How Do Media Companies Gain Legitimacy? An Experimental Study on the (Ir)Relevance of CSR Communication

ABSTRACT Due to structural changes in the media industry, the topic of CSR has gained more and more attention among media companies. Our research question was whether media companies can gain legitimacy through CSR disclosures. There is reason to assume that CSR disclosures both directly increase and indirectly decrease a media company’s legitimacy. On one hand, CSR is regarded as a means of strengthening legitimacy; on the other hand, stakeholders might become skeptical and distrust disclosures about generous deeds. The experimental study detailed here considers both possibilities by using five CSR disclosures of a fictional media company as the stimuli, ranging from low- to high-communicated CSR engagement (single-factor between-groups design, 274 participants). According to the results of the Structural Equation Model (SEM), both assumptions are incorrect: CSR is not the crucial factor in determining whether or not stakeholders perceive a media company as legitimate, but rather its corporate credibility.

The conceptualization of CSR as multirelational ascription allows us to specify the 'extent of CSR,' ranging from mandatory to discretionary. In line with Carroll's pyramid (1991), we speak of mandatory CSR when the involved norms are scarcely negotiable, which is most notably the case with respect to the economic principle to make products and generate profits, the legal principle to obey the law, and the ethical principle to respect human rights. In contrast, we speak of discretionary CSR when the addressed norms are beyond standards and thus not compulsory, which is-for instance-the case with respect to philanthropic donations.
Direct positive effect Tost (2011) expounds that there are two research strands on corporate legitimacy: From an institutional perspective, legitimacy is understood as a generalized perception or assumption that determines whether or not a company survives and prospers in society (e.g., Deephouse & Suchman, 2008;Meyer & Rowan, 1977;Suchman, 1995). From a socio-psychological perspective, legitimacy is understood as an individual's perception or attitude with regards to how a company attempts to reach desirable goals or outcomes (instrumental dimension), how it treats others with dignity and respect (relational dimension), and how it acts consistent with moral and ethical standards (moral dimension) (Tost, 2011;further Tyler, 1997;2006). Although these two perspectives are complementary rather than contradictory, we focus on the socio-psychological perspective because this study analyzed how CSR disclosures are processed and evaluated by individual stakeholders.
By conceptualizing corporate legitimacy as an individual's perception, we can justify the assumption that companies strengthen their legitimacy through 'discretionary CSR.' A company can express, by means of disclosures on discretionary CSR, that it induces to a large extent desirable outcomes for stakeholders (instrumental dimension), that it treats stakeholders with dignity and respect (relational dimension), and that its actions exceed moral and ethical standards by going beyond profitability and legal requirements (moral dimension). This perception leads individuals to an attitude regarding whether the company should be sanctioned positively or negatively in terms of survival and acquiring resources. In this respect, we propose that H1. The greater the extent of a media company's CSR disclosure, the greater its corporate legitimacy is perceived by its stakeholders.

