Method

Zusammenfassung

Four different research questions guide this research project. A mixed-method research design with qualitative and quantitative empirical methods was chosen since it enables an in-depth understanding of the phenomenon of corporate diplomacy. This thesis used a triangulation of methods to study corporate diplomacy on the organizational, media, and general audience levels. This chapter presents each method and the procedure of every study in detail.

Four different research questions guide the current research project (see section 1.2, Figure 1). A combination of qualitative and quantitative empirical methods was chosen to answer the research questions since it enables an in-depth understanding of the phenomenon of corporate diplomacy. Mixed-method designs are assumed to add breadth and complexity to the study, enriching the knowledge and insights gained from its findings (Denzin, 2012). Particularly the mix of qualitative and quantitative methods is increasingly used in social sciences since it allows the application of multiple perspectives, resulting in meta-interpretations (Olsen, 2004). This thesis used triangulation of methods to study corporate diplomacy on the organizational, media, and general audience levels. This chapter presents each method and the procedure of every study in detail.

6.1 In-Depth Interviews: The Organizational Perspective on Corporate Diplomacy and Legitimacy

This chapter is derived in part from an article published in Journal of Public Relations Research by Marschlich, S., & Ingenhoff, D. (2021a). The Role of Public Relations in Corporate Diplomacy: How Relationship Cultivation Increases Organizational Legitimacy. Journal of Public Relations Research, available online: https://www.tandfonline.com/doi/full/10.1080/1062726X.2021.1981332 and in Public Relations Review as Marschlich, S., & Ingenhoff, D. (2021b). Stakeholder Engagement in a Multicultural Context: The Contribution of (Personal) Relationship Cultivation to Social Capital. Public Relations Review, 47(4), 102091, available online: https://doi.org/10.1016/j.pubrev.2021.102091.

To answer the following research questions: (1) To what extent and how is corporate diplomacy in the UAE performed as engagement with its social environment? and (2) To what extent and how is corporate diplomacy in the UAE used to gain organizational legitimacy? — this research took a qualitative approach. A qualitative approach allowed for exploring in depth how practitioners present and interpret their experiences concerning corporate diplomacy, its challenges and particularities related to the specific country context of the UAE, and the role of relationships with actors in the organizational environment. Corporate legitimation is a complex process of social constructions and qualitative research enabled the discovery of how individuals make meaning of actions. Furthermore, it allowed for demonstrating how social patterns, such as institutional logic concerning organizational legitimacy building, evolve and are established (see Berg, 2007; Woods, 1992). Since this thesis is interested in exploring corporate diplomacy embedded in a specific political, cultural, and economic context, a qualitative approach enhanced gaining profound perspectives into the unique conditions and circumstances of corporate diplomacy in the UAE (see Corbin & Strauss, 2008). In-depth interviews were chosen as a qualitative research method as this technique facilitated direct interactions with individuals and allowed for asking open-ended questions and further questioning in case additional insights were necessary (Rubin & Rubin, 2005).

6.1.1 Sample

Both theoretical and convenience sampling techniques were applied to draw the sample. Concerning the theoretical sampling, in the first step, the European countries with the highest investment volumes in the UAE in recent years were selected, which include Switzerland, Germany, France, the United Kingdom, and the Netherlands (The Arab Investment & Export Credit Guarantee Corporation, 2016). In the next step, the 30 largest companies in market capitalization from each country were considered. Thirdly, MNCs that do not operate in the UAE, i.e., that do not have local management hubs in the UAE, were omitted. This procedure led to a total of 83 corporations¹, which were requested to participate in an interview. The heads of the public relations departments were contacted via the email addresses made publicly available on the company websites. In cases where no contact information concerning the public relations department’s head was available, a person was reached that was stated as the contact on the MNCs’ websites. The interview request included specific information on the research study. It stated that the study aimed to interview executives responsible for planning and implementing the company’s societal engagement in the UAE to learn more about MNCs’ societal engagement in the UAE. This remark was made to ensure the interviewees’ suitability to give comprehensive insights into the MNCs’ corporate diplomacy practices. In total, 21 MNCs agreed to take part in the interviews.

Moreover, in terms of convenience sampling, it was decided to request the participation of the chief executive officers (CEOs) of leading international public relations agencies operating in the UAE for interviews. This decision was made since some interviewees from the European MNCs claimed that they were supported by public relations agencies in their conception and execution of local societal activities. Seven major public relations agencies could be identified through initial research and were contacted via email, asking for participation in the interview study. As a result, four CEOs of international public relations agencies agreed to take part in the interview. By combining theoretical and convenience sampling, several different views on corporate diplomacy in the UAE were possible, increasing external validity (Lincoln & Guba, 1985).

