Mediated crises and strategic crisis communication of third sector organizations: a content analysis of crisis reporting in six countries

Abstract

Purpose

This study aims to contribute to strategic crisis communication research by exploring international media representations of third sector crises and crisis response; expanding the range of crisis types beyond transgressions; and developing a framework that integrates framing and crisis communication theory.

Design/methodology/approach

Quantitative content analysis was applied to identify patterns in crisis reporting of 18 news media outlets in Canada, Germany, India, Switzerland, UK and US. Using an inductive framing approach, crisis coverage of nonprofit organizations (NPOs) and intergovernmental organizations (IGOs) between 2015 and 2018 was analyzed across a wide range of crises, including but not limited to prominent cases such as Oxfam, Kids Company, or the Islamic Research Foundation.

Findings

The news media in six countries report more internal crises in the third sector than external crises. The most frequent crisis types were fraud and corruption, sexual violence/personal exploitation and attacks on organizations. Exploratory factor analysis revealed three components of crisis response strategies quoted in the media, conditional rebuild, defensive and justified denial strategies. Causal attributions and conditional rebuild strategies significantly influenced media evaluations of organizational crisis response. Three frames of third sector crises were detected; the critique, the damage and the victim frame. These frames emphasize different crisis types, causes, crisis response strategies and evaluations of crisis response.

Originality/value

The study reveals the particularities of crises and crisis communication in the third sector and identifies factors that influence mediated portrayals of crises and crisis response strategies of nonprofit organizations (NPOs) from an international comparative perspective. The findings have relevant implications for crisis communication theory and practice.

Citation

Schwarz, A. and Diers-Lawson, A. (2024), "Mediated crises and strategic crisis communication of third sector organizations: a content analysis of crisis reporting in six countries", Corporate Communications: An International Journal, Vol. 29 No. 4, pp. 567-591. https://doi.org/10.1108/CCIJ-08-2023-0117

Introduction

Many nonprofit organizations (NPOs) and nongovernmental organizations (NGOs) are highly influential actors at local, national and global public levels influencing policy and awareness for multiple social, political, environmental, cultural and economic issues (Lewis, 2003). In many countries they are a significant contributor to the national economy (Chapman et al., 2021b). Together with certain intergovernmental organizations (IGOs) such as UNICEF or WHO, many are key providers to development aid, humanitarian relief, poverty alleviation, public health and environmental protection. Yet others arguably advocate for less desirable causes such as promoting the distribution and possession of arms (e.g. NRA) or the denial of human-made climate change (e.g. The Heartland Institute).

In recent decades, the nonprofit sector has experienced several high-profile scandals and crises that have the potential to adversely affect the third sector as a whole (Chapman et al., 2022a). Well-known cases include different Red Cross organizations (Long, 2016), Oxfam and Save the Children (Scurlock et al., 2020), the Catholic Church (Maier, 2005), or FIFA (Manoli and Anagnostou, 2022). These crises garner significant news coverage in the mainstream media (MSM) and are subsequently discussed across communication platforms. NPO crises have the potential to damage the organizations’ reputation and trust with a variety of key stakeholders including donors, volunteers and partner organizations (Sisco et al., 2010). Although a sector-wide crisis of trust in NPOs has not (yet) been empirically confirmed (Chapman et al., 2021a), the risk cannot be ignored because trust in NPOs has been found to significantly influence charitable giving (Chapman et al., 2021b). This may also affect certain IGOs, which are entities created by multilateral treaties between several nations with different purposes, such as a number of UN agencies with missions comparable to international NGOs (e.g. UNHCR, UNICEF) and donations as an additional source of funding. Thus, limiting the loss of trust in NPOs/IGOs in the context of crises is mission critical for these organizations’ public relations (PR) efforts because publics hold them to higher moral and ethical standards compared to commercial organizations (Chapman et al., 2022a).

Crisis communication and PR research of NPOs/IGOs is scarce (Sisco, 2012; Schwarz and Pforr, 2011) and scattered across many disciplines (Chapman et al., 2022a). Moreover, while the NPO communication management literature identifies the MSM and journalists as important stakeholders (González et al., 2018), there are few studies evaluating MSM coverage of third sector crisis response despite the MSM’s influence on public opinion (Gibelman and Gelman, 2001, 2004) and the importance of media relations of NPOs in crisis contexts reported by communication practitioners (Zerfass et al., 2013). Most of this research centers on scandalized NPOs, the nature of allegations and their consequences. Crises where NPOs are the crisis victim are mostly disregarded (e.g. kidnapping of NPO workers, terrorist attacks, government oppression), although NPO leaders clearly include them in the list of major threats (Gilstrap et al., 2016). To our knowledge, media evaluations of NPOs’ crisis response and quoted strategies have not been studied in general.

This study contributes to strategic crisis communication research in several ways. First, while the nature and impact of MSM content on corporate reputation has been extensively studied, we lack an understanding of the unique features of crisis reporting about NPOs. This study addresses this gap by analyzing a wide range of third sector crises and the ways in which MSM frame crisis attributions, crisis response strategies and reputation assessments in countries that are considered major hubs for NGOs/IGOs. The findings will facilitate a better understanding of the role of MSM in reputation constitutions/loss in crisis contexts, with implications for crisis communication practice in the third sector in several countries. Second, in contrast to previous (case) studies on NPO crises, this study does not limit the analysis to transgressions or scandals, but includes all types of crises that NPOs experience according to the media. This will provide a more complete picture of how MSM evaluate the crisis response and reputation of NPOs depending on the specific context of crisis type, crisis causes and national context of countries with a large third sector. Third, to analyze and compare the aforementioned factors we apply and test assumption of previous crisis communication research and develop a framework that integrates framing and crisis communication theory for the analysis of mediated NPO/IGO crisis responses. This framework can be used for future research on mediated crisis communication and allows for comparisons across industries or long-term monitoring of crisis-induced reputation dynamics in MSM.

Although the importance of social media has increased in terms of news consumption (Scheffauer et al., 2023) and communication activities by NPOs (Santos et al., 2020), our study focusses on crisis representations of the third sector in MSM, for the following reasons. First, social media and news on social media are generally perceived as less credible compared to MSM (Scheffauer et al., 2023; Park et al., 2020). Second, crisis communication research suggests that crisis information in MSM has potentially more harmful effects on organizational reputation and triggers more secondary crisis reactions (etc. word of mouth, boycott, etc.) (Xu, 2020; Schultz et al., 2011; Utz et al., 2013). Thus, despite the increasing relevance of social media in crisis communication, MSM content seems to be more significant for the constitution or loss of reputation of organizations (Eisenegger and Imhof, 2008).

Crisis communication theory and framing

Literature review studies of crisis communication research have shown that situational crisis communication theory (SCCT), image restoration theory (IRT), attribution theory (AT) and framing are commonly applied frameworks (Ha and Boynton, 2014; Diers-Lawson, 2017). While SCCT, IRT and AT were mainly employed to study strategic crisis response from an organizational perspective in the business sector, framing was mainly used to investigate media constructions of crises; yet research that integrates these frameworks remains under-developed.