Indirect negative effects
We further assume that the extent of CSR disclosure has indirect negative effects on a company's legitimacy. Among communication scholars in particular, "stakeholder skepticism" is used as a generic term to encompass a heterogeneous group of mediator effects that are suspected to suppress or even counteract the positive effects of CSR communication. In this article, we distinguish two kinds of mediating effect when it comes to stakeholder skepticism. First, we assume that the perceived "content credibility" is a mediator variable that suppresses the positive effect of extensive CSR communication on a media company's legitimacy. Stakeholders are seldom present when CSR engagement actually takes place; instead companies inform about CSR via corporate webpages, as Capriotti (2011) describes in detail. Stakeholders therefore might have the impression that CSR disclosures are "decoupled" from actual CSR actions (Christensen, Morsing, & Thyssen, 2013). Although stakeholders have no reason to doubt a media company's communication about its mandatory CSR in terms of making profits and media products, it is more delicate for them to trust disclosures on discretionary CSR. In this respect, we propose that H2. The extent of CSR disclosures is negatively related to its overall content credibility.
H3. The overall content credibility of CSR disclosures is positively related to the media company's legitimacy.
Second, we assume that stakeholders may have the impression that communication from companies is inherently persuasive. In this context, Holladay and Coombs (2013) coined the term "public relations literacy" (Holladay & Coombs, 2013). When stakeholders perceive CSR as merely strategic and persuasive, they "develop negative feelings when they feel an organization is promoting its CSR efforts too vigorously" (Coombs & Holladay, 2009, p. 99). This assumption is in line with empirical research about "persuasion knowledge" and "psychological reactance" (Quick et al., 2013). The "persuasion knowledge model" states that individuals learn to cope with strategic communication (Friestad & Wright, 1994)-including CSR disclosures (Pomering & Dolnicar, 2009;Pomering & Johnson, 2009).
H4. The extent of CSR disclosure is positively related to the stakeholders' perception of persuasion intent.
In turn, a perceived persuasion intent evokes reactance, which is defined as a "motivational state that is hypothesized to occur when a freedom is eliminated or threatened with elimination" (Brehm & Brehm, 1981, p. 37). Reactance is a psychological reaction to everything that threatens personal freedoms, including persuasive communication (Brehm & Brehm, 1981, pp. 121-150;Koch & Zerback, 2013a, 2013bMeirick & Nisbett, 2011;Moyer-Gusé, Jain, & Chung, 2010;Quick et al., 2013). In line with reactance theory, we expect that H5. The stakeholders' degree of perceived persuasion intent is positively related to the stakeholders' degree of reactance.
According to Brehm and Brehm (1981), "boomerang effects," like the self-promoter's paradox or the catch 22 in the case of CSR communication, "represent the best evidence for reactance theory" (Brehm & Brehm, 1981, p. 38). Individuals that receive persuasive messages preserve their personal freedoms by derogating the communicator's credibility (consistent with Koch & Zerback, 2013a, 2013bMeirick & Nisbett, 2011;Pomering & Johnson, 2009, p. 425). We therefore assume that H6. The stakeholders' degree of reactance is negatively related to the media company's credibility.
According to Newell and Goldsmith (2001), corporate credibility is a specific variant of a communicator's credibility with "trustworthiness" and "expertise" as subdimensions. A company needs to be perceived as trustworthy and skilled in order to be perceived as credible (Newell & Goldsmith, 2001, p. 237; see also ;Hovland & Weiss, 1951). Consistent with credibility research (see, for instance, Koch & Zerback, 2013a, 2013b, we formulate that H7. The media company's credibility has a positive effect on the overall content credibility of its CSR disclosures. Furthermore, we assume that stakeholders who perceive a media company as trustworthy and skilled also perceive its actions as desirable, proper, and appropriate (Suchman, 1995;Tost, 2011). Hence, we propose that H8. The media company's credibility has a positive effect on its legitimacy. Figure 1 provides a summary of all hypotheses. Accordingly, the direct positive impact of a media company's extensive CSR disclosures on its legitimacy is suppressed by a decreased content credibility, and counteracted by the activation of persuasion perception and reactance, which lowers the media company's credibility. A lower corporate credibility-in terms of trustworthiness and expertise-leads to a lower content credibility of the CSR communication as well as to a lower legitimacy (Koch & Zerback, 2013a, p. 11, 2013b, for a similar model).

Participants
Students of social sciences from a Swiss university (274 in total, 169 women, 104 men, 1 unknown; Mage = 22.1 years, SD = 2.44, age range: 19-40 years) were asked to fill in a questionnaire. Although students have a relatively high educational level and may thus be expected to be important stakeholders of media companies that produce journalism (e.g., Elvestad & Blekesane, 2008, pp. 437-439;Jandura & Brosius, 2011, p. 200), we cannot draw descriptive conclusions about different stakeholder groups from this sample (Falkheimer & Heide, 2016, p. 164). However, the purpose of this study is not to analyze real stakeholders of media companies. Our aim is to advance theory by testing general causal assumptions about how CSR communication is processed and evaluated. These processes are independent of specific groups (consistent with Falk & Heckmann, 2009, p. 537).

