UvA-DARE (Digital Narrative persuasion by corporate CSR messages: The impact of narrative richness on attitudes and behavioral intentions via character identification, transportation, and message credibility

The current study examines the persuasiveness of narrative richness in messages about acts of corporate social responsibility (CSR). Specifically, we apply theory about narrative persuasion to the domain of corporate communication. Focusing on Coca-Cola ’ s clean water project, a cross-national experiment ( n = 659) was conducted in which the narrative richness and the source cue of a CSR message were manipulated, and the effects on (a) message processing (specifically via character identification, transportation, and credibility), (b) attitude towards the company, and subsequent (c) behavioral intention were measured. Considering the global nature of Coca-Cola , the experiment was simultaneously fielded in the United States and the Netherlands to verify the generalizability of our findings. Robust results in both countries suggest that narratively rich CSR messages can indirectly improve the attitude towards the company and thereby encourage behavioral intentions via character identification and transportation, irrespective of the source cue (company vs . news media). Importantly, message credibility was not compromised by the narrative richness, while being the strongest predictor of company attitude.


Introduction
Because humans are storytelling creatures (Gottschall, 2012), narrative forms of communication are very impactful and encouraged to play a larger role in conceptualizing and creating public relations initiatives (Coombs & Holladay, 2018). It has indeed long been demonstrated that narratives offer corporate communicators unique opportunities to engage audiences, build relationships, and shape corporate reputations (Heath, 1992;Kent, 2015). As corporate messages are typically designed with the intent to persuade audiences in support of a particular issue, value, or policy stance (Heath, 1992;Kent, 2015), PR professionals might benefit from what LaMarre (2017) termed strategic storytelling to increase public understanding of complex issues and build support for particular viewpoints.
The "use of storytelling for organizational branding and reputation" (Janssen et al., 2012, p. 32) suggests that organizations are harnessing the power of morally appealing narratives to manage their reputations. Likewise, companies are employing narratives to engage social media publics (Zhao et al., 2018), communicate crises (Clementson, 2020), and shape policy opinion (McBeth et al., 2014). Growing on this body of research, empirical studies are beginning to measure the effects that such narratives might have on attitude towards a corporation-which is conceivably an important determinant of behavioral intentions (Jung & Seock, 2016;Lii & Lee, 2012).
The current study examines whether the persuasiveness of corporate social responsibility (CSR) messages is conditional upon the narrative richness of such content. CSR has become an important element of strategic management with the aim of improving corporate reputation (Wood & Logsdon, 2019), but is often understood by different stakeholders in divergent ways (Sethi, 1975). While often presented as voluntary and philanthropic engagement, one could alternatively understand CSR as pro-social corporate rhetoric or discourse aimed at legitimizing and (pre) serving corporate power structures (Raman, 2007). From this perspective, it is important that the CSR story is expressed well: The current manuscript explores whether narratively rich content is especially suitable when communicating about CSR activities.
Herein, we specifically focus on Coca-Cola's "Clean Water Project". To verify the robustness of our findings, we conduct the study in two

Theoretical framework for narrative communication effects
Definitions of what a narrative entails include multiple elements, such as plot, scene, highly identifiable characters, temporal order, and sequence of "connected events and characters that convey messages about the topic being addressed" (Kreuter et al., 2007, p. 222). Practically speaking, most CSR messages or news items about CSR initiatives carry a plot, identifiable characters, descriptions of the scene and a temporal order. Still, these might be emphasized to a different degree. In other words, some narrative content about an organization's CSR will be relatively more compelling, offering more narrative details (e.g., about the main characters or the scenes). By comparison, others are more thematic, less detailed, or relatively more abstract (yet, the narrative is still present in the background). Thus, CSR messages may differ in "narrative richness". Accordingly, the current study does not examine the difference between narratives and non-narratives, but rather focuses on differences in narrative richness: We distinguish "rich narratives" in which the narrative elements are more compelling and play a relatively more central role from "muted narratives", which focus more on a factual account of the same issue (while still having some narrative elements present in the content).
The role of narrative persuasion is well-documented in a variety of corporate communication contexts, such as crafting different types of corporate messages (e.g., Kent, 2015;Yang et al., 2010), heightening employee engagement (Gill, 2015) and communicating as part of a crisis response strategy (Heath, 2004). The persuasive impact of narratives is explained by Dahlstrom (2012), who refers to narrative persuasion as a "covert mode of persuasion, where information is generally accepted first and only scrutinized later" (p. 502). Considering that narratives tend to reduce reactance (Moyer-Gusé, 2008), decrease message scrutiny (Dahlstrom, 2012;Green & Brock, 2000) and increase affective disposition towards the message (Raney, 2004), they potentially offer a compelling way to strengthen relationships and positively impact attitudes and behavior. However, narrative strategies are not without controversy (Dahlstrom & Ho, 2012). Mixed results suggest that, in some instances, non-narrative messaging is more effective (Clementson, 2020;Zhao et al., 2018).

