The effects of crisis response strategy, tone of voice and culture on consumers’ responses to organizational crisis communication on Instagram.

The effects of crisis response strategy, tone of voice and culture on

consumers’ responses to organizational crisis communication on Instagram

A Master’s thesis presented to The Department of Communication and Information Studies, Radboud University

Freya Gommans Radboud University Nijmegen

Master International Business Communication, Faculty of Arts Supervisor: Brigitte Planken

Assessor: Warda Nejjari Date: 3July, 2020

Abstract

This study aimed to research online crisis communication cross-culturally. The research question asked about the effects of crisis response strategies (individual accommodative vs. collective accommodative) and tone of voice (conversational human voice vs. corporate voice) in crisis communication on Instagram on Dutch and Spanish consumers’ perceived organizational reputation, word-of-mouth intention, and purchase intention. The importance of this study lies in the little number of cross-cultural studies that exist on crisis communication, but also on the quick development of social media platforms used nowadays for crisis communication. In an online experiment with a 2x2x2 between-subjects design, 124 Dutch and 139 Spanish consumers reacted to crisis communication messages in which a cosmetics company reacted to an organizational crisis, using Instagram. The respondents were exposed to one of four messages in which either a conversational human voice or a corporate voice, and either an individual accommodative or a collective accommodative crisis response strategy was used. It was expected that Dutch respondents, scoring higher on Hofstede’s individualism scale, would generate more positive consumer responses to an individual accommodative response strategy and a conversational human voice. In contrast, Spanish consumers would react more positively to a collective accommodative response strategy and a corporate voice. While preliminary analyses showed that Dutch consumers were more individualistic than Spanish consumers, no effects of culture, crisis response strategy, or tone of voice were found on consumers’ perceived organizational reputation, word-of-mouth intention, and purchase intention. However, these results should be interpreted carefully since manipulation checks showed that respondents did not perceive significant differences between the different conditions. Still, it could be concluded that previous studies on online crisis communication have shown mixed results. These findings, and the constant development of crisis communication possibilities highlighted the importance of further research on contextual factors that could influence consumers’ responses, such as the social media platform used or cross-cultural differences between consumers.

Keywords: Crisis communication, Instagram, cross-cultural, tone of voice, crisis response strategies, Dutch, Spanish

Introduction

In 2017, skincare company Shea Moisture faced an organizational crisis after it released an advertisement in which only one out of four women was black, although the brand is known for its target group of black women. The company was called out on social media and responded on Instagram: ‘Wow, okay – so guys, listen, we really f-ed this one up. Please know that our intention was not – and would never be – to disrespect our community, and as such, we are pulling this piece immediately because it does not represent what we intended to communicate.’ (Shea Moisture, 2017).

Many organizations nowadays face crises, which can negatively affect an organization’s reputation (Benoit, 1997). Therefore, it is highly important for organizations to communicate the right message and word it correctly so that the organizational threat can be reduced. On top of that, in contrast to more traditional media, social media platforms, such as Instagram, allow disseminating these messages easily to stakeholders (Allen, 2014). Therefore, organizations nowadays more often switch from traditional media to more modern social media platforms, such as Instagram, to communicate organizational crises to stakeholders (Guidry et al., 2014). However, not much research has been conducted on how crisis responses should be adapted to the nature of social media or Instagram in particular.

Research on organizational crisis responses has shown that a reputational threat can be mitigated when the type of response matches the nature of the crisis (Ma & Zhan, 2016). An accommodative strategy (e.g. apologies) can be more effective when the organization is responsible for the crisis whereas a defensive strategy (e.g. denial) can work better when the responsibility of the crisis lies outside the organization (Ma & Zhan, 2016).

Moreover, a more personal style of crisis communication, such as a conversational human voice, has been shown to lead to more positive brand evaluations (Van Noort & Willemsen, 2011) and a lower reputational threat (Park & Cameron, 2014). Given the informal nature of social media, a conversational human voice might be relevant for effective crisis responses on Instagram.

Most studies to date have researched the effects of different crisis response strategies monoculturally while organizations often operate in multiple countries. It is plausible that, during a crisis, companies may need to adapt their crisis responses to stakeholders from different cultures (Coombs & Laufer, 2018) as their reactions might depend on cultural factors

reactions to organizational crisis communication on Instagram differ depending on the type of crisis response strategy and the tone of voice used in the message.

Theoretical framework

Crisis communication and its effects on consumers’ responses

An organizational crisis is an unexpected event that can threaten the organization’s operations financially and reputationally (Coombs, 2007). Since how a company reacts to crises affects consumers’ brand-related judgments (Dutta & Pullig, 2011), a company’s crisis response should be appropriate to the situation, for example in terms of using a crisis response strategy that is in line with the nature of the crisis. An often-used framework for crisis communication is the Situational Crisis Communication Theory (SCCT), which suggested that reputational threat increases when stakeholders attribute more responsibility to the organization (Coombs, 2007). Accordingly, the model identified three clusters of responsibility, from the lowest to the highest level of responsibility: the victim cluster, the accidental cluster, and the preventable cluster. The victim cluster concerns crises where the organization is also seen as a victim of the crisis, such as a natural disaster. The accidental cluster involves crises that are unintentional on the part of the organization, such as crises that are caused by a technical error. Finally, during a preventable crisis, such as human-error product harm, the organization is aware of the fact that it is placing people at risk and is thus highly responsible for the crisis (Coombs, 2007). Coombs (2007) suggested that crisis response strategies should be adapted to the type of crisis, which has been confirmed by a meta-analysis of studies investigating the effects of attributed responsibility and crisis response on organizational reputation (Ma & Zhan, 2016). Results showed that crisis response strategies matching the level of attributed responsibility to the organization protected the organization’s reputation (Ma & Zhan, 2016).

Based on the organization’s level of responsibility, the SCCT model identified several crisis response strategies and placed them on a defensive-accommodative continuum. More defensive strategies put the organizational interests first, such as blaming an actor outside the organization for the crisis. In contrast, more accommodative strategies prioritize victims, such as offering them compensation (Coombs, 2007). Coombs (2007) suggested that more accommodative strategies would be most effective for crises with a high level of organizational responsibility. These assumptions have been confirmed by Mattila (2009), who studied the effects of crisis response strategy on consumer perceptions. Denying the organization’s responsibility (i.e. defensive strategy) led to more negative attitudes and behavioral intentions

Similarly, Lyon and Cameron (2004) examined the effects of response strategy on consumers’ attitudes and behavioral intentions after reading news stories about organizational crises with an internal responsibility. Results showed that apologizing for the crisis (i.e. accommodative strategy), rather than defensive responses, had a positive effect on consumers’ attitudes towards the company and purchase intentions (Lyon & Cameron, 2004).

