• No results found

The Effect of Valence and Arousal of NPO Communications on Electronic Word-of-Mouth (eWOM)

N/A
N/A
Protected

Academic year: 2021

Share "The Effect of Valence and Arousal of NPO Communications on Electronic Word-of-Mouth (eWOM)"

Copied!
47
0
0

Bezig met laden.... (Bekijk nu de volledige tekst)

Hele tekst

(1)

The Effect of Valence and Arousal of NPO Communications on

Electronic Word-of-Mouth (eWOM)

Morven Piers 10232206 Thesis: Business Studies

Supervisor: Hsin-Hsuan (Meg) Lee Academic Year: 2013-2014

(2)

Abstract

Non-profit organisations are recently realising the growing importance of marketing for their causes, campaigns and organisations. Due to the restrictive budget these types of organisations usually have, innovative and diverse ways of reaching their target markets are crucial. Electronic word-of-mouth (eWOM) for non-profit organisations is a unique way of creating awareness and receiving donations. Marketers for these organisations need to know how the communications they present to consumers will drive them to engage in eWOM, in order to effectively tailor the message to promote this. Therefore, this study examines how the valence (positive and negative) and arousal (high and low) in non-profit communications affects consumers’ willingness to engage in eWOM. In order to do this, an internet-mediated experiment questionnaire was conducted. The results show that consumers are more likely to spread positive rather than negative communications, and the level of arousal is not a predictor of likelihood of engaging in eWOM. Furthermore, valence affects how consumers perceive the organisation, but not how they perceive the campaign. Arousal does not affect either of these variables. An exploratory study showed that the perceptions of the organisation and the likelihood of engaging in eWOM are linked. Researchers should look further into what drives

consumers to share non-profit communications, as there are a number of potential other factors influencing this. Marketers can use these findings to accordingly adjust the messages they are communicating, as well as to make themselves aware of the potential impact of their communications on customer perceptions and actions.

(3)

Table of Contents

Abstract 2 1. Introduction 5 2. Literature Review 8 2.1 Electronic Word-of-Mouth 8 2.2 Message Framing 9 2.3 Conclusion 11 3. Conceptual Framework 12 4. Methodology 14 4.1 Research Design 14 4.2 Sample 16 4.3 Measures 17 4.3.1 Message Framing 17

4.3.2 Perceptions of the organisations, issues and campaigns 18

4.3.3 User willingness to engage in eWOM 18

4.3.4 Control variables 18

4.4 Data Analysis 19

5. Results 20

5.1 Descriptive Statistics & Reliability 20

5.2 One-Way ANOVA 22

5.2.1 Differences in the valence of the message 23

5.2.2 Differences in the arousal of the message 23

5.3 ANCOVA 23

5.4.1 Differences between groups for NPO perceptions 24 5.4.2 Differences between groups for campaign perceptions 25

5.4 Exploratory Correlation Analysis 25

6. Discussion 27

6.1 Differences in relation to valence 27

6.2 Differences in relation to arousal 28

(4)

6.4 Differences in NPO campaign perceptions 29

6.5 Theoretical Implications 29

6.6 Managerial Implications 30

6.7 Limitations of the Study 31

7. Conclusion 33

Bibliography 35

Appendix 38

Appendix I: Self-Assessment Mannikin 38

(5)

1. Introduction

With the increased power and use of social media and online platforms today, it is

essential that marketers know how to properly convey messages to their target consumers on the Web. Today, almost every person is involved in social networking (online and offline) (Allsop et al., 2007), and approximately 58% of Internet users are involved in an online social networking site (Social Networking Statistics). These users have been known to post opinions, reviews, experiences, knowledge, etc. online, which has been proven by a number of authors (Hennig-Thurau, Gwinner, Walsh and Gremler, 2004; Subramani and Rajagopalan, 2003; Chiou and Cheng, 2003). Online discussions are important today due to the increased and persuasive use and reliance on the Internet. Hennig-Thurau et al. (2004, p. 39) define electronic word-of-mouth (referred to as eWOM hereafter) 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”. Besides products or companies, Internet users may also discuss social issues online. The number of internet users using social media to convey their opinion is rising, such as highlighted in KPMG’s report on the increasing use and importance of social media relating to healthcare issues (Britnell, 2011), and it is important that non-profit organisations (NPOs) are aware of how perceptions of their organisations and campaigns influence eWOM, and how best to use this new aspect of communication.

Williams & Buttle (2013, p. 285) highlight the increased interest in the NPO and academic worlds of the “influence WOM has on organisational outcomes”. Thus far, little investigation has been made as to the influence of message framing on consumer

perceptions of NPOs. Williams & Buttle (2013) help to change the perspective of how NPOs should be viewed, claiming that they are service organisations with an intangible product. It is clear that traditional WOM, and increasingly electronic WOM, have been heavily analysed from a marketing perspective, however, far less emphasis has been placed on NPOs. Therefore, research should additionally be conducted into how

consumer perceptions of NPOs affect the extent to which consumers engage in eWOM, especially as eWOM could become a key (and almost free) way for NPOs to

(6)

engender passionate support for their work.

In many NPO communications, negative emotions, such as sadness or guilt, are typically the focus of the message. However, as White, MacDonnell & Ellard (2012, p. 114) point out, this is usually ineffective, due to the “curvilinear relationship between the degree of need and helping”. Researchers also have contrasting views on this, as Faseur and Geuens (2010, p. 499) state that the “use of emotional advertising appeals can be very effective to persuade people to donate money for the good cause that is advertised or to promote other helping behaviour”. There is clearly a need for NPO marketers to know how to appropriately focus on emotions, whether it be positive or negative, in order to reach and appeal to consumers. WOM, in particular in emotional appeals, also has one of the greatest impacts on consumers’ perceptions of a company, demonstrating how

important this is, both for literature and in practice (Allsop et al., 2007).

Pilion Trust, a small non-profit organisation campaigning against poverty, recently conducted and filmed a social experiment to see how people on a London high-street would react to two vastly different messages (Smith, 2014, 8 April). The first message contained an offensive slogan directed at the poor, while the second was

requesting help for the poor. In the first case, it is visible that members of the public were angry and vocalised this to the representative, however, upon presenting the second message in the same location, the representative was almost completely ignored. The message Pilion Trust was trying to convey was that the public does care about poverty, and how a more shocking message seemed to be much more thought-provoking. This campaign has gone viral over YouTube and Facebook.

Possible reasons for the virality of this campaign can also be found in academic research. Research has shed light on the fact that higher arousal content, in Pilion Trust’s case – anger, is more likely to be shared (Berger & Milkman, 2012). Past literature has taken contradicting stances as to whether positive or negative content is more likely to become viral, with both sides being argued for (e.g., Park & Lee, 2009; Berger & Milkman, 2012). Although these may have an effect on virality, Berger & Milkman (2012, p. 10) suggest that it is necessary to look at more than just valence (whether the message is positive or negative), and that there is a “causal relationship between activation and social transmission”. Dillard & Peck (2000) add that valence is simply a

(7)

way of distinguishing affects, and not explaining why these occur.

This research aims to close the gap in research by answering the following research questions: How do valence and arousal in framing of communication messages from NPOs influence consumers’ perceptions of the campaign and company? How do these varied perceptions in turn affect user willingness to engage in eWOM?

