From zero to villain: Applying narrative analysis in research on organizational reputation

What narratives accompany the emergence of a negative reputation? I combine research on public organizations’ reputation with narrative analysis. Narratives offer multiple benefits to reputational research, playing an important role in human cognition and comprising social constructions of both organizations and other actors. Organizations profit from insights of narrative analysis concerning their reputation management. I apply the Narrative Policy Framework to the Swiss Child and Adult Protection Agencies (CAPA). A quantitative analysis of 667 narratives in mass media shows that the emergence of the CAPA’s negative reputation was accompanied by villain depictions early on, with narratives assuming a different quality after an implementation scandal. Also the CAPA’s target groups underwent marked changes in their depictions, most notably with problem causers being cast as heroes in several narratives, thus questioning the CAPA’s basic legitimacy as problem-solving organization. These findings provide the CAPA with concrete starting points to amend their reputation.

A quantitative analysis of 667 narratives in mass media shows that the emergence of the CAPA's negative reputation was accompanied by villain depictions early on, with narratives assuming a different quality after an implementation scandal. Also the CAPA's target groups underwent marked changes in their depictions, most notably with problem causers being cast as heroes in several narratives, thus questioning the CAPA's basic legitimacy as problemsolving organization. These findings provide the CAPA with concrete starting points to amend their reputation.

K E Y W O R D S
child and adult protection policy, narrative policy framework, organizational reputation, quantitative media analysis, theories of the policy process

| REPUTATION THROUGH THE LENS OF NARRATIVES
What narratives accompany the emergence of a negative reputation? The article at hand aims to answer this research question.
Public organizations' reputation matters. On the one hand, an organization's reputation is crucial in determining whether its employees are motivated and whether qualified personnel can be acquired (Alon-Barkat & Gilad, 2016;Busuioc & Lodge, 2016). On the other hand, reputation influences the organization's interactions with its environment. Whether cooperation with other organizations is feasible (Busuioc, 2016) and whether an organization is able to act autonomously within its field of jurisdiction (Hinterleitner & Sager, 2018) both depend on the organization's reputation. Thus, public organizations have a vital interest to know about and protect their reputation.
Research on organizational reputation in public administration has boomed in the last decade (Maor, 2020). Among others, scholars have conceived measurement tools for reputation (Lee & van Ryzin, 2019), they have focused on effects of reputation (Maor & Sulitzeanu-Kenan, 2013) and on attempts to influence a reputation (Busuioc & Rimkutė, 2020).
A topic that reputational scholars have overlooked so far are narratives. I argue that the study of narratives offers three important benefits to research on organizational reputation. First, narratives play an important role in human cognition. Thus, their analysis provides important clues concerning the formation of an organization's reputation with their audiences. Second, narratives comprise social constructions. Therefore, they may serve as an informational source when determining the state of an organization's reputation. Third, narratives often do not only focus on one actor but also focus on several, detailing relationships between them. Accordingly, reputational scholars may learn about an organization's environment and about relational dynamics by studying narratives. The analysis exemplifies this last benefit by zooming in on organization's target groups, an audience whose relationship with an organization so far has not been systematically conceptualized in reputational research. Thus, this article provides multiple contributions to the burgeoning literature on organizational reputation (Bustos, 2021) and answers recent calls to advance "the narrative turn" in public administration research (Barker, 2021;Nowell & Albrecht, 2019).
Knowledge retrieved from the analysis of narratives is also relevant for public organizations. It may serve as a basis both to evaluate the current reputation and how it emerged and to devise strategies to reach a desired reputation (Waeraas & Sataøen, 2014). For example, being aware of a particularly harmful narrative may enable an organization to create a useful counter-narrative.
I conduct a first application of narrative analysis in reputational research by investigating the narratives that accompanied the emergence of a negative reputation. Relying on the Narrative Policy Framework (NPF) as a prominent variant of narrative analysis , I focus on the case of the Swiss Child and Adult Protection Agencies (CAPA). The CAPA are interesting both from a theoretical and from a practitioner's perspective. From a theoretical point of view, they allow for the observation of an organizational reputation from their inception, further exhibiting a considerable reputational development within only 3 years. Thus, researchers may examine a large within-case variation while holding the context constant. For practitioners, the CAPA constitute a remarkable case as they experienced what may be considered a reputational worst-case scenario (Luoma-aho, 2007;Malay & Fairholm, 2020). Instead of exploiting the opportunity of a new name to achieve a positive reputation, the CAPA quickly became "Switzerland's most hated authority" 1 (Dorer, 2017). Therefore, an in-depth examination provides valuable learnings for other organizations.
Lastly, the CAPA are also an interesting case for researchers committed to the NPF's advancement. Examining the CAPA provides the opportunity of exploring narratives in an organizational context and allows for an application of the framework in a non-U.S. area-a call frequently made by NPF scholars (Shanahan et al., 2017).
The article is structured as follows. First, I substantiate the research gap and show how the NPF may be combined with research on reputation by formulating propositions for the empirical analysis. Next, I introduce the research design by presenting the case of the CAPA and by providing information on data and methods. 667 narratives are collected by means of a quantitative content analysis of mass media between 2013 and 2016. Following up, I present the results of the analysis. Findings indicate that narrative depictions of the CAPA as villain accumulated over time and assumed a different quality after an implementation scandal that spurred the CAPA's negative reputation. Also, the CAPA's target groups underwent marked changes in their depictions, most notably with problem causers being cast as heroes in several narratives, thus questioning the CAPA's basic legitimacy as problem-solving organization. Such narratives questioned the CAPA's legitimacy. In the remainder of the article, I reflect on the findings' broader implications and highlight opportunities for future research.

