Corporate social irresponsibility : the influence of CSR reputation on corporate social irresponsibility

(1)

Corporate social irresponsibility:

The influence of CSR reputation on corporate social irresponsibility

---

Masterthesis

Author: Wieteke Kluiwstra (10445080)

University of Amsterdam, Faculty of Economics and Business Under supervision of: Dr. M. Vock

Secondary supervision: Drs. L.T. Moratis June 29, 2015

(2)

Abstract

While more companies engage in and communicate about their corporate social responsibility (CSR), incidents of corporate social irresponsibility (CSIR) seem to increase when firms globalize. Although prior research has made some early attempts to explore CSIR behaviour, it remains unclear how consumers react to positive and negative CSR reputations in case of subsequent CSIR behaviour. This study aimed to examine the influence of CSR reputation and CSIR domain on consumer emotions and attitude toward the company. Survey data were collected from a sample of 387 Dutch consumers. Results indicated that CSR reputation has a significant effect on emotions and attitude toward the company. A negative CSR reputation leads to more negative emotions and lower attitudes toward the company than a positive CSR

reputation or no CSR reputation. There is evidence that CSIR domain has a significant effect on emotions, but no interaction effect with CSR reputation is found. Implications of these findings are discussed.

Keywords: Corporate social responsibility, corporate social irresponsibility, CSR reputation,

(3)

3

1. INTRODUCTION

The expansion of the role of corporate social responsibility (CSR) the last decades has led to a shift in doing business. As governments and business recently become more aware of the difficult environmental, social and governance challenges that have come along with the rise of globalization and the power shift from government to corporations, the role of CSR in business has strengthened and contact between business and society have become more extensive. A study by Becker-Olsen, Cudmore and Hill (2006) shows that over 80% of the consumers believe that companies should engage in social initiatives. It seems like it is not the question whether to embrace CSR in business, but how (Smith, 2003). Not only does pressure from consumers, employees and other stakeholders force business to address social issues, more firms realise that embracing CSR gives them an opportunity to gain a competitive advantage while addressing global problems (Bielak, Bonini, & Oppenheimer, 2007). Companies seem to be encouraged by scholars establishing a link between a company’s CSR and positive consumer responses, such as positive brand evaluation (Klein & Dawar, 2004), product evaluation (Brown & Dacin, 1997), attitude toward the company (Brown & Dacin, 1997; Sen & Bhattacharya, 2001) and purchase intentions (Creyer & Ross, 1997; Mohr & Webb, 2005). A responsible image becomes a strategic tool which can be used by companies to differentiate themselves from competitors (McWilliams & Siegel, 2001). Tony Chocolonely, The Body Shop and Ben & Jerry’s are just a few examples of brands that use their CSR focus to position themselves. Additionally, there is an increase of public information about CSR activities of firms and in response to the growing attention about negative firm behaviours more companies are making CSR an important strategic objective (Wagner, Lutz & Weitz, 2009).

While companies are frequently positioning themselves as corporate responsible, they often fail to avoid irresponsible actions. Despite the promising CSR statements of companies, consumers are still often exposed to the firm’s contrary – or inconsistent – actual behaviour (Wagner, Lutz & Weitz, 2009). For example, American retail corporation Walmart has proclaimed in their

(5)

5 sustainability campaign (2005) to cut the energy use and reduce waste. Walmart improved its image, but recently it came out that over the past years greenhouse gas emissions have continued to grow and it only increased its environmental footprint (Mitchell, 2012). Think about how that news would be perceived by their customers. For a good consumer-brand relationship it is essential to keep your promises (Aaker, Fournier, & Brasel, 2004). In this way a firm’s effort to get a good CSR reputation might actually backfire when customers start to

question the firm’s motives (Schendler, 2007). As a well-known quote of Warren Buffett goes: “It takes 20 years to build a reputation and five minutes to ruin it. If you think that, you’ll do things differently” (Buffett, n.d.).

The extent to which a CSR initiative will lead to favourable attitudes toward the company depends on the company’s existing CSR reputation (Bhattacharya & Sen, 2004). When people have to evaluate morally questionable behaviour, they do not only consider the action itself but also the moral track record of the actor (Effron & Monin, 2010). A positive CSR reputation might excuse the company from negative consequences when confronted with corporate social

irresponsibility (CSIR) (e.g. Eisingerich, Rubera, Seifert, & Bhardwaj, 2011; Vanhamme, Swaen, Berens, & Janssen, 2014). However, a positive CSR reputation can also backlash if it activates perceptions of corporate hypocrisy on consumers, which on its turn has a negative impact on attitudes toward the firm (Wagner, Lutz, & Weitz, 2009). A negative CSR reputation on the other side probably won’t excuse the company from negative consequences. However, consumer expectations for a company with a negative CSR reputation might be different than consumer expectations for a company with a positive CSR reputation. So, as these conflicting results illustrate, it is still unclear whether a positive or a negative CSR reputation might help or hurt the company in case of subsequent irresponsible corporate behaviour. This thesis will address the question if and when a positive CSR reputation can help a company in case of subsequent irresponsible corporate behaviour.

(6)

6 Further, the effect of a positive or negative CSR reputation might also depend on the domain of the corporate social irresponsible action. Most of the time companies participate in multiple CSR activities which are in different domains, but they tend to focus on one domain in particular. Tony Chocolonely for example is known for its slave free chocolate (employee domain), but they are also active in a less communicated project to reduce their ecological footprint

(environmental domain). Consumers might accept that companies cannot accelerate in every domain. Therefore, CSIR in the same domain as the positive reputation (i.e. related

transgression) (e.g. Tony Chocolonely involved in a child labour scandal) might be a bigger problem than CSIR in a different domain (i.e. unrelated transgression) (e.g. Tony chocolonely not paying enough tax). This effect might be different for companies with negative CSR reputations. Corporate social irresponsibility in a domain related to the negative CSR reputation (i.e. related transgression) might be expected. Whereas CSIR in a domain unrelated to the CSR reputation might come on top of the negative CSR reputation and makes it worse.

The objective of the present thesis is to research the emotions and attitudes of consumers that face corporate social irresponsible actions. In an experimental study I will try to get an answer if a positive reputation can save the company from negative emotions and eventually from

negative attitudes toward the company. To get a good picture of the impact of reputation I will compare positive and negative reputations. Secondly, this research will try to get an answer if it is worse for a company to be confronted with CSIR in a domain related to the CSR reputation and compare this with CSIR in domains unrelated to the CSR reputation. This leads to the following research question:

Research question:

To what extent does the CSR reputation (positive versus negative) and the CSIR domain (i.e. related versus unrelated to the CSR reputation domain) have an influence on consumers’ emotions and their attitude toward the company when confronted with CSIR?