Jin, Liu, and Austin (2014) also studied the role of crisis origin and response strategy on consumers’ reactions towards crisis communication on social media. For crises with an internal origin, an accommodative, versus a defensive, response led to a higher acceptance of the response. However, for publics perceiving an external crisis origin, a defensive response had a more positive effect on the consumers’ acceptance of the response (Jin, Liu, & Austin, 2014). In their study, Jin, Liu and Austin (2014) introduced the Social-mediated Crisis Communication Model (SCCM), which suggested that the appropriateness of a crisis response not only depends on the crisis response strategy and the responsibility of the organization, but also on other contextual factors such as the message form, and the source.

Thus, while results from previous studies have mainly suggested that accommodative strategies generally lead to more positive consumer responses than defensive strategies, the effectiveness of crisis response strategies also seems to depend on situational circumstances like, for example, the internal or external origin of the crisis (Jin, Liu, & Austin, 2014) or the level of attributed responsibility to the organization (Ma & Zhan, 2016). Therefore, more research on these specific situational circumstances that might affect consumers’ responses to crises, or crisis response strategies particularly, is needed, especially on newly developing platforms, such as Instagram. None of the studies above have focused on Instagram as a crisis communication platform, although a systematic review of research on online crisis communication showed that stakeholders perceive different social media platforms diversely (Eriksson & Olsson, 2016). Moreover, 70% of studies that investigated crisis communication on a social media platform researched the context of Twitter, and only 2% studied crisis communication on Instagram (Eriksson & Olsson, 2016). Therefore, more research is needed on effective social media crisis communication in other contexts than Twitter. For this reason, the present study extended theory on crisis communication to investigate the effects of different communication strategies on a relatively new, and not much researched, social media platform: Instagram.

Tone of voice and its effects on consumer’s responses to crisis communication

In addition to choosing an appropriate crisis response strategy, research has shown that the wording of the message may also affect consumers’ reactions to crisis responses. For example, consumers often want organizations to communicate about a crisis in a personal way, especially on social media which is known for its dialogic nature (González-Herrero & Smith, 2008).

When investigating the effects of tone of voice, researchers have distinguished conversational human voice (CHV) and corporate voice (CV). CHV is a style of communication in which the organization, as an individual, engages in a natural and interactive way with the public in a personalized way (Kelleher, 2009). Various methods can be used to create messages that reflect CHV, such as addressing the public personally by using second-person pronouns, using abbreviations, employing emotional language, and starting a dialogue (Van Noort et al., 2014). In contrast to CHV, organizations can also choose to use CV in their messages, where the position of the organization with the stakeholder is more formalized. An example is writing from an organizational perspective and using third-person pronouns to address the public (Kerkhof, Beugels, Utz, & Beukeboom, 2011).

Kelleher (2009) studied how using CHV in messages on organizational blogs affected publics’ relational outcomes. Using CHV led to higher levels of consumers’ trust, satisfaction and commitment than not using CHV (Kelleher, 2009). Similar results were found in studies that investigated tone of voice in crisis communication contexts. For example, Sweetser and Metzgar (2007) found that the use of CHV in organizational blogs about crises positively affected consumers’ perceptions of the organization’s efforts to maintain stakeholder relationships. Moreover, in a study on the most effective strategies to react to consumers’ online complaints in a crisis context, consumers’ brand evaluations after web care were more positive when a company employed CHV in their web care communication than when it did not (Van Noort & Willemsen, 2011). Thus, it is likely that the use of CHV in crisis response on the Internet generally leads to more positive outcomes for the organization than not using CHV or using CV. However, no studies have especially investigated whether this is the case on Instagram, which is increasingly used for crisis communication nowadays (Guidry et al., 2014). Previous studies on online crisis communication have mostly focused on the use of CHV or CV on organizational blogs. However, Instagram tends to be a highly interactive platform (Ting, Ming, de Run, & Choo,2015) and might, therefore, ask for a rather informal, natural style of

voice by investigating crisis communication on Instagram, which is a highly interactive social media platform.

Crisis response strategies and tone of voice

Few studies have investigated the influence of both crisis response strategy and tone of voice on consumer reactions in a crisis communication context. While crisis response strategies tend to focus on the content of the message, tone of voice is more concerned with how the message is worded. The usefulness of studying crisis response strategies and tone of voice together lies in the fact that it might be important to determine what should best be communicated, but also how it should be in terms of language use.

One rare example in which the potential interaction of crisis response strategies and tone of voice were studied is the study of Park and Cameron (2014). The study investigated the effects of tone of voice and type of response strategy in crisis communication via blogs on consumers’ acceptance of the response, organizational reputation, and behavioral intentions. Their experiment, where students were exposed to a blog post written in CHV or CV, indicated that the use of CHV increased participants’ purchase intentions and intention of word-of-mouth (Park & Cameron, 2014). However, no effect of tone of voice on organizational reputation or publics’ acceptance of the response was found (Park & Cameron, 2014). Concerning the type of crisis response, participants who were exposed to a message with an accommodative instead of a defensive strategy found the organizational crisis responses to be more acceptable (Park & Cameron, 2014). Nevertheless, no significant effects of response type on perceived reputation and purchase intention emerged.

Another study investigating crisis strategy and tone of voice, but in the specific context of responses to online complaints, was done by Huibers and Verhoeven (2014). First, a content analysis examined how Dutch organizations use web care strategies and CHV on Twitter. Dialogues between organizations and consumers were coded for CHV and web care strategies, based on crisis response strategies identified by previous research, such as the SCCT (Coombs, 2007). Subsequently, these strategies were categorized as individual accommodative (IA), collective accommodative (CA), or defensive. IA strategies try to solve the problem of the individual stakeholder that complains about the company (Huibers & Verhoeven, 2014). By using CA strategies, the organization aims to solve the problem for all stakeholders and thus focuses more on the group (Huibers & Verhoeven, 2014). Moreover, in a defensive strategy,

the organization does not acknowledge its responsibility for the crisis. Based on their corpus, they classified information, sympathy, and compensation strategies as IA, whereas apology and corrective action were classified as CA. Moreover, they found that CHV was only used in 24,1 % of Twitter dialogues investigated. The second study investigated the effect of different web care strategies and of CHV on consumers’ perceptions of a corporation’s reputation. Participants saw a tweet in which the company reacted to a customer’s complaint. While tone of voice did not affect perceived reputational threat, IA web care strategies led to a smaller threat on corporate reputation than CA web care strategies, and defensive web care strategies (Huibers & Verhoeven, 2014).

Thus, previous research on crisis response strategies and tone of voice has given contrasting effects of tone of voice and response strategy on consumers’ reactions to crisis communication. This suggests that the effectiveness of both factors might not only depend on the message, but contextual factors such as type of crisis or the platform the message is posted on might also play a role in how consumers react to an organization’s crisis response. Therefore, research needs to keep investigating the effects of crisis response strategies and tone of voice in different contexts, such as on social media platforms that have not been investigated yet. Thus, as accommodative strategies have been shown to positively affect consumer responses to a crisis on Twitter (Huibers & Verhoeven, 2014), it might be useful to extend these findings to other social media platforms, such as Instagram. However, consumers’ responses to organizational crisis communication might not only depend on factors related to the message, such as what or how it is said or on which media it is distributed. It might be the case that the consumers’ personal background characteristics, such as their culture, might also play a role in how they react to a crisis communication message. Therefore, the next section discusses research on cultural differences in consumers’ reactions to crisis communication.

