The Public’s Personality: Towards Tailored Crisis Communication
An Experimental Investigation into the Moderating Role of Personality on
Secondary Crisis Communication Behaviour
Master’s Thesis
Graduate School of Communication
Master’s programme: Corporate Communication
Cyril Schaub 12352713 Supervised by dr. Toni van der Meer
Abstract
Communicating effectively during crisis is crucial for organisational survival in disruptive environments. By knowing who their stakeholders are in terms of personality, organisations could tailor their communication selectively, conforming with individual stakeholders’ affective, behavioural and cognitive preferences.
To explore the moderating influences of personality traits on effects of crisis communication, this study conducted a 2 (crisis response strategy: diminish vs. rebuild) times 2 (organisational type: profit oriented vs. non-profit oriented) between-subjects factorial experiment among N = 191 participants. To examine the influence of the Big Five as well as the Dark Triad personality traits on aspects of affective, behavioural, and cognitive crisis communication outcomes, anger was included as a mediating factor in explaining secondary crisis communication intention and content of participants. Main effects of independent variables on secondary crisis communication intention were not found, crisis response strategy however influenced the content of post-crisis communication, while organisational type did not have an influence. The effect of crisis response strategy was mediated by anger, which was strongly related to all outcome measurements, confirming previous crisis communication research. The Big Five personality traits did not show any moderating influence, while the Dark Triad personality traits all indicated to be weakly related to main effects on secondary crisis communication intention and content.
The absence of influence of organisational type as well as the promising results of malevolent personality traits are discussed, and implications warranting further examinations especially of malevolent personality traits in crisis communication research are explained.
Introduction
In March 2018, Facebook published an official statement that it would not allow third party data analytics firm Cambridge Analytica on their platforms anymore (Grewal, 2018). The data analytics firm was able to estimate voters’ personalities through data gathered by questionable means from Facebook and offered their services to candidates of the 2016 presidential election in the USA (Collins & Dance, 2018). By knowing their target audience, Cambridge Analytica was able to communicate selectively and influence political behaviour effectively (Confessore & Hakim, 2017).
Just like political marketing, crisis communication is concerned with communicating effectively and influencing attitudes and behaviours during a crisis. Attribution Theory, as one of the foundations of crisis communication research, posits that individuals will attribute responsibility for crises depending on their affective, behavioural and cognitive processes, which is equivalent to personality at the aggregate level (McCrae & Costa, 1987, 1995; Weiner, 1985, 2006, 2010; Wilt & Revelle, 2015). Reactions to a crisis are then the consequence of those attributions made. In understanding crisis outcomes, research has been concerned with exploring various organisational aspects and their consequences on post-crisis attitude and behaviour (e.g., Billings, Milburn, & Schaalman, 1980; Brummette & Fussell Sisco, 2015; Coombs & Holladay, 1996, 2008; Feinglass, 2005).
But what if, disregarding ethical considerations for the moment, organisations would be aware of their stakeholders’ personalities and thus be able to tailor their communication to stakeholders’ personalities? Although psychology clearly informs us that affect, behaviour and cognition as they influence post-crisis behaviour are dependent on individual personality, crisis communication research has mostly been concerned with an organisational centred perspective. That is, most examinations explored organisational behaviour and how it affects crisis outcomes, neglecting characteristics of the audience of crisis communication. Only recently, scholars have also started to take into account public based factors such as what emotions stakeholders experience during a crisis (e.g., Jin, Pang, & Cameron, 2007, 2012).
An audience centred approach of how characteristics of the public influence crisis outcomes does not exist in crisis communication research yet. This examination will overcome this gap by studying how the effects of crisis response strategies and organisational context on crisis communication outcomes are dependent on personality of the public. To do so, Coombs’ Situational Crisis Communication Theory (SCCT) (2007) will serve as a framework to integrate psychological understanding of personality in examining crisis communication outcomes. To
investigate the role of personality, an online experiment was conducted (N = 191), testing the influence of personality traits on the individual level on the effects of crisis response strategy (diminish vs. rebuild) and organisational type (profit vs. non-profit organisation) on secondary crisis communication outcomes. Since behaviour and cognition are strongly interrelated with affect (e.g., Weiner, 2010; Winter & Kuiper, 1997), the mediating role of anger was considered in linking personality and secondary crisis communication behaviour.
Theoretical Framework Crisis as a Reputational Threat
Crises are considered disruptive and unexpected events, posing a threat to organisational reputation through re-evaluations by the public of organisational behaviour (Coombs, 2007; Wartick, 1992). Stakeholders will re-evaluate their expectations of an organisation and its behaviour based on information they receive. Although organisations also communicate directly with the public, most stakeholders will learn about a crisis through the media, or through word-of-mouth (Coombs, 2007). Because a crisis is by definition disruptive, those re-evaluations during a crisis can lead to a loss of reputation and have unfavourable outcomes such as loss of legitimacy (Carmeli & Tishler, 2005).
Consequently, information about a crisis plays a crucial role in how big the reputational threat is. Apart from the content, the way in which this information is framed plays role as well. The concept of framing describes how information can be viewed from different perspectives and thus have different implications (Chong & Druckman, 2007). For an organisation in crisis this means that the possible threat a crisis poses to reputation is dependent on what information is spread but also on how it is framed (Coombs, 2007).
Crisis Communication
Crisis communication, after considering ethical priorities such as preventing further possible harm, is concerned with protecting and/or shielding an organisation’s reputation during a crisis. Key factors in assessing reputational threat are crisis severity and responsibility for it (Benoit, 1997). However, depending on how information about a crisis and the organisation is framed, both severity of and responsibility for the crisis can change.
Situational Crisis Communication Theory (2007) provides crisis response strategies to mitigate negative re-evaluations of the organisation and prevent reputational losses. Next to
prior reputation and crisis history, SCCT considers an organisation’s crisis responsibility in assessing the reputational threat, considering that it is dependent how a crisis is being framed. In doing so, SCCT recognises the possible influence of the media or of word-of-mouth on the reputational threat a crisis poses (Coombs, 2007).
SCCT groups crises into three clusters depending on the amount of attributed responsibility to an organisation and provides three corresponding crisis response strategies. The purpose of crisis communication strategies is to protect reputation by shaping components of the crisis: attributions of the crisis (what the crisis is) and the perception of the organisation in crisis (how much responsibility is attributed to the organisation) to mitigate negative outcomes (Benoit, 1997; Coombs, 2007).
Secondary crisis communication. A strength of SCCT is its understanding of crises
within a communicational context and how framing influences perceptions of a crisis. Depending on this, reputation of the organisation may change, which in turn can lead to behavioural intentions such as negative word of mouth (Coombs, 2007). However, SCCT does not consider that such communication by the media or stakeholders, referred to as secondary crisis communication (SCC), is outside the organisation’s reach and can thus apply different communicational frames. Actors such as the media or stakeholders might further change attributions and perceptions of the crisis through their communication, which in turn can affect organisational reputation in a negative way (Tucker & Melewar, 2005).