| REPUTATION AND NARRATIVES
This section substantiates the research gap and provides propositions for a combination of reputational research and the NPF.

| Establishing the gap, or how narratives fit in with research on reputation
Research on the reputation of public organizations commonly bases on the definition by Carpenter and Krause (2012, 26), who conceive of organizational reputation as "a set of beliefs about an organization's capacities, intentions, history, and mission that are embedded in a network of multiple audiences." A crucial feature of this definition is the conception of a reputation not being formed by the organization itself, but by the audiences observing it; reputation is the conglomerate of individual beliefs. Individuals in turn are not neutral observers, but interpret information both cognitively and affectively (Capelos et al., 2016).
A systematic literature review on organizational reputation shows that research on the affective side of reputation is scarce (Bustos, 2021, 25). This is a considerable research gap since preliminary evidence shows the centrality of affective information processing for reputation formation. Several studies on the effects of organizational communication demonstrate through survey experiments that symbols and brands may outweigh substantial information on organizations (Alon-Barkat, 2020; Teodoro & An, 2018). However, symbols and brands are not the only elements capturing individuals' attention and influencing their cognitive processes. Linguistic elements such as frames (Chong & Druckman, 2007) and narratives (Berinsky & Kinder, 2006) have also been shown to crucially influence human cognition (Ertas, 2015;Jones & Song, 2014).
Narratives are heuristic devices that convey a variety of information in condensed form (Jones & McBeth, 2010). An important difference between brands and narratives is the fact that the latter are not the intellectual property of organizations. Everybody can create narratives to convince others of their opinion, and while common decency may demand that one should not disseminate lies, recent experiences with "alternative facts" and "post-truth" have taught us that this standard is not always adhered to (Jones & McBeth, 2020). Thus, narratives are both interesting and dangerous for public organizations, and their analysis helps filling an important gap in reputational research.
As mentioned in the introduction, a further benefit of narratives is that they contain social constructions. Individuals, groups, and organizations are not portrayed neutrally in narratives but evaluated normatively, for example, when framed as the narrative's villain. Hence, examining the social constructions of a public organization in narratives may contribute to the assessment of an organization's reputation. While recently developed survey instruments such as the ones by Lee and van Ryzin (2019) or Overman et al. (2020) undisputedly provide more nuanced information on reputation, the study of narratives bears two advantages. First, to arrive at an assessment of an organization's overall reputation-the conglomerate mentioned above-via surveys, extensive and representative samples are necessary. This may not always be within the scope of researchers interested in reputation. Second, the ability of surveys to establish insights about the past is limited through the restricted memory of human beings and the ensuing danger of distortions. As narratives are often retained in written form such as newspaper articles or social media posts, studying such sources allows for estimations of past reputational dynamics. This is a feature sought after by public organizations (Bustos, 2021, 26).
Finally, narratives offer the opportunity to study relational dynamics in reputation. In his systematic literature review, Bustos(2021, 25) highlights the need for more research on organizations' audiences. I agree and propose to go one step further: It is not only audiences themselves that matter but also depictions of the relationships between an organization and its audiences. Such depictions provide an important informational basis for individuals forming their reputational beliefs. In particular, the relationship of an organization with its target groups is relevant in this respect, as it is the target groups who directly experience an organization's actions (Mintrom & Williams, 2015). Target groups have not yet received systematic attention in research on reputation. Either they are treated as one of many audiences and stakeholder groups (e.g., Boon et al., 2020) or they are directly mentioned with regard to the policy field under examination (e.g., pharmaceutical firms as regulatees of the U.S. Food and Drug Administration in Carpenter, 2010). I propose a systematic conceptualization of the relationship between an organization and its target groups (cf. following section). To conduct an empirical examination of this conceptualization, I take advantage of narratives' relational characteristics.

