Reactions towards random acts of
kindness
_______________________
Master thesis
Msc. in Business Studies – Marketing Track
University of Amsterdam, Faculty of Economics and Business
Author:
Bart Goes
Student number:
10459235
Date of submission: 7-7-2014
Abstract
Random acts of kindness are increasingly successful performed by companies nowadays. The author of this research has made an initial attempt to show how (different) consumers react to random acts of kindness and whether they are as positive as has been widely suggested. Consumer reactions are measured in terms of electronic word-of-mouth (eWOM) and brand attitude. In addition, this study investigates how personality, act of kindness type (random versus strategic) and sincerity influence the impact of acts of kindness on consumer reactions. The developed conceptual model was tested using an online experiment in which participants who read a news article about a fictitious airline had to evaluate the brand and its act of kindness. The results showed that acts of kindness are not always as positive as assumed as it does not generate positive eWOM and sharing behavior, although it is consistent with the characteristics of viral content. Both types of consumers (entity and incremental theorists) did hold very positive brand attitudes after reading about a brand’s act of kindness, but it is unclear to what extent the act of kindness contributed to this positive attitude as participants who read a general description of the brand (without an act of kindness) evaluated the brand just as positive. The results further reveal the mediating role of sincerity in the relationship between acts of kindness and reactions suggesting consumers use sincere attributions to determine their willingness to share and to form subsequent attitudes towards the brand. Finally, no support was found for the suggested moderating roles of personality type and act of kindness type, although this study provides encouraging support for the notion that both may help explain how different consumers react in different ways. So did entity theorists report higher scores on brand attitude when the act of kindness was strategic and were incremental theorists more willing to share truly random acts of kindness. This research concludes with theoretical and managerial implications followed by avenues for future research.
Table of contents
1. Introduction ... 3
2. Literature review ... 7
2.1 Random acts of kindness ... 7
2.2 Implicit theories of personality ... 10
2.3 Random versus strategic acts of kindness ... 13
2.4 Attribution theory and sincerity ... 16
3. Research design ... 19 3.1 Procedure ... 19 3.2 Manipulation check ... 20 3.3 Pre-test ... 21 3.4 Sample ... 22 3.5 Measures ... 22 3.5.1 Dependent variables ... 22 3.5.2 Control variables ... 23 3.5.3 Manipulation check ... 26 3.5.4 Moderator ... 26 3.5.5 Mediator ... 28 4. Results ... 29 4.1 Manipulation check ... 29 4.2 Correlations matrix ... 30 4.3 Tests of hypotheses ... 31 4.3.1 Dependent variables ... 31 4.3.2 Moderation effects ... 34 4.3.4 Mediation effect ... 41 5. Discussion ... 42
5.1 Discussion of the results ... 42
5.2 Theoretical implications ... 47
5.3 Managerial implications ... 49
5.4 Limitations and future research ... 50
6. Conclusion ... 53
7. References ... 55
8. Appendix ... 61
8.1 Experiment ... 62
1. Introduction
Random acts of kindness were featured as one of the biggest consumer trends for 2011 (Trendwatching, 2011). Companies are picking up the trend and with social media as an appropriate platform companies are even more performing random acts of kindness nowadays. One of the major reasons for these companies’ kindness actions is to receive massive positive eWOM (Trendwatching, 2011). Soon a number of companies successfully performed random acts of kindness and received a lot of positive ‘buzz’ while just
experimenting with this new phenomenon. A great example is the ‘How Happiness Spreads’ campaign of the Dutch airline company KLM (Airlinetrends, 2010). In order to turn
passengers’ boredom while waiting for takeoff into happiness, they formed a KLM ‘Surprise Team’. They identified customers who checked in on Foursquare at one of KLM’s venues or posted messages on Twitter that they were flying with KLM. The team did as much research about that person as they can to find a gift that is customized to that particular customer in order to surprise them with a small appropriate gift just before departure. The campaign was a big success as consumers reacted very positive towards KLM’s campaign as well as to the brand itself. Besides, it was massively shared among the web: over one million direct
impressions on Twitter, 5.000 new Facebook friends and their online video was discussed on hundreds of blogs and media (e.g. BBC) as a best practice case on how brands can use social media in a relevant way (Boondoggle, 2014).
However, there is only one study so far (Baskerville et al., 2000) that focused on random acts of kindness, which was performed in a personal interaction context without any involvement of brands. This is interesting as random acts of kindness suggest to be positive as consumers are surprised with a gift or special service treatment. Besides, brand’s kindness actions like KLM’s one are likely to go viral as previous studies (Berger & Milkman, 2012;
Botha & Reyneke, 2013; Eckler & Bolls, 2011; Henke, 2013) have shown that positive content and content that evokes high-arousal emotions are more viral. Also, Derbaix and Vanhamme (2003) demonstrate that there is a significant relationship between surprise and word-of-mouth. They argue that the likelihood of inducing social sharing is high for surprised customers. But, brand’s kindness actions have been widely researched in corporate social responsibility (CSR) literature which shows that good intention behavior is not a guarantee for positive results. So, it is interesting to study if random acts of kindness are always as positive as assumed.
Besides this widely held assumption little is known about the reactions of observers of random acts of kindness. We know that most direct receivers of random acts of kindness logically become happy, but what about people who read messages about it? Are people willing to spread this kind of content? Besides being kind eWOM is a major goal of brands’ random acts of kindness, but companies do not really know the consequences; is it really as positive as they think? We already know from CSR literature that when companies do good consumers not necessarily become happier or more positive (Forehand & Grier, 2003; Kim, 2011; Yoon, Gurhan-Canli, & Schwarz, 2006). Therefore, it is important to know if random acts of kindness actually work. This research fills this gap by determining how different people react to random acts of kindness.
A study by Dweck, Chiu and Hong (1995b) already shows that different types of people react differently to things. Therefore, to be able to determine reactions to random acts of kindness of different people two theories are involved in this study. First, implicit theories of personality are used to divide consumers in two groups to be able to measure reactions of different types of people. Entity theorists believe that personality traits are fixed and expect a high degree of consistency in behavior (Dweck et al., 1995b). In contrast, incremental
theorists believe that personality is malleable and view behavior as varying, either over time
or across situations (Dweck et al., 1995b). Second, attribution theory is used to determine consumers’ reactions to random acts of kindness as the theory suggests that consumers assign causes for managerial actions (Folkes, 1988). In other words, different people may attribute different reasons or motivations to infer the brand’s kindness action.
Elaborating on these theories, this research makes a number of theoretical
contributions. First, I shed light on how consumers react to brands’ random acts of kindness. This research enriches the theoretical understanding of implicit theories of personality and attribution theory. The first by providing insights into how different types of people (entity versus incremental theorists) react differently to random acts of kindness, the latter by determining how those different people attribute different motivations to a brand’s kindness action. Second, this research shows which consequences those reactions might have for consumers’ attitudes towards a brand. It also complements recent studies by expanding our knowledge about the type of content being shared online, specifically for random acts of kindness. Additionally, this research provides the first empirical look into reactions of observers towards random acts of kindness and its consequences.