Accordingly, since Coombs introduced SCCT (2007), various scholars have recognised the importance of examining not only organisational reputation as outcome of crisis communication, but also SCC behaviours to understand how post-crisis communication is affected by organisational crisis behaviour. For instance, Jin and Liu (2010) and Liu, Austin and Jin (2011) introduced the Social-mediated Crisis Communication Model (SMCCM), which integrates offline and online communication among the public into organisational crisis communication. Further, Schultz, Utz and Göritz (2011) also measured the effect of crisis responses on individuals’ secondary crisis communication and reactions, incorporating a behavioural component into crisis outcomes as well.
To further explore how organisational behaviour may affect SCC, this study considers two aspects of secondary crisis communication behaviour as relevant outcomes. Specifically, following Liu et al. (2011) and Schultz et al. (2011), individuals’ intention to talk about a crisis will first be considered. Second, because it does not only matter if individuals communicate about a crisis but also how, the content of secondary crisis communication will be measured. In
an innovative approach, individuals will be asked to provide messages about a crisis and hence their attitude towards an organisation. SCC content can therefore also be considered a proxy for organisational reputation. The assumed relations between crisis response strategies and organisational context as independent variables and SCC behaviour as outcome with personality and emotion as moderating and mediating factors are conceptualised in Figure 1. Following, the theoretical assumptions will be described in detail.
Figure 1: Conceptual model of crisis response, organisational type and their effect on SCC behaviour. Personality as moderating, anger as mediating factors.
Note. c = direct effect, c’ = total effect. N = Neuroticism, E = Extraversion,
A = Agreeableness, C = Conscientiousness, O = Openness, MACH = Machiavellianism, PSY = Psychopathy, NAR = Narcissism
Crisis type and crisis response strategies. The biggest threat for organisational
reputation poses a preventable crisis because a high amount of responsibility is attributed to the organisation, as assumed by SCCT (Coombs, 2007). Preventable crises are characterised by the organisation willingly neglecting regulations and/or taking inappropriate risks (Coombs, 2007). Appropriate crisis responses are even more important during a preventable crisis because the (reputational) stakes are higher compared to crises with lower responsibility. Thus, preventable crises have often been studied in crisis communication research and have been linked to the most negative effect on organisational reputation (Claeys et al., 2010).
During a preventable crisis is a rebuild strategy that accepts responsibility, followed by a diminish strategy that downplays crisis responsibility are the best responses (Coombs, 2007).
Rebuild strategies consist of an apology and/or offers of compensations to the victims of the crisis, while diminish strategies consist of justifications for organisational misbehaviour and/or dispute of organisational control over the event (Coombs, 2006). Rebuild strategies have been consistently linked to the most favourable outcomes during preventable (e.g., Claeys & Cauberghe, 2014; Claeys et al., 2010; Fussell Sisco, 2012) with Schultz et al. (2011) also linking rebuild strategies to the most favourable SCC outcomes. In trying to replicate those findings, Hypotheses 1a and 1b assume that a rebuild strategy during a preventable crisis will lead to more favourable SCC behaviour compared to a diminish strategy.
H1a. A rebuild strategy causes less intention to engage in secondary crisis communication than
a diminish strategy. (path c1, see Figure 1)
H1b. A rebuild strategy causes less negative content in secondary crisis communication than a
diminish strategy. (path c2)
Organisational context in crisis. Coombs stated that SCCT provides a comprehensible
framework for a “variety of organizational forms” (2007, p. 164), specifically refraining from the word corporation because it might imply understandings of profit-oriented (PO) organisations. Indeed, research has shown that SCCT can be applied within the context of non-profit organisations (NPOs) as well (Fussell Sisco, 2012).
For NPOs however, crisis communication gains in importance. NPOs rely heavily on their relations with the public because donations are a big factor of an NPO’s operations. Reputational damages therefore represent an existential threat to NPOs (Feinglass, 2005; Fussell Sisco, 2012; Fussell Sisco, Collins, & Zoch, 2010). Successful shielding of reputation during a crisis becomes thus a necessity.
Research suggests that reputation consist of two dimensions: First, a corporate ability (CA) aspect which is linked to the organisation’s capability of delivering products and/or services. Second, a corporate social responsibility (CSR) dimension which is related to the organisation’s behaviour and character concerning societal issues (Brown & Dacin, 1997). While both, POs and NPOs seem capable of maintaining favourable CA relations with their public through meeting expectations of quality of products, NPOs are more easily associated with favourable CSR relations than POs. Solely through operating, a NPO strives to contribute to societal welfare, establishing CSR relations with the public (Sohn & Lariscy, 2014). Hence, an NPO is more easily able to establish favourable pre-crisis relations with its stakeholders than a PO through its nature of business.
Next to responsibility for a crisis, SCCT considers reputational capital which refers to the stock of positive relations an organisation has assembled with its stakeholders over time (Fombrun & van Riel, 2004). Reputational capital can function as a buffer during an organisational crisis, because an organisation with favourable pre-crisis reputation has more reputational capital to spend compared to organisations with unfavourable or neutral pre-crisis reputation. Hence, an organisation with wit higher reputational capital will emerge from a crisis with stronger post-crisis reputation (Coombs, 2007).
Consequently, all else being equal an NPO should have better post crisis reputation and hence more favourable SCC outcomes than a PO because of its advantage in CSR reputation. So far, a direct comparison between POs and NPOs in crisis communication outcomes does not exist. This examination will fill this gap by comparing a PO’s and a NPO’s post-crisis outcomes in experimental conditions. Because of the notion of reputational capital and an NPO being more easily able to establish it through strong CSR relations, an NPO’s crisis response will lead to more favourable SCC outcomes compared to a PO’s crisis response strategy.
H2a. A NPO’s crisis response causes less intention to engage in SCC than a PO’s crisis
response. (path c3)
H2b. A NPO’s crisis response causes less negative content in SCC than a PO’s crisis response.
(path c4)
The Public’s Side
Understandably, crisis communication research has mostly been concerned with organisational behaviour in crisis. In doing so, crisis communication research has granted the organisation agency over its audience in communicative processes, neglecting characteristics of the public to a large extent.
SCCT as one of the most influential paradigms in the field is based on Attribution Theory, which posits that individuals search for causal relations for events and attribute responsibility to actors or circumstances subsequently (Coombs, 2007; Weiner, 1985, 2006, 2010). Those attributions then build the foundation for behavioural and affective responses.
The understanding of such processes is subject to personality psychology which “has long been concerned with identifying the fundamental ways in which people differ from each other” (Wilt & Revelle, 2015, p. 478). Specifically, personality research is concerned with describing coherent patterns on individual level of affective (“conditions such as moods, emotions, feelings, feeling-like states and preferences” (Wilt & Revelle, 2015, p. 479)),
behavioural (“verbal utterances (…) or movements that are potentially available to careful observers using normal sensory processes” (Furr, 2009, p. 372)) and cognitive (e.g. attention, memory, knowledge, beliefs, appraisals, interpretations, expectations (Wilt & Revelle, 2015)) processes. Returning to Attribution Theory as bases for SCCT, it only seems logic to consider processes of affect, behaviour and cognition (ABC) on individual level to understand outcomes of crisis communication.