Communicating corporate social responsibility messages with narratives: A synergy?
By communicating the values and morals of a company, corporate social responsibility (CSR) initiatives strengthen corporate identity (Bravo et al., 2012;Hildebrand et al., 2011), and hence become an important element of corporate public relations strategies. Still, with dozens of possibilities in today's media environment (Birth et al., 2008), companies struggle to find the most effective way to communicate their CSR initiatives (Morsing et al., 2008).
CSR arguably offers a suitable context for the effective use of narratively rich content. Theoretically, CSR narratives are similar in form and structure to other strategic narratives, including elements such as plot, characters, and temporal order of events (Kreuter et al., 2007). Because strategic narratives utilize classic literary elements (e.g., protagonist, antagonist, plot, moral), it is likely that audiences process narratively composed CSR messages in ways similar to traditional stories (Slater & Rouner, 2002); thereby strengthening the potential influence on attitudes and behavioral intentions more effectively than standard means of communicating CSR messages (i.e., non-narratives or muted narratives). We draw upon media psychology literature to understand the process by which strategic storytelling potentially improves attitudes toward and behavioral intentions that support the organization.
CSR messages often convey an organization's ethics, morality, and corporate character. As such, rich CSR narratives in which moral arguments play a central role in resonating with one's beliefs are likely to be effective in persuading audiences and changing behaviors (Haidt, 2012). Put differently, given the important role that morality plays in CSR messages (Frederiksen, 2010) as well as in persuasive story-telling, there is a certain synergy between these two types of messaging. As such, rich narratives offer a relatively compelling way to situate the organization as the protagonist with strong corporate (moral) character compared to muted narratives, which describe the CSR initiative in more abstract or factual terms. Accordingly, we predict that narratively rich CSR messages will have a stronger beneficial effect on attitude towards the company and subsequent behavioral intentions than a muted CSR narrative about the same topic.
More concretely, we predict that higher narrative richness leads to (a) more positive attitude, which subsequently causes (b) stronger behavioral intentions to support the company, such as increased purchase intention or interest in joining the corporation as an employee (Fombrun et al., 2000;Lii & Lee, 2012). These hypothesized relationships align with the Theory of Planned Behavior (Ajzen, 1991), which predicts that attitude (together with perceived behavioral control and subjective norms) precedes behavioral intentions (Fishbein & Ajzen, 1977). Thus, as people have a more positive attitude towards a corporation (Schwaiger, 2004), they are more likely to support it through their behavior (Jung & Seock, 2016;Lii & Lee, 2012). Hence, we expect the following: H1a-b. CSR messages expressed in a rich narrative positively affect (a) attitude and subsequently (b) positively affect the behavioral intentions towards the company compared to CSR messages expressed in a muted narrative.

Underlying mechanisms of the narrative richness effect
To understand how rich narrative CSR messages may have an effect compared to muted narrative CSR messages, it is important to illuminate the "black box" between cause (i.e., narrative richness) and effect (attitude, behavioral intentions) by focusing on specific underlying psychological processes. Specifically, we will analyze the role of three mediating variables: (a) character identification, (b) transportation and (c) message credibility.
A first path through which narrative richness may accomplish an effect is through the mediating process of identification with story characters (Cohen, 2001;Green & Brock, 2000). Identification has been defined as "a process that consists of increasing loss of self-awareness and its temporary replacement with heightened emotional and cognitive connections with a character" (Cohen, 2001, p. 251). Specifically, emotionally and cognitively identifying with a story character leads to more narrative-consistent attitudes (Raney, 2004;Wirth, 2006).
Extending this to public relations contexts, LaMarre and Cohen (2017), examined how audiences make sense of organizational-level characters in public relations narratives. They found that audiences simplify narratively structured messages (i.e., CSR stories) about large organizations to the basic literary elements that humans were raised to understand (hero, villain, plot, moral dilemma, see Gottschall, 2012). Applying this to the Coca-Cola example, we expect that audiences process the narratively rich CSR message in a manner similar to traditional story-telling. With a more developed storyline and relatively more details about the main characters, rich narratives will arguably encourage more character identification than muted narratives.
Because narratively structured messages activate existing story schemas more strongly, we expect this type of identification to occur more readily in a rich narrative message when compared to a muted narrative CSR message. As such, rich narratives that centrally cast the corporation as the protagonist in a message about how its CSR acts benefit the environment or local communities will likely allow for stronger character identification (i.e., identifying with the protagonist) than more factually focused muted narrative messages: H2. CSR messages expressed in a rich narrative cause stronger character identification during message processing when compared to CSR messages expressed in a muted narrative.
High levels of identification include seeing oneself as aligned with the character (usually the story protagonist), such that they would likely make the same choices or act in similar ways with said protagonist (LaMarre & Grill, 2019). In the context of CSR messages, we expect that as individuals experience stronger levels of identification, they will subsequently hold a more positive attitude toward the company involved in the story. Ostensibly, this occurs as a process wherein one views the CSR narrative and finds themselves cognitively and emotionally aligning with the organization's goals and objectives. This high level of character identification cues more positive appraisals of the character and their actions (i.e., positive attitude toward Coca-Cola), which will lead to a stronger intention to support the brand (LaMarre & Grill, 2019).
In essence, such narratives-that create identification with protagonists (i.e., Coca-Cola) resolving moral issues (i.e., providing clean water to communities)-activate positive appraisals and reduce scrutiny of the organization (Van Laer et al., 2014). Accordingly, stronger identification with an organization that engages in pro-social CSR acts should result in more favorable attitudinal and behavioral outcomes in the narratively rich conditions (Green & Brock, 2000;LaMarre & Cohen, 2017):

