Overcoming negative Outcomes of Corporate Social Responsibility Communication: The Role of CEO Behavior and Stakeholder Involvement
By Silvio Kunz 10602305
Master’s Thesis
Graduate School of Communication
Master’s program Communication Science: Corporate Communication Supervisor Dr. Piet Verhoeven
Abstract
Corporate social responsibility (CSR) communication is a widely discussed topic in academia and practice. When done well, CSR communication can enhance stakeholders’ evaluations of a company. However, communicating about CSR programs can go wrong, too. Companies are often faced with negative outcomes. Organizational members such as the Chief Executive Officer (CEO) likely influence the outcomes of the company’s CSR program through her or his actions. It can be argued that stakeholders who perceive that a CEO does not intrinsically care about a CSR issue, which is targeted by the company’s CSR program might develop skepticism and unfavorable attitudes. Therefore, a better understanding of the role of organizational members with regard to CSR communication is needed. This research sets out to investigate how perceived inconsistency in CEO behavior in relation to the company’s CSR program affects stakeholders’ unfavorable attitudes toward the organization and what roles skepticism and involvement in the CSR cause play. A three-group posttest-only experiment was conducted with a fictional organization, a fictional CEO and a fictional CSR program. 376 participants were randomly assigned into one of the three conditions: Inconsistent CEO behavior in relation to the CSR program, consistent behavior and the control group. The findings show that stakeholders who are exposed to inconsistent CEO behavior develop significantly higher levels of skepticism and unfavorable attitudes toward the organization than stakeholders who are exposed to consistent behavior. Specifically, perceived inconsistency in CEO behavior is positively related to unfavorable attitude toward the organization and skepticism mediates the relationship. Furthermore, involvement is shown to strengthen the relationship between skepticism and unfavorable attitude toward the
organization. Overall, the findings suggest that management should carefully choose specific CSR causes to incorporate in the company’s activities because the CEO, and likely other organizational members, need to truly get behind the CSR cause.
Introduction
Corporate Social Responsibility (CSR) and its communication has become of great importance for organizations. In an article in The Economist (Do it right., 2008) it was stated that “[...] it is almost unthinkable today for a big global corporation to be without one [CSR policy].” But how do stakeholders perceive a company and its CSR program if core
organizational members such as the Chief Executive Officer (CEO) do not show their commitment to a CSR cause that their company incorporated into its CSR strategy? Imagine you are a customer of a fashion company and you saw on its website that the company aims to reduce its CO2 emission in the production of the clothes. One day, you read in an article that the CEO privately does not care about reducing the CO2 emission because she or he travels by plane between geographically very close cities instead of taking the train, which would take the same amount of time, but that would emit far less CO2. Would you become skeptical toward the CSR program? Would your attitude toward the company change because you perceived that the CEO’s behavior is inconsistent with the CSR program?
Mohr and Webb (2005) argued that some companies do not initiate CSR programs into their activities because the response of consumers to CSR was not reliable and strong. There has been a rise of skepticism toward CSR programs of organizations and in particular toward CSR communication (Skarmeas & Leonidou, 2013). It was shown in empirical studies that skepticism likely leads to negative word of mouth, lower resilience to negative
information and lower retailer equity (Skarmeas & Leonidou, 2013) as well as to a negative attitude toward the organization and lower purchase intention (Elving, 2013). Overcoming skepticism was argued to be a key challenge of CSR communication (Du, Bhattacharya, & Sen, 2010). Research provided evidence that a positive prior organizational reputation (Elving, 2013) and CSR frames from organizations in socially stigmatized sectors
negative outcomes of CSR communication would be valuable for organizations and consequently for society at large. This is because it could likely encourage companies to incorporate social and environmental issues into their business activities and to communicate about these.
However, academia is far away from fully understanding how to reduce skepticism in CSR communication in order to achieve non-financial outcomes such as a favorable attitude toward the organization. In specific, there is a lack of knowledge about how the perception of the behavior of organizational members such as the CEO in relation to the company’s CSR program influences stakeholders attitudes. Elving and Kartal (2012) tested in an experiment whether the (in-) consistent behavior of a CEO in relation to the CSR program of the organization influences the attitude toward the organization, purchase behavior and
skepticism among stakeholders. However, their results showed no significant changes in the outcome variables. The present experimental study builds on Elving and Kartal’s (2012) experiment and reinvestigates the role of perceived inconsistency in CEO behavior in relation to the CSR program in reducing skepticism and decreasing unfavorable attitudes toward the organization. In particular, this study aims to investigate how perceived inconsistency in CEO behavior in relation to the company’s CSR program affects stakeholders’ unfavorable
attitudes toward the organization, and what roles skepticism toward the CSR program and involvement in the CSR cause play.
Theoretical Framework
There has been a shift from a traditional view under which companies were expected to solely make profits to a society-based model: Organizations have to gain legitimacy to operate from the larger society in which they are embedded (Cornelissen, 2011). In recent definitions, the breadth of the CSR concept differs in regard to which issues they include.
While scholars such as Vanhamme and Grobben (2009) defined CSR narrowly by including “[…] legal, economic, ethical, and discretionary responsibilities […]” (p. 273), others defined it more broadly by incorporating environmental issues, too. For example, CSR in a broader sense was defined by Ihlen, Bartlett and May (2011) as an organizational attempt to negotiate the relationship between an organization and its stakeholders and the public in which CSR focuses at a minimum on “[…] economic, social and/or environmental issues” (p. 8). To meet expectations posed upon organizations by stakeholders, companies develop CSR programs. These are organizational attempts to fulfill the economic, social and/or environmental obligations across all stakeholder groups and the public and involve all aspects of an organization (Pirsch, Gupta, & Grau, 2006). Explanations for why organizations aim to be socially and environmentally responsible can be given using the theory of property rights and the stakeholder theory. First, the theory of property rights states that managers have
responsibilities toward all stakeholder groups and cannot solely act in favor of shareowners (Donaldson & Preston, 1995). Consequently, a company does not have unlimited rights, but has obligations to society at large. Second, the stakeholder theory posits that companies must consider both economic and non-economic objectives to survive and to be successful. Thus, they need to take interests of all stakeholder groups into account (Argandoña, 1998; Freeman, 1984; Pirsch et al., 2006).
CSR communication is how organizations inform stakeholders and the public about their CSR programs: It is “[…] the ways that corporations communicate in and about this process [CSR]; it is the corporate use of symbols and language regarding these matters.” (Ihlen et al., 2011, p. 8). Moreover, CSR can be understood within an instrumental or functionalistic approach as the actions and the communication about those actions (Golob, Podnar, Elving, Nielsen, Thomsen, & Schultz, 2013; Schultz, Castelló, & Morsing, 2013). In doing so, it was argued that CSR communication has to be consistent in order to be
considered as trustworthy and legitimate (Schultz et al., 2013). The field of integrated communications is based on the rationale that in today’s markets, stakeholders are
increasingly skeptical, which makes it inevitable for companies to communicate consistently (Christensen & Langer, 2009). Therefore, organizations need to align “[…] symbols,
messages, procedures and behaviours across formal organizational boundaries.” (Christensen, Firat, & Torp, 2008, p. 423). Van Riel and Fombrun (2007) further argued that consistency likely increases stakeholders’ trust in the organization. In contrast, other scholars argued for more diversity and openness between words and actions (e.g., Christensen & Langer, 2009; Christensen, Morsing, & Cheney, 2008; or Christensen, Morsing, & Thyssen, 2013).