Finally, this research also has an important managerial implication. Whereas until now marketers used random acts of kindness as mere experiments, this study deepens marketers’ understanding of this phenomenon, what type of reactions one might expect when using campaigns like this and its influence on brand evaluations.
The research question is:
‘Do consumers respond differently to random acts of kindness dependent on their type of personality?’
This paper begins with a literature review in the areas of online sharing behavior, implicit theories of personality and attribution theory. Based on this research a set of hypotheses are formulated. Next, the methodology for testing the hypotheses is described, followed by the results and discussion. The paper concludes with the implications of the findings and suggestions for future research as well as possible limitations to the study.
2. Literature review
To answer the formulated research question this research makes use of some theoretical concepts. Implicit theories of personality, attribution theory and sincerity form the theoretical background for hypotheses regarding reactions towards random acts of kindness.
2.1 Random acts of kindness
Random acts of kindness can be generally defined as “a something one does for an unknown other that they hope will benefit that individual, like giving a stranger a flower” (Baskerville et al., 2000). As mentioned before, random acts of kindness were ranked as the number one consumer trend for 2011 by trend firm Trendwatching. Their more specific (marketing) definition describes random acts of kindness as “any acts of kindness by brands that will be gratefully received by consumers long used to (and annoyed by) distant, inflexible and self-serving corporations. For brands, increasingly open communications both with and between consumers (especially online), means that it has never been easier to surprise and delight audiences with random acts of kindness: whether sending gifts, responding to publicly expressed moods or just showing that they care” (Trendwatching, 2011). According to
Trendwatching (2011) this trend is becoming of great interest as audiences publicly disclosing more and more personal information on for example Facebook. They provide information about their lives, moods and whereabouts, and more consumers are sharing their experiences with friends and wider audiences on social networks, which means that random acts of kindness can spread far beyond the original recipients. Brands hope to benefit from multiple positive consequences that can be made possible by brand’s kindness actions. For example, as with CSR initiatives, brands may want random acts of kindness to contribute to successfully change their images or enhance reputations. Certainly in today’s world of massive social
media usage, companies are provided with opportunities to evoke more favorable attitudes towards their brands. Surprised and happy consumers will positively talk about the brand, spread the word on the web and the brand itself can for example send out a single message or video in which they show what they are doing. In this way, acting kind to consumers can lead to positive eWOM and viral messages which enhances the brand’s image. Besides, random acts of kindness can also help to develop positive relationships with consumers and other stakeholders. Studies from CSR literature like Creyer and Ross (1997) show that as a result of kindness actions consumers are more willing to give incentives to these companies.
Eventually, this may lead to switching to a brand or buy products from it simply because of its kindness actions (Smith & Alcorn, 1991).
The remaining question, however, is whether random acts of kindness are as positive as assumed. A prediction of its positive effects in terms of eWOM can be made by looking at previous studies that focused on general types and characteristics of content that goes viral. There is consensus among researchers that positive and emotional content is likely to go viral. So did Berger and Milkman (2012) find that positive content and content that evokes high-arousal emotions, whether positive (e.g. awe) or negative (e.g. anger), are more viral.
Deactivating emotions (e.g. sadness), on the contrary, are less viral. In a similar vein, Henke (2013) found that individuals who experience flow, defined as emotional engagement, are significantly more likely to pass along, download or purchase content, whether the content is pleasant or unpleasant.
It is interesting to see whether these results are consistent among different types of content. Therefore, results of some other studies are provided to see whether the same
characteristics of viral textual content apply to content that includes for example videos. So do Eckler and Bolls (2011) suggest that video’s with a positive emotional tone (i.e. those viewed as pleasant) result in favorable attitudes towards the ad and the brand, and intent to forward
the viral ad. Ads with mixed emotional or negative tones lead to less favorable attitudes. A study by Botha and Reyneke (2013), conducted in the same context, suggests two
determinants of sharing content: emotions and familiarity with the content. They conclude that emotions may not always be the first determinant to share content. They argue that in case of specific content, viewers’ feelings about a video firstly depend on their familiarity with the content, where familiarity leads to stronger emotional reactions and unfamiliarity to little, if any, emotional reaction. So, according to Botha and Reyneke emotion is the second determinant of sharing videos when viewers are confronted with specific content. On the other hand, when the content of the video is general, emotions are the first determinant of sharing. In line with previous studies they conclude that a positive emotional reaction
increases the likelihood to share content, but their conclusion that negative ones decrease the likelihood to share is a bit contradicting with for example Berger and Milkman’s (2012) findings.
Remember the focus in this research is on reactions regarding random acts of kindness. Looking at the type of content and characteristics of these kindness actions, prior research suggest that in general people will respond positively towards these actions. First of all because random acts of kindness are positive: a consumer receives a present, special service treatment or something like that. In addition, as consumers are suddenly approached by a brand and receive something positive, a high-arousal emotion is evoked: surprise. Derbaix and Vanhamme (2003) show that surprise leads to social sharing and the studies just mentioned show that this type of content is likely to go viral in an online context, which means that observers are likely to pass along content like this. So, based on previous research that shows what type of content goes viral, this research hypothesizes:
H1: Random acts of kindness will lead to positive eWOM.
It is not only hypothesized that random acts of kindness in general will lead to positive eWOM, but it is also expected to lead to positive brand attitudes. CSR studies (Brown & Dacin, 1997; Handelman & Arnold, 1999) provide evidence revealing that consumers create goodwill towards companies who perform campaigns based on good intention behavior. As mentioned previously, consumers are more willing to give incentives to companies with kindness actions. Besides, studies (Brown & Dacin, 1997; Handelman & Arnold, 1999; Smith, 1996) show that consumers are more likely to buy from a company that is socially responsible, positive associations about social responsible companies can enhance product and brand evaluations and marketing actions with a social dimension generate consumers’ support for the organization. The same positive effects regarding brand attitude are in general expected for random acts of kindness as it is similar to CSR initiatives by performing
kindness actions and to some extent being socially responsible. Therefore, this research hypothesizes the following:
H2: Random acts of kindness will lead to positive brand attitudes.
2.2 Implicit theories of personality
Although random acts of kindness may infer by its name and definition it is truly random and focused on surprising consumers, these actions as well serve other purposes just as any marketing campaign. In line with other kindness actions like CSR initiatives, companies’ good intention behavior does not solely serve the public good but also stems from self-centered motives. This results in consumers having different evaluations of companies’ kindness actions (Kim & Lee, 2012). Therefore, implicit theories of personality are used to divide consumers into two groups to measure those different evaluations of different types of people. First, some information is provided about this theory in general.
Implicit theories of personality are lay beliefs about the malleability of traits and attributes regarding the self and the environment (Poon & Koehler, 2008; Skowronski, 2002). Those beliefs about the malleability of personality play a key role in social judgments and reactions (Dweck, Hong, & Chiu, 1993). Previous studies (Dweck et al., 1995b; Poon & Koehler, 2008) propose that implicit theories develop as a result of socialization practices, accumulated personal experiences that highlight trait versus situational forces and everyday contextual cues. Implicit theories can be activated through changes in real-life occurrences, exposure to persuasive arguments or media evidence that advocates a particular theory view (Dweck et al., 1995b; Levy, Chiu, & Hong, 2006). Implicit theories influence how people use personality trait information in order to make inferences about the causes of behaviors
(Dweck et al., 1995b; Plaks, Levy, & Dweck, 2009). Moreover, they affect the way people form judgments of others from a particular instance of behavior (Plaks et al., 2009).
