Master Thesis Marketing Management

(1)

UNIVERSITY OF GRONINGEN

The Effect of Personality Matching in Online

Reviews

Master Thesis Marketing Management

(2)

The Effect of Personality Matching in Online

Reviews

Master Thesis Marketing Management

Florentine Sterkel

June 2018

University of Groningen Faculty of Economics and Business

Department of Marketing

PO Box 800, 9700 AV Groningen (Netherlands)

Berkelstraat 24 9725GV Groningen (Netherlands) Mobile: +49 (0) 1575414524 E-mail: f.a.sterkel@student.rug.nl Student number: s3497577 Supervisors: University of Groningen First Supervisor: R.P. (Roelof) Hars

(3)

Management Summary

True to the well-known adage of “birds of a feather flock together”, dating services believe to have the perfect technique to bring people together but also other companies should take advantage of this phenomenon as matching like-minded people seems to improve sales. Focusing on the personality as an unique differentiator of an individual, one can find indications that emphasizing the personality of the consumer, and sending him or her personalized messages in line with his or her personality, might have a positive effect on the consumer’s willingness to buy. Especially opinions posted online have an enormous impact on consumer’s shopping behavior, and through the use of online reviews, firms might be in the position to positively affect consumers’ purchase decisions by matching the personality of reviewer and review reader. Moreover, it is assumed that if the reader’s actual personality matches his or her aspirational personality,the relationship between similarity in personality and willingness to buy gets stronger. In consequence, this study aims to uncover if product recommendations are more effective if done by someone who closely matches the actual and aspirational personalities of the review reader. Reflecting on the review reader’s aspirational personality is a new contribution to the marketing literature in the online domain.

An answer to the research’s aim was achieved by firstly investigating the importance of online reviews. Secondly, prior research about personality, social influence and similarity attraction was taken into account. In line with the hypotheses derived from the literature review, a quantitative study based on an online questionnaire (n = 202) examined the effect of personality in online reviews, focusing on the Big-Five personality dimension of extraversion. Results demonstrate that the willingness to buy the product mentioned in the extraverted review is 29% higher than in the introverted review. Moreover, the review reader significantly likes the extraverted reviewer better, and has in doing so an increased willingness to buy. Additionally, this research found out that individuals aspire to be extraverted, and that language use in text can reflect one’s personality. However, review readers were not significantly attracted to reviewers with more similar personalities taking into account the actual and aspirational personalities of the reader.

(4)

(5)

Abstract

Many purchase decisions are made based on online reviews. Yet little is known about the social psychology of online reviews. Especially, there is a lack of the extent to which personality and primarily similarity in personality influence the reader’s willingness to buy. Therefore, this study investigates if online product recommendations are more effective if done by someone who closely matches the actual and aspirational personalities of the review reader. This question was examined by focusing on the Big-Five personality dimension of extraversion, doing an online survey (n = 202). Results indicate that the willingness to buy the product mentioned in the extraverted review is 29% higher than in the introverted review. This effect is mediated by liking the reviewer: By reading a review of an extraverted reviewer, the reader significantly likes the reviewer and has an increased willingness to buy. Additionally, this study discovered that individuals aspire to be extraverted. However, review readers were not significantly attracted to reviewers with more similar personalities in regard to the reader’s actual and aspirational personalities. Based on the findings obtained, it is proposed that the reviewer’s personality plays an important role in the buying process of the reader.

(6)

Preface

“We don’t make money when we sell things. We make money when we help customers make purchase decisions.” - Jeff Bezos, founder of Amazon.com, Inc.

In the past, Bezos, the founder of Amazon.com, Inc. was criticized for permitting customer reviews. Today, 90% of consumers read online reviews before entering a business, 67,7% of consumers rest their purchase decision upon reviews, and 74% of consumers state that positive reviews make them trust a local business more (Erskine, 2017). It is not deniable that online reviews have a big impact on consumers’ purchase decisions, and companies should take advantage of it by adapting online review configurations in a way that it positively affects the reader’s buying decision. The personality of review reader and reviewer might play an important role as examined in this study.

I would like to express my appreciation and gratitude to my friends and family, as well as to my advisory supervisor Roelof Hars who greatly supported me especially during the last semester.

I hope you enjoy reading my thesis! Florentine Sterkel

(7)

1. Introduction

Today’s consumers do not only rely on personal, face-to-face product information from friends, family or salespersons, but also trust in recommendations from strangers or influencers who share their opinions and experiences online, using Web 2.0 tools like social media platforms, discussion forums and consumer review sites (Gupta and Harris, 2010; López and Sicilia, 2014). In its Global Trust in Advertising Survey, answered by 30,000 online respondents in 60 countries, Nielsen (2015) found out that 66% of the survey respondents trust online posted consumer opinions. Regarding traditional media sources, only 60% trust ads in newspapers and 58% trust magazines. Notably, traditional media is trusted less than online recommendations (Nielsen, 2015). Thus, opinions posted online have an enormous impact on the consumer’s shopping behavior and many purchase decisions are made based on online recommendations. Through online reviews, consumers on the one hand are able to inform themselves via different information sources before the purchase, but on the other hand, Chen and Xie (2008) note that an increased cognitive effort is required in order to find personally relevant information within the great amount of information in comparison to traditional media.

In order to stimulate sales, firms use consumer reviews as a marketing tool spreading

additional information online, sponsoring influencers and even manipulating

recommendations to make them more favorable(Dellarocas, 2006). Hence, companies spend

(9)

In general, word of mouth (WOM) is known as “the act of consumers providing information about goods, services, brands, or companies to other consumers” (Babić Rosario, Sotgiu, De Valck and Bijmolt, 2016, p. 297). If such information is communicated through the internet, it is known as electronic word of mouth (eWOM). Hennig-Thurau, Gwinner, Walsh and Gremler (2004) define eWOM as “any positive or negative statement made by potential, actual, or former customers about a product or company, which is made available to a multitude of people and institutions via the Internet” (p. 39). Primary reasons for individuals to leave reviews online are social benefits, economic advantages, altruism and extraversion (Hennig-Thurau et al., 2004; Park and Park, 2008). Mainly, eWOM consists of opinion-seeking behavior and opinion-giving behavior (López and Sicilia, 2014).

(10)

Denissen, Van Aken and Meeus, 2010). In this study, similar levels of extraversion, agreeableness and openness acted as predictors for friendships. Thus, similarity in personality has a big impact on the interaction of individuals. This knowledge should not only remain in the literature of social psychology but might also have a big impact on the domain of marketing. Consumers may be more receptive towards messages from reviewers with a similar personality.