| Combining the Narrative Policy Framework with research on reputation
In the following, I present the Narrative Policy Framework (NPF) and formulate propositions that aim at leveraging the substantial advantages of the NPF for research on reputation.
The NPF was introduced by Jones and McBeth in 2010 and has since found remarkable resonance in the public policy research community (e.g., Dunlop et al., 2021;Gjerstad and Fløttum, 2021;Kuhlmann and Blum, 2021;Schlaufer et al., 2021;Tosun and Schaub, 2021;Vogeler et al., 2021). The most frequent elements are the narrative characters; a narrative may contain villains who cause a problem, victims who suffer from a problem, and heroes who solve a problem. Furthermore, narratives often include a plot that establishes a temporal and causal order, and a moral of the story, that is, a normatively charged ending (Shanahan et al., 2017, 176). These recognizable and repeating elements render different narratives comparable with each other. The person who constructs a narrative fills the elements with policy-specific actors and information that fits their strategic goal (Shanahan et al., 2017, 177). For example, if somebody wants to convince others to close a country's borders for migrants (moral of the story), they might construct a narrative that depicts migrants as villains, the domestic population as victims, and border guards as heroes.
An NPF analysis captures all institutions, groups, and individuals that appear in a specific policy field's narratives. Researchers decide themselves whom they want to investigate and which categories to use on them, which renders the NPF's applicational possibilities diverse and flexible. For the present purpose, I focus on a public organization and on its target groups.
The following propositions are formulated considering the phenomenon of interest as defined in the research question, that is, the emergence of an organization's negative reputation. Furthermore, the propositions base on the premise that the depiction of an actor as hero is equivalent with a positive social construction, while the depiction as villain implies a negative social construction. This premise has been established by previous NPF research (Crow & Wolton, 2020, 663).
A first proposition reads as follows: Proposition 1 A public organization's emergence of a negative reputation is associated with an increase in narratives depicting the organization as a villain.
Postulating a connection of a negative reputation with villain depictions appears evident, and indeed, Terry (1997) illustrates how former US president Ronald Reagan systematically used villain depictions to legitimize the dismantling of bureaucracy. Nonetheless, as the interaction between reputation and narratives constitutes a new research area, an empirical examination is advisable.
Coming to the propositions regarding target groups, as mentioned above, a systematic conceptualization is needed. The causal model of public policies (Sager et al., 2017, 47-49) is useful in this regard. Public organizations are the implementers of public policies. To solve a societal problem, policies and their implementers may target two different groups: On the one hand, they can provoke behavior changes with the problem causers, for example, with firms that are monopolizing the market. On the other hand, they may alleviate suffering by assisting the persons affected by problems, for example, by granting unemployment benefits to individuals without jobs. Thus, public organizations' target groups may be divided into problem causers and problem affected. 2 Concerning social constructions, narrative depictions of problem causers as villains and problem affected as victims may be assumed as standard, since a policy has undergone a democratic process of societal legitimation before being adopted. However, an emerging negative reputation of the organization implementing the public policy may cause shifts in these depictions. For problem causers, two scenarios are conceivable. With the organization turning villain, they might be left to themselves and continue to cause the problem. Therefore, no shift in their narrative depiction would become visible. Another possibility is that the villainous organization turns on the problem causers in a nonappropriate way, thus provoking a shift in the depiction of problem causers to victims. Due to this theoretical ambiguity, I formulate two diverging propositions:

Proposition 2a
A public organization's emergence of a negative reputation is not associated with a change in the depiction of problem-causing target groups as villains. Proposition 2b A public organization's emergence of a negative reputation is associated with an increase in narratives depicting the problem-causing target groups as victims. Finally, I propose the following for the target groups affected by the problem: Proposition 3 A public organization's emergence of a negative reputation is not associated with a change in the depiction of problem-affected target groups as victims.
This last proposition bases on the consideration that as the organization supposedly helping the affected has turned into a villain, the affected are left alone with the continuing problem.

| RESEARCH DESIGN
The following section is divided into two parts. First, I introduce the case of the Swiss Child and Adult Protection Agencies (CAPA) and describe the phenomenon of interest, that is, the emergence of their negative reputation. Second, I present the methods and data applied in the analysis.