The final sample consisted of 25 interviews with public relations executives from seven German corporations, five British MNCs, three Swiss corporations, three Dutch corporations, three French MNCs, and four public relations agencies. Since data collection reached the point of information saturation after these 25 interviews and no new ideas had appeared, it was decided to finish the data collection after the 25 interviews. The interviewed MNCs were part of different sectors (including banking, food and beverages, healthcare, construction, and automotive). The interviewees varied in gender and had different cultural backgrounds. Except for two respondents, all interviewees were non-Emiratis. Further information on the interviewees is not presented at this point. Due to interviewees’ expressed feelings of insecurity in the UAE’s given research context, the interviewees were assured high confidentiality and full anonymization, which did not allow for additional identity-related questions. For this reason, it is also not disclosed which companies were interviewed.

6.1.2 Procedure and Data Collection

Before the interviews started, the interviewees’ informed consent was obtained², and the interviewees were asked for permission to record the interviews. Furthermore, the participants were debriefed on the interview structure, and it was explained what the interviewer was referring to by “local, societal engagement” and other terms used in the interview to ensure external validity (Lincoln & Guba, 1985). Finally, the interviewer emphasized the openness of the interview situation, assuring high confidentiality and the full anonymization of the interview data. This step aimed at providing a comfortable interview atmosphere, allowing the respondents to express whatever they think and feel, thereby enhancing internal validity (see Lincoln & Guba, 1985).

The interviews were guided by an interview protocol³, which was discussed beforehand with academic experts in public relations and strategic communication from a large Swiss university and a large university in the UAE. Moreover, the interview guide was tested and refined with practitioners before the data collection started. Hence, the development of the research instrument included different academic and professional insights from both a Western country and a UAE perspective, increasing the instrument’s validity and reliability. The interview guides for the public relations executives from the MNCs and the CEOs of the public relations agencies differed slightly regarding their reference object (companies: “your company”; agencies: “multinational corporations”).

The interview guide consisted of five parts. The first part included questions on the interviewee’s position, allowing for a short introduction. In the second part, the interviewees were asked about the MNC’s societal engagement in the UAE, i.e., topics and motivations to engage in these activities. The questions did not refer to the term “corporate diplomacy,” as previous research has shown that MNCs are often unfamiliar with this term (White & Fitzpatrick, 2018). Next, the interviewer raised questions concerning the addressed social groups in the UAE and their expectations and asked about the role of engaging with the social groups indicated by the interviewee. The fourth part of the interview focused on the public relations activities employed. The final part of the interview guide included questions on the peculiarities of the company’s societal engagement in the given country context and the role of the MNC’s home country. At the end of the interview, the respondents were asked for further comments concerning the interview topic or the interview situation itself. At this point, and during the whole interview, when new issues emerged, the interviewer raised further questions to fully understand the interviewee’s thoughts. All questions were open-ended to encourage the interviewees to self-report as much in detail as possible. Moreover, all interviews were conducted in English. Since the interviewees and the interviewer have different language backgrounds and English is, in addition to Arabic, the main business language in the UAE (Dorsey, 2018), this approach seemed suitable.

The author of this thesis conducted the interviews that took place in the two most important business hubs in the UAE, Dubai and Abu Dhabi, between January and June 2019. All the interviews were conducted face to face. Except for two interviews, the interviews were recorded and transcribed. For the interviews that were not allowed to be recorded, the interviewer took notes in a detailed manner, which were transcribed immediately after the interview to ensure that most memories were still recallable. The interviews lasted between 37 and 79 minutes and yielded 231 single-spaced pages of data. The audio data and the transcripts were fully anonymized and stored only in the interviewer’s notebook, secured with a password and only accessible by the interviewer. This procedure sought to ensure the confidentiality of the information related to the interviewees.

6.1.3 Data Analysis

The analysis of the interview transcripts followed the approach of qualitative content analysis according to Mayring (2000, 2008). The analysis aimed to summarize the main content of the interviews to investigate how corporate diplomacy is performed in the UAE, what role the engagement of different actors in corporate diplomacy activities plays, how this engagement process looks like, and whether and how companies use corporate diplomacy to gain legitimacy in the host country. The analysis was conducted using the qualitative analysis software MAXQDA.

