Personality traits and emotions: why do we write
reviews on the Internet?
Do personality traits and different negative emotions, formed after a consumer experience, have an effect on the different motives to engage in electronic word-of-mouth?
Nicole Tuinsma Student ID: 10658505
Final draft submitted on: January 27, 2015
MSc. in Business Administration – Marketing Track University of Amsterdam
2 Statement of originality
Statement of originality
This document is written by Student Nicole Tuinsma who declares to take full
responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and
that no sources other than those mentioned in the text and its references have
been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision
of completion of the work, not for the contents.
3 Table of contents
1. Introduction 6
2. Theoretical framework 7
2.1. From word-of-mouth to electronic word-of-mouth 7
2.2. Consumer experience, Emotions and Motives 8
2.2.1. Engaging in eWOM 10 2.2.2. Motives eWOM 13 2.3. Personality traits 15 2.3.1. Agreeableness 16 2.3.2. Extraversion 18 2.3.3. Neuroticism 19
2.4. Negative emotions: anger versus sadness 20
2.5. Hypotheses overview 23
3. Methodology 24
3.1. Survey design 24
3.1.1. Personality traits 25
3.1.2. Emotions induction 26
3.1.3. Motives to engage in eWOM 27
3.2. Sample 27
3.3. Statistical procedure 29
4. Results 31
4.1. Reliability 31
4.2. Descriptives and normality 37
4 4.4. Direct effect 39 4.5. Moderation 41 4.6. Hypotheses overview 43 5. Discussion 44 5.1. Key findings 44 5.2. Managerial implications 49
5.3. Limitations and future research 50
6. Conclusion 52
References 54
Appendix 1 61
Appendix 2 64
5 Abstract
Since the raise of the World Wide Web, consumers started sharing experiences and thoughts on the Internet. To better understand the motives of why people write about these experiences on the Internet, a closer look at the factors that influence the different motives should be taken. The purpose of this study is to investigate the influence of personality traits and some specific negative emotions (anger and sadness) on the different reasons people have when engaging in electronic word-of-mouth (eWOM). Multivariate general linear model analyses were used to get a better insight in these influences. The results of this study show that there is a direct effect between agreeableness and the motive helping others, between extraversion and the motives self-enhancement and sense of belonging, and between neuroticism and the motive sense of belonging. However, this is not enough evidence to say that personality traits do have a significant influence on the different motives. The results of this study also show us that sadness and anger do not have a moderating effect on the relationship between personality traits and the motives to engage in eWOM. Moreover, in this study there is not enough evidence found to support all hypotheses. Personality traits and specific negative emotions (anger and sadness) cannot be seen as main influencers. Though, managers should keep in mind that all individuals are different, in a way that they have different personality traits, and that they could have different motives to engage in eWOM because of this.
6 1. Introduction
In the past couple of years, communicating via the Internet (social media, such as Facebook and Twitter) has become a very big trend. Social media enabled people to communicate with other people all around the world, without leaving their house(s). Hence, this trend caused a shift in marketing. Where marketers before were in control of everything that was brought into the market, after the introduction of social media the consumers took over control. Hennig-Thurau, Hofacker and Blocking (2013) compare the shift in marketing to the games bowling and pinball. Before the Internet influenced marketing, marketing could be seen as a game of bowling. Bowlers (Marketers) would throw a bowling ball (chose specific marketing instruments) on the bowling alley (via mass media) to hit as many as pins (customers) as possible. Though, since the rise of social media, marketers have lost control over their own marketing. Marketing can now be compared to a pinball game. Marketers still control the ball (marketing instruments), but there are some difficulties to keep the ball in control. There are bumpers (social media and online reviews) to overcome and keep the ball as long as possible in the game. The only way to keep control over the ball is to use the flippers (Hennig-Thurau et al., 2013). To control the ball in the pinball game, marketers should be aware of the various reasons why people decide to communicate via social media.
So why do we communicate via social media? Imagine you are going to buy a new iPhone. In the shop you experience something very negative; the salesman is very rude and he does not listen to your wishes. To share your experience with other people, you are going to write a review on the Internet. But, why? Do you write reviews because you want to help other consumers by making a decision or do you want to warn them for that specific brand? Or do you want to take revenge, because the sellers treated you very bad? Consumers could have a great number of different reasons of why they want to communicate via social media.
7
These reasons could also be influenced by different factors. Age and gender could have an influence, but also brand loyalty or the buying environment. Furthermore, the reasons of why people communicate via social media could vary between consumers because of their values and preferences, which are often reflected in personality traits (Chen, 2008). Personality traits can be defined as certain characteristics of individuals that stay the same over their lifespan (Hamel, Shaffer & Erdberg, 2000). This current research looks into the influence of personality traits and the following research question will be answered: Do personality traits
and different negative emotions (anger and sadness), formed after a consumer experience, have an effect on the different motives to engage in electronic word-of-mouth?
To answer this research question, this study will start with a theoretical framework in which all fundamental concepts will be discussed; of-mouth (WOM), electronic word-of-mouth (eWOM), motives to engage in WOM and eWOM, and personality traits. The theoretical framework will be followed by the methodology of the study and the eventual results. The results will be explained and discussed, and an overview will be given. In the end a discussion, the managerial implication and direction for future research will be given. This study finished with an overall conclusion.
2. Theoretical framework
2.1 From word-of-mouth to electronic word-of-mouth
For a great number of years, companies (especially their marketers) had control over what happened in the consumer market, explained as the game of bowling. The marketers had all the power, because they were the ones to decide what kind of information to bring to the market. Consumers had only to belief the marketers or could only discuss certain companies face-to-face with other people, what is called mouth. Arndt (1967) defined
word-of-8
mouth (WOM) as “oral, person-to-person communication between a perceived non-commercial communicator and a receiver concerning a brand, a product, or a service offered for sale” (p. 190). Thus, when consumers share their personal experiences with a certain brand, product or service with other consumers, WOM takes place (Brown, Barry, Dacin & Gunst, 2005). Since consumers were only able to reach a small part of the population via word-of-mouth, marketers were the ones who spread a message and, therefore, were in power. Though, since the raise of the World Wide Web, companies and their marketers lost their power and the game of bowling changed into a game of pinball (Hennig-Thurau et al., 2013). The Internet provided consumers with platforms, which were easily accessible for everyone. Consumers were now able to communicate with one another. The opportunity for consumers to share their thoughts and assessing their thoughts from others changed the market dramatically, such that the consumers were now the once with all the power. The phenomena of exchange via the Internet is called electronic word-of-mouth. Hennig-Thurau, Gwinner, Walsh and Gremler (2004) explained electronic word-of-mouth (eWOM) communication 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). The World Wide Web is not just a platform providing access to information, but it is also being thought of as playing the role of a social technology providing its users with platforms to communicate and taking over the market power (Hamburger & Ben-Artzi, 2000).