As mentioned before, implicit theories of personality classify people in two groups: entity theorists and incremental theorists. People who believe that personality is fixed and determined by static traits are referred to as entity theorists. They suppose that a behavior observed in a specific situation is a reliable indicator of a corresponding personality trait (Chiu, Hong, & Dweck, 1997). Those who believe that personality is malleable are referred to as incremental theorists. They think that behavior and traits can change over time and across situations. Thus, they base their judgments not solely on one specific trait or situation but more on contextual information.
This research expects a moderating role for implicit theories of personality, which has already been showed by a study that shows that the theory moderates the effect of brand anthropomorphization (Puzakova, Kwak & Rocereto, 2013). They find that entity theorists view anthropomorphized brands that undergo negative publicity less favorably than
nonanthropomorphized brands. In contrast, incremental theorists are less likely to devalue an
anthropomorphized brand from a single instance of negative publicity. In line with the premise that different types of people (entity versus incremental theorists) react differently to things (Dweck et al., 1995b), this research expects implicit theories of personality to play a moderating role between random acts of kindness and reactions in terms of willingness to share and attitude towards the brand. This research expects consumers with different theories of personality to make sense of random acts of kindness in different ways. Specifically, entity theorists are likely to view a brand’s random act of kindness as an indicator of stable positive brand characteristics. They view a brand’s kindness action as corresponding to the brand and thus perceive the brand as positive and kind (Chiu et al., 1997). Their perception is fixed and does not change so they are likely to stay positive and are valuable in the long run. They are likely to consider a single positive action of a brand as a stable and ongoing type of behavior of that brand on which they can rely to make predictions about future positive brand
performance (Ross & Nisbett, 1991). Therefore, given that entity theorists perceive a brand’s kindness action as positive and a stable manifestation of the brand’s characteristics, they are likely to evaluate the brand better. On the other hand, incremental theorists do not base their judgements on one trait or situation and expect variation in future behaviors after being exposed to a brand’s single kindness action (Dweck et al., 1995b; Levy, Plaks, Hong, Chi, & Dweck, 2001). This means that they are likely to see a brand’s random act of kindness as unstable and temporal. They perceive a random act of kindness as a one-off; only performed this one time, which means it does not correspond to the brand. So, although they might perceive the brand’s kindness as positive, incremental theorists are not likely to rely on a single, positive act to form an overall judgment of the brand, instead they need more examples of behavior to render a robust judgement (Plaks, Stroessner, Dweck, & Sherman, 2001). So, they are not likely to evaluate the brand better because of its single random act of kindness. Following this line of argument, this research hypothesizes the following:
H3: Personality will be a moderator of the relationship between random acts of kindness and reactions towards it in terms of willingness to share and brand attitude. H3a: Entity theorists are more positive towards willingness to share and brand attitude. H3b: Incremental theorists are less positive towards willingness to share and brand attitude.
2.3 Random versus strategic acts of kindness
As already assumed random acts of kindness do not solely serve the public good as brands may as well serve their own interests with their kindness actions. This means random acts of kindness do not have to be necessarily random. They can be part of a bigger, surrounding marketing campaign and thus may in that case be better referred to as a strategic act of kindness (to avoid confusion, from now on random acts of kindness in general are referred to as acts of kindness, truly random ones as random acts of kindness and strategic ones as strategic acts of kindness). Just like KLM describes their act of kindness as looking for ways to surprise their customers and make them happy, and eventually, discover how this happiness spreads (SocialTimes, 2011). In other words, they are constantly looking for ways to interact with customers. This is an indication that acts of kindness are not necessarily random and in some cases they may be seen more as a strategic move. Kim and Lee’s study (2012) already shows that the existence of firm-serving motives regarding companies’ kindness actions results in different evaluations of that particular company.
An interesting question for companies is whether consumers are willing to accept firm-serving motives for their kindness actions. CSR literature has focused on this issue and studies yield different results. Some studies suggest that consumers’ perceptions of firm-serving motives negatively influence a company’s CSR effectiveness and hence an
unfavorable evaluation of the company, while public-serving motives positively influence its
effectiveness (Campbell & Kirmani, 2000; Ellen, Mohr, & Webb, 2000; Webb & Mohr, 1998). This may be explained as consumers use the existence of firm-serving motives as a cue to their attitude towards the company. Consumers would ideally like to see pure
public-serving motives behind a company’s actions and any deviation from that is viewed negatively resulting in a less positive reaction (Forehand & Grier, 2003). An alternative explanation is that the negative evaluation of firm-serving motives may also be caused as companies’ strategies seem deceptive or manipulative (Campbell, 1995; Forehand, 2000). Whether consumers perceive firm-serving motives is not so important but rather whether the perceived motives are contradictory to the company’s stated motives.
On the other hand, results of other studies indicate that consumers are willing to accept and give reputational credit for firm-serving motives, as long as they also perceive that the companies are sincere in serving public interests (Ellen, Webb, & Mohr, 2006; Kim & Lee, 2012; Sen, Bhattacharya, & Korschun, 2006). Even though the salience of firm-serving motives negatively influences consumer responses in general, such impact should be considered in relation to the apparently greater salience of public-serving motives. This indicates that as consumers learn more about CSR and companies’ motivations, they are increasingly willing to accept the main purpose of business (having firm-serving motives) in society. They believe that kindness actions like CSR initiatives can and should serve both the needs of society and companies. In line with this, Forehand and Grier (2003) argue that consumers do not respond negatively to firm-serving motives per se, but rather to marketing strategies whereby the company is being deceptive about its true motives. Any discrepancy between consumers’ perceived CSR motives and a company’s publicly stated motives will trigger scepticism and feelings of being deceived, which in turn will drive negative reactions to the CSR activities. Forehand and Grier (2003) show that when companies also
communicate their firm-serving motives besides their public-serving motives, companies can
inhibit scepticism, enhance the credibility of its CSR message and generate goodwill. So, if companies are forthright about their motives and publicly state the potential self-serving benefits of their actions, consumers are willing to accept firm-serving motives.