Research design and stimulus
The research design is a single-factor between-groups design with five different CSR disclosures of a fictional media company named "Tell-Media Inc." All texts of this company were presented in the layout of the corporate's website. Participants were randomly assigned to one of five groups. All of the participants read a self-description of the company (corporate webpage 1/2; "about us"), which indicated that Tell-Media Inc. is a profit-oriented media company: "With its daily and weekly newspapers, magazines, online platforms as well as printing facilities, Tell-Media Inc. is one of the leading media corporations in Switzerland. Tell-Media is an independent media company, which has its roots in the German-and French-speaking parts of Switzerland. We stand for freedom of speech, quality journalism and excellent entertainment. Tell-Media Inc. has been listed on the Swiss Stock Exchange since 2004." Afterwards, each participant read one of five versions of the media company's self-ascribed CSR (corporate webpage 2/2; "our social responsibility"). These stimuli served as the independent variable and increased by degrees in terms of the extent of the company's self-ascribed CSR (see Figure 2).
The control group (CG) read solely the following text about the media company's mandatory CSR, namely to make profits and media products: "As a large Swiss media company we carry social responsibility. We make our contribution to democracy in shaping public opinion through fair, critical, and independent reporting. We also provide plenty of talkingpoints with entertaining stories from every area of life. The Tell-Media Inc. is a profit-orientated company. Only with the highest profitability and economic efficiency can we contribute to the formation of public opinion today and in future." The experimental groups read additional texts on Tell-Media's discretionary CSR, which goes beyond the core functions of media companies. The first experimental group (EG1) read an additional text in which the media company self-imposes internally covenanted obligations: "The ethical basis for our actions is written down in the Tell-Media 'code of conduct'. The ethical guidelines of this 'code of conduct' are binding for all employees-regardless of their ranking in the company or their area of business. With these guidelines, we ensure respect in interpersonal dealings within all of our journalistic and entrepreneurial activities. We have pledged to conduct ourselves ethically and responsibly." The second experimental group (EG2) read-besides the aforementioned texts-an additional text in which the media company self-imposes externally covenanted obligations: "We take our social responsibility seriously with our charity program 'Tell-Media: Culture & Education'. We are committed long-term to promote culture and education in Switzerland. We organize a number of concrete projects, such as dance festivals, theatre performances, exhibitions of young artists, language courses for teenagers and much more. Every six months we provide information on how the 'Tell-Media: Culture & Education' program keeps its promises." Besides the preceding texts, the third experimental group (EG3) read an additional text about the media company's voluntary donations within its production chain: "In addition, we assume social responsibility by supporting charitable organizations and foundations occasionally. They can ask us to waive media advertising costs or printing costs for publications. Our internal committeecomprised of members of the management as well as chief editors-decides on the approval of applications." The fourth experimental group (EG4) read, in addition, a text about the media company's voluntary donations beyond its production chain: "We know about tragic events such as wars, natural disasters and famines through our reports and stories. We want to do more than merely informing about the harsh realities of people from all over the world. As a media company, we know first-hand where help is most needed worldwide. That's why we give project-related funds to recognized charities when people are in distress and need quick and professional help." The extent of CSR disclosure is thereby increased by degrees, from mandatory CSR, with the media company's undeniable core functions in terms of making products and profits (CG), to selfimposed internally (EG1) and externally covenanted (EG2) obligations in terms of a code of conduct and a charity foundation, to voluntary donations inside (EG3) and outside (EG4) the production chain. To increase the external validity, all texts are based on real website texts from Swiss media companies (e.g., NZZ-Mediengruppe, Ringier Inc., Tamedia Inc.).
This experimental design has two advantages. First, the text-building block concept makes a manipulation check of the independent variable unnecessary because it is certain that the extent of CSR disclosure increased (O'Keefe, 2003, p. 257). The manipulation is twofold: in terms of form (text length) and content (from mandatory to discretionary CSR). Second, in most social science experiments, stimulus variables have merely nominal level. Possibilities for statistical data analysis are thus restricted. Although we do not wish to deny that there are reasonable doubts about the permissibility of "treating ordinal variables as if they were continuous" in general (Winship & Mare, 1984, p. 522), this study's text-building block concept allows us to understand the experiment's stimulus variable at ordinal level because it clearly scales the "extent of CSR disclosure" from low to high (Hayes, 2005, pp. 20-23). Moreover, for practical reasons we can grasp it at quasi metric level in order to run regression based analyses. We come back to this point in the findings.