In order to provide an answer to these research questions, an experiment will be conducted among participants of differing genders, ages, issue involvement levels and trust levels. Considering the results of the survey, and the analysis herein, NPOs can use the presented results to frame their future messages and campaigns in order to positively influence consumers. Additionally, marketers will be able to acknowledge how different perceptions might cause consumers to engage in eWOM, and how to manage it.

The remainder of this paper is structured as follows. Firstly, a literature review is conducted, analysing past literature. Secondly, the conceptual framework is presented, from which the research will be based. Thirdly, the methodology will explain the survey and the components of research. Fourthly, the results will be presented as well as the analysis. This will lead to a discussion and short conclusion, where the research question will be answered and explained.

(8)

2. Literature Review

The following section elaborates on the literature and studies that should be considered in relation to the research problem, as well as highlighting the gap in literature that this study aims to fulfil.

2.1 Electronic Word-of-Mouth (eWOM)

Electronic Word-of-Mouth (eWOM) was defined above in Section 1 (Introduction), using the definition presented by Hennig-Thurau et al. (2004). Bickart and Schindler (2001) highlight the increasing importance of online discussions to marketers and consumers, stating that it is “more powerful than marketer-generated information in stimulating product category interest” (p. 37). Gruen, Osmonbekov and Czaplewski (2005, p. 449) support Bickart and Schindler (2005). They express that word-of-mouth (WOM) “has been shown to have a significant impact on consumers’ choice as well as post-purchase product perceptions”. Furthermore, Hennig-Thurau et al. (2004) expresses that because of the unique characteristics the Internet has offered consumers, such as the ability to

communicate directly to the target audience, the unlimited time-frame that

communications can be available, as well as the possible anonymity of communicators, eWOM is more important than ever before. This not only applies to products and for-profit companies, but also to services through non-for-profit organisations. Williams and Buttle (2013) stress the need for non-profit organisations to manage WOM, in relation to the intangible service that NPOs offer.

Traditional word-of-mouth (tWOM) definitions typically include sharing product or service experiences with others in a face-to-face setting (Richins, 1984). eWOM research and theories have been compared continuously to traditional WOM research, which Hennig-Thurau et al. (2004) says are highly relatable, especially in regard to motivations for participating. One big difference between the two, presented by Shin, Song and Biswas (2013, p. 2) is that the Internet “encourages people to disseminate the kind of information that does not travel very well in the tWOM contexts”. Bickart and Schindler (2001, p. 37) see eWOM as greatly beneficial due to the written format in which it is presented, which “offers the consumer the ability to acquire the information at his/her own pace”. As in a vast amount of previous studies, as well as due to the fairly

(9)

new concept of eWOM, this study will also take into consideration evaluations of general WOM principles and apply such results to eWOM. This is validated by results of studies including a comparison the two WOM forms. For example, in Bickart and Schindler’s 2001 research, they show that the characteristics (credibility, relevance and ability to evoke empathy) that prove online forums to hold higher influential status than company websites, share the same characteristics needed for forms of tWOM communication to be influential.

Despite the wide range and quantity of research on WOM and currently on eWOM, very few studies have looked into eWOM for non-profit organisations (NPOs). Williams and Buttle (2013) present a view regarding how NPOs are managing or can better manage WOM communication, however, little information is provided to the electronic form, as well as why WOM occurs. As an NPO is offering an intangible service (Williams and Buttle, 2013), and the organisation operates in a few similar ways to for-profit companies, the literature is also somewhat applicable.

2.2 Message Framing

Chang and Lee (2012, p. 2912) explain that framing “refers to the presentation of one of two equivalent value outcomes to different decision makers, where one outcome is presented in ‘positive’ or ‘gain’ terms, and the other is presented in ‘negative’ or ‘loss’ terms”. Throughout the literature, the focus within message framing has been heavily placed on the gains and losses aspect of message framing (i.e., Puto, 1987; Bettman and Sujan, 1987; Qualls and Puto, 1989; Kahn and Meyer, 1991; Ganzach and Karsahi, 1995). Kahneman and Tversky (1979) also focus on this theory, explaining it in terms of the prospect theory, as the majority of people are risk-adverse when the decision problem is presented as what one can ‘gain’, and risk-prone when the problem is angled at what could be ‘lost’. This highlighted that humans exhibit loss aversion. Ganzah and Karsahi (1995) add the element of involvement in their gain/loss research, finding that loss framing was more effective when the individual had high involvement, and gain framing was a better option when involvement was low.

Valence (i.e. positive or negative) messaging framing has been looked at to a lesser extent. Dillard and Peck (2000, p. 462) claim, “valence is an important means of

(10)

distinguishing affects”, although their study focused on valence with discreet emotions, in order to conclude the affects regarding persuasion. Chang and Lee (2012) highlight messages promoting charitable donations as being influenced by valence, concluding that negatively framed messages are more effective in donation intentions than positively framed messages. However, Berger and Milkman (2012) prove that positively framed messages are more effective (concluded through the likelihood that the individual would share the message). Huang (2001, p. 245) suggest that “positive and negative emotions are found to be independent in marketing, and the presence of positive emotions does not preclude the absence of negative emotions”. This indicates that message framing must not be viewed as one extreme or another, and researchers need to take this into

consideration in multiple areas of their studies (experimental design, results and analysis). An additional way of message framing is the level (low-high) of arousal evoked as a consequence of the communication. Lang, Dhillon and Dong (1995, p. 314) define arousal as “a continuous response ranging from ‘energised, excited and alert’ to ‘calm, drowsy or peaceful’”. Smith and Ellsworth (1985) suggest that there are two dimensions of emotions, pleasantness and arousal (level of activation). Berger (2011) explains that the level of arousal can be a factor in the behaviour of the consumer following exposure to the communication. Furthermore, Lang et al. (1995, p. 324) finds that “the arousal level of a message may be as much a determinant in whether it is remembered, as the valence of the appeal”, demonstrating the significance of arousal for organisations.

Additionally, arousal can influence whether the recipient will share the message or not, as Berger (2011, p. 892) finds: “Physiological arousal can plausibly explain transmission of news or information in a wide range of settings… regardless of whether they are positive or negative in nature”.

There are a number of reasons that have been previously addressed as to why negatively framed messages are seemingly more effective than positively framed messages. Firstly, “negatively framed messages tend to arouse viewers’ self relevance, consciousness and sympathy regarding the serious consequences if no action is taken” (Chang and Lee, 2012, p. 2912). This theory shows that the emotions that the messages evoke are what cause the viewers to act on the messages, whether this is through sharing or purchase intentions. Secondly, consumers’ have been shown to desire more

(11)

information about the potential consequences if no action is taken, and “ways to avoid their occurrence” (Chang and Lee, 2012, p. 2912), when the message is framed

negatively. As the message evokes arousal, consumers are looking for ways to balance this arousal, hence the search for additional information (Chang and Lee, 2012).

Guy and Patton (1989; referenced in Fraseur and Geuens (2010)) explain two categories of potential factors that may influence a decision process to donate money to a charitable cause. The first (internal factors) involves demographics, personality traits, mood, etc. The second (external factors) relates more to how the appeal is framed, who is involved, and so forth. Guy and Patton (1989) elaborate on these factors, claiming that the external factors are those which drive helping behaviour.