H3a-b. CSR messages expressed in a rich narrative cause stronger character identification and therefore have a positive indirect effect on (a) attitude and (b) behavioral intentions toward the company compared to CSR messages expressed in a muted narrative.
A second underlying process that explains the effect of muted versus rich narrative CSR messages goes through the mediating variable transportation (Slater & Rouner, 2002). Transportation is a construct originating from literary (Gerrig, 1993) and social-psychological research (Green & Brock, 2000), which was extended to media psychology research as a means of understanding how individuals process media content. Generally defined as being "primarily engaged in the storyline, rather than in one's immediate environment" (Moyer-Gusé, 2008, p. 409), narrative transportation includes, "emotional reactions, mental imagery, and a loss of access to real-world information" (Green & Brock 2000, p. 703).
The term transportation was originally developed by Gerrig (1993) to encapsulate the key difference audiences experienced in response to narrative and non-narrative forms of communication. Gerrig (1993), along with Green and Brock (2000), contends that the prominence of narrative elements induces a unique state in which individuals' thoughts, emotions, and visualizations take them into the narrative world. Extending this to CSR messages, narratively rich messages should lead audiences to become more mentally and emotionally engaged than those exposed to similar CSR messages in a muted narrative. Hence, we predict: H4. CSR messages expressed in a rich narrative cause stronger transportation when compared to CSR messages expressed in a muted narrative.
Transportation influences attitude and behavior in a congruent direction with narratively structured messages (Green, 2004). The more transportation that a positive CSR message achieves, the more likely it becomes that audiences will hold favorable opinions of the organization (Barbour et al., 2016). High levels of transportation correlate strongly with more narrative enjoyment (Raney, 2004), which in turn evokes greater feelings of empathy (Grabe et al., 2017) and reduces message scrutiny. Put simply, when people are transported into the story they are more emotionally engaged and less motivated to critically scrutinize every element of a message (Young, 2008), conceivably because it would lessen their enjoyment. As such, it is unlikely that highly transported message recipients develop counterarguments that are incongruent with the narrative message (Boukes et al., 2015;Slater & Rouner, 2002, p. 180) because they are relatively more focused on following the storyline.
Overall, increased felt enjoyment coupled with reduced reactance (e. g., counter-arguing) to embedded messages (Moyer-Gusé, 2008), offers potential explanations for why transportive narratives can produce relatively stronger effects than non-narrative messages (Lee & Jahng, 2020) or muted narratives. Applying this to Coca-Cola's clean water program, narratively rich messages about communities benefitting from the water projects should produce higher levels of transportation (Green & Brock, 2000;Slater, Rouner, & Long, 2006) when compared to muted narratives. Accordingly, more positive evaluations of the Coca-Cola corporation and stronger behavioral intentions to support the corporation are expected when the CSR message is presented in a rich narrative compared to a muted narrative. This indirect effect is hypothesized as:

H5a-b. CSR messages expressed in a rich narrative cause stronger transportation and therefore have a positive indirect effect on (a) attitude and (b) behavioral intentions toward the company compared to CSR messages in a muted narrative.
It should be noted, however, that these potential narrative effects might be lessened if audiences perceive the narratively rich CSR messages as having lower credibility than the muted narratives (Kopfman et al., 1998). Clementson (2020), for instance, found non-narrative information to have a higher trustworthiness in the public's view than narrative information: It is possible that this occurs as a function of reduced perceived credibility of rich narratives. Framing theory, for example, suggests that when episodically presenting a single exemplar (as happens in a narrative), people may perceive the message as an isolated case that is not generalizable (Iyengar, 1991). However, evidence opposite to this pattern has also been found (see, e.g., Boukes, 2021): Narrative style in court testimony, for example, does not decrease complainant credibility (Westera et al., 2015). Still, entertainment psychology literature suggests that people might discount narratively structured information as "just entertainment" or "meaningless stories" (Nabi et al., 2007). Extending this to a public relations perspective, it is possible that audiences perceive rich narrative messages as less convincing when compared to more factually focused muted narratives (LaMarre, 2017). Accordingly, the public might not believe that the CSR initiative is a broader reflection of a company's societal engagement when told in the form of a rich narrative compared to a muted narrative. Although the literature is somewhat mixed, we test the following hypothesis: H6. CSR messages expressed in a rich narrative are perceived as less credible than CSR messages expressed in a muted narrative.
When people perceive a CSR message as non-credible, one would expect this to harm the attitude towards the organization (Green & Donahue, 2011). Following this line of reasoning, if rich narratives are perceived as less credible (than muted narratives), then the potential positive impact of narratively rich CSR messages on attitude and/or behavioral intention might be mitigated by a message discounting process. Based upon the message discounting processes found in entertainment psychology literature (Boukes et al., 2015;Nabi et al., 2007), we concede that lower credibility of narratively rich CSR messages could dampen its potential benefits. More specifically, running opposite of the two positive indirect effects outlined above (i. e., character identification and transportation), we expect the following negative indirect effect:

H7a-b. CSR messages expressed in a rich narrative cause lower message credibility and therefore have a negative indirect effect on (a) attitude and (b) behavioral intentions toward the company compared to CSR messages expressed in a muted narrative.
In summary, Fig. 1 recapitulates our conceptual model and visualizes the hypotheses about the effects of muted versus rich CSR narratives that are investigated in this study. This overarching model will be empirically tested with a structural equation modelling approach.

Narratives in corporate communication: moderation by source cue
One interesting artifact of how people evaluate CSR initiatives relates to perceived sincerity (Yoon et al., 2006). CSR narratives fall into the master plot category of "transformation" (Kent, 2015). If in a text, an organization helps a community overcome social ills (e.g., economic hardship, lack of clean water), then the narrative should be perceived as a transformative story (Tobias, 1993). As one of the master plots used in stories, audiences already have schema in mind to easily process transformative stories (Kent, 2015). While such schema should result in positive appraisals of the organization involved in the transformation, this might backfire if the message is seen as self-serving.
Put simply, it is quite plausible that if organizations tell their own positive message about how "good" they are, they might be viewed as less sincere and be perceived as having self-serving motives (Cole & Greer, 2013). In such case, attitudes might actually be harmed (Yoon et al., 2006), or at least be less likely to improve, because audiences perceive them as untrustworthy or hypocritical (Shim & Yang, 2016). In contrast, having an independent source, such as a news outlet, convey the transformative story could be viewed as more sincere (Pornpitakpan, 2004). As the audience will expect more objectivity from a journalistic source than from a company source, a transformative CSR narrative could be more influential when produced by a non-involved objective actor (e.g., journalist). Accordingly, we test the assumption that when a journalistic source shares the narratively rich transformative CSR narrative, it is likely to produce relatively stronger positive effects on message processing, attitudes and behavior than when the organization touts its own story: H8a-d. CSR messages expressed in a rich narrative cause higher levels of (a) transportation, (b) character identification, (c) attitude and (d) behavioral intentions compared to CSR messages in a muted narrative, but these effects are stronger for messages published by journalistic news media than by the company itself.
As there is no clear theory to predict how message credibility of a muted versus narratively rich CSR message will be affected differently when produced by a journalistic outlet or by the company itself, we explore the following research question: RQ 1 : Does message source (journalistic vs. company) significantly affect the relationship between narrative richness (muted vs. rich) and message credibility?

Method
An online experiment was conducted using a sample of individuals obtained from survey research company Research Now SSI. To increase generalizability of findings, data were collected in two countries: Coca-Cola's home country (U.S.; n = 321) and a European country where Coca-Cola products are ubiquitous (the Netherlands; n = 338). No differences between the two countries were expected regarding hypothesized effects, so both samples were pooled in the data analysis. Sample quality was assured with a quality check question (i.e., two plus two is [4]) and two attention checks (asking whether people remembered the topic or company of the stimuli): People answering these questions wrongly were dropped from the dataset prior to data analysis. The sample was diverse and close-to-representative on several characteristics: age (M = 45.82, SD = 14.14, Min. = 19, Max. = 67), gender (52.4 % female), and education (44.8 % higher educated).

Stimuli
A randomized 2 × 2 between-subjects factorial design was employed for this study (see Fig. 2). As such, no pre-test measures were needed that Note: For reasons of clarity, we did not visualize the specified co-variances between exogenous variables and the specified covariances between error terms of mediating variables, although they were included in the model. Dashed lines are insignificant effects, light-grey lines are the effects of control variables. Hypotheses (H) are identified together with their predicted effect direction (positive or negative effect).  (2021) 102107 5 could prime participants with the purpose of the study. 1 In all conditions, participants were exposed to a text that positively described Coca-Cola's water sustainability initiative in Thailand. The text was manipulated on two aspects: narrative richness (muted vs. rich) and source cue (corporate website vs. news outlet). The American and Dutch samples received the same text (all of similar length); the original English texts were literally translated into the Dutch for the latter group. Online Appendix A shows the English-language stimuli. 2