A content-structuring content analysis following the approach of Mayring (2000, 2008) was conducted, aiming to identify and conceptualize content aspects and systematize them in the material. Accordingly, the different aspects of a topic build the categories and subcategories. Following Mayring (2000, 2008), the content-structuring content analysis included the following steps: 1) deriving main categories from the research question and the interview guide, 2) determining units of analysis, 3) developing new categories and subcategories and their definitions, 4) applying and modifying categories, 5) coding of the entire body of material with the final categories, and 6) presenting the results and interpretation. Hence, to develop the categories, this study followed a deductive-inductive approach (Mayring, 2000, 2008). This procedure allows for identifying emerging topics not covered by the deductively formed categories and finding subthemes.

In particular, the deductive development of the categories was guided by the research questions, the themes of the interview protocol, and the theoretical work, including “corporate diplomacy initiatives and topics,” “motives and objectives,” “social groups addressed/involved,” “societal expectations,” “engagement,” and “particularities in the UAE.” In addition, categories and subcategories were developed inductively based on the data material, including main categories such as “partnerships,” “challenges of engagement,” and “alignment with expectations” and subcategories such as “employee engagement” and “community engagement” as subthemes of “engagement” or “internal expectations” and “external expectations” as subthemes of the category “societal expectations.” In this way, the main categories represent broader themes, whereas subcategories represent aspects of the themes.

Text passages relevant to the research interest were marked in the transcripts, representing units of analysis to assign and identify existing and emerging categories. Text passages could be of varying length and comprise one or more sentences. While using the software, similar text passages were marked in the same color, whereas passages with different references or meanings were marked differently. In addition, memos were placed at marked locations to record ideas about the meanings and to reflect on them later. By re-reading the transcripts, focusing on the marked passages, and comparing them across all transcripts, categories could be revised and refined, and new categories and subcategories could be developed. In this step, the formulation and definition of the categories and their abstraction were frequently considered to ensure internal reliability (Mayring, 2008). When necessary, categories were modified, resulting in a final set of categories and subcategories. Subsequently, the entire body of data was (re-)coded, i.e., each text passage relevant to the research interest was assigned to a category or a subcategory.

In the next step, using MAXQDA, code matrices were created that included the categories and subcategories, the anonymized company name, the text passages, and the memos. The final step was the scientific abstraction of the data material to find similarities and differences in the data and, in this way, to identify specific patterns and indications of different corporate diplomacy approaches, i.e., engagement and legitimation strategies. For the sake of clarity, exemplary, directly cited statements are displayed in the results section.

6.2 Content Analysis: The Media Perspective on Corporate Diplomacy and Legitimacy

To answer the third research question—to what extent and how can the media coverage of corporate diplomacy contribute to organizational legitimacy?—a quantitative content analysis was employed. This method allows for investigating patterns in the media coverage of corporate diplomacy and, in this way, analyzing particular categories that form frame elements, which together build a media frame (see Matthes & Kohring, 2008, for this approach). Moreover, quantitative content analysis has been previously applied to organizational legitimacy research in the media (e.g., Deephouse, 1996; Marberg et al., 2016; Schultz, Marin et al., 2013; Vergne, 2011) showing appropriate applicability.

6.2.1 Sample

Since this study sought to investigate organizational legitimacy constructions from the perspective of the MNC’s host country, in this case, the UAE, the most relevant media outlets in the UAE were chosen. Particularly regarding corporate news, daily newspapers are regarded as a primary information source (see Williams & Delli Carpini, 2011). Thus, the newspapers with the highest reach in the UAE were identified, which are Khaleej Times and Gulf News (Arab Media Outlook 2016–2018, n.d.). Due to their reach, it can be assumed that the chosen newspapers reflect and influence organizational legitimacy perceptions (see Deephouse & Suchman, 2008). Both newspapers are based in Dubai and have published local and international news daily since 1978 (Gulf News, n.d.; Khaleej Times, n.d.). Due to the investigator’s language competencies, the English-speaking versions of the two newspapers were considered. This approach seems appropriate since English is, in addition to Arabic, the UAE’s main language of business affairs (Dorsey, 2018). Moreover, comparing Arabic-speaking and English-speaking media outlets in the UAE, Duffy (2013) has found that the news stories hardly differ except for sensitive issues, which barely apply to corporate diplomacy.