In this research, reactions regarding random acts of kindness are measured among entity and incremental theorists. It is expected that the previous argumentation has to some extent an impact on their reactions. In line with H3a, this research assumes that entity
theorists react positively towards random acts of kindness. They see the brand’s kindness as a consistent characteristic of and corresponding to the brand, which leads to a positive
evaluation of the brand. Their perception will not change when a brand’s kindness action turns out to be a strategic move because entity theorists believe personality traits are fixed (Dweck et al., 1995b). It is expected that they will accept firm-serving motives as long as the brand serves the public good. As long as they do that, entity theorists will believe the kindness action is a reflection of the brand’s personality. For incremental theorists this research
assumes that they will react less positively towards truly random acts of kindness. They perceive the brand’s single kindness action as a one-off which is not corresponding to the brand. They need more examples of behavior to form an overall judgement of the brand (Plaks et al., 2001), which therefore means that they do not evaluate the brand better. However, their perception is expected to change when a brand’s act of kindness turns out to be strategic. A temporarily campaign in which the brand is acting kind in order to serve also its own interests, is in line with incremental theorists’ line of thinking that personality is malleable and behavior varies, either over time or across situations (Dweck et al., 1995b). When a brand’s act of kindness is strategic, it can indeed be seen as a one-off, which means that this corresponds whit incremental theorists’ perception that it does not stand for the brand and thus that personality can change. Besides, they are likely to value the brand’s firm-serving motives in order to enhance business. It is just one campaign that is part of a long-term
strategy in order to improve the brand’s image for example, and to reach that goal the brand may perform different campaigns with varying approaches.
So, in general this research hypothesizes the following:
H4: Types of acts of kindness (random versus strategic) moderate the moderated relationship of H3 where personality is the moderator between acts of kindness and reactions towards it.
For entity theorists, this research hypothesizes the following:
H4a: Entity theorists will react positively when a brand’s act of kindness is truly random.
H4b: Entity theorists will react positively when a brand’s act of kindness is strategic.
For incremental theorists this research hypothesizes the following:
H4c: Incremental theorists will react less positively when a brand’s act of kindness is truly random.
H4d: Incremental theorists will react more positively when a brand’s act of kindness is strategic.
2.4 Attribution theory and sincerity
How people react to acts of kindness may not solely depend on whether they see a brand’s characteristics and behavior as fixed or malleable. In literature, one key mediator found before between consumer reactions towards companies’ good intention behavior is sincerity. So did for example Yoon et al. (2006) stress the mediating role of perceived sincerity of company’s motives underlying CSR activities in determining its effectiveness. They show that sincere
motives improve a company’s image, insincere motives hurt a company’s image and when the sincerity of motives is ambiguous, CSR activities are ineffective.
One way to explain sincerity’s mediating role is through attribution theory. Although most attribution theory research has been conducted in social psychology, applying attribution theory to consumer behavior issues is quite common. It appears to be particularly helpful in understanding a receiver's interpretation of a sender's motives for communicating certain information. In marketing research, attribution theory suggests that consumers assign causes for managerial actions (Folkes, 1988). Processes by which individuals evaluate the motives of others are addressed and the theory explains how these perceived motives influence
subsequent attitudes and behavior (Forehand & Grier, 2003). Previous studies (Campbell, 1995; Campbell & Kirmani, 2000; Ellen et al., 2000; Forehand, 2000; Webb & Mohr, 1998) have found that consumers draw inferences about marketer motives and that attributions of marketer motives impact subsequent evaluations of the company.
In this research, sincerity may play a crucial role regarding the type of motivations people attribute to a brand’s act of kindness. Literature in CSR already revealed that when companies do good consumers not necessarily become happier or more positive (Forehand & Grier, 2003; Kim, 2011; Yoon et al., 2006). These studies have shown that sincere consumer perceptions regarding a brand’s kindness are very important as insincere perceptions lead to negative evaluations of the brand. So, in general, positive evaluations regarding random acts of kindness are only possible when the brand’s kindness action is perceived as sincere. In contrast, perceived insincerity does not lead to favorable evaluations. Moreover, perceived insincerity may even backfire, leaving the brand with more negative evaluations than would have been the case without its kindness action (Yoon et al., 2006). In addition, research in psychology (Klein & Dawar, 2004; Sen, Bhattacharya, & Korschun, 2006; Yoon et al., 2006) has shown that publics’ perceived attributions about the sincerity of the CSR purposes is most
important for the success of a CSR program. This means that when consumers do not perceive a brand’s act of kindness as sincere, they will not be able to evaluate the brand’s kindness as favorable. Therefore, sincerity will mediate consumers’ reactions regarding acts of kindness.
Hence, this research proposes the following hypothesis:
H5: Sincerity is the mediator between acts of kindness and reactions towards it in terms of brand attitude and willingness to share.
Figure 1 presents the overall conceptual model of this research.
Figure 1: Conceptual model
Personality (entity vs. incremental)
Acts of kindness
(random vs. strategic) Sincerity
Willingness to share Attitude towards the brand 18
3. Research design
3.1 Procedure
Data is gathered through an online experiment, primarily because it requires far less time to design and conduct, which is important given the limited time period of this research. Besides, the target group (students) is easily approachable online. The experiment is further
characterized by a between-subjects design. Here, participants are either part of the control group or the treatment group, but they cannot be part of both. This design fits this research as it consists of three different groups.
More specifically, this research is an experimental vignette study. Vignette studies use short descriptions of situations or persons (vignettes) that are usually shown to respondents within surveys in order to elicit their judgments about these scenarios (Atzmüller & Steiner, 2010). This approach fits this research as its aim is to show short descriptions of three
different situations. These situations are each formed by a news article representing one time a general description of the brand and two times a brand performing an act of kindness. The two versions in which an act of kindness is performed are further distinguished by the type of kindness action: random or strategic. So, version 1 gives a general description of the brand and its activities, without mentioning any act of kindness (the control group). In version 2 the act of kindness is truly random, focused on surprising customers and making them happy because the brand really cares. Finally, version 3 describes a strategic act of kindness, focused on the brand intentionally developing and performing a campaign to surprise customers. Participants were randomly assigned to one of the three versions.
The brand involved in this research is a fictitious airplane company, called Universal Airlines. It is a fictitious brand so participants do not hold existing or previous attitudes towards the brand which may bias them. Besides, this research has chosen to use an airline
company because the service approach used in this research is frequently used in this business-to-consumer industry (Accenture, 2009). Service is very important in the airline industry, a study by Airline Quality Rating (2012) reports that customer service is even the most important consideration for air travel consumers. Airline companies need to significantly improve their customer experience and often use activities around their services to boost customer loyalty (Accenture, 2009). Therefore, an airline company seems to be the most appropriate brand for this research.
3.2 Manipulation check
The approach of the manipulation in this research is similar to that applied by Puzakova et al. (2013), who study the effects of brand anthropomorphization on consumers’ brand attitudes and brand performance. The authors manipulated the anthropomorphism of the brand by print advertisements and its accompanying text. In this research, the text of the news article is manipulated. More specifically, this research manipulates the salience of firm-serving benefits in the news article about the brand’s act of kindness. The style, design and length of the three news articles are all the same (see appendix 1 for all news articles). So, the only difference is the focus of the article in the last few sentences (no act of kindness versus random act of kindness versus strategic act of kindness).
In short, the news article is about the airline company that faces a travelling peak during the Easter holidays. The article starts with some general information about the busyness like the total number of passengers. Next, the marketing manager talks about how the brand copes with the busy period emphasizing their high service, like adding more staff to accommodate all passengers with caring service and comfort. From this point onwards the articles differ. First, version 1 (control group) continues to focus on the brand’s everyday attempt to deliver their service professionally, like navigating passengers as smoothly as they
can. This text presents a general picture of the brand as a high service company, also in busy days like the Easter holidays. Version 2, including a random act of kindness, continues that the brand ‘at random surprised their customers with some small, personalized gifts’. The focus here is on the fact that the brand initiated the gift-giving spontaneously (without a plan) and that it pleases the brand to make their passengers happy with unexpected kindness. Finally, version 3, including a strategic act of kindness, continues that the brand as usual offered gifts during the Easter holidays. The focus is on the tradition of gift-giving of the brand during this period and that it pleases the brand to make their passengers happy with this tradition of theirs.