Narrative richness: muted vs. rich
To assure external validity, stimuli were inspired by an original Coca-Cola video that explained the corporation's water sustainability program with the exemplar of a Thai farmer who personalized the benefits of the program that he and many others experienced. 3 Closely following the content of this video, a narratively rich written version was created using the exact same story elements, including Na Mao (the farmer in the video), the exact same setting (rural Thai village of Limthong), same social issue (i.e., lack of water described by Na Mao as, "We had two straight years where it never rained. We couldn't grow any crops. We were forced to borrow money to allow us to eat and have a place to live."), and the same events as described in the video (i.e., "Then a woman came from Coca-Cola to tell us they wanted to start a water retention project" … "But the water retention project changed everything. It fulfilled all our expectations. It's like the water is overflowing for us."). The real world Coca-Cola Clean Water Project story was converted into a text version for two reasons: (1) to ensure that we properly isolated the muted-versus-rich narrative manipulation and eliminated any confounding factors known to trouble video based stimuli in prior media psychology experiments (Green & Brock, 2000;Landreville & LaMarre, 2013); and (2), so that we could test the muted versus rich narrative's effect in multiple nations using the same (translated) stimuli.
For the muted narrative conditions, this text was carefully rewritten to make the narrative elements less prominent and less central in the message, while keeping the information equal. Instead, it provides a comparably more factual and abstract summary of Coca-Cola's water sustainability program. Concretely, it does not write about particular characters but about how "the Love Water program has already benefited over 1 million Thai people." Moreover, where the rich narrative explained the personal struggles of Na Mao (i.e., the social issue), the muted narrative version generally explained "in times of drought, the production of rice is severely limited, resulting in extreme economic hardships." Finally, the plot from the rich narrative was replaced in the muted narrative condition with a factual account of how the program works. For instance, the muted narrative stated that "Through this program, the local people have increased their incomes and improved their quality of life. Children have returned to school and the local farmers report that the living conditions are improving daily" (see Online Appendix A for exact stimuli materials and these quotes).
Hence, while we controlled for tone (written in equally positive terms) and key facts (information presented as similar as possible) to maintain internal validity, the narrative richness of the CSR message clearly differed. A manipulation check asking what kind of text people had been exposed to (i.e., written true story or written summary of facts) confirmed the success of the manipulation, χ 2 (2) = 231.79, p < .001.

Message source: corporate website vs. news outlet
The source of the message was manipulated by varying the top part of the visuals in which the texts were presented: Either the banner of the Coca-Cola website was used or that of a well-known and trusted journalistic outlet. Thus, for the corporate communication conditions it obviously was a text from the Coca-Cola website; whereas for the news outlet conditions the text was clearly published on a journalistic news website. We used the Dutch public broadcaster NOS.nl for the sample from the Netherlands, because it is the most trusted newscaster in the country (Costera Meijer & Groot Kormelink, 2020). Due to the politically polarized context of the United States, we avoided inherent partisan interpretations and selected the well-known and highly trusted BBC news for the American sample. Moreover, the header of the message was minimally edited to be written in either a first-person style (corporate website; e.g., "We have launched these efforts") versus a third-person style (news outlet; e.g., "The company states that they have launched these efforts"). A manipulation check confirmed that people easily recognized the source of the text accurately, χ 2 (2) = 50.86, p < .001.

Measurements
Independent variables were operationalized as dichotomous dummy variables to make these suitable for independent samples t-tests and structural equation modelling: Regarding the narrative richness variable, the muted (0) versus rich (1) conditions were distinguished; regarding the source, we distinguished the corporate website (0) versus the news outlet (1) conditions. Below, we describe the operationalization of the mediating variables and dependent variables. Principal axis factoring with oblique rotation confirmed that all multi-item scales loaded only on one latent component (max. one eigenvalue > 1).

Character identification
How strongly participants identified with the character and aligned with its emotions and thoughts was measured with five disagree-agree Likert-scale items (adopted from Cohen, 2001) instruction was given about who or what should be regarded as the "main character" to make the survey items suit all experimental conditions. In the Discussion, we further reflect on this decision and its potential limitations; however, the high Cronbach's α reflects an overall high internal consistency of the measurement, which suggests that participants had one consistent character in mind (although it is not clear who/what).

Transportation
Narrative engagement with the text (i.e., transportation) was measured with four survey items tapped on seven-point Likert disagreeagree scales (adopted from Green & Brock, 2000): (a) I did not notice the events occurring around me during the story; (b) I was fully focused on the story; (c) The story affected me emotionally; and (d) I found myself thinking of ways the story could have turned out differently. The scale was reliable (Cronbach's α = .84, M = 3.33, SD = 1.35).

Message credibility
In contrast to the transportation variable, the variable "message credibility" measured on a more cognitive dimension how people assessed the CSR message to which they had been exposed. How much people trusted the credibility of the CSR message was tapped with three disagree-agree Likert-scale items (adopted from Lock & Seele, 2017) 1 Randomization checks showed that conditions did not differ significantly on age (p = .818), gender (p = .098), nor education (p = .612); thus, randomization was successful.