2.3 Conclusion

In summary, electronic word-of-mouth (eWOM) is a crucial way for marketers to engage with customers as well as gaining publicity about their cause, campaign, organisation, etc. It is critical that marketers know how to best frame a message in order to encourage online sharing. The main focus of message framing literature has been to the gain and loss frame. A lot of attention has also been drawn to the valence of a message, but less so to the arousal. In the non-profit sector, the way a message is framed has barely been considered.

The research conducted in this paper aims to close an important gap for non-profit marketers, as it will analyse how message framing (presented with different types of valance and arousal) of NPO communications will affect eWOM. This will give marketers a way to target consumers with the right messages to encourage their

participation in and promotion of the organisation, with the aims to increase awareness, donations, sign-ups, or other goals.

(12)

3. Conceptual Framework

This section discusses the specific application of these theories and ideas to the research being conducted in this paper, as well as presenting the hypotheses of the study.

Past literature has taken contradicting stances as to whether positive or negative content is more likely to become viral, with both sides being argued for (e.g., Park and Lee, 2009; Berger and Milkman, 2012; Godes et al, 2012). Furthermore, common practice in NPO communications is to frame messages as negative, fear appeals, as proven in a content analysis by Dillard and Peck (2000), which seemed far less effective. However, Berger and Milkman (2012) prove that positive communications are more likely than negative messages to be shared. They test this using multiple methods, including evaluation of articles that were on the “most emailed” list for the New York Times, and laboratory controlled experiments. Although this has not been explicitly tested in a non-profit setting, the reasoning and results of Berger and Milkman (2012) are expected to transfer to NPOs. Therefore, for NPO communications, it is expected that consumers will be more inclined to share positive communications rather than negative.

Hypothesis 1: Consumers are more likely to engage in eWOM for positive rather than negative NPO communications.

In his 2011 paper, Berger finds that “arousal-inducing content should be shared more than content that does not induce arousal” (p. 892). Additionally, Faseur and Geuens (2010) explain that emotional appeals can be an effective source of stimulating donation (time or money) intentions. Furthermore, Berger and Milkman (2012, p. 200), prove that “content that evokes more anger or amusement (high arousal) is more likely to be shared, and this is driven by the level of activation it induces”, one of the first papers that directly links arousal to eWOM. This research will follow the line of evidence that Berger and Milkman (2012) have presented, hypothesising that consumers are more likely to engage in eWOM when communications are high rather than low arousal.

Hypothesis 2: Consumers are more likely to engage in eWOM for high arousal rather than low arousal NPO communications.

(13)

The messages presented to respondents will differ in valence (either positive or negative) and arousal (either high or low arousal). Dillard and Peck (2000, p. 488) show that the level of emotion in the communication influences how the individual evaluates the effectiveness of the communication, and in turn the company on a whole. Buda and Zhuang (2000) also find that consumers’ attitudes differ according to the valence of the message. Additionally, Allsop et al. (2007) found that “emotional appeal (trust, good feelings, respect) consistently has the strongest influence on corporate reputation”. This means that the framing of the message will influence the organisational reputation in the consumers’ mind.

Hypothesis 3a: Consumers’ perceptions of the NPO will differ based on the valence and arousal of the communication.

Hypothesis 3b: Consumers’ perceptions of the campaign will differ based on the valence and arousal of the communication.

(14)

4. Methodology

This research aims to provide a tool for NPOs to use with regard to their marketing communication. In the previous section, specific constructs and hypotheses were presented. The following section explains and justifies the choices made in the research design, sample, measurement of variables, survey design and analysis.

4.1 Research Design

For this research, experiments by means of questionnaires were conducted. The

questionnaires were distributed through e-mail and social media platforms (Facebook and Twitter), and the questionnaire was hosted on “Qualtrics.com”, a commonly used survey platform via the University of Amsterdam. This automatically allowed respondents to see the research was conducted for the University of Amsterdam, under the faculty of

Economics and Business. Upon clicking on the questionnaire link, respondents were randomly allocated (through the computer programme software) one of four message types. The participants were first shown the communication (consisting of a short text and an image), each with different valence and arousal combinations from a fictitious non-profit organisation, “Mothers Against Leukemia (MAL)”.

The participants were asked to read the communication carefully and answer the questions following, regarding perceptions of the NPO, likelihood to engage in eWOM and control questions (familiarity, trust, age, gender etc.). All of the questions were identical, with the only variable factor being the communication (text and image) presented at the beginning. The main character, setting and issue were also the same throughout all the communications. One of the versions of the questionnaire (from the manipulation ‘Positive, high arousal’) can be found in Appendix II. It was mentioned in the introductory passage that the responses given would solely be used for the purpose of this research, and all their responses and information provided was anonymous.

An internet-mediated questionnaire was chosen due to large number of

participants needing to be reached (Saunders et al., 2009). Additionally, the survey was open to be completed by anybody who had access to a computer, which meant almost all individuals that would open the link to the survey would be eligible participants, and allowed for mass distribution. As time for conducting the research was relatively limited,

(15)

it would be unfeasible to conduct telephone interviews or postal questionnaires, thus the choice for a self-administrated internet-mediated option (Saunders et al., 2009). This method of questionnaire distribution also meant that the process of collecting data was almost autonomous. As there are four different types of messages being researched (namely, positive high arousal, positive low arousal, negative high arousal and negative low arousal), an experiment must be conducted, with participants being randomly assigned a message type to adapt their responses for, similar to the technique used in Berger and Milkman (2012).

Although an internet-mediated questionnaire seems most apt to the research discussed, there are some limitations of this type of research method. Firstly, the questionnaire is limited in length. It has been advised to limit the number of questions and pages, due to the low completion rate otherwise. Furthermore, Saunders et al. (2009, p. 364) finds a very low response rate of 11% for internet-mediated questionnaires, compared to around 50 – 70% for telephone and structured-interview questionnaires. A pilot study was conducted using four participants for each message type in order to ensure the questionnaire worked sufficiently well, and the communications were manipulated in an appropriate manner. The feedback from this pilot study was taken into consideration before officially distributing the survey. After this pilot study and the appropriate changes were made, the questionnaire was distributed. Unfortunately the respondents did not view the manipulations (especially that of arousal) as expected with the larger (N = 139) number of participants, and the first official version of the study was deemed invalid. Thereafter, the questionnaire was adjusted again; another small pilot study (N = 14) was performed, and the questionnaire was distributed.

In order to adequately adjust the text to fit the message frame desired, a subset of emotions were used for representation. These emotions were deemed appropriate due to their nature and the common practices of NPO communications. Further more, the negative emotions were taken from Berger and Milkman’s 2012 study, where they too used anger for high arousal and sadness for low arousal. The emotions used to manipulate the texts and images are as follows:

Awe (positive; high arousal), contentment (positive; low arousal), anger (negative; high arousal), sadness (negative; low arousal)

(16)

The communications created used the same character (Chanda, a young girl from India), issue (juvenile leukemia), campaign (One Child at a Time) and organisation (Mothers Against Leukemia). Each of the texts were subsequently adjusted in order to fit the emotions targeted, named above. An example of one of these texts (positive and high arousal; attempting to be awe-inspiring) is provided in the questionnaire example in Appendix II.

4.2 Sample

In the sample used, there were 125 completed questionnaires, of which 37% were male and 63% female. All these participants were technology literate, as the survey was conducted over the Internet (either through Facebook or email).