Crisis communication research has indeed started to consider public based factors in studying crisis outcomes. Next to considering emotions as mediators (e.g., Schultz, Utz, & Glocka, 2012; Utz, Schultz, & Glocka, 2013; van der Meer & Jin, 2019; van der Meer & Verhoeven, 2014), there are studies that also consider the public’s side as an agent over the effect of crisis communication: Claeys et al. (2010) examine individuals’ loci of control and find a moderating impact on the effect of crisis response strategy on reputation. Choi and Lin (2009a) and Claeys and Cauberghe (2014) show that individuals’ involvement with the organisation in crisis or their products impacts crisis perceptions and finally, Wan (2008) shows evidence that resonance with the public’s cognition and affect in crisis responses has an influence on SCC outcomes.
Understandings of Personality. Despite promising results, there are no examinations
that have integrated a more wholesome understanding of personality in crisis communication despite personality psychology providing numerous taxonomies of personality that describe and summarise differences in affect, behaviour and cognition (ABC) on the individual level (Wilt & Revelle, 2015).
The Big Five. Most of the personality research field has settled on the so called Big Five
personality traits as overarching framework for understanding personality. The Big Five “represent personality at the broadest level of abstraction” (John & Srivastava, 1999, p. 109), which is also the reason for their name.
Specifically, the Big Five traits are Extraversion (E), Agreeableness (A), Conscientiousness (C), Neuroticism (N) and Openness (to experience) (O) (McCrae & Costa, 1987). Since the Big Five taxonomy has been studied for almost a century, there exist varying conceptualisations of the traits including different version of their names. For instance, Neuroticism is often named as being opposed to Emotional Stability (ES) and a substantial part of the field believes the fifth personality trait should rather be Intellect/Imagination (I) instead of Openness (John & Srivastava, 1999). The conceptualisation within this examination will further be addressed during the discussion of the applied methodology.
If as assumed by Attribution Theory, individual processes of ABC are the underlying mechanisms in post crisis behaviour, personality, as manifestation of differences in ABC, will have an effect on crisis outcomes. The question is, what personality trait will have what effect on which aspect of crisis outcome?
To answer this it is crucial to understand what personality traits actually entail (Pytlik Zillig, Hemenover, & Dienstbier, 2002; Wilt & Revelle, 2015), meaning, to what extent do the Big Five personality traits describe affect, behaviour or cognition? Imagine a person is categorised as being agreeable (i.e., scoring high in Agreeableness). To what extent will this describe which dimension of this person’s personality? That is, to what extent does agreeableness manifest itself through emotions, though behaviour or through attitude?
Since a multitude of different personality inventories exist, it would be easy to assume that different inventories will emphasize different ABC components per personality trait, shifting the problem to a question of justification and adequacy depending on interest. However, Pytlik Zillig et al. (2002) and Wilt and Revelle (2015) find clear divergence between the personality traits and astonishing convergence throughout different measurements of the Big Five concerning their components of A, B and C, suggesting that the Big Five consist different components of ABC on a conceptual level. To make assumptions about the influence of the Big Five on crisis communication, the five traits and what they actually entail will be briefly discussed in the following. A complete overview of the personality traits can also be found in Appendix A.
Affective trait of the Big Five. Neuroticism is characterised by “negative emotionality, such as feeling anxious, nervous, sad, and tense” (John & Srivastava, 1999, p. 30). Both Pytlik Zillig et al. (2002) and Wilt and Revelle (2015) find that Neuroticism throughout all inventories investigated is predominantly associated with affective aspects of personality. If a person is characterised as being high in Neuroticism, this mostly concerns how the person feels. More specifically, Neuroticism has also been linked strongly to the experience of anger in particular (Martin, Watson, & Wan, 2000).
As briefly mentioned above, crisis communication has given heightened attention to the examination of emotions as mediators on crisis outcomes as well. Especially anger, which plays a crucial role in Attribution Theory explaining negative reactions in behaviour and attitude (Weiner, 1985, 2010), has been extensively studied. Consistent findings suggest that anger mediates the effects of organisational behaviour on negative crisis outcomes (e.g., Choi & Lin, 2009b; Coombs & Holladay, 2005, 2007). To debunk those connections between crisis
response/organisational type, Neuroticism, anger and SCC behaviour, first the mediation of crisis response (H1a & H1b) and organisational type (H2a & H2b) on SCC behaviour through anger will be tested. As research has shown, anger is assumed to have a negative effect on SCC outcomes:
H3a. The influence of crisis response on SCC intention and content is mediated by anger in
such a way that anger will lead to higher intention to engage in SCC and more negative content. (paths a1 - b1/c’1 and a2 - b2/c’2)
H3b. The influence of organisational type on SCC intention and content is mediated by anger
in such a way that anger will lead to higher intention to engage in SCC and more negative content. (paths a3 - b3/c’3 and a4 - b4/c’4)
Next, because individuals high in Neuroticism are more likely to experience anger, a diminishing crisis response during a preventable crisis will lead to individuals experiencing even more anger if they are high in Neuroticism compared to those low in Neuroticism:
H4a. Neuroticism moderates the by anger mediated influence of response strategy on the
intention to engage in SCC and its content (H3a, paths a1 - b1 and a2 - b2) in such a way that individuals high in Neuroticism will report more anger and thus higher intention to engage in SCC and more negative SCC content than individuals low in Neuroticism.
Accordingly, if an organisation does have lower reputational capital (i.e., PO), a preventable crisis will lead to individuals experiencing more anger compared to an organisation with higher reputational capital (i.e., NPO), especially if individuals are high in Neuroticism:
H4b. Neuroticism moderates the by anger mediated influence of organisational type on the
intention to engage in SCC and its content (H3b, paths a3 - b3 and a4 - b4) in such a way that individuals high in Neuroticism will report more anger and thus higher intention to engage in SCC and more negative SCC content than individuals low in Neuroticism.
Behavioural traits of the Big Five. Further, Pytlik et al. (2002) and Wilt and Revelle (2015) associate Extraversion, Agreeableness and Conscientiousness mostly with behavioural personality. Agreeableness is described by pro-social behaviour, trust, mindfulness and cooperation, while Extraversion characterises individuals as sociable, affectionate, active and talkative. Finally, Conscientiousness entails controlled behaviour, goal orientation, adherence to norms and rules or planning and organising (John & Srivastava, 1999; McCrae & Costa,
1987). As some of those descriptions by John and Srivastava and McCrae and Costa respectively also indicate, all three traits are mostly describing how people behave. Thus, Extraversion, Agreeableness and Conscientiousness will have an influence on individuals’ behaviour following a crisis. As behaviour, this examination considers individuals’ intention to engage in SCC as crisis outcome.
Hence, if an organisation in a preventable crisis responds with a diminishing strategy, individuals will be more likely to talk about the crisis if they are high in Extraversion (because they are outspoken and talkative), high in Conscientiousness (because the organisation does not follow rules and norms) and low in Agreeableness (because they do not conform with the organisation’s behaviour or do not feel obliged to cooperate with it) than if an organisation follows a rebuild response strategy.
H5a. Extraversion, Conscientiousness, and Agreeableness moderate the influence of response
strategy on the intention to engage in SCC (path c1) in such a way that individuals high in Extraversion and Conscientiousness and low in Agreeableness will report higher intention to engage in SCC than individuals low in Extraversion, Conscientiousness and high in Agreeableness.