| The CAPA and their negative reputation
The Swiss Child and Adult Protection Agencies (CAPA) are regional public organizations that are charged with the implementation of the Child and Adult Protection Policy (CAPP). The policy's goal is to help persons in need that are unable to assume their rights and control their assets (Germann et al., 2017), such as orphans, abused children, or elderly persons affected by dementia-in terms of the established target group categorization, they are the problem affected. Since such persons often find themselves in relationships of intensive and asymmetrical dependency with others, many CAPA interventions do not only involve the problem affected, but also involve other persons in their immediate environment, such as parents or close relatives. While those relatives are not always responsible for the suffering of the problem affected, in many cases they contribute to the problem constellation. For example, if a person with dementia lives in a family with intense asset conflicts, somebody might try to take advantage of the affected person's state and arrange an alteration of the will. In such a case, the CAPA may get active to withdraw the problem causer's guardianship rights and install a neutral representation for the affected person. Thus, the CAPA have both problem-causing and problemaffected target groups. 3 While throughout the twentieth century, local lay authorities had been in charge of implementing the CAPP, the Swiss national parliament judged in 2008 that this task was too complex for laypersons and that forthwith, experts such as lawyers, psychologists, and social workers should take over (Hildbrand et al., 2020). Organized in more than 140 CAPA situated all over Switzerland, these experts took up their work in January 2013.
Within as short a period as 3 years, the CAPA's negative reputation emerged. By 2016, civil society groups had formed, vouching to fight the CAPA's "despotism" (Gisler, 2015), politicians had launched popular initiatives to abandon the CAPA, and facades of CAPA offices had been sprayed with graffito that read "Murderers!". Several CAPA employees had even to be put under temporary police protection. While in the last few years, the situation has calmed down again, the CAPA are still famous for being "Switzerland's most hated authority" (Dorer, 2017). A popular initiative that wanted to curb the CAPA's power gathered 96,000 signatures (Steiger, 2019), and until today, CAPA employees report difficulties in approaching their target groups.
A major implementation scandal played a decisive role in this development. On New Year's Day 2015 in Flaach, a small village near Zurich, a woman killed her 5-year-old son and 2-year-old daughter. The local CAPA had placed the children in a children's home in November 2014 after the police had arrested the parents on suspicion of fraud. Although the mother was released after a few days, the CAPA decided that the children should remain in the children's home until further notice. The mother could take the son and daughter to her place for the holiday season. Still, she was to return them to the children's home on January 2 and await the investigation's conclusion. Instead of doing so, she committed the double infanticide and tried to kill herself. Police prevented the suicide and arrested the mother (Kappeler, 2015).
With the children's grandparents and later also the mother herself accusing the CAPA of having caused the infanticide through their insensitive and inhumane behavior, the situation escalated rapidly into a veritable firestorm that spread all over the German-speaking part of Switzerland and affected not only the local, but all CAPA. While administrative examinations later would discharge the CAPA involved in the double infanticide of being responsible for the children's death (Vögeli, 2016), the recent developments delineated above indicate that this acquittal had little to no effect on the CAPA's reputation.