As it would be very unrealistic to sample the whole (technology literate) population, a sample needed to be derived in order to achieve results that can be

generalised. As it was not possible to use probability sampling, due to time and financial constraints, the sample was selected through non-probability sampling (Saunders et al., 2009). Convenience sampling and self-selection sampling were used as individual cases are relatively easy to identify, the sample needed to be relatively large, and the variation in the population could be controlled for (Saunders et al., 2009).

Unfortunately, in both convenience and self-selection sampling, there is a low probability that the sample is representative of the population (Saunders et al., 2009). As previously mentioned, the variations in the population that would ensure the

generalisability of the sample were measured through control variables (i.e. age, gender, issue involvement, trust and familiarity). The combination of convenience and self-selection sampling were conducted through e-mail and social media platforms (i.e. Facebook and Twitter) through personal networks.

The sample consisted generally of respondents familiar with social media, due to the platforms on which the questionnaire was presented. As the majority of respondents were drawn from personal networks, there were a large number of students. This could be an area for improvement in the future, in order to achieve a more representative

distribution of ages. However, for the purpose of this study, age is relatively insignificant, and seen as a control variable.

(17)

Through this means of sampling, and considering that there are four different message types that needed to be considered, at least 30 responses were required in each category. This is in line with Stutely’s (2003) minimum threshold in order to statistically analyse the data. Therefore, a minimum of 120 respondents was needed for the 2x2 questionnaire.

4.3 Measures

In the questionnaire, the following methods were used in order to measure the variables, ensuring the validity of the research. Simple language or pictures were used and

definitions were provided where determined necessary, in order to ensure all participants understand the questions in the intended way (Saunders et al., 2009). All four

questionnaires were identical, the only difference being the picture and text to read in the beginning, for the “NPO communication”.

4.3.1 Message Framing

To examine the affects of message framing, four different texts and images were used in the questionnaire, each containing one of the valence/arousal pairs, represented by one of the four emotions, as described earlier.

The emotions used to manipulate each of the four messages were controlled for using the Self-Assessment Manikin (SAM) (Lang, 1980), which uses pictures to measure the arousal, pleasure and dominance of emotions in response to a situation (see Appendix I). This method allows participants to be able to connect the pictures and their feelings in a visual manner, as arousal is a relatively difficult emotion to pinpoint with such a short message, as well as when using the internet-mediated questionnaire method. This method was found to be more successful than the polar questions (e.g. How do you feel after reading this text on a 7-point polar scale from Mellow to Active), which were asked in the first attempt of this questionnaire (taken from Berger and Milkman’s 2012 study). Due to the nature of this research, the dominance aspect of the manikin (seen in

Appendix I on the 3rd, bottom, row) was excluded from the data collection, and therefore only pictures representing valence and arousal were presented to participants.

(18)

4.3.2 Perceptions of the organisations, issues and campaigns

To measure this opinion variable (Saunders et al., 2009), a series of questions were asked following one of the four texts to which the participants were required to respond based on a seven-point Likert scale. These questions mainly consisted of “agreement” ratings with a scale from “strongly agree” to “strongly disagree” (Saunders et al., 2009, p. 380). To avoid too many questions, a fake company (Mothers Against Leukemia (MAL)) was used in order to cancel out perceptions users may have had of the company before the experiment. Therefore, the questionnaire only asked for users perceptions of the organisation after they have read the text. These questions included statements adapted from Heller and Reitsema (2010).

4.3.3 User willingness to engage in eWOM

In order to assess how willing the user would subsequently be to engage in eWOM after they had stated their perceptions from the text, seven-point likelihood Likert-scales were used, ranging from “extremely” to “not at all” (Saunders et al., 2009, p. 380). The same method can be seen in Berger and Milkman’s (2012) studies 2 and 3, asking participants “how likely would you be to share it with others” (p. 200). This was also adapted to include two more questions, asking if they believed by sharing the communication it would help the organisation and/or campaign, in order to gain insight as to why one would share the communication.

4.3.4 Control Variables

Additionally, there are a number of control variables that were required throughout the experimental questionnaire. These were phrased as closed category questions (age, gender) and seven-point Likert-scale questions (familiarity, issue involvement, trust). Familiarity questions included how familiar the participant was with the organisation, campaign and issue before the questionnaire. Due to the NPO and campaign both being fake, this was to ensure participants were answering honestly to the questions, as well as controlling familiarity as a factor affecting perceptions and/or eWOM. The issue

involvement questions were adapted from Maheswaran and Meyers-Levy (1990, p. 364), asking “how interesting, involving and personally relevant the material was”. Trust was

(19)

another control variable included in the study, which also asked participants the extent to which they trusted the NPO, campaign and issue. In order to confirm the communication manipulation was accurate and plausible, two questions, adapted from Mohr and Webb (2005), were included to check the credibility of the scenario, including if they could image the situation occurring in real life.

4.4 Data Analysis

The data analysis was conducted using a few techniques: After looking at the data descriptive statistics, and ensuring that the data was reliable (using Cronbach’s Alpha), both a one-way ANOVA and ANCOVAs were performed. In order to test Hypotheses 1 and 2, a one-way ANOVA was conducted. This tested whether there were significant differences between the means of the group. To run both tests at the same time, planned contrasts accompanied the ANOVA.

Furthermore, to test Hypotheses 3a and 3b, an ANCOVA test was run to compare all the groups, using the control variables of familiarity, trust and issue involvement as covariates in the analysis. This test took into account both valence and arousal (therefore, all groups), and used two dummy variables, “Positive Dummy” (marking positive groups as 1 and negative groups as 0) and “Arousal Dummy” (high arousal groups will be given a score of 1, and low arousal groups a score of 0), in order to clearly distinguish between the groups for analysis. Levene’s Test of Equality of Error Variances was also run in order to confirm the data was not violating the homogeneity of variances assumption needed to perform an ANCOVA test.

(20)

5. Results

5.1 Descriptive Statistics & Reliability

In order to analyse the data in an effective manner, it is crucial to look at the descriptive statistics of the sample used, in order to have an overview of the sample, which could influence the results. As mentioned earlier, there were more female respondents to the survey than male (63% to 37%). The average ages of the survey participants can be seen in Table 1.

Table 1: Ages of Respondents Age (years) Frequency Percentage

Under 18 3 2.4 18 – 24 49 39.2 25 – 34 20 16.0 35 – 44 12 9.6 45 – 64 38 30.4 65+ 3 2.4 Total 125 100.0

As visible in the table, the main bulk (69.6%) of the participants came from the age categories “18 – 24” and “45 – 64”. This is most likely due to the large number of willing participants being friends and students from my personal network (making up the

majority of the “18 – 24” age category), as well as family members’ personal networks (hence the large percentage in the “45 – 64” age category).

The following table, Table 2a, shows the descriptive statistics for the data. These statistics take into account the main dependent variables, company perception, campaign perception, intent and eWOM. Each variable involved a number (3-4) of questions, which were merged together (using the mean of each response category), in order to create one variable, shown in the table. As described in the methodology, the questions were asked using a seven-point Likert-scale, which ranged from strongly disagree (1.00) to strongly agree (7.00). Neither agree nor disagree (4.00) is considered the middle of the scale. For

(21)

NPO and campaign perceptions; the higher the number, the more positively they are evaluated. Regarding intentions; the higher the number, the more likely the participant is to act. The higher the value for eWOM, the more likely the participant is to spread the communication.