In the first step, this study used the company names as keywords to find news articles about the largest European MNCs operating in the UAE. Several steps were necessary for selecting the largest European MNCs, described in detail in section 6.1.1. The corporation selection process yielded a total of 83 corporations originating from France, Great Britain, Germany, Switzerland, and the Netherlands.⁴ The names of these companies were used as keywords, and due to the number of keywords, each company was searched separately. To access the newspaper articles, the electronic database “LexisNexis” was used, searching for each keyword for the period from January 1, 2014, until December 31, 2019. In the second step, the retrieved newspaper articles were read entirely, and those articles not referring to corporate diplomacy were excluded. As a result, the final sample consisted of 385 newspaper articles reporting on (at least) one of the selected corporations and its corporate diplomacy efforts. The newspaper article represents the unit of analysis.

6.2.2 Categories and Coding Procedure

A coding scheme was developed following an inductive-deductive approach, building on the news articles and previous literature on media frames, legitimacy, and corporate diplomacy. The coding scheme was divided into formal and content categories⁵, and each category was defined in detail, giving coding examples to increase the research instrument’s reliability. The formal categories included the name of the newspaper, the year, and the company name. The content categories were separated into two parts distinguishing between categories analyzing socio-political organizational legitimacy (moral, pragmatic, and regulative legitimacy) and cognitive organizational legitimacy.

First, the coding scheme included several categories reflecting the four frame elements, according to Entman (1993), to analyze the newspaper articles regarding their contribution to socio-political legitimacy. These frame elements are problem definition, causal interpretation, moral evaluation, and treatment recommendation, together building a frame (Entman, 1993). This approach was suggested and applied by Matthes and Kohring (2004, 2008) and is used in an adapted form for this research. Each frame element comprised one or two categories. The frame element “problem definition” consisted of categories coding the corporate diplomacy topic and involved actors conceived as institutional linkage. For instance, the subcategories included public health and education for the category “corporate diplomacy topic” and government or national companies within the category “institutional linkage.” As outlined before, an institutional linkage is conceived as “any explicit reference to or association with state institutions (government, judiciary, law enforcement), international institutions (global institutions and international governments), and/or economic institutions (for-profit institutions)” and can contribute to the socio-political legitimacy of an organization (de Souza, 2010, p. 482). Again, the subcategories were coded with “1” for the occurrence of the topic or linkage or “0” if the news article did not refer to information related to one of the subcategories.

Second, the frame element “causal interpretation” was analyzed through categories coding the evaluation of organizational legitimacy and the dimension of legitimacy. It was coded whether the corporate diplomacy initiative and the corporation were endorsed or challenged and, if so, on which level of legitimacy (moral, pragmatic, or regulative). Following Suchman (1995), when the news article presented the company as contributing to the wider local community, and its corporate diplomacy efforts were regarded as the “right thing to do,” moral legitimacy was coded. When the article outlined that the MNC and its corporate diplomacy initiative provided individual and collective actors with favorable exchanges, it was coded as contributing to pragmatic legitimacy (see Foreman & Whetten, 2002). Lastly, regulative legitimacy was coded when the company and its corporate diplomacy activity were presented as complying with governmental rules or expectations (Diez-Martin et al., 2019).

Third, to analyze the frame element “moral evaluation,” the news article was reviewed regarding who was displayed as a beneficiary of the presented corporate diplomacy initiative, for instance, local community members. Fourth, the last frame element, “treatment recommendation,” was analyzed by the category “suggestion,” which coded whether the news article was supportive or critical of the corporate diplomacy initiative. For instance, when the news article promoted the initiative by encouraging readers to take part in the activity, it was coded as supportive. Following previous scholars, all categories were coded only if applicable since certain frame elements can be absent (Entman, 1993; Matthes & Kohring, 2008). Hence, if an article did not refer to, for instance, a beneficiary, all of the subcategories of “benefit attribution” were coded with “0.” Simultaneously, multiple subcategories could be coded with “1” if more than one subcategory could be applied. Furthermore, all subcategories belonging to the categories of each frame element were developed inductively, building on an initial coding of 20% of the entire body of data.

Moreover, the second part of the coding scheme consisted of categories analyzing cognitive legitimacy. These included the position of the company’s name in the newspaper article and the position of a quotation from a company’s representative. The position of the company’s name and any quotations can increase the visibility of a company, contributing to its cognitive legitimacy (Kennedy, 2008; Lamertz & Baum, 1998). Both categories comprised subcategories that referred to the position in the text (e.g., the first third of the main text) and were coded with “1” if the company was named or quoted in the respective section or with “0” if not. Since the company could be named or quoted in different positions, each subcategory was coded separately.