In general, authors demonstrated proof for similarity attraction in terms of attitudes, values, and beliefs as people’s view of the world is confirmed by meeting a similar other, and because similar beliefs end up in less incongruities and conflicts (Berscheid, 1985; Byrne, 1971; Klohnen and Luo, 2003). However, limited research in a marketing context has concentrated on the topic whether similarity in personality traits leads to attraction (Klohnen and Luo, 2003). Authors claim a lack of knowledge about the extent to which principles of social psychology carry over into the online domain and little is known about how one can use those principles in activities such as community building, e-commerce, marketing, and personalization of services (Li and Chignell, 2010).

In response, the underlying research investigates online reviews from a personality based perspective. Focusing on the importance of the reviewer’s personality, this study aims to uncover if product recommendations are more effective if done by someone who closely matches the personality of the reader. To this end, this study explores if the personality match between the eWOM sender and eWOM reader increases the reader’s willingness to buy a certain product or service mentioned in the online consumer review. As stated before, homophily positively influences the eWOM effectiveness as each person validates, enriches, or strengthens his or her self-concept of the similar other (e.g., Byrne, 1971; Clore and Byrne, 1974; Dryer and Horowitz, 1997), and therefore, by matching the personalities of eWOM sender and reader, it is hoped to increase the review effectiveness.

(11)

self-descriptions. Therefore, in the subsequent analysis, I will not only take into account personality matching in terms of the review reader’s actual personality, but I will also reflect on the aspirational personality of the reader. This is a new contribution to the marketing literature in the online domain as until now, research has operationalized similarity in personality only in terms of the actual personality. For instance, Li and Chignell (2010) found out that Blog readers were more attracted to blog writers with more similar personalities but only have taken into account the actual personality of the reader.

With this research study, I want to provide insights to the similarity in personality and eWOM literature not only for academics but also for practitioners. All in all, this paper adds value to the incomplete academic research examining how firms can best organize their social networks to meet strategic goals (Naylor, Lamberton and West, 2012). Specifically, this paper contributes to a more efficient and effective targeting of customers with eWOM by taking into account his or her personality. This study might find support for the recent idea to establish personality-based recommendation systems and to define personality-based communication strategies (Kosinki, Bachrach, Kohli, Stillwell and Graepel, 2014; Chou, Picao-Vela and Pearson, 2013), emphasizing the personality of the reader and sending him or her personalized messages, depending on the reader’s personality. Furthermore, firms could greatly expand their online platform user understanding, and improve their services and the user experience (Kosinki et al., 2014). Finally, this leads me to the following research question:

RQ: Are online product recommendations more effective if done by someone who closely matches the actual and aspirational personalities of the eWOM reader?

To examine the influence of the personality match on the purchase decision of consumers, I will first take a closer look at the literature itself, analyzing previous research about online reviews, the Big-Five personality traits, social influence and similarity attraction. Then, I will present the conceptual model, and point out the methodology used in the present research. After evaluating the results, I will discuss the findings and give recommendations for further research.

2. Literature Review

(12)

2.1 Online Reviews

Understanding how online reviews affect consumers’ purchase decisions and how companies can take advantage of this moment of truth in the customer journey is vitally relevant to companies that rely on eWOM to spread information about their services and products (Zhu and Zhang, 2010). Nowadays, firms compete in an internet-based environment. Opinions, recommendations and experiences posted online can be found in a notable variety of contexts: restaurants, hotels, movies, and many more mostly benefit from consumer reviews. Individuals increasingly use online reviews in their decision-making, and the popularity of websites such as Yelp.com, TripAdvisor.com, or Amazon Customer Reviews is growing constantly. Numbers are even more alerting: 90% of consumers read online reviews before entering a business, 67,7% of consumers rest their purchase decision upon reviews, and 74% of consumers state that positive reviews make them trust a local business more (Erskine, 2017).

This increased usage and trust in online product reviews and product evaluation websites has not escaped the notice of companies who started to encourage consumers actively to share personal recommendations about products and services with the internet community (Floyd et al., 2014). eWOM in online reviews can therefore stimulate consumers to buy a certain product and service. As stated before, Babić Rosario et al. (2016) highlight in their meta-analytic review that “eWOM is positively correlated with sales (.091) but its effectiveness differs across platform, product, and metric factors” (p. 297). According to these authors, eWOM on social media platforms is more effective when eWOM readers can associate with similar eWOM senders on the basis of homophily details such as username, profile picture, location and profile page. Another possible homophily detail can be personality. Therefore, having in mind the purpose of the research, the question arises how personality is defined, and out of which factors it is composed.

2.2 The Big-Five Personality Traits

(13)

been labeled as follows: (I) extraversion (talkative, assertive and energetic), (II) agreeableness (good-natured, cooperative and trustful), (III) conscientiousness (orderly, responsible and dependable), (IV) neuroticism versus emotional stability (not calm, neurotic and easily to upset), and (V) openness (intellectual, imaginative and independent-minded) (Cattell, 1943; Goldberg, 1990; John and Srivastava, 1999). Interestingly, in earlier years, the Big-Five research was dominated by factor analysts and is now shifting towards an investigation of the predictive power of the Big-Five dimensions (Cuperman and Ickes, 2009). For instance, already in the 1990s, authors found out that differential relations between the personality dimensions can forecast one’s work performance (Barrick and Mount, 1991). Additionally, the Big-Five traits are said to be predictors of behavior (Cuperman and Ickes, 2009). Ozer and Benet-Martínez (2006) linked the Big-Five dimensions to individual, interpersonal, and social outcomes such as, inter alia, happiness, criminality, physical health and volunteerism. In this study, the predictive power of the Big-Five dimensions will be evaluated by analyzing the effect of personality in online reviews.

The following section of this research will discuss how individuals get socially influenced by their environment, and what effect this influence has, as online reviews can be seen as a form of social influence that has an enormous impact on the purchase behavior of consumers.

2.3 Social Influence and Liking

Fennis and Stroebe (2017) argue that individuals get influenced by around 1000 commercial stimuli a day. Considering online reviews as one of these stimuli, consumers face a great amount of information online and struggle to find personally relevant information (Chen and Xie, 2008). Moreover, individuals do not only get influenced by marketing sources, but also by non-marketing sources like friends, family and the environment in general. This influence can take place online or offline. Because of the amount of influence and information, individuals are often overwhelmed and cannot manage making rational decisions. To this end, researchers investigated that social influence plays an essential part in consumer behavior (Cialdini and Goldstein, 2004). Cialdini (2009) defines six “weapons of influence” to stimulate individuals to agree to a request: (I) reciprocity, (II) commitment and consistency, (III) social proof, (IV) authority, (V) scarcity, and (VI) liking.