(7)

7

Theoretical and managerial contributions

Although research about the effect of corporate social responsibility (CSR) on consumers has appeared frequently over the years (e.g. Klein & Dawar, 2004; Brown & Dacin, 1997; Mohr & Webb, 2005), the influence of corporate social irresponsibility (CSIR) is not sufficiently

addressed in an explicit manner in the discussion about CSR. Early attempts to explore CSIR and reputation can be found in studies by Eisingerich et al. (2011), Grappi et al. (2013), Vanhamme et al. (2014), Wagner et al. (2009) and Yoon et al. (2006). Grappi et al. (2013) investigate the effect of CSIR on emotions and how they activate negative responses toward the corporation, but they do not consider the CSR reputation and transgression domain. Wagner et al. (2009) look into companies’ CSR statements and the effect of CSIR on attitude toward the company, but they only consider the related transgressions which results in perceived hypocrisy. Eisingerich et al. (2011) and Vanhamme et al. (2014) both study the effect of a good CSR reputation when confronted with CSIR, but they do not investigate bad CSR reputations and related vs. unrelated transgression domains. Yoon et al. (2006) examine good and bad CSR reputations, but they only study the effect of consecutive CSR statements and do not look into the effect of transgressions (CSIR). Finally, the psychological study of Effron and Monin (2010) examines the effect of prior good deeds and the relatedness of the transgression domain, but they do not take prior negative deeds into account. Also their research is limited to human actors and did not consider the effects of business reputations and consumer responses to the company. This research will contribute by addressing both positive and negative reputations in case of a corporate social irresponsible action, and takes difference between related and unrelated transgression domains into account. Furthermore, this research will address the recent call by Grappi et al. (2013) to study the effect of felt emotions on attitudes.

From a managerial perspective, this research will try to answer the question if it is

recommended to invest in a strong positive CSR reputation or not. Companies are investing a huge amount of money in responsible investments. For example, Target’s $107.8 million

(8)

8 donation in CSR represents 3.6% of its pre-tax profits, General Motors’ $51,2 million donation represents 2.7% of its pre-tax profits and Merck’s $921 million donation represents 11.3% of its pre-tax profits. These are just a few examples of the large amount of money companies invest in CSR to enhance their CSR reputation (Luo & Bhattacharya, 2006). In 2011 according to The Forum for Sustainable and Responsible Investment the US spend $3.74 trillion on responsible investments (US SIF, 2012). It is very important to know if consumers react differently on corporate social irresponsible actions depending on the CSR reputation, because in that case managers can anticipate their responses to potential transgressions. In addition, to avoid perceptions of hypocrisy it is valuable for managers to know if they should invest more in the domain of their current CSR reputation or should spread their investments over different domains. It is particularly important for managers to know how emotions will affect attitudes toward the firm, because it has been shown that consumer attitudes affect consumers’ actual purchase patterns (Lichtenstein, Drumwright, & Braig, 2004; Pan & Zinkhan, 2006).

The remainder of this thesis is organized as follows. In the following section, I review relevant theory about corporate social irresponsibility, CSR reputations, the difference between related and unrelated transgression domains and I propose a set of formal hypotheses. Next, I will present the conceptual framework and describe the method and the results for my studies. I will end with a discussion of my findings, examination of the managerial implications and limitations of this research.

(9)

9

2. LITERATURE REVIEW + HYPOTHESES DEVELOPMENT

2.1 Corporate social irresponsibility

Definition: In this research I define corporate social irresponsibility (CSIR) as unethical corporate actions that negatively influence one or more stakeholders of a firm. This definition is consistent

with prior literature. Lin-Hi and Müller (2013) who have explored all the different articles which talk about corporate social irresponsibility until then conclude that: “Although the current discussion of CSIR is characterized by different positions and approaches, their common denominator is that CSIR is seen as immoral and/or illegal corporate actions with negative consequences for others” (Lin-Hi & Müller, 2013, p. 1932).‘Others’ refers to all the different stakeholders of a firm, so corporate actions at the expense of employees, consumers, suppliers, the environment, other organizational stakeholders, and even society at large (Pearce & Manz, 2011). Corporate social irresponsible actions can also be seen as moral transgressions. Moral transgressions are a violation of the implicit or explicit norms that guide relationship

performance and evaluation. However, transgressions are more focused on individual actions. This research will focus on corporate actions and thus will focus on corporate social

irresponsibility instead of moral transgressions. According to Lange & Washburn (2012, p. 300) corporate social irresponsibility (CSIR) is “the opposite side of the responsibility coin – that is, the failure to act responsibly”. Examples of CSIR include environmental pollution, using child labour, price-fixing scandals or corruption scandals. Even if a corporation does not break the law, corporate actions can still negatively affect others (Lin-Hi & Müller, 2013). Especially, when corporations employ activities at locations where there are fewer rules or laws.

While more companies engage in and communicate about their corporate social responsibility, incidents of irresponsible behaviour seem to increase when firms globalize. Especially firms operating in different countries are faced with an enormous challenge to act responsible across different cultures and business practices. Complex value creation processes make it very difficult

(10)

10 for a company to completely exclude the possibility that they become involved in an

irresponsible action. Some well-known examples are sweatshop and child labour practices at GAP and Nike plants in Asia, the aggressive marketing of baby-milk in Africa by Nestlé and the Niger Delta controversy by Shell Oil. While there are situations known where consumers trade their ethical norms for personal gain and do not care about irresponsible behaviour (Brunk & Blümelhuber, 2011) and in some cases firms are even rewarded for CSIR activity (Groening & Kanuri, 2013), managers need to be aware of irresponsible behaviour, because consumers are more sensitive to negative information (e.g., Sen & Bhattacharya 2001; Brunk & Blümelhuber 2011). In most cases customers do care about irresponsible behaviour and consumer reactions to irresponsible behaviour can be extremely harmful for the company. Consumers expect firms to work according the legal standards and confer basic moral norms. It is the absolute minimum they expect and failing to do so can have a detrimental effect. Not only can irresponsible

behaviour result in boycotting behaviour (Lindenmeier, Schleer, & Pricl, 2012), it can also evoke negative word of mouth and protest behaviour which goes further than only boycotting (Grappi, Romani, & Bagozzi, 2013). Effects of consumer boycotts and protest behaviour can be so

dramatic that they disturb the business bottom line (Lindenmeier, Schleer, & Pricl, 2012). In the long run it can harm the firm’s reputations, brand images and, as in the baby-milk case of Nestlé, cause a spill-over effect to other unrelated brands and products in their portfolio (i.e. Nescafé) (Brunk, 2010).

