Tracking down the interplay
An investigation of media reputation- and topic characteristics contingencies in first level corporate agenda building
Jara Ruth Strebel 10855351
Master`s Thesis
Graduate School of Communication Master's program Communication Science
Research Master Communication Science
Supervisors: Dr. Jeroen Jonkman, Dr. Anne Kroon
Abstract
The purpose of this study is to investigate the agenda interplay between the media and the organization, and how the relation changes depending on media reputation and topic. For this study, 3,863 press releases of Dutch organizations over three years were collected as well as 73,566 newspaper articles reporting on them in the same time span. Media reputation was measured with the help of different automated sentiment analyses, whereby the present topics were depicted via LDA topic modeling. By investigating agenda building via VAR time series modeling, it was possible to depict causality in the reciprocal relationship between the media- and the organizational agenda, per topic. Researching the interplay separately for each depicted topic greatly contributes to current literature, which has often ignored the possible contingency of topic factors or solely investigated one topic per sample. No interplay was found between the media and the organization, nor when measuring separately for high- and low reputed organizations. However, the results indicate that the agenda-building relationship between the two players are dependent on the two investigated contingencies: media
reputation and topic. This paper contributes to existing literature by serving as a reference for future research to break down agenda-building theory by investigating player- as well as topic characteristics.
Keywords: corporate communication, agenda-building interplay, topic modeling, automated sentiment analysis, time series
Tracking down the interplay
An investigation of media reputation- and topic characteristics contingencies in first level corporate agenda building
Media salience influencing the audience’s awareness about issues has been conceptualized as agenda setting, and is rooted in political communication research. The social scientists McCombs and Shaw (1972) pioneered the idea of agenda setting by demonstrating the correlation between the issue importance of the media agenda and the public agenda during the 1968 U.S. pesidential election. The authors found a high correlation between the
emphasis news media placed on an issue and undecided voters’ belief of that issue’s importance. This finding indicates that by reporting on certain topics more frequently, the media was capable of influencing the salience of topics on the public agenda. Presently, agenda-setting theory is considered one of the most widely employed and investigated frameworks in communication research (Tewksbury & Scheufele, 2007).
Over the last four decades, agenda setting has been researched through a two-step approach. On the first level, the theory describes the transfer of object importance from the media to the public. Within agenda-setting theory, objects are the concepts (e.g. issues, actors, organizations) about which one can form an opinion. The additional transfer of the object attributes (e.g. the accompanied tone related to the object) to the public constitutes the second level of agenda setting (Matsaganis & Payne, 2005).
In the 1970’s, scholars started to shift the focus in agenda-setting research by investigating how the media agenda is determined. This stream of research, labeled agenda building, is described as an extension of agenda-setting theory (Berkowitz, 1992). Initially, the notion of agenda building was used to explain how the media agenda was set; now, agenda building is mostly focused on the effects of different institutional entities, with a strategic interest in influencing the media agenda in order to gain an advantage (Kroon &
Van der Meer, 2018). Agenda setting commonly refers to an information transfer between the media and the public, while agenda-building scholarship focuses primarily on the interplay between multiple agendas– especially focusing on the reciprocal relationship between the media agenda and other institutional agendas (e.g. Aelst, Sehata, & Dalen, 2010; Kiousis, Popescu, & Mitrook, 2007; Vliegenthart, & Walgrave, 2008), such as the corporate agenda (Kroon & Van der Meer, 2018). Although the majority of the agenda-building studies were in a political communication context, a body of corporate agenda-building research has been established in recent years (Kiousis et al., 2007).
Carroll and McCombs (2003) conceptualized agenda building and defined corporate actors as “(an) aggregate of discourse by a firm including attempts through public relation strategies to influence the media with a variety of information subsidies” (p. 42). Although the stated definition, as well as many others, point to the influential effect of the corporate agenda on the media agenda, agenda building, in fact, refers to the reciprocal relationship between institutional agendas and the media agenda (e.g. Kroon & Van der Meer, 2018). The strategy behind a press release is to present an organization as positively as possible to the public, whereby the media aims to influence the corporate agenda to serve the public by demanding answers in situations where misbehavior threatens public norms (Eriksson & Östman, 2013; Maat, 2007).
Thus, agenda-building efforts cannot simply be seen as unidirectional processes of one actor influencing the other. Due to gatekeeping processes on both (e.g. Carroll & Deephouse, 2014) the media- and the organizational sides of the spectrum, the actors are in control of the extent and the manner that they adapt to issues presented by the other
(Strömbäck, 2008; Tushman & Katz, 1980). It is only natural that organizations have to adapt and discuss topics which are salient in the media in order to keep up their legitimacy
media will inform the public on the reaction of the organizations as well as further comment on it. In consequence, discussion arises between the two players, in which both have an interest in building the other’s agenda according to their own. Organizations want to increase their business returns, whereby the media aims to fulfil its watchdog position and report on newsworthy information (e.g. Eilders, 2006).