(14)

often rely on the heuristic rule that the more an individual likes someone, the greater should be the willingness to comply with the request; the rule is used automatically and unwittingly also when the request comes from unknown people (Burger, Soroka, Gonzago, Murphy and Somervell, 2001). In this regard, “compliance refers to a particular kind of response — acquiescence — to a particular kind of communication — a request” (Cialdini and Goldstein, 2004, p. 592). In line with Cialdini (2009), liking can be seen as a predictor of the behavior to comply. By liking a person, the individual is more likely to perform a task the liked individual has recommended. This can be transferred into the context of online reviews: When liking the reviewer, the eWOM reader might be more likely to buy the recommended product or service mentioned in the review (Cialdini, 2009). An important predictor of liking a person (here the reviewer) is the personality since individuals like others who they believe to have desirable personality traits, and dislike people who seem to have undesirable personality traits (Tenney, Turkheimer and Olkmanns, 2009). Prior researchers have focused on the salesman’s personality traits and its effect on the consumer’s purchase behavior (Lamont and Lundstrom, 1977; Martin, 2011). These researchers tried to identify the combination of personality traits of top sales persons who get the customers to buy their products. Results indicate that personality dimensions directly have an impact on top salespersons’ selling style and finally their success in selling the product or service (Martin, 2011). Putting this into the context of the present research, the reviewer’s personality plays an important role in the purchase decision of the review reader. Thus, it depends on the reviewer’s personality whether the reader likes the reviewer or not, and this has an impact on the review reader’s willingness to buy. When liking the reviewer, the reader might be more likely to purchase the product or service mentioned in the review. Liking the reviewer clarifies the relationship between the personality of the reviewer and willingness to buy, and can be seen as the mediator of this relationship.

Therefore, I propose the following first hypothesis:

Hypothesis 1: There is a positive relationship between the reviewer’s personality and the review reader’s willingness to buy, and this relationship is mediated by liking the reviewer.

(15)

consumers’ names than when it does not (Brendl, Chattopadhyay, Pelham and Carvallo, 2005). Accordingly, in the following chapter it will be examined if, how and why individuals might be attracted by the similar other.

2.4 Similarity Attraction

The two well-known adages of “opposites attract” and “birds of a feather flock together” are deeply discussed by researchers and the general public (Klohnen and Luo, 2003). Dating services believe to have the perfect technique to bring people together but also other firms should take advantage of it as matching like-minded people seems to improve sales.

In general, it is widely investigated that similarity influences attraction. In 1965, Byrne and Nelson propose the so called “law of attraction” through which the authors found out that “attraction toward a stranger is a positive linear function of the proportion of that stranger's attitudes which are similar to those of the subject” (Byrne and Griffitt, 1966, p.699). The law of attraction assumes that similarity between two persons is satisfying because each person validates, enriches, or strengthens the self-concept of the similar other (Byrne, 1971; Clore and Byrne, 1974; Dryer and Horowitz, 1997). Importantly, Berger (1975) announced that the similarity effect is “one of the most robust relationships in all of the behavioral sciences” (p. 281), and Layton and Insko (1974) affirmed that similarity attraction is “one of the best documented generalizations in social psychology” (p. 149). Therefore, it is not surprising that a meta-analysis, taking into account 313 similarity studies, detected that similarity generates a positive, moderate effect on attraction (Montoya, Horton, and Kirchner, 2008). Bearing in mind the results of this meta-analysis and prior studies (e.g. Berger, 1975; Layton and Insko, 1974), research focusing on complementary attraction will not be discussed in this paper as one can find little evidence that the opposite attracts.

(16)

For sure, similarity attraction cannot only be examined by looking at personality similarity of two persons, but also by taking into account, for instance, attitude similarity (Byrne, 1971). For example, Hendrick and Page (1970) discovered that people sharing similar attitudes are also more influenced by them. However, authors claim that a lot is known about similarity in terms of attitudes, values, and beliefs, but little is known whether similarity in personality characteristics also leads to liking (Klohnen and Luo, 2003). McLaughlin (1971) assumes that the similarity attraction effect works in a comparable extent for personality traits as it does for attitudes. In his study, similarity was found to affect attraction. Nevertheless, this was in a non-Marketing setting. A more recent study, having a similar frame to the underlying research, found out that Blog readers were more attracted to blog writers with more similar personalities (Li and Chignell, 2010). These authors concluded that personality metrics may be useful in supporting online social interaction and should be considered in the design of content recommendation systems. Results were found through two studies being structured as laboratory experiments.

Prendergast, Ko and Yuen (2010) argue that similarity between a forum’s topic and a user’s interests has a positive influence on consumer’s willingness to buy products presented in the forum. In their first experiment, Nass and Lee (2001) found out that participants were more likely to buy a book when a computer-synthesized voice on a book-buying website matched the participant’s personality. Additionally, as already discussed, individuals are more likely to comply with a request and to purchase an advertised product or service when feeling similar to the salesman (Cialdini, 2009). Further, individuals like persons who are similar to themselves as individuals tend to like themselves (Allport, 1961). Thus, similarity in personality might positively influence the willingness to buy a product or service.

(17)

being attracted to the similar reviewer should result in liking the reviewer and finally, should positively influence the willingness to buy the product or service mentioned in the review. Hence, I assume that there is a positive relationship between the reviewer’s personality and the reader’s willingness to buy if the review reader’s actual personality matches the reviewer’s personality, and this relationship is positively mediated by liking the reviewer. The strength of the relation between the reviewer’s personality and liking the reviewer depends on the reader’s actual personality which can be seen as a moderator of this relationship.

Consequently, taking the mentioned prior literature review into account, I arrive to the following hypothesis:

Hypothesis 2: The mediation effect becomes stronger when the review reader’s actual personality matches the reviewer’s personality.

(18)

to the eWOM sender when the reader’s actual and aspirational personalities closely match the personality of the reviewer. To this end, in the present research, I examine the personality match not only in terms of the actual personality of the eWOM reader but also in terms of the aspirational personality. This is a new contribution to the marketing literature in the online domain as until now, research has operationalized similarity in personality only in terms of the actual personality. For instance, Li and Chignell (2010) found out that Blog readers were more attracted to blog writers with more similar personalities but only have taken into account the actual personality of the reader.