Table 2a: Descriptive Statistics

N Min. Max. Mean SD Alpha NPO Perceptions 127 2.00 7.00 5.186 1.024 0.785 Campaign Perceptions 127 1.67 7.00 4.945 1.109 0.821 Intentions 127 1.00 7.00 4.073 1.421 0.879

eWOM 127 1.00 7.00 3.869 1.636 0.907

To test whether the measurements can be used in further analysis of the data, a reliability check must be conducted. As a number of new variables were computed, these reliability checks are crucial in determining whether the variables can be valid and used in the analysis. Cronbach’s alpha is a reliability check that controls the “consistency of a measure” (Field, 2009, p. 681). Unfortunately, the smaller number of items included in the scale, the more likely the alpha is to being insignificant and small (Pallant, 2002), which is something to pay attention to in this research, as generally between two to four items were used for each variable. Field (2009) suggests that an alpha value of 0.7 is the minimum significance the variable can be to be reliable, although the aim is to have the alpha value as high as possible.

In the last column of Table 2, the Cronbach’s alphas for each of the variables can be found. As it can be seen, all the alphas are found to have a relatively high reliability (greater than 0.7). These alphas are for the data as a whole. Upon looking at the

individual manipulations separately, as seen in Table 2b, all alphas were indeed above the 0.7 cut-off, apart from “NPO perceptions” for “Positive; Low Arousal” (PL). The

reliability of this measure was somewhat lower, at 0.657. As it was not possible to increase the alpha by deleting items, this variable was included in the analysis, and the consequences of this are taken into account in the discussion, section six.

(22)

Table 2b: Cronbach’s Alpha per manipulation PH PL NH NL NPO Perceptions 0.787 0.657 0.816 0.741 Campaign Perceptions 0.811 0.838 0.850 0.785 Intentions 0.879 0.879 0.879 0.879 eWOM 0.875 0.926 0.920 0.892 N 37 35 31 24

There are already a number of inferences that could be made looking at this data (Tables 2a and 2b). For example, on average, participants that view messages that are framed negatively and induced low arousal will have less positive perceptions of the NPO than any of the other message combinations. Although this could be assumed by viewing this data, various tests are necessary in order to statistically validate such observations. The following sections will provide statistical evidence needed in order to accept or reject the hypotheses made in Section 3 (Conceptual Framework).

5.2 One-Way ANOVA

Hypothesis 1 theorises that consumers will be more likely to spread positive eWOM than negative eWOM. In order to statistically test this theory, a one-way between subjects ANOVA was conducted, comparing the effect of valence on eWOM in positive and negative conditions. The same one-way between subjects ANOVA will be able to simultaneously compare the effect of arousal on eWOM in high arousal and low arousal conditions. This tests Hypothesis 2, which states that high arousal NPO communications are more likely to be shared than low arousal NPO communications. By performing the tests together, the risk of increasing the type I error is avoided.

Although the initial ANOVA test found that the effect of valence and arousal (together) was not significant at p<0.05 level [F(3, 123) = 2.227, p = 0.088], it was necessary to perform orthogonal contrasts, as it was relatively close (0.038) to the 0.05

(23)

margin. The results of the contrasts are described separately for each independent variable in the following two sections.

5.2.1 Differences in the valence of the message

The first orthogonal contrast was performed in order to test the effect of valence on eWOM. This test compares the means of the groups for positive message framing with the groups for negative message framing, in order to check whether there is a significant difference between the two. The results found that the participants that read the negative communication were significantly less likely to engage in eWOM than those who received the positive communication (p = 0.011), supporting Hypothesis 1. 5.2.2 Differences in the arousal of the message

The second hypothesis, that consumers shown higher arousal messages will be more likely to share than those shown lower arousal messages, was also tested using the

ANOVA contrasts. As previously mentioned, the contrasts were conducted using level of arousal (high condition versus low condition) as the independent variable, and likelihood of engaging in eWOM as the dependent variable. Upon running the contrast tests, the results showed that participants who were in the high arousal conditions were not

significantly more likely to engage in eWOM than those in the low arousal conditions (p = 0.890). This evidence does not support the theory of Hypothesis 2.

5.3 ANCOVA

Hypotheses 3a and 3b aim to prove that message framing will affect consumer

perceptions of the organisation, and of the campaign. In order to do this, a number of 2x2 between-groups analysis of covariance (ANCOVA) tests were performed to assess the influence of valence and arousal combinations on consumers’ perceptions of the NPO and the campaign. The independent variable was the framing of the message (valence and arousal), and the dependent variable was consumer perceptions of the NPO or consumer perceptions of the campaign. The tests were performed taking the control variables (familiarity, issue involvement and trust) into account as covariates.

(24)

order to control for this factor, Levene’s Test of Equality of Error Variances is conducted. If this test is significant (i.e. p < 0.05), the assumption has been violated and the

ANCOVA test cannot be performed.

5.3.1 Differences between groups for NPO perceptions

Hypothesis 3a states that consumer’ perceptions of the NPO will differ according to the framing of the message (level and combination of valence and arousal). This was tested using ANCOVA, with the results shown regarding consumer perceptions of the NPO in Table 3.

Table 3: Test of Between-Subject Effects for NPO Perception Source Mean Square F Significance Levene’s Test Sig.

Corrected Model 9.624 15.440 0.000 0.168 Intercept 12.071 19.366 0.000 Familiarity 0.006 0.009 0.924 Trust 32.375 51.943 0.000 Involvement 2.741 4.397 0.038 Valence 4.505 7.228 0.008 Arousal 0.687 1.102 0.296

Firstly, it is clearly shown in the right column of Table 3 that the Levene’s test has a significance of 0.168, which is greater than 0.05, indicating the assumption of

homogeneity of variances has not been violated. Furthermore, Table 3 shows the output of the ANCOVA test ran using NPO perceptions as the dependent variable. It is clear that the model is found to be significant for valence (F = 7.228, p < 0.01) but not for arousal (F = 1.102, p > 0.05). This confirms that there is a difference between the types of framed messages and consumers’ perceptions of the NPO. Although, the separation of the results for valence and for arousal shows and confirms yet again the significant effect of valence and not of arousal.

(25)

5.3.2 Differences between groups for campaign perceptions

These tests were repeated for the dependent variable of campaign perceptions,

considering Hypothesis 3b, which claims perceptions of the NPO’s campaign will differ according to the framing of the message read. This ANCOVA test followed the same method as testing for differences in consumers’ perceptions of the NPO. The Levene’s Test of Equality of Error Variances had a significance of 0.293, ensuring that the null hypothesis (homogeneity of variances) was not rejected. This means that the ANCOVA test could proceed. Table 4 provides the outcome of the ANCOVA test.

Table 4: Test of Between-Subject Effects for Campaign Perceptions Source Mean Square F Significance Levene’s Test Sig.

Corrected Model 7.968 8.884 0.000 0.293 Intercept 15.431 17.205 0.000 Familiarity 1.100 1.226 0.270 Trust 29.062 32.404 0.000 Involvement 5.687 6.341 0.013 Valence 0.001 0.001 0.974 Arousal 1.170 1.303 0.256

As seen at the bottom of the “significance” column, the model was found not to be of significance as F = 0.001 (Valence) and F = 1.303 (Arousal) and p > 0.05 for both. Using these results, it cannot be concluded that there are differences between the various framed messages as to how consumers evaluate the NPO campaign.