Accordingly, if a PO that has lower reputational capital suffers a crisis, individuals high in Extraversion and Conscientiousness and low in Agreeableness will be more likely to talk about the crisis than when an NPO with higher reputational capital suffers a crisis.
H5b. Extraversion, Conscientiousness, and Agreeableness moderate the influence of
organisational type on the intention to engage in SCC (path c3) in such a way that individuals high in Extraversion and Conscientiousness and low in Agreeableness will report higher intention to engage in SCC than individuals low in Extraversion, Conscientiousness and high in Agreeableness.
Cognitive trait of the Big Five. Finally, Pytlik Zillig (2002) and Wilt and Revelle (2015) identify the fifth Big Five trait Openness as mostly consisting of cognitive aspects of personality, thus relating it to processes of thinking and sense making. Openness is often associated with intelligence, with the trait also showing strong correlations with its alternative conceptualisation Intellect/Imagination (McCrae & Costa, 1987). As the description already indicates, individuals high in Openness are open to new experiences and often appear more accepting of others or their behaviour.
Because Openness manifests itself mostly through cognitive processes, individuals high in Openness will thus be less likely to change their cognition of an organisation in crisis based on limited information and will be more open to multiple perspectives (McCrae & Costa, 1987; Pytlik Zillig et al., 2002). Within this examination, Openness will have a positive influence on individuals’ SCC content as a manifestation of one’s attitude towards the organisation. Openness will mitigate the main effects of crisis response (H1b) and organisational type (H2b) on SCC content.
H6a. Openness moderates the influence of response strategy on the content of SCC (path c2)
in such a way that individuals high in Openness will report less negative SCC content than individuals low in Openness.
H6b. Openness moderates the influence of organisational type on the content of SCC (path c4)
in such a way that individuals high in Openness will report less negative SCC content than individuals low in Openness.
The Dark Triad. Although widely used, the Big Five taxonomy is not the sole option
for measuring personality. While the Big Five aspire to enable a wholesome description of personality capturing countless individual attributes within a comprehensible framework, there are other taxonomies that measure more distinct features of personality (John & Srivastava, 1999). As such, the Dark Triad, comprising of Machiavellianism, Psychopathy and Narcissism, aims to describe a person’s malevolence (Furnham, Richards, & Paulhus, 2013; Williams & Paulhus, 2002). In particular, the Dark Triad consists of overall tendencies of “emotional coldness, aggressiveness, and self-promotion” (James, Kavanagh, Jonason, Chonody, & Scrutton, 2014, p. 211) and is correlated with low Agreeableness (Jakobwitz & Egan, 2006; Williams & Paulhus, 2002).
Crisis communication has so far mostly been focused on how crises elicit negative outcomes because attributions lead to negative affective responses. However, negative reactions are not necessarily motivated by negative affect. Rather, personality research suggests that the Dark Triad personality traits are strongly related to the experience of schadenfreude, that is pleasure at misfortune or harm that others experience (James et al., 2014; Leach, Spears, Branscombe, & Doosje, 2003). A crucial factor in the experience of schadenfreude is that individuals are not personally affected and consider those suffering from harm as an outgroup (Leach et al., 2003; Smith, Powell, Combs, & Schurtz, 2009).
As Attribution Theory suggests, affective reactions to attribution of responsibility to an organisation build the foundation for behavioural and cognitive responses (Weiner, 2006,
2010). Schadenfreude has also been positively linked to organisational responsibility during crises (Coombs & Holladay, 2005). Since all three personality traits of the Dark Triad have been found to promote schadenfreude, it follows that both the effects of crisis response and the type of organisation in crisis on SCC behaviour are influenced by the Dark Triad personality traits in such a way that people high in the Dark Triad traits will show more negative SCC behaviour.
H7a-b. The Dark Triad personality traits moderate the influence of response strategy on (a) the
intention to engage in SCC (path c1) and (b) the content of SCC (path c2) in such a way that individuals high in the Dark Triad traits will report (a) higher intention to engage in SCC and (b) more negative SCC content than individuals low in the Dark Triad traits.
H7c-d. The Dark Triad personality traits moderate the influence of organisational type on (c)
the intention to engage in SCC (path c3) and (d) the content of SCC (path c4) in such a way that individuals high in the Dark Triad traits will report (c) higher intention to engage in SCC and (d) more negative SCC content than individuals low in the Dark Triad traits.
Methods
To test the influence of crisis response strategy, organisational type and the moderating impacts of personality as well as mediation through anger on SCC, an online experiment was conducted. The experiment applied a 2 (response strategy: diminish vs. rebuild) x 2 (organisational type: profit vs. non-profit) between-subjects factorial design. Personality regarding eight different traits (the Big Five and the Dark Triad) was measured as moderator of the main effects on SCC behaviour. Because of practical considerations and limited resources, a convenient sample of 195 respondents was recruited through personal contacts of the author, social network sites, as well as survey exchange platforms. Four participants were excluded because they answered the same on all dependent measures, including the manipulation check items. Thus, the final sample consisted of N = 191.
Sample and Procedure
Participants were between 17 and 68 years old (M = 28.79, SD = 11.44), with 65.4% percent being female (n = 126), opposed to 33.5% male (n = 64). The rest reported other or non-binary
sexes (n = 2). Respondents were mostly Swiss (35.6%), followed by German (19.9%) and British (9.9%), with no geographical limitations for participation.
Successful randomisation was first confirmed through a series of Chi-Square test for gender (x2(6) = 3.29, p = .772), nationality (x2(90) = 91.41, p = .439), residence (x2(42) =
37.02, p = .689) and education (x2(9) = 14.29, p = .112). Because the two experimental
organisational types were concerned with different industrial sectors, randomisation for the respondents’ preferences for the two industries (environmental protection/finance) as well as the topic of the crisis (i.e. racial & gender inequality) was tested through a series of one-way ANOVAs in addition to respondents’ age. Results are indicated in
Table 1 and show that interest in racial equality was not randomly distributed across conditions. A post hoc test revealed that respondents in the diminish/profit condition (M = 4.85, SD = 1.72) were significantly less interested than respondents in the diminish/non-profit condition (M = 5.72, SD = 1.06), p = .034. Hence, interest in racial equality was included as a control variable in the analysis.
After giving their informed consent and demographic information as well as indicating their interest in above mentioned issues, participants’ personality was measured before they were assigned into one of four experimental conditions (diminish/profit, diminish/non-profit, rebuild/profit, rebuild/non-profit) and exposed to the stimulus. Subsequent, dependent variables were measured before participants were finally debriefed and informed that the articles as well as the incident they were about, including the organisations, were invented for the purpose of the examination.
Table 1: Randomisation check for age, interest in industries of organisations and interest in crisis topic. N = 191.