Measures
In line with the paths of our 'CSR dilemma model,' we first measured the constructs "content credibility," "perceived persuasion intent" and "reactance" on one page, and then "corporate legitimacy" and "corporate credibility," with its subdimensions "expertise" and "trustworthiness," on another. All items were judged on 7-point Likert scales (1 = "strongly disagree" and 7 = "strongly agree"). The order of items was randomized to control for possible side effects. We avoided items that were supposed to measure the same construct appearing immediately after each other.

Content credibility of the CSR communication
We measured this construct with one item derived from Zerback (2013a, 2013b) and two items derived from Kohring (2004; see also Rössler, 2011, p. 355): "I consider the statements from the Tell-Media Inc. about their social responsibility as credible" (cre1), "The statements made by the Tell-Media Inc. about taking on social responsibility would withstand a verification" (cre2), and "The website of the Tell-Media Inc. about its social responsibility provides the facts as they are" (cre3) (Cronbach's α = .703; M = 4.05).

Perceived persuasion intent
In order to measure this construct, we adapted three items from Koch and Zerback (2011, 2013a, 2013b: "I have the feeling that the Tell-Media Inc. wants to convince the reader of their leading role with respect to social responsibility" (per1), "The Tell-Media Inc. deliberately tries to convince me that they are socially responsible" (per2), and "The Tell-Media Inc. wants to convince me of their social responsibility" (per3) (Cronbach's α = .772; M = 5.50).

Reactance
Like Zerback (2013a, 2013b), we relied on items developed by Herzberg (2002) in order to measure reactance as a one-dimensional construct. We used the following three items: "It bothers me very much that the Tell-Media Inc. presents itself as a role model" (rea1), "I would like to object to the Tell-Media Inc. with regard to its social responsibility" (rea2), and "I strongly resist the Tell-Media Inc.'s attempts at manipulation" (rea3) (Cronbach's α = .726; M = 3.50).
Corporate credibility (trustworthiness and expertise) We used Newell and Goldsmith's (2001) measurement scale in order to measure corporate credibility as a two-dimensional construct of trustworthiness and expertise. Trustworthiness was measured by four items: "I trust the Tell-Media Inc." (tru1), "The Tell-Media Inc. makes truthful claims" (tru2), "The Tell-Media Inc. is honest" (tru3), and "I do not believe what the Tell-Media Inc. tells me" (tru4) (Cronbach's α = .875; M = 4.01). To measure expertise, we used the following items: "The Tell-Media Inc. is skilled in what they do" (exp1), and "The Tell-Media Inc. has great expertise" (exp2) (Newell & Goldsmith, 2001, p. 237). Additionally, we relied on two items developed by Ingenhoff and Sommer (2010, p. 352) to measure expertise: "The Tell-Media Inc. stands for high product quality" (exp3), and "The Tell-Media Inc. works with comprehensive attention to detail" (exp4) (with regard to all four expertise-items, Cronbach's α = .782; M = 4.27).