For management, perceived benefits of CSR were found to be a reduction of employee turnover, an increase in customer satisfaction and the inference of positive characteristics of the organization (Galbreath, 2010). Experimental research has confirmed that the outcome of CSR is a more positive firm evaluations (Brown & Dacin, 1997; Mohr & Webb, 2005) and a better attitude toward the brand (Lafferty & Goldsmith, 1999). In addition, a meta analysis of CSR literature and research by Aguinis and Glavas (2012) revealed that especially positive firm evaluation and reputation are consistent outcomes of CSR.
Although CSR was found to have positive outcomes for organizations, Mohr and Webb (2005) argued that some companies do not initiate CSR programs because consumers’ response to CSR was not reliable and strong. In achieving the goals of a company’s CSR program, CSR communication is of great importance (Golob et al., 2013). However, a controversial topic in academic literature is whether or not to explicitly communicate about CSR because stakeholders might develop distrust when being exposed to CSR
communication (Schultz & Wehmeier, 2010) and might accuse such companies of
greenwashing (Schultz et al., 2013). According to the self-promoter’s paradox, organizations likely face skepticism from their stakeholders when promoting CSR programs (Ashforth &
Gibbs, 1990; Morsing & Schultz, 2006). Organizations seek legitimacy from stakeholders, but at the same time stakeholders are suspicious because they disregard the legitimation attempts (CSR programs) as manipulative or self-serving (Ashforth & Gibbs, 1990). Ashforth and Gibbs (1990) stated that “The greater the need for legitimation, the more suspect of legitimation attempts are constituents [stakeholders].” (p. 186). Overall, it was argued that there was a rise in skepticism toward CSR communication (Skarmeas & Leonidou, 2013), and overcoming skepticism was a key challenge for CSR communication (Du et al., 2010;
Waddock & Googins, 2011).
CEO Behavior
The CEO symbolizes to stakeholders what the organization stands for - she or he is the face of the company (Elving & Kartal, 2012; Treadway, Adams, Ranft, & Ferris, 2009). Elving and Kartal (2012) argued that the behavior of the CEO in relation to the company’s CSR program is related to stakeholders’ attitudes toward the organization, and Treadway et al. (2009) proposed that CEO reputation is positively related to corporate reputation. The argument of Elving and Kartal (2012), which states that stakeholders’ perception of an inconsistent behavior of the CEO in relation to a company’s CSR program likely affects outcomes of CSR communication such as the attitude toward the organization, can be underlined with literature of corporate reputation. Stakeholders’ attitudes toward the
organization and corporate reputation are closely related (Meijer & Kleinnijenhuis, 2006). A person’s attitude toward an object such as an organization “[…] is a psychological tendency that is expressed by evaluating a particular entity with some degree of favor or disfavor.” (Eagly & Chaiken, 1993, p. 1). Evaluations are both a critical feature, and an observable consequence of attitudes (Eagly & Chaiken, 1993). On the other hand, corporate reputation can be defined as an evaluation or assessment of a company (Barnett, Jermier, & Lafferty,
2006), such as by Gotsi and Wilson (2001). Organizations with strong reputations are characterized by visibility, distinctiveness, authenticity, transparency and consistency (Fombrun & van Riel, 2004). Therefore, according to the scholars, companies that are perceived as consistent with their communications and actions likely have a favorable reputation.
The inside-out approach further underlines the importance of organizational members. According to the approach, employees are the key component in building trustworthiness (Morsing, Schultz, & Nielsen, 2008), which on the other side can be regarded as a component of corporate reputation (Wang, Berens, & van Riel, 2012). In the field of corporate branding, Christensen et al. (2008) argued that employees as brand ambassadors are expected to align their personal values with the identity of the corporate brand. The importance of consistency was additionally highlighted by Brønn and Vrioni (2001): “Only a consistent, believable contribution to a cause can build brand image and brand equity.” (p. 218).
Attribution theory, which was originally developed by Heider (1944), describes how people make causal inferences to explain the motives of others’ behavior. It can be used to explain how stakeholders interpret and react to CSR communication (e.g., Ellen, Webb, & Mohr, 2006; Forehand & Grier, 2003; Parguel, Benoît-Moreau, & Larceneux, 2011; or Yoon, Gürhan-Canli, & Schwarz, 2006) and CEO behavior (Elving, 2013). Forehand and Grier (2003) distinguished between public-serving and firm-serving motives that are attributed to organizations’ actions. Whereas public-serving motives refer to focusing on the well-being of people outside and inside of the organization, firm-serving motives solely relate to needs of the company. The scholars argued that consumers prefer the motives behind a company’s actions to be public-serving rather than firm-serving. In line with this reasoning, if an
organization is attributed firm-serving motives, consumers perceive the action negatively and are likely to have a more unfavorable attitude toward the organization (Forehand & Grier,
2003; Du et al., 2010) and to develop skepticism (Du et al., 2010; Elving, 2013; Skarmeas & Leonidou, 2013). Following this line of argument, Elving and Kartal (2012) reasoned that if stakeholders perceive inconsistent CEO behavior in relation to the CSR program, they likely attribute firm-serving motives to the organization’s CSR program because they likely believe the organization does not incorporate the CSR program to do good, but to increase
stakeholders’ attitudes or purchase intention. On the other hand, Elving and Kartal (2012) argued that if stakeholders perceive consistent behavior, they likely develop public-serving motives.
Elving and Kartal (2012) tested the relationship between consistency in CEO behavior and stakeholders’ attitude toward the organization, but did not find significant results.
However, hypocrisy– an inconsistency between CSR communication and an organization’s action – was found to be negatively related to the attitude toward an organization (Wagner, Lutz, & Weitz, 2009), and it was shown in a recent literature review that CSR itself was consistently found to predict positive firm evaluations (Aguinis & Glavas, 2012).
Nevertheless, no previous research was found that supports the relationship between CEO consistency and attitude toward the organization.
Nonetheless, based on the above arguments, it can be reasoned that stakeholders who perceive that a CEO does not behave in line with the CSR program likely have a more unfavorable attitude toward the organization. When adapting the above arguments from corporate branding to CSR, it can be reasoned that if employees, as brand ambassadors, and in particular the CEO do not align personal values about a cause with the CSR program, stakeholders likely perceive inconsistent behavior. This, in turn, likely affects outcomes of CSR communication such as the attitude toward the organization. Furthermore, by adapting the arguments from corporate reputation literature to CSR, a perceived inconsistency in CEO behavior is likely related to an unfavorable reputation – or an unfavorable attitude toward the
organization. Applying the attribution theory to CSR, it can be assumed that when
stakeholders perceive inconsistent CEO behavior, they think the CSR program is not credible and the organization’s motives are not value driven or public-serving, which in turn likely leads to a more unfavorable attitude. This leads to the following hypotheses:
H1a: Stakeholders who are exposed to inconsistent CEO behavior in relation to the organization’s CSR program have a more unfavorable attitude toward the organization than stakeholders who are exposed to consistent CEO behavior.