This huge impact of negative information can be explained by psychological studies which demonstrate that humans in general show more thorough cognitive, affective and emotional reactions to negative information than to positive information (e.g., Baumeister, Bratslavsky, Finkenauer, & Vohs, 2001; Taylor, 1991). The so called negative asymmetry effect (e.g. Taylor, 1991) or negativity effect (Skowronski & Carlston, 1989) state that negative behaviour carries more weight than positive behaviour. There are various theoretical explanations for the

(11)

11 explanations by Fiske (1980), Peeter’s and Czapinski’s (1990) behavioural adaptation model, Peeters’s (1971) mushroom model, and the category diagnosticity approach by Skowronski and Carlston (1989). This research will draw further on the category diagnosticity approach

(Skowronski & Carlston, 1989). This model explains that immoral (negative) behaviours are generally more diagnostic of negative traits than moral (positive) behaviours are of positive traits (Skowronski & Carlston, 1989). For example, a person robbing a bank will be perceived more easily as dishonest, than a person returning too much change can be categorized as honest. According to Skowronski and Carlston (1989) the diagnosticity of negative and positive

behaviour differs because of the usefulness people belief that the different kind of behaviours as information has: “People generally expect more inconsistency from those perceived as

possessing negative traits than from those perceived as possessing positive traits” (Skowronski & Carlston, 1989, p. 137). Even a person committing a robbery will not constantly perform immoral behaviour but sometimes act in an honest way. On the other hand, honest people will almost exclusively act in an ethical way. In this way, a person committing unethical behaviour can be more easily categorized as bad person than a person who behaves ethically. Furthermore the prospect theory, a model that describes individual choice, supports the assumption that people care more about negative than about positive outcomes (Kahneman & Tversky, 1979). The prospect theory suggests that people are risk-averse. According to Kahneman and Tversky (1979) losses have more impact on the value function than gains. To be positively evaluated, good behaviour must be performed continuously while one irresponsible action can be established to a negative perception. Likewise, Mohr and Web (2005) demonstrate that this negative asymmetry effect also applies to evaluations of organizations. Their research shows that (negative) CSIR information has a bigger effect on the evaluation of a firm and purchasing intention of customers than (positive) CSR information. In other words, the evaluation of a firm’s behaviour is not just a sum of all the positive and negative CSR information, but one action can determine the overall direction (Brunk & Blümelhuber, 2011).

(12)

12 According to Brunk (2010) there are 6 domains and 36 subdomains that are capable of inducing un/ethical perceptions in consumers mind. She identified the following domains: consumers, employees, environment, overseas community, local community & economy and business community (see figure 1). Because there are so many different CSR domains, companies participate in multiple CSR domains, but they tend to focus on one (or a few) domain(s) in particular. The Body Shop for example has an active approach towards protecting the

environment, supporting the community and defending human rights. However, their core focus is ‘no animal testing’. You could actually say that they have a good reputation for CSR in the environmental domain, because according to Brunk (2010) the subdomain animal protection is part of the domain environment.

Figure 1: Consumer perceived ethicality – domains of origin (Brunk, 2010, p. 258)

However, Brunk (2010) is not the only one who explored the different domains in CSR. Wagner, Bicen and Hall (2008) identified specifically which business practices lead to perceptions of

(13)

13 corporate social irresponsibility in retailing. They identified 14 different factors which represent perceptions of CSIR in retailing: natural environment, local business, foreign economies, local employment, societal rules, employee benefits, employee wages, local working conditions, employee discrimination, foreign labour, sales practices, dishonesty, offensive material, pricing policies. These domains regarded as socially irresponsible behaviour correspond to a great extent with the domains described by Brunk (2010).

2.1.1 The effect of CSIR on Attitudes toward the company

Academic literature researching consumers’ attitudes toward the company suggest that they are influenced by two factors: ethical behaviour and product attributes (e.g. Brown & Dacin, 1997; Folkes & Kamins, 1999). Consumer evaluations of companies’ good or bad behaviour is not only based on the perfomance of the products, but also on the firms’ (ir)responsible behaviour. Irresponsible behaviour can have a major impact on attitudes toward the company. An indication of the way consumers evaluate firms’ (ir)responsible behaviour comes from

psychological research about the diagnosticity of negative and positive information described in the preceding paragraph. Similar to evaluations of organizations, ethical and unethical

information influences attitudes toward the company asymmetrical (Folkes & Kamins, 1999). This means that information about the firms’ irresponsible behaviour should provide more proof of the firm’s characteristics and influences attitudes toward the company stronger, compared to information about the firms’ responsible behaviour. For example, information about a company caught in a sweatshop scandal will provide more evidence that the company acts irresponsible than information about a company avoiding sweatshop practices will give evidence that the company acts responsible. As explained in the previous section, even bad organizations do not constantly behave irresponsible. For this reason people consider irresponsible behaviour as more diagnostic than responsible behaviour. Although there are studies that demonstrate how CSR positively affects consumer’s attitudes toward the company and its products (Brown & Dacin, 1997; Creyer & Ross, 1997; Ellen, Mohr, & Webb, 2000) CSIR

(14)

14 practices will have a much bigger impact on attitudes (Folkes & Kamins, 1999). Corporate social irresponsibility strongly reduces customers’ attitudes toward the company regardless of

product attribute level (Folkes & Kamins, 1999). Even superior product quality is not able to reduce the negative effect of unethical behaviour in influencing attitudes toward the company. Consistent with previous research, this research will assume that CSIR will strongly reduce customers’ attitudes toward the company.

2.1.2 The effect of CSIR on emotions

Until today still little is known about how irresponsible behaviour of a company is processed in the consumer’s mind. Recent studies of Lindenmeier, Schleer and Prici (2012) and Grappi, Romani and Bagozzi (2013) are one of the first that examine the underlying mechanisms driving consumer responses toward CSIR. Both studies investigate the effect CSIR has on emotions. They elaborate further on basic research in moral psychology which found that ethical and social transgressions stimulate negative emotions at people. Psychologists discovered that emotions are the connection between perceived violations of moral standards and moral behaviours. In other words, the felt emotions are an incentive to act in negative ways to unethical behaviour. The studies of Lindenmeier, Schleer and Prici (2012) and Grappi, Romani and Bagozzi (2013) have confirmed this mediating effect of emotions on behaviour in response to corporate social irresponsibility. According to Lindenmeier, Schleer and Prici (2012) emotional dimensions of consumer outrage have an impact on boycotting intentions. Grappi, Romani and Bagozzi (2013) found evidence that contempt, anger and disgust (other condemning emotions) were felt in response to ethical or social harm done by corporations. These emotions activate, in

combination with other-regarding virtues, extremely negative responses which will harm the corporation (Grappi, Romani, & Bagozzi, 2013). Prior literature has suggested the influence of emotions on attitudes (e.g. Yoo, Park, & MacInnis, 1998; van den Hooff, Schouten, & Simonovski, 2012). Because negative emotions are a direct reaction to CSIR behaviour, this research expects that emotion is a main psychological mechanism that effects attitudes toward the company directly.

(15)

15

2.2 CSR reputation

Definition: A corporate reputation is frequently defined as a stakeholder’s overall evaluation of company’s past and future actions compared to other leading rivals (Fombrun, 1996; Gotsi &

Wilson, 2001). Following this definition, with regard to the more specific concept of CSR

reputation, the evaluation of a company focusses specifically on all the CSR related activities and consumer perceptions of those activities. While most of the literature discusses reputation only as a positive aspect, it is important to recognize the positive and negative perceptions. So, in the context of this study companies with a positive CSR reputation are positioned as highly engaging and supporting CSR activities. They are known by customers for their values and efforts. On the other side, companies with a negative CSR reputation have a bad image in terms of their CSR policies. They are known for their social irresponsible behaviour and consumers have lower expectation in terms of their CSR.