(19)

ideal and actual self is small, and it is hard to distinguish the ideal from the actual self (Herbst,Insko and Gaertner, 2003). This corresponds to the findings of LaPrelle et al. (1990). However, there are few opposing notions, distancing themselves from the idea that similarity attraction is mainly affected by the ideal self. Herbst et al. (2003) note that similarity to the ideal self may not always lead to attraction as there might be opposing reactions to the ideal other. Seeking to be like a certain individual and aiming to create a match between the ideal self and the self of the apparently perfect other can be inspiring and exciting, but at the same time the comparison with somebody who is perfect in the eyes of the viewer might be devastating (Herbst et al., 2003). The described mechanism is called upward social comparison which can lead to negative emotions such as jealousy and envy, and can evoke the wish to distance the self from the aspiring individual. (Herbst et al., 2003; Salovey and Rodin, 1991; Tesser, Millar, and Moore, 1988). In general, people often compare themselves to others. They perceive others as being better, more beautiful, more intelligent, being in a better shape, or being more liked. Therefore, upward social comparison may be ego-deflating as people equate being better with being good (Festinger, 1954). Hence, individuals might dislike online consumer reviewers that only appeal to their ideal and aspirational self as they might feel threatened in their actual self-concept, and consequently want to distance themselves from the reviewer. Nevertheless, there is evidence that upward comparison does not always come along with the ego-deflating experience. Collins (1996) argues that “upward comparison helps people to achieve and maintain superiority by serving self-evaluative and self-improvement functions” (p. 67). For instance, dieters often pin pictures of people in a better shape on their refrigerator. This can be seen as upward comparison because it reminds the dieter of his or her own (disliked) shape. The admired person is used as a goal that guides the dieter’s eating behavior, and offers the motivation/inspiration needed to meet that goal (Collins, 1996). The individual engages in upward social comparison hoping that self-enhancement and self-improvement will occur. In terms of the present study, the review reader might use the reviewer as a motivation to be, for instance, more extraverted/introverted in the future, aspiring to be like the reviewer.

(20)

actual and the aspirational personalities of the review reader, the reader reaches to be like his or her aspiring self, and this effect strengthens the relationship between similarity in personality and willingness to buy. Finally, the strength of the relationship between the personality match and willingness to buy, which is influenced by liking the reviewer, depends on the match between the reader’s actual and aspirational personalities.

Therefore, I propose my last hypothesis:

Hypothesis 3: The moderating effect of the review reader’s actual personality becomes stronger when there is also a match between the actual and aspirational personalities of the review reader.

2.5 Conceptual Model

By taking into consideration the findings of the prior literature review and the subsequent hypotheses, the following moderated mediation conceptual model (figure 1) was developed. The dependent variable consists of the willingness to buy the product or service mentioned in the review, the independent variables displays the personality of the reviewer, liking the reviewer acts as a mediator of the relationship, and the actual and aspirational personalities of the eWOM reader function as moderators. The framework proposes that the relationship between the reviewer’s personality and the willingness to buy the product or service mentioned in the online review is mediated by liking the online reviewer, as an individual is more likely to buy a product or service from a person they like (Cialdini, 2009). Furthermore, it is assumed that the mediation effect becomes stronger when the actual personality of the review reader matches the personality of the reviewer which is in line with the similarity attraction literature. Finally, the framework displays that this effect is even stronger when the reader’s actual personality matches the aspirational personality. The reader’s actual and aspirational personalities affect the strength of the relation between the reviewer’s personality and liking the reviewer.

(21)

3. Methodology

The following chapter elaborates on the methodology used to test the proposed hypotheses on the basis of the conceptual model (figure 1). Therefore, the research method will be explained, followed by the data collection, taking into account the participants, as well as the procedure and measurements. Finally, the plan of analysis will be discussed.

3.1 Research Method

In order to test the proposed hypotheses developed in the literature review, and which are also included in the research model, a self-administered online questionnaire was conducted as “surveys are the main means of collecting quantitative primary data” (Mooi and Sarstedt, 2011, p.52). The survey was developed using the software qualtrics.com and contained 26 questions. It took participants around 5 minutes to complete the questionnaire online. Before launching the questionnaire, it was tested with a pre-sample of 10 persons. Measurements were based on existing scales. The final questionnaire can be found in appendix I.

3.2 Data Collection

Regarding the data collection, the data was collected by spreading the online survey via social media like Facebook and WhatsApp, and via e-mail making use of the snowball sampling procedure. Therefore, the link of the online questionnaire was sent out to a couple of people who were asked to forward the link to other persons, or to share it on their social media platform. Snowball sampling is especially useful when the access to individuals is difficult and can improve the response rate significantly. In general, this sampling procedure is used frequently, can be easily executed, and does not involve high costs (Mooi and Sarstedt, 2011). Moreover, through the snowball sampling procedure, a heterogeneous sample that reflects the population in general was hoped to be achieved. Consequently, it was an appropriate procedure for the underlying study as through this sampling procedure, the response rate was expected to be improved (friends shared the link with their friends). As a goal, a sample size of at least 150 respondents was required.

3.2.1 Participants

(22)

survey. Having at least a good command in English is crucial to guarantee that participants understand the online survey questionnaire completely. This research adopted the fact that individuals must be 14 years old or older from the terms and conditions of taking part in a GfK survey (GfK, 2018).

3.2.2 Procedure and Measurements

As stated before, the questionnaire was distributed in English. All the items, except for the variable reviewer’s personality, were rated on a 5-point Likert scale. The variables included in the conceptual model as depicted in figure 1 were tested or, in case of the independent variable (reviewer’s personality), manipulated. In total, the five variables consisted of: actual personality, aspirational personality, reviewer’s personality, liking the reviewer and reader’s willingness to buy. To improve the response rate, participants were forced to answer the questions. However, participants had the option to indicate “neither agree nor disagree” or “neither likely nor unlikely” for each item.

(23)

extraversion and therefore, extraverted individuals are characterized as sociable, talkative and enthusiastic, whereas introverted persons are known to be more reserved, shy and less self-confident (Costa and McCrae, 1992; Goldberg, 1990). Furthermore, extraverted individuals can more often be found in conversations, use a greater amount of words, express positive emotions through laughing and spend less time alone compared to introverts (Mehl et al., 2006).