H1b: Perceived inconsistency in CEO behavior in relation to the organization’s CSR program relates positively to unfavorable attitude toward the organization.
Mediating Role of Skepticism
Skepticism was defined by Obermiller and Spangenberg (1998) as a tendency toward disbelief and by Boush, Kim, Kahle and Batra (1993) as the overall tendency to question. Pomering and Johnson (2009) argued that skepticism was one response to advertisements. It can be divided into situational skepticism which varies with regard to the situation a person is in (Mohr, Eroǧlu, & Ellen, 1998) and dispositional skepticism which is seen as a personality trait (Boush et al., 1993; Obermiller & Spangenberg, 1998).
Stakeholders respond with (situational) skepticism when they perceive that an organization intends to develop a favorable reputation instead of to improve social and environmental issues (Elving & Kartal, 2012). Following the arguments of the attribution theory (Heider, 1944), scholars argued that if consumers attributed firm-serving motives behind a company’s actions, they likely develop skepticism (Du et al., 2010; Elving, 2013; Skarmeas & Leonidou, 2013). A recent study provided evidence that the attribution of egoistic (similar to firm-serving) motives increased consumer skepticism and that the
attribution of value driven (similar to public-serving) motives lowered consumer skepticism (Skarmeas & Leonidou, 2013).
Recently, studies have focused on skepticism in relation to CSR communication (Elving, 2013; Elving & Kartal, 2012). Elving and Kartal (2012) did not find significant results for the relationship between inconsistency in CEO behavior and skepticism. However, in a later study, the relationship between (consumer) skepticism and attitude toward an organization was supported (Elving, 2013).
Heider (1958) described triadic relationships in the balance theory, which can be used to explain skepticism as a mediator of the relationship between perceived inconsistency in CEO behavior and attitude toward the organization. A mediator such as skepticism is a construct that explains a relationship – the independent variable can affect the dependent variable because of the generative mechanism of the mediator skepticism (Baron & Kenny, 1986). The central assumption of the balance theory is that people want to have balanced relationships. Imbalance causes tension: Hence, people try to balance it (Heider, 1958). Applied to CSR, the theory could explain the relationship between a stakeholder, the perception of a company’s CEO and the CSR program (Elving & Kartal, 2012). When perceiving consistent CEO behavior, a balanced relationship is predicted: A stakeholder is assumed to like both the CSR program and the CEO because she or he behaves in line with the CSR program. Consequently, they likely do not develop skepticism and their level unfavorable attitude toward the organization is low. However, in case of perceived
inconsistent behavior, an imbalanced relationship occurs: The stakeholder is expected to like the CSR program, but not the CEO because of the inconsistent behavior. In this situation, the stakeholder tries to balance the relationship based on the central assumption of the balance theory (Heider, 1958). In doing so, she or he either decides that the inconsistent behavior of the CEO was not as bad as initially thought, or that the CSR program was not as great as
previously assumed. It can be argued, however, that stakeholders do not pursue the first tactic because if they like the CSR program, an inconsistent behavior can hardly be downplayed. Therefore, when choosing the latter balancing tactic, stakeholders are predicted to become skeptical toward the CSR program and are likely to develop a more unfavorable attitude toward the organization.
Based on the above reasoning and on the findings of previous research, it can be reasoned that inconsistent behavior of a CEO is likely to lead to a more unfavorable attitude toward the organization because it increases the level of skepticism. Following the argument of the attribution theory (Heider, 1944), it can be predicted that if stakeholders perceive firm-serving motives such as solely striving for profit without concerning about the cause, they will likely be more skeptical. Hence, stakeholders who perceive inconsistent CEO behavior will have a more unfavorable attitude toward the company because they are more skeptical. The above reasoning leads to the following hypotheses:
H2a: Stakeholders who are exposed to inconsistent CEO behavior in relation to the organization’s CSR program have a higher level of skepticism toward the CSR program than stakeholders who are exposed to consistent CEO behavior.
H2b: The relationship between perceived inconsistency in CEO behavior in relation to the organization’s CSR program and attitude toward the organization is mediated by
skepticism toward the CSR program.
Involvement
The Elaboration Likelihood Model (ELM) proposed by Petty and Cacioppo (1981) and the heuristic-systematic model by Chaiken (1980) explain why people’s attitudes change, namely due to their motivation and ability to process information (Petty & Cacioppo, 1996). The scholars distinguished between the central/systematic route and the peripheral/heuristic
route of information processing. People who are able and motivated to process information follow the central route, while others who are either not able or unmotivated follow the peripheral route. In the central route, strong arguments lead to favorable thoughts; however, weak arguments elicits counterarguments (Petty & Cacioppo, 1996). On the other hand, people who follow the peripheral route due to a lack of motivation or ability process judgment cues that are easily observable and do not demand much cognitive processing (Chaiken & Trope, 1999).
Applied to CSR communication, it can be argued that stakeholders are able to process information. However, they likely differ in their motivation because they have different levels of involvement in the CSR cause. People who are involved in the cause and therefore
motivated to process information can be assumed to develop more unfavorable attitudes toward the organization when they have a higher level of skepticism because they are likely to develop counterarguments. By contrast, stakeholders who are not involved in the CSR cause and therefore not motivated to process information follow the peripheral route and likely perceive the company’s mere incorporation of CSR activities as positive because they only perceive easily observable cues and do not invest energy in processing the information, despite of their levels of skepticism. In turn, the perception of easily observable cues likely results in a less unfavorable attitude toward the organization. This reasoning leads to the following hypothesis:
H3: Involvement in the CSR cause strengthens the relationship between skepticism toward the CSR program and unfavorable attitude toward the organization.
In sum, the theoretical framework laid out in the above reasoning is two-fold. First, stakeholders who are exposed to inconsistent CEO behavior are predicted to differ in their scores on skepticism and unfavorable attitude toward the organization compared to
it is predicted that skepticism toward the CSR program mediates the relationships between perceived inconsistency in CEO behavior and unfavorable attitude toward the organization. In addition, involvement in the CSR cause moderates the effect between skepticism toward the CSR program and unfavorable attitude toward the organization.
Figure 1. Theoretical model.
Method
To test the theoretical framework, an online experiment was conducted through an online survey in Qualtrics.com. In the experiment, consistency in CEO behavior in relation to the organization’s CSR program was manipulated, and the independent variable perceived inconsistency in CEO behavior in relation to the CSR program, the dependent variable unfavorable attitude toward the organization, the mediator skepticism, and the moderator involvement were measured. In particular, a three-group posttest-only design was applied with the between-subject variable exposure to (in-) consistent CEO behavior. No pretest was included because first, it was assumed that participants would identify the purpose of the study when being asked to rate statements regarding the central variables. Second, participants were exposed to information about a fictional organization, CEO and CSR program. This decision was made to prevent participants’ variations in knowledge about the organization, CEO and CSR program. Therefore, the conducted study implemented a
quasi-Involvement Unfavorable attitude toward the organization Skepticism Perceived inconsistency in CEO behavior
experimental design in which all the components of the classical experimental design were present except the pretest. Participants of the inconsistent CEO behavior group were exposed to behavior in which the CEO did not behave in line with the CSR program while participants of the consistent CEO behavior group were exposed to behavior in which the CEO did behave in line with the CSR program. On the other hand, participants of the control group were not exposed to CEO behavior in relation to the company’s CSR program. In the experiment, participants were randomly assigned into a group. The consistent CEO behavior group served as a base group against which the inconsistent CEO behavior group was compared in the analysis. However, participants of the control group were neither exposed to inconsistent nor consistent CEO behavior in case non-significant results were found. In such a case, it could be tested whether or not participants that were exposed to inconsistent CEO behavior differed in their scores on the dependent variables compared to participants who were in the control group.