2.2. Consumer experience, Emotions and Motives
To understand and to influence this eWOM communication, researchers have investigated what happens prior to eWOM – the emotions that are formed when experiencing
9
a certain situation – and what the goals of eWOM are – the motives to engage in eWOM that are created by these emotions. When consumers experience certain situations, either positive or negative, the experiences can be elicited into certain emotions (such as anger, happiness, sadness, or fear) towards a specific company (Zeelenberg & Pieters, 2004). Hence, these emotions drive consumers to share their experiences; which is called word-of-mouth (WOM) (Verhagen, Nauta & Feldberg, 2013). Therefore, WOM is a consequence of the emotions that are driven by experiences (Derbaix & Vanhamme, 2003; Söderlund & Rosengren, 2007). Consumers want to communicate their experiences and their emotions openly with other people, because it is a way to get help and support, to get a closer bond, to get attention and to arouse empathy (Rimé, Philippot, Boca & Mesquita, 1992; Rimé, 2009). Other reasons for consumers to share their experiences could be to draw attention to the cause of their dissatisfaction/satisfaction, prevent/ensure others for experiencing the same, and encourage a company to improve (Thøgersen, Juhl & Poulsen, 2009; Litvin, Goldsmith & Pana, 2008; Zaugg & Jäggi, 2006). In previous research about word-of-mouth it was found that negative emotions (such as anger and sadness), as a results of a negative experience, contributed to the probability that consumers engage in negative WOM (Nyer, 1997). Also, evolving negative emotions during an experience are directly linked to engaging in negative WOM (Zeelenberg & Pieters, 2004). On the contrary, evolving positive emotions during an experience is expected to reduce the chance of consumers to engage in negative WOM (Nyer, 1997).
To see if this also counts for online WOM, Verhagen et al. (2013) focused on electronic word-of-mouth. Indeed, they found that when consumers have negative emotions, because they experienced something negative, they express themselves negatively on the Internet. When experiencing positive emotions, negative eWOM is negatively influenced. It is also found that when consumers express themselves negatively on the Internet about a specific product or service, and their motive is not only to express their feelings but also to
10
help others, the chance of them switching to another company will be more likely (Verhagen et al., 2013). It is indicated that negative eWOM could have a very strong effect on consumer behaviour. This drives companies to make use of webcare teams. The aim of these teams is to reduce the change that negative opinions will be spread via the Internet (Van Noort & Willemsen, 2012).
Knowing that consumers translate their emotions into eWOM, one should know why people engage in eWOM by looking at different motives. Even though eWOM and WOM differ in many ways – speed, scalability, accessibility, persistence and measurability – (Cheung & Lee, 2012), the concept of eWOM communication is very close to that of the traditional WOM communication; the motives that are identified as relevant for traditional WOM can also be expected to be relevant for eWOM (Hennig-Thurau et al., 2004). Table 1 list the motives for WOM communication suggested by the most prominent literature that have been a basis for subsequent studies.
2.2.1. Engaging in eWOM
A prominent study about eWOM communication is by Hennig-Thurau et al. (2004). They started a research with the purpose to find out what motives consumers have to articulate themselves on the Internet, and thus engage in eWOM. In their research they focus on eWOM via Web-based consumer-opinion platforms for several reasons; these platforms are the most widely used and consumers are stronger influenced by these platforms than by other means. Hennig-Thurau et al. (2004) build upon a framework of social interaction utility introduced by Balasubramanian and Mahajan (2001) and use this framework to identify the motives that consumers have to engage in eWOM via Web-based consumer-opinion platforms.
11
Table 1: Motives to engage in word-of-mouth
Dichter (1966) Engel, Blackwell
and Miniard (1993)
Sundaram, Mitra and Webster (1998)
Description
Other-involvement Concern for others Altruism (positive WOM)
The need to give something to the receiver; a genuine desire to help a friend or relative make a better purchase decision
Product-involvement
Involvement Product involvement
Level of interest or involvement in the topic under consideration serves to stimulate discussion; personal interest in the product, excitement resulting from product ownership and product use
Self-involvement Self-enhancement Self-enhancement Enhancing images among other consumers by projecting themselves as intelligent shoppers; recommendations allow person to gain attention, suggest status, give the impression of possessing inside information, and assert superiority Helping the
company
Desire to help the company
Anxiety reduction Easing anger, anxiety, and frustration
Vengeance To retaliate against the company associated with a negative consumption experience
Advice seeking Obtaining advice on how to resolve problems
Message-involvement
Message intrigue Entertainment resulting from talking about certain ads or selling appeals; Refers to discussion which is stimulated by advertisements, commercials, or public relations
The three types of utility are: focus-related utility, consumption utility and approval utility.
-‐ Focus-related utility can be described as the utility consumers receive when adding value to the community through his or her contributions (Balasubramanian & Mahajan, 2001). In the traditional WOM communication the motives helping the
12
company, social benefits, exerting power and concern for other consumers can be
related to focus-related utility (Hennig-Thurau et al., 2004).
-‐ With consumption-utility one means the value consumers obtain “through direct consumption of the contribution of other community constituents” (Balasubramanian & Mahajan, 2001, p. 125). In WOM/eWOM communication context, individuals consume when they read reviews and comments about a product or service written by others on the Internet, which could be a motivation for those individuals to write a review or comment as well. Sharing information in the Internet about a product or service can be seen as post purchase advice seeking (Hennig-Thurau et al., 2004). -‐ Approval utility is concerned with a consumer’s satisfaction that comes “when other
constituents consume and approve of the constituent’s own contributions” (Balasubramanian & Mahajan, 2001, p. 126). When an individual writes about a specific product or service on the Internet and others praise this individual (either informal via comments or formal via contribution rankings), individuals feel approved. In the traditional WOM communication the motives self-enhancement and
economic reward are related to approval-utility (Hennig-Thurau et al., 2004; Dichter,
1966; Engel, Blackwell & Miniard, 1993).
Additional to these three utilities, Hennig-Thurau et al. (2004) include two other utilities, namely moderator-related utility and homeostase utility. Moderator-related utility happens when there is a third party that makes the complaint act easier for members of the community/platform. Convenience and problem-solving support are specific eWOM communication motives that work as a moderator. The last utility, homeostase utility, refers to the idea that people have the basic desire to have balanced lives (Zajonc, 1971). With this utility, WOM communicating motives are expressing positive emotions and venting negative
13
and the motives to engage in eWOM communication are building upon the study done by Hennig-Thurau et al. (2004), some studies find other motives and different outcomes. Table 2 gives an overview about the motivations individuals could have for engaging in eWOM from the point of view of several studies.
