the arguments contained in a communication in more detail (Bless, Bohner, Schwarz & Strack, 1990). When people’s attitude towards advertising is unfavorable due their high skepticism, consumers engage in silent personal discussions with the communicator (Greenwald, 1968; Williams, 2012). People then generate counterarguments against the presented message, resulting in reduced persuasion (Festinger & Maccoby, 1964). According to the cognitive response theory, cognitive responses are the stimulator of those counterarguments (Festinger & Maccoby, 1964). It is these cognitive responses rather than the arguments themselves, which are remembered later in purchase situations. Prior re-search has challenged cognitive responses mainly to explain the production of counterargu-ments (Festinger & Maccoby, 1964). In terms of persuasion, unfavorable cognitive responses, — that result in counterarguing — seem to have a negative effect on persuasion.
Furthermore, Obermiller and Sprangenberg (1998) suggest that consumers with very high skepticism may even be impossible to persuade by means of systematic processing (i.e., intensively judge information on argumentation). Highly skeptical consumers would not be-lieve any stated claims, they may be persuasible by other means, such as noninformation ap-peals. They claim that the resistance to the persuasive effect of advertising that comes from advertising skepticism, increased counterarguing that would result in a tendency to disbelieve the claims.
However, as mentioned earlier on, source credibility, especially in systematic pro-cessing, impairs one’s ability to counterargue. A highly credible source, — according to Har-mon and Coney (1982) — reduces cognitive response activation. In this way, individuals will be limited to produce counterarguments. In situations in which individuals are very skeptical, such as consuming eWOM, high source credibility is expected to result in active self-control depletion (e.g., resisting influence). Source credibility — because of the reduced resistance — would then lead to more persuasion than a low credible source does (Tomala, et al., 2006). The favorability of cognitive responses in this light is likely to influence the effects of source cred-ibility.
H5. The positive, persuasive effects of a highly credible source are stronger among people
16 2.5 Conceptual Model
The following model (figure 1.) shows the relationship between source credibility (IV) on the attitude towards the product (DV). In this model, eWOM adoption (ME1) is the mediating
variable. I namely suggest that this variable accounts for the direct effect in that relationship. The moderating variable in this model is the favorability of cognitive responses (MD). This is
the variable which affects the strength or nature of the relationship between source credibility and attitude (Dawson, 2014).
Furthermore the expectation is that both dimensions of source credibility (i.e., trustwor-thiness and expertise) are mediated by differential types of trust. Trustwortrustwor-thiness is mediated by affective trust (ME2), whereas expertise is mediated by cognitive trust (ME3). I hypothesize
that the antecedents of source credibility have differential, be it, positive effects on one’s atti-tude towards the product. Subsequently, I suggest that eWOM adoption accounts for the posi-tive effects of source credibility on the attitude towards the product. Lastly, I expect to see unfavorable cognitive responses to have a positive influence on the relationship between the independent and dependent variable. Namely that unfavorable cognitive responses will strengthen the persuasive effects of source credibility on attitude.
Figure 1:
The mediating effect of eWOM adoption on source credibility and attitude, including the moderating effect of cognitive responses and the mediating effects of affective and cognitive trust.
+ + +
+
+ + + + (IV) Source credibility Trustworthiness cue Expertise cue (ME1) eWOM adoption (MD) Unfavorable cognitive response (DV) Attitude towards product (ME2) Affective trust (ME3) Cognitive trust
17
3. METHOD
3.1 Participants and design
Participants
156 Participants voluntarily participated in this experiment, and were randomly assigned across the conditions. After deleting four outliers using Mahalanobis distance measurements to account for odd scores, and deleting 49 unfinished cases, 103 cases remained. Participants were personally recruited to participate in this research. The data was collected through an online survey. Participants were asked to complete a five minute online survey that was designed with the online survey platform Qualtrics. Potential respondents either received a link via Linked-in or Whatsapp Messenger to fill in the anonymous survey, or were asked to fill in the digital survey in person. For motivation purposes, I raffled a €25,- gift card to one randomly picked respondent.
Design
Four conditions (expertise vs. trustworthiness vs. expertise + trustworthiness vs. control group) were used to test the hypotheses. Source credibility was the manipulated between-sub-jects variable, attitude towards the product was the dependent variable, eWOM adoption was the mediating variable and one’s skepticism towards advertising was the moderating variable. Such as stated in the conceptual model, affective and cognitive trust were also two mediating variables (figure 1).
