The framing of the Dieselgate crisis
A case study using semantic network analysis
Master Thesis
Graduate School of Communication, University of Amsterdam Master’s programme Communication Science
Corporate Communication
Laura Plank
Student-ID 12266337
Supervisor: Dhr. Dr. Piet Verhoeven Submitted: 26.06.2019
Abstract
This study seeks to identify the most frequent implicit frames, which appear as word-occurrences and are used by the domains automotive organizations’ PR and news media in the Dieselgate crisis. Furthermore, it examines frame alignment between the two different
domains throughout the crisis. Frame theory and frame alignment theory within crisis communication serve as theoretical foundation.
The study was conducted based on the model of semantic network analysis, which enables the detection of various frames within a certain discourse. By using this method, 197 news media articles and 96 press releases, published by automotive organizations were analyzed. The study period is divided into an initial, second and final crisis phase, all dated in between 19th September 2015 and 30th March 2019.
The results indicate frame alignment over time and provide insights about the most frequent implicit frame types used by the news media and the organizations’ communication professionals.
The findings of this study contribute to the theoretical framework of semantic network analysis to detect implicit frames in crisis situations. Moreover, it provides further evidence that crisis frames among news media and PR domains align over time.
Introduction
“The scandal of the year. The Dieselgate transformed our admiration for the
automotive organizations into shame [...].“, runs the headline of the German business journal Handelsblatt (Iwersen, 2018). The decades-long positive reputation of German carmakers experienced a threat within the past four years. Precisely because of the so called
“Dieselgate”, which became public in 2015 and continues to impact the automobile industry as we know it today (Kerler, 2018). This marks a major crisis case, affecting the whole automotive industry, since most of the leading carmakers, including Volkswagen, Audi, Daimler and BWM were involved (Leggett, 2017).
The organizations have been accused of actively manipulating the emission control software of millions of diesel models, in order to fit the standards required by the U.S. Environmental Protection Agency (EPA) , 2016). As the standards for diesel engines in the U.S. are rather high, it constitutes a challenge for whole automotive industry. This is why the affected cars’ software systems were intentional constructed so that they can detect external control programs. Hence, every time the authorities conducted emission controls, results fell under the limitations, but as soon as the car was back on the road, emission rates rose
tremendous (Bovens, 2016).
After the exposure of the emission scandal, the organizations faced legal charges in the United States of America and in Europe and were confronted with moral allegations worldwide (Bovens, 2016). The crisis entailed a public debate about diesel models and its’ negative effect on the environment (Kerler, 2018). Hence, the topic experienced a large international media coverage.
In particular the crisis communication of the organizations, above all Volkswagen’s public reaction, was largely criticized by the media. The Financial Times for instance deprecates in December 2016: “Volkswagen blunders through communications over
emissions scandal – Customers and investors have been left exasperated by carmaker’s public statements”. The article continues to blame the company’s executive management and its communication strategy, as especially the press releases’ and public statement’s lack transparency and are misguiding (Milne, 2015). It therefore appears that Dieselgates’ crisis communication marks an interesting, even though not necessary successful case of crisis communication.
There has been a number of previous research papers about the Dieselgate scandal. Whytas (2016) for instance, analyzed tweets in order to detect the crisis’ impact on
Germany’s national branding. Although she focused on the social media domain and only included one automotive brand, namely Volkswagen in her research (Whytas, 2016). Therefore, an expansion to other domains like media and public relations, as well as more brands are of further interest. Other studies took a closer look into the green washing aspect (e.g. Siano, Vollero, Conte & Amabile, 2017), as well as customer’s perception affected by organizational crisis response (Bowen, Freidank, Wannow & Cavallone, 2018). Siano et al. (2017) analyzed Volkswagen CSR reports and U.S. newspaper headlines addressing the emission scandal. Bowen et al. (2018) also only focused on Volkswagen in their study, but collected data from both the U.S. and Germany. Thus, other automotive organizations than Volkswagen have not been included in previous research. It can be further noticed that solely German news media and press releases from several involved organizations, not just one, have never been analyzed before.
Furthermore, there is no research which addresses the framing of the emission scandal and possible frame alignment between the two domains of organizational Public Relations and news media. Frames are carriers of certain meanings, being used for sense making and reduce the complexity of a crisis situation (Van Der Meer, Verhoeven, Beentjes &
Vliegenthart, 2014). Framing has yet been a central research topic within crisis
communication, with a number of studies having been conducted about frame usage during different crises (e.g. Van Der Meer et al. 2014).
For that reason, the question is whether to identify frame types and possible frame alignment, which will allow for characterizing the framing of the Dieselgate crisis. Thereby the focus lies on implicit frames, which refer to latent patterns of words and are more commonly used in crisis situations (Van Der Meer, Verhoeven, Beentjes & Vliegenthart, 2014).
Therefore the overall research question is the following:
What are the most frequent implicit frames used by automotive organizations and newspapers in the Dieselgate crisis and to what extend do the implicit frames used by the automotive organizations and the news media align?
Theoretical Framework
Dieselgate as a CSR-based challenge crisis
Bowen, Wannow and Cavallone describe the emission scandal as “one of the most far-reaching organizational crises that quickly spread worldwide“ (2018, p.222). Moreover, Dieselgate is categorized by previous research as a scandal, since “the origin is rather obscure, it has no immediate victims and a company can hardly deny responsibility for the cause“, even though one could argue, that the owners of the manipulated cars can be seen as direct victims (Bowen, Wannow, & Cavallone, 2018, p.224). Furthermore, scandals are seen as transgressions with internal and intentional causes and constitute a threat to organizational reputation and stakeholders’ attitudes towards the organization (Bowen, Wannow, &
Cavallone, 2018). Because of the negative effect on stakeholder’s perceptions and reputation, an unexpected crises can harm the financial resources of an organization and its long-term success (Coombs, 2007).
Coombs defines crisis communication “as the collection, processing and dissemination of information required to address a crisis situation“ (2010, p.20). This domain has always been a focal area in communication science and communication management, as effective communication influences the outcome of organizational crises (Van Der Meer et al., 2014). That is why it is important to identify effective and ineffective strategies for crisis response. Besides, it is undisputed that the news media play a crucial role in this field and possess the ability to prevent certain situations from escalation (Van Der Meer & Verhoeven, 2013).
Corporate Social Responsibility (CSR) is an essential part of organizational
management and is officially defined by the European Commission as “the responsibility of enterprises for their impacts on society“ (Coombs & Holladay, 2015, p.146). CSR is linked to organizational crises, especially with regards to reputation, as one of the central pieces of crisis communication (Coombs & Holladay, 2015). In “The Ethics of Dieselgate“, Bovens emphasizes that Dieselgate clearly damaged the reputation of the involved organizations and was a threat to their CSR (2016). Hence, the crisis can be categorized as a corporate social response (CSR)-based challenge. This issue is mainly triggered by groups of stakeholders stating that a corporation is acting in an irresponsible way (Lerbinger, 1997). The emission scandal represents a CSR-based challenge, which eventually lead to a crisis because of its failing organizational crisis management (Coombs & Holladay, 2015). Coombs and Holladay (2015) argue that the linkages between CSR, reputation and crisis enable this escalation.
Framing theory
Framing is an essential part of crisis communication (Gerken et al., 2016). The
concept of framing was first defined by Goffman in the 1970s, who argued that the media use specific perspectives and interpretations to frame messages for its audience (Vlieger & Leydesdorff, 2011). Thereby the - conscious or unconscious - selection by the authors or journalists plays a crucial role (Vlieger & Leydesdorff, 2011).
The central foundation of frames are words, which are selected and presented in a certain way and therefore build a meaningful discourse around the crises. Hence, the co-occurrences and co-absence of words, which build the base of implicit frames, plays an important role (Jonkman & Verhoeven, 2013).
Frames play a key role in sense making and reduce the complexity of crises (Van Der Meer et al., 2014). Van Der Meer et. al (2014) list the formation of an organizational
reputation, the prevention of crisis escalation and the avoidance of public confusion as the main functions of frames during crises concerning organizations.