Corporate legitimacy
As noted previously, there is no established measurement scale for this construct as a perception yet (Hudson, 2008;Vergne, 2011). To measure corporate legitimacy, we derived the following three items from our definition of legitimacy (Deephouse & Suchman, 2008;Suchman, 1995;Tost, 2011): "The Tell-Media Inc. makes an important contribution to the general public" (leg1), "It would be a great pity if the Tell-Media Inc. were to go bankrupt" (leg2), and "The Tell-Media Inc. should be retained in Switzerland" (leg3) (Cronbach's α = .776; M = 4.31).

Statistical analyses
To test the hypotheses we applied structural equation modeling (SEM). With regard to our experimental data, the main advantages of this statistical method are that it allows us to determine the effect of the stimulus (the independent variable: "extent of CSR disclosure") on the output measure (the dependent variable: "corporate legitimacy") and-at the same time-to take mediating variables ("content credibility," "perceived persuasion intent," "reactance," and "corporate credibility") into account (Bollen & Pearl, 2013, pp. 316-318;Pearl, 2014).
We chose Maximum Likelihood (ML) and Bootstrapping (5,000 samples) as the estimate. These settings allow the testing of SEM even if distributions are not normal (Preacher & Hayes, 2008, p. 880). All calculations were run using Mplus 7.3.

Two full Structural Equation Models: Trustworthiness and expertise
Given that corporate credibility is a two-dimensional construct, we ran two full SEMs (see Figures 3a and 3b as well as Table 1), one with "trustworthiness" and the other with "expertise" as a mediator variable (consistent with Newell & Goldsmith, 2001, p. 245). The upside of this approach, where each subdimension is calculated independently, is that it makes the models more parsimonious.
Extent of CSR disclosure has no direct effect on legitimacy Both SEMs reveal that the media company's extent of CSR disclosure has no direct effect on the perceived corporate legitimacy-which leads to a rejection of H1. The standardized path coefficients are small (β = .09 and β = .04, respectively) and nonsignificant (p = .14 and p = .84, respectively). We did not expect this result and thus ran an additional analysis of variance (ANOVA, Bootstrapping with 2,000 samples) with the extent of CSR disclosure as the independent and the perceived legitimacy as the dependent variable. This additional test also counteracts possible criticism around the statistical permissibility of treating the independent variable as if it were metric.   (Hayes, 2005, p. 373). Post hoc analyses were performed using the Scheffé test-again, no differences were found (p = .96). This means that a media company that limits its social responsibilities to the making of profits and media products (CG) is perceived to be as legitimate as a media company that additionally signs a code of conduct, runs a charitable foundation, waives advertisement costs for nonprofit organizations, and donates money to people in need (EG4).
Content credibility has no suppressing effect H2 assumed a negative effect from the extent of CSR communication on the content credibility. Although there is a significant relation in the trustworthiness-model (p = .02), there is none in the expertise-model (p = .28). However, regardless of the significance, the path coefficients are positive in both models (β = .12 and β = .06, respectively), meaning that-contrary to expectation-the extent of CSR communication is not negatively related to its content credibility, which leads to the rejection of H2. Furthermore, H3 must also be rejected: The content credibility of the CSR disclosure has no effect on legitimacy (β = -.01, p = .98 and β = .01, p = .98, respectively). The rejections of H2 and H3 imply that the content credibility of the CSR communication has no mediating effect. Hence, content credibility is not a suppressor variable-it cannot explain the preceding result that the extent of CSR communication has no effect on legitimacy.