3.2 Procedure and stimuli
Procedure
The study began with a short briefing about the purpose of the study, after which I asked for the participants’ gender, age, education level and familiarity with online purchases. Follow-ing this, participants were presented a scenario in which they were told to imagine that they visited a webpage with a product description in the course of the pre-purchase information search for a specific product. Subsequently, participants were exposed to the product descrip-tion and the corresponding online review.
18
involvement product was chosen because of the increased probability they would consume the online review. Consumers’ involvement namely plays an important role in determining the im-pact of eWOM content on purchase intention (Park & Kim, 2008; Doh & Hwang, 2009).
After being presented to the product page, participants were asked to answer questions that measured the variables of interest including a manipulation check to ensure internal valid-ity. Finally, I thanked the participants for their time and effort.
Manipulation of source credibility
Source credibility, that consists of two antecedents (i.e., trustworthiness and expertise), was the between-subjects factor that has been manipulated with a distinction between its ante-cedents. Participants were randomly exposed to a webpage with a product description including one online product review with either a neutral (N = 23), trustworthy (N = 23), expert (N = 27), or a combined (N = 30) credible communicator, see appendix A.
Pretest
To account for external validity, a pretest was conducted. In this pretest, I asked fourteen participants to which extent presenting only one review was sufficient to effectively form in-ferences about the product. I showed the participants the product and the product description with one, two, or three reviews. The pretest was conducted on the basis of a 5 point Likert scale (scale: 1 = not sufficient, 5 = completely sufficient). The results of this pretest indicated (M = 4.36, SD = 0.63) that presenting one review was indeed sufficient to form reliable inferences about the product. Thus, only one review was presented in this study.
A second pretest was conducted in order to make reliable inferences regarding the in-ternal validity of this experiment. The same group of fourteen respondents were asked to rate manipulated reviews on their perceived trustworthiness (k = 4) and perceived expertness (k =
4) of the communicator. The pretest was conducted on the basis of a 5 point Likert scale (scale:
1 = low in trustworthiness/expertise, 5 = high in trustworthiness/expertise). The corresponding review with the highest score on trustworthiness (for trustworthiness) and expertise (for exper-tise) were eventually selected for this experiment (trustworthiness: Mtrustworthiness = 4.36, Mexpertise = 4.43, SDtrustworthiness = 0.51, SDexpertise = 0,65).
Manipulating trustworthiness
19
appendix A. A ‘’helpful reviewer’’ credential is information on a review over which a profile owner has no control. An individual who has proven to be a helpful source based on peer rat-ings, can be recognized as a ‘’helpful reviewer’’ by the product website. Shan (2016) shows that this “helpful reviewers” status serves as a reputation cue and helps the customer to form the perceptions of the reviewer’s trustworthiness. The level of trustworthiness can be indicated by other people investing trust in the reviewer in terms of a rating. The common design for attributing trustworthiness to a review source is that users can directly invest trust in other members based on their posting histories and helpfulness (Cheung et al. 2009).
Manipulating expertise
Expertise can be made explicit by the administrator in terms of a caption that implies expertness by a professional title. To underscore the validity of the professional title, the ad-ministrator usually verifies the validity of the communicator by a ‘verified’ badge. This badge provides clarity over the source that his/her profession was well-grounded (Williams, 2012). This variable manipulation was adopted from Williams’ study (2012). The expertise condition in this way was displayed with a verified badge in combination with a professional title, see appendix A.
Manipulation check
Additional to the manipulations as described above, a manipulation check was conducted in the experiment. Ohanian’s (1990) scale as used in Shan’s (2016) study was applied to measure the perceived expertise of the review. This five-item seven-point Likert was anchored with “Not an expert/Expert,” “Experienced/Inexperienced”, “Knowledgeable/Unknowledgeable”, “Unqual-ified/Qualified”, and “Skilled/Unskilled” (Mexpertise = 4.73, Mtrustworthiness = 3.83, SDexpertise =
1.58, SDtrustworthiness = 1.03, α = .95). Results show that — as expected — participants in the
ex-pertise condition rated the source significantly higher on exex-pertise than respondents did in the trustworthiness condition (t(48) = 2.32, p = .025).