2.2.1 Positive CSR reputation as insurance

While people tend to care more about negative and thus irresponsible behaviour, prior research suggest that positive CSR activities can offer companies protection against crisis (e.g., Klein & Dawar, 2004; Eisingerich, Rubera, Seifert, & Bhardwaj, 2011). In this way, CSR activities might function as an insurance policy against potential CSIR behaviour. The extent to which a CSR initiative will lead to favourable attitudes toward the company depends on the company’s existing CSR reputation (Bhattacharya & Sen, 2004). Kasper Ulf Nielsen, executive partner at Reputation Institute, explains the importance of CSR: “It is a core element of reputation and can be used to help establish trust and goodwill amongst stakeholders. Almost half of people’s willingness to trust, admire, and feel good about a company is based on their perceptions of the company’s corporate social responsibility efforts, so this is a key tool for companies to use to improve support from stakeholders including consumers, regulators, the financial community, and employees” (Nielsen, 2013, p. 6).

(16)

16 Businesses like Ben & Jerry’s, The Body Shop and Tony Chocolonely excel in their positioning on the CSR platform and they are known for their efforts to go beyond charitable giving. They integrate the CSR thought wholly through their business and position themselves in terms of their CSR. This positioning might actually be beneficial in case of accusations of irresponsible behaviour. When consumers have to judge about the morally dubious behaviour of a company, they do not only consider the action itself but also a firm’s moral track record. Building a strong positive CSR reputation might be a very effective strategy to buffer against later accusations of irresponsible behaviour (Vanhamme, Swaen, Berens, & Janssen, 2014). This research will define buffering as a decrease in negative emotions and an increase of attitudes toward the company when the company had a positive CSR reputation compared to when it had a negative CSR reputation or no CSR reputation. According to consumer-company identification literature consumers are willing to forgive a company’s irresponsible behaviour if it happens occasionally and possibly unintended (Bhattacharya & Sen, 2004). Identification is an important factor that addresses the need of individuals for self-definition and social identity. When consumers believe that the identity of a company is enduring, distinctive and contributes to their self-esteem, they are more likely to identify with that company. Identification with firms that have a positive CSR reputation can enhance consumers’ self-esteem and can establish some kind of connection to the organization. When consumers perceive negative information about the company they identified with, this can affect their self-esteem. For this reason consumers having a strong identification with the firm are relatively resistant to moderately negative publicity (Einwiller, Fedorikhin, Johnson, & Kamins, 2006). A positive CSR reputation will generate a strong identification with the firm and higher levels of goodwill at consumers. This goodwill can be used as a buffer or insurance when the company is accused of corporate irresponsible behaviour. Also Eisingerich, et al. (2011) support these findings that CSR initiatives can protect companies from negative information about CSR practices. However a positive CSR reputation does not protect the firm from consumers responses to negative information about the firm’s core service offering

(17)

17 (Eisingerich, Rubera, Seifert, & Bhardwaj, 2011) or extreme negative information (Einwiller, Fedorikhin, Johnson, & Kamins, 2006).

Similarly, the psychological research of Effron and Monin (2010) studied the effect of an actor’s prior good deeds on observers’ willingness to excuse – or license – its morally dubious

behaviour. Their study is in the context of individuals and not on business, but they exhibit the same results. They found evidence that someone’s prior good deeds can change the way the observer will interpret the subsequent morally dubious behaviour. According to the moral credits model (Nisan, 1991) good deeds could be used as moral credits to balance out the subsequent morally ambiguous behaviour (moral debits). The positive and negative behaviour of an actor creates an overall impression. In this way someone’s prior good deeds can make it more justifiable for individuals to engage in morally questionable behaviour and they are more willing to excuse – or license – its morally dubious behaviour.

However, as explained in the first paragraph irresponsible behaviour has a bigger impact on consumers than responsible behaviour and they are not equally balanced to an overall

impression. To be positively evaluated, good behaviour must be performed continuously while one irresponsible action can be established to a negative perception. So, the positive CSR reputation might not able to balance out the irresponsible behaviour and can rather seem hypocritical in the eyes of the consumer and hurt the company.

2.2.2 The opposite side of positive CSR reputations

Although buffering and licencing suggest a positive effect of a good CSR reputation,

inconsistency theories and expectancy violation theories suggest that positive CSR reputations might increase, rather than decrease, negative emotions and attitudes. This means instead of decreasing the damage of potential irresponsible behaviour, it suggest that a positive reputation will backlash and hurt the company. According to the negative asymmetry effect (Taylor, 1991)

(18)

18 irresponsible behaviour has a bigger influence on consumers than responsible behaviour. Brunk and Blümelhuber (2011) demonstrate that evaluations tend to emerge as holistic impressions. Negative information is perceived in relation to other pieces of information (e.g. CSR reputation) and cannot be produced from evaluation of those information components in isolation. When consumers are confronted with inconsistency their evaluation will be based on the perceived diagnosticity of the available information. In the case of inconsistent information it is very possible that consumers base their evaluation on the later negative information, because negative behaviours are generally more diagnostic of negative traits than positive behaviours are of positive traits. Likewise, the study of Wagner et al. (2009) shows that positive CSR information, if accompanied by corporate social irresponsible behaviour, can be destructive to its desired effect. Inconsistent CSR information about firms negatively affect consumers’ evaluations. Due to inconsistency in CSR information, perceptions of corporate hypocrisy are activated at a consumer, which on its turn has a negative impact on CSR beliefs and attitudes toward the firm. Hypocrisy can be defined as “the belief that a firm claims to be something that it is not” (Wagner, Lutz, & Weitz, 2009, p. 79). It prompts negative assumptions about the prior positive CSR information, which consequently seems insincere, more selfish and calculated (Effron & Monin, 2010). Positive statements and positioning as highly engaging in and supporting CSR activities followed by negative behaviour (proactive strategy) increases hypocrisy and thus negative attitudes. Thus instead of buffering, positive CSR reputations will likely have an opposing negative effect on emotions and attitudes.

Additionally, the expectancy-violation theory shows that unanticipated violations of positive expectations will lead to extremitized negative evaluations. Expectations are a very important tool used by consumers as basis for comparing performance outcomes. Expectations can be hopes, wishes and anticipations. In the context of this study an expectation is an “anticipation of future consequences based on prior experience, current circumstances, or other sources of information” (Tryon, 1994, p. 313). Psychological literature demonstrates that meeting

(19)

19 expectations will lead to positive affects (Mandler, 1975), whereas violations of expectations will generate arousal and negative emotions (Bobes, Valdes-Sosa, & Olivares, 1994; Clary & Tesser, 1983). Customers do not expect a failure, so the initial reference point is no failure (Berry & Parasuraman, 1991). For brands with positive CSR reputations consumers have high

expectations of their CSR policies and expectancy violation suggests that CSIR damage the whole meaning on which this partnership was based. This does not apply to brands with negative CSR reputations because for these brands consumers have lower expectations and CSIR is less damaging.