Cultural differences in consumers’ reactions to crisis communication

Although the effect of crisis response strategies, sometimes in combination with the effect of tone of voice, has been researched, many previous studies adopted a single-country approach. Nevertheless, cultural characteristics might also play a role in how these responses are received. Especially now that organizations often operate in multiple countries and disseminate their messages to diverse stakeholder groups, there is a great need for more research in the area of

whether stakeholders in different countries react diversely to an organization’s crisis response (Coombs & Laufer, 2018).

A rare study that has researched consumers’ reactions to crisis response strategies cross-culturally is that of An, Park, Cho, and Berger (2010), who studied the effects of organizational crisis response strategies in the United States and South Korea on consumers’ anger, attitudes towards organizational crisis response and impressions towards the organization. These two countries score differently on Hofstede’s individualism-collectivism scale, which shows to what extent individuals are integrated into groups (Hofstede, 2003). Participants responded to hypothetical news stories about a crisis scenario. Four types of responses from the organization were created based on two factors: causal/treatment and individual/organizational response. The causal response focused on communicating the cause of the crisis, which could either be an individual or the organization. In the treatment response, an emphasis was placed on how the crisis would be handled, either by punishing individual employees or improving the systems of the organization as a whole. Results showed that in messages with an individual causal and treatment response, South Korean participants had more negative attitudes, emotions, and impressions towards the message and the organization. However, in the conditions with an organizational causal and treatment response, the opposite effect was found. According to the researchers, these differences can be explained by the countries’ differing scores on Hofstede’s individualism-collectivism scale (2003). Individualist societies, such as the US, are known for the individual’s focus on themselves, whereas people in collectivist societies, such as South Korea, are taught to be caring for the group (Hofstede, 2003). Similarly, it might be assumed that other countries that score differently on this scale might react in diverse ways to more individually or collectively focused crisis response strategies.

Two European countries that differ in terms of individualism are Spain and the Netherlands. Whereas Spain is placed mid-scale on the individualism-collectivism scale, it can be considered to be more of a collectivistic culture (51) than the Netherlands, which is seen as relatively individualistic (80) (Hofstede, 2003). These differences in cultural characteristics might impact consumers’ reactions to crisis communication. Previous studies on cross-cultural crisis communication have mostly researched cultures that are geographically further away from each other, with a focus on the United States and Asia (e.g. An et al.,2010). Therefore, to fill in the research gap of cross-cultural studies on crisis communication in other continents and contexts (Eriksson & Olsson, 2018), the current study investigated two countries within Europe that are geographically located relatively close to each other, namely the Netherlands and Spain.

Similarly to crisis response strategies, Spanish and Dutch levels of individualism might also impact their preference for CHV or CV in crisis communication. As CHV depicts the organizations more as an individual that consumers can hold a conversation with, it might create more positive consumer responses in a more individualist culture like the Netherlands than in a more collectivist culture like Spain. However, no research has been conducted on the effectiveness of CHV for Dutch and Spanish consumers. Therefore, the present study not only aimed to investigate cross-cultural consumer reactions to various crisis response strategies but also to the use of different tones of voice on Instagram.

Research questions and hypotheses

While previous research has shown that the use of different crisis response strategies and tones of voice can impact consumers’ reactions towards crisis communication differently (e.g. Lyon & Cameron, 2004; Jin, Liu, & Austin, 2014; Sweetser & Metzgar, 2007; Van Noort & Willensen, 2011), a limited number of studies have investigated the impact of these factors combined. Moreover, as individualism-collectivism has been shown to affect consumers’ responses towards crisis responses differently across countries (e.g. An et al., 2010), cross-cultural differences between reactions might arise in the present study as well. Additionally, individualism-collectivism as a factor in crisis communication has not yet been researched online and, while crisis response strategies and tone of voice were investigated on Twitter and organizational blogs, they have not been researched in the context of Instagram even though it is nowadays an often used platform for crisis communication. As an organizational crisis can threaten the organization both in a financial and a reputational way, it is important to investigate the effects of these factors, discussed above, both on dependent variables related to reputation, such as perceived organizational reputation and word-of-mouth intention, but also on a variable that might be indirectly concerned with financial performance, such as purchase intention. Therefore, similarly to Park and Cameron (2014), this study researched consumers’ responses in terms of perceived organizational reputation, word-of-mouth intention, and purchase intention. For these reasons, the following research question was formulated:

RQ: To what extent do crisis response strategy (accommodative at the individual level vs. accommodative at the collective level) and tone of voice (conversational human voice vs.

crisis communication on Instagram, in terms of perceived organizational reputation, purchase intention and word-of-mouth intention?

To investigate the research question above more clearly, several hypotheses were formulated.

According to Hofstede’s individualism-collectivism scale, Spain has a more collectivistic culture while the Dutch are rather individualistic. The results of the experiment by Huibers and Verhoeven (2014), researching the effect of web care strategies on organizational reputation, suggested that Dutch consumers perceived a lower reputational threat when a more individual approach was employed in the crisis response strategy. A possible explanation is that Dutch respondents value individualism and thus respond favorably to communication strategies that fit with their culture, such as individual accommodative crisis response strategies. However, as Spain is a more collectivist culture, Spanish consumers might react more positively when a strategy is used that focuses on the well-being of the group, such as a collective accommodative crisis response. Therefore, the following hypotheses were proposed:

H1: Spanish consumers will (a) perceive the organizational reputation as more positive, (b) have a higher purchase intention and (c) a greater word-of-mouth intention after being exposed to collective accommodative crisis response strategies than Dutch consumers will.

H2: Dutch consumers will (a) perceive the organizational reputation as more positive, (b) have a higher purchase intention and (c) a greater word-of-mouth intention after being exposed to individual accommodative crisis response strategies than Spanish consumers will.

Compared to traditional media, social media is known for its informal and personal communication practices. Based on previous research (González-Herrero & Smith, 2008; Sweetser and Metzgar, 2007; Van Noort & Willemsen, 2011), CHV might, therefore, be more effective in crisis communication on Instagram than CV:

H3: Based on the dialogic nature of social media, the use of a conversational human voice in an organizational crisis response on Instagram will lead to (a) a more positive perception of organizational reputation, (b) a higher purchase intention and (c) a greater word-of-mouth intention, regardless of culture.