Sample
In December 2014, 611 people were recruited within 11 days to take part in the experiment. They were recruited through (1) the personal network of the graduate student, (2) posts in Dutch Facebook groups (see Appendix A), and (3) through the dissemination of flyers (see Appendix B) at the University of Amsterdam, the Free University of Amsterdam, and the University of Utrecht. Hence, the resulting sample was a convenience sample. As an incentive to take part in the survey, one randomly selected participant was given 100 Euros. There was a dropout rate of 26.51% (162) during the survey; most of them stopped before being exposed to the stimulus materials. Another 11.95% (73) were removed because (1) they were exposed to the experimental material for less than 10 seconds (27 participants), (2) they indicated to currently not live in the Netherlands (26 participants), (3) they indicated to know
the fictional company (8 participants), (4) they had a clear response pattern (6 participants), (5) they knew the aim of the study (3 participants), and (6) they answered the majority of the questions which tested if they read the stimulus materials wrong (3 participants). Hence, the sample consisted of 376 participants, from which 65.7% (247) were women, 33.5% (126) men, 0.5% (2) chose the option ‘other’, and one participant did not indicate the gender (N = 376). Participants were on average 24.68 years old (SD = 6.16) and the majority of them held or pursued a Master’s degree from a research university (n = 118). The 376 participants were evenly assigned into conditions: 128 participants were assigned into the inconsistent CEO behavior group, 126 participants were assigned into the consistent CEO behavior group and 122 participants were assigned into the control group.
Procedure
Once participants entered the survey (see Appendix C), they were randomly and evenly assigned into one of the three groups. Then, they were introduced to the experiment, which was described to them as aiming to study information processing. Next, participants had to read and accept an informed consent. The research study was conducted under the responsibility of the Amsterdam School of Communication Research (ASCoR), and its ethical rules applied: (1) participants’ anonymity was guaranteed, (2) participants took part in the study voluntarily, and (3) participants were not exposed to offensive material or deliberately mislead. By accepting the informed consent, participants agreed on these terms and were willing to take part in the experiment. After that, they were introduced to the fictional fashion company Fynn through an online newspaper article, which was produced for the present experiment (see Appendix D). They read that the Finnish company plans to enter the Dutch market after being successful in the Scandinavian and U.S. markets. Additionally, the article contained background information about Fynn, its mission and a paragraph about its CSR
program. A fashion company was chosen because it was assumed that clothes were a relevant product category for all participants. After having read the article, participants were
introduced to Fynn’s CSR program through a screenshot of its website, which was constructed for the purpose of the present experiment (see Appendix E). It contained information on the program that aimed to reduce Fynn’s carbon footprint. The choice for communicating CSR through Fynn’s website was made because the data of the national Reputation Quotient surveys in 2005 showed that Danes and Norwegians indicated that communicating about CSR through minimal releases such as annual reports or websites was the most appropriate way (Morsing & Schultz, 2006). In addition, Ashforth and Gibbs (1990) argued that communicating about CSR programs through minimal releases is preferred by stakeholders because, according to the self-promoter’s paradox, conspicuous communication decreases legitimacy. Furthermore, an environmental CSR cause was chosen because it could be directly related to the fashion company. Before participants were exposed to the stimulus material in form of another online newspaper article in which CEO behavior was manipulated (see Appendix F), they answered several questions to test if they had read the article and screenshot of the website. After being exposed to the stimulus material, participants answered two questions that tested whether they had read the article and in addition, they were asked to state the aim of the study. Then, participants of the inconsistent and consistent behavior groups were asked to take part in the perceived inconsistency in CEO behavior measure. The control group was not presented with the measure because they were neither exposed to consistent nor inconsistent CEO behavior. Afterwards, all participants rated statements and answered questions that measured skepticism, involvement and unfavorable attitude toward the organization. Next, participants were asked to provide demographic data. In the last part of the survey, participants who wanted to take part in the lottery to win 100 Euros and who wished to receive a summary of the results of the study could provide their e-mail addresses.
Before leaving the survey, participants were debriefed and thanked for their time. About one month after data collection, a summary of the study’s main insights was sent to the
participants and 100 Euros were paid to a randomly selected participant.
Stimulus Material
The exposure to inconsistency in CEO behavior in relation to the company’s CSR program was manipulated in an online newspaper article that reported the opening of the first Fynn store in Amsterdam (see Appendix F). All participants read the same article, with the exception of one paragraph consisting of three sentences in which the private travel plan of the CEO was described (see Table 1). Participants of the inconsistent CEO behavior group read that the CEO would fly from Amsterdam to Paris after the opening presentation and the ones of the consistent CEO behavior group read that she would take the train. Additionally, both groups read two sentences that highlighted the (in-) consistent behavior. The choice of exposing participants to information about a private travel plan of the CEO was made because it relates to CO2 emission, which was the target of the company’s CSR program. Participants were assumed to know that the travel time between Paris and Amsterdam by plane and train
Table 1
Manipulation of CEO behavior
Inconsistent CEO behavior group Consistent CEO behavior group “Shortly after [the CEO’s presentation], she ended the
opening presentation and was off to catch her flight to Paris for a short family vacation.
Jeroen Visser, a journalist from the fashion, arts and lifestyle Magazine Blend, said at the opening that in private, Mrs. Laine does not live up to the company’s mission to reduce its Carbon Footprint. She could have gone by train to Paris, but instead chose to travel by plane which emits far more CO2 per passenger than traveling by train.”
“Shortly after [the CEO’s presentation], she ended the opening presentation and was off to catch her train to Paris for a short family vacation.
Jeroen Visser, a journalist from the fashion, arts and lifestyle Magazine Blend, said at the opening that also in private, Mrs. Laine lives up to the company’s mission to reduce its Carbon Footprint. She could have flown to Paris, but instead chose to travel by train which emits far less CO2 per passenger than traveling by plane.”
was approximately the same because it is a popular destination for people living in the Netherlands. In addition, it was highlighted in the third sentence of the manipulation that travelling by plane emits more CO2 per passenger than travelling by train. As previously elaborated on, the control group was not exposed to CEO behavior in relation to the company’s CSR program that was communicated in these three sentences. They only read that the CEO would travel to Paris for a short family vacation after the opening presentation.