Attitude towards company
In a similar format as previous studies (Bae & Cameron, 2006;Lii & Lee, 2012)

Behavioral intentions
To thoroughly explore behavioral intentions, we extended beyond intentions to purchase products to also include other known key behaviors, such as engaging in positive word-of-mouth (Lii & Lee, 2012) and expressing interest in working for Coca-Cola (Fombrun et al., 2000). As such, a three-item scale was created that tapped the intentions to act in ways that would support the company (Cronbach's α = .88, M = 2.35, SD = 1.81): (a) share this Coca-Cola story with others; (b) seek employment with Coca-Cola company; and (c) buy more Coca-Cola products.

Attitude and behavior (H 1 )
Hypothesis 1 a predicted positive effects of CSR messages expressed in a rich narrative compared to CSR messages expressed in a muted narrative on attitude toward the company. However, no main effects of narrative richness were yielded on the attitude. Although the attitude was slightly more positive in the rich narrative conditions (M = 4.23, SD = 1.42) than in the muted narrative conditions (M = 4.15, SD = 1.45), an independent samples t-test showed this effect was not significant, t (577) = − 0.75, p = .456. Thus, no support was found for Hypothesis 1 a . A significant difference between both conditions was not found regarding behavioral intentions either, t(577) = 0.44, p = .663.

Character identification (H 2 )
The second hypothesis predicted that CSR messages in a rich narrative increase the identification with the main character. Results indicate that the rich narrative conditions indeed increased character identification compared to the muted narrative conditions, t(577) = 4.04, p < .001. Participants more strongly identified with the character in the rich narrative conditions (M = 4.18, SD = 1.10) than in the muted narrative conditions (M = 3.76, SD = 1.21). Thus, evidence is found in support of Hypothesis 2.

Transportation (H 4 )
Hypothesis 4 predicted that CSR messages in the rich narrative would lead to higher levels of transportation compared to the muted narrative. The analysis finds that people were indeed more narratively engaged with the message in the rich narrative conditions compared to the muted narrative conditions, t(577) = 2.33, p = .020. People who were exposed to the rich narrative text experienced more transportation into the message of Coca-Cola's water sustainability program (M = 3.46, SD = 1.37) than those in the muted narrative conditions (M = 3.20, SD = 1.32). This finding supports Hypothesis 4.

Credibility (H 6 )
The sixth hypothesis predicted that the narratively rich CSR message would be perceived as less credible than the muted narrative. However, no significant differences were yielded between conditions regarding message credibility, t(577) = 0.74, p = .461. Thus, credibility of the message was not harmed nor enhanced by having a richer narrative in the message about Coca-Cola's water sustainability program compared to conditions with muted narrative elements. The importance of this finding is addressed in the Discussion section.

Indirect effects on attitude and behavioral intentions
Structural equation modelling (SEM) was used to test the overall model, and specifically the intertwined set of indirect effects that narrative forms of communication may have on attitude and behavioral intentions. To control for external factors, analyses used country, age, and gender as exogenous control variables. Narrative richness (dichotomous variable: muted vs. rich) was an exogenous independent factor. 5 Transportation, character identification, and message credibility were simultaneously used as mediating variables.
SEM analysis is especially useful because it tests for the indirect effect of each mediating variable while controlling for the (indirect) effect of the other mediating variables-which is important because mediating variables often correlate (in this case: Pearson's r ranged from 59, to .69). As such, this analysis provides a more robust understanding of the role that each variable plays during media processing. Attitude and behavioral intentions were the dependent variables, of which the first was allowed to predict the latter. Fig. 1  The SEM analysis replicates the effects that were found earlier, thereby re-confirming H 2 and H 4 , while again rejecting H 6 : Positive effects of the rich narrative compared to the muted narrative were yielded on transportation (b* = .11, p = .004) and character identification (b* = .19, p < .001), whereas message credibility did not differ significantly between conditions in this model either (b* = − .02, p = .646).
Character identification had a positive effect on attitude (b* = .14, p < .001). As people identified more with the character in the text, they assessed Coca-Cola more positively. In the same model, attitude was positively affected by how much people were transported into the message (b* = .11, p = .005). Thus, as people were more transported into the CSR message about Coca-Cola's water sustainability program, they developed a more positive perception of the corporation. Although not affected by the narrative richness itself, message credibility had the strongest effect on attitude (b* = .52, p < .001). Put differently, CSR message credibility was not negatively affected by the message's narrative richness; however, the people that perceived the CSR message as more credible (irrespective of the experimental condition) developed a more positive attitude towards Coca-Cola.
In line with Hypothesis 1 b , behavioral intentions were affected by the attitude towards the company (b* = .20, p < .001): As people had a more positive perception of Coca-Cola, they were more likely to act in ways that support the company. Moreover, the behavioral intentions were positively affected by transportation (b* = .29, p < .001); message credibility was also a significant and positive predictor of behavioral 4 We alternatively measured audience's perceptions of their relationship with Coca-Cola with the following five disagree-agree Likert-scale items adopted from the relationship scales developed by Hon and Grunig (1999) The effects of narrative richness on and mediated via this "relationship quality" variable are highly similar (i.e., effect direction, significance) to the direct and indirect effects on/of attitude. 5 The (interaction) effect of source cue was not included in the model to keep the SEM as parsimonious as possible. Notably, source cue didn't influence any of the mediating variables nor outcome variables (see analysis below). 7 intentions (b* = .27, p < .001). Behavioral intentions were not directly impacted by character identification (p = .728).