5.4 Exploratory Correlation Analysis

Although not stated in the hypotheses, the study has also raised the issue of whether perceptions of an NPO are linked to engagement in eWOM. A correlation test was performed to see if there was any link between these two variables in the data gathered. As both were dependent variables in this study, it is impossible to conclude whether or not NPO perceptions influence eWOM, or vice-versa. It can only be inferred from the correlation test whether or not the variables co-vary. The normality test showed that the

(26)

data was not normally distributed (according to the Shapiro Wilks W test, p > 0.05), the variables are ordinal (measured on a 7-point Likert scale), and the variables have a

positive monotonic relationship. Therefore, a Spearman’s Rank Order correlation test was run. The test found a strong, positive correlation between perceptions of the NPO and likelihood to engage in eWOM (rs = 0.587, p < 0.001). This indicates that the variables do in fact co-vary, which should be taken into account for future research (mentioned in Section 7).

(27)

6. Discussion

This section discusses the interpretations that can be concluded of the results presented in section five, including whether the hypotheses mentioned in section three (conceptual framework) are supported or not, as well as possible reasons for this. Furthermore, the theoretical implications, as well as how NPO managers can use these results in practice will be considered. Lastly, a review of the limitations of this research will be conducted. 6.1 Differences in relation to valence

The significant relationship found in section 5.2.1 regarding the valence of a message, implies that positive NPO communications are more likely to be shared. This supports hypothesis 1, and draws the same conclusion as Berger and Milkman (2012) did. This result shows that there may be some relationship between eWOM theories for non-profit and for-profit organisations, in regard to message framing. The findings also contradict the commonly used non-profit advertising practice of using negative messages to draw consumers’ attention. Though there was a significant relationship between positively framed messages and eWOM, the overall one-way ANOVA model was found not to be significant.

Although the finding of non-significance of the effect of arousal on eWOM mostly caused the non-significance of the ANOVA model, as proven in the subsequent contrasts, it must not automatically be assumed that valence is a large predictor of eWOM. Other possible predictors of engaging in non-profit eWOM may be the

reputation of the website (Park and Lee, 2009), the relationship consumers have with the organisation and how that organisation assists them in advocating for them (Godin, 2009; in Miller, 2009) and the gender of the consumer (Moosmayer and Fuljahn, 2010). Warner, Abel and Hachtmann (2014, p. 11) suggest that marketers need to know their audience and engage with them accordingly, instead of sending out mass messages. They also find that NPOs must provide “content that is relevant, valuable and actionable to key

audiences”, which encourages engagement in eWOM. It is possible that this is another major predictor of eWOM.

(28)

6.2 Differences in relation to arousal

Section 5.2.2 proved that there was not a significant relationship between the level of arousal and likelihood of sharing the communication (engaging in eWOM). The results indicate that hypothesis 2 was not supported. This might be caused by the emotion (sadness) of the “negative, low arousal” communication, as some may feel they would be equally likely to share this message as the “negative, high arousal” communication (channelling the emotion of anger). The findings were not in line with Berger and

Milkman’s (2012) outcomes, although the same emotions (anger and sadness) were used in this study and in their research. This could be due to the disparity between for-profit messages and non-profit messages.

As previously mentioned, consumers may have felt that both situations were equally worthy of being shared, which can be seen in looking at the raw mean data: 3.83 for high arousal and 3.92 for low arousal (1 being very unlikely to engage in eWOM and 7 being very likely to engage in eWOM). Additionally, as mentioned in detail in section 6.1, there may be a number of other factors influencing eWOM, which were not

specifically looked at in this experiment. 6.3 Differences in NPO perceptions

The evidence presented in section 5.3.1 supports hypothesis 3a, that there is a difference between NPO perceptions depending on the communication type (valence and arousal combination) that the participants read. This implies that the way a message is framed can impact the consumers’ perceptions of the organisation. It also means that it is possible to predict organisation perceptions from the variables used. The findings are in line with reasoning from Allsop et al. (2007), and this is further proven by the

significance of all covariates besides familiarity for the model. The differences from the post-hoc test between positive and negative high arousal messages further add to the statistical reliability that positively framed messages are more effective than negatively framed messages. In this case, the positive high arousal messages would have a more positive impact on perceptions of NPOs than negative high arousal messages.

(29)

Familiarity would most likely not have been significant for the model as the organisation and campaign were fictionally created for the purpose of this research, and therefore it was expected that no one would have been familiar with either of these factors. The only factor of familiarity that was relevant to the analysis was that of the issue at hand, which was more suited for the “issue involvement” category.

6.4 Differences in NPO campaign perceptions

Section 5.3.2 states and proves that communication type is not a significant predictor of campaign perceptions (p > 0.05). This means that the way a message is framed will not affect the perceptions consumers have of a campaign. Reasons for the non-significance of this may be because there is very little knowledge of the campaign. Participants read a text, which they may have attributed directly to the NPO, and not so much the campaign. Additionally, it was quite obvious where the money would go to when one would donate to the organisation, and it may have not been clear what role the campaign actually plays within the NPO. Further research could have been done into finding out how the

consumer linked the campaign and organisation, and what they assumed their roles were within the organisation, as both were made-up for the purpose of this study.

Another reason for these findings may be due to the low Cronbach’s alpha, as mentioned in section 5.1. The campaign perceptions alpha for the “Positive, Low Arousal” manipulation was less than 0.7. This could have been a factor in the analysis. Adding more items to the scale, which would have entailed asking the participants more questions relating to campaign perceptions, may have increased the alpha, making the variable more reliable. Alternatively, there may be other reasons why NPO campaign perceptions were not influenced by message type, such as lack of clarity of the campaign, no distinction between the campaign’s purpose and the NPO and the look or feel of the campaign’s presentation not being pleasing for the participant personally.

6.5 Theoretical Implications

This research provides additional data for the debate on whether positive or negative messages influence the probability of engaging in eWOM. It was clearly found that consumers are more likely to spread positive NPO communications than negative. This

(30)

supports the findings of Berger and Milkman (2012) from their study of for-profit companies, but contradicts the results Chang and Lee (2012) found in their study of communications requesting donations for charities. A possible reason for this could be that Chang and Lee’s (2012) emphasis was placed on intent to donate, whereas Berger and Milkman (2012), as well as this study, focused on eWOM. The results of this study can start to bridge the theories of the two articles, creating a path for a new line of reasoning for this area of research.

The statements by Dillard and Peck (2000) concerning how consumers evaluate the brand based on the level of emotion in the communication are also supported by these findings. As it has been shown, perceptions of NPOs differ based on the valence and arousal of the communications, which can directly link to the claims Dillard and Peck (2000) made.