Method

Design

To investigate the effects of tone of voice (CHV vs. CV), crisis response strategy (IA vs. CA) and culture (Spain vs. the Netherlands) on consumers’ responses to organizational crisis posts on Instagram, an experimental study was conducted. A 2x2x2 experimental design with as between-subjects factors tone of voice, crisis response strategy, and culture was used. The dependent variables in this study were perceived organizational reputation, word-of-mouth intention, and purchase intention. Table 1 presents the distribution of respondents based on culture and experimental condition they were exposed to.

Table 1. Distribution of respondents based on culture and experimental condition (crisis response strategy and tone of voice).

Culture Crisis response

strategy Tone of voice n Dutch IA CHV 28 Dutch CA CHV 33 Dutch IA CV 28 Dutch CA CV 35 Spanish IA CHV 40 Spanish CA CHV 37 Spanish IA CV 32 Spanish CA CV 30 Total 263 Materials

The materials created for the experiment consisted of two parts: an online article from a national newspaper that reported objectively about an organizational crisis, and an Instagram post showing the organization’s reaction to the crisis. A fictional crisis about too high concentrations of chemical ingredients in a day cream of a non-existing cosmetics company, Skin Cosmetics,

formed the context for the materials and stimuli. As Instagram is often used as a PR platform by the cosmetics sector, a crisis in this sector was chosen.

The newspaper article told the story of the Better Skin day cream, a new product of a fictitious cosmetics company named Skin Cosmetics. All respondents, regardless of the condition to which they were assigned, were exposed to this introductory text. To ensure that the crisis referred to both men and women, the case included a unisex facial cream. It was explained that customers experienced severe skin rashes and irritations after they had used the day cream, which had been caused by too high concentrations of chemical ingredients in the product. It was made clear that the organization knowingly did not test the product enough before releasing it and decided to manufacture the product in a country where legislation on the use of chemical ingredients in cosmetics is limited. Thus, the organization’s responsibility for the crisis was explicitly shown to be high and the crisis, therefore, belonged to the intentional cluster. In this way, it was hoped that the strongest response would be elicited from the respondents in this study. The Dutch newspaper article can be found in Appendix A, the Spanish newspaper article can be found in Appendix B.

Both in Spanish and in Dutch, four Instagram posts in which the company reacted to the newspaper article were created. One employed an individual accommodative strategy (IA) in conversational human voice (CHV), one used an individual accommodative strategy (IA) in corporate voice (CV), another one a collective accommodative strategy (CA) in conversational human voice (CHV), and the last one used a collective accommodative strategy (CA) in corporate voice (CV).

Response strategy, the first independent variable, was operationalized as follows. According to Huibers and Verhoeven (2014), IA response strategies included the objective information, sympathy, and compensation strategies of the Situational Crisis Communication Theory (Coombs, 2007). These three crisis response strategies were combined in the IA response strategy condition. CA response strategies included apology and corrective action, which were combined in the CA response condition. More specifically, and following Huibers and Verhoeven (2014), the IA strategy was operationalized by providing objective information about the crisis (i.e. too high concentrations of chemical ingredients were found), showing sympathy for the individual victim (In Dutch: We leven met jou/u mee; In Spanish: Simpatizamos contigo/con usted), and offering compensation to the individual customer in the form of a reimbursement. On the other hand, in the CA response strategy, the organization deeply apologized for what happened (In Dutch: We willen onze oprechte excuses aanbieden;

In Spanish: Queremos ofreceros nuestras más sinceras disculpas), asked for complete forgiveness from the society as a whole, and said what steps were being taken to prevent a recurrence in the future so that no single client would ever experience a similar situation. Moreover, the IA condition addressed the client as an individual and used singular language while all clients at once were addressed in the CA condition (i.e. In Dutch: Skin Cosmetics heeft het beste met u voor vs. Skin Cosmetics heeft het beste met al haar klanten voor; In Spanish: Skin Cosmetics le desea lo mejor vs. Skin Cosmetics quiere lo mejor para todos sus clients).

Tone of voice, the second independent variable, was operationalized based on previous studies. As suggested by Van Noort et al. (2014), a CHV can be created through message personalization, informal speech, abbreviations, emotional language and dialogue. In the current study, in Instagram posts with CHV, first- and second-person personal pronouns such as ‘wij, jij’ and ‘nosotros, tú’ were used. The messages with CV were written in third person and addressed the customer in a formal way (E.g. In Dutch: Skin Cosmetics heeft het beste met u voor; In Spanish: Skin Cosmetics le desea lo major). Furthermore, the messages written in CHV asked for direct feedback by inviting consumers to send an e-mail in the case of complaints of problems, while this was not the case when corporate voice was used. In addition, more informal speech was employed in the CHV condition compared to the CV condition. For example, the customer(s) was/were addressed as ‘Dear customer’ (In Dutch: Geachte klant; In Spanish: Estimado cliente) in the CV condition while ‘Ok, dear followers’ (In Dutch: Oké lieve volger; In Spanish: Hola, queridos seguidores) was used in the CHV condition. Also, the Instagram posts with CHV closed the message with the name of the management team of Skin Cosmetics (In Dutch: XOXO Het Managementteam van Skin Cosmetics; In Spanish: XOXO El equipo de gestion de Skin Cosmetics), while the ones with CV ended with the name of the organization (Skin Cosmetics).

The messages in all four conditions presented the same factual information and it was made sure that they were of more or less the same length. The messages were written in either Spanish or Dutch. The Spanish texts were checked and fine-tuned by two native speakers of the Spanish language to make sure the Spanish translations of the four stimulus texts were as close as possible to the original Dutch. The Dutch version of the materials can be found in Appendix C and the Spanish version of the materials can be found in Appendix D.

Subjects

In total, 263 respondents participated in the study. More specifically, 124 Dutch (47.1%) and 139 Spanish consumers (52.9%) were found based on convenience sampling. Dutch respondents were found through personal connections, which was also the case for Spanish respondents. However, Spanish respondents were also found online through Facebook groups.

Age

Respondents were between the age of 18 and 71, with the average age being 30.63 (SD = 11.85). Dutch respondents were between the age of 18 and 71, with an average age being 30.19 (SD = 13.10). The age of the Spanish respondents ranged from 18 to 66 with the average age being 31.02 (SD = 10.64).

A one-way analysis of variance showed a significant difference between the mean age of the respondents viewing the four different conditions. F (3,259) = 3.01, p = .031. The mean age of the respondents in the IA CHV condition (M = 33.63, SD = 11.82) was significantly higher than the mean age of the respondents in the CA CV condition (p = .046, Bonferroni correction; M = 28.17, SD = 10.66). No significant differences were found between the other conditions in terms of age.

Within the Dutch group, however, a one-way analysis of variance showed no significant differences between the conditions in terms of respondents’ age (F (3,120) = 2.57, p = .06). In the Spanish group, a one-way analysis of variance showed there were no significant differences in respondent age between the conditions either (F(3,135) = 1.01, p = .392).