3.3 Pre-test
A pre-test among 30 participants (70% male, 90% aged between 17 and 35) was conducted to ensure that the experimental manipulation of the news articles worked. First, by means of a summarizing question (for more details see chapter 3.5.3) it was checked whether respondents indeed distinguished between the news article without an act of kindness (control group) and the two treatments with an act of kindness. All participants correctly summarized what they had been reading. Second, it was checked if the participants also distinguished between the random and strategic act of kindness. An independent samples t-test showed that the two treatments did not differ significantly at the .05 level (p = .081). Despite this finding, in agreement is decided to not adjust the manipulation as there is a limit to making seem initiatives totally random.
The pre-test was further used to check the reliability of the different scales. Reliability analyses in SPSS showed that all scales (brand attitude, willingness to share, sincerity,
perceived motives, manipulation check, credibility, realistic check, familiarity, likeability and personality) were reliable as all cronbach’s alpha (from now on α) scores were bigger than .7.
3.4 Sample
The sample consists of students and acquaintances. A total of 17 respondents were dropped out because of missing values or they misinterpreted the news article. So, the final sample size is 159 respondents (59.1% men and 40.9% women) of which 67.3% of all respondents fall into age group 17-25 (86.8% is aged between 17 and 35). The author chose to approach primarily students to participate in this research. The main reason is that students are most online and the heaviest users of social media (CBS, 2012). Research by CBS (2012) reveals that 93% of people between 12-18 years old use social media and people aged between 18-25 years old use social media even more (98%). In contrast, only 50% of people aged 55 and up uses social media. Hence, students best fit this research to collect the best possible results.
3.5 Measures
All the different variables are explained below. Before variables were added, a literature search was conducted in the hope of finding relevant existing measurement instruments. The measures used to operationalize the variables were adapted from relevant studies, when available, with changes in wording to fit the target context.
3.5.1 Dependent variables
After reading the news article, consumers’ attitude towards the brand was measured on a five-item, seven-point differential scale (1 = “negative, dislike, bad, unfavorable, unpleasant” and 7 = “positive, like, good, favorable, pleasant”) (Coulter & Punj, 2007; Puzakova et al., 2013). The items were averaged to form a total brand attitude score on this scale (α = .92).
Next, participants rated their willingness to share the article on a three-item, seven-point Likert scale (1 = “strongly disagree” and 7 = “strongly agree”). Specifically, two items were used from Eckler and Bolls’ (2011) research with small changes in wording to fit the
target context: “I will pass on this news article to others” and “I will recommend this news article to others”. The third item that was added to this scale was “I am willing to share this news article on my social media account (e.g. Facebook)”. The addition of this last item is important as this research is particularly interested in online sharing behavior because acts of kindness are performed in combination with social media. The three items were averaged to form a willingness to share index and together form a reliable scale (α = .85).
3.5.2 Control variables
Credibility
Participants indicated their level of agreement with two statements concerning the credibility of the information about the airline company. The two statements were adopted from research by Yoon et al. (2006) and are as follows: “I believe the information I have read about
Universal Airlines in the news article” and “The information about Universal Airlines is credible”. The two statements, measured on a seven-point Likert scale (1 = “strongly disagree” and 7 = “strongly agree”), were averaged to form a credibility index (α = .88).
Realistic check
Two questions were added to check whether respondents, especially in the act of kindness treatments, perceived the brand’s activities as realistic. Although a fictitious airline brand is used in this research, it is very important that participants perceive the setting as realistic. An unrealistic setting may tend respondents to give random answers as they do not believe the scenario described. That is why the news article is identical in its length and style to a news article of an existing source. The article also includes an existing source name (Dutch
newspaper “De Telegraaf”) and a marketing manager who is being interviewed. Because the scenario takes place during the Easter period, the article is even more realistic as the
experiment is actually performed one day after the Easter weekend, as if the news article was actually released one day after the Easter weekend in an existing Dutch newspaper.
In order to check respondents’ perceived reality of the story, two items in the control variables section are included. Existing measurement instruments from Ribbens and Malliet’s (2010) study were adapted, with changes in wording to fit the target context, and resulted in the following items: “The scenario described in the article is realistic” and “I can imagine the scenario described in the article happens in real life”. The two statements, measured on a seven-point Likert scale (1 = “strongly disagree” and 7 = “strongly agree”), were averaged to form a realistic check index (α = .88).
Familiarity
Participants were asked to indicate their level of familiarity with acts of kindness. This construct was measured as it may influence a respondent’s answers. For example, a
respondent that is very familiar with these activities may perceive it as a general activity, not very special or unique, while someone who is not familiar with acts of kindness may perceive it as an unique action by a brand. Familiarity was measured according to the experimental design suggested by Puzakova et al.’s (2013) research where they use the following two items: “I am familiar with companies that surprise their customers with gifts” and “I have seen companies surprising their customers with gifts before”. These items, measured on a seven-point Likert scale (1 = “strongly disagree” and 7 = “strongly agree”), were averaged to form a familiarity index (α = .86).
Likeability
The extent to which participants liked the airline’s act of kindness may also influence their answers. Participants who do not like it may have more negative attitudes and perceptions
regarding the brand than the ones who do like it. That is why this construct is measured on a seven-point Likert scale (1 = “strongly disagree” and 7 = “strongly agree”), consisting of the following items: “I like companies that surprise customers with gifts or special service treatments” and “I would like to be surprised by a company with a gift or special service treatment”. These two items were averaged to form a likeability index (α = .88).
Demographics
Three demographic questions were asked at the end of the experiment. These questions may help the author to determine what factors may influence a respondent’s answers, interests and opinions. First, participants were asked to indicate their gender and next they had to fill in their age. The author has chosen to classify age into six groups for the main reason that students are not of one specific age, but usually fall into the age category of 17-25. Besides, this group is of special interest for this research. The other age categories were “0-16”, “26-35”, “36-45”, “46-55” and “55+”. Finally, participants were asked to select their highest level of education completed by choosing one of the following: “Elementary school”, “High school”, “Intermediate vocational education”, “Bachelor” and “Master”.
Time spend online
The last question of the experiment asked participants to indicate how much time they spend online on average per day. Here, online does not only mean visiting websites but also
activities on Facebook, Twitter, etc. Respondents had to choose between the following five groups: “0-30 minutes”, “30-60 minutes”, “1-2 hours”, “2-3 hours” and “ >3 hours”.