Variable Results ANOVA
Age F(3, 187) = 1.61, p = .188, η2partial = .025
Environmental Protection F(3, 187) = 0.08, p = .973, η2partial = .001
Finance & Economics F(3, 187) = 0.38, p = .771, η2partial = .006
Racial Equality F(3, 187) = 3.19, p = .025, η2partial = .049
Gender Equality F(3, 187) = 0.97, p = .409, η2partial = .015
Manipulation
Participants were randomly divided into four experimental conditions and exposed to a fictitious online news article. Because two different organisational types were suffering the same crisis, an internal event was chosen to assure even plausibility for the crisis across the conditions. Further, to avoid respondents questioning why they had not heard about the crisis and thus the authenticity of the article, a crisis with low external threat caused by discrimination and sexual harassment, establishing high crisis responsibility of the organisation and corresponding to a preventable crisis cluster, was established (Coombs, 2007).
Articles were designed along The New York Times layout and were loosely based on two online articles about internal NGO crises caused through managerial misbehaviour but with no source given (Burch, Blinder, & Eligon, 2019; McVeigh, 2019). A pilot study (N = 12) to test the stimuli also asked for the perceived credibility of the articles (When you think of news that you usually come across online, how realistic was the article you just read?) on a 7-point Likert scale from (1) not realistic at all to (7) very realistic. Results indicated that participants perceived the article as acceptably realistic (M = 4.64, SD = 1.12).
Articles contained three main paragraphs, with the first introducing the respective organisation suffering from the crisis. The organisations were either a European financial services provider or a European non-profit organisation advocating for the protection of tropical rainforests. To manipulate organisational pre-crisis reputation, articles contained information about the organisations’ recent activities. The respective introduction paragraphs can be found in Appendix B.
The second paragraph was identical in all conditions and reported of an internal investigation commissioned after the suicide of an employee. The investigation reported practices of inappropriate behaviour and sexual misconduct of high-ranking employees within the organisation.
Finally, the articles were concerned with the crisis response strategy of the organisations. Citing a spokesperson of the organisation, responses pertained either to a diminish strategy (downplaying the incident and emphasising ongoing day to day business), or to a rebuild strategy (apologising and promising reparations as well as looking ahead). The response paragraphs can also be found in Appendix B.
Participants were exposed to the articles for a minimum of 30 seconds before they could advance to the measurement of the dependent variables.
Moderators
As moderating factors, personality was measured regarding the Big Five and the Dark Triad personality traits. To measure the Big Five traits, Donnellan, Oswald, Baird and Lucas’ (2006) Mini-IPIP inventory was used. With only 20 items (four per trait), the Mini-IPIP represents a good compromise between practicability and precision and is recommended for measuring personality in a concise matter. The scales have shown to yield practicable factor structure and reliability coefficients (Baldasaro, Shanahan, & Bauer, 2013; Cooper, Smillie, & Corr, 2010). The inventory consists of items such as I sympathize with others’ feelings or I like order to which respondents are asked to answer on a 7-point Likert scale from (1) disagree strongly to (7) agree strongly. Following Goldberg and Saucier (2002), approximately half of all items are negatively keyed throughout the entire inventory. Important to note is that the Mini-IPIP conceptualises the fifth trait, Openness, as Intellect/Imagination. Since the conceptualisation as Intellect/Imagination has shown to be highly correlated with Openness, this should not pose a limitation to interpretation of the results (John & Srivastava, 1999; McCrae & Costa, 1987). The complete inventory can be found in Appendix C.
Reliabilities for all five traits were not as high as expected and reported in previous studies. In fact, only two traits out of the five yielded acceptable Cronbach’s alphas: Agreeableness (α =.74) and Intellect/Imagination (α = .63). The remaining Big Five traits were measured unreliably, α = .49 (Extraversion), α = .46 (Conscientiousness), α = .33 (Neuroticism). This needs to be considered when interpreting results. The overall means as well as Cronbach’s alpha coefficients for all personality traits are summarised in.
To measure the Dark Triad personality traits, Jonason’s and Webster’s (2010) Dirty Dozen inventory was used. Like the Mini-IPIP, the Dirty Dozen uses four items per personality trait and has shown to be a valid tool to measure the Dark Triad (Jonason & Luévano, 2013; Webster & Jonason, 2013). Items include but are not limited to statements such as I tend to be cynical or I tend to lack remorse, which were answered on the same 7-point scale as the Mini-IPIP items. The complete inventory can also be found in Appendix C. The three latent scales yielded very satisfactory reliability coefficients (i.e., α = .84, α = .80, α = .84), which can also be found in Table 2.
Table 2: Reliability and overall means, standard deviations for all personality traits (N = 191).
Trait α M SD
Big Five Extraversion .49 4.77 0.86
Agreeableness .74 5.91 0.76
Conscientiousness .46 5.31 0.79
Neuroticism .33 4.58 0.78
Intellect/Imagination (Openness) .63 5.52 0.76
Dark Triad Machiavellianism .84 2.88 1.40
Psychopathy .80 2.55 1.25
Narcissism .84 3.69 1.44
Dependent Measures
Anger. Immediately after reading the articles, participants were asked to report their feeling of
anger (When hearing about the organisation's reaction, I feel angry) on a 7-point Likert scale from (1) disagree strongly to (7) agree strongly (M = 4.72, SD = 1.63).
Secondary crisis communication intention. Next, SCC intention was measured using an
adapted list of items from Jin et al. (2014). Six items were included measuring participants’ intention to communicate about the crisis on social media, asking for the likelihood of respondents to engage in the respective actions (see Table 3). Another two items were included measuring the intention for communication outside a social media context. All items were answered on a 7-point Likert scale from (1) very unlikely to (7) very likely. Participants who indicated that they were not using any social media at all were presented only the last two items.
Principal component analysis (PCA) followed by Varimax rotation revealed the scale to be multi-dimensional, with two components having Eigenvalues above 1.00 (4.34 and 1.17). The two factors accounted for a total of 43.08% of total variance, measuring SCC intention through social media and through private channels respectively. Loadings per factor ranged from .64 to .88. Reliabilities were α = .88 (communication through social media) and α = .80 (communication through private channels) and could not be improved by omitting items. SCC intention was therefore split into two dependent measurements, communication intention on social media (M = 2.60, SD = 1.41, n = 179) and communication intention through private means (M = 4.52, SD = 1.81, n = 191). Table 3 contains all items and their respective loadings on the two factors.
Table 3: Rotated component loadings of PCA using Varimax rotation and reliability coefficients of secondary crisis communication intention (N = 191).
Item Factor
1 2
Like a post or video about the crisis on the internet .81
Retweet or share a post or a video about the crisis .85
Tag someone in a post or video about the crisis .64 .45 Share a post or video about the crisis with someone on Social Media .68 .41 Comment on an existing post or video about the crisis .81
Write a post or article about the crisis on the internet yourself .69 .26 Text someone you know privately about the crisis (SMS, Email, WhatsApp or
any other messenger) .23 .88
Talk to someone you know in person or on the telephone about the crisis .88
Cronbach’s alpha .88 .80
Note. Loadings > |.20| are displayed, strongest loadings per item boldfaced. Factor 1 = Intention to
engage secondary crisis communication through social media, Factor 2 = Intention to engage in secondary crisis communication through private channels.