Jonkman and Verhoeven (2013) distinguish framing from agenda-setting by relating the former to the topics the public focuses on and the salience of some crises. However, framing describes the way different stakeholders perceive a crisis and how their opinions on it are shaped (Coombs, 2007). Hellsten, Dawson and Leydesdorff (2010) argue that frames, as carriers of specific meanings, change and shift over time. By focusing on certain aspects, frames can influence individuals’ attitudes towards a crisis and have an impact on the public’s perception. Thus, in addition to the selection process, the emphasis and the presentation of the crisis becomes an essential part of framing (Gerken, Van der Land & Van der Meer, 2016).
Each crisis is characterized by certain frame types used by organizational crisis managers and by the news media (Coombs, 2007). Generally, these frames tend to differ between the domains of organizational PR and the media (Van Der Meer, Verhoeven,
Beentjes & Vliegenthart, 2014). Organizations’ crisis response strategies aim to establish and reinforce frames. News media frames are crucial, since this is primarily how stakeholders receive information about the organization’s crises (Coombs, 2007). Therefore, news media coverage is essential to the success of crisis response strategies (Gerken, Van der Land & Van der Meer, 2016).
Implicit frames
Most research analyzes frames, which are embedded within a text. Vlieger and Leydesdorff (2011) explain that these textual frames are created through key words and the relations between them. Furthermore, in this context it can be differentiated between implicit and explicit frames.
The directly observable, but less common type are explicit frames (Van Der Meer, Verhoeven, Beentjes & Vliegenthart, 2014). These refer to word choices made by authors to frame issues in a specific way (Jonkman & Verhoeven, 2013). Opposed to this, implicit frames are described as latent patterns of words and are not directly observable manually (Gerken, Van der Land & Van Der Meer, 2016). They “co-occur between sentences and paragraphs in one media text or between different texts in different media outlets or at different moments in time.“ (Jonkman & Verhoeven, 2013, p.2).
Furthermore “implicit frames shed light on the latent dimension of communication content and on the meaning that these co-occurring words create“ (Jonkman & Verhoeven, 2013, p.2). Therefore, implicit frames are the main field of interest, as the research on these frames allows to depict frame changes within certain time periods (Van der Meer et al., 2014).
Previous literature in communication science introduces framing as „a leading theoretical concept for studying latent meanings of observable messages in their context“ (Vlieger & Leydesdorff 2011, p.2). Analyzing the latent dimensions in electronic messages, such as online news media, press releases, social media, etc. has already been the focus of former research (Vlieger & Leydesdorff, 2011). This becomes especially important with regards to the rise of modern communication technologies, which causes a “duality of structure“, as well as that the “analyst shifts his / her attention from the communication of information in observable networks to the communication of meaning in latent dimensions“ (Vlieger & Leydesdorff 2011, p.2). To identify implicit frames as word clusters within the digitally published press releases and online articles, the method of semantic-network is useful, as it enables to detect various frames within a discourse (Jonkman & Verhoeven, 2013).
Semantic-Network Analysis
There are different empirical methods that can effecitovely analyze implicit frames. One of them is the computer-assisted approach (Vlieger & Leydesdorff, 2011).
The automated semantic network analysis is part of the computer-assisted approach. Computer programs are used to detect word occurrences, creating semantic maps with shared meaning. Latent structures in a number of texts can be discovered using semantic networks. This can be done without manually coding and guarantees more objectivity of the research (Vlieger & Leydesdorff, 2011). The methodological approach enables the coder to analyze the development of implicit frames, as well as crisis framing and is a derivative of the social-network analysis (Hellsten, Dawson & Leydesdorff, 2010). Semantic-social-network analysis is established within the domain of crisis communication and was successfully used in previous research (e.g. Van der Meer et al., 2014).
The method proceeds without predefining frames and rather operates in an explorative manner, which has the benefit of detecting unexpected implicit frames as word clusters (Van der Meer et al., 2014). This is especially suitable to identify implicit frames, since they are more difficult to detect (Gerken, Van der Land & Van Der Meer, 2016).
“The semantic maps approach measures the meanings of words in their contexts, and results in visualizations of word networks“ (Hellsten, Dawson & Leydesdorff, 2010). An essential part of the semantic network analysis are word/document matrices. These matrices can be visualized as networks, which present the interrelationships among the word frames (Vlieger & Leydesdorff, 2011). Hence, the method enables the coder to structure a network on the basis of shared meaning and builds a map model of related words (Van der Meer et al., 2014). These words (co-)occurrences, presented as maps in semantic fields, allow a
comparison between implicit frames (Hellsten, Dawson & Leydesdorff, 2010).
Frame alignment
The ability to analyze if, when and where the observed frames align is essential for the comparison of implicit frames. Crisis-frame alignment is said to “imply an increase in frame linkage, congruency and complementarity between the domains“ (Van Der Meer, Verhoeven, Beentjes and Vliegenthart, 2014, p.752). Moreover, this phenomenon is a possible indicator of crisis-response effectiveness (Gerken, Van der Land & Van Der Meer, 2016).
The framework used by Van Der Meer, Verhoeven, Beentjes and Vliegenthart (2014) identifies three different crisis phases, the initial, the second and the final phase.
The focus lies on implicit frames, which refer to latent patterns of words used by the automotive organizations and by the news media (Van der Meer et al., 2014).
The frame alignment theory by Van der Meer et al. (2014) suggests a shift in the frame alignment throughout the different crisis phases. In the first phase, just after the crisis
happens, the frames used by the different domains are highly likely to vary. As the news media domain and the domain of the organizational PR “may use the same information or codes, but they differ in terms of their criteria which underpin the selection of relevant information and meaning provisions to words“ (Van der Meer et al., 2014, p.752). Collective sense making plays a significant part in the second crisis phase, after the domain specific frame-building takes place (Van der Meer et al., 2014). With the aim to reduce complexity and uncertainty, frame alignment eventually emerges. In the third and final crisis phase, the domain-specific frames dissociate again and a de-alignment is observable (Van der Meet et al., 2014).
Overall, research shows that the implicit frames of PR, news media and public domains align over time (Gerken, Van der Land & Van Der Meer, 2016).
Previous studies denote the alignment as “temporal necessity“, launched by the event of an organizational crises (Van der Meer et al., 2014, p.760). After the collective sense making the frames get back to the domain-specific concerns and de-alignment takes place (Van der Meer et al. 2014).
The first set of sub-questions aims to identify frame types, used by the automotive organizations and the news media. Moreover, these questions will try to assess shifts among the frame types used in the three different crisis phases, defined by Van der Meer et al. (2014).
RQ 1.a Which frame types used by automotive organizations and news media are present in the first phase of the Dieselgate crisis?
RQ 1.b Which frame types used by automotive organizations and news media are present in the second phase of the Dieselgate crisis?
RQ 1.c Which frame types used by automotive organizations and news media are present in the third phase of the Dieselgate crisis?
The second set of sub-questions will analyze frame alignment in the three different crisis phases of the Dieselgate.
RQ 2.a Do the frames used by automotive organizations and news media vary in the initial phase of the Dieselgate?
align in the second crisis phase of the Dieselgate?
RQ 2.c Do the frames used by automotive organizations and news media differ in the final phase of the Dieselgate?
Method Sample
The sample consists of press releases from different automotive organizations and articles about the Dieselgate published by several news media. The chosen automotive organizations are Volkswagen, Audi, BMW and Daimler. All organizations were actively involved in the crisis and represent a major part of the German automotive industry.