The age of the Spanish and Dutch respondent groups overall did not significantly differ either (F(1,261) < 1).

Gender

In total, 22.8 % of respondents were male, while 76.8% were female and 0.4% preferred not to indicate their gender. Of the Dutch respondents, 18.5% were male and 18.5% indicated to be female. Of the Spanish respondents, 26.6% indicated to be male,72.7% were female and 0.7% preferred not to say.

A Chi-square test showed no significant relation between condition and gender (χ2 ₍₆₎

= 9.57, p = .144) for the respondent group as a whole. Moreover, a Chi-square test showed no significant relation between condition and gender for the Dutch group (χ2 _{(3) = 6.14, p = .105).}

Similarly, a Chi-square test showed no significant relation between condition and gender for the Spanish group (χ2 _{(6) = 5.32, p = .503). Furthermore, a Chi-square test showed no significant}

relation between nationality and (χ2 _{(2) = 3.42, p = .181).}

Educational level

In total, most respondents (63.9%) indicated WO or Educación superior universitaria to be their educational level. Furthermore, 12 respondents (4.6%) indicated secondary education as their educational level, 25 respondents (9.5%) indicated MBO or Título de técnico superior to be their educational level, 58 respondents (22.1%) had HBO or Educación superior no universitaria as educational level. A Chi-square test showed no significant relation between condition and educational level for the respondent group as a whole (χ2 _{(9) = 4.48, p = .877).}

Of the Spanish respondents, 12 respondents (8.6%) indicated secondary school, 7 respondents (5.0%) filled in título de técnico superior, 18 respondents (12.9%) indicated Educación superior no universitaria and 102 (73.4%) indicated Educación superior universitaria as their educational level. A Chi-square test showed no significant relation between condition and educational level for the Spanish group of respondents (χ2 _{(9) = 2.47, p = .982).}

Of the Dutch respondents, 18 respondents (14.5%) indicated MBO as their educational level, 40 respondents (32.3%) filled in HBO to be their educational level and 66 respondents (53.2%) indicated to have finished WO. A Chi-square test showed no significant relation between condition and educational level for the Dutch group of respondents (χ2 _{(6) = 8.59, p =}

.198).

However, a Chi-square test showed a significant relation between nationality and educational level (χ2 _{(3) = 32.15, p < .001). Significantly more Spanish respondents (100%)}

indicated secondary education as their educational level than Dutch respondents did (0%). Moreover, significantly more Dutch respondents (72%) indicated MBO as their educational level than Spanish respondents (28%) indicated the similar level Título de técnico superior. Furthermore, significantly more Dutch respondents (69%) indicated HBO to be their educational level than Spanish respondents filled in Educación superior no universitaria (31%).

Lastly, significantly more Spanish respondents (60.7%) indicated their educational level to be Educación superior universitaria than Dutch respondents (39.3%) indicated WO.

Instruments

A questionnaire was used to collect participants’ data about the proposed dependent variables and demographic variables. Based on Park and Cameron (2014) and Huibers and Verhoeven (2014), this study measured perceived organizational reputation, purchase intention, and word-of-mouth intention. All scales were translated into Dutch and into Spanish. The entire questionnaires can be found below in Appendix E (in Dutch) and F (in Spanish).

Based on Sung and Yang (2008), perceived organizational reputation was measured using five 7-point Likert-type items. Examples are: ‘This organization looks like an organization with strong prospects for future growth’ and ‘This organization is socially responsible’. The reliability of ‘perceived organizational reputation’ comprising five items was good: α = .84. Consequently, the mean of all five items was used to calculate the compound variable ‘perceived organizational reputation’, which was used in the further analyses.

Additionally, word-of-mouth intention was measured using two items on a 7-point Likert scale adapted from Brown, Barry, Dacin, and Gunst (2005). The scale used by Brown et al. (2005) focused on services, while the organization in the current study only sold products. Therefore, ‘services’ was changed into ‘products’ for the questionnaire of this study. An example is ‘I would encourage friends to use products from this organization’. The reliability of ‘word-of-mouth intention’ comprising two items was good: α = .98. Consequently, the mean of all two items was used to calculate the compound variable ‘word-of-mouth intention’, which was used in the further analyses.

Purchase intention was measured as in the study of Park and Cameron (2014), using five 7-point Likert scale items adapted from Coyle and Thorson (2001). Examples are ‘I would purchase products from this company’ and ‘I would try products from this company’. The reliability of ‘purchase intention’ comprising five items was good: α = .96. Consequently, the mean of all five items was used to calculate the compound variable ‘purchase intention’, which was used in the further analyses.

Also, a manipulation check was done for crisis response strategy and tone of voice to ensure that the manipulation of these factors is strong enough in all four conditions. The manipulation of crisis response strategy was checked using five 7-point Likert scale items based

on Huibers and Verhoeven (2014). An example is: ‘De organisatie toont sympathie’. Separate reliability analyses were conducted for ‘individual accommodative strategy’ and ‘collective accommodative strategy’. The reliability of ‘individual accommodative strategy’ comprising three items bordered on acceptable: α = .60. Similarly, the reliability of ‘collective accommodative strategy’, comprising two items, bordered on acceptable: α = .58. Therefore, the items of ‘collective accommodative crisis response strategy’ were recoded so that their direction was the same as the items of ‘individual accommodative crisis response strategy’: a high score on the scale represented a strong perception of an individual accommodative strategy. Consequently, the mean of all five items was used to calculate the compound variable ‘crisis response strategy’, which was used in further analyses.

To check the manipulation of tone of voice, five items on a 7-point Likert scale adapted from Kelleher and Miller (2006) were used. For example, ‘De organisatie nodigt uit tot interactie’. The reliability of ‘tone of voice’ comprising five items was good: α = .88. Consequently, the mean of all five items was used to calculate the compound variable ‘tone of voice’, which was used in the further analyses.

Furthermore, Spanish and Dutch participants were tested to test the assumption they differed in terms of their positions on the individualism-collectivism scale, and to assess if the Dutch are indeed more individualistic than the Spanish consumers. Sivadas, Bruvold, and Nelson (2007) presented a 14-item scale measuring the individualist/collectivist dimensions of culture, which was adapted and slightly shortened for the current study. Six items on a 7-point Likert scale were used. An example of an item is: ‘I usually sacrifice my self-interest for the benefit of my group’. The reliability of ‘individualism’ comprising six items was questionable: α = .58. However, after deleting the item ‘Without competition it is not possible to have a good society’ (In Dutch: ‘Zonder concurrentie is het niet mogelijk om een goede samenleving te hebben’; In Spanish: ‘Sin competencia no es posible tener una buena sociedad’), the reliability of ‘individualism’ comprising five items was near acceptable: α = .65. Therefore, the remaining five items were used to calculate the compound variable ‘individualism’ for further analyses. Finally, the participant was asked some demographic questions such as nationality, age, gender, and educational level. Moreover, the respondents were asked how often they normally use facial cream, cosmetics, and Instagram.