Pilot Study
A pilot study was conducted in which 22 participants who were recruited from the personal network of the graduate student were asked to take part in the online experiment described above. Participants were asked to leave suggestions for improvements or comments in every part of the experiment. The pilot study was conducted for the following reasons: (1) to test whether the two online newspaper articles and the screenshot of Fynn’s website containing information of its CSR program were understood by participants, or whether certain parts needed rephrasing, (2) to assess whether the measures of the central variables were reliable, or whether certain items needed to be rephrased, and (3) to test whether participants of the inconsistent CEO behavior group indicated a higher perceived
inconsistency in CEO behavior score than participants of the consistent CEO behavior group. Out of the 22 participants, four were removed because they did not complete the survey. Therefore, the sample consisted of 18 participants who were randomly assigned into one of the three groups. Based on their comments and suggestions, the following adjustments to the online newspaper articles were made: First, the line which separated the title of the articles from the text was removed because a participant indicated newspaper articles usually did not include this line. Second, the last sentence of the introduction article was divided into
two because it was indicated by a participant that it was not a journalistic writing style.1
Third, the word caps was exchanged with winter hats in the articles that manipulated the CEO behavior because a participant commented that the latter word was more frequently used. Fourth, the CEO’s (in-) consistent behavior was highlighted in the article that manipulated her behavior through a direct quote from a journalist who attended the opening. A participant suggested to use an indirect quote instead because she or he did not think it would be directly quoted in a real newspaper article. Thus, an indirect quote was used. In addition, the
following improvements were made to Fynn’s website: The font color of the text was
changed from grey to black in order to increase the readability and the quote at the top of the page was rephrased in order to enhance the understandability.2
The measure perceived inconsistency in CEO behavior was found to be reliable (α = .93). Nevertheless, in the third item, the word strive was exchanged with effort because a participant suggested that it might not be understood by all participants. Skepticism was found to be a reasonably reliable measure (α = .80) and no adjustments were made.
Additionally, involvement in the CSR cause was found to be a reliable measure (α = .88) and unfavorable attitude toward the organization was found to be a reasonably reliable measure (α = .69). However, in order to achieve a higher reliability of the latter two measures, the
following adjustments were made: First, the items of the semantic differential scales were flipped around in the experiment indicating favorable attitudes and more involvement on the right side and unfavorable attitudes and less involvement on the left side of the bipolar scale.
1 From “The company changed its manufacturing for the aim of reducing the Carbon Footprint of its
clothes – the total amount of CO2 and other greenhouse gases that are emitted as part of its manufacture” to “The company changed its manufacturing for the aim of reducing the Carbon Footprint of its clothes. The Carbon Footprint represents the total amount of CO2 and other greenhouse gases that are emitted.”.
2 . From “It’s important to remember that greenhouse gas emissions aren’t just transport fumes. They
come from all over the supply chain. To make one tonne of nitrogen fertilizer takes one tonne of oil, one 2 . From “It’s important to remember that greenhouse gas emissions aren’t just transport fumes. They
come from all over the supply chain. To make one tonne of nitrogen fertilizer takes one tonne of oil, one hundred tonnes of water and creates seven tonnes of CO2. Going organic could do more good than electric transport.” to “It’s important to remember that greenhouse gas emissions aren’t just transport fumes. Emissions originate from all over the supply chain. Making one ton of nitrogen fertiliser takes one ton of oil, one hundred tons of water and creates seven tons of CO2. Going organic in the fashion industry could do more good than electric transport already does.”.
One item of the attitude measure and two items of the involvement measure were not flipped in order to be able to assess whether participants read the items and not simply chose the box on either one side. Second, whereas participants were presented with 5-point Likert scales in the attitude and involvement scales in the pilot study, 7-point Likert scales were used in the experiment for the sake of consistency.
Finally, the six participants of the inconsistent CEO behavior group (M = 6.38, SD = 0.51) had a significantly higher average score on perceived inconsistency in CEO behavior compared to the six participants of the consistent CEO behavior group (M = 3.50, SD = 1.80),
t(10) = -3.74, p = .004, 95% CI [-4.59, -1.16]. Therefore, it could be reasoned that the
manipulation of the exposure to (in-) consistent CEO behavior in relation to the company’s CSR program was successful.
Measures
Perceived inconsistency in CEO behavior. To measure the independent variable, a new measure was constructed and tested in the pilot study. Participants of the consistent and inconsistent CEO behavior group had to state the extent to which they agreed with four questions on a 7-point Likert scale ranging from (1) strongly disagree to (7) strongly agree. Before the analysis, the items were reversely coded to reflect the underlying concept of perceived inconsistency in CEO behavior. In order to enhance reliability, one of the four questions was not included in the measure. A Principal Component Analysis (PCA) confirmed that the resulting three items (see Table 2) build an uni-dimensional scale (eigenvalue 2.69), which accounts for 89.75% of the variance. Therefore, the measure perceived inconsistency in CEO behavior was constructed by averaging participants’ scores on the three items (α = .94), with low scores indicating the perception of consistent behavior and high scores representing the perception of inconsistent behavior, M = 3.95, SD = 1.95.
Skepticism. The degree to which participants were skeptical about the organizations’ CSR program was measured with a 4-item measure developed by Skarmeas and Leonidou (2013). Participants had to indicate the level of agreement with statements on a 7-point Likert scale ranging from (1) strongly disagree to (7) strongly agree. In addition to the four
statements of the original scale, a fifth item that incorporated the word skeptical was added. A PCA confirmed that the five items (see Table 2) build an uni-dimensional scale (eigenvalue 2.92) that accounts for 58.32% of the variance. The resulting measure, which was composed by averaging the five items, was found to be reliable (α = .82), M = 3.49, SD = 1.12.
Involvement. The level of involvement in the CSR cause was measured based on a bipolar 1-7 semantic differential scale originally developed by Zaichkowsky (1985) and adapted by Parsa, Lord, Putrevu and Kreeger (2015). Participants were presented with the sentence “Buying products that are manufactured in an environmentally friendly way is…” and had to indicate their position between opposing adjectives (see Table 2). One item was excluded from the measure in order to enhance its reliability. A PCA showed that the three opposing items build an uni-dimensional scale (eigenvalue 2.68) which accounts for 89.31% of the variance. The final measure averaged the scores of the three items and was found to be reliable (α = .94), M = 5.08, SD = 1.36.
Unfavorable attitude toward the organization. The dependent variable was measured by combining semantic differential scales used by Becker-Olsen (2003) and Rodgers (2004). As stated by Bruner (2009), the source of the two scales is unknown. Participants were presented with the sentence “My overall impression of the Fynn company is…” and had to indicate their position on bipolar 1-7 semantic differential scales between two opposing adjectives (see Table 2). Additionally, the items were reversely coded with high values indicating an unfavorable attitude and low values a favorable attitude. A PCA
explains 73.53% of the variance. The resulting measure, which was composed by averaging the scores of the items, was found to be reliable (α = .91), M = 2.73, SD = 0.88.
Table 2
Measurement of variables
Measure Items Factor
loadings Eigen-value % of variance α Perceived inconsistency in CEO behavior 2.69 89.75 .94
Does the CEO Jenna Laine privately…
(1)…follow Fynn’s mission to reduce the emission of CO2 when travelling to Paris?
.96
(2) …show through her behavior that she stands behind Fynn’s mission to reduce the emission of CO2?