The selected press releases and statements were distributed among the three crisis phases. To get insights into the development of the implicit frames, the dates of the crisis phases had to be defined first. Van der Meer et al. (2010) determine the implicit crisis phase as the time period, where news media and PR cover the issue for the first time. The first crisis phase of Dieselgate started on 19th September 2015, the day when Volkswagen was publicly accused of manipulating emission data and ended on 28th September 2015, after a statement of the former VW CEO Winterkorn, followed by his resignation and the confession that other brands, for example Audi are affected as well (NDR, 2018). The second crisis phase occured between 29th September 2015 and 1st March 2018. This phase includes “extensive crisis communication by all [...] domains“ (Van der Meer et al., 2014, p.755). During the second period further allegations against VW, Audi, BMW and Daimler occurred. For that reason, both the news media and the organizational PR published a major amount of articles, press releases and statements. The final crisis period marked the time after 1st of March 2018 until 30th March 2019 (NDR, 2018).
In order to collect the organizations’ sample, press releases and statements, associated with the Dieselgate from September 2015 the beginning of the crisis until March 2019 -which demonstrates the final phase - , were systematically tracked. The text material was collected from the organizations’ websites, in particular from their press and media homepages.
In total 96 press releases and statements, as well as transcribed management speeches were found. The main criteria was that the text materials cover the Dieselgate or a topic related to the crisis.
The sample of news media coverage consists of newspaper articles about the emission scandal published by quality national newspapers and journals with the largest circulation in Germany. Only German newspapers and journals were chosen, simply because all automotive organizations affected by the Dieselgate are based in Germany. Therefore, German news media were closest to the source of the crisis. US newspapers for instance often referred to German news media articles when they cover the topic (e.g. Kerler, 2018). The selected news media were the daily national newspapers Die Zeit, Die Welt and Süddeutsche Zeitung. Additionally, the content of two quality journals Der Spiegel and Wirtschaftswoche were chosen.
All news media either have a special economy department or are specialized on economic topics in general (e.g. Wirtschaftswoche). The selected articles were found on the newspapers’ and journals’ publicly accessible webpages. The final sample consisted of 197 news media articles, dated from September 2015 until the end of March 2019. The articles were distributed between the three phases of the crisis, which have been defined previously. The articles were selected through random sampling. The main criteria was that all stories cover the Dieselgate or a topic related to the crisis.
Overall, the final sample was made up of a total of 293 news media articles and organizational press releases and statements, all written in German and freely available online.
Research Method
The chosen method of analysis was semantic network analysis, allowing the
researcher to analyze the implicit frames used by the automotive organizations and the news media during the Dieselgate coverage. The purpose of choosing the semantic-network analysis was to reveal similarity between words, followed a factor analysis of the similarity matrix. The aim was to visualize a map showing the positions and the strength of words linkages in the respective network (Hellsten, Dawson & Leydesdorff, 2010).
The semantic network analysis helped to identify implicit frames as word clusters. The goal was to build a network based on shared meaning (Van der Meer et al., 2014). The
semantic network analysis helped to measure frame alignment between organizations’ PR communication and newspaper articles by following several steps which are explained in the following. The analysis was conducted in accordance to the method described by Vlieger and Leydesdorff (2011).
After collecting the sample material, the text files had to be combined into a separate text documents for each domain (organizational PR and news media) and research period (initial, second and final phase), which made a total of six text documents. By doing so, word frequency lists for each domain and research period could be created with the computer program FrequencyList. Hence, six different lists of the most frequent words of each domain and research period emerged.
Literature suggests not to include more than 75 words into one frequency list, since this would make the visualization of the network too difficult and unclear (Jonkman & Verhoeven, 2013). A list of stop words was used to detect and remove irrelevant words, which would negatively influence the results (Vlieger & Leydesdorff, 2011).
In the next step, each frequency list, together with the sample text files of the
respective research period and domain was inserted into the software program FullText. This allowed the creation of word/document occurrence matrices for each phase of the Dieselgate crisis (Hellsten, Dawson & Leydesdorff, 2010). Subsequently, the matrices were used to conduct a factor analysis on SPSS (Jonkman & Verhoeven, 2013). This shows the linkage between the most frequent words and components (Vlieger & Leydesdorff, 2011). Words with a variance of zero were previously excluded and not used for the factor analysis.
As a final step, Vlieger and Leydesdorff recommend to limit the output to six factors in order to simplify the visualization (2011) as a final step. Thus, six components for each domain and research period were derived from the factor analysis. Cronbach’s alpha was used to test the reliability of the components. Components which showed a smaller score than .60 were eliminated (Jonkman & Verhoeven, 2013). The identified components could be
categorized into one dominant frame, with the highest EigenValue (EV) and remaining sub-frames (Van der Meer et al., 2014).
Because there is not a more objective way to label the frames, a qualitative approach was necessary (Leydesdorff & Welbers, 2011). The frames were named after a general term, which connects all words within one frame. The labeling is mainly based on the terms used by Jonkman and Verhoeven’s research , as they have used a similar approach (2013).
In order to visualize the word-occurrence matrix, the network visualization program Pajek creates semantic fields was the last step of the analysis (Jonkman & Verhoeven, 2013). The semantic maps were presented as two-dimensional word networks. Each of the
components, which emerged through the factor analysis, portrays a network of words or in this case nodes. Because there was no other, more objective option, the frame labels had to be
assigned by the researcher (Leydesdorff & Welbers, 2011). To be able to allocate the words to the different frames, the colors of the nodes refer to certain frames. Additionally, all grey nodes represent words that could not be assigned to any frames. The size of the nodes
indicates the frequency of the words. Hence, large nodes refer to more frequently used words. In general, frame alignment can be analyzed by a quantitative, as well as qualitative method (Van der Meer et al., 2014). However, the quantitative approach is fairly rare in communication research so far and was recently introduced by Van der Meer et al. (2014) because of its lack of elaboration this research only focused on the qualitative method.
Since the word clusters were defined through factor analysis and present different implicit frames, the labeling of these frames enables a qualitative comparison of the implicit frames.
Results
The collected press releases and news media articles were not equally distributed among the three crisis phases. Hence, the first crisis phase features only 28 news media articles and press releases, which is significantly less than the number of texts in the other periods. This is mainly due to the relatively short duration of the initial phase, that only lasted ten days. This was especially apparent during the second period, but also during the third period the news media coverage, as well as the organizational communication was more extensive. Beyond that, the frame usage varied significantly when comparing the three phases and the two domains.
The results were presented by creating six semantic maps, one for each domain and research period. For clarification purposes, table 1 provides an overview of the different implicit frames used by the organizations and the media throughout the crisis phases.
A more detailed consideration of the organizational PR’s frame usage in the first period demonstrates that the primary strategy was denial, as the most dominant frame. This frame includes words such as fulfill and lawful requirements. Opposite to this, the second most frequent frame was confession, containing words like aware, consequent, purpose and manipulation. The organizational sub debates were identified as responsibility (with words such as responsible and transparent), as well as external (with words like trust and customers) and internal trust (using the words employees and work), focusing either on customers or employees (see Fig. 1).
News media focused on the incident frame, initially after the crisis happened, by using words like manipulation, USA and manufacturer. This dominant frame is followed by
informing the audience with public information and focusing on words such as knowledge, case and affected. Sub debates covered a variety of domains, that were all affected by the crisis. These could be identified as judicial (words like accusation and rights of use), technical (using e.g. software, specification and emission test) and financial (words such as million and dollar). The sixth sub debate involved industrial words such as automobile manufacturers. Furthermore this frame introduced the term emission scandal (Fig. 2).
The second crisis phase includes a total of 151 press releases and news media articles, published between 29th September 2015 and 1st March 2018.
The most dominant frame used by the automotive organizations during this time period was an economic-ecological frame. This includes the words humans, world,
production and future. The empirical data shows that the sub debates are similar to the ones used by the news media in the first phase, by focusing on technical (words like electrical, liter) and financial (words such as Euro and billions) aspects. Additional sub debates include vocabulary covering the aspects of responsibility, USA / international, as well as public information (Fig. 3).
The results of the factor analysis identified the technical frame, with words like
software and switch-off device) as the main frame used by the news media in the second crisis period. In contrast to the automotive organizations, news media focused on topics about national policy, rather than on international ones, with words such as Germany, German Government and (German) Ministry of Transport. However, most of the sub debates are similar to the ones of the organizational PR. These were assigned as financial, morality and responsibility frames. A topic that was mostly ignored by the organizations’s PR officers, but could be identified as a news media’s sub debate was detection of illegal lab experiments in January 2018. This was framed as an incident using the words monkeys, humans and study) (Fig. 4).