As another important characteristic, extraverted individuals can be seen as being more dominant, and introverted persons can be seen as more submissive. This declaration is based on the two-dimensional interpersonal circumplex theory of interpersonal behavior in which especially the traits extraversion and agreeableness are organized along two orthogonal axes (McCrae and Costa, 1989; Wiggins, 1982). More in specific, McCrae and Costa (1989) found out that the interpersonal circumplex is defined by the two Big-Five traits of extraversion and agreeableness. Therefore, the extraversion dimension ranges from the dominance to submissiveness axis and is commonly discussed as the power or control factor (McCrae and Costa, 1989; Nass, Moon, Fogg, Reeves, and Dryer, 1995). The agreeableness dimension ranges from the warmth to hostility axis and is usually referred to as the affiliation or warmth factor (Lorr and McNair, 1965; Nass et al., 1995). An individual scoring high in dominance is characterized as leading a group or a person, directing and giving orders. Contrarily, when scoring low in dominance, an individual avoids directing, and permits others to take the lead (Lorr and McNair, 1965). Finally, in this study, when discussing the dimension extraversion, I also refer myself to the dimension of dominance by taking into consideration the interpersonal circumplex theory in personality. Other authors also have based themselves on the dimension of dominance when bringing up extraversion as for instance Nass et al. (1995) or Moon (2002).

3.2.2.1 Control Variables

In the first survey block, participants were asked to give answers regarding socio-demographic questions like gender, country of origin, age, education and English skills. Importantly, these control variables were not of primary concern for the survey.

 Gender was measured by one item, asking the participants “What is your gender?” The response categories were 0 (male), 1 (female).

(24)

 Furthermore, age was measured through the question “What is your year of birth?” Participants needed to enter their year of birth.

 Education was evaluated using one item: “What is the highest degree or level of school you have completed?” Response categories were: 1 (No schooling completed), 2 (High School), 3 (Trade/technical/vocational training), 4 (Bachelor degree), 5 (Master degree), 6 (PhD or higher).

 Finally, English skills were measured using one item: “How would you describe your English skills?” Response possibilities were: 1 (No English skills/Beginner), 2 (Basic communication skills), 3 (Good command), 4 (Very good command), 5 (Native speaker) (LMU, 2018).

3.2.2.2 Actual Personality

Afterwards, the next objective was to gather information about the respondent’s actual personality (e.g. “who I actually am”). The actual personality was measured by asking each participants to fill in the questions regarding extraversion (or introversion) of the Big-Five Inventory (BFI) questionnaire (Benet-Martinez and John, 1998; John, Donahue and Kentle, 1991; John, Naumann and Soto, 2008), which asks respondents to estimate on a 5-point Likert scale (ranging from disagree strongly to agree strongly) how well their personality matches a series of descriptions (Mairesse et al., 2007). The BFI is a self-report inventory used to measure Big-Five traits. It contains short sentences with comparatively accessible vocabulary in 44 items in total. Here, 8 items in total questioned a person’s actual degree of extraversion (or introversion): I see myself as someone who… a) is talkative; b) is reserved; c) is full of energy; d) generates a lot of enthusiasm; e) tends to be quiet; f) has an assertive personality; g) is sometimes shy, inhibited; and h) is outgoing, sociable (Benet-Martinez and John, 1998; John et al., 1991; John et al., 2008). Notably, statement b), e) and g) were reverse-scored items which means that scoring high on those items indicated a low degree of extraversion or a high degree of introversion.

3.2.2.3 Aspirational Personality

(25)

(Benet-Martinez and John, 1998; John et al., 1991; John et al., 2008). All items were rated on a 5-point Likert scale.

3.2.2.4 Reviewer’s Personality - Manipulation of Message Style

Researchers found out that language use can reflect one’s personality (Boyd and Pennebaker, 2017; Pennebaker and King, 1999; Yarkoni, 2010). Thus, statements in text do not only contain semantic content but also convey a great amount of information about the sender such as the sender’s personality traits which can be considered as the most essential aspect of variation between humans (Mairesse et al., 2007). The way individuals talk and write can be seen as stamps of individual identity (Pennebaker and King, 1999). Therefore, in appendix II, a summary of identified language cues for extraversion is provided (Mairesse et al., 2007). Furthermore, in appendix III, words, phrases, and topics which have the strongest correlations to extraversion or introversion are displayed (Park, Schwartz, Eichstaedt, Kern, Kosinski, Stillwell, Ungar and Seligman, 2015). Those words, phrases, and topics were found by analyzing the written language from 66732 Facebook users. On the one hand, evidence for high extraversion is language reflecting positive emotion (e.g., love, :) ), enthusiasm (e.g., best, pumped), and sociability (e.g., party, hanging, dinner with). On the other end, evidence for introversion is language reflecting inward focus (e.g. I don’t, I should), relatively greater interest in things than in people (e.g., computer, chemistry), and tentativeness (e.g., probably, suppose, apparently) (Park et al., 2015).

Regarding the survey, after having indicated the actual and aspirational personalities, each participant was randomly assigned to either an extraverted book review or an introverted book review (between-subject design). One review was supposed to be written by an extraverted person and therefore, written in an extraverted way whereas the other was written by an introverted individual. The elaboration of both reviews was in style of Moon (2002) who puts extraversion with dominance on a level and introversion with submission. In his research, the extraverted or dominant review condition consisted of strong language containing assertions. On the contrary, the introverted or submissive review condition consisted of weaker language containing suggestions and questions.

(26)

regarding the style of presentation contained in the message claims. Still, as indicated in appendix II, the reviews demonstrated that introverts use many tentative words in contrary to extraverts (e.g. perhaps, maybe, could versus definitely, absolutely). Moreover, the message style was more formal regarding the introverted or submissive review condition than for the extraverted or dominant review condition (e.g. hello versus hi, cool). Beyond that, the extraverted review contained more positive words whereas the introverted review was predominantly written in a more negative style (e.g. “It’s easy to read” versus “It’s not too hard to read”). Furthermore, words, phrases, and topics which have the strongest correlations to extraversion or introversion, indicated in appendix III, were included (see underlined words in the reviews below).

In order to be sure that the texts were written in an extraverted and introverted style, and therefore by an extraverted (introverted) reviewer, the IBM Personality Insights service was used. IBM developed models to gather scores for Big-Five dimensions from written texts based on an open-vocabulary approach. The mentioned models rely on investigations in the fields of psychology, psycholinguistics, and marketing (IBM Watson Developer Cloud, 2018). Developing the Personality Insights service, IBM based itself on personality surveys to establish factual data obtained through direct observation for personality inference marketing (IBM Watson Developer Cloud, 2017). Regarding the book reviews in the present research, the extraverted review condition had a score of 92% in extraversion which stands for a highly extraverted person. The introverted review condition had a score of 0% in extraversion which stands for a highly introverted person (low degree of extraversion). The scores of the IBM Personality Insights service indicated that the reviews are very likely to be written by an extraverted (introverted) person.