One coder conducted the content analysis of the newspaper articles, and the intra-coder reliability for each variable was calculated. The study relied on Cohen’s kappa coefficient, as it is considered a standard technique in content analysis for analyzing the proportional reduction of error to measure the reliability of the dichotomous measures (see Wimmer & Dominick, 2000). The reliability of each variable was medium to high, ranging from .64 to 1.0,⁶ with an average coefficient of .82. According to Banerjee et al. (1999), Cohen’s kappa coefficients of .75 or higher indicate excellent agreement, .40 to .75 is fair to good, and a value below .40 implies a poor agreement. The kappa values for the variables in this study indicate an excellent or good agreement (see Banerjee et al., 1999) and imply good-to-high intra-coder reliability (Neuendorf, 2009).

6.2.3 Data Analysis

Since the overall aim of the quantitative content analysis was to find patterns in the news coverage on corporate diplomacy and to examine whether specific media frames exist that contributes to organizational legitimacy building in the media, the data analysis followed an explorative approach. According to Entman (1993), a media frame consists of four frame elements. These elements may be combined in certain ways, resulting in the construction of media frames (Matthes & Kohring, 2008). Following previous studies (Baumann et al., 2003; Semetko & Valkenburg, 2000), exploratory factor analysis and cluster analysis were conducted using SPSS 26. Both statistical methods allow for uncovering structures in data and, in this regard, investigating media frames as a combination of frame elements that share the same characteristics among the analyzed categories.

The analysis of the media frames sought to explore corporate diplomacy news articles and their contribution to socio-political legitimacy types. Socio-political legitimacy reflects how the media cover and evaluate a certain organization and its activities. In contrast, cognitive legitimacy rests on the an organization’s public recognition and visibility in the media (Deephouse et al., 2017; Suchman, 1995). Moreover, individual legitimacy judgments on a cognitive level are related to socio-political legitimacy and can hardly be regarded separately (Bitekine, 2011). For this reason, the media frames identified in the first step of the data analysis were further analyzed regarding their link to cognitive legitimacy. A cognitive legitimacy score variable was created as the sum of the categories “position of the company” and “position of the quotation.” Hence, the more often the company’s name was mentioned in different positions and the more frequently it was quoted, the higher the score of the newly developed “cognitive legitimacy score” variable. Building on the results of the cluster analysis, a new category was created that included the number of the cluster assigned to each case (news article). Next, a one-way variance analysis (ANOVA) was conducted with the cognitive legitimacy score as the dependent variable and the cluster number (representing media frames 1, 2, or 3) as the independent variable. The ANOVA sought to measure the differences between the cognitive legitimacy ascriptions of the identified media frames.

6.3 Experimental Survey: The Audience’s Perspective on Corporate Diplomacy and Legitimacy

This chapter is derived in part from an article published in Public Relations Review by Marschlich, S., & Ingenhoff, D. (2022). Public-Private Partnerships: How Institutional Linkages Help to Build Organizational Legitimacy in an International Environment. Public Relations Review, 48(1), 102124, available online: https://doi.org/10.1016/j.pubrev.2021.102124.

This study examined the influence of corporate diplomacy news with or without institutional linkages on organizational legitimacy by asking the research question to what extent and how do institutional linkages with governmental institutions influence the effects of corporate diplomacy on organizational legitimacy? Applying an experimental design study with a survey, this study is the first to analyze the effects of corporate diplomacy communication on organizational legitimacy. Experimental design studies are an appropriate research design to test hypotheses on the causality between two or more variables and have been widely applied in previous public relations research (e.g., Einwiller et al., 2017; Jiménez-Castillo, 2016; Tao & Song, 2020). Moreover, surveys allow access to individual judgments of an organization (Fombrun, 2007) and have been used to measure organizational legitimacy through an experimental design by previous studies (Bachmann & Ingenhoff, 2016, 2017).

6.3.1 Experimental Design, Procedure, and Sample

This study employed a one-factorial (corporate diplomacy news without/with governmental, institutional linkages) experimental design embedded within a survey. In the first step, the respondents were given information on the research study, including remarks on the full anonymization and confidentiality of the survey data, and were asked to sign an informed consent form.⁷ Then, the participants received information concerning the setting, in which they were asked to imagine reading an article in the newspaper they usually read to avoid any bias regarding the choice of newspaper for the stimulus. In the next step, the participants were randomly assigned to one of the groups, either receiving a newspaper article about corporate diplomacy without or with institutional linkages between the MNC and the local government. After that, the participants were exposed to a questionnaire that included several statements on the perception of organizational legitimacy on different levels (moral, pragmatic, and regulative legitimacy) and the intervening variables (governmental legitimacy, issue legitimacy, and media credibility). Moreover, the questionnaire included a statement on whether the initiative in the newspaper article was performed as a collaboration with the government (manipulation check) and asked for demographic information.