In sum, agenda influences are often investigated in a static manner, from one player to the other, while the notion of interplay is often neglected (Kroon & Van Der Meer, 2018). Moreover, by only using relatively simple correlation measurements - as done in the lion’s share of agenda-building research - it is not possible to statistically depict causality. This calls for more in-depth research that is indeed able to account for a causal interplay over time between the organizational/corporate agenda on the one hand, and the media agenda on the other.
A notable exception to previous research is a recent project conducted by Kroon and Van der Meer (2018) that investigates the relationship between the corporate- and the media agenda with a time series approach. This study found an interplay-effect between the
organizational- and the media agenda by studying the reciprocal effect between topics in corporate press releases and media coverage in the U.K. More precisely, it was found that the organizational- and the media agenda are intertwined and are influencing each other in a reciprocal manner with regard to topic salience.
In addition, prior agenda-building research from the field of political communication suggests that agenda-building processes can differ regarding the topic at stake. For example, Lee and Riffe’s (2017) research of U.S. press releases and U.S. media content showed that journalists are less likely to use press statements as a source, depending on the sensibility of the topic. Notably, this preference for third party sources in sensible news reporting indicates that the interplay differs depending on topics. Some isolated studies have investigated
corporate agenda building regarding specific situations - such as Ragas, Kim, and Kiousis (2011), who explored the interdependencies regarding financial information in a case study of a U.S. company. Such projects indicate that the interplay between the media and the
organization is worth investigating separately for various kinds of issues.
Hence, the main aim of this paper is to investigate the reciprocal relationship between the media and the organizational agenda across different topics. In comparison to existing literature, this study will investigate not merely one isolated issue but also collate the interdependencies of different issues within one sample. The strength of this research lies in its inductive approach, which will result in making a contribution to current research by serving as a reference for the growing body of literature that investigates contextual factors as contingencies in the agenda interplay. While doing so, the paper will also add to literature explaining the interdependencies of the specific agenda-building interplay between the media and the organization.
Further, it must be noted that agenda-building interdependencies may differ contingent upon the power dynamic of the different actors involved (Kiousis et al., 2007). Therefore, next to investigating the agenda interplay between the media and organizations, this study also aims to shed light on what characteristics of an organization ultimately determine which of the two players takes the lead in the relationship. More precisely, the differences in the interplay of media- and organizational agendas may depend on
organizational media reputation (Deephouse, 2000).
Media reputation represents the opinion that news entities hold over an organization when reflecting on all accessible information over the organization’s previous interactions with its stakeholders (Deephouse, 2000). The media reputation determines the predominant sentiment in the news portrayal of an organization. Arguably, journalists utilize
they hold in their compiler. In this paper, media reputation is used as a proxy for the level of trust the media holds in an organization. Vice versa, organizations may react differently on negative or positive media portrayals, respectively. A low media reputation may indicate negative press representation, whereas high reputation signals positive news coverage. Positively reputed companies may therefore be more successful with their agenda-building efforts than poorly reputed firms.
To investigate the first level agenda-building interplay between organizations and the media for a variety of topics as well as for different media reputations, the following two research questions are formulated:
RQ1: To what extent is the interplay between the media agenda and the organizational agenda different across topics?
RQ2: To what extent is the interplay between the media agenda and the organizational agenda on the level of topics different for (a) high and (b) low media reputation firms?
Theoretical Framework
Organizations have several ways to communicate with their stakeholders. The most
routinized one being via press releases. Managers have realized how crucial the news media can be to their success and therefore, hire public relations (PR) staff to manage the media relations. PR is traditionally, and often still, referred to as a strategic organizational function, used to generate complimentary publicity through news coverage (Sleurs, Jacobs, & Van Waes, 2003). The PR employees’ ultimate goal is for organizational press releases to be taken over by journalists, considering the topic- as well as the sentiment salience (Smith, 2013). In a second step, the organization then hopes that the successful agenda building towards the media can further influence the public agenda in their favor (Curtin, 1999). Organizations are aware of the gatekeeping processes within a news agency and therefore
prepare their PR statements in a way that fits the so-called media logic (e.g. Schafraad, Van Zoonen, & Verhoeven, 2016). Exemplarily, Schafraad et al. (2016) found that press releases are most likely to be picked up by the media if they include news factors or news values (e.g. Eilders, 2006) such as “controversy”, “negative consequences”, “surprise”, or “elite
organizations.” The study also showed that journalists are most likely to adapt topics
surrounding the financial performance of an organization, or topics regarding employees and the management. By constructing press releases according to such findings, PR professionals try to strategically build the media agenda (Schafraad et al., 2016). However, not only the content of a PR statement is strategic; the formal characteristics, such as writing from a third person perspective, are also a part of it. Formal news requirements are incorporated into R statements with the intention that the journalist will not have to adapt the text, and can simply publish it as it is (Maat, 2007).