3.5.3 Manipulation check
Two manipulation checks were included. The purpose of the first and most important one was to control if participants distinguished between the control group (no act of kindness) and the two other treatments (including a random or strategic act of kindness). Immediately after reading the news article, participants were asked to summarize in one or two sentences what they thought the article was about. In the case of the treatments including an act of kindness, responses had to be related to the act of kindness, mentioning for example the gift-giving itself or a marketing campaign or something like that. Responses that were not related, for example those only mentioning the busyness at the airport because of the Easter period, were dropped out because they misinterpreted the message of the news article.
The purpose of the second manipulation check was to verify whether participants in the two treatments including an act of kindness distinguished between a random and a
strategic act of kindness. The following four statements formed a scale for this manipulation: “Universal Airlines’ act of kindness is delivered spontaneously”, “Universal Airlines’ act of kindness is delivered strategically”, “Universal Airlines’ act of kindness is random” and “Universal Airlines’ act of kindness is planned”. The first and third item, corresponding to the randomness of the act of kindness, were recoded. In this way, a high score indicates a stronger belief in the strategicness of the act of kindness. The four items were measured on a seven-point Likert scale (1 = “strongly disagree” and 7 = “strongly agree”) and together formed a reliable scale (α = .79).
3.5.4 Moderator
The moderating effect of implicit theories of personality has already been showed by
Puzakova et al.’s (2013) study where it moderates the effect of brand anthropomorphization. In line with the premise that different types of people (entity versus incremental theorists)
react differently to things (Dweck et al., 1995b), this research expects implicit theories of personality to play a moderating role between acts of kindness and reactions towards it in terms of willingness to share and brand attitude. Belief in entity or incremental theories of personality was assessed using the Implicit Persons Theory Measure (Levy, Stroessner, & Dweck, 1998). Respondents rated the extent to which they agree on eight statements, four statements representative of entity theory (E) and four representative of incremental theory (I). The eight statements, measured on a seven-point Likert scale (1 = “strongly disagree” and 7 = “strongly agree”), are as follows:
• Everyone is a certain kind of person, and there is not much that they can do to really change that. (E)
• The kind of person someone is, is something basic about them, and it can’t be changed very much. (E)
• People can do things differently, but the important parts of who they are can’t really be changed. (E)
• As much as I hate to admit it, you can’t teach an old dog new tricks. People can’t really change their deepest attributes. (E)
• People can change even their most basic qualities. (I: reverse coded)
• Everyone, no matter who they are, can significantly change their basic characteristics. (I: reverse coded)
• People can substantially change the kind of person who they are. (I: reverse coded) • No matter what kind of person someone is, they can always change very much. (I: reverse coded)
Disagreement with the entity items represents agreement with the incremental items, and the other way around (Levy et al., 1998). Responses were averaged to these eight items to
create an implicit person theory index for each participant (α = .92). The responses to the incremental statements were recoded, hence a high score indicates a stronger belief in entity theory.
3.5.5 Mediator
Sincerity has been recognized as a mediator between consumer reactions towards companies’ good intention behavior in many disciplines, including CSR (Yoon et al., 2006) and
psychology (Klein & Dawar, 2004; Sen, Bhattacharya, & Korschun, 2006; Yoon et al., 2006). The focus in this study is on sincerity referring to whether or not the brand really cares about its customers and thus has an honest interest in the kindness action (Yoon et al., 2006). Sincerity was measured on a seven-point Likert scale anchored by “extremely unlikely” versus “extremely likely”. Participants indicated attributions about the sincerity of the
company’s purposes for pursuing the act of kindness through responses to the following three statements: “Universal Airlines has genuine concerns for its customers and their happiness”, “Universal Airlines sincerely cares about its customers and their happiness” and “Universal Airlines is an honest company”. These measures were averaged to form a sincerity index (α = .82).
Also, participants indicated their attributions about the airline company’s motives for pursuing an act of kindness. This variable, called perceived motives, was measured according to the experimental design suggested by studies of Ellen et al. (2006) and Yoon et al. (2006). The three statements, measured on the same Likert scale as sincerity, were as follows:
“Universal Airlines tried to make a good image of the company”, “Universal Airlines tried to improve its existing image” and “Universal Airlines wants to attract more customers through their act of kindness”. These three measures were averaged to form a perceived motive index (α = .85).
4. Results
4.1 Manipulation check
To make sure that the manipulation of the act of kindness worked, participants had to summarize what they read in one or two sentences. Participants in the control group focused their summary of the news article on the high service of the airline company performed in an extremely busy period, which was as intended. Besides, almost all participants in the act of kindness treatments summarized their news article (in which an act of kindness was
performed) correctly. Everyone recognized in their own way a marketing campaign and most of them directly referred to the gift-giving. Only two participants (both in one of the act of kindness treatments) were dropped out because of a misinterpretation of the news article. They completely ignored the marketing campaign and gift-giving and instead solely focused on how busy the period was for the airline company. So, only two participants were dropped out of an initial sample size consisting of 176 participants. This means that 98.9% of the whole sample correctly summarized the message of the article. Given this high score, it can be concluded that participants indeed distinguished between the news articles of the control group (without an act of kindness) and the two other treatments (including a random or strategic act of kindness). Hence, the main manipulation check worked.
On the other hand, participants did not fully distinguish between the randomness and the strategicness of an act of kindness. An independent samples t-test did not show a
significant difference between the random (M = 5.22, SD = 1.28) and strategic act of kindness (M = 5.47, SD = .93; t(90) = -1.14, p = .256). An explanation for this might be that consumers critically evaluate the randomness of a marketing campaign. They may believe that a
marketing campaign is always prepared. Besides, familiarity may also provide an explanation here. The average score (M = 4.35) on the scale that measures the extent to which consumers
are familiar with acts of kindness is bigger than half the scale (3.5), suggesting that more consumers are familiar with acts of kindness than they are not. This might increase
consumers’ perceptions of the strategicness of the act of kindness and decrease the possibility that they perceive it as unique as they have seen or heard about it before.
Finally, participants viewed the articles and its story as realistic. Descriptive statistics of the realistic check showed that participants perceived the scenario described as realistic and they can imagine the scenario happening in real life (M = 5.18, SE = 1.11).
4.2 Correlations matrix
Descriptive statistics and correlations between the variables are presented in table 1. The first noteworthy correlation for this study is that between sincerity and both willingness to share (r = .37, p < .01) and brand attitude (r = .55, p < .01). In line with H5, sincerity would play a mediating role and suggests that when consumers perceive an act of kindness as sincere, more positive reactions in terms of willingness to share and brand attitude are the result.
The dependent variables together, willingness to share and brand attitude, also show a positive correlation (r = .25, p < .01). This suggests that a positive brand attitude increases consumers’ willingness to share and one who is willing to share the act of kindness probably has a more positive brand attitude.
Table 1 further sows that credibility positively correlates with brand attitude (r = .37, p < .01) and sincerity (r = .35, p < .01). This makes sense in a way that when consumers
perceive a brand’s act of kindness as credible this leads to a more positive brand attitude and more sincere perceptions.
Finally, noteworthy are several correlations of the extent to which consumers like the act of kindness. The more one likes a brand’s kindness action, the more positive brand
attitudes, sincere and credible perceptions are evoked. These correlations are not surprising and seem to make sense.