| Data and methods
The analysis was conducted using the Child and Adult Protection Policy (CAPP) dataset . The CAPP dataset comprises all policy narratives surrounding the CAPP that were published in the six major newspaper outlets of both the German-and French-speaking region of Switzerland (Blick,Neue Zürcher Zeitung,Tages Anzeiger,24 Heures,Le Matin,Le Temps) between 2013 and 2016. The newspaper selection is to ensure representativeness in terms of both circulation and political positioning, featuring the biggest tabloids (Blick and Le Matin), as well as central-leftist (Tages Anzeiger, Le Temps) and conservative quality newspapers (Neue Zürcher Zeitung, 24 Heures). The focus on print rather than digital media arises from the intention to capture narratives that had the potential to achieve broad societal impact. During the period of investigation, print media reached a total population coverage of approximately 50% (Vogler & Hauser, 2015, 48).
A keyword search in the news database Factiva delivered a total of 848 newspaper articles. These articles were analyzed by two researchers, applying an established codebook from NPF research (cf. Appendix 1). Paragraphs served as unit of analysis.
One researcher coded the newspapers from the German-speaking region, the other the ones from the French-speaking region. Coding units were defined thematically (Krippendorff, 2003, 107-9) and captured narratives uttered by specific speakers (the narratives' "authors") in a text. The thematic categorization implied that one article could yield several narratives and one narrative could span more than one sentence. An intercoder reliability test on a subset of 90 articles displayed solid results for the villain code (Krippendorff's α = 0.907). By contrast, the codes of hero and victim displayed low values (α = 0.565 and α = 0.663, respectively), prompting the researchers to redo the analysis of these codes together in the entire dataset. When conflicts of opinion occurred, the researchers discussed the specific observations until a consensus emerged.
The CAPP dataset comprises 1,088 narratives. The analysis at hand bases on a subset of 667 narratives, which corresponds to all narratives published in German-speaking newspapers 4 between January 1, 2013 (when the CAPA took up their work), and February 5, 2016 (when the reports on the press conference about the administrative examination ceded). Figure 1 provides an overview of newspaper articles mentioning the CAPA over time. To make the dataset amenable to the analysis of temporal developments, I computed a LOESS model (span = 0.15, cf. Figure 1 blue line) and subdivided the period of investigation into seven phases by tracing the curves' minimum turning points (cf. Figure 1 red lines, Table 1). Compared with segmenting the investigated timespan according to fixed slots of, for example, three-month length, the LOESS model provides the advantage of capturing trends in the data. Thus, each phase represents a "wave" of newspaper articles that feature the CAPA. The phases' start and end dates, and the number of articles and narratives within these phases are listed in Table 1.
To make the narratives amenable to further analysis, I subsequently categorized all coded narrative characters according to whether they represent the CAPA, 5 a representative of the CAPA's problem-affected 6 or problem-causing 7 target groups (cf. Appendix 1). The evaluation of propositions primarily bases on counts of these categories over time but is complemented with in-depth qualitative analyses of narrative content.

| FINDINGS
I now turn to the assessment of the theoretically devised propositions.

| Assessment of proposition 1: Narrative depiction of the CAPA
Out of the 667 coded narratives, 240 (35.98%) feature the CAPA as a narrative character, with the CAPA appearing as either a hero, villain, or victim. Table 2 ("Proposition 1") and Figure 2 give evidence of the steady build-up of narrative CAPA depictions as a villain from their inception until phase F I G U R E 1 Newspaper articles mentioning the CAPA over time. Key: X-axis displays the period of investigation, and Y-axis displays the frequency of newspaper articles mentioning the CAPA. Blue line: LOESS model, span = 0.15. Red lines: phase borders generated on the basis of LOESS model. Please note that Sundays were eliminated from the calculation, as none of the investigated newspapers appears on this day of the week 4, which includes the time of the double infanticide in Flaach. The frequency of narratives featuring the CAPA generally decreases afterward, indicating a loss of public attention and a certain saturation effect. Nevertheless, the relative share of the CAPA as a villain remains high. An in-depth examination of narratives featuring the CAPA as a villain reveals a wide range of accusations. Recurring themes are, for example, overpriced measures, incomprehensible communication, or flawed decisions due to allegedly lacking engagement of CAPA employees with their target groups. While many of these accusations resurface time and again throughout the entire period of investigation, the double infanticide marks a pronounced turning point in the quality of CAPA villain depictions that is best understood by recurring to the concept of reputational dimensions. According to Carpenter (2010, 46-47), an organization's reputation may be evaluated by referring to its performance, procedures, technical expertise, or moral standing. Before January 2015, the CAPA are mostly depicted as villains based on criticisms concerning their performance-they are described as slow, malfunctioning, etc.-or concerning procedural aspects such as a lacking inclusion of implementation partners. After the double infanticide, the moral dimension of their reputation becomes dominant, F I G U R E 3 Development of problem-causing target group narrative depictions. Key: X-axis displays the seven phases of the investigated period (cf. Even though after several weeks, attention subsides as the number of narratives about the CAPA decreases, criticism based on moral aspects remains prevalent in the rest of the dataset.

T A B L E 1 Phases' start and end dates, number of articles and narratives
Overall, proposition 1 is partly corroborated by the data. The emergence of the CAPA's negative reputation was associated with an increase in narratives depicting them as a villain. However, a loss of attention after the double infanticide led to a drop in the frequency of CAPA villain depictions.