This study provides an initial view on message framing and eWOM for non-profits. This can relate Bickart and Schindler’s (2001) statement regarding the importance of online discussions to the non-profit sector of business. Although message framing and eWOM have been considered together in a variety of contexts, this is one of the first studies to look specifically at valence and arousal effects on eWOM in NPOs. This study did not support Berger (2011) and Lang, Dhillon and Dong’s (1995) claims that arousal explains why a message is shared, nor that high arousal messages are more likely to be shared than low arousal messages (Berger and Milkman, 2012). Due to the little previous research on the subject in this study, many of the theories and findings of studies of more general message framing and eWOM were used. Therefore, the results shed light onto the differences between non-profit and for-profit marketing, and how the two cannot be used or theorised interchangeably; an important note to be made for future researchers in the field. Furthermore, the proof that neither arousal explains why NPO communications are shared, nor that one level of arousal is more likely shared than the other (in NPO

marketing), means that there are other factors or predictors that researchers need to find. 6.6 Managerial Implications

The outcome of this study can be crucial to marketers for NPOs, especially given the human and financial resource limitations typically found in such organisations. Marketers

(31)

must first of all be very careful when framing a communication, as it is proven to affect the perceptions consumers’ have of the organisation, and can also play a role in whether the consumer will participate in eWOM, a vital tool for marketers to have access to and influence over. Marketers should try to frame their communications in a positive way, as it is more likely for such messages to be passed on. Contrasting to the commonly used practice of negatively framed messages, these are statistically proven not to be as

effective for non-profit organisations, when the goal is for consumers’ to share messages with their friends. This could achieve a number of objectives such as increased awareness for the organisation (or issue), more regular followers of the organisation’s activities, or donations to the organisation’s work. Furthermore, more information is needed as to whether high or low arousal is more effective in communications. Although it was found not to be significant in this research, marketers should not ignore the potential effect it may have while they are creating communications.

Managers should also take into consideration how a message may influence consumers’ perceptions of the organisation. As the results show, the way the message is communicated is related to the views consumers have of the NPO. Moreover, marketers should focus on building trust with consumers, as well as staying aware that consumers more involved with the issue at hand will have an influence over the way the organisation is perceived.

6.7 Limitations of the Study

The biggest limitation of this research was controlling and measuring arousal. Initially, the measures of arousal were the same as used in Berger and Milkman (2012), asking on a polar scale whether the participants were feeling mellow, active, high energy, low energy, etc. However, upon running the survey in this manner, participants were

continuously evaluating these feelings in the middle (with a level of 3), possibly as they were unsure of the meaning. Therefore, another attempt was made, using the SAM technique (as explained earlier), which seemed to work out much better, however the extremes of the scales were not quite what was expected either. Although this meant the results could be used, as the texts had been manipulated in an accurate way to trigger such feelings, it seems quite difficult to arouse participants with such a short text and

(32)

image combination. This also leads back to the limitation of an internet-mediated

experiment, which, as mentioned earlier, cannot be too long as otherwise participants will not be willing to complete it. Nevertheless, it is impossible to say whether a longer

communication would have been able to arouse participants more. Therefore, future researchers may consider conducting the research through another means, such as a laboratory experiment where participants need to complete the entire survey, and there are more ways to control for and manipulate arousal (e.g., having participants jump up and down prior to reading the communication).

Another limitation, as previously highlighted, is the low Cronbach’s alpha for NPO perceptions for Positive, Low arousal communications. This may be due to the small number of items used to calculate the items (Pallant, 2002). As deleting items would not increase the alpha at all, it is possible that three items were not enough and reliable to evaluate consumers perceptions of the NPO for this message.

(33)

7. Conclusion

The main purpose of this study was to find the impact of NPO message framing, in terms of levels of valence and arousal, on electronic word-of-mouth. Additionally, the study was looking at how organisation and campaign perceptions were influenced by message framing. The findings show that positive messages have an impact that encourages the consumer to engage in eWOM. It was concluded that level arousal was not able to predict eWOM activity for NPO communications. Further, consumer perceptions of the NPO were shown to be significantly different according to the way the message was framed (in terms of arousal and valence). The positively, high arousal framed messages were more influential than the negatively, high arousal framed messages.

Although not all characteristics of message framing were proven to be influencers of eWOM engagement, the results provide both researchers and managers with a step forward in NPO eWOM research. Due to the relatively unresearched area this study is exploring, the results found proves that one cannot be hasty in applying eWOM theories related to for-profit companies to non-profit organisations.

Furthermore, researchers could also take into account links between for-profit and non-profit eWOM theories. Research in this direction may have been of benefit to this study, in order to create awareness regarding which theories are applicable to NPOs and which are not. Due to the limits of research in this area thus far, this study assumed that the eWOM theories could be applied to NPOs, however, as seen in the case of high and low arousal, there is a discrepancy.

Another possible area of research could be to consider what drives consumers to share NPO communications in general. As shown in the results, the significant

relationships found between valence of a message and eWOM only accounts for a very small percentage of factors driving consumers to share. Therefore, it may be of interest to other researchers to explore additional drivers of engaging in eWOM, such as those mentioned in section 6.1 (including website reputation, gender, and personalised or non-personalised messages). Some of these suggestions were taken from eWOM literature regarding cause-related marketing (CRM), as well as other corporate social responsibility (CSR) initiatives, and therefore should be tested to see this also applies to NPOs. As there are a variety of other possible influencers, researchers should also take into account

(34)

familiarity as a control. As the organisation and campaign were fictional in this study, it was impossible that the participant would already be familiar with these. Further studies could work with established organisations to see how familiarity plays a role in likeliness to share.

(35)

Bibliography

Allsop, D. T., Bassett, B. R., & Hoskins, J. A. (2007). Word-of-mouth research: principles and applications. Journal of Advertising Research, 47(4), 398. Berger, J. (2011). Arousal increases social transmission of information. Psychological

science, 22(7), 891-893.

Berger, J., & Milkman, K. L. (2012). What makes online content viral?. Journal of Marketing Research, 49(2), 192-205.

Bettman, J. R., & Sujan, M. (1987). Effects of framing on evaluation of comparable and noncomparable alternatives by expert and novice consumers. Journal of

Consumer Research, 14(2), 141.

Bickart, B., & Schindler, R. M. (2001). Internet forums as influential sources of consumer information. Journal of interactive marketing, 15(3), 31-40.

Britnell, M. (2011). Increasing Importance of Social Media in Healthcare. KPMG Issues Monitor. Retrieved June 23, 2014, from

http://www.kpmg.com/CH/en/Library/Articles- Publications/Documents/Sectors/pub_20120207_issues-monitor-healthcare-social-media-in-health_en.pdf

Buda, R., & Zhang, Y. (2000). Consumer product evaluation: the interactive effect of message framing, presentation order, and source credibility. Journal of Product & Brand Management, 9(4), 229-242.

Chang, C. T., & Lee, Y. K. (2009). Framing Charity Advertising: Influences of Message Framing, Image Valence, and Temporal Framing on a Charitable Appeal.

Journal of Applied Social Psychology, 39(12), 2910-2935.

Chiou, J. S., & Cheng, C. (2003). Should a company have message boards on its web sites?. Journal of Interactive Marketing, 17(3), 50-61.

Dillard, J. P., & Peck, E. (2000). Affect and Persuasion Emotional Responses to Public Service Announcements. Communication Research, 27(4), 461-495.

Faseur, T., & Geuens, M. (2010). Communicating the right emotion to generate help for connected versus unconnected others. Communication Research, 37(4), 498-521. Field, A. (2009). Discovering statistics using SPSS. Sage publications.

Ganzach, Y., & Karsahi, N. (1995). Message framing and buying behavior: A field experiment. Journal of Business Research, 32(1), 11-17.

Gruen, T. W., Osmonbekov, T., & Czaplewski, A. J. (2006). eWOM: The impact of customer-to-customer online know-how exchange on customer value and loyalty. Journal of Business research, 59(4), 449-456.