Situational crisis communication and attribution theory (RQ1 and RQ2)

SCCT built on IRT’s focus on image repair discourse (Benoit, 1997) by evaluating the role of attribution, or blame, in crisis response in combination with attribution theory (Coombs and Holladay, 2004; Coombs, 1995). It offers a prescriptive framework of crisis communication including an empirically validated differentiation of crisis response strategies by implied acceptance of crisis responsibility and a typology of crisis types according to the level of perceived crisis responsibility among stakeholders of an organization (Coombs, 2019). A content analysis of 51 studies employing IRT or SCCT between 1991 and 2009 revealed that the most frequently used strategy was bolstering, followed by denial, corrective action and mortification (Kim et al., 2009). Most of the authors of these studies evaluated full apology as the most effective strategy followed by mortification and corrective action. Denial was the strategy evaluated as least effective. A meta-analysis of crisis response research identified corrective action as the most effective, and denial as the most ineffective response strategy, while denial was used most often by organizations (Arendt et al., 2017).

SCCT-related research also resulted in categorization of crisis types as preventable (high blame), accidental (low blame) and victim crises (limited or no blame). Studies found that matching crisis response strategies with crisis types yields different effects on emotions, reputational evaluations and word of mouth towards organizations among stakeholders. Therefore, SCCT’s approach to crisis response is dependent on the crisis type or attributions of crisis responsibility to an organization because research consistently found that blame negatively affects reputation and/or trust (Coombs and Holladay, 2004; Ma and Zhan, 2016; Schwarz, 2012). These findings reflect SCCT’s roots in attribution theory (Coombs and Holladay, 1996; Härtel et al., 1998). Attribution theory suggests that events – like crises – that are personally relevant or threatening are likely to trigger attributions of blame as a way to reduce the uncertainty produced by the events (Weiner, 1985; Galloway et al., 2019), allowing people to feel more confident about events or behavior they observe (Heider, 1958).

In addition to modifiers added to SCCT like crisis and relationship history, further extensions of the framework applied Kelley’s (1967, 1972) covariation principle and found that when in crisis, an organization’s stakeholders use observations of the organization’s behavior over time to decide whether blame should be attributed internally, externally, or to circumstances (Schwarz, 2012; Kim et al., 2022). These judgments are made based on covariation information, more specifically consensus, distinctiveness and/or consistency information. Consensus information explains behavior in a certain crisis context in comparison to other social entities’ behaviors in similar crisis contexts. Distinctiveness information explains behavior in a certain crisis context compared to that organization’s behavior in other crisis contexts different from the focal situation. Consistency information tells about behavior over time in similar crisis situations. Covariation information helps stakeholders judge the likely causes of a crisis which, in turn, influences their attitudes and reputational evaluations of social entities. Observers will attribute crisis causes more likely to internal factors of the organization if (1) the organization is the only one involved in a certain crisis context (low consensus); (2) that organization is facing other crisis issues in addition to the focal crisis (low distinctiveness); and (3) that organization has faced similar crises repeatedly in the past (high consistency). On the other hand, observers will attribute crisis causes more likely to external or circumstantial factors, if (1) an organization is one of several being involved in a certain crisis context (high consensus); (2) that organization is facing no other crisis issues in addition (high distinctiveness); (3) and that organization has not faced similar crises in the past (low consistency).

Based on this review of attribution theory and SCCT, we know that attributions are crucial for stakeholder perceptions of organizations in crisis, but there remains a dearth of knowledge about media coverage of blame attributions across crisis types related to NPOs/IGOs; therefore, we pose our first research question.

RQ1.

How do media reports on NPO/IGO crises attribute causes, causal antecedents (covariation information) and organizational responsibility to these organizations in crisis?

Second, it is also clear from the research related to SCCT that crisis-related attributions have a significant impact on organizational reputation, yet that research is seldom conducted on the third sector. Therefore, our second research question is.

RQ2.

How do media reports on NPO/IGO crises evaluate the organizations’ reputation and crisis response depending on attributions of cause and responsibility as well as crisis response strategies?

Moreover, in line with previous research reviewed here, we can hypothesize.

H1.

The more the causes and causal antecedents of NPO/IGO crises are attributed internally to the organization as opposed to external factors the more the media will attribute crisis responsibility to the organization.

In addition to crisis attributions, it is yet unclear which patterns the media follow when evaluating an NPO’s/IGO’s reputation and crisis response, e.g. depending on their selection of attribution information as well as quoted crisis response strategies. Reputation has been defined as “collective social judgment regarding the quality or capabilities of a focal actor within a specific domain” (Boivie et al., 2015, p. 188), with the MSM playing a crucial role in constituting and/or damaging an organization’s reputation (Eisenegger and Imhof, 2008; Jonkman et al., 2020). Systematic reviews of crisis communication research have found strong effects of attributions and weak or moderate effects of crisis response strategies on reputation (Ma and Zhan, 2016; Arendt et al., 2017). However, the third sector was only rarely studied in this regard, which is why we posed RQ2.

Previous research suggests that internal causal attributions are more harmful for organizational reputation (Coombs and Holladay, 2004; Ma and Zhan, 2016) since stakeholders perceive higher levels of crisis responsibility of organizations (Schwarz, 2012). In addition, crisis response strategies implying low responsibility acceptance of organizations are the least effective in protecting reputation in crisis contexts (Arendt et al., 2017; Kim et al., 2009). Given these findings, it is of interest to understand whether journalists follow a similar pattern when attributing responsibility in their reports and quoting an organization’s crisis response strategy. Therefore, we hypothesize.

H2.

The more a crisis is attributed to internal causes and internal responsibility of an NPO/IGO and the more the organization is quoted denying responsibility (e.g. external scapegoating) the more negative the media will evaluate the organization’s reputation and its crisis response.

Media framing (RQ3)

In addition to testing for basic relationships between crisis attributions, quoted crisis response strategies and media evaluations of NPOs/IGOs, our goal is to understand whether these characteristics of media reports cluster systematically into recurring patterns of mediated NPO/IGO crises and whether these patterns can be found in MSM across countries. In media research, such patterns have been frequently studied by using the framing approach (D’Angelo and Kuypers, 2009).

Framing is useful in interdisciplinary contexts to analyze communication processes. Framing is a process that highlights information in such a way that makes some aspects of that information more salient than others, which guides how that information is interpreted (Scheufele, 2003). Entman (1993) explained that: “To frame is to select some aspects of a perceived reality and make them more salient in a communicating text, in such a way as to promote a particular problem definition, causal interpretation, moral evaluation, and/or treatment recommendation” (p. 52). Based on Entman’s framework, four dimensions of frames (i.e. problem definition, causal interpretation, moral evaluation, treatment recommendation) have been identified in media coverage resulting from the way that journalists select and process information (Matthes and Kohring, 2008).

Attribution theory provides the bridge between crisis communication theory and framing. We therefore assume that when media report about NPO/IGO crises and their impact (problem definition), they are likely to offer causal explanations (causal interpretation), evaluate focal NPOs/IGOs in terms of moral responsibility, reputational dimensions and behavior (moral evaluation) consistent with previous attributions and identify potential solutions (treatment recommendation). Thus, for each of the four framing dimensions theory-driven indicators (SCCT, attribution theory) were developed for content analysis of NPO/IGO crisis reporting, which were subsequently used in a cluster analysis to detect recurring patterns in the media coverage (i.e. frames). This procedure is aligned with Matthes’ and Kohring’s (2008) method of frame detection, which is widely regarded as valid and reliable. Thus, we pose our third research question connecting cross-national media coverage of third sector and IGO crises.