| Assessment of propositions 2a and 2b: Narrative depiction of problemcausing target groups
The depiction of problem-causing target groups features marked changes over time (cf. Table 2, "Propositions 2a and 2b," and Figure 3). While at the beginning of the investigated period, problem causers do not appear in the public discourse, already the second phase (June 2013 to July 2014) marks an increase with villain depictions clearly dominating. While the remaining period of investigation features considerable fluctuations in the number of narratives casting problem causers as villains, they continue to overweigh in the overall depiction of this target group. A frequent example in this category is the mother who committed the double infanticide in Flaach: Yet here, it was not the CAPA, but the mother who committed grave injustice against her children.
(Blick, 7 January 2015) Also less drastic examples make up this development, such as parents who subject their children to permanent surveillance or single parents who manipulate their children to the detriment of their ex-partners. Overall, these narratives correspond to the CAPA's basic policy problem definition: Asymmetrical power relations, as they are frequently found in family constellations, may lead to abuse.
Over time, however, we see the emergence of alternative problem causer depictions. First, at the time of the double infanticide in Flaach, there is a surge of victim depictions. In these narratives, parents or near relatives are not causing a problem, but suffering from it: They are put under stress and harassed, their rights are curtailed. In almost 50% of these narratives, the villains are the CAPA, being responsible for the problem causers' suffering.
In a third stage, problem causers assume the role of heroes. Once again coupled in half of all instances with the CAPA as villains, we now see parents fighting for their families or elderly siblings "saving" each other from the CAPA. A case that reached prominence in this context is the story of a father whose children were placed in a home by the CAPA. In a cloak-and-dagger operation, he brought the children and their mother to the airport, where they boarded a plane to the Philippines, the mother's country of origin. Thus, the father removed his children from the CAPA's area of jurisdiction. A journalist reports:

For his resistance against the Child and Adult Protection Authorities (CAPA), on the internet the father is celebrated as a hero. His own Facebook page is flooded with friend requests, and in a new Facebook group fans document the case in real time. He gets comments such as "Awesome dad! Hero," the authorities by contrast are addressed with a "Fuck CAPA!"
(Tagesanzeiger, 29 July 2015) Similar to the CAPA villain depictions, the depiction of problem causers experiences a significant drop in the last phase of investigation.
The consecutive emergence of problem causer depictions first as victims and then as heroes corresponds to a revaluation and an empowerment, as the public seemed to perceive problem causers-at least in part, the continuing high amount of villain depictions also provides evidence of other opinions-as taking matters into their own hands. This entails a negation of the underlying policy problem: If the problem causers are heroes, then the underlying problem does not exist. Therefore, the hero depictions do not merely amend the evaluation of these individuals but question the very basis of the CAPA's interventions.
To conclude, propositions 2a and 2b both fail to capture the observed developments. While the depiction of problem causers as villains continuously dominates the examined narratives, both victim and hero depictions contest this social construction. The emerging negative reputation of the CAPA is accompanied by complex and changing dynamics in the depiction of the problem-causing target group.

| Assessment of proposition 3: Narrative depiction of problem-affected target groups
As Table 2 ("Proposition 3") reveals, the depiction of the CAPA's problem-affected target groups is roughly stable over time and predominantly casts them as victims. However, a non-negligible minority of narratives features villain depictions, peaking in phase 3 and phase 5 (cf. Figure 4). A qualitative examination reveals that the peaks are caused by two events that each sparked considerable media attention. In phase 3, six children from an Eritrean refugee family in need of support were depicted Problem affected as villain Problem affected as hero Problem affected as victim in several newspaper articles as overstraining the services of the village they lived in. The social construction of refugees has been ambivalent in Switzerland, with the right-wing populist Swiss People's Party regularly denouncing them as exploiting Swiss hospitality. Phase 5 was dominated by the case of a man running amok who had escaped a psychiatric clinic in which the CAPA had temporarily placed him. His killing spree results in several narratives depicting him as a villain. Thus, while the first case is characterized by a social construction from another policy field (migration) that superposes the construction common for the CAPA's policy field, the second case constitutes a rare example of a problem-affected person that has become a danger to society. In consequence, proposition 3 is partly corroborated. While the overall continuously dominating depiction of problem-affected target groups as victims during the emergence of the CAPA's negative reputation may be confirmed, the distinctive amount of villain depictions shows that the original proposition of problem-affected target groups being seen as victims is over-simplified. Table 3 provides a summary of the propositions' empirical assessment. In the following, I evaluate the application of narrative analysis to research on public organizations' reputation by harking back to the three benefits of narratives mentioned in this article's introduction.