The sample comprised UAE residents that have lived in the UAE for at least five years. The decision on this requirement was made due to the specific population in the UAE, which comprises 80 to 90% expatriates. Hence, it was assumed that the assessment of whether an organization contributes to the local community (in terms of moral legitimacy) or individual interests (in terms of pragmatic legitimacy) or whether it aligns with governmental expectations (in terms of regulative legitimacy) was more likely to be valid if the individuals have lived in the country for several years. A market research company based in Dubai performed the data collection by individually surveying the participants using a tablet-assisted system. The data collection took place in the UAE’s three largest cities, i.e., Abu Dhabi, Dubai, and Sharjah, in November and December 2019.

In total, 199 individuals participated in the study, fully completing the questionnaire. The number of participants among the two experimental groups was almost the same, with 99 individuals randomly receiving the corporate diplomacy news manipulation without governmental, institutional linkages (control group) and 100 participants receiving fictitious corporate diplomacy news with governmental, institutional linkages (experimental group). Overall, participants were between 18 and 60 years old⁸ (24.6% 18–29 years, 30.2% 30–39 years, 30.2% 40–49 years, 15.1% 50–60 years), and 50.3% of the sample were female. Furthermore, most of the participants were expatriates (87.9%), coming from Asia (70.4%), Western countries (12.6%), and Arab countries (5.0%), while the rest of the participants were Emirati (12.1%). Lastly, most participants had lived in the UAE for five to 10 years (45.7%). Around one-third of the participants had lived in the UAE for 11 to 25 years (34.2%), and around one-fifth (20.1%) had been in the UAE longer than 25 years or were born there.

6.3.2 Development of the Stimulus

Given that this study sought to investigate the effects of corporate diplomacy (without/with governmental, institutional linkages) on the organizational legitimacy perceptions of the general audience, a fictitious newspaper article was created.⁹ Newspapers are one of the major sources for business-related content (see Williams & Delli Carpini, 2011) and of high importance for the formation of public legitimacy perception and judgments (Deephouse et al., 2017). Building on the research of news content on the corporate diplomacy activities of European MNCs as part of the content analysis (see section 7.2), it was decided to include a corporate diplomacy initiative of the MNC “Danone,” which engages in the societal issue of public health education and nutrition. Danone is a food and beverage company that is popular across the UAE (see Arabian Gazette, 2018), and researching for news on Danone on the electronic database LexisNexis showed that between 2014 and 2019, no negative news had been disclosed in the Emirati newspapers Gulf News and Khaleej Times. The topic of public health education and nutrition was chosen for two reasons. First, public health is among the UAE’s top priorities (UAE Government, 2018) and is equally important for males and females of several ages. In the UAE, many citizens face diabetes and obesity (Gulf News Report, 2019). Therefore, it can be assumed that public health and nutrition are considered relevant among different socio-demographic groups. Second, the chosen issue of health and nutrition is part of Danone’s core business, and, thus, it can be presumed that the stimulus material seems plausible for the study’s participants. The procedures concerning the chosen context of the stimulus sought to increase the study’s validity.

Embedded within the news story were manipulations presenting either that the corporate diplomacy activity was initiated and performed by the MNC (Danone) alone (the control group without governmental, institutional linkages) or together with the UAE Government as a partnership (the experimental group with governmental, institutional linkages). To manipulate the institutional linkages with the government, the subtitle and the main text included four references to Danone’s collective engagement with the UAE Government in the initiative “Danone for Healthier Life,” as follows:

Ministry of Health collaborates with Danone Middle East celebrating ‘Danone for Healthier Life’ week

Roger Miller, business executive officer at Danone Middle East, commented: “In partnership with the Ministry of Health, our ‘Danone for Healthier Life’ initiative again reached thousands of UAE residents.”

Within the project, initiated by Danone Middle East together with the UAE Government, participants had the opportunity to attend workshops to learn about diabetes.

As Miller added, “By showing how to eat and live healthily, Danone, in collaboration with its partner the Ministry of Health, is committed to supporting the local community to achieve the highest standards of health.”