.95
(3) …behave in line with Fynn’s effort to reduce its Carbon Footprint when travelling?
.94
Skepticism 2.92 58.32 .82
(1) It is doubtful that Fynn is an environmentally friendly retailer.
.79
(2) It is uncertain that Fynn is concerned to reduce its CO2 emission.
.74
(3) It is sure that Fynn follows high environmentally friendly standards.
-.84
(4) It is unquestionable that Fynn acts in an environmentally responsible way.
-.68
(5) I am skeptical whether this is an environmentally responsible retailer.
.76
Involvement 2.68 89.31 .94
Buying products that are manufactured in an environmentally friendly way…
(1) is unimportant to me / is important to me .95
(2) does matter to me / matters to me .96
(3) is irrelevant to me / is relevant to me .93
Unfavorable attitude toward the organization 3.68 73.53 .91 My overall impression of the Fynn company is…
(1) bad / good .87
(2) unfavorable / favorable .88
(3) disliked / liked .89
(4) unsatisfactory / satisfactory .85
Manipulation Check
Participants were not asked to answer an explicit question regarding the manipulation of the CEO behavior. However, they took part in the perceived inconsistency in CEO
behavior measure. No additional manipulation check was incorporated because it would have been a repetition for participants, which could have led to a higher dropout rate. However, the time participants spent reading the stimulus material was measured and participants were asked questions regarding the content of the online newspaper article which indicated whether or not they had read it. As previously mentioned, 27 participants were removed because they were exposed to the stimulus material for less than 10 seconds and three participants were removed because they answered the majority of the questions wrong. Therefore, it can be reasoned that the manipulation of CEO behavior was successful.
Results
The analysis of the data was carried out with the IBM SPSS software. It was
conducted with and without the control variables gender, age, level of education and country of origin. However, the control variables did not influence the results, and therefore, all the results presented below derive from the analysis without the inclusion of control variables.
Mean Differences of Central Variables
The average score of participants (N = 376) on the 7-point skepticism scale was 3.49 (SD = 1.12), participants indicated a favorable attitude toward the organization on the 7-point scale ranging from 1 = very favorable attitude to 7 = very unfavorable attitude (M = 2.73, SD = 0.88) and they indicated to be involved in the company’s CSR cause on the 7-point scale (M = 5.08, SD = 1.36). In addition, the distributions were normally distributed.
Two one-way analyses of variance (ANOVAs) were conducted to test whether the mean of unfavorable attitude toward the organization and skepticism differed significantly between participants of the consistent CEO behavior, inconsistent CEO behavior and the control group. In doing so, the assumptions of ANOVAs described by Field (2009) were met.3
It was predicted that participants who were exposed to inconsistent CEO behavior showed a higher average score on the unfavorable attitude toward the organization measure than participants who were exposed to consistent CEO behavior (Hypothesis 1a). Overall, the average scores between groups showed small differences. Participants who perceived
inconsistent CEO behavior (M = 2.91, SD = 0.78) indicated the most unfavorable average attitude, followed by participants of the control group (M = 2.69, SD = 0.93). The least average unfavorable attitude was indicated by participants of the consistent CEO behavior group (M = 2.59, SD = 0.89). As visible in Table 3, a one-way ANOVA was carried out to test whether the means differed significantly. The analysis showed a significant difference between the group means, F(2, 373) = 4.76, p = .009. The Bonferroni post-hoc test revealed a significant difference between participants’ means who were exposed to inconsistent and those who were exposed to consistent CEO behavior (Mdifference = 0.33, p = .008). This result was furthermore supported by planned contrasts: The average unfavorable attitude toward the organization mean of participants who were exposed to consistent or inconsistent CEO behavior did not differ compared to those of the control group, t(373) = 0.61, p = .543. However, participants who were exposed to consistent CEO behavior differed on the
unfavorable attitude toward the organization measure compared to those who were exposed to
3 First, the observations were independent and the dependent variables were measured on an interval
level. Second, the assumption of homogeneity of variance was tenable for skepticism, Levene’s F(2, 373) = 1.19, p = .307, as well as for unfavorable attitude toward the organization, Levene’s F(2, 373) = 1.70, p = .183. Finally, the assumption of normal distribution within groups of skepticism was met, with skewness values ranging from 0.28 to 0.41 and kurtosis values ranging from -0.25 to -0.14. The distribution of unfavorable attitude toward the organization within groups revealed relatively high values of skewness and a leptokurtic distribution for the consistent CEO behavior group, skew = 0.59, kurtosis = 0.71, and for the control group, skew = 0.73, kurtosis = 1.78. However, the probability-probability plot (P-P plot) which plots the actual against the expected z-score (Field, 2009) showed that the variable unfavorable attitude toward the organization was roughly normally distributed within groups. Therefore the assumption of normal distribution within groups was met, too.
inconsistent CEO behavior, t(373) = 3.02, p = .003. Thus, participants who were exposed to inconsistent CEO behavior had a significantly more unfavorable attitude toward the
organization than those who were exposed to consistent CEO behavior. Therefore, hypothesis 1a was supported.
Secondly, it was expected that participants in the inconsistent CEO behavior group had a higher level of skepticism toward the CSR program than participants in the consistent behavior group (Hypothesis 2a). Overall, there were relatively small differences of the average scores on the skepticism measure between the different groups. Participants who were exposed to consistent CEO behavior (M = 3.40, SD = 1.11) showed a lower score than participants who were exposed to inconsistent CEO behavior (M = 3.70, SD = 1.05), whereas participants of the control group showed the lowest score (M = 3.38, SD = 1.18). A second one-way ANOVA was carried out to assess whether the skepticism means differed
significantly between the groups. As can be seen in Table 3, a significant difference between group means was found, F(2, 373) = 3.31, p = .037. The Bonferroni post-hoc test indicated non-significant differences between means of participants in the inconsistent and consistent CEO behavior condition (Mdifference = 0.30, p = .093), between participants of the inconsistent CEO behavior condition and control group (Mdifference = 0.32, p = .070), and between
participant of the consistent CEO behavior condition and control group (Mdifference = 0.02, p = 1.000). Planned contrasts were conducted to support the findings of Bonferroni’s test and revealed that participants who were exposed to consistent or inconsistent CEO behavior did not differ significantly on the skepticism measure compared to those in the control group,
t(373) = 1.38, p = .168, but participants who were exposed to consistent CEO behavior
significantly differed compared to those being exposed to inconsistent CEO behavior, t(373) = 2.17, p = .031. Hence, whereas the difference was relatively small, participants who were exposed to inconsistent CEO behavior did show a significantly higher score on the skepticism
measure than participants who were exposed to consistent CEO behavior. Thus, hypothesis 2a was supported.
Table 3
Results of one-way ANOVAs
Sum of Squares df Mean Square F
ANOVA 1: Dependent variable unfavorable attitude toward the organization
Between groups 7.14 2 3.57 4.76* Within groups 279.68 373 0.75
Total 286.82 375 ANOVA 2: Dependent variable skepticism
Between groups 8.16 2 4.08 3.31* Within groups 459.34 373 1.23
Total 467.61 375 Note. * p < .05, ** p < .01; N = 376.