| DISCUSSION
With narratives constituting a central influence on human cognition, I first postulated that through their study, we may gain insights concerning the formation of an organization's reputation. The analysis at hand demonstrated this benefit, both leading reputational researchers' focus toward the affective side of reputation (Bustos, 2021) and enlarging their perspective to include persuasive communication stemming from other sources than the organization (Maor, 2016). The emergence of the CAPA's negative reputation was accompanied by a multitude of narratives that underwent varied dynamics. While the analysis was restricted to narrative characters, examining other narrative elements such as frequently mentioned problems and solutions in CAPA narratives or prevalent plot types-for example, if the CAPA are a villain, do they harm their victims purposefully or inadvertently? (Stone, 1989)might enable further insights.
Next, I pointed to the social constructions contained in narratives as an informational source for the state of an organization's reputation. The analysis shows that in quantitative terms, this statement must be restricted (cf. Table 3, "Proposition 1"). Whereas the growing negative reputation did coincide with an increase in narratives depicting the CAPA as a villain, the general loss of public attention after the implementation scandal entailed a decrease in villain narratives. Therefore, more accurate quantitative predictions would necessitate the inclusion of research on general attention and media dynamics (e.g., Baumeister et al., 2001). By contrast, in qualitative terms, the narratives indeed inform knowledge about the CAPA's reputation. The villain depictions emerged early on and focused on different aspects of the CAPA, with a marked change from questions of performance and procedure to morality issues after the double infanticide in 2015. As opposed to general assertions concerning different reputational dimensions, the examination of narratives allows us to go one step further and identify the exact aspects that were relevant for the emergence of the CAPA's reputation. For example, the main points of criticism concerning the CAPA's performance were not excessive interventionism or exuberant passivity, but overpriced measures and flawed decision making.

T A B L E 3 Summary of proposition assessments
Thus, while survey instruments (Lee & van Ryzin, 2019;Overman et al., 2020) are more precise measures of reputation, the analysis of social constructions in narratives is a useful tool to gain complementary information on the discourse surrounding an organization. Such information may serve as a basis for communicational countermeasures.
The third advantage of narratives announced in the introduction concerns their ability to capture relationships between multiple actors. The analysis at hand demonstrated this feature by focusing on depictions of the relationship between a public organization and its target groups. This analytic extension provided valuable information by revealing that the emergence of the CAPA's negative reputation was accompanied by a partial reframing of its target groups. An increasingly positive depiction of family members that are often considered to be problem causers interfered with the CAPA's basic legitimacy as a problem-solving organization. The fact that I did not anticipate such a development theoretically emphasizes the necessity to take into consideration such fundamental questionings of public organizations in societal debates.
Hence, the reputational approach benefits from an expansion toward relational dynamics between multiple actors. Theoretically, a multitude of scenarios are conceivable. While the dynamic in the CAPA case corresponded to a trade-off between the organization's and the target groups' reputation, in other cases, mutual uplifting or down-pulling dynamics might occur. Reputational scholars have investigated similar phenomena for multiple organizations sharing a common reputation, for example, under the labels of the "bad apple" or "least common denominator" effect (Carpenter & Krause, 2012;Waeraas, 2015). By contrast, to my knowledge so far, no studies exist that investigate dynamics between multiple reputations of different actors. This constitutes an interesting avenue for future research.
Last, this article introduced the conceptualization of target groups as distinguished between problem causers and problem affected to research on reputation. In the case at hand, this differentiation proved useful, as the narrative dynamics varied considerably between the two groups. Both the distinction and additional reflections concerning the properties or social constructions of a public organization's target groups (Schneider & Ingram, 1993) may prove valuable for research on reputation, as they equip scholars with an enhanced understanding on an organization's behavior toward its audiences. Furthermore, this analysis shows that target groups' power should not be underestimated: Although formally, they are the subject of an organization's actions, this power balance may be challenged in a reputational context. Drawing back on Carpenter's seminal insights (2010), it is therefore not only organizations that leverage power through their reputation but also other actors such as organizations' target groups.