RQ3.

What are dominant media frames of NPO/IGO crises and how do they differ across countries?

Material and methods

We conducted a quantitative content analysis of 18 news media outlets in six countries. This adds to the body of research on strategic communication in the third sector where mediated crisis response and cross-national differences have rarely been studied.

Study sample

The countries included were Canada, Germany, India, Switzerland, the US, the UK These countries represent world regions where the third sector is most developed and most of the larger NGOs and IGOs are headquartered. In addition, some of the most important donors (US, Germany, UK, Canada, Switzerland) and the largest recipient (India) of development aid are included in the country sample (OECD, 2022). An additional selection criterion was the variance of trust in NGOs. According to the Edelman Trust Barometer (Edelman, 2018), the sector is highly trusted in India (67% trust NGOs) and Canada (50%), is moderately trusted in the US (49%) and less likely to be trusted in the UK (46%) and Germany (37%). Switzerland was not in the Edelman survey, but headquarters numerous IGOs as well as many of the larger NGOs.

The media sample included highly circulated national daily newspapers published in English or German language. In each country, three of the most circulated newspapers in English or German were selected that represent a variety of political leanings (liberal/left-wing and conservative/right-wing) and reporting styles (tabloid and quality). The selection of media outlets reflects the news media landscape of each country, where, for example, tabloids are important in terms of circulation/impact in some cases (e.g. UK and Germany) and less relevant in others (e.g. Canada and India) (Newman et al., 2018). In most cases, the selected media outlets were among the top five in each country in the time period of this study (circulation) (e.g. ABC Audit Bureau of Circulations, 2017; Ponsford, 2017; Sieber, 2019). In a few cases, slightly lower circulated media outlets were included (e.g. taz in Germany, Blick in Switzerland) because of limited database access or to attain the goal of diverse political leanings and reporting styles for each country (Table 1).

The period of analysis ranged from January 1, 2015 to May 1, 2018. That way, we were able to include a broad range of crisis cases in the NPO sector including but not limited to NPOs like Oxfam, the Kids Company and the Islamic Research Foundation. Thus, we avoided reporting bias due to extraordinary events such as the COVID-19 pandemic that started in 2020. To identify relevant media reports on these crises, we applied a complex search string of keywords to search headlines and lead paragraphs of articles. IGOs working primarily in the military field (e.g. NATO) and supranational organizations with executive and legislative functions (e.g. European Union) were excluded. The search term combined keywords that capture different types of crises and NPOs/IGOs. These terms were identified through a previous literature review (e.g. Gibelman and Gelman, 2001) and a preliminary review of media reports. English and German keywords of crisis types included, for example, mismanagement, scandal, fraud, allegations, accident, attack, killing, kidnapping, sexual harassment, deficit, illegal, cyberattack, etc. Keywords to capture articles that focus on NPOs included for example nonprofit organization, charities, nongovernmental organizations, third sector, association, humanitarian, foundation, religious, Church, etc. For IGOs, we also included specific names since the term itself (IGO) is rarely used by journalists (e.g. UN, United Nations, UNDP, UNICEF, UNHCR, World Health Organization, etc.).

Coders used headers and lead paragraphs to check whether an article dealt substantially with a crisis of an NPO, NGO, or IGO. Reports dealing with movies or books were excluded. Articles mentioning NPOs/IGOs in the context of natural disasters or wars without referring to a specific crisis event for the organization itself were also excluded (e.g. undisrupted relief operations). The selected articles were then reviewed completely to determine to what extent the crisis is among the main topics. If at least 20% of the text referred to such crises, the article was included.

The keyword search returned 5,141 articles. After screening of headlines and the articles’ text, we identified 1,057 news items focusing on specific organizations (NPO/IGO) and specific crises. All articles from all sections were included from each media outlet except for three newspapers (The Guardian, Daily Mail, Times of India) where more than 100 media reports were identified. From these three media outlets, we drew a simple random sample of 40% resulting in a final overall sample of 691 articles from 18 media outlets for further detailed analysis.

Codebook

The development of the codebook was guided by Entman’s (1993) framing theory and the four frame elements of problem definition, causal interpretation, moral evaluation and treatment recommendation (Matthes and Kohring, 2008). Since the framing literature points to a broad range of heterogeneous measures for frame elements (Guenther et al., 2023; Dan and Raupp, 2018) that are not based on crisis communication theory, we incorporated concepts from attribution theory as well as crisis communication theory to meet the goals of our study. These concepts were matched to the four frame elements as shown in Table 2.

For the frame element problem definition, we coded information for up to the three most salient NPOs/IGOs addressed in the article including the name of the organizations in crisis and its field of activity (e.g. humanitarian, environmental, etc.). We coded the consequences of the crisis for the organization itself and its stakeholders with a total of 20 indicators. Moreover, the crisis type was coded based on an exhaustive list that was generated from previous nonprofit research (Gibelman and Gelman, 2004; Gilstrap et al., 2016) and SCCT (Coombs and Holladay, 2004).

For the remaining three frame elements, we only coded categories for the main organization identified as being “in crisis” in the article. The dimension of causal attribution included indicators of causal antecedents. Low distinctiveness, for example, was coded when an organization was involved in other crises besides the main crisis event. High consistency was coded when the article mentioned the organization had experienced a similar crisis in the past. Consensus referred to references to other NPOs/IGOs, aside from the main one, in the article experiencing similar crises at the time. Concerning the actual causes of the focal crisis, we coded whether the article mentioned crisis causes from within (internal) and/or outside of the organization (external).

The frame element of moral evaluation included indicators of moral responsibility of the organization or other entities for the main crisis. We coded the crisis response strategies of the focal organization as cited within an article since those strategies usually imply some level of acceptance or rejection of moral responsibility for a crisis. These strategies were derived from SCCT (Table 2). Further indicators measured the evaluation of crisis response and management such as timeliness, transparency and message consistency (Coombs, 2019; Gilstrap et al., 2016). The reputation of the organizations was coded by evaluation of actions, expertise, efficiency, social responsibility and attractiveness of the organization (Eisenegger and Imhof, 2008) on a three-point scale (positive, ambivalent, negative).

Mentions of treatment recommendations included several kinds of recommended actions during or after a crisis for the organization, politics/government, legal action, or the media. In addition, the sender of up to three treatment recommendations was coded.

To assess the codebook’s reliability, all seven coders were assigned to a random sample of 20 articles. Intercoder reliability using Gwet’s AC1 resulted in an acceptable mean score of 0.86. The aggregated reliability scores for each frame element are shown in Table 2. News media articles in German language were coded by native speakers.

Data analysis

Besides descriptive statistics, multiple linear regressions, binomial logistic regressions, one-way ANOVA as well as chi-square tests were used to test the hypotheses (RQ1, RQ2) and explore further relationships between variables. Effect sizes were reported as Cramer’s V (chi-square tests), adjusted R-squared, standardized beta (regression), odds ratio (OR) (logistic regression) and eta-squared (ANOVA). The interpretation of effect sizes followed Cohen’s (1988) recommendations.