Guy, B. S., & Patton, W. E. (1989). The marketing of altruistic causes: understanding why people help. Journal of Consumer Marketing, 6(1), 19-30.

Heller, N., & Reitsema, K. (2010). The Use of Brand Alliances to Change Perceptions of Nonprofit and Private Organizations. Journal of Applied Business & Economics, 11(4).

Hennig‐ Thurau, T., Gwinner, K. P., Walsh, G., & Gremler, D. D. (2004). Electronic word‐ of‐ mouth via consumer‐ opinion platforms: What motivates consumers to articulate themselves on the Internet?. Journal of interactive marketing, 18(1), 38-52.

(36)

Psychology, 16(2), 239-247.

Kahneman, D., & Tversky, A. (1979). Prospect theory: An analysis of decision under risk. Econometrica: Journal of the Econometric Society, 263-291.

Kahn, B. E., & Meyer, R. J. (1991). Consumer multiattribute judgments under attribute-weight uncertainty. Journal of Consumer Research.

Lang, A., Dhillon, K., & Dong, Q. (1995). The effects of emotional arousal and valence on television viewers’ cognitive capacity and memory. Journal of Broadcasting & Electronic Media, 39(3), 313-327.

Lang, P. J. (1980). Self-assessment manikin. Gainesville, FL: The Center for Research in Psychophysiology, University of Florida.

Maheswaran, D., & Meyers-Levy, J. (1990). The influence of message framing and issue involvement. Journal of Marketing Research (JMR), 27(3).

Miller, B. (2009). Community fundraising 2.0—the future of fundraising in a networked society?. International Journal of Nonprofit and Voluntary Sector Marketing, 14(4), 365-370.

Mohr, L. A., & Webb, D. J. (2005). The effects of corporate social responsibility and price on consumer responses. Journal of Consumer Affairs, 39(1), 121-147. Moosmayer, D. C., & Fuljahn, A. (2010). Consumer perceptions of cause related

marketing campaigns. Journal of Consumer Marketing, 27(6), 543-549. Pallant, J. (2002). SPSS survival manual: A step by step guide to data analysis using

SPSS. McGraw-Hill International.

Park, C., & Lee, T. M. (2009). Information direction, website reputation and eWOM effect: A moderating role of product type. Journal of Business Research, 62(1), 61-67.

Pilion Trust's Blog | 7th April 2014. (2014). Retrieved April 13, 2014, from http://piliontrust.com/

Puto, C. P. (1987). The framing of buying decisions. Journal of Consumer Research. Qualls, W. J., & Puto, C. P. (1989). Organizational climate and decision framing: An

integrated approach to analyzing industrial buying decisions. Journal of Marketing Research (JMR), 26(2).

Shin, D., Song, J. H., & Biswas, A. (2013). Electronic word-of-mouth (eWOM) generation in new media platforms: The role of regulatory focus and collective dissonance. Marketing Letters, 1-13.

Smith, M. (2014). Pillion Trust shock charity appeal goes viral with f-bomb - Mirror Online. Retrieved April 13, 2014, from http://www.mirror.co.uk/news/uk-news/pillion-trust-shock-charity-appeal-3388120

Smith, C. A., & Ellsworth, P. C. (1985). Patterns of cognitive appraisal in emotion. Journal of personality and social psychology, 48(4), 813.

Social Networking Statistics | Statistic Brain. (2014). Retrieved from http://www.statisticbrain.com/social-networking-statistics/

Stutely, M. (2003) Numbers Guide: The Essentials of Business Numeracy. London: Bloomberg Press.

Subramani, M. R., & Rajagopalan, B. (2003). Knowledge-sharing and influence in online social networks via viral marketing. Communications of the ACM, 46(12), 300-307.

(37)

INFORMATION. Advances in consumer research, 11(1).

Warner, T., Abel, A., & Hachtmann, F. (2014). Empowered and engaged: Exploring social media best practices for nonprofits. Journal of Digital & Social Media Marketing, 1(4), 391-403.

White, K., MacDonnell, R., & Ellard, J. H. (2012). Belief in a just world: consumer intentions and behaviors toward ethical products. Journal of Marketing, 76(1), 103-118.

Williams, M., & Buttle, F. (2013). Managing Word-of-Mouth: A Nonprofit Case Study. Journal of Nonprofit & Public Sector Marketing, 25(3), 284-308.

(38)

Appendix I

Self-Assessment Manikin (Lang, 1990):

Please note that the bottom row of pictures, representing dominance, was excluded from the study.

(39)

Appendix II

Questionnaire Example (Positive; High Arousal)

I am Morven Piers, a final year of BSc student of Economics and Business at the University of Amsterdam. I greatly appreciate you taking the time to participate in my questionnaire, which is a crucial part of my final year research. The research I am conducting regards the effects of perceptions of Non-Profit Organisations (NPOs) on electronic word-of-mouth (eWOM).

The questionnaire is very short: It should take you 5-10 minutes to complete. The information you provide will be treated with the strictest confidence: This research will solely be used for my thesis, and all your answers and information is anonymous and will not be distributed.

If you have any questions, or feedback, I would be delighted to respond to these, just contact me at morven.piers@student.uva.nl.

Please read the instructions below carefully, as the following questions will be based on the information.

Imagine you came across a piece of information on the usual website that you browse. It has come from a non-profit organisation called "Mothers Against Leukemia (MAL)". This charity supports children and mothers in both rich and poor countries in the fight against juvenile leukemia.

Donations will go to funding the medical treatment for the child, as well as help for their families while they have to give up time working to take care of the children.

Chanda is an inspiration to us all. She lives in a small village in the South of India. She is only 9, but she has spent the last 2 years campaigning for better education in her area, and free school meals for poor children. She speaks from the heart and she is determined to have a good education. She hopes to be a doctor when she's older!

Last year, Chanda was diagnosed with Juvenile myelomoncytic leukemia, a rare form of cancer. This made her even more determined to make a difference. She writes letters to the city hall, talks to children around her neighbourhood, and engages with local businesses to make everyone's lives safe and happy.

Thanks to Mothers Against Leukemia and our One Child at a Time campaign, Chanda was able to get the treatment she needed and her parents were not able to afford, and we are proud to say, her cancer has gone!

Referenties

GERELATEERDE DOCUMENTEN

`The Difference` van de organisatie `Kirabo`. Er is gekozen om maar één campagne te gebruiken en deze dan zelf in een kalme en opwindende campagne te

As discussed above, the results of Experiment 1 suggest that irrespective of pleasure level, affective arousal impacted auditory masked sensitivity as measured by the

Due to the fact that we did find a significant relationship for mastery, and that we didn't find one for achievement goals means that we could say that mastery goals are a better

Voor de gemeente zou het dan ook gemakkelijker zijn om haar eigen doelstellingen te halen omtrent 'meer en beter groen', aangezien uit dit

Overall, we posit that the theory that all threats (i.e., both existential and non-existential) cause similar arousal responses ( Tritt et al., 2012 ; Jonas et al., 2014 ) was

Subjective rating methods performed on the data acquired by these experiments show that the moments of additional heart rate are indeed caused by emotional arousal during

The GS1 data pool offers a vehicle to enable this process, see the IT architecture depicted in Figure 2. This figure gives an overview how product information is shared in FMCG

Twenty-Third European Conference on Information Systems (ECIS), Münster, Germany, 2015 2 users can be cheated and waste their time and money; (b) owners of web pages