Thus, this research expects that a positive CSR reputation can not balance out the subsequent irresponsible behaviour because of the huge impact negative behaviour has on consumers. Furthermore, hypocrisy and violations of expectations will have a tremendous effect on companies with a positive CSR reputation. Corporate social irresponsible behaviour from companies with a negative CSR reputation will, on the other hand, not appear hypocritical and inconsistent with their reputation. Therefore negative emotions will be higher and attitudes toward the company will be lower for companies with a positive CSR reputation relative to companies with no CSR reputation. On the other side, companies with a negative CSR reputation will not significantly differ from no CSR reputation, because the negative CSR reputation does not suffer from hypocrisy and expectancy violation. This research thus propose the following:

H1: The CSR reputation of the company involved in CSIR affects consumers’ negative emotions and attitudes toward the company. Specifically, a) negative emotions are higher and b) attitudes toward the company are lower, when companies with positive CSR reputations face CSIR than when companies with no CSR reputations face CSIR. Companies with negative CSR reputations will not significantly differ from companies with no CSR reputation.

(20)

20 While this research suggests that a positive CSR reputation will violate expectations and hurt the company when facing CSIR, it is not totally excluding the buffering impact of positive CSR

reputations. Instead it will focus on the interaction between CSR reputation and different CSIR conditions in which buffering or hypocrisy effects is most likely. The next paragraph will examine further on this interaction effect.

(21)

21

2.3 Domain of CSR reputation and CSIR – related versus unrelated

Consistent with prior psychological research of Effron and Monin (2010) this research suggests that the effect of the CSR reputation (increase or decrease of positive emotions and attitudes) discussed in the previous paragraph is depending on the relatedness of the CSIR domain vs. the domain of the CSR reputation. Therefore this research suggests that the CSIR domain interacts with the CSR reputation on negative emotions and attitudes and thus acts as a moderator. Companies with a positive CSR reputation are positioned as highly engaging and supporting CSR activities. They are known by customers for their values and efforts. Corporate social

irresponsibility in a domain related to the positive CSR reputation will as a result seem inconsistent with the companies’ values and positioning and extremely unexpected. Doing the opposite from what is claimed will be perceived inconsistent and hypocrite (Wagner, Lutz, & Weitz, 2009). As discussed earlier hypocrisy can be defined as “the belief that a firm claims to be something that it is not” (Wagner, Lutz, & Weitz, 2009, p. 79). People have a general aversion to inconsistent and unexpected behaviour. As discussed in the previous paragraph inconsistency elicits perceptions of hypocrisy and negative attributions about the prior CSR reputation, which consequently seems insincere, more selfish and calculated. Also according to the expectancy-violation theory (Bobes, Valdes-Sosa, & Olivares, 1994) the expectancy-violations of positive expectations will lead to extremitized negative evaluations. Therefore it seems logical to assume that CSIR in a domain related to the positive CSR reputation will appear hypocritical and unexpected and thus make consumers unwilling to buffer or licence. On the other hand CSIR in a domain unrelated to the positive CSR reputation would most likely not seem that hypocritical, since they are not directly inconsistent with the domain of the positive CSR reputation. Furthermore they will be less contrary to the positive expectations. In this case I expect buffering and licencing will protect the company from negative emotions and attitudes.For example, Tony Chocolonely is known for its slave free chocolate and so they have a positive CSR reputation in the employee domain. If there is subsequent information that they actually treat their employees unfair

(22)

22 (related to the positive CSR reputation), it will probably be considered more hypocritical and evoke more negative emotions and attitudes then when they would harm the environment (unrelated to the positive CSR reputation). People might accept that companies cannot excel in all the different CSR domains. However, when people know the company for their excellent employee treatment and the employees are actually not treated very honest they would probably be more suspicious and produce more negative emotions and attitudes. So, I expect that corporate irresponsibility in a domain related to the positive CSR reputation has a negative impact on emotions and attitudes toward the company. On the other side, I expect that corporate social irresponsibility in a domain unrelated to the positive CSR reputation will be somewhat protected by the positive CSR reputation and buffer the company from negative emotions and attitudes toward the company.

H2: The CSIR domain interacts with the CSR reputation on negative emotions and attitudes toward the company. Specifically, a) negative emotions are higher and b) attitudes toward the company are lower, when companies with positive CSR reputations face CSIR in a related domain (to the CSR reputation) than when companies with positive CSR reputations face CSIR in an unrelated domain (to the CSR reputation).

On the other hand, companies with a negative CSR reputation have a bad image in terms of their CSR policies. They are known for their social irresponsible behaviour and consumers have lower expectation in terms of their CSR. It is likely that people do not want to identify themselves with these companies and buffering or licensing will not protect these companies. However,

controversially to companies with positive reputations, corporate social irresponsibility in a domain related to the negative CSR reputation will not be perceived inconsistent with the

company’s reputation and will be expected by consumers. These companies will not act different from what is expected and for this reason not perceived inconsistent and hypocrite. As stated earlier, meeting expectations will lead to positive affects (Mandler, 1975). As a result, it seems

(23)

23 likely to propose that CSIR in a domain related to the negative CSR reputation will evoke less negative emotions and attitudes than CSIR in an unrelated domain (to the negative CSR reputation). CSIR in an unrelated domain to the negative CSR reputation will be less expected than CSIR in a related domain because the related CSIR (to the negative reputation) is already known by the consumers. Unrelated CSIR (to the negative reputation) will be something extra that was not known before by the consumers and likely act as an add on to the negative reputation. For example, Nike is known for its sweatshop practices in Asia (negative CSR reputation in the employee domain). Thus if there is subsequent information that they treat their employees unfair (related to the negative CSR reputation) it will be more expected than when they would harm the environment (unrelated to the negative CSR reputation). In this way CSIR related to the negative CSR reputation will likely evoke less negative emotions and

attitudes than CSIR unrelated to the negative CSR reputation.

H3: The CSIR domain interacts with the CSR reputation on negative emotions and attitudes toward the company. Specifically, a) negative emotions are higher and b) attitudes toward the company are lower, when the companies with negative CSR reputations face CSIR in an unrelated domain (to the CSR reputation) than when companies with negative CSR reputations face CSIR in a related domain (to the CSR reputation).

H4: Negative emotions mediate the influence of CSR reputation and CSIR domain on attitude toward the company.

Figure 2: conceptual framework CSR Reputation Pos./neg./ no CSR Emotions Attitude toward the company CSIR domain Related /unrelated

(24)

24

3. METHOD

In the following section the research design and data collection will be described. More specifically, the manipulation of the independent variable and the measurement of the dependent variables are outlined. Also information about the participants and results of the pretest is given.

3.1 Research design

To test my propositions an online survey with an experimental design was considered the best approach. The experimental design allows to create 9 different conditions (see Table 1 for the experimental conditions) and observe the effect of the manipulation of the independent variable on the dependent variables. Advantages of this approach are the ability to study causal links between variables and the reduced chance of biases. The combination with an online survey makes it possible to collect a large amount of data in a cost-effective manner. Standardized questions make it is easy to compare responses. Disadvantages of this approach are the forced choices respondents have to make, low response rate of online surveys and lower

generalizability of experiments to real-life settings (Saunders & Lewis, 2012).

To test the hypotheses I employed a 3 (CSR reputation: positive CSR reputation, negative CSR reputation and a control condition in which no information about CSR was provided) x 3 (CSIR domain: related CSIR domain; unrelated CSIR domain: environment; unrelated CSIR domain: animal) between-subjects design of CSIR information (see Table 1).