Secondary crisis communication content. Finally, to measure the content of possible SCC,
respondents were asked to imagine writing a post on Twitter (for participants using social media) or a text message to someone they knew (for participants not using social media). This innovative approach was taken for two reasons: First, although organisational reputation has been used widely to capture individuals’ attitude towards an organisation, the concept lacks an overarching definition and thus commonly accepted measurement instruments. In addition, evidence suggests that reputation is rather issue specific than general, complicating the validity of a general instrument further (Caruana & Chircop, 2000; Walker, 2010).
Instead, asking respondents to provide content concerning an event or the organisation themselves, enables comparisons between different individual attitudes towards the organisation. A disadvantage of this method is the increased effort asked by participants which may lead to higher dropout rates. However, promising 83.77% (n = 160) of all respondents provided meaningful messages.
Further, to ensure that messages were comparable between respondents, messages were limited to 280 characters for tweets and 300 characters for text messages respectively. To obtain a single indicator for the content, messages were coded for overall sentiment, considering both,
language and content on a 5-point scale from (-2) very negative to (+2) very positive. An inter-coder reliability test to determine the precision of instructions was conducted after a short inter-coder training session. The author and an associate coded a total of 13 messages (8.12% of total messages obtained, N = 160). Kripendorff’s alpha coefficient proved the instructions to be reliable (kalpha = .72) (Hayes & Krippendorff, 2007). The data was then coded by the author, with messages scoring -1.01 on average (SD = 0.75, n = 160). Instructions for the coding of overall sentiment contained in the codebook can be found in Appendix D.
Manipulation Checks
To test respondents’ perception of the stimuli, a number of manipulation check items were included at the end of the survey. First, the perception of the crisis as preventable was tested (When you think of the crisis, how much is it the organisation's fault in your opinion?), answered on a 7-point Likert scale from (1) not the organisation’s fault at all to (7) absolutely the organisation’s fault. The overall mean 5.25 (SD = 1.31) suggests that the manipulation was successful, with respondents attributing a high amount of responsibility to the organisation.
Further, participants were asked about the organisation’s response strategy (Would you say the organisation is sorry about what happened?) on a scale from (1) not sorry at all to (7) very sorry. An ANOVA was conducted to determine means between diminish and rebuild conditions. Results show that participants perceived the responses as intended, with the overall mean in the diminish conditions (M = 1.88, SD = 1.10) being clearly lower than in the rebuild conditions (M = 3.71, SD = 1.38), F(1, 179) = 97.73, p < .001, η2partial = .353.
Finally, the perception of the different types of organisation (profit vs. non-profit) was tested by asking about participants’ evaluation of the organisation’s goal (Would you say the organisation's goal is helping society or making profit?) from (1) to help society to (7) to make profit. Again, an ANOVA was conducted (F(1, 179) = 55.01, p < .001, η2partial = .235),
indicating that participants who were exposed to a profit stimulus (M = 5.99, SD = 1.18) judged the organisation to be focused rather on making profit than participants in the non-profit conditions (M = 4.32, SD = 1.79). Hence, all manipulations were successful.
Analysis
To test Hypotheses 1a through 2b, the effects of different crisis response strategies and organisational types respectively on SCC behaviour, a series of ANOVAs was conducted to determine differences between groups. For Hypotheses 3a and 3b as well as 4a and 4b, Hayes’
PROCESS-macro (Hayes, 2013) was applied to test mediation of the effect of the independent variables on SCC behaviour by anger and a moderation of it through Neuroticism. Finally, to test Hypotheses 5a through 7d, multiple linear regression models were run to test moderation of main effects on SCC behaviour through the remaining Big Five personality traits and the Dark Triad traits. Because SCC intention showed to be multi-dimensional, two analyses were run for all hypotheses including SCC intention. Finally, participants’ interest in racial equality was included as a control variable in all analyses.
Results
The results of all analyses are subsumed in Figure 2Figure 2: Conceptual model with significant paths between crisis response, organisational type and secondary crisis communication and personality as moderating, anger as mediating factors., with significant paths highlighted. Following, the results are discussed in detail.
Figure 2: Conceptual model with significant paths between crisis response, organisational type and secondary crisis communication and personality as moderating, anger as mediating factors.
Note. + = p < .10, * = p < .05, ** = p < .01, *** = p < .001. Significant mediation of crisis response on SCC content through anger in red, remaining significant effects of anger on dependent variables in blue.
Hypotheses 1a and 1b assumed a rebuild response to lead to less intention to engage in SCC and less negative content than a diminish response. Accordingly, Hypotheses 2a and 2b assumed a NPO’s response to lead to more favourable SCC outcomes than a PO’s response. The assumptions were tested through a series of one-way ANOVAs. Since SCC intention proved to be multidimensional, two separate ANOVAs were run for Hypotheses 1a and 2a respectively. Only the effect of crisis response strategy on SCC content (H1b) yielded significant results, F(1, 157) = 5.04, p = .026, η2partial = .031. Tweets and messages written by
respondents in the diminish conditions were more negative (M = -1.14, SD = 0.61) than those written by participants in the rebuild conditions (M = -0.86, SD = 0.85), as Table 4 shows.
Interestingly, no effects were found on the intention to engage in SCC at all. A qualitative look at respondents’ messages indicates that a single article might not have been a strong enough stimulus to manipulate respondents’ SCC intentions (e.g. “I would not write a comment based on the very limited information provided by this single article”; “I cannot imagine anything I can write about it. I'm unlikely to ‘report’ it to anyone (…)”).
Additionally, three two-way ANOVAs were conducted to test a possible interaction effect between crisis response and organisational type on SCC behaviour. Results show that there is no significant interaction. Hypotheses 1a, 2a and 2b are therefore rejected, Hypothesis 1b is supported, with no interaction effects found either. Results of all ANOVAs can be found in Table 4.
Table 4: Results of ANOVAs for all main and interaction effects, Hypotheses 1a - 2b.
DV: SCC intention social media
IV: Crisis response strategy (H1a) F(1, 176) = 2.09, p = .150, η2partial = .012 diminish (n = 87) M = 2.80, SD = 1.54
rebuild (n = 92) M = 2.42, SD = 1.25
IV: Organisational type (H2a) F(1, 176) = 0.17, p = .677, η2
partial = .001
profit (n = 87) M = 2.80, SD = 1.54
non-profit (n = 92) M = 2.42, SD = 1.25
IV: Interaction Crisis response*Organisational type F(1, 174) = 2.07, p = .152, η2
partial = .012 diminish/profit (n = 40) M = 2.75, SD = 1.58 diminish/non-profit (n = 47) M = 2.84, SD = 1.52 rebuild/profit (n = 47) M = 2.24, SD = 1.28 rebuild/non-profit (n = 45) M = 2.61, SD = 1.21 DV: SCC intention private
IV: Crisis response strategy (H1a) F(1, 188) = 0.63, p = .427, η2
partial = .003 diminish (n = 94) M = 4.78, SD = 1.81
rebuild (n = 97) M = 4.38, SD = 1.81
IV: Organisational type (H2a) F(1, 188) = 0.98, p = .324, η2partial = .005
profit (n = 94) M = 4.57, SD = 1.83
non-profit (n = 97) M = 4.48, SD = 1.80
IV: Interaction Crisis response*Organisational type F(1, 186) = 1.29, p = .258, η2partial = .007 diminish/profit (n = 47) M = 4.50, SD = 1.90
diminish/non-profit (n = 47) M = 4.85, SD = 1.71
rebuild/profit (n = 47) M = 4.64, SD = 1.78
rebuild/non-profit (n = 50) M = 4.13, SD = 1.82
DV: SCC content
IV: Crisis response strategy (H1b)
F(1, 157) = 5.04, p = .026*, η2partial = .031
diminish (n = 83) M = -1.14, SD = 0.61
rebuild (n = 77) M = -0.86, SD = 0.85
IV: Organisational type (H2b) F(1, 188) = 0.98, p = .324, η2partial = .005 profit (n = 79) M = -0.95, SD = 0.86
non-profit (n = 81) M = -1.06, SD = 0.62
IV: Interaction Crisis response*Organisational type F(1, 155) = 1.27, p = .197, η2partial = .011 diminish/profit (n = 42) M = 4.50, SD = 1.90
diminish/non-profit (n = 41) M = 4.85, SD = 1.71
rebuild/profit (n = 37) M = 4.64, SD = 1.78
rebuild/non-profit (n = 40) M = 4.13, SD = 1.82
Note. Interest in racial equality in all ANOVAs included as covariate. Significant p-values marked with
Hypotheses 3a and 3b assumed the main effects of crisis response and organisational type on SCC intention and content to be mediated by anger. To test mediation of the effects, Hayes’ PROCESS macro was used (Hayes, 2013).