While framing research using Entman’s four frame elements traditionally has used mainly hierarchical cluster analysis (Matthes and Kohring, 2008), we applied two-step cluster analysis for frame detection (RQ3). This technique has been reported to produce reliable and robust cluster solutions compared to latent class analysis or hierarchical cluster analysis (Benassi et al., 2020; Kent et al., 2014). Two-step cluster analysis has several advantages compared to other techniques such as the ability to include nominal and interval-scaled variables, being less sensitive to the violation of normal distribution assumptions, and, most importantly, determining the number of clusters based on statistical measures (AIC, BIC) instead of arbitrary choices based on visualizations or other criteria (Benassi et al., 2020). Especially for specifying the number of media frames in data sets, the researcher’s previous assumptions may lead to subjective decisions. Since the procedure was reported to work best with continuous variables (Bacher et al., 2004) we dummy coded variables and/or used standardized factor scores (regression) by using principal component analysis (PCA).

Results

Of the1,057 total news items focusing on NPOs/IGOs in crisis, the UK published the most (43%), while Canadian and Swiss news media published the fewest (Figure 1). After applying the previously described sampling procedure, a final sample of 691 articles was analyzed. That sample slightly overrepresents the US and Germany and underrepresents the UK articles compared to the initial population of 1,057 news reports. UK-based media emphasized much more on crises of NPOs than any other country. They reported intensively on cases of sexual exploitation at Oxfam (25%) as well as the allegation of mismanagement against the Kids Company (15%). Out of all crisis reports, 24.3% were published in 2015, 28.4% in 2016, 20.3% in 2017 and 27.1% in 2018. Given that we only included the media coverage of the first four months of 2018, that year represents a significant increase in mediated NPO/IGO crises.

Overall, these finding suggest news media covered a wide variety of crises. Most of the articles reported about NGOs, foundations, or associations (77%) ranging from service organizations like Oxfam or Save the Children to political foundations like the Donald Trump Foundation. IGOs (mainly UN agencies) were covered substantially less often (12.7%). The remaining nonprofit crises were reports about churches, hospitals, or universities (5.8%). The field of activity of most organizations was humanitarian relief and development (47.2%), followed by community development (17.2%), religious or spiritual (9.1%), politics/business (3.5%), human rights (2.9%), environment (2.6%) and sports (1.2%). Surprisingly, 72% of all IGO crises were found in German media, while the other five countries’ media had a share of 3%–8% of IGO crises. Thus, most of the crisis reporting in each country except Germany referred primarily to the third sector.

Crisis types and attributions of cause and responsibility (RQ1)

In terms of crisis types, cases of fraud and corruption were by far the most frequent (26%). This was followed by cases of sexual violence or exploitation of persons in need of support (12.6%). Another 9.4% of articles addressed attacks on NPOs/IGOs (e.g. terrorism). The political influence of governments to change or inhibit an organization’s activities was reported in 8.4% of the articles, followed by sexual violence at the workplace (6.2%), legal misconduct (5.9%) and safety issues due to the rise of armed conflicts or terrorism (5.2%). The remaining categories pertain to the long tail with less than 3%.

Most of the crises were internal (60.6%) compared to external (33.6%) crisis types. Additionally, many journalists attribute specific internal factors (37.2%) as causing the crisis compared to external factors (31.4%). However, 33% of the articles did not mention any specific causes. There was a high negative correlation between internal and external attributions excluding articles without mention of any causes (r(459) = −0.94, p < 0.001). This suggests that journalists framed the crises as either internal or external crises instead of pointing to more complex causal analysis by mentioning internal and external factors simultaneously leading to the crisis (only 13 articles).

In terms of causal antecedents, we found somewhat substantial mentions of high consensus (22%), low distinctiveness (8.8%) and high consistency (17.2%). These factors were recoded into dummy variables (0 – not mentioned, 1 – mentioned) for further analysis. To test the influence of causal attributions and causal antecedents on responsibility attributions (H1), we used the receiver of crisis responsibility as dependent binary variable differentiating between internal responsibility of the NPO/IGO for the crisis (1) versus external or no attribution of responsibility (0). Causal attributions were computed as the difference between the mention of external causes (0/1) and internal causes (0/1), resulting in a three-point scale where 1 means exclusive external attribution, 0 no attribution or mention of both internal and external causes and −1 emphasis of only internal causes in an article.

A binomial logistic regression model with internal crisis responsibility of an NPO/IGO as dependent variable was statistically significant, χ²(4) = 245.5, p < 0.001, resulting in a moderate amount of explained variance (Nagelkerke’s R² = 0.4). The percentage of accuracy in classification was 78.3%, with a sensitivity of 71.0% and a specificity of 84.5%. All predictors had significant effects on crisis responsibility of NPOs/IGOs (Table 3). Causal attribution had a large negative effect (OR = 0.23). Thus, more emphasis of external versus internal causes of a crisis decreased the likelihood of crisis responsibility. Low distinctiveness had a large positive effect (OR = 5.45) and high consistency had a small positive effect on responsibility (OR = 2.15). In other words, pointing to other ongoing crises of an NPO/IGO besides the main focal crisis at the time as well as mentioning similar crises of that NPO/IGO in the past increased the likelihood of the media to assign the moral responsibility for that crisis to the NPO/IGO. High consensus had a moderate positive effect on crisis responsibility (OR = 3.06). Hence, including the occurrence of similar crises at other NPOs/IGOs in the media coverage increased an article’s attribution of crisis responsibility to the focal organization. H1 was partially confirmed since internal causal attributions as well as low distinctiveness and high consistency were significantly related to internal crisis responsibility of NPOs/IGOs. For high consensus information an opposite relationship found since it was expected to be negatively related to internal crisis responsibility.

The attribution of internal crisis responsibility to an NPO/IGO was more frequent for internal than for external crisis types (χ²(9) = 171.27, p < 0.001, Cramer’s V = 0.50), which also confirms H1. Most frequent internal responsibility attributions were found for cases of sexual violence towards persons in need of support, fraud/corruption and legal misconduct of an NPO/IGO. The least amount of responsibility attributions was found for safety issues in war zones or terrorism contexts as well as attacks on NPOs/IGOs (Figure 2). In cases of government interference on NPOs/IGOs, the share of internal responsibility attributions was relatively high (43.1%) considering the fact it is an externally caused crisis. Vice versa, internal responsibility attributions were found less often than expected (46.5%) for cases of sexual violence at the workplace of an NPO/IGO, although it represents an internal crisis type.

Mediated crisis communication and evaluations of NPOs/IGOs (RQ2)

We examined crisis response strategies that were quoted directly or indirectly and how journalists evaluated them when reporting about NPO/IGO crises (Figure 3). Frequencies showed that denial was the most quoted strategy (20.5%). Further frequent response strategies also mainly implied no or low acceptance of crisis responsibility such as justification (10%), attack the accuser (8%), or external scapegoating (8%). Rebuild strategies with organizations accepting a fair amount of crisis responsibility such as apology (5%) and compensation (1%) were less common. A little more than half of the media coverage (52.8%) did not contain any of the crisis response strategies, whereas 22.6% referred to one, 13.3% to two and 7.7% to three of the strategies.