(25)

25

Table 1: Experiment design

CSR reputation CSIR domain N

Condition 1 Positive Related (employee domain) 45 Condition 2 Positive Unrelated (environment domain) 40 Condition 3 Positive Unrelated (animal domain) 34 Condition 4 Negative Related (employee domain) 38 Condition 5 Negative Unrelated (environment domain) 43 Condition 6 Negative Unrelated (animal domain) 46 Condition 7 No CSR reputation Related (employee domain) 46 Condition 8 No CSR reputation Unrelated (environment domain) 46 Condition 9 No CSR reputation Unrelated (animal domain) 49

3.2 Measures

3.2.1 Independent variables

In an introductory text, participants received background information about the fictitious fashion firm - Fashion Loft - and its CSR reputation. This research makes use of a fictitious company to limit the effects of participants having prior knowledge about the company. Although the company was fictitious, texts are all based on real issues in the fashion industry.

CSR reputation

Depending on the experimental conditions, the information about Fashion Loft included information about the positive CSR reputation, negative CSR reputation or did not include information about the CSR reputation. This control group that did not receive any information about the CSR reputation is used as a comparison group. Apart from the information about the CSR reputation the participants in the control group had the same information as the

participants in the other experimental conditions. By using this control group as a comparison group this research is able to isolate the independent variable (CSR reputation) and look at the impact it has on emotions and attitude toward the company. The positive CSR reputation and negative CSR reputation conditions both described Fashion Loft’s CSR reputation regarding the employee domain. In the positive CSR reputation condition, respondents read that Fashion Loft stands out with regard to fair working conditions and is known as the industry leader when it

(26)

26 comes to CSR. In the negative CSR reputation condition, respondents read that Fashion Loft stands out with the regard to unfair working conditions and is known as the industry laggard when it comes to CSR (See appendix 1 for the full text of the manipulations).

After reading the introduction, respondents rated the company on perceived ethicality as a manipulation check. Three items of Wagner et al.‘s (2009) study are used. This included the following questions: In my opinion Fashion Loft “is a socially responsible company”, “is concerned to improve the well-being of society”, “follows high ethical standards” (this study’s Cronbach’s α = .95). Questions are asked at a seven-point rating scale (from 1, completely disagree till 7, completely agree).

CSIR domain

The introductory text was followed by information about corporate social irresponsible behaviour of Fashion loft. Depending on the experimental conditions, the CSIR domain was related or unrelated to the CSR reputation domain (i.e. working condition of employees). The related condition contained CSIR information about the employee domain, namely sweatshop labour (related to the CSR reputation). Two other conditions, with information about

environmental pollution and animal abuse, serve as the unrelated CSIR domains (relative to the CSR reputation). The different domains (environmental, animal and employee domain)

represent CSIR domains that are identified in prior research (e.g. Wagner, Lutz & Weitz, 2009; Handelman & Arnold, 1999). I used the Wagner et al. (2009) method to instruct participants to imagine two weeks have passed in between the two pieces of information (the introductory text with CSR reputation and information about irresponsible behaviour). The full text of the

(27)

27

Pretest

Prior to the data collection I conducted, together with two other thesis students, 3 pretests of the manipulations. In the first pretest we assessed the effectiveness of the CSR reputation

manipulation. In total 24 Dutch respondents volunteered to participate in the first pretest. They answered three questions about the perceived ethicality across the CSR reputation conditions (Cronbach’s α = .98). In Table 2 you will find the means and standard deviations of the two different CSR reputation conditions on perceived ethicality. An ANOVA revealed that

participants in the positive CSR reputation condition exhibited significantly higher perceptions of ethicality than those in the negative CSR reputation condition (F(1, 22)= 203.11, p < 0.01). So the CSR reputation manipulation seems to work.

Table 2: Perceived ethicality by CSR reputation

CSR Reputation Mean SD N

Positive 5,86 0,87 12

Negative 1,53 0,59 12

Total 3,69 2,33 24

We ran a second and third pretest to ensure the validity, credibility and realism of the CSIR domain manipulations. In the second pretest 33 Dutch respondents participated. We included one related CSIR domain (employee domain) and 4 (semi) unrelated CSIR domains

(environmental crisis, animal crisis, community crisis, health care scandal). In table 3 and 4 you will find the means and standard deviations of the different CSIR domain manipulations on ethicality, harmfulness, responsibility, credibility and realism. Overall respondents perceived the CSIR domain manipulations as unethical (on a 1, unethical to 7, ethical scale), harmful (on a 1, not harmful at all to 7, extremely harmful scale), credible (on a 1, not credible to 7, very credible scale), realistic (on a 1, not realistic to 7, very realistic scale) and they held the company

responsible (on a 1, deserves no blame at all to 7, deserves very much blame). However,

respondents did not perceive the health care domain as unethical, harmful, credible and realistic as the other CSIR domain manipulations. Therefore, we choose to eliminate the health care

(28)

28 domain manipulation. Also the community and animal CSIR domain manipulations were not as unethical, harmful, responsible, credible and realistic as the employee and environment CSIR domain manipulations. Therefore we made some changes in these two manipulations to improve this. We made the animal CSIR domain manipulation more harmful, unethical and the company more responsible by adding a sentence in which we described how the fur has been plucked while the rabbits are still alive. We did the same for the community CSIR domain manipulation by adding a sentence in which we described that Fashion Loft would rather destroy its clothes than give them to the homeless. Both additions are based on real life examples (see appendix 1).

Table 3: Perceived ethicality, harmfulness and responsibility by CSIR domain (pretest 2) CSIR

domain

Ethicality Harmfulness Responsibility

N Mean SD Mean SD Mean SD

Employee 7 2,04 1,02 5,38 1,70 5,13 1,41 Environment 8 1,92 0,75 5,88 0,35 5,50 0,76 Animal 7 2,48 0,85 4,43 1,51 4,43 1,27 Community 6 2,33 0,75 3,86 1,51 4,29 1,47 Health care 6 2,56 0,81 3,50 1,38 4,67 1,21

Table 4: Credibility and realism by CSIR domain (pretest 2)

CSIR domain Credibility Realism

N Mean SD Mean SD Employee 7 4,88 1,57 4,75 1,60 Environment 8 6,00 0,76 5,50 0,75 Animal 7 5,14 1,70 4,29 0,69 Community 6 3,00 1,47 2,86 2,50 Health care 6 4,00 1,72 3,83 1,89

(29)

29 In a third pretest, with 24 Dutch respondents, we tested the CSIR domain manipulations again to ensure the improvements. The mean of the animal CSIR domain manipulation improved on ethicality, harmfulness, responsibility and realism. Realism was still rather low compared to the employee and environment conditions, but above the mid-point (i.e. 4). With this improvement the animal manipulation was comparable to the employee and environment CSIR domain manipulations. However, the community CSIR domain manipulation was still not seen as

unethical, harmful and responsible as the other three CSIR domain manipulations. Therefore, we choose to eliminate the community domain and only use the employee, environment and animal domain for our research. These three CSIR domain manipulations are comparable to each other. Credibility and realism are high across all three conditions. Perceptions of perceived ethicality on all the different CSIR domains are low and perceptions of harmfulness and responsibility were very high. See appendix for the full pretests.