Ohanian’s (1990) scale as used in Shan’s (2016) study was also applied to measure the perceived trustworthiness of the review. This five-item seven-point Likert was anchored with “Dependable/Undependable,” “Dishonest/Honest,” “Reliable/Unreliable,” “Sincere/Insin-cere,” “Trustworthy/Untrustworthy” (Mtrustworthiness = 4.88, Mexpertise = 4.84, SDtrustworthiness =
1.22, SDexpertise = 1.12, α = .93). Participants in the trustworthiness condition rated the source
20 trustworthiness did not have the desired effect.
Importantly, since no differential effects were found in attitude change by manipulating reviews, the self-reported trustworthiness and expertise scores (i.e., high ‘’> 6’’ vs. low ‘’< 6’’ score on the concerned dimension) were eventually used to make more reliable inferences in testing the stated hypotheses. I therefore grouped the participants scoring high on trustworthi-ness together, and grouped the participants scoring high on expertise together. The independent variable in this way comprises the perceived trustworthiness or expertise, with perceived ex-pertise (N = 17), trustworthiness (N = 27) and control (N = 69).
Measurement one’s attitude towards the product
I measured one’s attitude towards the product as the main dependent variable on the basis of the five-item scale developed by Spears and Singh (2004). The five-item scale was scored on a seven-item Likert scale ranging from strongly disagree (In the Dutch survey trans-lated: ’sterk oneens’1) to strongly agree (in Dutch: ’sterk eens’). People described their overall
feelings about the product that was shown in the webpage on the basis of five items. The mean of the items is taken for an overall score (M = 5.34, SD = 0.95, α = .91). Sample items of the scale were to rate the product on the seven-point Likert scale as: 1. Unappealing/appealing 2. Bad/good, see appendix A.
Measurement eWOM adoption
eWOM adoption, as the mediating variable, was measured by the five item scale adopted from Cheung et al.'s (2009) study. This scale has also been used in previous research regarding eWOM and online reviews. The five items were scored on a seven-item Likert scale ranging from strongly disagree to strongly agree. The mean of the items is taken as an overall score (M
= 5.25, SD 1.12= α = .92). All items were positively worded. Sample items of the scale were
as follows: “This review has enhanced my effectiveness in making purchase decision” and “To what extent do you agree with the review” , see appendix A.
Measurement cognitive response
One’s favorability of the cognitive response, as the moderating variable, was measured by the Obermiller and Spangenberg (1998) scale, measuring the extent of skepticism towards advertising. The extent of skepticism comprises the favorability of cognitive responses.
1 In order to acquire substantial data, I created a completely Dutch version of the survey. The respondents
21
Increased skepticism namely has been found to negatively affect perception of the communica-tion (Janssen & Fennis, 2017). The use of skepticism to infer the favorability of the cognitive responses is adopted from the Janssen and Fennis (2017) study. In this study, the authors created an advertisement disclaimer to manipulate the favorability of the cognitive responses. In the Janssen and Fennis (2017) study, the authors made the assumption that respondents in the dis-claimer condition to show unfavorable cognitive responses. Thus, in this study, participants with high scores on the skepticism scale are also believed to show unfavorable cognitive re-sponses.
This scale consists of nine items, operationalized on a seven-point Likert scale ranging from strongly disagree to strongly agree. The mean of the items is taken as an overall score (M
= 3.34, SD = 0.95, α = .86), with higher scores representing higher skepticism. Sample items
of the scale were: “We can depend on getting the truth in most advertising’’ and “Advertising’s aim is to inform the consumer.’’, see appendix A.
Measurement cognitive trust
The four items measuring cognitive trust were adopted from Xu (2014) and Johnson and Grayson’s (2005) study. The four items were scored on a seven-item Likert scale ranging from strongly disagree to strongly agree. Sample items include ‘‘I have to be cautious about acting on the advice of this reviewer because his opinions are questionable’’ and ‘‘I can rely on this reviewer for suggesting a product that I am interested in’’, see appendix A. The mean of the items was taken as an overall score (M = 4.32, SD = 1.27, α = .83). One item was deleted to ensure higher internal validity. Note: item 1 of this scale was reversed recoded afterwards in order to calculate the mean score.