Book reviews were chosen as books are considered to be a neutral object towards which individuals do not have specific preferences. Furthermore, this corresponds to the research by Nass and Lee (2001) in which participants heard a synthesized voice (extravert or introvert) about a book review and had to answer questionsregarding the book and review quality. Hereafter, the extraverted and introverted book reviews are presented.

Extraverted or dominant review condition:

(27)

extremely well and the wonderful story is certain to be of great interest to you. Truly, hanging out with my loved ones on the weekend and reading this extraordinary story makes my day! You should read it while chilling at the beach with your friends or family. You’ll absolutely enjoy reading it! :)

Introverted or submissive review condition:

Hello, perhaps you should buy this book. It appears to be written in a nice style. Apparently, it’s not too hard to read, which could make it a convenient book for summer holidays. The author describes the main characters pretty well and it’s possible that this story will be of interest to you. I’ve finished reading the book last week. Actually, I downloaded an e-version of the book in order to read the book on my tablet. The download won’t take much time. I don’t think that it’s a weird story. Maybe you will enjoy reading it? xd

3.2.2.5 Liking the Reviewer

After being confronted with one of the two book reviews, participants were asked to indicate if they like the reviewer or not. A lot of investigations on the relation of similarity to liking have made use of Byrne’s Interpersonal Judgement Scale (IJS) to measure the degree of liking (Nesler, Storr and Tedeschi, 1993). Therefore, participants rated liking the reviewer using a two-item measure: “I feel that I would probably like this reviewer.”; “I believe that I would like working with this reviewer.” (Byrne and Nelson, 1965). In the end, as indicated by Byrne and Nelson (1965), the two-item ratings were added up. Byrne and Nelson (1965) suggested measuring the items on a 7-point Likert scale but in order to not confuse the survey’s participants and in order to make the results more comparable, I decided to use a 5-point Likert scale.

3.2.2.6 Willingness to Buy

(28)

3.3 Plan of Analysis

Data analysis was performed using SPSS Statistics 23. First of all, the sample characteristics were explained by taking into account the control variables. Second, the(Pearson) correlation coefficient was calculated to test whether the items measuring actual personality, aspirational personality and liking the reviewer positively correlate. Third, regarding the items measuring the actual personality, aspirational personality and liking the reviewer, internal consistency was measured. As the dimensions were already known, internal consistency was measured by assessing reliability in order to know if variables measure the same latent variable (concerning the personality items, statement b), e) and g) are supposed to measure introversion and therefore needed to be reverse coded). If variables which are supposed to measure extraversion or liking the reviewer strongly and positively relate to one another, there is a substantial degree of internal consistency. Here, the Cronbach’s Alpha coefficient was used as it is one of the most popular measures of internal consistency (Mooi and Sarstedt, 2011). The Cronbach’s Alpha coefficient varies from 0 to 1, and in this study, the agreed lower limit for the coefficient is 0,7 which means that the Cronbach’s Alpha should be higher than 0,7 (Mooi and Sarstedt, 2011). Having a Cronbach’s Alpha > 0,7, items were summed up in order to build one average-variable (factor), measured on a Likert scale.

Moreover, the main analyses were conducted, testing the hypotheses. First, the relationship between the personality of the reviewer, liking the reviewer and willingness to buy was analyzed by doing a mediation analysis (hypothesis 1). Then, the moderated mediation analyses were performed taking into consideration hypothesis 2 and 3. In the following paragraphs, the methods used are discussed. Generally, a mediation analysis as well as a moderated mediation analysis are types of a regression analysis. Consequently, the linear regression analysis and the requirements that need to be fulfilled to conduct such an analysis will be explained first, as the requirements also hold for the mediation and moderated mediation analyses.

(29)

number of observations is at least 50 + 8 * k (k is the number of independent variables). As I have four independent variables (taking into account the mediator and moderators), a minimum of at least 82 respondents was required. Furthermore, collinearity should not be present among the independent variables. By calculating the VIF (Variance Inflation Factor), “a tolerance of below 0.10 indicates that (multi) collinearity is a problem” (Mooi and Sarstedt, 2011, p. 166). In addition, VIF scores greater than ten indicate collinearity issues. Simple mediation analysis: As mentioned, a mediation analysis is a regression analysis which helps to solve the question how X (independent variable) transmits its effect on Y (dependent variable) (Hayes, 2009; Hayes, 2013). Therefore, it is any causal system in which at least one causal antecedent X variable is assumed to influence an outcome Y through a single intervening variable M (Mediator) (Hayes, 2013). To prove a significant mediation effect (see figure 2), the bootstrapping method is used to consider the effect of a * b by using Hayes PROCESS Macro Model 4 for SPSS (Hayes, 2013). The non-parametric method bootstrapping is based on resampling with replacement and is gaining popularity (Hayes, 2009; Hayes, 2013; Hayes and Preacher, 2014).

Figure 2. Mediation Model.

(30)

seen in figure 3. The macro provides bootstrapping confidence intervals to test if the reviewer’s personality (X) leads though liking (M) to a higher willingness to buy (Y) when the actual personality of the review reader (W) matches the personality of the reviewer (X), and if this effect is even stronger when the reader’s actual personality (W) matches his or her aspirational personality (Z).

Figure 3. Model Templates for PROCESS for SPSS (Model 11).

Source: Hayes (2013)

4. Results

In this chapter, the results of this study will be presented. First, the sample characteristics will be discussed by making use of descriptive statistics. Then, the questions’ correlations and reliability will be checked by using the Pearson correlation index and Cronbach’s alpha. Finally, the proposed hypotheses will be tested by conducting, inter alia, a moderated mediation analysis.

4.1 Data Cleaning and Sample Characteristics

(31)

Regarding the country of origin, the majority of the sample consisted of 109 (54%) participants who came from Germany, 30 (14,9%) from Mexico, 9 (4,5%) from the U.S. and 9 (4,5%) from Brazil. Concerning the highest degree of school completed, the majority consisted of 85 (42,1%) participants who indicated having a Bachelor’s degree, and 72 (35,6%) respondents indicated having completed high school as their highest degree of school.

All in all, the goal of collecting a minimum of 150 answers of survey respondents was achieved and surpassed. Furthermore, with regard to Green (1991), meaningful regression analyses could be conducted as one could find a sufficiently large sample size (202 > 82).