The third and final crisis phase is dated between 1st of March 2018 until 30th March 2019 and presents 114 analyzed texts. In this period the organizational press releases frame the Dieselgate topic - with a focus on future development by using words such as
transformation, development, e-mobility and future. Sub debates are framed in terms of economic (words like organization and segment), ecological (using words such as
sustainability and air quality) as well as a combination of both, assigned as economic-ecological (including words like fuel consumption and service).
Other sub debates cover financial (net cash flow and return on sales) and responsibility (Fig. 5).
The news media discourse in the third phase clearly focused on the events concerning the Audi corporation with words such as Audi CEO, Rupert Stadler, custody and image. For background information, former Audi CEO Rupert Stadler was accused of fraud and arrested in June 2018 and released later in October 2018. The sub debates could be identified as industrial (words like automotive industry and manufacturer), national policy (e.g. German federal court) and judicial (words like prosecutor and compensation). Hence, news media primary covered legal domains in this final period. Further sub debates concerned
responsibility and financial topics (Fig. 6).
Figure 1 organizational discourse initial phase. Yellow nods are the main debate, consecutive sub-debates are in green, red, blue, turquoise and orange. Grey nods are words not connected to a frame according to the results of the factor analysis.
Figure 2 news media discourse initial phase. Yellow nods are the main debate, consecutive sub-debates are in green, red, blue, turquoise and orange. Grey nods are words not connected to a frame according to the results of the factor analysis.
Figure 3 organizational discourse second phase. Yellow nods are the main debate, consecutive sub-debates are in green, red, blue, turquoise and orange. Grey nods are words not connected to a frame according to the results of the factor analysis.
Figure 4 news media discourse second phase. Yellow nods are the main debate, consecutive sub-debates are in green, red, blue, turquoise and orange. Grey nods are words not connected to a frame according to the results of the factor analysis.
Figure 5 organizational discourse third phase. Yellow nods are the main debate, consecutive sub-debates are in green, red, blue, turquoise and orange. Grey nods are words not connected to a frame according to the results of the factor analysis.
Figure 6 news discourse third phase. Yellow nods are the main debate, consecutive sub-debates are in green, red, blue, turquoise and orange. Grey nods are words not connected to a frame according to the results of the factor analysis.
The results indicate, that many frames occurred more than once throughout the crisis. The most commonly used are the frames considering responsibility and financial terms, both of them appeared in each crisis phase and were applied by the two domains of organizational PR and news media. Frames concerning economic and ecological topics were utilized
frequently as well. News media used the incident frame in each of the three phases. In order to answer the second part of the research questions, which aims to detect possible frame alignment between the two domains of organizational PR and news media, the assigned word clusters had to be compared. Therefore a qualitative approach, by analyzing if the previously labeled frames align, was used.
The comparison of the labeled frames in the first crisis phase shows that there was no implicit frame alignment in the time period between 19th – 28th September 2019. While the organizational domain focused on crisis response strategy denial or confession, as well as addressing internal and external stakeholders, the frames used by the news media differed strongly. The focus lay on informing the public about the incident. However, the second period showed that half of the implicit frames were commonly used by both domains. These frames were named as technical, financial and responsibility. Hence, the results identify frame alignment after the first crisis phase. In the third crisis phase two out of the six frames shared the same labeling. These were the frames covering financial and responsibility aspects.
Figure 7 frame alignment between the two domains over time ; x-axis indicates crisis phase, y-axis indicates number of aligned frames
In conclusion the qualitative comparison of the results generated by factor analysis indicates frame alignment over time between the two domains of organizational PR and news media (Fig. 7).
Discussion and Conclusion
The overall goal of this study was to gain knowledge about organization’s PR and news media framing of the Dieselgate, by identifying the most frequent frames, as well as frame alignment. Through analyzing 96 press releases, statements, and transcribed
management speeches, as well as 197 news media articles, the results provide practical as well as theoretical implications.
The semantic network analysis enabled a characterization of the Dieselgate crisis through identifying the most frequently used word clusters as implicit frames. The results present a number of dominant and sub dominant frame types for each crisis phase and domain.
The first crisis phase was shaped by a process of sense making of the automotive organizations and the news media. In accordance to Van der Meer et al.’s (2014) research, initially after the incident happened, both domains framed the emission scandal differently, although using more or less the same information. As the results of this research show, the organization’s strategy after the incident of software manipulation was revealed, was denial. Surprisingly, the strongest sub-debate was confession, which could indicate either a rapid shift in the crisis response strategy, or date from the fact that the analyzed press releases originate from several organizations, which might have used different strategies.
On the other hand, the news media mainly framed the beginning of the crisis in an informative way by providing facts, names and numbers. The introduction of the terms emission scandal and emission affair, which emerged in the early days of the Dieselgate crisis, is significant. Moreover the application of vocabulary, such as manipulation and threat, indicates a negative perception towards the involved organizations.
After the initial information processing, a more collective framing took place. Here, a major part of the frames used by the two domains aligned. In comparison to the semantic maps of the initial (Figure 1 and Figure 2), the second crisis phase (Figure 3 and Figure 4), presented more structured networks, with the main debates of both domains being defined as technical, financial and responsibility.
A recognizable variation was the different geographical focus. While the
organizational PR presented the incident in an international, especially US-American context, the news media centered on Germany. This could be explained by the fact that the generally, the German news media mainly approaches a German audience and therefore national topics are more relevant than issues concerning the USA. The organizations on the other hand operate internationally, plus the United States were the main location of the Dieselgate crisis. Another observable object was the increasing number of automotive organizations being involved in the crisis case. In addition to Volkswagen, BMW and Audi, the second phase further mentioned the Daimler, Opel and Porsche being involved in the crisis case.
In the final phase the implicit frames shifted again, with each domain focusing on different topics, although still aligning in two out of six frames. Future development was the main frame of the automotive organizations. The dominant organizational frames in the third crisis period confirm that the organization’s focus lay on future improvement, closely tied to sustainability. This is shown by frequently used words like electro mobility, transformation as well as air quality and improvement. This result may indicate that they wanted to move forward and drive stakeholder’s attention away from the crisis towards a progressive
movement and an aspiring future. The sub discourses were all about economic, ecological and financial topics.
On the other hand, the organizational behavior was highly criticized by the news media, as the media coverage concentrates on the legal dispute. Significant for the news media is the negative vocabulary, including words such as affair, scandal and manipulation. This indicates what a risk the emission scandal for the involved organizations is and how it could possible damage their reputation on long-term.
The news media set the focus in the third period on legal issues about the crisis,
especially in terms of the allegations against the Audi management. Former Audi CEO Rupert Stadler played a central role here, with his name co occurring together with expressions like custody, image and investigations. Together they built the dominant frame, which indicates the high significance level of the allegations against the Audi management in the third crisis phase. The news media’s sub debates show that after revealing and discussing the issue in the initial and second phase, the crisis case comes to an end with its legal consequences for all involved parties.
With regards to these findings, it is noticeable, that the organizations seemed to skip addressing judicial proceedings, but only mentioning aspiring future developments, rather
than current legal issues. This clear de-alignment is very characteristic for the last crisis phase, as Van der Meer et al.’s (2014) findings indicate as well.
Overall the research question of this study could be answered by identifying the most frequent frames used by automotive organizations and newspapers in the Dieselgate crisis, as well as analyzing the frame alignment between the domains and crisis phases.
The findings of this study match the results of previous research using automated semantic network analysis. Similar to Van der Meer et al.’s (2014) conclusion, a frame alignment over time between the two domains of organizational PR and news media could be discovered, even though this was solely proved by a qualitative, rather than a quantitative approach.