2.2.2. Motives eWOM
A great number of motives why people engage in eWOM have been found. This study will build upon previous research and will explore several motives for eWOM communication: the motive of self-enhancement, helping others, sense of belonging, and reciprocity. When individuals engage in eWOM because they want to gain attention, feel like a pioneer, gain positive recognition on his or her judgement by others or when they just want to gain status, the motive is self-enhancement (Dichter, 1966; Hennig-Thurau et al., 2004). Self-enhancement is a motive for eWOM when individuals want to increase their own welfare (Cheung & Lee, 2012) and when they want to enhance their image by presenting themselves as intelligent shoppers (Hennig-Thurau et al., 2004). Helping others, often also referred to as concern for others, relates to the need individuals have to help others (Dichter, 1966) and to share experiences with others to help them make better decisions (Engel, Blackwell & Miniard, 1993). Individuals that want to serve the public good by sharing experiences on the Internet, engage in eWOM because they want to help others (Cheung & Lee, 2012; Yoo & Gretzel, 2007). Individuals’ need for a sense of belonging means that an individual wants to be part of a group or community (Christodoulides, Jevons and Bonhomme, 2012). When individuals see themselves as one with the community and they participate actively, they are willing to do something beneficial for that community, even though others do not necessarily do that (Hennig-Thurau, 2004; Hars & Ou, 2012). Reciprocity means doing something for
14 D es cr ipt ion: Indi vi dual s wr it e r ev ie ws , be caus e … The y be li eve tha t w ri ti ng vi a a pl at form w il l be m ore c onve ni ent tha n doi ng i t a not he r w ay T he y w ant to ge t s om e a nge r of f t he ir c he st T he y w ant to he lp ot he r i ndi vi dua ls T he y w ant to s how ot he r i ndi vi dua ls tha t t he y a re int el li ge nt c ons um ers T he y w ant to fe el be longi ngne ss ; t he y w ant to be pa rt of a group T he y ge t re w ards for w ri ti ng t he m T he y w ere ve ry c ont ent w it h t he c om pa ny a nd w ant to prom ot e i t T he y hope ot he r pe opl e c an gi ve the m s om e advi ce /m ore inform at ion a bout the c om pa ny or produc t Re ci proc it y N ega ti ve re ci proc it y: re ve nge W ha t ot he rs di d for t he m (pos it ive )/ to m e (ne ga ti ve ), t he y w il l do ba ck Co-c re at ion T he y s ee the m se lve s a s s om eone w ho a dds va lue H enni g-T hur au e t al . (2004) Che ung & L ee (2012) Yap, Soe tar to & Swe ene y (2013) G re tz el & Y oo (2007) H o & D em ps ey (2010) Chr is todoul ide s e t al . (2012) B ronne r & de H oog (201 1) P la tform a ss is ta nc e V ent ing ne ga ti ve fe el ings V ent ing ne ga ti ve fe el ings V ent ing ne ga ti ve fe el ings P ers ona l/ ve nt ing ne ga ti ve fe el ings Conc ern for ot he rs E nj oym ent of he lpi ng ot he rs /c ol le ct ivi sm H el p ot he r c ons um ers Conc ern for ot he rs cons um ers N ee d t o be al trui st ic /pe rs ona l grow th S oc ia l c onc ern S el f-e nha nc em ent Re put at ion S el f-e nha nc em ent S el f-e nha nc em ent N ee d t o be di ffe re nt S el f-c onc ept P ers ona l/ S el f-enha nc em ent S oc ia l be ne fi ts S ens e of be longi ng S oc ia l be ne fi ts S oc ia l be ne fi ts N ee d t o be pa rt of a group Com m uni ty S oc ia l/ group com m it m ent E conom ic inc ent ive s E conom ic inc ent ive s Tabl e 2: Mot iv at ions for e ngagi ng i n e le ct roni c wor d-of -m out h H el pi ng t he c om pa ny A dvi ce s ee ki ng A dvi ce s ee ki ng E conom ic inc ent ive s H el pi ng t he c om pa ny H el pi ng t he c om pa ny H el pi ng t he c om pa ny
15
somebody, either negative or positive, with the idea that somebody would do the same back for you (Cheung & Lee, 2012). Individuals that share experiences on the Internet because they expect that other individuals will share their experiences in return in the future will engage in eWOM because of the motive reciprocity.
2.3. Personality traits
In the previous paragraphs it is explained that consumers could perceive certain situations in a different way than other consumers. When experiencing a situation, specific emotions are created (Zeelenberg & Pieters, 2004). These emotions can form different motives to engage in electronic word-of-mouth (Verhagen et al., 2013). A great number of researches have already written about that relationship, but not much is written about factors that can influence that relationship. The motives to engage in electronic word-of-mouth (sense of belonging, reciprocity, helping others, and self-enhancement) could vary between consumers because of people’s different values and preferences, which are often reflected in personality traits (Chen, 2008). Personality traits can be defined as certain characteristics of individuals that stay the same over their lifespan (Hamel, Shaffer and Erdberg, 2000). Famous personality traits are “The Big Five”, which are agreeableness, conscientiousness, extraversion, neuroticism, and openness to experience. Because not al five of them will have a significant influence on the motives why people engage in eWOM, this research will only focus on agreeableness, extraversion and neuroticism. The traits openness to experience and conscientiousness are not further used in this research, because not much evidence was found in existing literature that showed that these traits could have a significant influence on the motives why people engage in eWOM. Table 3 summarises the different definitions of the used personality traits from different authors.