Procedure

The experiment was conducted via an online survey. This happened around the first weeks of May 2020. After the respondents clicked on an anonymous link, a screen would pop up where the topic of the study was shortly explained and respondents were asked to agree with the thesis terms and conditions. Then, a short story about the fictious organization and an explanation of the crisis situation were shown. The respondents read the fictional story about the crisis and were randomly exposed to one of the four conditions, after which they would fill in the questionnaire. Afterwards, they were thanked for their participation, and contact information was presented for possible questions or remarks about the experiment. Completing the questionnaire took approximately 7 minutes. No incentives were given to the participants for taking part in the study.

Statistical treatment

To analyze the effects of tone of voice, crisis response strategy among Spanish and Dutch participants on perceived organizational reputation, purchase intention and word-of-mouth intention, three three-way univariate analyses of variance were conducted. In the case of any interaction effects, further one-way analyses of variance were done to interpret the interaction.

Results

Preliminary analyses

An independent samples t-test showed no significant difference between the conditions with CHV and the conditions that used a CV with regard to the extent to which respondents experienced a CHV (t (262) = 0.25, p = .805). Respondents who were exposed to a message in which a CHV was used (M = 4.71, SD = 1.42) did not perceive the message to be more interactive or humane than respondents who were shown a message with a CV (M = 4.66, SD = 1.44).

An independent samples t-test showed no significant difference between the conditions with an IA crisis response strategy and the conditions that used a CA crisis response strategy with regard to the extent to which respondents experienced an IA crisis response strategy (t (261) = 1.75, p = .081). Respondents who were shown an IA crisis response strategy message (M = 4.65, SD = 1.14) did not differ from respondents who were shown a CA crisis response strategy (M = 4.40, SD = 1.18) in terms of the extent to which they perceived the message to contain an IA crisis response strategy.

Thus, it can be concluded that both the manipulations of Voice and Crisis response strategy were not successful. The differences between the conditions of the independent variables were not sufficient enough and respondents did not perceive the different conditions to be sufficiently different. These findings should be taken into account when interpreting the results of the main analyses in this study.

Further analyses tested the extent to which Dutch and Spanish respondents showed individualistic or collectivistic cultural characteristics. An independent samples t-test showed a significant difference between Dutch and Spanish respondents with regard to the extent to their individualism-collectivism (t (256.54) = - 4.90, p < .001). As assumed for this study, Spanish respondents showed higher levels of collectivism (M = 4.23, SD = 1.09) than Dutch respondents (M = 3.64, SD = 0.85).

On average, respondents used facial cream between once a week to three to four times a week (M = 5.49, SD = 2.31). Moreover, respondents generally used cosmetic products on average once a week to three to four times a week (M = 5.47, SD = 2.23). Similarly, respondents indicated to use Instagram three to four times a week on average (M = 5.66, SD = 2.17). Based on these results, it could be concluded that the crisis case used in this study, concerning a facial

frequently by the respondents, Instagram can be considered relevant for these respondents. This contributes to the ecological validity of this study.

The effects of Voice, Crisis response strategy, and Culture on perceived organizational reputation

A three-way univariate analysis of variance with Voice (CHV vs. CV), Crisis Response Strategy (IA vs. CA) and Culture (Spanish vs. Dutch) as factors showed no significant main effect of Voice on perceived organizational reputation (F(1,255) < 1). No significant main effect of Crisis response strategy was found either (F(1,255) < 1). Culture was not found to have a significant main effect on perceived organizational reputation (F(1,255) < 1). Moreover, no significant interaction effect of Voice and Crisis response strategy on perceived organization reputation was found (F(1,255) < 1). No significant interaction effect of Crisis response strategy and Nationality was found (F(1,255) < 1). Furthermore, there was no significant three-way interaction effect of Voice, Crisis Response Strategy and Nationality on perceived organizational reputation (F(1,255) < 1). Table 2 shows means and standard deviations of perceived organizational reputation in function of Crisis response strategy, Voice, and Culture. While a weak, significant interaction effect of Voice and Nationality on perceived organizational reputation was found (F (1,255) = 4.48, p = .035, η2 = .017), follow-up one-way analyses of variance, for the Dutch and the Spanish respondents separately, showed no significant effect of Voice on perceived organizational reputation for the Dutch respondents (F (1,122) = 3.27, p = .073, η2 = .026) or for the Spanish respondents (F (1,137) = 1.97, p = .163, η2 = .014).

Table 2. Means and standard deviations (between brackets) of perceived organizational reputation per Crisis response strategy, Voice, and Culture (1 = negative perceived organizational reputation, 7 = positive perceived organizational reputation)

Dutch (n = 124) Spanish (n = 139) IA CA IA CA CHV 3.84 (0.98) n = 28 3.98 (1.16) n = 33 3.77 (1.25) n = 40 3.69 (1.25) n = 36 CV 3.71 (1.03) n = 28 3.48 (0.85) n = 35 4.05 (1.59) n = 32 4.05 (1.24) n = 31

The effects of Voice, Crisis response strategy, and Culture on Word-of-mouth intention For the three-way univariate analysis of variance with Voice (CHV vs. CV), Crisis Response Strategy (IA vs. CA) and Nationality (Spanish vs. Dutch) as factors, Levene’s test of equality of variances turned out to be significant (F (7,255) = 3.45, p = .002). However, due to sufficient and relatively equal sample sizes of Dutch and Spanish respondents, homogeneity of variances can still be assumed. The three-way analysis of variance showed no significant main effect of Voice on word-of-mouth intention (F(1,255) < 1). No significant main effect of Crisis response strategy on word-of-mouth intention was found (F(1,255) < 1). Nationality was not found to have a significant main effect on word-of-mouth intention (F(1,255) < 1). Moreover, no significant interaction effect of Voice and Crisis response strategy on word-of-mouth intention was found (F(1,255) < 1). No significant interaction effect of Crisis response strategy and Nationality was found (F(1,255) < 1). Furthermore, no significant three-way interaction effect of Voice, Crisis Response Strategy and Nationality on word-of-mouth intention was found (F(1,255) < 1). Table 3 shows means and standard deviations of word-of-mouth intention in function of Crisis response strategy, Voice, and Culture.

While a weak significant interaction effect of Voice and Nationality on word-of-mouth intention occurred (F (1,255) = 4.41, p = .037, η2 = .017), subsequent one-way analyses of variance showed no significant effect of Voice on perceived organizational reputation for the Dutch respondents (F (1,122) = 3.94, p = .050, η2 = .031) or of Voice on perceived organizational reputation for the Spanish respondents (F (1,137) = 1.45, p = .231, η2 = .010).