In addition, the PROCESS macro that was developed by Andrew F. Hayes was conducted in SPSS to further test the role of involvement and skepticism in the relationship between exposure to the different conditions and unfavorable attitude toward the organization. The macro, which implements bootstrapping and Monte Carlo confidence intervals, can be used to conduct conditional process analyses such as to test moderated mediation models (Hayes, 2013). PROCESS’s model 14 was carried out to test the moderated mediation effect of involvement and skepticism, respectively, on the relationship between the different conditions and the unfavorable attitude toward the organization. In it, the interaction term of the mediator and moderator variable was included. The results revealed a significant
moderated mediation relationship of skepticism and involvement between the exposure to different groups and the independent variable, SE = 0.01, 95% CI [-0.00, 0.02].
Regression Analyses
In the following step, hypotheses 1b, 2b and 3 had to be tested. However, it first had to be assessed whether participants who were exposed to consistent or inconsistent CEO
behavior differed in perceived inconsistency in CEO behavior. The control group was not exposed to CEO behavior. Therefore, only the consistent and inconsistent CEO behavior groups performed the perceived inconsistency in CEO behavior measure (n = 254).
An independent-means t-test was conducted and the assumptions of the t-test described by Field (2009) were met.4 It revealed that the average score on the perceived inconsistency in CEO behavior scale of participants who were exposed to consistent CEO behavior (M = 2.27, SD = 0.98) differed significantly compared to participants of the
inconsistent CEO behavior condition (M = 5.60, SD = 1.05), t(252) = 26.14, p = .000, 95% CI [-3.58, -3.08]. Hence, participants who were exposed to inconsistent behavior of the CEO in relation to the company’s CSR program actually did perceive it as more inconsistent than participants who were exposed to consistent behavior. Therefore, it can be assumed that perceived inconsistency in CEO behavior was a valid measure of the underlying concept.
The hypotheses were tested with single and multiple regression analyses and the mediation effect additionally with Sobel’s Z. Hypothesis 1b proposed a positive relationship between the independent variable perceived inconsistency in CEO behavior and the
dependent variable unfavorable attitude toward the organization. As displayed in Table 4, regression model 1 significantly predicted 10% of the variation in the dependent variable,
F(1, 252) = 27.03, p = .000. Perceived inconsistency in CEO behavior had a significant,
positive and moderately strong association with unfavorable attitude toward the organization. A person who perceived a higher level of inconsistency in CEO behavior likely had a 0.14
4 First, the scores were independent and the dependent variable was measured on an interval. Second,
the assumption of homogeneity was tenable, Levene’s F(252) = 1.91, p = .17. Third, the P-P plot showed that the data were roughly normally distributed, even though the distribution was leptokurtic with a kurtosis value of -1.366. According to the central limit, the sampling distribution is normal under the condition that the data are roughly normally distributed. Hence, the assumption of normally distributed sampling distribution was met.
score higher unfavorable attitude toward the organization. Thus, the results supported hypothesis 1b.
Hypothesis 2b proposed that skepticism toward the CSR program mediated the relationship between the independent variable perceived inconsistency in CEO behavior and the dependent variable unfavorable attitude toward the organization. A condition for a
mediation relationship is that the independent and dependent variables were associated (Baron & Kenny, 1986). As described in the previous analysis, model 1 in Table 4 showed a
significant relationship between perceived inconsistency in CEO behavior and unfavorable attitude toward the organization. Thus, the first condition was met. A second condition for a mediation relationship is that the separate relationships between the independent variable and the mediator, as well as between the mediator and the dependent variable were significant (Baron & Kenny, 1986). Therefore, these associations had to be tested. First, regression model 2 (see Table 4) in which perceived inconsistency in CEO behavior was included to predict skepticism, was found to be significant and accounted for 5% of the variation in skepticism, F(1, 252) = 14.41, p = .000. The regression model revealed that perceived inconsistency in CEO behavior was a significant, positive and weak predictor of skepticism. Second, regression model 3 (see Table 4) that proposed skepticism as a predictor of
unfavorable attitude toward the organization significantly predicted 35% of the variance in unfavorable attitude toward the organization. Skepticism was found to be a significant, positive and strong predictor of unfavorable attitude toward the organization. Consequently, the second condition for a mediation relationship was met, too. A multiple regression analysis could then be carried out to test the mediation effect of skepticism. Regression model 4 (see Table 4) incorporated skepticism and the independent variable perceived inconsistency in CEO behavior to predict unfavorable attitude toward the organization. The model was found to significantly predict 33% of the variance in the dependent variable. Controlling for
Table 4
Results of regression analyses
F (df, df) ⊿R 2 b b* t- value 95% Confidence Intervals Model 1: Predicting unfavorable attitude
toward the organization
27.03*** (1, 252) .10
Constant 2.22 19.29*** [1.99, 2.44]
Inconsistency in CEO behavior 0.14 .31 5.20*** [0.08, 0.19] Model 2: Predicting skepticism 14.41***
(1, 252) .05
Constant 3.04 20.16*** [2.74, 3.33]
Inconsistency in CEO behavior 0.13 .23 -3.80*** [0.06, 0.20] Model 3: Predicting unfavorable attitude
toward the organization
203.00*** (1, 374) .35
Constant 1.11 9.27*** [0.87, 1.34]
Skepticism 0.47 .59 14.25*** [0.40, 0.53]
Model 4: Predicting unfavorable attitude toward the organization
62.55*** (2, 251) .33
Constant 1.03 6.46*** [0.72, 1.35]
Inconsistency in CEO behavior 0.09 .20 3.68*** [0.04, 0.13]
Skepticism 0.39 .50 9.42*** [0.31, 0.47]
Model 5: Predicting unfavorable attitude toward the organization
71.98*** (3, 372) .36 Constant 2.74 75.79*** [2.66, 2.81] ZSkepticism 0.50 .58 13.85*** [0.43, 0.58] ZInvolvement -0.09 -.10 -2.49* [-0.16, -0.02] Interaction term 0.07 .08 1.98*
Note. * p < .05, *** p < .001; model 1, 2 and 4 N = 254, model 3 and 5 N = 376.
skepticism, perceived inconsistency in CEO behavior was a significant, positive and weak predictor of unfavorable attitude toward the organization. To detect whether skepticism was a mediator in the sample, the unstandardized coefficients (b) of perceived inconsistency in CEO behavior between model 1 and model 4 were compared: Although the coefficient decreased in
magnitude when controlling for the mediator variable in model 4, it remained significant. Therefore, skepticism was found to be a partial mediator between perceived inconsistency in CEO behavior and unfavorable attitude toward the organization in the sample. In addition, Sobel’s Z was significant, Sobel’s Z = 3.80, p = .000; and thus, skepticism toward the CSR program was found to be a mediator in the population. Overall, the results supported hypothesis 2b.