Because of the co-occurrence of response strategies in articles indicated by substantial correlations and for purposes of further statistical analysis, we conducted a PCA on all response strategies except compensation, excuse and ingratiation because of very low frequencies. This resulted in a satisfactory component solution (KMO = 0.63; Bartlett’s test p < 0.001). Further analysis yielded empirical justification for retaining three factors which accounted for 47.65% of the total variance (Table 4). The varimax-rotated factor solution revealed three interpretable components of mediated crisis response. The first factor combines apologies and corrective action with expressions of regret (e.g. sympathy for victims), but also internal scapegoating (blaming individuals within the NPO/IGO instead of the whole organization). We termed this set of strategies conditional rebuild strategies. The second component comprised rather defensive strategies such as external scapegoating and attacking accusers of the NPO/IGO in combination with describing the NPO/IGO as a victim itself of the crisis (victimage). This component was termed defensive strategies. The strategies denial, justification and reminder loaded on the third factor. This suggests that media coverage highlights this NPO/IGO responses when they deny that the situation is serious, downplay the impact and remind stakeholders of their past good deeds. We termed this factor justified denial strategies. The factor scores of the three components were used for further analysis (regression method).

To test the influence of quoted crisis response strategies (factor scores) on the media’s reputational evaluation of an NPO/IGO and its crisis response, we conducted another PCA to reduce correlated variables to broader factors and to avoid multicollinearity in subsequent analyses. All reputation evaluation and crisis response evaluation variables (three-point scales with 1-negative, 2-ambivalent and 3-positive evaluation) were included in the PCA except evaluation of social responsibility because of its low communality (0.23). This resulted in a satisfactory component solution (KMO = 0.66; Bartlett’s test p < 0.001). Two interpretable factors were retained which accounted for 45.41% of the total variance. The first factor reflects the media’s evaluation of an NPO’s/IGO’s reputation (four indicators) and the second the evaluation of crisis response (four indicators). The according factor scores (regression) were used for subsequent analysis with higher values indicating a more positive evaluation.

Multiple linear regression models were calculated to test the influence of quoted crisis response strategies (factor scores) in comparison to causal antecedents, causal attributions and attributed internal crisis responsibility. For the media’s evaluation of the NPOs’/IGOs’ reputation, a significant (p < 0.05) marginal influence of conditional rebuild strategies (B = 0.11) along with causal attributions (B = 0.13) was found. Thus, the application of conditional rebuild strategies and the emphasis on external versus internal causes increased positive reputational evaluations of NPOs/IGOs. However, the explained variance for the overall model was neglectable (R²_adj. = 0.02). Hence, H2 was only partially accepted and the results have to be treated with caution because of the poor model fit. While conditional rebuild strategies (high acceptance of responsibility) and causal attribution had significant small effects on reputation in the expected direction, other crisis response strategies, internal responsibility and causal antecedents had no influence.

For the media’s evaluation of the NPOs’/IGOs’ crisis response as dependent variable, a well fitted regression model (R²_adj. = 0.24) was calculated. Conditional rebuild strategies had a significant and negative influence on crisis response evaluation (B = −0.11). Thus, quoting these strategies increased negative evaluations of crisis response. Crisis responsibility (internal), high consistency and low distinctiveness increased negative evaluations of crisis response, while emphasis on external causal attributions increased positive evaluations (Table 5). Crisis responsibility and causal antecedents were of more predictive importance than crisis response strategies. Thus, H2 was only partially accepted since conditional rebuild strategies did not improve, but worsen crisis response evaluations, whereas denial and defensive strategies yielded no effect. With the exception of high consensus information, all attribution variables had significant effects in the expected direction.

Media frames of NPO/IGO crises (RQ3)

To identify media frames, we conducted a two-step cluster analysis employing the BIC/Schwarz Criterion. Besides producing a statistically satisfactory cluster solution, the objective was including theoretically relevant indicators for each of the four frame elements (Entman, 1993) as well as achieving a meaningful and interpretable frame structure. This tradeoff between statistical robustness and theoretical and/or qualitative assessment has been practiced and addressed in previous framing research using an inductive approach to frame detection (Matthes and Kohring, 2008).

For the frame element of causal interpretation, we entered variables related to causal attribution and crisis responsibility. Causal antecedents (consensus, distinctiveness, consistency) were removed since their inclusion did not produce a robust cluster solution. However, we conducted a post-hoc analysis (ANOVA) to assess whether the use of covariation information differed between clusters. For moral evaluation, crisis response strategies as well as media evaluations of crisis communication and reputation were included (PCA factor scores). Treatment recommendations were only scarcely mentioned in the media coverage in general. Only for recommendations concerning internal action or changes within the NPO/IGO in a crisis context, we found a substantial frequency (13.6%) and included it in the cluster analysis. For problem definition, we included variables that measure the consequences and the impact of the crises as described by the media. Out of 20 indicators, 16 were included in a PCA. Four indicators were removed because of low frequencies and/or low communalities. The resulting factor solution was satisfactory (KMO = 0.63; Bartlett’ test p < 0.001; total explained variance = 53.33%) comprising six components (internal backlash and dismissals, financial impact, license to operate issues, damage of NPO/IGO reputation and sector reputation, stakeholder damage, government issues). Because of cross-loadings of items on several components, the sixth component (government issues) was not included in the cluster analysis. Other variables were not included because of low predictor importance for building clusters and/or because no satisfactory and robust cluster solution was produced (silhouette coefficients ≤0).

The best cluster solution containing indicators for all four frame elements according to Entman (1993) comprised 13 variables and resulted in an acceptable cluster structure (silhouette coefficient = 0.3) of three clusters, which henceforth were interpreted as media frames (Table 6). A follow-up one-way ANOVA showed that most clustering variables except for license to operate issues, stakeholder damage and evaluation of the NPO’s/IGO’s reputation contributed significantly to separate clusters. The most important variables (large effects with η² > 0.14) were internal backlash and dismissals, causal attribution, crisis responsibility, conditional rebuild strategies, justified denial strategies, evaluation of the NPOs’/IGOs’ crisis response and organizational treatment recommendation. Hence, indicators for all four frame elements were statistically crucial for detecting and describing media frames of NPO/IGO crises.

First, the critique frame, was identified in 142 articles (20.5%). This frame focuses on describing the organization and its actions negatively. For example, the frame defines the problem as internal including employee backlash and/or dismissals as reactions to management compared to the other two emergent frames. Articles using the critique frame directly blame the organization for the crisis. Moreover, in this frame, there is little the NPO/IGO does that is positive with both crisis response and reputation evaluated more negatively than in the remaining two frames. Interestingly, accommodative strategies like the conditional rebuild strategies were quoted significantly above average in this cluster. This was also the only frame where substantial treatment recommendations were included, stressing the need for internal change within the organization.

Second, the damage frame, (N = 253; 36.6%) highlights the outcomes of the crisis, emphasizing financial and reputational damages to the focal NPO/IGO and/or the whole sector. While the crisis causes were mainly internally attributed, the internal focus was less severe than in the critique frame and the organizations’ crisis responses were slightly more positively evaluated. The primary response strategies quoted were justified denial strategies. Treatment recommendations were seldom offered.