Since only one significant main effect was found (crisis response strategy on SCC content), only one model is discussed here (H3a). Overall, the model proved to be not a great fit explaining only 6% percent of the variance in SCC content, R2 = .06, F(2, 157) = 4.66, p = .011. Confidence intervals calculated through bootstrapping for the direct (95% CI [-.11, .36]) and the indirect effect of response strategy on SCC content (95% CI [.05, .24]) indicate that the effect of response strategy on SCC content is mediated by anger (b = .14). Further, anger is significantly related to all outcome variables, as beta coefficients (column b) indicate. Hypotheses 3a is therefore supported and Hypotheses 3b rejected, since organisational type failed to have a significant effect on the experience of anger. Results of all models calculated with the PROCESS macro can be found in Table 5.
Table 5: Mediation models for main effects through anger, Hypotheses 3a and 3b. Independent and dependent
variables a1 b1
Indirect
effect direct(c’1) total (c1)
DV: SCC intention social media IV: Crisis response strategy
(diminish vs. rebuild) (H3a)
R2 = .13, F(2, 176) = 15.48
-1.01(.22)*** .27(.07)*** -.27(.10) [-.48, -.12]
-.01(.20) -.29(.20)
IV: Organisational type
(profit vs. non-profit) (H3b) R2 = .12, F(2, 176) = 12.39 .05(.23) .27(.06)*** -.02(.07) [-.11, .15] .07(.19) .08(.20) DV: SCC intention private IV: Crisis response strategy
(diminish vs. rebuild) (H3a)
R2 = .09, F(2, 188) = 15.48
-.96(.21)*** .19(.09)* -.18(.09) [-.39, -.03]
-.02(.26) -.20(.25)
IV: Organisational type
(profit vs. non-profit) (H3b) R2 = .09, F(2, 188) = 9.60 -.01(.22) .19(.08)* .00(.05) [-.10, .09] -.25(.25) -.25(.25) DV: SCC content
IV: Crisis response strategy
(diminish vs. rebuild) (H3a)
R2 = .06, F(2, 157) = 4.66
-.94(.24)*** -.14(.04)** .14(.05) [.05, .24]
.13(.12) .26(.12)*
IV: Organisational type
(profit vs. non-profit) (H3b)
R2 = .03, F(2, 157) = 2.32
-.20(.25) -.16(.08)*** .03(.04) [-.05, .11]
-.11(.11) -.08(.12)
Note: * = p < .05, ** = p < .01, *** = p < .001. Unstandardised regression coefficients are given, with
standard errors in parentheses. Confident intervals (95%) are provided for indirect effects. Interest in racial equality in all models included as covariate.
1a and b refer to the effect of independent variable on anger and the effect of anger on the dependent variable respectively. c' and c refer to the direct and total effects. Corresponding paths are indicated in Table 1 and Table 2.
On the basis of Hypotheses 3a and 3b, Hypothesis 4a and 4b assumed the mediation of main effects through anger to be moderated by Neuroticism. Although only one main effect yielded significant results (crisis response on SCC content mediated by anger), moderated mediation models were calculated for the effect of both independent variables on SCC intention (social media and private means) and content through the PROCESS macro. The models are summarised in Table 6, showing overall model fits and moderated mediation indexes for
Neuroticism. As interaction effects show, no significant interactions between independent variables and Neuroticism were found. Hence, no moderated mediation occurs, also not for the significant mediation of anger on the effect of crisis response on SCC content found in Hypothesis 3a (R2 = .00, F(1, 155) = .53, p = .469). Hence, Hypotheses 4a and 4b are rejected. Table 6: Moderated mediation models calculated through PROCESS macro for Hypotheses 4a and 4b.
Independent and dependent
variables Interaction effect
1 Moderated mediation index Neuroticism
DV: SCC intention social media IV: Crisis response strategy
(diminish vs. rebuild) (H4a)
R2 = .00, F(1, 174) = 1.00, p = .319
-.08 (.08) [-.25, .09]
IV: Organisational type
(profit vs. non-profit) (H4b)
R2 = .00, F(1, 174) = 0.33, p = .567
.05 (.09) [-.12, .24]
DV: SCC intention private IV: Crisis response strategy
(diminish vs. rebuild) (H4a)
R2 = .01, F(1, 186) = 1.44, p = .232
-.06 (.07) [-.23, .04]
IV: Organisational type
(profit vs. non-profit) (H4b)
R2 = .00, F(1, 186) = 0.27, p = .602
.03 (.07) [-.08, .19]
DV: SCC content
IV: Crisis response strategy
(diminish vs. rebuild) (H4a)
R2 = .00, F(1, 155) = .53, p = .469 .03 (.05) [-.07, .14]
IV: Organisational type
(profit vs. non-profit) (H4b)
R2 = .00, F(1, 155) = .53, p = .455 -.04 (.05) [-.15, .07]
Note. The moderation index column contains unstandardised regression coefficients for Neuroticism in
the overall model, with standard errors in parentheses and 95% confident intervals in brackets. 1The interaction effect refers to the interaction effect between independent variable and Neuroticism on
the mediating variable anger.
Hypotheses 5a-b, 6a-b and 7a-d were all concerned with the moderation of the main effects of crisis response and organisational type on SCC intention and content respectively. Multiple linear regression models including independent variable, respective personality trait and their interaction term were calculated to determine the moderating effects of personality. All models can be found in Table 7, containing standardised regression coefficients for the interaction terms and overall model fit.
As the table shows, no personality trait of the Big Five had significant influence on any main effect. Hence, Hypotheses 5a and 5b, which assumed Extraversion, Agreeableness and Conscientiousness to have an influence on main effects on SCC intention, were both rejected.