Indirect effects via character identification (H 3 )
We used a bootstrapping approach to test the significance of the indirect effects, because bootstrapped estimates have been found to be more precise than the traditional Sobel-test (Hayes & Scharkow, 2013). Bootstrapping is a statistical technique that resamples the dataset (in this case 10.000 times) to create many simulated samples and, thereby, constructs confidence intervals without necessarily meeting the normality assumptions. In our case, the indirect effects of the muted narrative (0) versus the rich narrative (1) CSR message were analyzed using 10.000 bootstrapped estimations of the indirect effects' bias-corrected confidence intervals. We

Indirect effects via transportation (H 5 )
We

Indirect effects via credibility (H 7 )
Because narrative richness did not significantly influence message credibility (i.e., the muted narrative was perceived as equally credible as the rich narrative), no significant indirect effect could exist of narrative richness via message credibility on attitude (p = .664) nor behavioral intentions (p = .653): Thus, Hypothesis 7 ab is rejected. This, however, does not mean that message credibility is irrelevant. In contrast, the direct effects of credibility on attitude and behavioral intentions signal its indispensable importance in cultivating a positive attitude towards the company. Notably, however, credibility does not significantly vary between the two different narrative richness conditions (muted vs. rich). As such, it is possible for rich narratives to achieve as much credibility as muted narratives; an important finding discussed in more detail below.

Moderated effects by source cue (H 8 )
No main effects were yielded of message source (i.e., corporate website vs. journalistic website) on any of the mediating and dependent variables: character identification (p = .269), transportation (p = .926), message credibility (p = .704), attitude (p = .321), nor on behavioral intentions (p = .559). However, we were mainly interested in how source cue moderates the effects of narrative richness.
Opposite to our expectations (see Hypothesis 8), the source (Coca-Cola or news) did not moderate the effects of narrative richness on any of the predicted mediator variables or dependent variables. The interaction between muted-versus-rich narrative CSR message and source cues was insignificant for transportation (p = .545), character identification (p = .658), message credibility (p = .786), attitude (p = .652), and behavioral intentions (p = .189). Put simply, the persuasive effects were of equal strength, irrespective of who the source of the message was. Hypothesis 8 a-d was thus not supported.

Country effect
Because Coca-Cola is a global organization, additional analyses were conducted to verify the generalizability of effects across nations. No twoway or three-way interaction effects were found when we included the dichotomous country variable (USA vs. Netherlands) in the models. This demonstrates the generalizability of our findings. Thus, the yielded effects were not conditional upon geographic context.