Persuasion intent and reactance have no boomerang effects
There is neither a significant effect of the extent of CSR disclosure on persuasion intent (H4: β = .14, p = .06 in both models) nor on reactance (H5: β = .13, p = .11 and β = .10, p = .23, respectively). Therefore, H4 and H5 need to be rejected as well. Perceived persuasion intent and reactance are not counteracting variables-they also cannot explain the preceding result that the extent of CSR disclosure has no effect on legitimacy. H6 can be confirmed because there is a strong effect from reactance on corporate credibility in both models: Reactance decreases the media company's perceived trustworthiness (β = -.72, p = .01) as well as expertise (β = -.63, p = .01). However, it is not the extent of CSR communication which triggers perceived persuasion intent and reactance. Hence, we do not know what triggers reactance.
The importance of corporate credibility Corporate credibility has a strong effect on the overall content credibility of the CSR disclosure (in both models β = .83, p = .01). Moreover, there is also a strong effect of trustworthiness on corporate legitimacy (β = .76, p = .01) as well as of expertise on corporate legitimacy (β = .78, p = .01). Hence, H7 and H8 are confirmed.
In summary, the extent of CSR disclosure has no effect on corporate legitimacy. This absence of a direct effect can neither be explained by the mediator variable "content credibility of CSR disclosure" nor the mediator variables "perceived persuasion intent" and reactance." The results indicate no suppressing effects. However, the full models reveal that the "corporate's credibility," with its two dimensions of trustworthiness and expertise, explains a corporate is perceived as legitimate.

One restricted Structural Equation Models
To get a clearer picture of the effects of trustworthiness and expertise on the perceived corporate legitimacy, we ran a restricted model without the mediator variables "content credibility," "perceived persuasion intent," and "reactance" (see Figure 4 and Table 1).
The global fit indices of this restricted model show that the model fits the empirical data (χ 2 = 92.95, df = 51, p = .01; RMSEA = .06 with 90% C.I. between .04 and .07, p = .31; CFI = .97; TLI = .96; SRMR = .04). It confirms the previous result that the extent of CSR communication does not effect corporate legitimacy (β = .08, p = .14). In contrast, the two subdimensions of corporate credibility-trustworthiness and expertise-have direct effects on legitimacy (β = .45, p = .01 and β = .36, p = .03 respectively) and explain about 61% of its variance. The results of this restricted SEM underline the unimportance of extensive CSR communication and the importance of corporate credibility-with its two subdimensions trustworthiness and expertise-to explain a media company's legitimacy as perceived by its stakeholders.

Discussion
The surprising result of our experimental study and answer to our RQ is that the extent of CSR communication has neither a direct positive effect nor indirect negative effects on a media company's legitimacy. Instead, the perceived corporate credibility-with its two subdimensions trustworthiness and expertise-is the crucial factor, which explains best whether a media company is perceived as legitimate by its stakeholders.
Do we then still have a CSR dilemma? At least with regard to the media industry, our findings challenge two widespread beliefs among CSR scholars and practitioners: Namely the "business case" argument that discretionary CSR is in the interest of a firm because it strengthens its legitimacy (e.g., Carroll & Shabana, 2010, pp. 99-100;Dowling & Pfeffer, 1975, p. 133;Lee, 2008), as well as the "catch 22" argument that extensive CSR communication triggers stakeholder skepticism, which in  turn decreases a firm's legitimacy (Morsing et al., 2008;further Brønn, 2011;Morsing & Schultz, 2006;Pomering & Dolnicar, 2009;Pomering & Johnson, 2009;Pomering et al., 2013). We assumed that media companies are rewarded for discretionary CSR in terms of a higher legitimacy. However, according to our results, a media company that limits its CSR disclosure to mandatory obligations, in terms of providing good media products and making a profit, is perceived to be as legitimate as a media company that additionally communicates about its discretionary CSR activities. We further assumed that stakeholder skepticism suppresses or even counteracts the positive effects of extensive CSR communication on legitimacy.
However, according to our results, a media company's CSR communication that goes beyond its core obligations in terms of making profits and products has neither a lower content credibility nor increases the perceived persuasion intent, which in turn might activate reactance. We could not identify mediator variables that have significant suppressing or counteracting effects. Adapting the quote from Bartlett (2011, p. 81), we can sum up the first main finding as follows: Media companies are neither damned if they communicate about discretionary CSR nor if they don't. The analyses of the experimental data also reveal that the perceptions of a media company's credibility, with the two subdimensions trustworthiness and expertise, have strong and significant effects on the perceived legitimacy. The second main finding is that a media company's credibility is the main driver for its legitimacy.
Most hypotheses are not supported by the data. Accordingly, our 'CSR dilemma model' can be considered empirically inadequate. What are the possible explanations for the null results? We assume that news media companies, such as the fictional "Tell-Media Inc.," are already perceived as legitimate, regardless of their CSR efforts. Although this assumption is reinforced by the results of a more recent experimental study we have conducted, where we used a company from the raw material sector as the stimulus (Bachmann & Ingenhoff, 2016), we must address several limitations.