This research can be seen as another additional example for using the concept of semantic network analysis to detect implicit frames, as well as frame alignment within crisis communication. It supports the findings of previous studies (e.g. Hellsten, Dawson & Leydesdorff, 2010) and proves that this method is suitable to use within framing theory.
For this reason, semantic network analysis can be recommended to use in future studies, concerning implicit frames and crisis communication.
The results align with the findings of previous studies about the Dieselgate crisis. Especially with the findings of Siano et al., who discovered a “a new type of green washing hereby defined as descriptive manipulation” (p. 33, 2017). Siano et al.’s (2017) content analysis of U.S. newspaper headlines detects that the concept of corporate fraud and
manipulations plays a main role in news media coverage of the Dieselgate. These results can be confirmed through the present study as well, as terms such as fraud or manipulation occurred frequently within the news media coverage.
Moreover the results match as well with Whytas’ (2016) research. In addition to the significant relationship between crisis topic and location, her study identified Germany’s reputation, Germany’s economy, the German government’s role and lastly the German car industry as some of the main topics of the Dieselgate debate (Whytas, 2016). All of these themes could be detected in this study in form of implicit frames, too. In particular the news media paid a lot of attention towards Germany’s involvement in the crisis by addressing the German government and legal system frequently. Thus, the important role of Germany as the country of origin of the crisis, is undisputed and could be further studied in future research.
First of all, it is necessary to mention that the sample size includes just 293 news media articles and press releases and is therefore rather small and could be increased in the future. This would provide more accurate results.
The main limitation of this research can be found in the lack of being able to
statistically evaluate the frame alignment. Using only the qualitative method causes a risk of interpretative bias (Van der Meer et al., 2014). This is why future research should use a quantitative approach in addition, which would generate more objective results.
Furthermore, this study only focuses on the two domains of organizational PR and news media. A recommendation for studies about framing of the Dieselgate crisis would be to include the public domain as well. This is especially important as most of previous studies regarded the public domain as rather important (e.g. Van der Meer et al., 2014).
To further test the method of semantic network analysis and elaborate the concept of framing in crisis communication theory, additional research about different crisis cases is essential, since this study only addressed one crisis case.
Lastly, some practical implications and recommendations can be made on the basis of this research. Crisis communication is essential to control and solve organizational crises by maintaning good relationships with stakeholders, including the news media. PR professionals have the duty of communicating frequently and transparently to the public. As the results of this study show, even though the implicit frames of the different domains might differ, a collective sense making and frame alignment over time is observable.
Hence, it is essential for organizations to be aware of the communication methods and strategies of the different domains, especially news media and react to those in a strategic manner. By doing so this could improve crisis-response effectiveness and might reduce the negative impact on organizational reputation.
References
Bovens, L. (2016). The Ethics of Dieselgate. Midwest Studies in Philosophy, 262-283.
Bowen, M., Freidank, J., Wannow, S. & Cavallone, M. (2018). Effect of Perceived Crisis Response on Consumers' Behavioral Intentions During a Company Scandal: An Intercultural Perspective. Journal of International Management, 222-237.
Coombs, W. T. (2007). Protecting Organization Reputations During a Crisis: The
Development and Application of Situational Crisis Communication Theory. Corporate Reputation Review, 10(3), 163-176.
Coombs, T., & Holladay, S. (2015). CSR as crisis risk: expanding how we conceptualize the relationship. Corporate Communications: An International Journal, 144-162.
Gerken, F., Van Der Land, S. & Van Der Meer, T. (2016). Crisis in the air: An investigation of AirAsia’s crisis-response effectiveness based on frame alignment. Public Relations Review, 42(5), 879-892.
Hellsten, I., Dawson, J., & Leydesdorff, L. (2010). Implicit media frames: Automated analysis of public debate on artificial sweeteners. Public Understanding of Science, 19(5), 590-608.
Iwersen, S (2018). Fahrverbot statt Fahrspaß: Dieselgate und die weitreichenden Folgen. Retrieved from https://www.handelsblatt.com/unternehmen/management/menschen- des-jahres-2018/skandal-des-jahres-fahrverbot-statt-fahrspass-dieselgate-und-die-
weitreichenden-folgen/23750210.html?ticket=ST-1690001-jkcv0mBzzrEeRPIwbDbU-ap4
Jonkman, J. & Verhoeven, P. (2013). From risk to safety: Implicit frames of third-party airport risk in Dutch quality newspapers between 1992 and 2009. Safety Science, 58, 1-10.
Kerler, W. (2018). You thought Dieselgate is over? It’s not. Retrieved from https://www.theverge.com/2018/9/18/17876012/dieselgate-volkswagen-vw-diesel-emissions-test-epa-german-auto-industry-mercedes-benz-bmw
Leggett, T. (2017). BMW, Volkswagen and Daimler face EU diesel emission probe. Retrieved from https://www.bbc.com/news/business-45558588
Lerbinger, O. (1997). The crisis manager: Facing risk and responsibility.
Leydesdorff, L. & Welbers, K. (2011). The semantic mapping of words and co-words in contexts. Journal of Informetrics, 5(3), 469-475.
Milne, R. (2015). Volkswagen blunders through communications over emissions scandal – Customers and investors have been left exasperated by carmaker’s public statements. Retrieved from https://www.ft.com/content/b9f35440-98ed-11e5-bdda-9f13f99fa654 NDR (2018). Die VW Abgas Affäre: Eine Chronologie. Retrieved from
https://www.ndr.de/nachrichten/niedersachsen/braunschweig_harz_goettingen/Die-VW-Abgas-Affaere-eine-Chronologie,volkswagen892.html
Siano, A., Vollero, A., Conte, F. & Amabile, S. (2017). “More than words”: Expanding the taxonomy of greenwashing after the Volkswagen scandal. Journal of Business Research, 27-37.
Van Der Meer, T., Verhoeven P., Beentjes, H. & Vliegenthart, R. (2014). When frames align: The interplay between PR, news media, and the public in times of crisis. Public Relations Review, 40(5), 751-761.
Van Der Meer, T., & Verhoeven, P. (2013). Public framing organizational crisis situations: Social media versus news media. Public Relations Review, 39(3), 229-231. Vlieger, E. & Leydesdorff, L. (2011). Visualization and Analysis of Frames in Collections of
Message: Content Analysis and the Measurement of Meaning.
Whytas, K. J. (2016). Impact of Brand Crisis on Nation Branding: An Analysis of Tweets about VW’s Emssions Crisis.
Appendix
Appendix A. Factor loadings for factor analysis Phase 1, organizational PR. Only the 75 most used words were used as variables for the analysis.