Discussion
Media psychology studies have demonstrated the persuasive power of narrative compared to non-narrative messages (Moyer-Gusé, 2008;Slater & Rouner, 2002). Still, the influence of narrative richness in the context of public relations and corporate communications might be more nuanced and requires more empirical investigation (Heath, 2004). Clementson (2020), for instance, found that narrative information decreased the trustworthiness of information in spokesperson interviews during crisis times. The current study focuses on the effectiveness of narrative richness in CSR messages; arguably a domain of public relations that allows a strong synergy with narrative richness due to the shared emphasis on morality. After all, morality is an important pillar both under narrative construction (with a prominent role for protagonists, and the fight for good over evil, see LaMarre & Cohen, 2017) as well as under acts of corporate social responsibility (Frederiksen, 2010). Moreover, a narratively engaging message might be more appreciated under regular circumstances, whereas this could seem inappropriate in the context of urgent crisis situations (Clementson, 2020).
Our study focused on Coca-Cola, a well-known and iconic company with high profile reputation issues. Still, this is a long standing, global company. As such, participants might have already had a relatively stable attitude toward Coca-Cola, which potentially explains why the single CSR message in our experiment did not have strong main effects on attitude and behavior. In the absence of these direct effects, however, we found evidence that CSR messages expressed in a rich narrative increased character identification and transportation compared to a muted narrative. This suggests that relatively more compelling storytelling offers public relations professionals an effective way to share CSR initiatives with key audiences. When compared to a muted CSR narrative, the narratively rich CSR message could thereby indirectly improve attitudes toward Coca-Cola. In line with the Theory of Planned Behavior (Ajzen, 1991;Fishbein & Ajzen, 1977), a more positive attitude increased positive behavioral intentions towards the company. Studying the effects of narratively rich CSR messages on attitudes and behaviors towards less renowned corporations than Coca-Cola might shed light on the generalizability of our findings and under which conditions CSR narratives are most persuasive. Important, and in contrast to what was hypothesized, the credibility of the CSR message was not compromised by narrative richness. As such, our study provides empirical evidence that narratively rich CSR messages are perceived equally credible as muted narratives about CSR activities. Its strong positive effect on attitude towards the Coca-Cola corporation and on behavioral intentions demonstrates that credibility is an indispensable condition to enhance views about a company with CSR messages. Message credibility avoids negative perceptions of an organization being self-interested (Bae & Cameron, 2006) and/or insincere (Yoon et al., 2006), which are known to harm a company's reputation. Because it has often been assumed that stories with more prominent narrative elements will be discounted as entertainment, and therefore rendered less credible (see Nabi et al., 2007), our research provides a key finding that highlights the potential of narratively rich content to effectively communicate about CSR activities without decreasing message credibility.
Interestingly, message source did not elicit any significant effects. In fact, we rejected all hypotheses that messages published by the company would be less impactful than those published by a respected journalistic outlet. In contradiction to what is known about pure commercial messages (e.g., Cole & Greer, 2013), we conclude that the effect of narrative richness in corporate CSR messages does not depend on whether it appears in journalistic news media or in publications by the company itself. This could be due to the nature of CSR initiatives, which focus on a company's positive contributions to society and thereby might reduce skepticism among a lay audience. Hence, we provide public relations professionals with empirical evidence that touting their own CSR initiatives is potentially as effective as having journalists tell the story for them. Knowing that message credibility is both important to fostering positive attitudes and not hindered by narratively rich storytelling, this provides public relations professionals some assurance that social media or corporate websites are good resources to share their CSR successes. It would be naïve to assume that rich narratives always lead to positive effects; we were interested in better understanding which underlying mechanisms, if any, would create positive attitudinal or behavioral outcomes. By exploring the mediating paths of various psychological processing variables (e.g., identification, transportation, credibility), we shed light on the black-box between cause (narrative richness of CSR message) and effect. Specifically, transportation (Green & Brock, 2000) and character identification (Cohen, 2001) are key pathways to achieving the desired narrative effects. However, whereas transportation positively influenced both the attitude and behavioral intentions, character identification only improved the attitude but not the behavioral intentions: The reason could be both theoretical and methodological. Potentially, character identification occupies less cognitive resources and less mindful attention than message transportation. Accordingly, it could influence the attitude, while not affecting the more robust and stable behavioral intentions (e.g., Nabi & Krcmar, 2004). Alternatively, our measurement might perhaps not have precisely nor fully captured the concept of "character identification". First, the survey items measuring character identification did not specify who the "main character" was. As such, it is possible that participants answered the questions with different "characters" in mind (e.g., Coca-Cola, the farmer). Ideally, future research breaks down this concept by distinguishing between (a) identifying with the CSR beneficiaries and (b) identifying with the company of which CSR acts are presented. If future work can unpack whether identifying with the organization (as protagonist) as being a good citizen is key or if identifying with the beneficiaries of the program matters more, then the narratives can be designed for maximum effectiveness. Second, by relying on an existing scale (Cohen, 2001), we might not have entirely captured the full meaning of what character identification (in theory) would be, and therefore not relate strongly enough to the behavioral intentions measurement. Future research could extend the existing scale and include new survey items that more directly tap into important elements of character identification, such as empathy or feeling concerned about the main character.
It is important to mention that we only investigated the effects of exposure to positive CSR messages. Although positive messages will dominate in the communications of companies about their own CSR acts, journalists may instead present negative views, such as CSR being a legitimating discourse (Raman, 2007) or how local communities are often skeptical about CSR initiatives (Chaklader & Gautam, 2013). When reporting such a negative perspective through a rich narrative, the effects could potentially be more harmful than when it is expressed in a muted narrative, because the same narrative persuasion processes will be activated. Negative narratives might actually be even more powerful than positive narratives because people are more responsive to negative information in general (Soroka, 2014), and towards negative news about companies in particular (Jonkman et al., 2020).
In conclusion, we recommend that when developing narrative CSR messages special attention be paid to creating a narratively rich text with highly identifiable characters and transporting (engaging) story lines while maintaining high levels of credibility. This combination should produce positive effects on key audiences, such as increasingly positive attitudes towards the company and eventually behavioral intentions to support this company. In all, narrative strategies seem to theoretically fit well with the CSR initiatives (due to their shared emphasis on morality), offering an engaging way to convey corporate character. Our work confirms that narratively engaging messages offer corporate communicators an effective way to positively influence audiences with acts of CSR.

Funding
This work was supported by the Netherlands Organisation for Scientific Research (NWO) with a VENI grant under project number: 016. Veni.195.123.

Declaration of Competing Interest
The Authors declare that there is no conflict of interest.