Table 1: Means, standard deviations and correlations
4.3 Tests of hypotheses
4.3.1 Dependent variables
In this section are the results of both dependent variables presented, starting with willingness to share followed by brand attitude.
Willingness to share
This research proposed that acts of kindness in general (so both random and strategic ones) lead to positive eWOM (H1). To test this hypothesis, an analysis of covariance (ANCOVA) was performed with act of kindness as independent variable, willingness to share as
dependent variable and credibility, likeability and personality as covariates. Familiarity was dropped out as a covariate because preliminary tests showed the variable to be dependent on the independent variable. This is in contrast with the assumption of the independence of covariate, which means that familiarity cannot be used as covariate to reduce the error
Variables M SD 1 2 3 4 5 6 7 8 1. Brand attitude 5.46 .79 (.92) 2. Willingness to share 2.39 1.17 .25** (.85) 3. Sincerity 4.58 1.03 .55** .37** (.82) 4. Perceived motives 5.83 .83 .20* .09 -.02 (.72) 5. Credibility 4.75 1.06 .37** .14 .35** -.07 (.88) 6. Familiarity 4.35 1.67 .06 -.06 .06 .11 -.03 (.86) 7. Likeability 5.69 1.12 .54** .11 .45** .09 .35** .19* (.88) 8. Personality type 4.01 1.17 -.01 .04 -.11 .10 .12 .07 -.04 (.92)
**. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed).
Note: Elements in parentheses on the diagonal represent cronbach's alpha scores.
variance as overlapping covariates do not contribute to a reduction in error variance (Mayers, 2013). The violation of this assumption by familiarity applies to all hypotheses, hence
familiarity is dropped out as covariate for each ANCOVA. Table 2 presents the mean scores of both dependent variables and the included covariates.
Table 2: Dependent variable (DV) and covariate (CV) scores, by act of kindness group
No act of kindness (N = 59) Act of kindness (N = 100)
Mean SE Mean SE Willingness to share (DV) 2.19 .15 2.50 .12 Brand attitude (DV) 5.43 .10 5.47 .08 Credibility (CV) 4.79 1.08 4.73 1.05 Likeability (CV) 5.61 1.22 5.74 1.06 Personality (CV) 3.82 1.08 4.12 1.21
Note: This table includes the actual mean scores.
Next, an independent one-way ANOVA was used to show that there was no
significant difference in willingness to share scores between the control group (without an act of kindness) and the group in which an act of kindness (random or strategic) was performed, F (1, 157) = 2.635, p = .107. When the covariates where applied in ANCOVA, it indicated that willingness to share scores were poorer for the control group (M = 2.19 versus M = 2.50 for the act of kindness group), but it was not significantly poorer, F (1, 154) = 2.512, p = .115. Thus, H1 is not supported.Table 3 presents the ANCOVA results for willingness to share including the covariates. It shows none of the covariates to be significant.
Additional ANCOVA analyses suggested that variance in willingness to share scores may be shared with variance in familiarity. Statistically, this finding suggests that they may be measuring the same construct (Mayers, 2013). However, these two constructs are very
different, so an alternative explanation might be that the respondent starts the experiment with
reading the news article about an act of kindness and thus becomes more familiar with it, more easily recognizes it or more easily compares it to a similar past experience. This may influence responses in a way that may lead to higher familiarity scores when the participant at the end answers how familiar he/she is with acts of kindness.
Table 3: ANCOVA results for H1
Dependent Variable: Total score on willingness to share scale Source Type III
Sum of Squares
df Mean Square
F Sig. Partial Eta Squared Corrected Model 8.675 a 4 2.169 1.597 .178 .040 Intercept 5.881 1 5.881 4.332 .039 .027 CredTOT 2.414 1 2.414 1.778 .184 .011 LikeTOT .848 1 .848 .624 .431 .004 PersonTOT .029 1 .029 .022 .883 .000 ActofKindness 3.410 1 3.410 2.512 .115 .016 Error 209.075 154 1.358 Total 1124.333 159 Corrected Total 217.750 158
a. R Squared = .040 (Adjusted R Squared = .015)
Brand attitude
In addition to H1, this research also proposed that acts of kindness in general lead to positive a brand attitude (H1). Again, an analysis of covariance (ANCOVA) was performed with act of kindness as independent variable, brand attitude as dependent variable and credibility, likeability and personality as covariates (as mentioned before, familiarity was dropped out as covariate in every ANCOVA). First, using an independent one-way ANOVA, it was shown that there was no significant difference in brand attitude scores between the control group (without an act of kindness) and the group in which an act of kindness (random or strategic) was performed, F (1, 157) = .092, p = .762. When the covariates were applied in ANCOVA, it
indicated that brand attitude scores were almost the same for the control group (M = 5.43) and the act of kindness group (M = 5.47). The minimal difference was not significant, F (1, 154) = .009, p = .923. Thus, this finding does not provide support for H2. Table 4 presents the
ANCOVA results for brand attitude including the covariates.
Table 4: ANCOVA results for H2
Dependent Variable: Total score on brand attitude scale Source Type III
Sum of Squares
df Mean Square
F Sig. Partial Eta Squared Corrected Model 32.771 a 4 8.193 19.063 .000 .331 Intercept 28.827 1 28.827 67.078 .000 .303 CredTOT 3.515 1 3.515 8.180 .005 .050 LikeTOT 18.890 1 18.890 43.956 .000 .222 PersonTOT .020 1 .020 .045 .831 .000 ActofKindness .004 1 .004 .009 .923 .000 Error 66.182 154 .430 Total 4830.920 159 Corrected Total 98.953 158
a. R Squared = .331 (Adjusted R Squared = .314)
As illustrated in table 4, two covariates are significant: credibility (p = .005) and likeability (p = .000). However, this is not such a surprising finding as it is straightforward that the more one likes the act of kindness and perceives it as credible leads to a more positive brand attitude.
4.3.2 Moderation effects
In this section are the results of both moderator hypotheses presented, starting with personality as a moderator followed by type of act of kindness.
Personality
H3 posits that personality type moderates the effect of acts of kindness on consumers’ willingness to share and brand attitude. Testing this hypothesis includes an ANCOVA with act of kindness and personality type as independent variables, willingness to share and brand attitude as dependent variables and credibility, likeability and personality as covariates. Table 5 shows the mean scores for both dependent variables and the included covariates for entity and incremental theorists under the two different act of kindness conditions.
Table 5: Dependent variable (DV) and covariate (CV) scores, by act of kindness group and
personality type
No act of kindness (N = 15) Act of kindness (N = 37) Entity Incremental (N = 5) (N = 10) Entity Incremental (N = 20) (N = 17) Willingness to share (DV) 1.73 1.63 (.28) (.85) 2.93 2.80 (1.27) (1.19) Brand attitude (DV) 5.28 5.86 (.70) (.91) 5.82 5.65 (.65) (.65) Credibility (CV) 4.70 5.05 (.27) (1.36) 4.85 4.26 (.97) (1.26) Likeability (CV) 5.50 6.10 (.50) (1.10) 5.90 5.88 (.85) (1.11) Personality (CV) 5.68 2.23 (.57) (.35) 5.66 2.19 (.34) (.56) Notes: This table includes the actual mean scores. Standard deviations are in parentheses.