Component 1 2 3 4 5 6 DIESELANTRIEB ,994 JAHR ,987 BMW ,974 ,115 GROUP ,974 ,115 VORGABEN ,974 ,115 EUROPA ,974 ,115 ABWEICHUNGEN ,956 -,108 -,180 ERFLLEN ,956 -,108 -,180 EU ,956 -,108 -,180 ANFORDERUNGEN ,956 -,108 -,180 GESETZLICHEN ,956 -,108 -,180 GESAMTEN ,848 -,118 ,147 ,472 EINSATZ ,848 -,118 ,147 ,472 AUTO ,773 -,142 ,209 ,557 WELTWEIT ,733 ,168 -,611 FESTGESTELLT ,564 ,282 -,306 ,348 ,135 -,538 KEINERLEI ,533 ,177 -,189 ,201 ,243 -,523 AUDI -,117 -,979 STEFAN -,117 -,979 MOSCH -,117 -,979 ARBEITNEHMERVE TRETER -,117 -,979 DIESELMOTOREN ,117 ,979 KNIRSCH -,117 -,979 ENTWICKLUNG ,126 -,977 HUBER -,175 -,912 -,291 AUFKLRUNG -,239 -,911 ,179 -,160 BERTHOLD -,195 -,846 -,106 -,416 PRESSEKONTAKTE -,648 ,734 MANNSCHAFT -,246 -,559 -,163 ,166 ,329 ,463 KONZERN ,533 ,442 -,316 MITGLIEDER -,107 ,110 ,970 AUFSICHTSRATE -,107 ,110 ,970 AUFGABE -,107 ,110 ,970 KENNTNI -,136 -,135 ,956
BEWUSST -,149 ,103 ,926 WINTERKORN -,185 ,169 ,921 ,132 -,103 PROFESSOR -,159 ,168 ,907 ,141 UNTERNEHMEN ,316 -,156 ,887 -,257 ABGASWERTEN -,185 ,181 ,862 ,101 -,344 MANIPULATION -,185 ,181 ,862 ,101 -,344 STELLEN ,148 -,210 ,726 ,208 -,157 ,466 SITZUNG -,217 ,207 ,649 ,120 -,609 ,134 KONSEQUENZEN -,261 -,163 ,628 ,139 -,610 ,113 FFENTLICHKEIT -,452 ,506 -,542 GENERATION -,190 ,143 -,154 -,931 ,111 MARKE -,190 ,143 -,154 -,931 ,111 PKW -,190 ,143 -,154 -,931 ,111 BETROFFEN -,190 ,143 -,154 -,931 ,111 FAHRZEUGE ,380 ,170 -,189 -,851 ,111 -,108 AKTUELLEN ,525 ,176 -,173 -,713 -,313 ,113 ABZUWENDEN -,186 ,371 -,435 ,553 ,512 -,101 TRANSPARENT -,186 ,371 -,435 ,553 ,512 -,101 BEHRDEN ,172 ,409 -,469 ,524 ,133 ,439 ZUSTNDIGEN ,372 ,426 ,188 ,519 ,480 HOCHDRUCK ,247 -,261 -,519 ,311 -,511 AUFSICHTSRAT -,252 -,256 ,125 -,860 ,165 NOTWENDIGEN -,213 ,198 ,209 ,121 -,845 ,187 AUFARBEITUNG -,225 -,307 -,132 ,121 -,832 ,159 VERTRAUEN -,368 ,465 ,380 ,646 ,188 KUNDEN -,441 ,571 -,127 ,577 ,136 TECHNISCHEN -,239 -,324 -,389 -,250 ,564 UNREGELMIGKEIT EN -,177 ,274 ,374 ,317 ,550 ERKLRUNG -,451 -,301 ,365 ,321 -,511 -,133 SCHADEN -,315 -,374 ,345 ,460 ,503 ,169 PERSNLICH -,280 ,295 -,288 ,371 -,429 -,200 ARBEIT -,227 ,210 -,253 ,277 ,819 MITARBEITER ,429 ,208 -,228 ,188 -,323 ,722 INFORMIERT -,212 ,372 -,397 -,230 ,265 -,657 MENSCHEN -,252 ,251 ,370 ,495 ,652 BITTEN -,291 ,276 ,431 ,268 -,216 ,632 SOFTWARE ,184 -,191 ,245 ,300 -,613 VORSTAND -,257 -,102 ,102 ,474 ,159 -,599 FEHLVERHALTEN ,481 -,138 ,273 ,576 ARBEITEN -,309 ,280 -,357 -,237 ,493 ,559
Appendix B. Factor loadings for factor analysis Phase 1, news media. Only the 75 most used words were used as variables for the analysis.
Component 1 2 3 4 5 6 ICCT ,879 ,279 TEST ,803 -,276 -,291 PRFSTAND ,774 ,206 ,130 ,273 PASSAT ,735 -,202 ,280 STICKOXIDE ,693 -,155 ,118 AUTO ,671 -,143 -,253 ,473 USA ,657 -,126 -,115 ,188 ,290 -,424 FAHREN ,644 ,431 -,152 MANIPULIERT ,634 ,206 ,575 -,230 KILOMETER ,594 -,288 ,124 ,155 -,370 ,126 HERSTELLERN ,591 ,222 ,362 EUROPA ,567 -,136 -,122 ,362 ,142 ABGASWERTE ,534 -,110 ,393 -,121 BMW ,519 -,289 -,177 FEEDBACK ,498 -,197 ,479 -,163 ,309 MANIPULATION ,490 ,185 -,166 ,389 -,248 IMAGE -,399 -,264 -,138 -,194 MODELLE ,381 -,139 -,281 -,127 ,193 DIESELMOTOREN -,321 ,227 -,317 ,165 WINTERKORN -,112 ,874 -,264 -,108 -,147 KONZERN ,829 -,154 ,132 -,240 BETROFFENEN ,817 ,377 ,125 KUNDEN ,792 ,169 ,107 ,132 WISSEN ,122 ,757 ,207 -,195 ,253 ,126 ZWEI ,717 ,174 -,205 VOLKSWAGEN ,111 ,672 -,355 SKANDAL ,268 ,641 ,192 ,571 ,130 WEITERE ,154 ,622 ,404 -,516 ,166 FALL -,184 ,621 -,289 -,147 ,234 UNTERNEHMEN -,230 ,610 -,347 -,281 -,326 -,210 TAGEN -,147 ,584 -,376 -,513 VW ,346 ,491 -,210 ,354 -,373 ABGASAFFRE -,192 ,383 ,310 -,103 DEUTSCHE -,102 -,322 ,204 -,249 ANZEIGE ,459 ,789 -,146 ,155 HERSTELLER ,275 ,745 -,225 ,361
NUTZUNGSRECHT E ,532 ,691 -,142 ,273 KOMMENTIEREN ,525 ,689 -,143 ,145 -,300 ZEIT -,105 ,682 -,162 NEUE ,144 ,677 ,519 -,170 DIESEL -,160 -,147 ,650 -,207 -,170 WARUM ,196 ,105 -,574 ,266 -,210 JETZT -,379 ,104 ,528 ,364 -,296 -,103 DEUTSCHEN -,191 ,515 -,192 -,230 ,176 FOLGEN ,154 ,155 ,494 ,207 ,334 -,267 PRO ,391 -,389 -,414 ,187 -,236 -,126 SAGT -,203 -,206 ,361 JAHR ,105 -,312 ,224 -,279 -,222 ANGABEN ,171 ,895 ,164 MSSEN -,112 ,349 ,473 ,689 -,139 ,155 EURO -,194 -,117 ,680 ,135 ABGASTEST ,450 ,216 -,164 ,673 ,302 -,244 KRAFTFAHRTBUND ESAMT ,121 ,617 ,599 SOFTWARE ,182 -,340 ,592 ,516 ,101 JAHREN -,261 ,580 ,105 ,147 AUTOBAUER -,509 ,561 -,337 -,130 AUDI -,263 ,470 ,449 ,432 PROZENT -,108 ,241 ,280 -,154 -,151 MILLIONEN -,270 ,191 -,266 ,195 ,687 DOLLAR ,148 ,654 -,392 MILLIARDEN -,106 ,502 -,119 ,612 -,188 ALLE -,180 ,599 ,239 -,602 HARNSTOFF ,559 ,297 ,579 FAHRZEUGE ,228 ,145 ,213 ,552 ,259 DROHEN ,152 ,183 ,398 -,377 BUNDESREGIERUN G -,362 FAHRZEUG ,120 -,114 -,232 ,300 ,754 ABGASSKANDAL ,199 -,152 ,711 BETROFFEN ,360 ,320 ,627 AUTOHERSTELLER ,100 -,416 ,522 EPA ,198 -,234 -,185 ,253 ,210 -,504 USUMWELTBEHRD E ,256 -,317 ,393 ,118 -,460 CLEAN ,359 -,305 -,249 -,385 DEUTSCHLAND ,115 ,132 -,105 -,177 ,360 BEHRDE ,117 -,300 -,143 -,156 ,344
Appendix C. Factor loadings for factor analysis Phase 2, organizational PR. Only the 75 most used words were used as variables for the analysis.