Table 3. Means and standard deviations (between brackets) of word-of-mouth intention in function of Crisis response strategy, Voice, and Culture (1 = low word-of-mouth intention, 7 = high word-of-mouth intention)

Dutch (n = 124) Spanish (n = 139) IA CA IA CA CHV 2.57 (1.25) n = 28 2.74 (1.48) n = 33 2.53 (1.56) n = 40 2.38 (1.42) n = 36 CV 2.30 (1.00) n = 28 2.17 (1.08) n = 35 2.91 (1.91) n = 32 2.65 (1.45) n = 31

The effects of Voice, Crisis response strategy, and Culture on purchase intention

A three-way univariate analysis of variance with Voice (CHV vs. CV), Crisis Response Strategy (IA vs. CA) and Nationality (Spanish vs. Dutch) as factors showed no significant main effect of Voice on purchase intention (F(1,255) < 1). No significant main effect of Crisis response strategy on purchase intention was found (F(1,255) < 1). Nationality was not found to have a significant main effect on purchase intention (F(1,255) < 1). Moreover, no significant interaction effect of Voice and Crisis response strategy on purchase intention was found (F(1,255) < 1). Neither was a significant interaction effect found for Voice and Nationality (F (1,255) = 1.37, p = .243, η2 = .005) or Crisis response strategy and Nationality (F(1,255) < 1) on purchase intention. Finally, no significant three-way interaction effect of Voice, Crisis Response Strategy and Nationality on purchase intention was found (F(1,255) < 1). Table 4 shows means and standard deviations of purchase intention in function of Crisis response strategy, Voice, and Culture.

Table 4. Means and standard deviations (between brackets) of purchase intention in function of Crisis response strategy, Voice, and Culture (1 = low purchase intention, 7 = high purchase intention) Dutch (n = 124) Spanish (n = 139) IA CA IA CA CHV 2.77 (1.27) n = 28 2.50 (1.40) n = 33 2.55 (1.36) n = 40 2.57 (1.28) n = 36 CV 2.44 (1.10) n = 28 2.27 (1.08) n = 35 2.76 (1.55) n = 32 2.58 (1.53) n = 31

Conclusion and discussion

The present study investigated the effects of crisis response strategy (individual accommodative vs. collective accommodative) and tone of voice (conversational human voice vs. corporate voice) in crisis communication messages on Instagram on Dutch and Spanish consumers’ perceived organizational reputation, word-of-mouth intention, and purchase intentions. To answer the general research question, several hypotheses were formulated.

The first hypothesis (H1) expected that Spanish consumers would generate a more positive perception of the organization’s reputation, would have a higher word-of-mouth intention and would pursue a higher purchase intention than Dutch consumers would when an organization used a collective accommodative crisis response strategy in an Instagram post. No support for this hypothesis was found. Spanish consumers did not react differently than Dutch consumers to collective accommodative crisis response strategies, although they were shown to be more collectivistic than Dutch respondents in this study.

The second hypothesis (H2) predicted a reverse effect, namely that individual accommodative crisis response strategies would lead to a more positive perceived organizational reputation, a greater word-of-mouth intention, and a higher purchase intention for Dutch consumers than they would for Spanish consumers. Results showed no support for this hypothesis. It was found that, although the Dutch respondents scored higher on the individualism scale, a more individual accommodative crisis response strategy did not lead to more positive outcomes on the dependent variables for them than for the Spanish respondents in this study.

These results are not in line with previous research. A study that was done by Huibers and Verhoeven (2014), among Dutch consumers, found that the threat to an organization’s reputation was smaller when an individual accommodative web care strategy was used than when a collective accommodative strategy was used. However, in the current study, an individual accommodative crisis response strategy did not lead to a more positive perceived organizational reputation than a collective accommodative crisis response strategy did for Dutch consumers. Neither did the Dutch elicited more positive consumer reactions when they were exposed to an individual accommodative strategy than Spanish respondents did. While comparisons between the study done by Huibers and Verhoeven (2014) and the present study should be done cautiously because different dependent variables were used (threat to

organization’s reputation in contrast to perceived organizational reputation), results could still be considered divergent.

In conclusion, looking at the first and second hypotheses, the present study showed that the degree of individualism/collectivism of Spanish and Dutch consumers, with the Dutch being more individualistic and the Spanish being more collectivistic, might not translate directly to more positive responses of Dutch or Spanish consumers when either an individual accommodative or a collective accommodative crisis response strategy is employed in a crisis communication message on Instagram.

However, a possible explanation for these results might also be the operationalization of the individual accommodative and the collective accommodative crisis response strategies in this study. More specifically, the manipulation for crisis response strategy might not have been strong or clear enough since respondents did not distinguish clearly between the two types of strategy. Another plausible explanation is that the classification of crisis response strategies into individual accommodative strategies (information, sympathy, and compensation) and collective accommodative strategies (apology and corrective action) might not have been sufficiently related to individualistic or collectivistic characteristics as they have been identified by Hofstede’s (2003) cultural dimensions. Since this study tried to link Dutch and Spanish individualism scores, based on Hofstede’s cultural dimension of individualism-collectivism, to individual and collective accommodative response strategies, it was expected that individual accommodative and collective accommodative crisis response strategies could be related to individualism-collectivism as defined by Hofstede (2003). A plausible explanation would be that respondents, for example, considered sympathy, where the organization showed sympathy for the stakeholder, and apology, where the organization deeply apologized for the situation the stakeholder is in, to be quite similar in the present study. It might be questionable to say that showing sympathy for the stakeholder is more individualistic than apologizing. Possible future studies that will relate individualism-collectivism with individual accommodative and collective accommodative crisis response strategies should research further whether, for example, a sympathy strategy is indeed more individualistic than an apology strategy.

Finally, the third hypothesis (H3) expected the use of conversational human voice, compared to corporate voice, to lead to more positive consumer responses towards organizational crisis communication, regardless of culture. Again, results showed no support for this hypothesis. Messages in which a conversational human voice was used, compared to a corporate voice, did not generate a more positive perceived organizational reputation, higher

word-of-mouth intention, or higher purchase intention. This was the case for both Spanish and Dutch consumers. This suggests that employing a conversational human voice in Dutch and Spanish crisis communication messages on Instagram might not be more effective than a corporate voice, or vice versa, in terms of perceived organizational reputation, word-of-mouth intention, and purchase intention. However, similarly to the first and second hypotheses researching crisis response strategies, the manipulation of tone of voice was insufficient. This should be taken into account when interpreting these results and makes it harder to draw conclusions about the effects of tone of voice on Dutch and Spanish consumers’ responses to crisis communication messages on Instagram.