In the stimulus of the control group, such references to the government were missing:

Danone Middle East is celebrating ‘Danone for Healthier Life’ week

Roger Miller, business executive officer at Danone Middle East, commented: “With our ‘Danone for Healthier Life’ initiative, we again reached thousands of UAE residents.”

Within the project, initiated by Danone Middle East, participants had the opportunity to attend workshops to learn about diabetes.

As Miller added, “By showing how to eat and live healthily, Danone is committed to supporting the local community to achieve the highest standards of health.”

Since the experimental design study examined the effects on organizational legitimacy, the newspaper article included several statements that emphasized the MNC’s dedication to social values and norms in the UAE and the corporation’s commitment to the country, demonstrated through the corporate diplomacy activity as well as the value of the initiative for individual interests. These statements reflected organizational legitimacy on the moral, pragmatic, and regulative levels (see Diez-Martin et al., 2019; Suchman, 1995) and were included in both the control and the experimental groups’ stimulus material.

6.3.3 Measurement, Pre-test, and Statistical Measurement Model Validation

Independent, Dependent, and Mediating Variables

The independent variable in this study was corporate diplomacy with or without institutional linkage. Respondents were randomly assigned to one of the two conditions. The variable was measured as a dummy variable, with “0” for the corporate diplomacy news article without institutional linkages (control group) and “1” for the corporate diplomacy news article with institutional linkages to the government (experimental group).

Three organizational legitimacy levels (moral, pragmatic, and regulative legitimacy) represented the dependent variables. Measures for moral legitimacy were used, slightly adapted from Bachmann and Ingenhoff (2016), and included four¹⁰ items, e.g., “From my point of view, the corporation makes an important contribution to UAE society.” Four newly created items measured pragmatic legitimacy since it had not been analyzed before in the conceptualization as to what degree the organizational actions contribute to individual self-interests (Bitekine, 2011). The newly developed items reflected the impression that the MNC’s initiative contributed to personal interests and benefitted the individual; for instance, “I think the corporation and what it does in its initiative has value for me.” Similar to pragmatic legitimacy, regulative legitimacy had not been measured before in how this thesis conceptualized the variable. Regulative legitimacy was conceived as the extent to which the organizational activities were perceived as fulfilling governmental demands and being accepted by the government (Diez-Martin et al., 2019). Following this, four new items were created, measuring the perception of the contribution of the MNC’s activity to governmental expectations and its assumed congruence with governmental rules, including “In my opinion, the corporation behaves in a way that complies with UAE governmental rules.” The items of the variables were rated on five-point Likert scales (1 strongly disagree, 5 strongly agree).

The measurement for issue legitimacy was developed following Chung et al. (2016), while media credibility was analyzed using a slightly adapted version of Finch et al.’s (2015) measurement. Lastly, governmental legitimacy was measured with a newly developed item, building on the conception of organizational legitimacy by Bitekine (2011), adapted to the government. All the items were rated on five-point Likert scales (1 strongly disagree, 5 strongly agree). Table 6.1 presents an overview of the latent variables, their items, and the sources.¹¹

Table 6.1 Variables with items and sources

Pre-test of Stimulus Material and Measurement

The stimulus material and the measurement were pre-tested in three rounds to increase the study’s validity. First, a group of students at an Emirati university (N = 14) was asked to participate in an item-sorting task, following Anderson and Gerbing’s (1991) approach. For this, the students read the description of each construct, using colloquial expressions and the items developed by the investigator. After that, they were asked to assign each item to the construct they believed matched best and, in the next step, to rate which items indicated the construct best. The item-sorting task is a useful means by which to assess the substantive validity of the measurement (Anderson & Gerbing, 1991). Building on this task, scores for each item were calculated, resulting in a list of good and less suitable items. Furthermore, the students and the investigator openly discussed the stimulus material concerning its clarity, length, and wordings.

Second, the pre-test involved another student sample from a Swiss university (N = 13) that was given the stimulus material, a description of the constructs, and a list of the items that belonged to each construct. Then, they were asked to evaluate each item regarding how easy it was to understand and how much they thought it reflected the construct. The students rated the clarity and suitability of each item individually. Moreover, the items and the stimulus material were openly discussed, and the notes of this discussion, together with the students’ evaluations of the items, provided the investigator with helpful comments and suggestions for improving the items and the stimulus material. In the last step, the pre-test included a discussion with academic experts (N = 17), who were asked to give feedback on the measurement and the stimulus material to assess face validity. For this step, a questionnaire was developed, assessing the items’ appropriateness and the stimulus material’s suitability. The questionnaire included the investigator’s conceptualization of the constructs, a list of the items for each construct, the stimulus material, and closed-ended and open-ended questions to examine the research instrument’s adequateness.