Measurement affective trust
22 3.3 Analysis plan
The raw dataset regarding the survey was downloaded from the Qualtrics server before it has been exported to the statistical analysis program SPSS. Subsequently, I accounted for outliers using the Mahalanobis distance measurement after which I structured the data in such a way that I was able to make inferences based on the data. To ensure internal validity, reliabil-ity and consistency, the Cronbach’s Alpha test was executed. By using established scales, no factor analysis was needed to ensure internal validity. After these preparation checks, I was able to test the stated hypotheses, based on differential statistical techniques.
Regarding the hypotheses testing, the analysis of variance (ANOVA) was used to test the main effects. Independent t-tests were performed to subsequently test the differential main effects across the antecedents of source credibility. After testing the main effects, the PROCESS Macro in SPSS was executed to test the indirect effects (i.e., moderation and mediation; Model 5, 5000 bootstrap samples). This model namely allows to test for moderated-mediation with up to ten mediators operating in parallel (Hayes, 2013). Thus by testing the complete model in the PROCESS Macro, — allowing for simultaneous tests of mediation and, moderation by multiple variables, and bootstrapping procedure to test for indirect effects — the hypotheses regarding the indirect effects were tested. Given the fact that the independent variable was a multicate-gorical variable, the categories were automatically dummy-coded via the PROCESS Macro. The control group was taken as the benchmark score.
4. RESULTS
4.1 Sample demographics
23 4.2 Results of hypotheses testing
Effects source credibility on attitude. With the experiment just been discussed, an attempt was made to find out whether or not source credibility shows a positive effect on atti-tude. I tested the following hypothesis:
H1: High source credibility, regardless of the extent of processing one engages in, will result in a significant positive attitude change to-wards the product or service.
In order to analyze whether or not the attitude towards the product differs when adding credibility to the source, I performed a One-way ANOVA of the self-reported perceived credi-bility of the source (i.e., scores > 6 on expertise and trustworthiness) on attitude. As expected, respondents who indicated the source as highly credible, reported a significantly higher atti-tudes towards the product (M = 6.01, SD = 0.77) compared to respondents who perceived the source to be low in credibility (M = 5.14, SD = 0.62), [F(1, 100) = 32.61, p < .001]. Thus, perceived source credibility does positively affect attitudes, see figure 2.
Differential effects dimensions source credibility. An attempt was made to find out whether or not the antecedents of source credibility show differential effects on attitude towards the product. I therefore tested the following hypothesis:
H2: The dimensions of the communicator’s source credibility (i.e., trust-worthiness and expertise) both have different, be it positive, effects in shaping consumers’ attitude towards a product or service.
To check for possible significant differences between the dimensions, a One-way ANOVA was performed. This One-way ANOVA showed that there are significant differences
24
between the dimensions of source credibility [F(2, 110) = 25.38. p < .001]. In order to check whether perceived trustworthiness or expertise differ in terms of attitude, an independent t-test was performed. This independent t-test showed that participants who rated the source as trust-worthy (M = 6.21, SD = 0.39) showed significantly higher attitudes than participants who rated the source as an expert (M = 5.85, SD = 0.75), [t = -2.15, df = 42, p = .037], see figure 3.
The PROCESS Macro in SPSS (Model 5, 5000 bootstrap samples) was subsequently executed to test whether both antecedents show differential effects. This analysis showed that both, perceived trustworthiness (b = 1.72, t(104) = 3.34, p < .001), and perceived expertise (b = 1.78, t(104) = 3.04, p = .003) are significant predictors of attitude. Overall, this model ex-plained a significant proportion of variance in the attitude score, R2 = .57, F(8,104) = 17.29, p < .001. Thus, both antecedents are equally significant predictors of attitude.
However, where the complete model shows that both antecedents significantly accounts for approximately the same proportion of variance in predicting attitude, seemingly, perceived trustworthiness is more important in shaping consumers’ attitude based on the analysis of var-iance. Reviews with perceived trustworthiness namely show more positive attitudes than re-views with the source perceived as an expert does.
Differential types of trust. To dive deeper into the differential effects of the anteced-ents of source credibility, I tested the following hypotheses:
H3a: Cognitive trust mediates the positive effects of the source’s perceived expertise in shaping consumers’ attitude towards a product or ser-vice.