| CONCLUSION
This article provided a first application of narrative analysis in research on organizational reputation. Specifically, I investigated what narratives accompany the emergence of a negative reputation. The Swiss Child and Adult Protection Authorities (CAPA) served as empirical case, with their reputation having developed "from zero to villain" within only 3 years after their inception. I conducted a quantitative content analysis of 667 narratives that had been extracted from mainstream news media by means of the Narrative Policy Framework (NPF).
Findings indicate that narrative depictions of the CAPA as villain were prevalent early on. They accumulated over time and assumed a different quality after an implementation scandal that spurred the CAPA's negative reputation. While the target groups supposed to be beneficiaries of the CAPA's interventions were continuously cast as victims, the depiction of target groups often addressed by the CAPA as problem causers underwent marked changes. The emergence of narratives portraying these groups as victims and eventually even as heroes fighting the CAPA shows that the CAPA's legitimacy was increasingly questioned.
The combination of research on the reputation of public organizations with narrative analysis is fruitful both from a theoretical and from a practical perspective. Scholars interested in reputation benefit from the examination of narratives, as they are central heuristics for human cognition and thus important in the formation of reputation. Furthermore, they contain information concerning the social construction of a public organization and of actors in its environment, thus allowing for inferences on the organization's state of reputation and on relational dynamics with others.
The potential of the combination between reputation and narratives is far from being exhausted. Specifically, I would like to highlight three avenues for future research. First, the inclusion of narratives' authors into the analysis might allow for inferences of different actors' rhetorical strategies visà-vis a public organization, thus providing the organization with further information on reputational threats and opportunities. Second, the examination of new data sources such as social media, parliamentary debates, or direct communication by political parties, interest groups, or NGOs could generate more specific information concerning the prevalent perceptions of an organization in different audiences. Third, comparisons between narratives circulating in public and an organization's actual implementation practices might reveal misperceptions that need to be tackled.
From a practical perspective, public organizations may use the knowledge generated through narrative analysis as a basis for their reputation management. They receive a detailed picture of the narratives that circulate about them in public. This allows organizations on the one hand to identify harmful narratives that should be combatted and on the other hand to detect useful narratives that might be reinforced. A narrative analysis may also be used as a starting point for the generation of new narratives.
Exemplifying this usage on the case at hand, two lessons may serve as starting point for adaptations in the CAPA's reputation management. First, the analysis suggests that the CAPA are firmly associated with the villain character, notably regarding the moral dimension of their reputation. To amend this, communication focusing both on the policy's moral basis-to help persons that are unable to assume their own rights-and on the measures taken to uphold this basis is needed. As narratives are effective rhetorical devices, these messages might be delivered in the form of specific case descriptions that ended in the alleviation of the problem-affected target's suffering. Additionally, portrayals of employees conveying their ethic convictions and daily efforts to help persons in need might be helpful to alter the depiction of the CAPA as an impersonal, villainous organization.
Second, the analysis has highlighted that the problem-causing target group is particularly prominent in public discourse and is depicted as either suffering from or defending itself against the CAPA. Here too, communication is key: The CAPA should describe their principles when in contact with problem causers, elucidate the reasons why somebody is conceived as a problem causer, and explain the target groups' rights during a procedure. Furthermore, the CAPA should reinvestigate the cases gone public to determine whether mistakes from their side occurred and draw their own lessons.

ACKNOWLEDGMENTS
My deepest thanks go to Fritz Sager for his useful advice and guidance throughout the last years. Further, I would like to thank Madalina Busuioc, Bettina Stauffer and the participants of the panel "The Politics of Bureaucratic Reputation" at the ECPR General Conference 2020 for their valuable inputs.
2 Please note that many policies target both groups simultaneously, for example, anti-tobacco policies that restrict tobacco companies' marketing possibilities and provide withdrawal programs for smokers. 3 Please note that the categorization of problem causers and problem affected refers to the normative evaluation imbued in the public policy and is not an individual judgment by the author. The socially constructed nature of problems necessitates a normative evaluation that may be derived from a policy's setup (Sager, Ingold, and Balthasar 2017, 11). 4 The analysis is restricted to the German-speaking part as it is only there that a negative reputation of the CAPA emerged. Why the CAPA in the French-speaking part were spared from this development is subject of Stauffer (2021). 5 That is, the agency itself, a specific local CAPA, or CAPA employees.
How to cite this article:  Narrative's author The person from whose perspective the narrative is told Schlaufer (2016) Hero "[…] actor(s) who plan to or fix, solve, assist, or seek to resolve past, current or future problem. Need to possess intention and/or agency" Crow and Berggren (2014) Villain "[…] actor(s) who create, cause, contribute, instigate, exacerbate, or plan to contribute to the problem. Need to possess intention and/or agency" Crow and Berggren (2014) Victim "[…] actors(s) who suffers, is targeted, is effected by the problem and/or Villain" Crow and Berggren (2014) Actor Categories:

CAPA
The CAPA as an agency type, specific local CAPA, CAPA employees Problem-affected target groups Persons affected by the problem the CAPA aims to solve, that is, persons that are unable to assume their rights and control their assets, such as orphans, abused children, or elderly persons affected by dementia Problem-causing target groups Persons considered by the CAPA as contributing to the problem constellation, for example, abusive parents or relatives of a person with dementia involved in asset conflicts