Component 1 2 3 4 5 6 MENSCHEN ,940 ,104 WELT ,873 ,156 ,138 DIESEL ,860 ,321 ,185 EUROPA ,848 ,242 ,275 ZUKUNFT ,784 ,245 ,423 JAHR ,765 ,426 ,337 ,129 PRODUKTION ,761 ,370 ,325 ,258 MITARBEITER ,755 -,146 ,271 MOBILITT ,751 ,175 -,112 ,458 KUNDEN ,750 ,255 ,222 ,449 BMW ,626 -,276 -,141 -,185 UNTERNEHMEN ,621 ,141 ,158 -,125 ,187 ,273 GENERATION ,616 ,290 ,430 ,191 -,205 WELTWEIT ,613 ,494 ,125 -,128 -,116 ,378 MODELLE ,572 ,550 ,103 -,131 AKTUELLEN ,564 ,237 -,135 DEUTSCHLAND ,552 ,366 -,132 ,111 -,203 -,107 JAHRE ,509 ,466 ,233 ,374 -,122 EUROPISCHEN ,497 ,292 -,201 BEHRDEN ,448 ,245 ,380 ,128 -,156 KOMBINIERT ,978 KM ,973 GKM ,970 ELEKTRISCH ,117 ,944 A8 ,164 ,939 ,112 -,101 LITER ,842 ,253 AUTO ,467 ,788 ,247 -,160 AUDI ,404 ,780 ,330 -,192 SCHRITT ,353 ,735 ,255 ,269 ,175 -,136 FAHREN ,633 ,725 ,103 ZIEL ,226 ,504 ,268 ,496 ,207 FAHRZEUG ,206 ,486 ,401 ERGEBNI ,881 -,110 OPERATIVE ,809 -,150 -,103 ,234 OPERATIVEN ,333 ,806 ,160 EURO ,290 ,717 -,122 -,106 ,145 MILLIARDEN ,294 ,693 ,276 ,126
ENTWICKLUNG ,352 ,206 ,667 ,126 ,280 MARKE ,220 ,566 ,629 MODELL ,334 ,205 ,618 -,239 PROZENT ,542 ,491 ,615 ,152 MILLIONEN ,301 ,515 ,325 ,271 STADLER ,389 ,428 -,325 ENDE ,103 ,398 ,293 ,159 ,364 QUARTAL ,353 ,210 TECHNISCHEN ,922 BETROFFENEN ,831 ,354 UMSETZUNG ,826 ,111 KBA -,110 ,815 MANAHMEN -,122 ,105 ,729 LSUNGEN ,269 -,100 ,688 ,155 SOFTWAREUPDAT E ,188 ,683 -,101 FAHRZEUGE ,108 -,218 ,632 ,158 INFORMIERT ,137 -,133 ,561 -,106 SOFTWARE ,154 ,392 VORSTAND ,296 -,136 -,375 HAUPTVERSAMML UNG ,173 -,111 -,164 -,289 BETROFFEN -,122 ,287 -,111 -,130 AUFSICHTSRAT -,101 -,133 -,260 ,128 USDOLLAR ,924 ZUSAMMENHANG ,154 ,167 ,838 VOLKSWAGEN -,106 ,179 ,758 ,552 RAHMEN ,213 ,750 ,240 AMERICA ,157 ,228 ,711 -,107 USA ,534 ,126 ,678 DIESELTHEMATIK ,265 -,165 ,639 FAHRZEUGEN -,122 ,394 ,416 ,178 UNTERSUCHUNG -,149 ,342 KONZERN ,195 ,919 MARKEN ,241 ,196 ,240 ,736 STRATEGIE ,619 ,225 ,631 MLLER ,628 ZWEITAUSENDFND UNZWANZIG ,344 ,337 ,332 ,586 PRESSEKONTAKTE -,328 -,198 ,304 ,287 ,397 AUFKLRUNG -,112 ,129 ,221
Appendix D. Factor loadings for factor analysis Phase 2, news media. Only the 75 most used words were used as variables for the analysis.
Component 1 2 3 4 5 6 OPEL ,891 ,105 ZAFIRA ,844 ,142 DIESEL ,797 ,327 ,237 GRENZWERTE ,762 ,271 ,125 ,150 ,234 ABGASREINIGUNG ,762 ABSCHALTEINRICH TUNGEN ,689 ,350 STICKOXIDE ,657 ,397 ,203 UMWELTHILFE ,645 ,193 -,157 AUTOINDUSTRIE ,624 ,254 ,108 -,132 ,151 HERSTELLER ,615 ,413 ,149 MESSUNGEN ,593 ,258 -,167 JAHR ,570 ,141 ,218 ,189 -,164 ,231 EXPERTEN ,545 ,401 -,172 -,133 DIESELAUTO ,541 ,280 ,218 -,163 SOFTWARE ,512 ,376 ,192 -,110 PRFSTAND ,508 ,123 ,505 MOTOR ,449 ,369 ,328 MODELLE ,438 -,102 ,207 ,351 DIESELFAHRZEUG E ,429 ,359 ,114 ,140 ,201 ABGASWERTE ,423 ,192 ABGASTEST ,399 MANIPULIERT ,349 -,109 -,187 ,119 ,186 INDUSTRIE ,329 ,855 AUTOMOBILINDUS TRIE ,853 ,101 ,236 BEHRDE ,819 ,237 REGIERUNG ,118 ,818 ,123 ,101 DEUTSCHEN ,219 ,811 BUNDESREGIERUN G ,360 ,773 DIESELSKANDAL -,111 ,600 ,354 -,110 ,117 DEUTSCHLAND ,275 ,484 ,415 ,182 DAIMLER ,263 ,482 ,143 ,265 AUTOHERSTELLER ,141 ,469 -,147 -,101 EUKOMMISSION ,437 -,126 -,117
SOFTWAREUPDAT E -,103 ,412 ,369 VERKEHRSMINISTE RIUM ,379 ,398 -,160 DOBRINDT ,368 ,397 MANIPULATIONEN ,119 ,341 ,153 ,184 -,174 BMW ,180 ,266 -,163 ,146 MILLIONEN -,114 ,798 ,134 EURO ,112 ,779 MILLIARDEN ,756 ,162 -,104 PROZENT ,186 ,381 ,651 KUNDEN ,538 ,389 ,194 EUROPA ,398 ,118 ,400 ,179 ,140 AUTO ,376 ,317 ,388 ,178 ,304 ANGABEN ,264 ,213 -,201 ABGASAFFRE ,122 ,149 ,101 BETRUG ,146 ,592 KONZERN -,159 ,426 ,582 ,127 VOLKSWAGEN ,542 ,558 MLLER -,170 ,359 ,107 ,548 ,118 VWKONZERN -,125 ,547 ,195 DIESELMOTOREN ,393 ,542 ,138 AUDI ,506 ,193 USA ,368 ,480 -,158 UNTERNEHMEN ,187 ,337 ,173 ,463 -,174 WINTERKORN -,106 ,117 ,426 -,153 -,111 SKANDAL -,137 ,303 ,417 BEHRDEN ,155 ,221 ,269 -,203 BETROFFENEN ,264 ,744 KRAFTFAHRTBUND ESAMT ,102 ,289 -,154 ,737 FAHRZEUGE ,248 ,201 ,173 ,721 RCKRUF -,133 ,113 ,680 KBA ,392 -,132 ,650 NUTZUNGSRECHT E -,123 -,395 ,447 PASSAT ,152 -,104 ,151 ,390 FEEDBACK -,102 -,252 ,389 -,126 MOTOREN ,380 ,109 -,260 ,242 ,385 ABGASSKANDAL -,194 ,219 -,139 PORSCHE ,147 ,182 AFFEN -,121 ,899 STUDIE -,126 ,864 MENSCHEN ,170 -,112 ,819
AUTOBAUER ,216 ,461 ,107 ,156 ,585
ERGEBNISSE ,242 ,120 ,420
Appendix E. Factor loadings for factor analysis Phase 3, organizational PR. Only the 75 most used words were used as variables for the analysis.