Accordingly, Hypotheses 6a and 6b assumed Openness to be influencing the effects of response strategy and organisational type on SCC content. No significant interaction effect was found either, leading to a rejection of Hypotheses 6a and 6b as well.
Hypotheses 7a through 7b assumed the Dark Triad traits to have a moderating effect on the effect of crisis response/organisational type on both, SCC intention and content. On a significance level of α = .10, all personality traits of the Dark Triad show some moderation effects: Machiavellianism is influencing the effects of response strategy on the intention to engage in SCC through private channels, on SCC content and shows moderation on the effect of organisational type on SCC content. However, respondents high in Machiavellianism tend to report lower intention to engage in SCC (b* = -.13, p = .074), which is opposite the assumed relation. Machiavellianism hence shows some support for Hypotheses 7b and 7d, but not for Hypothesis 7a. Further, Psychopathy moderates the effect of organisational type on SCC intention in a social media context (b* = .12, p = .082), as assumed by Hypothesis 7c and Narcissism moderates the influence of organisational type on SCC content (b* = -.15, p = .058) in support of Hypothesis 7d.
Hence, the Dark Triad traits show influence on main effects overall. However, only Machiavellianism proved to be affecting the sole significant main effect of crisis response on SCC content. Interestingly, Machiavellianism also showed to be mitigating the effect of response strategy on the intention to engage in SCC through private channels. Hence, Hypothesis 7a is clearly rejected without any evidence in support, while Hypotheses 7b, 7c and 7b are all partly supported.
Table 7: Multiple linear regression models for the moderating influence of personality traits. Independent and dependent variables Moderator b* t
DV: SCC intention social media
IV: Crisis response strategy (diminish vs. rebuild)
R2 = .15, F(4, 174) = 9.00 (H5a) Extraversion .11 1.57 R2 = .13, F(4, 174) = 7.61 Agreeableness .11 1.61 R2 = .13, F(4, 174) = 7.64 Conscientiousness -.02 -.22 R2 = .13, F(4, 174) = 7.77 (H7a) Machiavellianism .02 .22 R2 = .13, F(4, 174) = 7.86 Psychopathy -.02 -.32 R2 = .17, F(4, 174) = 10.07 Narcissism -.02 -.26
IV: Organisational type (profit vs. non-profit)
R2 = .14, F(4, 174) = 8.51 (H5b) Extraversion -.11 -1.57 R2 = .11, F(4, 174) = 6.48 Agreeableness -.04 -.60 R2 = .13, F(4, 174) = 7.58 Conscientiousness -.10 -1.36 R2 = .12, F(4, 174) = 7.19 (H7c) Machiavellianism .01 .14 R2 = .14, F(4, 174) = 8.07 Psychopathy .12+ 1.75 R2 = .16, F(4, 174) = 9.53 Narcissism .06 .93 DV: SCC intention private
IV: Crisis response strategy (diminish vs. rebuild)
R2 = .11, F(4, 186) = 6.69 (H5a) Extraversion .00 -.03 R2 = .09, F(4, 186) = 5.42 Agreeableness .06 .82 R2 = .09, F(4, 186) = 5.51 Conscientiousness -.05 -.74 R2 = .09, F(4, 186) = 5.68 (H7a) Machiavellianism -.13+ -1.80 R2 = .08, F(4, 186) = 5.12 Psychopathy .03 .39 R2 = .14, F(4, 186) = 8.40 Narcissism -.02 -.36
IV: Organisational type (profit vs. non-profit)
R2 = .12, F(4, 186) = 7.28 (H5b) Extraversion -.09 -1.24 R2 = .09, F(4, 186) = 5.44 Agreeableness -.07 -.97 R2 = .09, F(4, 186) = 5.74 Conscientiousness -.06 -.86 R2 = .08, F(4, 186) = 5.20 (H7c) Machiavellianism -.08 -1.08 R2 = .08, F(4, 186) = 5.19 Psychopathy -.02 -.29 R2 = .14, F(4, 186) = 8.42 Narcissism .00 .01 DV: SCC content
IV: Crisis response strategy (diminish vs. rebuild)
R2 = .04, F(4, 155) = 2.59 (H6a) Openness .02 .18
R2 = .05, F(4, 155) = 3.23 (H7b) Machiavellianism -.14+ -1.87
R2 = .04, F(4, 155) = 2.57 Psychopathy -.07 -.83
R2 = .06, F(4, 155) = 3.49 Narcissism -.07 -.94
IV: Organisational type (profit vs. non-profit)
R2 = .02, F(4, 155) = 1.97 (H6b) Openness -.12 -1.48
R2 = .03, F(4, 155) = 2.04 (H7d) Machiavellianism -.15+ -1.86
R2 = .01, F(4, 155) = 1.58 Psychopathy -.09 -1.11
R2 = .04, F(4, 155) = 2.79 Narcissism -.15+ -1.91
Note. + = p < .10, * = p < .05, ** = p < .01, *** = p < .001. Interest in racial equality in all models included as covariate. Cells contain either standardised regression coefficients for the interaction term of independent variable and respective moderator on dependent variable or their respective
Discussion
The aim of this study was to provide insights into the role of individuals’ personality in secondary crisis communication behaviour in perspective of enhancing crisis communication effectiveness through tailored communication. An experiment was conducted to investigate the moderating influence of personality traits on the effects of crisis response strategy and organisational context on subsequent affective, behavioural and cognitive reactions.
Results show that main effects of crisis response and organisational type on three different aspects of SSC behaviour (intention to engage in SCC through social media, intention to engage in SCC through private channels and SCC content) only yielded significant results for the effect of crisis response on SCC content, while organisational type did not have an influence on crisis outcomes at all. A possible explanation for the absence of differences in the intention to engage in SCC is the low external crisis threat contained in the stimulus, leading to participants not feeling affected. Similar results suggesting that reputational changes are low if individuals are not involved with an organisation or their products have been found by Claeys and Cauberghe (2014), pointing to the importance of identifying invested stakeholders during a crisis for communication professionals.
Meanwhile, the absence of influence of organisational type on SCC behaviour poses interesting insights. One explanation is that non-profit oriented organisations (NPOs) are not able to establish more favourable reputational relations than profit-oriented organisations solely through their operating towards societal welfare. An alternative reason for the absence might be found in the nature of the crisis: The internal crisis established can be categorised as a CSR crisis, and therefore it endangered an NPO’s core CSR reputational relations and represented a bigger violation of stakeholders’ expectations of the organisation’s behaviour, thus leading to a bigger reputational threat than for the PO (Coombs & Holladay, 2010; Sohn & Edwards, 2018; Sohn & Lariscy, 2014).
Concerning a more public based perspective in understanding crisis reactions, this study offers some new insights. First, the mediating role of anger in negative crisis responses found in various previous examinations (e.g., Coombs & Holladay, 2007; Utz et al., 2013; van der Meer & Verhoeven, 2014) was replicated and showed anger to be strongly related to all SCC outcomes, but no connection between the affective personality trait Neuroticism and anger was found. However, with two exceptions, the Mini-IPIP inventory yielded unreliable coefficients for the measurement of personality traits, with Neuroticism yielding the lowest Cronbach’s alpha coefficient (.33). This measurement might have obscured a possible effect of Neuroticism