4.2 Factor Building and Reliability of Variables

(32)

0,865. Here as well, Cronbach’s Alpha could not be considerably improved by deleting an item. Generally, all Cronbach’s Alpha values showed a high level of internal consistency and thus, those variables that also correlated significantly were summed up into a sum-variable and then divided by the amount of items to get a value within the 5-point Likert scale (=average). The 5-point Likert scale remained intact regarding the variables (factors) actual personality of the reader, aspirational personality of the reader and liking the reviewer. The mean and the standard deviation of all variables, but not the control variables or the independent variable, can be found in table 2.

Table 1. Measurement Instruments and Internal Reliability.

Construct Reference Number

of items

Scale Cronbach’s alpha Actual personality Benet-Martinez and John

(1998), John et al. (1991), John et al. (2008) 8 5-Point Likert scale 0,812

Aspirational Personality Benet-Martinez and John (1998), John et al. (1991), John et al. (2008) 8 5-Point Likert scale 0,719

Liking the reviewer (extraverted condition)

Byrne and Nelson (1965) 2 5-Point

Likert scale

0,756

Liking the reviewer (introverted condition)

Byrne and Nelson (1965) 2 5-Point

Likert scale

0,865

Table 2. Data Set Frequencies.

Variable Name Mean Standard Deviation

Actual personality 3,6238 0,69859

Talkativeness 3,98 0,985

Reverse-Coded: Being reserved 3,1634 1,14085

Full of energy 3,96 0,932

Enthusiasm 3,84 0,928

Reverse-Coded: Quiet 3,2574 1,23509

Assertive personality 3,81 0,954

Reverse-Coded: Shy, inhibited 3,0248 1,28302

Outgoing, sociable 3,96 0,982

Aspirational personality 4,2246 0,5177

Talkativeness 3,95 1,052

Reverse-Coded: Being reserved 3,6782 1,11544

Full of energy 4,64 0,72

Enthusiasm 4,56 0,711

Reverse-Coded: Quiet 3,8267 0,99485

(33)

Reverse-Coded: Shy, inhibited 4,1337 0,91801

Outgoing, sociable 4,43 0,833

Liking the reviewer (extraverted condition)

3,0347 0,99563

I feel that I would probably like this reviewer.

3,04 1,104

I believe that I would like working with this reviewer.

3,03 1,109

Willingness to buy (extraverted condition)

3,17 1,132

Liking the reviewer (introverted condition)

2,3614 0,99529

I feel that I would probably like this reviewer.

2,46 1,091

I believe that I would like working with this reviewer.

2,27 1,029

Willingness to buy (introverted condition)

2,46 1,145

Note: n = 202.

Generally, the independent variable (personality of the reviewer) displayed the degree of extraversion of the reviewer, as in the underlying study I have only focused on the personality dimension of extraversion. Therefore, to avoid misunderstandings, in the following, extraversion of the reviewer will be the name of the independent variable.

The degree of extraversion of the reviewer (independent variable) was presented by the two review conditions: 1 = High degree of extraversion of the reviewer, versus 0 = Low degree of extraversion (=introversion) of the reviewer. Hence, the independent variable was binary. The dependent variable, the reader’s willingness to buy, contained the values of both review conditions and was still measured on a 5-point Likert scale. Therefore, a regression analysis could be conducted as the dependent variable needs to be interval or ratio scaled (Allen and Seaman, 2007; Mooi and Sarstedt, 2011).

(34)

an overview over the data. By conducting the analyses, continuous variables for actual and aspirational personalities were used.

4.3 Influence of Control Variables

Analyzing whether or not the control variables as for instance gender of the respondents influence the willingness to buy (dependent variable), a regression analysis was performed with gender, country of origin, level of education, age and English level regressed on willingness to buy. The overall regression analysis was not significant, R2 = 0,021, F (1,201) = 0,837, p = 0,525. The control variables did not significantly influence the reader’s willingness to buy as the variables did not show a significant effect on the willingness to buy, p > 0,05 regarding all variables. Nevertheless, the control variables were included in the subsequent analyses in order to disqualify their effect on other variables.

4.4 Correlations of Independent Variables

Possible correlations between the independent variables were measured with the intention of assuring that multicollinearity is not an issue. The (Pearson) correlations are displayed in table 3. One could see some significant positive correlations between the independent variables but those were not too high. Between actual personality and aspirational personality, there was an overlap of 38% which was expected, and between reviewer’s personality and liking the reviewer, there was also an overlap of 32,2%. All in all, since the correlation indexes were in a reasonable rage and as expected, the analysis was continued as multicollinearity was not found to be an issue but still, the VIF scores were taken into account while performing the subsequent analyses.

Table 3. Correlations of Independent Variables.

Variable Name Reviewer’s personality Actual Personality Liking the Reviewer Aspirational Personality Reviewer’s personality 1 Actual Personality -0,39 1

Liking the Reviewer 0,322* 0,074 1

Aspirational Personality 0,075 0,38* 0,034 1

Note: n = 202.

(35)

After controlling the correlations between the independent variables, the effect of the independent variables on the dependent variable was examined by testing the aforementioned hypotheses.

4.5 Hypotheses Testing

With a view to the hypotheses and the model in general, the model will be divided into three smaller models to capture all possible effects. As I will start analyzing a mediation model, and subsequently add the moderator(s), I will not refocus on the results discussed in prior models as the results indicated similar effects. In the end, testing hypothesis 3, the complete model as depicted in figure 1 will be measured.

(36)

were computed for each of 5000 bootstrapped samples, and the 95% confidence interval was computed by determining the indirect effects at the 2,5th and 97,5th percentiles. The bootstrapped unstandardized indirect effect was 0,4861, and the 95% confidence interval ranged from LLCI = 0,2872 to ULCI = 0,7010. Thus, the indirect effect was statistically significant. This indicated that mediation has occurred. The mediator accounted for more than half of the total effect (a*b/c), PM = 0,6754. Control variables did not have any significant

effect on the included variables. Figure 4 displays the relationship graphically. In appendix IV, the complete results of the mediation analysis can be found.

Table 4. Mediation Results.

Relationship Unstandardized B t-stats p-value

c 0,7197 t (195) = 4,3541 p = 0,0000

a 0,6733 t (196) = 4,5397 p = 0,0000

b 0,7219 t (195) = 12,5052 p = 0,0000

c’ 0,2337 t (195) = 1,8253 p = 0,0695

Figure 4. Mediation Analysis (Model 4), Unstandardized B Coefficients.

Note: *Significant at p < 0.001.