Component 1 2 3 4 5 6 TRANSFORMATION ,906 ,325 JAHR ,875 ,313 ,109 ,239 MILLIONEN ,869 KUNDEN ,867 ,262 ,139 MILLIARDEN ,863 ,373 PRODUKTE ,850 ,163 -,145 -,124 AUTO ,842 -,136 ,268 -,157 WELT ,835 ,397 -,186 ARBEITEN ,829 ,139 ,274 ,161 MARKT ,826 ,157 ,307 ,243 ,146 ENTWICKELT ,808 ,263 ,148 -,132 -,102 MITARBEITER ,803 ,242 ,460 FAHREN ,792 ,359 ,216 ZUKUNFT ,765 ,389 ,306 ,250 VORJAHR ,748 ,284 ,281 ,225 ,222 PROZENT ,745 ,199 ,521 ZIEL ,732 ,190 ,616 DANKE ,732 -,131 ,616 ,118 WELTWEIT ,729 ,532 -,174 HERAUSFORDERU NGEN ,728 ,480 ,103 MARKE ,716 -,273 ,506 MODELLOFFENSIV E ,707 ,345 ,322 -,362 -,100 AUDI ,706 -,282 ,500 ,307 -,173 EURO ,654 ,148 -,124 ,458 MODELLE ,645 ,564 ,224 -,142 ,276 ELEKTRIFIZIERTE ,645 ,644 EMOBILITT ,637 ,533 -,157 AUSLIEFERUNGEN ,634 ,441 ,310 ,249 -,152 ENTWICKLUNG ,614 ,412 ,511 ,131 USA ,603 ,475 -,117 -,308 ,119 VORSTAND ,595 ,353 ,334 ,440 ,196
ANGEBOT ,528 ,160 ,123 ,119 ,181 DIESELKRISE ,478 -,230 ,313 ,477 -,242 SEGMENT ,920 ,125 AUTOMOBILE ,225 ,909 -,173 ,121 BMW ,876 -,162 ,250 ABSATZ ,845 -,164 -,141 MIO -,150 ,824 ,128 ,166 -,110 -,106 FAHRZEUGE ,770 -,206 -,148 MOBILITT ,282 ,705 -,190 ,394 UNTERNEHMEN ,264 ,700 ,330 ,429 -,127 -,124 CHINA ,657 ,679 ,184 VORLEISTUNGEN ,378 ,666 ,509 ,165 ,118 PRODUKTION ,422 ,660 ,521 EUROPA ,483 ,648 -,122 DEUTSCHLAND ,351 ,595 -,264 -,328 ,311 FAHRZEUG ,129 ,559 -,125 ,244 ,503 QUARTAL ,544 ,381 -,125 OPERATIVE ,259 ,892 ,146 UMSATZRENDITE ,109 -,110 ,870 ,233 SONDEREINFLSSE N -,116 ,865 -,104 NETTOCASHFLOW ,373 -,181 ,833 ,105 -,116 KONZERN ,246 ,812 -,120 ,183 ERGEBNI ,488 ,783 ,166 -,101 TRANSFORMATION SPLAN ,506 -,216 ,760 -,120 -,102 ZIELKORRIDOR ,159 ,482 ,724 ,285 GESCHFTSJAHR ,190 ,392 ,703 -,106 -,162 MERCEDESBENZ -,109 -,214 -,121 KONSEQUENT ,484 ,762 ,191 NACHALTIGKEIT ,546 ,741 ,121 MANAHMEN ,288 -,132 ,142 ,676 ,135 -,153 ELEKTROMOBILITT ,488 ,182 ,661 -,194 LUFTQUALITT -,178 -,204 ,120 ,740 -,189 VERBESSERUNG -,124 -,115 ,713 -,115 DEUTSCHEN ,139 ,496 ,712 ,264 NEUWAGEN -,100 ,707 BUNDESREGIERUN G -,108 -,131 -,157 ,671 HERSTELLER ,369 ,194 -,211 ,628 UMWELTPRMIE -,119 ,626 VOLKSWAGEN -,222 ,259 ,587 DIESEL ,182 ,364 -,239 ,171 ,408
DIESELFAHRZEUG E ,257 -,120 -,207 ,396 CO2EMISSIONEN -,120 ,896 SERVICE ,447 ,105 ,851 KRAFTSTOFFVERB RAUCH -,127 -,109 -,103 ,846
Appendix F. Factor loadings for factor analysis Phase 3, news media. Only the 75 most used words were used as variables for the analysis.
Component 1 2 3 4 5 6 RUPERT ,908 -,162 AUDICHEF ,872 -,147 STADLER ,858 -,151 MNCHEN ,816 -,114 -,137 -,102 UNTERSUCHUNGS HAFT ,810 -,168 -,129 STAATSANWALTSC HAFT ,764 AUDI ,713 ,121 ,177 ERMITTLUNGEN ,672 IMAGE ,662 ,239 ,149 ,221 BETRUG ,491 ,118 ,348 -,244 -,205 VERDACHT ,434 ,398 SOFTWARE ,392 -,109 ,136 -,223 EU ,866 -,114 ,125 DEUTSCHEN -,101 ,839 -,128 QUELLE ,156 ,779 -,102 -,138 AUTO ,761 AUTOBAUER ,102 ,742 -,125 ,117 HERSTELLER -,122 ,735 -,115 ,195 AUTOINDUSTRIE ,734 -,101 DEUTSCHLAND ,686 ,196 -,159 ,144 ANZEIGE ,570 -,178 ,206 -,207 ELEKTROAUTO -,120 ,477 -,153 -,156 -,129 ,336 EUROPA ,452 ,182 ,269 USA ,423 ,173 ,201 -,125
FAHRZEUGE ,406 ,210 ,283 DIESELAUTO -,168 ,287 ,186 ,132 -,135 DIESEL ,117 ,265 -,107 ,156 BMW ,110 ,217 -,199 KLAGE ,866 VERBRAUCHER ,117 ,861 -,103 MUSTERFESTELLU NGSKLAGE ,858 -,116 URTEIL -,180 ,621 ,164 -,106 GERICHT -,126 ,611 ,221 -,136 -,346 FALL ,552 ,425 -,123 -,250 VOLKSWAGEN ,512 ,302 -,174 ,327 UNTERNEHMEN ,284 ,474 ,303 BGH -,217 ,371 ,237 -,152 -,292 KUNDEN ,133 ,229 ,305 ,107 ,274 ANGABEN ,296 ABGASSKANDAL ,124 ,276 ,281 ,213 -,225 -,226 ZEIT ,167 FRAGE ,782 -,105 -,122 BRAUNSCHWEIG ,250 ,197 ,727 SCHADENERSATZ ,550 ,657 -,113 -,109 OBERLANDESGERI CHT -,160 -,105 ,290 ,652 -,197 -,240 KLGER ,204 ,351 ,619 ENTSCHEIDUNG -,222 ,583 -,143 -,210 STUTTGART -,174 ,574 DIESELSKANDAL ,500 ,321 VERFAHREN ,231 ,487 ,183 PORSCHE ,121 -,206 ,444 WINTERKORN ,306 ,101 ,358 -,242 -,202 ENDE ,270 ,265 ,329 ,239 SCHEUER ,117 -,101 ,126 ,764 RCKRUF ,744 -,120 ZETSCHE ,740 DAIMLER -,116 ,166 -,116 ,648 ,184 KBA -,142 ,633 -,184 BETROFFENEN ,401 -,116 ,613 ABGASREINIGUNG ,539 -,234 KRAFTFAHRTBUND ESAMT -,116 -,269 ,383 -,186 SOFTWAREUPDAT E -,125 ,340 MARTIN ,214 ,127 ,259 -,295 -,173 MITARBEITER ,205 -,153 -,260 ,115
EURO -,142 ,847 MILLIARDEN -,120 -,250 ,773 PROZENT -,175 ,442 -,116 -,139 ,617 JAHR -,161 -,130 -,143 ,602 MILLIONEN ,240 ,524 KOSTEN -,123 ,504 GRUND ,298 ,277 ,485 KONZERN ,403 ,158 ,204 ,413 ZAHLEN ,342 ,112 ,163 ,350 ABSCHALTEINRICH TUNG -,103 ,221 ,285 -,300 STELLEN -,170 -,153 -,201 ,282