These results contradict with most of the findings from the few previous studies that investigated the effects of a conversational human voice on consumers’ reactions to organizational crisis messages, which mostly showed positive effects of a conversational human voice on the dependent variables (e.g. Kelleher, 2009; Sweetser & Metzgar, 2007; Van Noort & Willemsen, 2011). While, in a study by Park and Cameron (2014), which researched crisis communication on blogs, a conversational human voice increased consumers’ word-of-mouth intention and purchase intention, it did not have an effect on perceived organizational reputation. In contrast, a conversational human voice led to a smaller threat to a corporation’s reputation in a study about crisis communication on Twitter (Huibers & Verhoeven, 2014). While it is difficult to draw conclusions about the results of this study due to the insufficient manipulation of the variables, the diverging results between the various studies done on online crisis communication might show that it might not be necessarily possible to extend studies on crisis communication on online platforms, such as organizational blogs and Twitter to other social media platforms, like Instagram. Since results from this study, the study of Park and Cameron (2014) and the research by Huibers and Verhoeven (2014) are mixed, it still remains unclear whether an organization in a crisis for which it is highly responsible should communicate in a humane and personalized manner or in a formal and distanced manner with its stakeholders on social media. Moreover, it should be considered that the studies from Park and Cameron (2014), Huibers and Verhoeven (2014), and the current study are difficult to compare since contextual factors of the studies, such as the sample, the type of crisis and the type of organization highly differed. Furthermore, while social media are seen as relatively informal and as media on which two-way communication between the organization and the stakeholder is said to be highly important (González-Herrero & Smith, 2008), the seriousness

such as a conversational human voice. Alternatively stated, this means that, while consumers would usually be likely to react more positively to a message with a conversational human voice on Instagram due to the dialogic and informal nature of social media, they might possibly expect a more formal approach, such as a corporate voice, from an organization when it communicates about a serious crisis that could affect the consumers seriously as well. Consequently, it might be likely that the positive effects of informal communication on social media might be diminished when an organization communicates about a serious situation, such as an organizational crisis, on a social media platform, such as Instagram.

By investigating these hypotheses, this study’s general research question could be answered as well. It was asked to what extent crisis response strategy (individual accommodative vs. collective accommodative) and tone of voice (conversational human voice vs. corporate voice) would have an effect on Dutch and Spanish consumers’ reactions to organizational crisis communication on Instagram, in terms of perceived organizational reputation, word-of-mouth intention and purchase intention. As discussed above, no significant effects of crisis response strategy, tone of voice, or culture on the dependent variables were found. In other words, Dutch and Spanish consumers’ perceived organizational reputation, word-of-mouth intention and purchase intention did not differ depending on which crisis response strategy or tone of voice was used in the Instagram post. Neither did the Dutch and Spanish consumers between themselves differ in their reactions towards the messages. Although the interaction effect of tone of voice and culture was found to be significant for perceived organizational reputation and word-of-mouth intention, separate analyses showed that the effects of tone of voice and culture on the dependent variables were not strong enough to conclude that a conversational human voice, which focuses more strongly on the individual, would work better in a more individualistic culture, like the Netherlands, and a corporate voice, which emphasizes the collectiveness of the organization, would lead to more positive consumer responses in a more collectivistic culture, such as Spain.

As discussed above, the manipulations of crisis response strategy and tone of voice were both insufficient in the present study, which limits drawing conclusions from the results of this study. However, it is also important to consider other plausible explanations for the insignificant effects of crisis response strategy and tone of voice on Dutch and Spanish consumers’ responses to crisis messages on Instagram. To begin with, the results of this study might be caused by choosing Instagram to post the crisis messages on, which is a platform that is largely based on individual profiles, and thus generally tends to be rather individualistic. Therefore, Spanish

consumers’ high levels of collectivism might be suppressed when they scroll on Instagram and they might not focus on a group feeling in this particular setting since it is a rather individualistic platform anyway. Moreover, Spain and the Netherlands are both part of the European Union. Since globalization has caused an increasing internationalism and cross-cultural influx and influence in both countries, it might be the case that the two countries are converging to some extent. Although this study has shown that Dutch consumers were significantly more individualistic than Spanish consumers, the differences in terms of individualism might not be large enough to result into different consumer responses to crisis communication on Instagram. However, this was the case in previous studies that researched countries geographically further away from each other, such as the one by An et al. (2010) which studied US and South Korean consumer responses to crisis communication messages.

Limitations

Like any research, this study also had some limitations. To start with, this study’s limitations included some methodological issues. Firstly, the reliability of the scales used for crisis response strategy (manipulation check) and individualism, measured by Cronbach’s alpha, was questionable. This could have possibly affected the accuracy of the results.

Secondly, as discussed above, the manipulation checks for crisis response strategy and tone of voice were found not to be significant, which suggested that the different conditions did not distinguish the different levels of the independent variables as intended. This could be a plausible explanation for the lack of significant effects in this study. The operationalizations of the variables were based on Huibers and Verhoeven (2014) and Park and Cameron (2014) and were proven to be effective in those studies. However, for both crisis response strategies and tone of voice, a difference was made between individualistic and collectivistic versions. More specifically, both conversational human voice and individual accommodative crisis response strategies focused on communicating directly with the individual stakeholder. For example, a message with a conversational human voice addressed the stakeholder personally and an individual accommodative strategy compensated the individual stakeholder. Similarly, corporate voice and collective accommodative crisis response strategies both emphasized the organization and the stakeholders as a group. For future studies, therefore, pretesting of the manipulations is recommended.

Furthermore, another potential limitation concerned the characteristics of the respondents of the present study. To begin with, Dutch and Spanish respondent groups were not equally distributed in terms of educational level. However, it should be considered that similar educational levels were chosen, but Dutch and Spanish educational levels are still difficult to compare. Moreover, the respondents viewing the four different conditions significantly differed in terms of age. Since the unequal distribution of respondents in terms of educational level and age might have impacted the results of this study, future studies should try to create equal distributions of participants among groups.

Moreover, some sampling limitations should be discussed. Many Spanish respondents were found through Facebook groups of Spanish people living in the Netherlands. Consequently, it might be the case that these respondents, although having the Spanish nationality, have adapted to Dutch culture since they might have been living in the Netherlands for a long period of time. In such a way, they might be used to Dutch, possibly more individualistic, communication styles.

Finally, while a fictitious organization and crisis were chosen to ensure that respondents did not have a prior reputation about the case, the respondents might not have felt affected by the crisis as much as they would when they would see the Instagram post in their real lives. Although the respondents in this study were shown to use cosmetics products frequently, they might have not experienced a similar situation before and possibly could not totally identify with the crisis situation. In future studies, pretesting could control for such confounds.

Suggestions for further research

To begin with, further research could solve this study’s problem concerning the manipulation of the independent variables tone of voice and crisis response strategies. In terms of individual accommodative and collective accommodative crisis response strategies, not many studies have investigated and operationalized these constructs. Future research could distinguish between these different types of accommodative strategies more clearly and could clarify how to operationalize or manipulate them. For example, instead of classifying information, sympathy and compensation as individual accommodative strategies, and apology and corrective action as collective accommodative strategies, it might be useful for future studies to distinguish solely between communicating to the individual stakeholder only or to the group as a whole. Thus, the company could compensate the individual in the individual accommodative condition or all