Limitations
We ran three SEM models, which all showed acceptable overall fit indices. Our results therefore have robust statistical evidence. However, the validity of this experimental study is limited for several reasons. One limitation relates to the stimulus material: We relied on a fictional media company, the "Tell-Media Inc." The participants probably perceived the materials as artificial, which lowers the external validity of our experimental study. There is reason to assume that psychological reactance and legitimacy could not be evoked, because participants were too indifferent to nonexistent organizations (Brehm & Brehm, 1981, pp. 59-61). In this case, the data would be inadequate to test the 'CSR dilemma model'. In other words: "Absence of evidence is not evidence of absence" (Altman & Bland, 1995, p. 485).
Other limitations are related to the measurements. The participants were aware of the survey situation, which means that they were probably more aware of persuasive attempts than in everyday situations. The level of perceived persuasion intent was high in the control group as well as in the four experimental groups (M Total = 5.5 on a 7-point Likert scale). Reactance is also difficult to measure via questionnaire items (Brehm & Brehm, 1981, p. 37;Koch & Zerback, 2013a, 2013b, p. 1004Quick et al., 2013, p. 171). Moreover, we operationalized the construct of corporate legitimacy based on the relevant literature. Although the items were reliable, we must take into consideration that the results rely on a nonvalidated scale.
Finally, our experimental study might not be fully applicable to other industries, although we believe that its results are particularly valuable with regards to the media industry. Considering that media companies face difficult market conditions and react with cost-cutting programs these days (Kaye & Quinn, 2010;Levy & Nielson, 2010;Russ-Mohl, 2011), we found that it might be reasonable for themin order to be perceived as legitimate by key stakeholders-to stick to their core obligations in terms of strengthening democratic processes through economically viable quality journalism instead of scattering their resources on discretionary CSR. Stakeholders may appreciate companies' extensive CSR in general, but according to our results, media companies are better off in terms of legitimacy when doing their core business in a credible manner. However, a replication of this study with a representative stakeholder sample would be necessary to draw this conclusion. The primary purpose of this article is to advance theory in strategic communication by analyzing CSR message effects.

Outlook
The proposed "CSR dilemma model" is about how corporate messages are processed and analyzed by individual stakeholders. This experimental study thus contributes to the body of strategic communication knowledge by laying a foundation for replication and extension in this area of strategic communication. We hope that our experimental study will stimulate further research where researchers might concentrate on companies from other industries. Furthermore, future research might also focus on optimizing the measurement constructs of perceived corporate legitimacy and validate the scale of perceived persuasion intent and reactance with regard to companies. A replication study could refer to Ham, Nelson, and Das (2015), who made an overview study on the measurement of "Persuasion Knowledge," which has been published since our experimental study was conducted.
We emphasize that we do not argue that media companies should not engage in and communicate about CSR beyond their core obligations of making profits and media products. From a normative point of view, we raise no objection to discretionary CSR. However, from an instrumental point of view, the findings of our experimental study may indicate that decision makers-at least from the private media sector-should be aware of the unimportance of extensive CSR disclosures and the importance of corporate credibility, with subdimensions trustworthiness and expertise, to gain corporate legitimacy.