The analysis of the responses, the item-sorting task of the student samples, and the comments of the academic experts led to minor changes in the wording of the items and the stimulus material. Second, the items evaluated as the least suitable were deleted, resulting in four items for each construct of moral, pragmatic, and regulative legitimacy. Lastly, the construct of cognitive legitimacy had been included in a prior version of the questionnaire. However, building on a reconsideration of the literature and the comments of the academic experts, it was decided to exclude the construct. The main reason for this decision was that cognitive legitimacy, reflecting a taken-for-grantedness and evolving over a relatively long time, is hard to measure in a one-time survey. Table 6.1 presents an overview of all items and shows which items were deleted as a result of the pre-test.

Measurement Model Validation

A confirmatory factor analysis (CFA) was conducted to test the specification of organizational legitimacy with three factors (moral, pragmatic, and regulative legitimacy) and the measurement model. The model was estimated with SPSS Amos Graphics 25, using maximum-likelihood (ML) bootstrapping on 5,000 samples.

First, the results of the CFA revealed that the one-factor model (all legitimacy items as one factor) did not fit the data well. The overall model fit (χ2(54) = 521.38, p < .001, CMIN/DF = 4.43), the root mean square error of approximation (RMSEA) = 0.13, the comparative fit index (CFI) = 0.80, the normative fit index (NFI) = 0.78, the Tucker–Lewis index (TLI) = 0.75, and the standardized root mean square residual (SRMR) = 0.84 were not adequate as the mentioned values were not in line with the recommended ones (see, Brown, 2015; Hu & Bentler, 1999). The three-factor model (four legitimacy items each factor with moral, pragmatic, and regulative legitimacy being one factor each) led to a more appropriate overall model fit compared to the one-factor model (χ2(22) = 226.26, p < .001, CMIN/DF = 1.8, RMSEA = 0.13, CFI = 0.93, NFI = 0.91, TLI = 0.90, and SRMR = 0.07). Hence, the three-factor model fit the data better and was further used for the data analysis.

Since the overall model fit of the three-factor model with four items per each latent construct (moral, pragmatic, and regulative legitimacy) cannot be regarded as sufficiently well-fitting in light of the presented model fit values (Brown, 2015; Hu & Bentler, 1999), the lowest standardized parameter estimates and the largest error covariances in the modification indices of the model were considered (Brown, 2015; Byrne, 2001). This inspection led to the deletion of the items (“mleg2,” “pleg2,” and “rleg3”) with the lowest standardized estimate in each of the three factors. Furthermore, building on the modification indices, error terms were related to the factors (the error term of “mleg 4” was related to the latent variable “pragmatic legitimacy,” and the error term of “rleg 1” was related to the latent variable “moral legitimacy”), which is reasonable due to the conceptual relationship of the legitimacy factors. As a result, a nested model was built that showed a good overall model fit (χ2(22) = 37.79, p = .011, RMSEA = 0.06, CMIN/DF = 1.8, CFI = 0.99, NFI = 0.98, TLI = 0.98, and SRMR = 0.03) (Brown, 2015; Hu & Bentler, 1999).

In the nested model, all items were significantly related to their latent variables (p < .001), showing high standardized regression weights (> .83), which indicates a high construct validity (Brown, 2015). Moreover, the latent constructs showed a high reliability (moral legitimacy: Cronbach’s α = 0.91; M = 4.08, SD = 0.79; pragmatic legitimacy: Cronbach’s α = 0.92, M = 3.73, SD = 0.91; regulative legitimacy: Cronbach’s α = 0.92, M = 4.20, SD = 0.76). Next, after the measurement model fit could be verified, discriminant validity was assessed, indicating that the correlations of the items with the items of the corresponding latent construct were higher than with the items of the other two latent variables. Therefore, discriminant validity was shown (Kline, 1998). Finally, convergent validity was examined by investigating the factor loadings of each item on the latent constructs, demonstrating that all items loaded significantly (p < .001) on the proposed factor. Thus, good convergent validity could be asserted (Anderson & Gerbing, 1988). See Table 6.2 for the correlation matrix and Table 6.3 for an overview of the final measurement, including the loadings of each item on its factor and the means and standard deviations for each factor.

Table 6.2 Correlation matrix of the variables

Table 6.3 Measurement model of the three organizational legitimacy factors with factor loadings, means, and standard deviations after validation through CFA