Table 5: Perceived ethicality, harmfulness and responsibility by CSIR domain CSIR

domain Ethicality Harmfulness Responsibility

N Mean SD Mean SD Mean SD

Employee 7 2,05 0,45 5,00 2,00 5,71 0,76 Environment 5 1,80 0,77 5,80 1,10 5,40 1,67 Animal 6 1,56 0,75 5,17 2,14 5,00 2,45 Community 6 2,27 0,71 4,17 1,33 4,17 1,72

Table 6: Credibility and realism by CSIR domain

CSIR domain Credibility Realism

N Mean SD Mean SD

Employee 7 4,86 1,86 5,86 0,69 Environment 5 5,00 1,87 5,20 2,05 Animal 6 4,83 1,94 4,50 2,07 Community 6 5,67 0,52 5,33 1,21

(30)

30

3.2.2 Dependent variables

To test my propositions I measured the dependent variables (i.e. emotions and attitude toward the company) in my framework. I made use of multi-item measures and seven-point rating scales. To measure the dependent variable, attitude toward the company, I used the four items of Wagner et al.’s (2009) study: In general, my feelings toward Fashion Loft are

unfavourable/favourable, bad/good, unpleasant/pleasant, negative/positive (this study’s Cronbach’s α = .92). To measure the dependent variable, negative emotions, I asked the

experience of the following negative emotions used by Grappi et al. (2013): contempt, anger and disgust (this study’s Cronbach’s α = .79). I added favourable as a positive filler of the construct and left them out of the analysis.

3.2.3 Demographics and control variables

The demographic section at the end of the questionnaire included questions about participants’ age, gender and educational background. Moreover, some control variables were integrated. Consumer perceptions of hypocrisy and expectancy violation are measured to check if there is a mediating relationship between CSR reputation and CSIR domain on attitudes toward the company through hypocrisy and expectancy. To measure hypocrisy, three items of Wagner et. al.’s (2009) study are used. This scale included the following questions: In my opinion Fashion loft “acts hypocritically”, “pretends to be something that it is not”, “keeps its promises” (this study’s Cronbach’s α = .72). Expectancy violation is measured by using three items of Afifi, Walid and Metts’ (1998) study: Fashion Loft’s recent behaviour “was completely expected/not at all expected”, “surprised me a great deal/only slightly”, “was only very slightly unexpected/ completely unexpected” (this study’s Cronbach’s α = .95).

Furthermore, credibility and realism were measured (despite earlier control in the pretest) as a second check if the results are not biased. Both control variables are measured by one item. To measure credibility and realism respondents were asked the following questions: “The

(31)

31 information I read about the company was credible” and “The situation described was realistic”. All measurement scales made use of multi-item measures and seven-point rating scales.

Finally, instead of a manipulation check for the CSIR domain, this research made use of an attention check at the end of the questionnaire to test if the respondents could remember which CSIR domain they had read about. This choice was made, because a manipulation check of CSIR domain (related vs. unrelated to the CSR reputation) would have been biased by respondents CSR reputation perceptions. Therefore we asked the respondents if the news item they read was about a sweatshop scandal, pollution scandal or animal abuse scandal.

3.2 Participants

In line with earlier research, participants were 387 Dutch consumers between 15 and 65 years of age. The average age of the sample was between 25-34 years old and 66% were female. I tried to balance the male/female respondents, because of previously detected gender effects on emotional (cognitive) processes (e.g. Lindenmeier, Schleer & Pricl, 2012; Wagner, Bicen & Hall, 2008). Ninety-nine percent had at least a high school diploma, with 28,8% holding a HBO degree. During a period of 28 days data was collected through an online experiment. Questionnaires were spread through an electronic link and respondents were randomly assigned to one of the nine experimental conditions (see Table 1 for the distribution of the respondents per condition). On average the questionnaire took 9 min to complete.

(32)

32

4. RESULTS

In Table 7 you will find an overview of the means, standard deviations and correlations of all the control variables (gender, age, education, credibility, realism, hypocrisy and expectancy) and dependent variables (emotions and attitude toward the company) used in this study. I excluded the missing data pairwise, so I analysed only the cases without missing data in each pair of variables being analysed. After analysing the correlation matrix, education was excluded as control variable in further analyses because it does not correlate significantly with the variables of the theoretical model. Negative emotions correlates negatively with gender (r = -.12, p < .05) and positively with age (r = .15, p < .01). This means that men will express less negative

emotions than women and when participants are older they will express more negative emotions. Both control variables (gender and age) do not correlate significantly with the dependent variable attitude toward the company, but they will be used as a control because of the significant correlation with negative emotions. As expected, the dependent variables negative emotions and attitude toward the company are negatively correlated (r = -.40) and statistically significant at the 0.01 level.

(33)

33

4.1 Manipulation checks

Before testing the hypotheses a manipulation check is conducted. The manipulation of the CSR reputation is tested on perceived ethicality with 3 items (Cronbach’s α = .95). The manipulations worked as intended. Participants in the positive condition rated Fashion Loft’s actions

significantly (p < .01) more ethical (M = 5.25 SD = 1.19 ; on a 1, unethical to 7, ethical scale) than those in the negative condition (M = 2.15, SD = 1.17). In addition to the CSR reputation

manipulation test an attention check about the CSIR domain is done. At the end of the

questionnaire respondents had to remember which CSIR domain they had been exposed to. A percentage of 95,6% participants answered the question regarding their CSIR condition correct. However, the one who did not correctly answer the question about the CSIR condition are not excluded from the analysis, because earlier information about employees in the CSR reputation manipulation could influence their answer on this question.

Furthermore, the different CSIR domain conditions were measured on credibility and realism as a second check after the pretest. An ANOVA revealed that there are no significant differences between the different CSIR domain manipulations on credibility (p =.83) and realism (P = .38). All CSIR domain manipulations scored above the mid-point (i.e. 4) on credibility and realism.

4.2 Impact of CSR reputation on emotions and attitude toward the

company (hypothesis 1)

In hypothesis 1 I expected a) negative emotions to be higher and b) attitudes toward the company to be lower, when companies with positive CSR reputations face CSIR than when companies with no CSR reputation face CSIR. I did not expect significant differences between companies with a negative CSR reputation and companies with no CSR reputation. When controlling for age and gender, a MANOVA revealed a significant main effect of CSR reputation on negative emotions and attitude toward the company F(4,338) = 18,87 p < .01 (using Wilk’s lambda). Separate ANOVAs on the outcome variables confirmed that emotions F(2, 338) = 5.11,

(34)

34

p < .01 and attitude toward the company F(2, 338) = 33.24, p < .01 were not equivalent across all

three CSR reputation conditions.