First, with willingness to share as dependent variable, the main effect of personality as moderator was tested (before including covariates) but no significant difference was found, F (1, 48) = .002, p = .967. When the covariates where applied in ANCOVA, it indicated that willingness to share differ for entity and incremental theorists both when an act of kindness was performed (M = 2.93 versus M = 2.80) as when no act of kindness was performed (M = 1.73 versus M = 1.63). However, the difference was not significant, F (1, 45) = .041, p = .841.
This finding is contradicting H3. Table 6 presents the ANCOVA results including personality as a moderator, willingness to share as dependent variable and the covariates.
Table 6: ANCOVA results for H3 – willingness to share
Dependent Variable: Total score on willingness to share scale Source Type III
Sum of Squares
df Mean Square
F Sig. Partial Eta Squared Corrected Model 17.796 a 6 2.966 2.290 .052 .234 Intercept 1.427 1 1.427 1.102 .299 .024 CredTOT 1.113 1 1.113 .860 .359 .019 LikeTOT .262 1 .262 .203 .655 .004 PersonTOT .290 1 .290 .224 .639 .005 ActofKindness 14.204 1 14.204 10.968 .002 .196 Personality .384 1 .384 .297 .589 .007 ActofKindness * Personality .053 1 .053 .041 .841 .001 Error 58.281 45 1.295 Total 407.778 52 Corrected Total 76.077 51
a. R Squared = .234 (Adjusted R Squared = .132)
In this model, act of kindness had a significant impact on willingness to share. Recall that no significance was found while testing the same relationship for H1 (for an explanation, see the discussion).
Next, the same ANCOVA was performed, only this time with brand attitude as dependent variable. The main effect of personality as moderator was tested (before including covariates) but no significant difference was found, F (1, 48) = 2.763, p = .103. When the covariates where applied in ANCOVA, it indicated that entity and incremental theorists also differ in terms of brand attitude. In the absence of an act of kindness, incremental theorists held a more positive brand attitude (M = 5.86 versus M = 5.28 for entity theorists), but when an act of kindness was performed entity theorists held a more positive brand attitude (M =
5.82 versus M = 5.65 for incremental theorists). Again, the difference was not significant, F (1, 45) = 1.209, p = .277. Hence, H3 is not supported. Table 7 presents the ANCOVA results including personality as a moderator, brand attitude as dependent variable and the covariates. Again, the covariate likeability is significant (p = .003). As noted before, this seems to make sense as the more one likes an act of kindness a more positive brand attitude is evoked.
Table 7: ANCOVA results for H3 – brand attitude
Dependent Variable: Total score on brand attitude scale Source Type III
Sum of Squares
df Mean Square
F Sig. Partial Eta Squared Corrected Model 8.648 a 6 1.441 3.840 .004 .339 Intercept 2.464 1 2.464 6.564 .014 .127 CredTOT .953 1 .953 2.539 .118 .053 LikeTOT 3.666 1 3.666 9.767 .003 .178 PersonTOT .481 1 .481 1.281 .264 .028 ActofKindness .321 1 .321 .856 .360 .019 Personality .606 1 .606 1.613 .211 .035 ActofKindness * Personality .454 1 .454 1.209 .277 .026 Error 16.892 45 .375 Total 1726.440 52 Corrected Total 25.541 51
a. R Squared = .339 (Adjusted R Squared = .250)
Types of act of kindness
H4 stated that the type of act of kindness (random or strategic) moderates the moderated relationship of H3, where personality is the moderator between acts of kindness and reactions towards it in terms of willingness to share and brand attitude. An ANCOVA was conducted with personality type and type of act of kindness as independent variables, willingness to share and brand attitude as dependent variables and sincerity, perceived motives, credibility,
likeability and personality as covariates. Table 8 shows the mean scores for both dependent variables and the included covariates for entity and incremental theorists under the two different act of kindness conditions.
Table 8: Dependent variable (DV) and covariate (CV) scores, by type of act of kindness and
personality type
Entity theorists (N = 20) Incremental theorists (N =17)
RAK SAK (N = 9) (N = 11) RAK SAK (N = 11) (N = 6) Willingness to share (DV) 2.48 3.30 (1.20) (1.26) 2.82 2.78 (1.22) (1.24) Brand attitude (DV) 5.60 6.00 (.70) (.57) 5.65 5.63 (.76) (.43) Sincerity (CV) 5.00 4.94 (1.14) (.84) 4.88 5.22 (1.02) (.54) Perceived motives (CV) 5.89 5.97 (1.07) (.72) 5.91 5.22 (.87) (1.22) Credibility (CV) 4.78 4.91 (1.42) (.44) 3.95 4.83 (1.46) (.52) Likeability (CV) 6.06 5.77 (.73) (.96) 5.91 5.83 (1.20) (1.03) Personality (CV) 5.71 5.63 (.40) (.30) 2.11 2.33 (.66) (.31) Notes: RAK = random act of kindness, SAK = strategic act of kindness. This table includes the actual mean scores. Standard deviations are in parentheses.
First, with willingness to share as dependent variable, the main effect of act of kindness type as moderator was tested (before including covariates) but no significant difference was found, F (1, 33) = 1.064, p = .310. When the covariates where applied in ANCOVA, it indicated that willingness to share differs per type of act of kindness for both entity and incremental theorists. Entity theorists were more willing to share when the act of kindness was strategic (M = 3.30) than when it was truly random (M = 2.48). In contrast, incremental theorists were more willing to share when the act of kindness was truly random
(M = 2.82) compared to the strategic act (M = 2.78). However, the differences were not significant, F (1, 28) = .820, p = .373. These results are contradictory to H4. Table 9 presents the ANCOVA results including types of act of kindness as a moderator, willingness to share as dependent variable and the covariates.
Table 9: ANCOVA results for H4 – willingness to share
Dependent Variable: Total score on willingness to share scale Source Type III
Sum of Squares
df Mean Square
F Sig. Partial Eta Squared Corrected Model 15.579 a 8 1.947 1.433 .227 .290 Intercept 1.190 1 1.190 .876 .357 .030 SincTOT 1.061 1 1.061 .781 .384 .027 MotiveTOT 5.980 1 5.980 4.400 .045 .136 CredTOT 1.723 1 1.723 1.268 .270 .043 LikeTOT .923 1 .923 .679 .417 .024 PersonTOT .262 1 .262 .193 .664 .007 Personality .139 1 .139 .103 .751 .004 AoKType 1.485 1 1.485 1.093 .305 .038 Personality * AoKType 1.115 1 1.115 .820 .373 .028 Error 38.055 28 1.359 Total 359.222 37 Corrected Total 53.634 36
a. R Squared = .290 (Adjusted R Squared = .088)
As illustrated in table 9, the covariate perceived motives has a significant impact (p = .045) on willingness to share in this model. This finding is somewhat surprising as it suggests that more perceived motives behind a brand’s act of kindness increase consumers’ willingness to share.
Next, willingness to share was replaced by brand attitude as dependent variable. The main effect of act of kindness type as moderator was tested (before including covariates) but