Finally, hypothesis 3 predicted that involvement moderates the relationship between skepticism toward the CSR program and unfavorable attitude toward the organization. To test the hypothesis, a multiple regression analysis was conducted in which standardized versions of skepticism, involvement and their interaction term were included in a regression model that predicted unfavorable attitude toward the organization. The resulting model 5 (see Table 4) was found to be a significant predictor of 36% of the variation in unfavorable attitude toward the organization. A significant, positive and strong main effect of skepticism and a
significant, negative and very weak main effect of involvement on unfavorable attitude toward the organization were found. Furthermore, while controlling for involvement and skepticism, the interaction term was found to be a significant, positive and very weak predictor of unfavorable attitude toward the organization. As illustrated in Figure 2,
involvement strengthened the relationship between skepticism toward the CSR program and unfavorable attitude toward the organization. Thus, the results provided evidence for a partial moderation effect of involvement and therefore, hypothesis 3 was supported.
Overall, the 5 regression models fit the data well: Despite models 1 and 2 having some outliers, they were not deleted because outliers do not influence the model as a whole when they have Cook’s distance values below 1 (Stevens, 1992). Furthermore, the average leverage values in models 3, 4 and 5 as well as the Mahalanobis distances in model 5 indicated that some cases could influence the model. However, the absolute standardized DFBeta values in
Figure 2. Graph of moderation.
all the models were above 1 and therefore did not indicate particular cases that substantially influenced the models’ parameters (Field, 2009). Thus, all cases were included in the analysis. Additionally, the 5 regression models can be generalized to the population: Even though the statistical assumption of conditional normality of errors in the estimation were violated in model 1, 3, 4 and 5, the models were nevertheless generalizable. This is because the assumption of homoscedasticity, which was met in all models is much more important in terms of generalization (Hayes, 2005). Finally, both the variance inflation factor (VIF) and tolerance statistics confirmed that the assumption of no perfect multicollinearity was met and the Durbin-Watson test supported the assumption of independent errors.
Discussion
The study aimed to answer the research question of how perceived inconsistency in CEO behavior in relation to a company’s CSR program affected stakeholders’ unfavorable attitudes toward the organization and what roles skepticism and involvement in the CSR cause played. It was expected that stakeholders who were exposed to inconsistent CEO
1 2 3 4 5 6 7
Low skepticism High skepticism
U n favor ab le atti tu d e tow ar d th e or gan iz ati on Low involvement High involvement
behavior had a more unfavorable attitude toward the organization and a higher level of
skepticism than stakeholders who were exposed to consistent behavior (Hypothesis 1a and 2a) and that perceived inconsistency in CEO behavior was positively related to unfavorable attitude toward the organization (Hypothesis 1b). In addition, it was proposed that the relationship between perceived inconsistency in CEO behavior and unfavorable attitude toward the organization was explained by skepticism (Hypothesis 2a) and that involvement strengthened the association between skepticism and unfavorable attitude toward the
organization (Hypothesis 3). An experimental study was conducted and its results supported the hypotheses. These findings were unique in three respects: First, whereas Elving and Kartal (2012) did not find support for the associations between CEO behavior and the outcomes attitude toward the organization and skepticism, the present study did establish these
relationships empirically. Second, whereas Elving (2013) found evidence for the association between skepticism and attitude toward the organization, the present findings contribute to that knowledge by showing that skepticism not only led to a more unfavorable attitude, but also partially explained why stakeholders who perceived inconsistent CEO behavior had a more unfavorable attitude toward the organization. Third, the reasoning of the Elaboration Likelihood Model (Chaiken, 1980; Petty & Cacioppo, 1981) was for the first time
successfully adapted in a research context investigating CSR communication (see below). In accordance with hypotheses 1a and 1b, inconsistency in CEO behavior was found to be positively related to unfavorable attitude toward the organization, and stakeholders who were exposed to inconsistent CEO behavior in relation to a company’s CSR program
developed a more unfavorable attitude than stakeholders who were exposed to consistent behavior. Even though the differences in the average attitudes were relatively small, they did differ significantly. Whereas the inside-out approach underlined the importance of
2008), the findings demonstrated that actions of organizational members need to be consistent with CSR communication, too. These findings further provided evidence for the relationship between hypocrisy and the attitude toward the organization tested by Wagner et al. (2009) and revealed that in particular the behavior of the CEO plays an important role. In addition, the findings revealed that the reasoning of the attribution theory (see Heider, 1944) can be used to interpret how stakeholders react to inconsistent CEO behavior. Whereas the data of an
experiment conducted by Elving and Kartal (2012) did not support the relationship, the present study showed that under a controlled environment, CEO behavior did affect stakeholders’ attitudes when they were confronted with CSR communication.
Consistent with hypothesis 2a, the findings revealed that stakeholders who were exposed to inconsistent CEO behavior on average developed a higher level of skepticism than stakeholders who were exposed to consistent behavior. Whereas Elving and Kartal (2012) did not find significant differences, the data of the present study revealed that the level of
skepticism did differ. In line with hypothesis 2b, skepticism was not only found to be directly related to both perceived inconsistency in CEO behavior and unfavorable attitude toward the organization, but partially explained the relationship. This means that stakeholders developed a more unfavorable attitude toward the organization when perceiving inconsistent CEO behavior because they felt more skeptical toward the company’s CSR program. However, other factors likely play a role in the relationship between perceived inconsistency in CEO behavior and unfavorable attitude because skepticism was found to only be a partial mediator and not a full mediator. In regard to theory, the findings provided further support for the reasoning of the attribution theory to explain skepticism (e.g., Du et al., 2010; Elving, 2013; or Skarmeas & Leonidou, 2013) and an unfavorable attitude toward the organization (e.g., Du et al., 2010; or Forehand & Grier, 2003) in CSR communication. In addition, the findings suggest that the mechanism of the balance theory (Heider, 1985) could be used to explain
why stakeholders developed skepticism and an unfavorable attitude toward the organization when being exposed to inconsistent CEO behavior.
In line with the ELM (Chaiken, 1980; Petty & Cacioppo, 1981), involvement was found to strengthen the relationship between skepticism and unfavorable attitude toward the organization. However, the moderator was found to be very weak in strength. Previous research has not investigated involvement in relation to CSR communication; therefore, the study set a starting point in focusing on a wider array of possible influencing factors. Specifically, the mechanism described in the ELM could be used to explain attitude differences among stakeholders who indicated to be skeptical toward a CSR program and others who indicated not to be skeptical. According to the ELM, people vary in the way they process information because of their different levels of ability and motivation to process it (Petty & Cacioppo, 1996). It was assumed that stakeholders are able to process the
information, and therefore the way they process it depends on their level of motivation, which was regarded as their level of involvement in the CSR cause in the present study. The results confirmed that highly involved stakeholders who were skeptical developed a stronger
association with unfavorable attitude toward the organization. This was because, according to the ELM, theses stakeholders developed counterarguments about the company’s CSR
program, while others who were not skeptical did not develop counterarguments which resulted in a weaker association with unfavorable attitude. Interestingly, stakeholders who were not involved showed a weaker relationship between skepticism and attitude toward the organization than involved stakeholders. An explanation for this effect can be provided with the reasoning of the ELM, too: Uninvolved or unmotivated stakeholders follow the peripheral route of information processing, which means they process easily observable cues and do not develop counterarguments (Chaiken & Trope, 1999). Following on from that, it can be argued that stakeholders perceive an organization’s engagement in CSR, which is an easily