16 2.3.1. Agreeableness
Agreeableness refers most of the time to individuals that are good-natured, tolerant, flexible and respectful (Yoo & Gretzel, 2011). Because people with high agreeableness are orientated towards others, their sense of belongingness is very important and they may choose the Internet as a platform to fulfil those needs (Seidmann, 2013). When comparing people
Table 3: Definitions of the different personality traits
Deng et al. (2013) Barrick et al. (2001) Zhang (2005) Judge et al. (1999)
Agreeableness The tendency of being
kind, considerate, likable, helpful, cooperative, and forgiving Can be described as cooperation, trustfulness, compliance and affability
People tend to be tolerant, trusting, accepting, and they value and respect other people’s beliefs and conventions
People that are
cooperative (trusting of others and caring) as well as likeable (good-natured, cheerful, and gentle)
Extraversion The tendency to be
social, active, and outgoing, and place a high value on close and warm interpersonal relationships Can be described as sociability, dominance, ambition, positive emotionality and excitements-seeking People tend to be sociable and assertive, and they prefer to work with other people
People tend to be socially oriented (outgoing and
gregarious), but also are surgent (dominant and ambitious) and active (adventuresome and assertive)
Neuroticism The tendency to be
anxious, self-conscious, and paranoid; it exhibits a lack of psychological adjustment and emotional stability Can be described as having anxiety, hostility, depression and personal insecurity People tend to experience such negative feelings as emotional instability, embarrassment, guilt, pessimism, and low self-esteem
Refers generally to a lack of positive psychological adjustment and
17
with high and low agreeableness, people with high agreeableness are more likely to be cooperative, supportive, caring and concerned about others (Zhai, Willis, O’Shea, Zhai & Yang, 2013). They prefer having a cooperative relationship instead of having a competitive relationship (Thompson, 2008). Hence, it can be said that agreeable people are more considered and more concerned about the well being of others (Wang, Jackson, Zhang, & Su, 2012). Furthermore, people with high agreeableness are also known of having a very trusting and helpful nature (Zhang, 2005). When people with high agreeableness decide to engage in eWOM, it is very likely that they do this because they want to help others. When taking these together with the fact that people with high agreeableness are concerned about others (Zhai et al., 2013) and feel the need to belong to a group (Den Hartog, De Hoogh & Keegan, 2007; Seidmann, 2013), one could argue that sense of belongingness plays a role as well. Individuals have a fundamental need for belongingness: they want to belong to a social group (Baumeister & Leary, 1995). In the past, agreeableness has also been positively linked to reciprocity (Buss, 1991; Ashton, Jackson, Helmes & Paunonen, 1998; Heineck & Silke, 2008; Graziano & Tobin, 2009). On the contrary, people who score low on agreeableness place self-interest higher than getting along with others (Graziano, Habashi, Sheese & Tobin, 2007). Engaging in eWOM could, therefore, be part of self-enhancement. They do not share their experience on the Internet to help others, but to enhance their self-esteem. They share because they feel good about themselves when telling their experiences to others (Hennig-Thurau et al., 2004). Taking these facts into consideration, the first hypothesis of this study is defined as:
Hypothesis 1: People who are agreeable do engage in eWOM because of the motives (a) helping others, (b) sense of belonging and (c) reciprocity and do not engage in eWOM because of the motive (d) self-enhancement.
18 2.3.2. Extraversion
People who are extrovert are social, talkative, assertive, and gregarious (Zhai et al., 2013). In contrast to introvert people, extrovert people enjoy being around others and are very chatty. Hence, they enjoy expressing themselves to others. One way to express your emotions (thoughts and feelings) and communicate with others is via eWOM. Yoo and Gretzel (2011) investigated if extraversion is related to self-enhancement. They found that extravert people are more motivated to create user generated content because of the energy they get out of communicating in public. Furthermore, extravert people seek to have personal contact with others, show them attention and be there for someone who needs help. They are socially oriented and tend to be gregarious (Judge, Higgins, Thoresen & Barrick, 1999). Gregarious means that people enjoy the idea of belongingness to a herd (Judge et al, 1999). These people want to feel a sense of belonging to a group. Engaging in eWOM, for example interacting with people via a forum or network, could bring that sense of belonging. Gaining personal benefits and a sense of belonging are not the only things extravert people tend to do; helping others is one as well. People, whom are extrovert, seek to have personal contact with others, show them attention and be there for someone who needs help (Seidmann, 2013). Therefore, hypothesis two is defined as:
Hypothesis 2: People who are extrovert do engage in eWOM because of the motives (a) helping others, (b) self-enhancement and (c) sense of belonging.
19 2.3.3. Neuroticism
Neuroticism reflects the extent to which people can be moody, insecure, self-conscious or anxious (Zhai et al., 2013). The opposite of people with neuroticism are people that are emotional stable. People that have a high level of neuroticism are very instable and often face depressions (Judge et al., 1999). Another characteristic of neuroticism is that people tend to be impulsive. Suls, Martin and David (1998) find that people with neuroticism especially are affected by negative life events. When neurotic people experience something negative (for example a negative interaction with a sales person), they can become moody, insecure, fearful and nervous (Ashton et al., 1998). These negative emotions discourage high levels of positive reciprocal behaviour, because they feel anger and resentment towards that individual (Ashton et al., 1998). Furthermore, neurotic people do not tend to have any care for other people, since dealing with their own problems takes a lot of energy already. Yoo and Gretzel (2011) find a negative influence between neuroticism and the concerns for others. Because neurotic people have an anxious nature and are very insecure, they are not able to help other people. And because of the anxiety and insecurity, they tend to seek help by others (Yoo & Gretzel, 2011). Other people could help them in certain situations to make them feel a bit more secure or just feel appreciated. The feeling of belonging helps neurotic people to become more stable. One way to feel you are part of a certain group is joining a forum on the Internet. People can share their thoughts and feelings, sometimes even anonymous. Hypothesis three of this study is defined as:
Hypothesis 3: People who are neurotic do not engage in eWOM because of the motives (a) helping others, (b) self-enhancement and (c) reciprocity and do engage in eWOM because of the motive (d) sense of belonging.
20 2.4. Negative emotions: anger versus sadness
The behaviour of people that respond to a certain situation will depend on the emotions they are feeling. Individuals that experience anger will most likely feel a sense of restlessness (Shields, 1984) and want to attack the cause of the anger (Mohammadi, Shafieyan, Azizi & Rahimi, 2014). Sad individuals, on the contrary, often turn towards the self and find themselves refuging from the situation (Mohammadi et al., 2014). When looking at the social transmission (sharing of content), emotions vary based on physiological arousal or activation (Smith & Ellswordt, 1985; Berger & Milkman, 2010). Even though anger and sadness both have the same valence (both negative emotions), they both create a different reaction. Where sadness is characterized by low arousal or deactivation, anger is characterized by heightened activation and high arousal (Barrett & Russel, 1998). Berger and Milkman (2010) suggest that differences in arousal (or activation) will play a significant role in social transmission. Because sharing information does require action, activation should have the same effects on social transmission; activation boosts the likelihood of consumers to engage in other action, so it should also boost the likelihood to share content with others (Berger and Milkman, 2010). It can be said that even though anger and sadness are both negative, the degree of transmission might be different. While anger increases the transmission (because its characteristic high activation), sadness decreases transmission (because its characteristic deactivation).
Personality traits play a role in the different responses coming from anger and sadness. Individuals that have higher levels of agreeableness generally report lower levels of anger (Watson, 2000), because they are more concerned with creating a positive and favourable impression to the outside world (Paulhus & John, 1998). Pearman, Andreoletti and Isaacowitz (2010) found that the more agreeable individuals are, the more they react to sadness.