Third, the victim frame, was the most frequent with 296 articles (42.8%) and focused on the NPO/IGO as a victim of circumstance or external interference. Therefore, few negative crisis impacts were discussed. The causes of the crisis were mainly attributed to external factors, and NPOs/IGOs were rarely blamed. Defensive response strategies were most often quoted with NPOs/IGOs blaming or criticizing external parties while presenting themselves as victims. Media evaluations of the NPO/IGO were typically positive compared to the other frames and treatment recommendations were not relevant.

We conducted follow-up analyses with causal antecedents to test whether they appear more likely in one of the three frames. We found no significant difference for high consensus information, F(2, 688) = 2.23, p = 0.11. However, information on low distinctiveness was more emphasized in the critique frame (M = 0.24, SD = 0.43) than in the damage frame (M = 0.09, 0.28) or victim frame (M = 0.02, SD = 0.13) (F(2, 688) = 32.15, p < 0.001, η² = 0.08). Similarly, high consistency information was more likely to be mentioned in the critique frame (M = 0.37, SD = 0.49) than in the damage frame (M = 0.13, SD = 0.33) or the victim frame (M = 0.11, SD = 0.32) (F(2, 688) = 27.29, p < 0.001, η² = 0.07).

To gain more qualitative insights into crisis types in each frame, we conducted a follow-up analysis using chi-square test. The results showed a significant pattern of crisis types within frames with a large effect size, χ²(18) = 302.76, p < 0.001, Cramer’s V = 0.47. Most of the reported cases on sexual violence towards persons in need of support (e.g. the Oxfam scandal) were found in the critique frame. Cases of fraud and corruption as well as legal misconduct mostly pertained to the damage frame. Attacks on the NPO/IGO or terrorism and war-related crises, were mostly part of the victim frame in addition to other external crises (Figure 4).

The comparison of how the three frames are distributed across countries showed that in most cases each country had its specific pattern of frame proportions with significant differences between countries, χ²(10) = 82.96, p < 0.001, Cramer’s V = 0.24 (Figure 5). In terms of the critique frame, we found very low shares of the frame in India and Canada whereas the remaining countries referred to it in between one-fifth (Germany) up to approximately one-third (Switzerland). Damage frames were most prevalent in Canada, the US and with lower proportions in the UK and India. The victim frame was dominant in Germany’s and India’s media, but also common in Switzerland. The US had the lowest share of that frame.

Subsequent pairwise comparisons between countries using chi-square tests revealed a significant difference between all country pairs, except Germany and Switzerland (Table 7). Comparing the effect sizes, the largest differences were found between India and Switzerland. Medium effect sizes also pointed to substantial differences between further seven country pairs, especially between Switzerland and Canada as well as the US and India. Only small differences were found for the UK compared to Switzerland, Germany, Canada and the US, for example.

Discussion and conclusions

The purpose of this study was to comprehensively analyze the role of the news media in the public framing of organizational crises in the third sector in six countries. Specifically, we aimed to identify how the news media attribute causes and responsibility for crises to NPOs (RQ1) and how these attributions affect the journalists’ evaluation of crisis response and reputation (RQ2). Finally, we sought to detect recurring patterns of crisis reporting (i.e. frames) about the third sector based on content elements known to influence reputation in crisis contexts (RQ3).

The results showed that most mediated NPO cases were internal crisis types such as fraud, corruption and sexual violence. Both causal attributions and covariation information had a significant impact on crisis responsibility, with internal causes increasing responsibility and external causes decreasing responsibility (RQ1). The media’s evaluation of an NPO’s reputation as positive was only weakly influenced by internal causal attributions and the quoting of conditional rebuild strategies. The media’s assessment of an NPO’s crisis response as negative was moderately influenced by the citation of conditional rebuild strategies as well as internal crisis responsibility along with covariation information (RQ2). Cluster analysis revealed three stable frames of NPO crises in the six countries’ news media. The critique frame (20.5%) mainly attributed crises internally, stressed the need for action/change and evaluated NPOs negatively, although conditional rebuild strategies were most frequently quoted. The damage frame (36.6%) emphasized negative outcomes of crises for NPOs and was more likely to attribute causes internally, with slightly more positive evaluations along with justified denial strategies cited by journalists. The victim frame (42.8%) was the most frequent and attributed NPO crises externally, with more positive evaluations and defensive response strategies more often cited. The comparison of frames between countries revealed that German and Indian news media used the victim frame most frequently and, thus, reported less negatively about NPO crises. The US and the UK media were the most critical in terms of NPO reputation since the critique and the damage frames clearly dominated. The following discussion addresses the implications of these findings for crisis communication research and practice in the third sector. Although we did not examine stakeholder perceptions, our findings on media framing of NPO crises can be used to discuss potential perceptual effects based on extensive previous crisis communication research.

Public crisis framing depends on the salience and composition of information disclosed by the media (van der Meer and Verhoeven, 2013). During crises, investigative journalism can amplify an issue’s salience ensuring sustained attention and pressure that compel institutions to change (Langer and Gruber, 2021). Within the context of mediated third sector crises and comparable IGO crises, our findings show that journalists’ choices regarding storytelling follow clear patterns. When crises are defined by the journalists as internally caused (i.e. the organization is to blame), for example, then the narrative focuses increasingly on damage and calls for institutional reform (critique frame) as opposed to narratives where NPOs are the victims of externally caused crises (victim frame).

What is noteworthy about the emergent frames is that they point to a potential public inoculation either for or against the NPO/IGO salient in a frame. Inoculation theory is designed to reinforce positive attitudes towards organizations, thereby protecting the organization from criticism during crises (Wan and Pfau, 2004). However, journalists positioned organizational responses to crises with a ready-made interpretation of the NPO/IGO responses. For example, in the critique frame where the organization was framed negatively, readers were potentially primed to reject pro-social response strategies from the NPOs/IGOs (e.g. apologies) because the statements were included and then evaluated negatively. Within the damage frame, the interpretation was left more open with a slightly less negative evaluation of the organization; however, readers were offered a narrative of organizations trying to justify their denial of responsibility or the severity of the crisis. While the interpretation of the response was less proscribed, the focus on the damage resulting from the internally caused crisis may subtly direct the reader to focus on damage (i.e. a negative evaluation). In contrast, the victim frame represents the narrative that the organization should not be held accountable for the crisis. In many cases, defensive crisis response strategies have become negatively evaluated by stakeholders in corporate contexts (Rhee and Yang, 2014; Einwiller and Steilen, 2015); however, these findings suggest that readers are potentially being inoculated to accept third sector defensive strategies that shift the blame or criticize external actors when they are framed as the crisis victim.

For crisis communication practitioners in the third sector, at least two relevant implications result from these findings. First, in cases of misconduct (internal crises), the use of standard apologies is not enough to facilitate more positive evaluations by the news media. More expressions of empathy for victims of a crisis (e.g. victims of sexual abuse) and more efforts of organizational renewal as part of the post-crisis discourse of NPOs may yield better results (Seeger et al., 2005). Second, when NPOs are victims of a crisis (e.g. attacks by terrorists or authoritarian governments), defensive crisis response strategies (e.g. attacking accusers, blaming perpetrators) may actually yield positive effects in the news media discourse. This contradicts the common wisdom of crisis communication research on business organizations and may represent a crucial difference compared to the specific context of NPO crises.