All in all, the mediation analysis has shown that the relationship between reviewer’s extraversion and reader’s willingness to buy was positively mediated by liking the reviewer. In consequence, the null-hypothesis (=the relationship between the reviewer’s personality and review reader’s willingness to buy is not mediated by liking the reviewer) could be rejected. However, the conceptual model does not only describe a mediated relationship between the independent variable and the dependent variable, but also incorporates moderated effects on the independent variable and the mediator. By incorporating the moderators, the second and third hypotheses will be tested.

(37)

was the moderator and willingness to buy acted as dependent variable. The relationship between the variables is depicted in figure 5.

Figure 5. Moderated Mediation Analysis (Model 7), Unstandardized B Coefficients.

Again, the values were mean centered and the control variables were included as covariates. Overall, regarding the effects on liking the reviewer, 12% of the variance (R2 = 0,1199) was explained in the model and it was overall significant, F = 3,9402, p = 0,0005. With regard to the moderator, there was neither a significant effect of the reader’s actual personality on liking the reviewer (p = 0,2472; t(194) = 1,1607; unstandardized B = 0,1243) nor a significant interaction effect (p = 0,8478; t(194) = -0,1922; unstandardized B = -0,0416). Therefore, the moderator did not have a significant effect on liking the reviewer. With regard to the prior mediation model, similar effects were found between reviewer’s extraversion and liking the reviewer. All results of the moderated mediation analysis are displayed in appendix V.

Concerning the effects on the dependent variable (willingness to buy), the overall analysis was significant, R2 = 0,4654, F = 33,4685, p = 0,000. All results were similar to the prior mediation model, however, in view of the moderated mediation effect, no significant effect was found relying on the bootstrapping results, indicated through the index of moderated mediation, LLCI = -0,3425, ULCI = 0,2590. The control variables did not have any significant effect on the variables included, p > 0,05 with regard to all variables. To sum it up, it could not be confirmed that the reviewer’s personality leads to more liking and to a higher willingness to buy when the review reader matches the reviewer’s personality. Therefore, the null-hypothesis could not be rejected.

(38)

Figure 6. Moderated Mediation Analysis (Model 11), Unstandardized B Coefficients.

Again, this analysis was conducted by using the PROCESS macro of Hayes (model 11) (Hayes, 2018). All variables were mean centered and the control variables were included in the model as covariates to control their effect. The results were similar to the prior mediation (model 4) and moderated mediation analyses (model 7), and can be found in appendix VI. The interaction effects did not show any significant effects, p > 0,05. Furthermore, the main interest lied on the index of the moderated moderated mediation which was shown through the bootstrapping results where the 95% confidence interval ranged from LLCI = -0,6308 to ULCI = 0,6895. As zero was included in the bootstrapping results, it could not be confirmed that the reviewer’s personality only leads to more liking and to a higher willingness to buy when the review reader matches the reviewer’s personality, and it could also not be confirmed that this unsupported effect is stronger when the reader’s actual personality matches the aspirational personality. All in all, the null-hypothesis could not be rejected.

5. Discussion and Conclusion

(39)

(40)

reviewer’s personality and the reader’s willingness to buy, it can be denied that extraverted reviews with stronger assertions are the same as more positive reviews. Liking the reviewer should not be influenced by the valence of the text: When the reviewer’s identity is not revealed, there is no significant difference between positive and negative online reviews either

in terms of perceived credibility or consumer trust (Kusumasondjaja, Shanka and

Marchegiani, 2012), and since there is generally a positive relationship between the extent to which the person is trusted by others and the person's likability (Rotter, 1980), the valence of a text is not assumed to influence liking the reviewer, as the level of trust remains the same even if the valence of the text changes. Finally, individuals do not like the reviewer because of his or her positive writing style but because of his or her high degree of extraversion. Another finding of this research is that people desire to be extraverted instead of being introverted. Taking a closer look at the aspirational personality of the respondents, 99% would like to see themselves as extraverted. This is an interesting finding which can be explained by taking into account the desired personality concept of our society: Individuals beingsociable, talkative and enthusiastic (Costa and McCrae, 1992; Goldberg, 1990) are perceived as better, more successful and more admirable than persons characterized as being more reserved, shy and less self-confident. Various studies have confirmed this notion: In their meta-analytic study, Newcomb, Bukowski and Pattee (1993) found evidence that traits related to extraversion can be seen as significant predictors of social acceptance. Furthermore, extraverted kids were more often found to be selected as friends by other children than kids low on this trait (Jensen-Campbell Adams, Perry, Workman, Furdella and Egan, 2002). Another study by Anderson (1968) ascertains that individuals possessing extraverted traits like being cheerful, good-humored and happy are rated as being more likable.

(41)

Furthermore, a more recent study, found out that Blog readers were more attracted to blog writers with more similar personalities (Li and Chignell, 2010). In addition, a meta-analysis, taking into account 313 similarity studies, detected that similarity generates a positive, moderate effect on attraction (Montoya, Horton, and Kirchner, 2008).

One reason for not having found a significant effect for similarity in personality on the reader’s willingness to buy is that the reader did not find him- or herself in the position to sufficiently identify with the reviewer. The reader may not have had enough information about the reviewer. Adding a profile picture, providing the country of origin, name, and age etc. of the reviewer may have helped the reader to identify him- or herself better with the reviewer. Nevertheless, this would have changed the setting of this research. By adding personal information, the review reader would not have “only” relied on the personality of the reviewer but also on additional details. Certainly, it can be said that adding additional information about the reviewer is in line with the real world; review profiles do not only contain the reviewer’s text and accordingly his or her personality, but also reveal additional personal information. In response, Babić Rosario et al. (2016) stress in their meta-analytic review that eWOM on social media platforms is more effective when eWOM readers can associate with similar eWOM senders on the basis of homophily details such as username, profile picture, location and profile page.

Another explanation for not having found significant effects regarding the hypotheses containing similarity in personality is that the study was not made in a research laboratory context. Prior researchers conducting similar studies, performed their data collection in laboratory environments making the review reader focus entirely on the review itself which allowed significant results for the similarity in personality hypotheses to be found (e.g. Li and Chignell, 2010; Nass and Lee, 2001). In a laboratory setting, participants experience standardized environments and are not distracted by external influences. In this research, participants might not have used their full attention while taking part in the survey.

Master Thesis Marketing Management

The Effect of Personality Matching in Online

Reviews

Master Thesis Marketing Management

The Effect of Personality Matching in Online

Reviews

Master Thesis Marketing Management

Florentine Sterkel

June 2018

Management Summary

Abstract

Preface

Table of Contents

1. Introduction

2. Literature Review

3. Methodology

4. Results

5. Discussion and Conclusion

6. Limitations and Further Research