21
Therefore, it is expected that individuals that are agreeable will react more to a sad situation than to an angry situation. In contrast with individuals with agreeable individuals, individuals that are neurotic are more likely to react to anger (Watson & Clark, 1984). Neurotic individuals feel easily irritated, annoyed and contempt (Watson, 2000), which makes them more reactive to angry making events (Bolger & Schilling, 1991). Therefore, it might be expected that individuals that have a personality that is more neurotic will react more to an angry situation than to a sad situation. When looking at extraversion, only little evidence is found about individuals responding to a sad or angry situation. Extravert individuals experience less negative affect when watching sad/angry films (Lischetzke & Eid, 2006) or when thinking about themselves in sad/angry situations (Larsen & Ketelaar, 1991), than other individuals (Steward, Ebmeier & Deary, 2005). In this study it is expected that extravert individuals do not react different in sad or angry situations and, therefore, no hypothesis is proposed. According to the previous, the following hypotheses are defined as:
Hypothesis 4: The evoked emotion moderates the relationship between agreeableness and (a) helping others, (b) sense of belonging, (c) reciprocity, and (d) self-enhancement, in a way that under a sad emotion the effect of agreeableness is higher.
Hypothesis 5: The evoked emotion moderates the relationship between neuroticism and (a) helping others, (b) sense of belonging, (c) reciprocity, and (d) self-enhancement, in a way that under an angry emotion the effect of neuroticism is higher.
22
23 2.5. Hypotheses overview
Table 4: Hypotheses overview
H1a People who are agreeable do engage in eWOM because of the motive helping others.
H1b People who are agreeable do engage in eWOM because of the motive sense of
belonging.
H1c People who are agreeable do engage in eWOM because of the motive reciprocity.
H1d People who are agreeable do not engage in eWOM because of the self-enhancement.
H2a People who are extrovert do engage in eWOM because of the motive helping others.
H2b People who are extrovert do engage in eWOM because of the motive self-enhancement.
H2c People who are extrovert do engage in eWOM because of the motive sense of belonging.
H3a People who are neurotic do not engage in eWOM because of the motive helping others.
H3b People who are neurotic do not engage in eWOM because of the motive
self-enhancement.
H3c People who are neurotic do not engage in eWOM because of the motive reciprocity.
H3d People who are neurotic do engage in eWOM because of the motive sense of belonging.
H4a The evoked emotion moderates the relationship between agreeableness and helping
others, in a way that under a sad emotion the effect of agreeableness is higher.
H4b The evoked emotion moderates the relationship between agreeableness and sense of
belonging, in a way that under a sad emotion the effect of agreeableness is higher.
H4c The evoked emotion moderates the relationship between agreeableness and reciprocity,
in a way that under a sad emotion the effect of agreeableness is higher.
H4d The evoked emotion moderates the relationship between agreeableness and
self-enhancement, in a way that under a sad emotion the effect of agreeableness is higher.
H5a The evoked emotion moderates the relationship between neuroticism and helping others,
in a way that under an angry emotion the effect of neuroticism is higher.
H5b The evoked emotion moderates the relationship between neuroticism and
self-enhancement, in a way that under an angry emotion the effect of neuroticism is higher.
H5c The evoked emotion moderates the relationship between neuroticism and reciprocity, in
a way that under an angry emotion the effect of neuroticism is higher.
H5d The evoked emotion moderates the relationship between neuroticism and sense of
24 3. Methodology
The required data for this research was obtained in the form of an online survey. The questionnaire has been distributed only in Dutch versions, because most likely most participants who would fill in the questionnaire would be Dutch. On October 31, 2014, a questionnaire has been send out via social media (Facebook and Twitter), via e-mail (friends, colleagues and acquaintances) and face-to-face with an iPad to fill in the questionnaire. Hence, people have been asked to forward the questionnaire to expand the range of participants. Four weeks later, on November 28, 2014, the survey was closed. This survey was distributed to obtain information about nine different variables. The independent variables are the three different personality traits: agreeableness, neuroticism, and extroversion. The dependent variables are the motives why people engage in eWOM: self-enhancement, helping others, reciprocity, and sense of belonging. The last variables are the moderators: anger and sadness. The following paragraphs will explain the survey design, the sample and the statistical procedure.
3.1. Survey Design
The questionnaire was set up to investigate three different objectives: personality traits, emotion induction, and motives to engage in eWOM. First the participants were asked questions to find out their different personality traits. Afterwards the participants were manipulated into two different groups to create different emotions. In the end, different questions were asked to find out more about the reasons why people engage in eWOM. An overview of the questionnaire can be found in appendix 1.
25 3.1.1. Personality traits
There are a great number of instruments (by means that they contain different amount of questions) to measure the different personality traits. Goldberg (1992) came up with a 100-item scale, Costa and McCrae (1992) used a 60-items scale, John, Donahue and Kente (1991) used the 44-item scale (big five inventory) and later the Ten Item Personality Inventory (TIPI) was introduced by Gosling, Rentfrow and Swann (2003). In this study the first reduced instrument, the big five inventory (BFI), was used to find out the different personality traits of the participants (John & Strivastava, 1991; John et al., 1991). The BFI is a 44-items questionnaire that measures the different dimensions of the big five personality traits (John et al., 1991; Goldberg, 1993). Even though time is limited for participants and questionnaires should not be too long (Herzog & Bachman, 1981), the BFI was chosen over TIPI. Because this study only focuses on agreeableness, neuroticism and extroversion, (and thus only questions about these traits were asked) I wanted to have a very clear measurement for the different traits per person. Furthermore, asking more questions would give more reliability and validity (John et al., 1991). The items are measured using a five-point Likert scale (Likert, 1932) ranging from 1, disagree strongly, to 5, agree strongly (see Appendix 1.1).
3.1.2. Emotion Induction
Emotion induction was used to create different emotions under the different participant. The emotion induction was done via scenarios written in the second person, which were either to create sadness or anger. Kahneman and Tversky (1982: in Huss, 1988) define a good scenario as “one that bridges the gap between the initial state and the target event by a series of intermediate events... A scenario is especially satisfying when the path
26
leading from the initial to the terminal state is not immediately apparent, so that the intermediate stages actually raise the subjective probability of the target event” (p. 35). Therefore, the scenarios in this study were divided into five small paragraphs/stages, each intermediate paragraph/stage making the scenario more intense and real (Keltner, Ellsworth & Edwards, 1993). The first three paragraphs were the same for both anger and sadness. These told a story about travellers that were ready to go on their honeymoon, but for whom the flight got cancelled. The last two paragraphs told a different story. In the sadness scenario the couple explained the sad story behind cancelling the plane and why this was so painful for the travellers. In de anger scenario you experience how bad the couple got treated by the ground personnel and how angry they became. To make sure the participants really tried to imagine they were the travellers, they were asked - before reading - to try to experience the event as vividly as possible (Keltner et al., 1993) (see Appendix 1.2). All participants were allocated to one of the two scenarios via randomization. According to Taylor (2006) randomization “gives the factorial survey the robustness of an experimental method” (Taylor, 2006, p. 1195). In this study randomization was created by the randomization function in Qualtrics.