H3b: Affective trust mediates the positive effects of the source’s perceived trustworthiness in shaping consumers’ attitude towards a product or service.
25
attitude. Step 4 of the analysis revealed that, including the mediator (cognitive trust) in the model, perceived expertise was still a significant predictor of attitude (b = 1.78, t(104) = 3.04,
p = .003). Perceived expertise, including the mediator, still has a direct influence on attitude.
Thus, incorporating the mediation variable in the regression model did not made the independent variable a non-significant predictor. Also the mediator was not a significant pre-dictor of attitude. Following the rules of mediation (Hayes, 2013), this analysis showed that no mediation effect has occurred (see figure 4.). Subsequently, the indirect effect was tested using a bootstrap estimation approach with 5000 samples. Incorporating the mediating variable in the model slightly decreased significance level, but the OMNIBUS-test showed that zero is in the range of LLCI and ULCI, which means that we can say with 95% certainty that also partial-mediation has not occurred (CI = -.3106, .3066).
As a side effect, there was found a significant effect of perceived expertise on affective trust (b = 0.79, t(110) = 2.88, p < .001). This means that ostensibly, perceived expertise also increases affective trust. However, inherent to cognitive trust, no mediation effect was found.
H3b. In step 1 of the mediation model testing H3b, the regression of the perceived trust-worthiness on attitude, ignoring the mediator affective trust, was highly significant (b = 0.95
t(110) = 4.20, p < .001). Perceived trustworthiness appears to directly influence attitude. Step
2 showed that the regression of the perceived trustworthiness on the mediator, affective trust, was also highly significant (b = 1.21, t(110) = 5.24, p < .001). Perceived trustworthiness is also directly influencing affective trust. Step 3 of the mediation process showed that the mediator (affective trust), controlling for the independent variable, was a non-significant predictor of attitude (b = 0.03, t(104) = 0.42, p = .676). Apparently, affective trust does not have any direct influence on one’s attitude. Step 4 of the analysis revealed that, including the mediator (affec-tive trust) in the regression, perceived trustworthiness was still a significant predictor of attitude
Cognitive trust
a: 1.90*** b: 0.03
Expertise c’: 1.72*** Attitude
c: 0.69***
Note. *** = p < 0.01 ** = p < 0.05 * = p < 0.10
26
(b = 1.72, t(104) = 3.34, p = .001). Thus, perceived trustworthiness, including the mediating variable, still has significant influence on attitude.
Incorporating the mediation variable in the regression model did not made the effect of the independent variable on the dependent variable non-significant. The mediator was also not a significant predictor of attitude. Following the rules of mediation (Hayes, 2013), this analysis showed that no mediation effects have occurred (see figure 5.). Subsequently, the indirect effect was tested using a bootstrap estimation approach with 5000 samples. Incorporating the mediat-ing variable in the model did not decrease the significance level of the direct effect on attitude and also the OMNIBUS-test showed that zero is in the range of LLCI and ULCI, which means that we can say with 95% certainty that also partial-mediation has not occurred (CI = -.1252, .1771).
As a side effect, there was also found a significant effect of perceived trustworthiness on cognitive trust (b = 1.44 t(110) = 6.21, p < .001). This means that ostensibly, perceived trustworthiness also increases cognitive trust. However, the same as with the perceived expert-ness, no further mediation effect was found.
eWOM adoption. To test if eWOM adoption accounts for the effects of source credi-bility in an online review context, I tested the following hypothesis:
H4: In an online review context, eWOM adoption mediates the relation-ship between source credibility and attitude towards a product or service.
The PROCESS Macro in SPSS (Model 5, 5000 bootstrap samples) was executed to test the hypothesis above. In step 1 of the mediation model, the regression of source credibility (i.e., perceived expertise and trustworthiness vs. not perceived) on attitude, ignoring the mediator eWOM adoption, was significant (b = 0.39 t(111) = 2.58 p = .012). Perceived source credibility
Affective trust
a: 1.21*** b: 0.03
Trustworthiness c’: 1.72*** Attitude
c: 0.95***
Note. *** = p < 0.01 ** = p < 0.05 * = p < 0.10