Regarding the main propositions of SCCT, with attribution theory being the main core, we confirmed the significant relationship between internal attributions of causes/responsibility and reputational evaluations of NPOs/IGOs within crisis news. Journalists seem to follow similar patterns as crisis stakeholders that were typically studied in experimental SCCT research (Ma and Zhan, 2016). However, the news media adapt these attributions more significantly to evaluations of crisis response than to general reputational evaluations of organizations. While this may in part result from journalistic norms of impartial reporting, evaluations of an organization’s crisis response can be incorporated in future experimental research in addition to other reputations measures that have been used differently in SCCT (Ma and Zhan, 2016). In terms of Kelly’s covariation principle, our findings confirmed that low distinctiveness and high consistency are aligned with internal responsibility attributions (Schwarz, 2012). However, the findings also suggest that consensus had inverse effects than expected. High consensus was correlated with internal attribution instead of external attribution; thus, suggesting that when an NPO/IGO experiences internally attributed crises, pointing to other organizations experiencing similar crises can result in more negative framing in a mediated context instead of being cited as a mitigating circumstance. It may even negatively affect the third sector as a whole. It was interesting to see that causal antecedents and attributions were more present and more strongly related to the media’s evaluation of NPO/IGO crises than the crisis response strategies known from IRT and SCCT. This should be taken into account in an NPO’s strategic crisis response by developing more information-based or attribution-based strategies (Schwarz, 2012; Zerfass et al., 2013) as well as strategies with more emphasis on empathy and commitment to change and renewal (Seeger et al., 2005).

The second SCCT proposition is that matching certain crisis response strategies to specific crisis clusters or perceived crisis responsibility yields better effects of protecting reputation than mismatching strategies. Past experimental research has produced mixed results in this regard with crisis response strategies having relatively weak effects on reputation (Ma and Zhan, 2016). Within the media reality of NPO/IGO crises, journalists seem to follow a different rationale than other crisis stakeholders. For example, SCCT would recommend that when the organization can be held accountable for the crisis (i.e. internal crises like harassment or sexual abuse), then apology and accommodative responses are the organization’s best response (Coombs, 1995, 2019; Coombs and Holladay, 2004). Yet, journalists did not find these strategies persuasive and cited them in a context of negative NPO/IGO evaluations. This may also be indicative of moral disillusionment effects in the specific context of NPO transgressions, which journalists judge more harshly in comparison to corporate scandals (Chapman et al., 2022b), regardless of the “right” or “wrong” response strategy.

Similarly, both IRT and SCCT would caution organizations about using defensive crisis response strategies (scapegoat, attacking accusers); yet in the victim frame journalists cited these strategies in a context of more positive reputational evaluations. Given the finding that the crises in the victim frame are primarily attacks on NPOs/IGOs (e.g. friendly fire, terrorism, kidnapping), responses such as attacking accusers or blaming external parties actually seem reasonable. Another mismatch in SCCT terms was found for the damage frame, where NPO crises were mainly categorized as internal/preventable crises (e.g. fraud, corruption) and were often cited with denial strategies; yet, journalists tend to evaluate NPOs/IGOs slightly positively compared to the critique frame. These findings point to the importance of more NPO-specific experimental research that analyses the impact of such scenarios and response strategies on NPO-specific crisis stakeholders (e.g. volunteers, donors).

Our data also suggest that cultural factors possibly influence media portrayals of NPOs/IGOs in crisis. There was substantial variation in framing patterns between countries with only Germany and German-speaking Switzerland being similar, which would make sense regarding the cultural proximity between the countries. Moreover, tone of coverage from negative to positive also varied substantially by country with the US coverage being the most negative and Germany being the most positive. Future research should explore media systems, national cultures, or different national levels of public trust in the third sector as possible explanations for these differences. On that note, another interesting finding was that in most cases, crisis coverage of the third sector centered in the country where the organization in crisis was headquartered suggesting limited transnational overspill. This would point to the importance of using decentralized and localized, but transnationally coordinated, PR models in international NPOs, which in many cases still seem to adhere to centralized or uncoordinated communication strategies (Schwarz and Fritsch, 2014).

These data demonstrate the importance of good media relations practice. Given the findings contradict traditional crisis communication thinking about response strategies and finding media coverage inoculating for or against an NPO/IGO crisis response strategy, our results support previous research suggesting that the news mediates crisis response (Kleinnijenhuis et al., 2015). In view of the strength and direction of the critique, damage and victim frames found, this study also shows that risks posed to NPOs’/IGOs’ reputation by media coverage is not constant; rather it is contingent on the crisis types as well as attributions of cause and responsibility. This supports previous research indicating that media influence on reputation was situationally dependent (Einwiller et al., 2010).

In terms of theory building, this study demonstrates the importance of combining a framework like Entman’s (1993) framing analysis with subject-specific theory to create a robust and theoretically grounded analysis of media framing. Our framework of crisis framing based on crisis communication theory and attribution theory proved to be useful for studying mediated NPO crises diachronically and comparatively. Such a framework can also serve as an approach to bridging frame building (NPOs’ strategic crisis communication), frame content (media frames of NPO crises) and framing effects (frame-based perceptions of crisis stakeholders). Framing in general can be a fruitful integrative approach to study the process of strategic communication (Verhoeven, 2016), but still requires domain-specific development for crisis communication in the third sector.

Limitations

Although the applied search string was rather complex and research-based, some NPO crises may not be captured. Some of the PCA had “mediocre” KMO values and a total explained variance below 50% and have to be treated with caution. However, factor loadings, communalities and the determinant (>0.0001) of the correlation matrix indicated a robust PCA. In general, although the news media are known to be powerful and credible sources that constitute and/or damage organizational reputation with the general public in crisis contexts, social media are increasingly used by NPOs to engage in two-way crisis communication with more specific stakeholder groups such as donors or volunteers. Future research should therefore study the interplay between news media and social media when NPOs respond to crisis situations.