In previous studies it has been shown that when participants rate their emotions after the induction, the induction will have a smaller effect because the participants will start thinking about their emotions (Keltner, Locke & Audrain, 1993). To overcome this problem, in this study the emotion inductions were pretested. The two scenarios and some emotion-related questions were send out to 20 people; 10 receiving the sad scenario and 10 receiving the anger scenario (see appendix 2). To analyse the results, two independent t-tests were performed. Where the first one compared the means of the sad emotions in both scenarios, the second one compared the means of the anger emotions in both scenarios. The results of this pre-test were significant (p < 0,05). Participants receiving the anger scenario felt a strong degree of anger, while the participants receiving the sad scenario felt a strong degree of
27
sadness. However, the participants receiving the sad scenario also felt angry emotions. Though, these angry emotions were of less degree than the sad emotions.
3.1.3. Motives to engage in eWOM
A great number of motives of why people engage in eWOM have been found. This study builds upon previous research and explores several motives for eWOM communication: the motive of self-enhancement, reciprocity, sense of belonging, and helping others. To test the different motives for consumers to engage in eWOM, the participants were asked eight questions (two for each motive to engage in eWOM). These questions were taken from existing literature and slightly modified to fit the scenarios. The motives for self-enhancement, helping others, sense of belonging and reciprocity were taken from Hennig-Thurau et al. (2004), Yap et al. (2013), and Cheung et al. (2012). In these papers the item loadings of all questions were measured via Cronbach’s alpha reliability. All questions had a significant item loading and it was therefore expected that this also would be the case for this study. See table 5 for an overview of the questions and their item loadings. Further, in this study the items are measured using a five-point Likert scale (Likert, 1932) ranging from 1, disagree strongly, to 5, agree strongly.
3.2. Sample
By the end of the distribution period, 310 people participated in the survey. However, from the 310 people who started the questionnaire, only 231 people indeed finished it. Taking all the 231 together, 90 males (39%) and 141 females (61%) participated. Age varied between the 15 and 85 years with an average age of 35,49 (SD = 16,3). Most participants are younger
28
than 25 years old (45%), 29% is between 25 and 45 years old, and 26% is older than 46. Furthermore, most participants have a Master as educational degree (34,2%) and only 5 participants (2,2%) have a PhD. Of these 5 participants, 4 are male. A summary of the demographics can be found in table 6.
Table 5: Questions concerning the motives to engage in eWOM
Motives Questions Item loading
Self-enhancement 1. I feel good when I can tell others about my
experience with the airline company
2. My contributions about the airline company show others that I am a clever customer
0,78
0,64
Helping others 1. I want to help other with sharing my own experience
about the airline company
2. I want to give others the opportunity to experience a great honeymoon instead of a bad one.
0,93
0,83
Sense of belonging 1. I want to belong to a group on the Internet
2. I want to share my experience with the airline company with others who experience that too
0,87 0,90
Reciprocity 1. When I write about my experience with the airline
company on the Internet, I believe that I will get an answer for giving an answer
2. When I write my knowledge about the airline company on the Internet, I expect to get back knowledge when I need it
0,77
29
Table 6: Summary of the demographics (N = 231)
Gender Male Female 90 (39%) 141 (61%) Age < 25 25 – 45 > 46 104 (45%) 67 (29%) 60 (26%)
Educational degree High school
MBO HBO University (Bachelor) University (Master) PhD 20 (8,7%) 9 (3,9%) 69 (29,9%) 49 (21,2%) 79 (34,2%) 5 (2,2%) 3.3. Statistical Procedure
After having collected the data, these data were investigated. Before analysing the data and testing the hypotheses, all participants who did not fill in the total questionnaire had to be deleted from that dataset. Because all questions in de questionnaire had to be answered (a function in Qualtrics was used to fulfil this condition), I only had to delete the participants who did not fill in anything (though, opened the questionnaire). Of the 310 participants who opened the questionnaire, 79 could be deleted. The remaining 231 participants filled in the questionnaire from the beginning to the end, resulting in perfectly usable values. Furthermore, I recoded the counter-indicative items for ‘agreeableness’, ‘neuroticism’ and ‘extraversion’’. Because some questions were negatively keyed items, I had to change them into positive ones. Afterwards an explanatory factor analysis was performed for all variables to measure their scale reliability.
The Statistical software Package for Social Sciences (SPSS) was used to analyse the data. The first three hypotheses investigate the direct relation between the three different personality traits and the motives to engage in eWOM. To analyse these hypotheses,
30
regression analyses were undertaken. The general linear model (GLM) was used for this regression analysis (Taylor, 2002). Because each hypothesis contains multiple dependent variables, I used the multivariate option of the GLM. This model helped to exclude some interdependencies between the dependent variables. The dependent variables are put into the dependent variables box and the independent variable is put into the covariate box. In the results, the Wilk’s Lambda explained if there is a significant different between the variables. If that is the case, the results of the follow up univariate analysis could be checked. The multivariate option of the GLM was also used for the last three hypotheses. These hypotheses investigate whether an evoked emotions (either anger or sadness) moderates the relation between the personality traits and the different motives to engage in eWOM. The moderators were coded as followed:
-‐ EMOTION à 1 as sad (n = 115) and 0 as not sad (n = 116) -‐ EMOTION2 à 1 as anger (n = 116) and 0 as not anger (n = 115)
To see the effect of the moderator, an interaction between the independent variable and the moderator should be created. The multivariate GLM has a function where you can create these interactions. The interaction will be as followed:
-‐ Hypothesis 4: agreeableness * EMOTION -‐ Hypothesis 5: neuroticism * EMOTION2
In the results, the Wilk’s Lambda explained if there is a significant difference between the variables. If that is the case, the results of the follow up univariate analysis could be checked. After the univariate analysis was checked, four separate multiple linear regressions per personality traits were performed. These multiple linear regressions were done to get an even more thorough look at the results of the regression. But before these regressions could be performed, a new variable – the moderator – was computed. To do this, the variables that
31
form the interaction/moderation (the interaction between the evoked emotion and the personality trait) were standardized. The standardization gave me the z-scores of the variables. After the standardization, the interaction/moderator was computed. By multiplying the z-scores of the two standardized variables (for example agreeablenessZ and emotionZ) the new variable was computed. To run the multiple linear regressions with the interaction/moderation, the independent variable (e.g. agreeableness), the evoked emotion (EMOTION), and the interaction/moderation were put in the regression as independent variables.
4. Results
Below, first the reliability of all the scales will be explained. Afterwards the descriptives, normality and correlation matrix will be outlined. The direct relations between the personality traits and the motives to engage in eWOM will follow. In the end the moderation analysis will be discussed. Paragraph 4.6 provides an overview of whether or not the sub-hypotheses were supported.
4.1. Reliability
To test the scale reliability of all scales consisting of multiple questions a factor analysis was used. A factor analysis is a multivariate statistical procedure that tests the reliability and the validity of scales. In this study I used an Explanatory Factor Analysis (EFA), because I have no expectations of the number of variables (Williams, Brown & Onsman, 2012). An overview of the Explanatory Factor Analysis outcomes is found in Appendix 3.1.