However, opposed to the predictions, planned contrasts revealed no significant difference between the positive CSR reputation (M = 5.02, SD = 1.12) and no CSR reputation (M = 5.04, SD = 1.24) on negative emotions (see figure 3). Furthermore, those assigned to the negative CSR reputation condition (M = 5.47, SD = 1.11) expressed significant (p < 0.01) higher number of negative emotions than participants exposed to the no CSR reputation condition. Participants exposed to the negative CSR reputation expressed significantly (p < 0.01) higher number of negative emotions than participants who were exposed to the positive CSR reputation.

Figure 3: Mean negative emotions by CSR reputation condition

In additional tests about attitude towards the company, planned contrasts revealed the same pattern. No significant difference between the positive CSR reputation (M = 3.17, SD = 1.12) and no CSR reputation (M = 3.06, SD = 1.14) was found. Participants expressed significant (p < 0.01) lower attitudes in the negative CSR reputation condition (M = 2.09, SD = 1.14) than participants who were exposed to the no CSR reputation condition. Finally, participants expressed significant

4,7 4,8 4,9 5 5,1 5,2 5,3 5,4 5,5 5,6 Positive CSR

reputation No CSR reputation(control) Negative CSRreputation

Ne gati ve e m otions Reputation condition

(35)

35 (p < 0.01) lower attitudes toward the company in the negative CSR reputation condition than in the positive CSR reputation condition (see Figure 4).

Figure 4: Mean attitude toward the company by CSR reputation condition

These results demonstrate that contrary to my predictions, participants did not express significantly more negative emotions and lower attitudes toward the company in the positive CSR reputation relative to no CSR reputation. Instead, participants in the negative CSR reputation condition expressed more negative emotions and lower attitudes toward the

company relative to the positive CSR reputation and no CSR reputation. Therefore hypothesis 1 is rejected.

4.3 Interaction effect of CSIR domain and CSR reputation on emotions

and attitude toward the company (hypothesis 2 & 3)

This research expected an interaction effect of the CSIR domain and the CSR reputation on negative emotions and attitude toward the company. Contrary to my predictions, a 2-way MANOVA revealed that there was a non-significant interaction effect between CSIR domain and CSR reputation on negative emotions and attitude toward the company F(8, 329 ) = 0.98, p = .45

0 0,5 1 1,5 2 2,5 3 3,5 Positive CSR

reputation No CSR reputation(control) Negative CSRreputation

A ttitu d e tow ar d th e com p any Reputation condition

(36)

36 (using Wilk’s lambda and controlling for gender and age). Using Wilk’s lambda, there was a significant main effect of CSIR domain on emotions and attitude toward the company, F(4, 329) = 4.79, p < .01. However, separate univariate ANOVAs on the outcome variables revealed only a significant main effect of CSIR domain on emotions F(2, 338) = 7.14, p < .01 and non-significant effect of CSIR domain on attitude toward the company F(2, 338) = .28, p = .76.

In hypothesis 2 I expected a) negative emotions to be higher and b) attitude toward the company to be lower when companies with positive CSR reputations face CSIR in a related domain (to the CSR reputation) than when companies with positive CSR reputations face CSIR in an unrelated domain (to the CSR reputation). Instead, negative emotions were slightly higher in the unrelated environmental CSIR domain (M = 5.10, SD = .94) and unrelated animal CSIR domain (M = 5.17, SD = 1.09) than in the related CSIR domain (M = 4.86, SD = 1.29). Comparing the two unrelated CSIR conditions versus the related CSIR condition with planned contrast revealed no significant differences between the related and unrelated CSIR domains for companies with positive CSR reputations (p = .21). Previous analysis already showed no

significant difference of CSIR domain on attitude toward the company (p = .76). So, hypothesis 2 is rejected.

In hypothesis 3 I expected a) negative emotions to be higher and b) attitude toward the

company to be lower when companies with negative CSR reputations face CSIR in an unrelated domain (to the CSR reputation) than when companies with negative CSR reputations face CSIR in a related domain (to the CSR reputation). Planned contrast determined a significant difference between related and unrelated CSIR domains for companies with negative CSR reputations (p = .01). Participants in the unrelated CSIR domains (environment: M = 5.50, SD =.98 and animal: M = 5.78, SD =.95 ) expressed significant higher negative emotions than participants in the related CSIR domain (M = 5.03, SD =1.33). However, previous analysis showed no significant difference

(37)

37 of CSIR domain on attitude toward the company. Therefore, hypothesis 3 is supported for a) negative emotions, but not supported for b) attitude toward the company.

Figure 5: Estimated mean of emotions by CSR reputation and CSIR domain

Planned contrast showed no significant difference between the unrelated environment domain and unrelated animal domain (p = .24).

4.4 Moderated mediating role of CSR reputation and CSIR domain on

attitude towards the company through emotions (hypothesis 4)

Hypothesis 4 proposed that the relationship between the CSR reputation and CSIR domain on attitudes toward the company would be mediated by negative emotions. In other words hypothesis 4 expects that the effect of CSR reputation on attitude toward the company through negative emotions depends on the CSIR domain. The SPSS macro of Preacher and Hayes, model 7, was used to test the hypothesized moderated mediation effects between the variables. “Preacher & Hayes (2008) prefer the use of bias-corrected and accelerated confidence intervals (BCa) (as opposed to regression coefficients or regular confidence intervals) as a basis for statistical conclusions because it adjusts for bias and skewness of the dataset” (Berger, 1999).

4,50 4,70 4,90 5,10 5,30 5,50 5,70 5,90 Positive CSR reputation reputationNo CSR (control) Negative CSR reputation N ega tive emot io ns CSR reputation

related (employee domain) unrelated (environment domain) unrelated (animal domain)

(38)

38 The independent variable CSR reputation was dummy-coded into two variables (see appendix 6). The negative CSR reputation was not explicitly coded and functioned as the reference group. This means that parameters in the model are quantifications relative to the negative CSR reputation condition. I added a second dummy-code for the proposed moderator CSIR domain and divided this one in related versus unrelated CSIR domain (see appendix 6). Previous analysis showed no significant difference between the unrelated environment domain and unrelated animal domain (p = .24). Therefore unrelated environmental CSIR domain and unrelated animal CSIR domain were coded into one unrelated CSIR group. Attitude toward the company is added as a dependent variable and negative emotions is added as a mediator. In Figure 6 you will find the estimated model coefficients. Contrary to my predictions, there was a non-significant interaction effect between CSIR domain and CSR reputation (p = .51). The bias-corrected and accelerated confidence interval (BCa95= [-.27; .12]) included zero, indicating that

no support can be found for the proposed moderated mediating model (see appendix 6 for the full SPSS output). Therefore, hypothesis H4, pertaining to the moderated mediation of the relationship between CSR reputation and CSIR domain on attitudes toward the company through negative emotions is rejected.

Figure 6: Estimated model coefficients resulting from dummy coding CSR reputation conditions using

negative reputation as the reference group (**_{p < .01)}

a2: -.47** a: -.46** CSR reputation (pos vs. neg) Emotions Attitude toward the company c’: +.93** b: -.36** c: +1.09** CSIR domain (Rel vs. Unrel)

Corporate social irresponsibility : the influence of CSR reputation on corporate social irresponsibility