Figures

Figure 1

Share of articles on NPO/IGO crises in the population after screening compared to the final sample

Figure 2

Frequencies of internal crisis responsibility by crisis type

Figure 3

Frequency of quoted crisis response strategies in the media (in %)

Figure 4

Distribution of mediated NPO/IGO crisis types across frames of (in %)

Figure 5

Distribution of mediated NPO/IGO crisis frames across countries (in %)

Table 1

Selected media outlets

Newspaper	Country	Style and political leaning
USA Today	USA	Tabloid, centrist
New York Times		Quality, liberal
Washington Post		Quality, liberal
The Guardian	UK	Quality, liberal-left
Daily Mail		Tabloid, conservative
Daily Mirror		Tabloid, left
Toronto Star	Canada	Quality, liberal
The Globe and Mail		Quality, conservative
National Post		Quality, conservative
The Telegraph	India	Quality, liberal
Times of India		Quality, conservative
Hindustan Times		Quality, conservative
Bild	Germany	Tabloid, conservative
taz		Quality, left
Die Welt		Quality, conservative
Neue Zürcher Zeitung	Switzerland	Quality, liberal-conservative
Tages-Anzeiger		Quality, liberal-left
Blick		Tabloid, liberal-left

Source(s): Table by authors

Table 2

Codebook categories (selection of main indicators)

Frame element/formal	Main categories	Reliability (Gwet’s AC1)
Problem definition	Affected organizationa Field of activitya Crisis typea Consequences for the NPO/IGO, stakeholders (e.g. funding cuts, decreasing donations, volunteers, dismissal of management, etc.)b	0.88
Causal interpretation	Covariation information (consensus, distinctiveness, consistency)c Causes (internal, external)b	0.93
Moral evaluation	Responsibility attributions (sender, object, receiver) Crisis response strategies (e.g. attack the accuser, denial, external scapegoat, justification, excuse, apology)b Evaluation of crisis communication/management (consistency, timeliness, transparency, operative crisis management)d Reputational evaluations (actions, competence, efficiency, social responsibility, attractiveness)d	0.89
Treatment recommendation	Sender of treatment recommendation (e.g. media, NPO/IGO, civil society actors, political actors, etc.) Type of treatment recommendation (organizational, political/government, legal, media reporting)b	0.94

Note(s): Scales: ^abased on extensive research-based lists of options, ^bmentioned/not mentioned, ^chigh/low/not mentioned, ^dpositive/ambivalent/negative

Source(s): Table by authors

Table 3

Binomial logistic regression with internal crisis responsibility as dependent variable

Predictor	B	SE	Wald	df	p	Odds ratio
Difference external - internal causal attribution	−1.457	0.130	124.892	1	<0.001	0.233
Consensus (high)	1.118	0.232	23.133	1	<0.001	3.058
Consistency (high)	0.765	0.280	7.496	1	0.006	2.150
Distinctiveness (low)	1.695	0.482	12.390	1	<0.001	5.447
Constant	−0.765	0.113	45.603	1	<0.001	0.465

Note(s): χ²(4) = 245.5, p < 0.001; Nagelkerke’s R² = 0.40

Source(s): Table by authors

Table 4

Principal component analysis (PCA) with crisis response strategies

	Component
	Conditional rebuild	Defensive	Justified denial
Scapegoat internal	0.685
Apology	0.669	−0.108	0.114
Corrective action	0.652		−0.139
Regret	0.600
Victimage		0.749	0.154
Scapegoat external		0.720	−0.245
Attack the accuser		0.680	0.295
Denial			0.738
Justification			0.667
Reminder	0.127	0.145	0.494

Note(s): Extraction method: Principal component analysis; varimax rotation with Kaiser normalization; rotation converged in 4 iterations; KMO = 0.63; χ² (45) = 510.0, p < 0.001

Source(s): Table by authors

Table 5

Multiple linear regression for evaluation of NPO crisis response

	Unstandardized coefficients		Standardized coefficients	t	Sig
	B	Std. Error	Beta
(Constant)	0.246	0.051		4.815	<0.001
Crisis responsibility (internal)a	−0.236	0.081	−0.118	−2.908	0.004
Conditional rebuildb	−0.108	0.037	−0.108	−2.921	0.004
Defensiveb	0.044	0.034	0.044	1.297	0.195
Justified denialb	−0.065	0.036	−0.065	−1.824	0.069
Difference external - internal causal attributionc	0.133	0.051	0.107	2.611	0.009
Consensus (high)a	0.074	0.083	0.031	0.896	0.370
Consistency (high)a	−0.477	0.096	−0.180	−4.959	<0.001
Distinctiveness (low)a	−0.724	0.132	−0.206	−5.472	<0.001

Note(s): R²_adj. = 0.24; F(8, 682) = 27.98, p < 0.001; Durbin–Watson = 2.00; ^adummy variables with 1 being the condition in brackets ^bfactor scores (regression); ^cscale: +1 (only external causes) to −1 (only internal causes)

Source(s): Table by authors

Table 6

Two-step cluster analysis media framing (BIC)

Frame element	Variable		Cluster (frames)
			1 Critique	2 Damage	3 Victim	Combined	η2	pa
Problem definition	Impact: Internal backlash and dismissals	M	0.81	−0.06	−0.34	0.00	0.18	<0.001
		SD	1.35	0.98	0.48	1.00
	Impact: Financial damage	M	−0.04	0.22	−0.17	0.00	0.03	<0.001
		SD	0.80	1.44	0.43	1.00
	Impact: License to operate	M	−0.14	0.07	0.00	0.00	0.01	0.12
		SD	0.79	0.88	1.17	1.00
	Impact: Reputation damage of the NPO/IGO and the whole sector	M	0.08	0.33	−0.32	0.00	0.08	<0.001
		SD	1.27	1.18	0.44	1.00
	Impact: Stakeholder damage	M	−0.17	0.09	0.01	0.00	0.01	0.05
		SD	1.05	1.16	0.80	1.00
Causal interpretation	Difference external (+1)/internal (−1) causal attribution	M	−0.73	−0.33	0.50	−0.06	0.39	<0.001
		SD	0.55	0.72	0.58	0.80
Moral evaluation	Crisis responsibility (internal)	M	0.84	0.71	0.06	0.46	0.49	<0.001
		SD	0.37	0.45	0.24	0.50
	Conditional rebuild strategies	M	1.05	−0.27	−0.27	0.00	0.28	<0.001
		SD	1.67	0.46	0.39	1.00
	Defensive strategies	M	−0.12	−0.11	0.15	0.00	0.02	<0.001
		SD	0.84	0.85	1.16	1.00
	Justified denial strategies	M	0.08	0.48	−0.44	0.00	0.17	<0.001
		SD	0.95	1.20	0.53	1.00
	Evaluation of the NPO’s/IGO’s crisis communication	M	−0.98	0.18	0.32	0.00	0.25	<0.001
		SD	1.78	0.42	0.29	1.00
	Evaluation of the NPO’s/IGO’s reputation	M	−0.11	−0.05	0.10	0.00	0.01	0.07
		SD	1.59	0.71	0.83	1.00
Treatment recommendation	Organizational/internal change	M	0.61	0.02	0.01	0.14	0.48	<0.001
		SD	0.49	0.12	0.12	0.34

Note(s): Silhouette coefficient = 0.3; ^aone-way ANOVA with three-cluster solution as factor

Source(s): Table by authors

Table 7

Pairwise country comparisons of frame distribution

	Germany	UK	Switzerland	Canada	USA	India
Germany		0.20**	n.s.	0.31**	0.34**	0.34**
UK			0.19*	0.16*	0.14*	0.31**
Switzerland				0.42**	0.30**	0.50**
Canada					0.21*	0.26*
USA						0.43**

Note(s): All coefficients are Cramer’s V effect sizes using chi-square tests with cluster solution (3 frames) and two countries of comparison as variables; *p < 0.05; **p < 0.001; effect sizes according to Cohen (1988) when df = 1: small 0.10 < 0.30, medium 0.30 < 0.50, large 0.50 or more

Source(s): Table by authors

Funding details: No external funding was received for this study.

Disclosure statement: The authors report there are no competing interests to declare.

Data availability statement: The data that support the findings of this study are available from the corresponding author upon reasonable request.