32 Explanatory Factor Analysis: Agreeableness
All 10 questions of the overall variable ‘agreeableness’ were put into an Explanatory Factor Analysis. The method Principal Component Analysis is used. Because correlation between the factors was expected, the rotation method Oblimin with Kaiser Normalization was used (Field, 2005). The Pearson correlation matrix was used to check the pattern of relationships. There are a great number of coefficients beyond 0,3 and no coefficients that are too high (R>0,9) or coefficients that are too low (R<0,0001). This means that all questions of ‘agreeableness’ correlate fairly well and, therefore, there is no need to consider eliminating any questions yet (Field, 2005).
The Kaiser-Meyer-Olkin (KMO) resulted in 0,874, which indicated that patterns of correlations are relatively compact. The Bartlett’s test of Sphericity is highly significant (<0,001), which indicates that there are relationships between the variables. These results show that a factor analysis was appropriate. The Rotated Component Matrix and the Scree plot showed that ‘agreeableness’ was made out of two factors. The first factor accounts for 38,935% of the variance and the second factor for 11,437%. The questions that make up the first factor (loadings >0,5) will be kept as the variable ‘agreeableness’. The other questions will be deleted. ‘Agreeableness’ will be made out of the questions “I am interested in people”, “I sympathize with others’ feelings”, “I am not interested in other people’s problems”, “I have a soft heart”, “I take time out for others”, “I feel others’ emotions” and “I make people feel at ease”. A Cronbach’s alpha reliability test was also performed to be even more sure of the internal consistency of the questions. For this scale the alpha value is 0,789. This is already high enough, though, after deleting the question “I have a soft heart”, the alpha value increased to 0,831.
33 Explanatory Factor Analysis: Extraversion
All 10 questions of the overall variable ‘extraversion’ were put into an Explanatory Factor Analysis. The method Principal Component Analysis is used. Because correlation between the factors was expected, the rotation method Oblimin with Kaiser Normalization was used (Field, 2005). The Pearson correlation matrix was used to check the pattern of relationships. There are a great number of coefficients beyond 0,3 and no coefficients that are too high (R>0,9) or coefficients that are too low (R<0,0001). This means that all questions of ‘extraversion’ correlate fairly well and, therefore, there is no need to consider eliminating any questions yet (Field, 2005).
The Kaiser-Meyer-Olkin (KMO) resulted in 0,887, which indicated that patterns of correlations are relatively compact. The Bartlett’s test of Sphericity is highly significant (<0,001), which indicates that there are relationships between the variables. These results show that a factor analysis was appropriate. The Rotated Component Matrix and the Scree plot showed that ‘extraversion’ was made out of two factors. The first factor accounts for 46,405% of the variance and the second factor for 10,944%. The questions that make up the first factor (loadings >0,5) will be kept as the variable ‘extraversion’. The other questions will be deleted. ‘Extraversion’ will be made out of the questions “I do not talk a lot”, “I feel comfortable around people”, “I start conversations”, “I have little to say”, “I talk to a lot of different people at parties” and “I am quiet around strangers”. For this scale the alpha value is 0,812.
Explanatory Factor Analysis: Neuroticism
All 10 questions of the overall variable ‘neuroticism’ were put into an Explanatory Factor Analysis. The method Principal Component Analysis is used. Because correlation between the factors was expected, the rotation method Oblimin with Kaiser Normalization
34
was used (Field, 2005). The Pearson correlation matrix was used to check the pattern of relationships. There are a great number of coefficients beyond 0,3 and no coefficients that are too high (R>0,9) or coefficients that are too low (R<0,0001). This means that all questions of ‘neuroticism’ correlate fairly well and, therefore, there is no need to consider eliminating any questions yet (Field, 2005).
The Kaiser-Meyer-Olkin (KMO) resulted in 0,818, which indicated that patterns of correlations are relatively compact. The Bartlett’s test of Sphericity is highly significant (<0,001), which indicates that there are relationships between the variables. These results show that a factor analysis was appropriate. The Rotated Component Matrix and the Scree plot showed that ‘neuroticism’ was made out of three factors. The first factor accounts for 40,820% of the variance, the second factor for 14,205% and the third factor for 11,340%. The questions that make up the first factor (loadings >0,5) will be kept as the variable ‘neuroticism’. The other questions will be deleted. ‘Neuroticism’ will be made out of the questions “I am easily disturbed”, “I get upset easily”, “I change my mood a lot”, “I have frequent mood swings” and “I get irritated easily”. For this scale the alpha value is 0,783.
Explanatory Factor Analysis: Helping others
The two questions of the overall variable ‘helping others’ were put into an Explanatory Factor Analysis. The method Principal Component Analysis is used. Because correlation between the factors was expected, the rotation method Oblimin with Kaiser Normalization was used (Field, 2005). The Pearson correlation matrix was used to check the pattern of relationships. The correlation coefficient goes beyond 0,3 and is not too high (R>0,9). This means that both questions of ‘helping others’ correlate fairly well and, therefore, there is no need to consider eliminating any questions yet (Field, 2005).
35
The Kaiser-Meyer-Olkin (KMO) resulted in 0,5, which indicated that patterns of correlations are relatively compact. The Bartlett’s test of Sphericity is highly significant (<0,001), which indicates that there are relationships between the variables. These results show that a factor analysis was appropriate. The Rotated Component Matrix and the Scree plot showed that ‘helping others’ was made out of one factor. The factor accounts for 75,9% of the variance. Both questions will be kept as the variable ‘helping others’. ‘Helping others’ will be made out of question Q1and Q2. For this scale the alpha value is 0,681.
Explanatory Factor Analysis: Self-enhancement
The two questions of the overall variable ‘self-enhancement’ were put into an Explanatory Factor Analysis. The method Principal Component Analysis is used. Because correlation between the factors was expected, the rotation method Oblimin with Kaiser Normalization was used (Field, 2005). The Pearson correlation matrix was used to check the pattern of relationships. The correlation coefficient goes beyond 0,3 and is not too high (R>0,9). This means that both questions of ‘self-enhancement’ correlate fairly well and, therefore, there is no need to consider eliminating any questions yet (Field, 2005).
The Kaiser-Meyer-Olkin (KMO) resulted in 0,5, which indicated that patterns of correlations are relatively compact. The Bartlett’s test of Sphericity is highly significant (<0,001), which indicates that there are relationships between the variables. These results show that a factor analysis was appropriate. The Rotated Component Matrix and the Scree plot showed that ‘self-enhancement’ was made out of one factor. The factor accounts for 66,2% of the variance. Both questions will be kept as the variable ‘self-enhancement’. ‘Self-enhancement’ will be made out of question Q1 and Q2. For this scale the alpha value is 0,388. This alpha value is far too low. A low alpha value could be due to interrelatedness between the items (a heterogeneous construct) or due to a low number of questions (Tavakol &
36
Dennick, 2011). Because internal consistency is a necessary but not sufficient condition for measuring homogeneity (Tavakol & Dennick, 2011), I still will use the questions in this study. The low alpha value will be discussed as a limitation for this study.
Explanatory Factor Analysis: Sense of Belonging
The two questions of the overall variable ‘sense of belonging’ were put into an Explanatory Factor Analysis. The method Principal Component Analysis is used. Because correlation between the factors was expected, the rotation method Oblimin with Kaiser Normalization was used (Field, 2005). The Pearson correlation matrix was used to check the pattern of relationships. The correlation coefficient goes beyond 0,3 and is not too high (R>0,9). This means that both questions of ‘sense of belonging’ correlate fairly well and, therefore, there is no need to consider eliminating any questions yet (Field, 2005).
The Kaiser-Meyer-Olkin (KMO) resulted in 0,5, which indicated that patterns of correlations are relatively compact. The Bartlett’s test of Sphericity is highly significant (<0,001), which indicates that there are relationships between the variables. These results show that a factor analysis was appropriate. The Rotated Component Matrix and the Scree plot showed that ‘sense of belonging’ was made out of one factor. The factor accounts for 69,6% of the variance. Both questions will be kept as the variable ‘sense of belonging’. ‘Sense of belonging’ will be made out of question Q1 and Q2. For this scale the alpha value is 0,592. Here again the alpha value is rather low. The low alpha value will be discussed as a limitation for this study.
Explanatory Factor Analysis: Reciprocity
The two questions of the overall variable ‘reciprocity’ were put into an Explanatory Factor Analysis. The method Principal Component Analysis is used. Because correlation
37
between the factors was expected, the rotation method Oblimin with Kaiser Normalization was used (Field, 2005). The Pearson correlation matrix was used to check the pattern of relationships. The correlation coefficient goes beyond 0,3 and is not too high (R>0,9). This means that both questions of ‘reciprocity’ correlate fairly well and, therefore, there is no need to consider eliminating any questions yet (Field, 2005).
The Kaiser-Meyer-Olkin (KMO) resulted in 0,5, which indicated that patterns of correlations are relatively compact. The Bartlett’s test of Sphericity is highly significant (<0,001), which indicates that there are relationships between the variables. These results show that a factor analysis was appropriate. The Rotated Component Matrix and the Scree plot showed that ‘reciprocity’ was made out of one factor. The factor accounts for 71% of the variance. Both questions will be kept as the variable ‘reciprocity’. ‘Reciprocity’ will be made out of question Q1 and Q2. For this scale the alpha value is 0,558. Here again the alpha value is rather low. The low alpha value will be discussed as a limitation for this study.
4.2. Descriptives and normality
Table 7 gives an overview of the descriptives of the variables that were used in this study (after performing reliability check). The mean, standard deviation and the variance are given. Also the normality of all variables is checked via skewness and kurtosis. Skewness is a measure of symmetry, or more precisely a concept that measures the lack of symmetry (Doane & Seward, 2011). Compared to normal distribution, skewness happens when the distribution is asymmetric and can be skewed to the left (negative) or to the right (positive) (Field, 2005). Kurtosis is a measure of whether the data are peaked or flat relative to a normal distribution (Field, 2005). When looking at skewness, it can be seen that almost all variables are approximately normal (skewness between -1 and 1) (Doane & Seward, 2011). However,
38
‘agreeableness’ is significantly negatively skewed. For kurtosis it can be seen that also almost all variables are approximately normal (kurtosis between -3 and 3) (Field, 2005). Again, the variable ‘agreeableness’ is significantly non-normal. Even though agreeableness is not normally distributed, I still will use it in the analysis. This will later be discussed as a limitation of the study.
Table 7: Descriptives and normality
Variable N Mean SD Variance Skewness Kurtosis
Agreeableness 231 4,0657 0,51496 0,265 -1,487 4,181 Extraversion 231 3,4504 0,62540 0,391 -0,456 0,264 Neuroticism 231 2,6292 0,62819 0,395 0,175 -0,349 Helping others 231 3,6752 0,87482 0,765 -0,824 0,490 Self-enhancement 231 2,6560 0,88395 0,781 0,223 0,269 Sense of belonging 231 2,3996 0,89799 0,806 0,717 0,763 Reciprocity 231 2,9615 0,89623 0,803 -0,145 -0,391
4.3. Correlation Analysis and Cronbach’s alpha
Table 8 gives an overview of the correlation analysis. The Cronbach’s alpha value is stated on the diagonal between brackets. A first observation derived from the table is that the personality trait agreeableness is significantly correlated with the motive helping others. Also is seen that extraversion is in coherence with the motives self-enhancement and sense of belonging. The personality trait neuroticism only significantly correlates with the motive sense of belonging. Further, this table tells that gender significantly correlates with age, educational degree and the personality traits agreeableness and neuroticism. The four different motives why people engage in eWOM are also significant correlated to each other. This also
39
counts for the three different personality traits, which are also significantly related to each other. Table 8: Correlation Variables 1 2 3 4 5 6 7 8 9 10 1. Gender - 2. Age -,285** - 3. Educational degree -,194** ,060 - 4. Agreeableness ,311** -,063 ,040 (0,831) 5. Extraversion ,068 -,026 ,016 ,344** (0,812) 6. Neuroticism ,207** -,228** -,134* -,127* -,205** (0,783) 7. Helping others ,120 -,133* -,054 ,212** -,075 ,048 (0,681) 8. Self-enhancement -,028 -,067 -,025 ,089 ,151* ,040 ,223** (0,388) 9. Sense of belonging ,123 -,051 ,015 -,041 -,241** ,145* ,236** ,367** (0,592) 10. Reciprocity ,007 -,106 ,085 -,048 -,034 ,098 ,237** ,364** ,439** (0,558) Note: N = 231
* Correlation is sig at the 0,05 level ** Correlation is sig at the 0,01 level Cronbach’s alpha between brackets
4.4. Direct effect
The first three hypotheses (11 sub-hypotheses) propose that there is a direct relation between the different personality traits and the different motives why people engage in eWOM. Since I am dealing with multiple dependent variables in each hypothesis, multivariate linear regression is used. The overall test of agreeableness (hypothesis 1) was significant: Wilks’ Lambda = 0,93, F (4,229) = 4,427, p < 0,05. Because the test was significant, I looked at the follow up univariate analysis (Manly, 2004). In this analysis it was seen that agreeableness does have a significant effect on the motive helping others: F (1,231) = 10,882; β = 0,212; p < 0,05 and that agreeableness explained 4,5% of variance in the motive helping others. Thus, hypothesis 1a is supported. Hypothesis 1b is not supported. There is no relation