For journalists, press releases are an important source of information, which otherwise would be difficult and possibly costly to obtain (Sleurs et al., 2003). Media professionals are often aware of the intentions of PR employees – they generally aim at creating a positive image of the organization (Curtin, 1999). In the end, however, it is in the hands of the journalists to decide what organizational information they want to present to the public, and to what degree (Strömbäck & Nord, 2006). First level agenda-building effects of a PR release occur when the journalists accept the topic proposed by the organization – for example, in a press release. It has been found that a journalist’s willingness to take over a topic depends on their judgement of relevance to the public. However, journalists do not only base their news reporting about organizations on information subsidies, but also raise topics themselves if they perceive it as relevant to the public (Haselmayer, Wagner, & Meyer, 2017).
Organizational news coverage may spark reactions of the affected organizations for different reasons. Livelihood and provability depend on the way the stakeholders perceive an
organization. Since most stakeholders get their information through the media, organizations promote positive news coverage as well as attempting to dissolve negative articles about themselves (Cornelissen, 2010).
Therefore, agenda-building efforts in the corporate communication context should not be investigated as a one-way action from an agenda to its opponent (Kroon & Van Der Meer, 2018). The interdependencies between the media- and the organizational agenda can be characterized as an ongoing discursive interplay between two partners, who each have their own interest in the partnership (Cornelissen, 2010). Both the organizations as well as the journalists control key resources upon which the other is dependent. The media control the image the public gets to see of the organization, while the organization possesses information that the news outlets needs in order to produce their content (Strömbäck & Nord, 2006).
However, most previous efforts to investigate the agenda-building process regarding the relationship between the organizational- and the media agenda fail to take the interplay into account and investigate agenda building as a unidirectional process (Kroon & Van der Meer, 2018). In the context of time series analyses in the agenda-building field, researchers speak of an interplay of a topic if both agendas influence the other significantly in the
interrogations regarding that topic (e.g. Meraz, 2011). Hence – taking this and the above into account the first hypothesis reads as follows:
H1: The extent to which the media agenda is influenced by the corporate agenda is equal to the extent to which the corporate agenda is influenced by the media agenda.
Van Zoonen et al. (2015) found that not only do the message factors have an influence on the agenda-building interdependencies, but also the agenda-building interactions differing between topics. Journalists generally report on topics that are relevant to the public. As such, the journalistic choice of source for information subsidies has found to differ regarding the sensibility of the topic (Lee & Riffe, 2017). Sensibility of a topic can be described as the
topic's potential to generate controversy in the public sphere as well as negative
consequences for the actors involved in it (Imhof, 1999). The more sensible a topic, the more often news outlets are likely to consult third party sources for their content creation instead of PR material. Consequently, organizations have less stirring power in topics that journalists perceive as too sensitive to consult the self-promoting material of the organization (Lee & Riffe, 2017. The sensitivity of a topic is heightened with the amount of people to which the topic appeals, which is often the case for topics about morality or social values (Denham, 2010). Furthermore, when the media covers topics from third party sources, this leads organizations to make statements about the same topics to try and also bring their point of view into the public sphere (Lee & Riffe, 2017). Such findings indicate that the agenda interaction between two players might differ among various topics. Seeing as the theoretical background on different topics in agenda-building research is yet to be established, it is difficult to make predictions. Therefore, sub-RQ1 was formulated:
RQ1: How does the interplay between the media- and the organizational agenda differ across topics?
Every organization has a different set of characteristics (Ledingham, Bruning,
Tomlinson, & Lesko, 1997). Stoker and Tusinski (2006) named equality as a requirement for an equal interaction between different parties. The authors elaborated that for organizations, it is especially difficult to create equality in their interaction with journalists. Many factors can influence such interactions, one of them being the trust conditions (Pettigrew & Reber, 2010). Therefore, organizations with different characteristics must have different kinds of interplays with the media. In some cases, the organization might possess a greater agenda building power than the media, which would result in the organization building its
Even though agenda building in the corporate context has been the center of a number of research projects, current literature has yet to establish a theory which characterizes the kind of interplay between the two parties (e.g. Kim, Kiousis, & Xiang, 2015; Lee & Riffe, 2017; Ragas et al., 2011; Schafraad et al., 2016).
Media Reputation
Media reputation is the overall representation of how an organization is portrayed in the media; this is commonly measured as the average sentiment of the media discourse on the organization (Deephouse, 2000). In general, having a high reputation suggests that the
stakeholder (e.g. a member of the general public or a journalist) trusts an organization (Xiong & Liu, 2005). According to management scholars, media reputation can be viewed as an asset, which helps an organization to a competitive advantage, because it creates a positive representation towards the public (e.g. Deephouse, 2000). Along those lines, a high media reputation would indicate that journalists trust an organization and, therefore, perceive their statements as more believable and accurate (Deephouse, 2000; Rindova, Williamson, & Petkova, 2010).
Media reputation commonly refers to the sentiment with which the organization is portrayed in the news. Notably, that sentiment reflects the opinion of the compilers
(Deephouse, 2000). Therefore, a high media reputation of the organization may well indicate that journalists have a more positive opinion about- as well as trust in that unity, more so than they would for an organization with a lower media reputation. Journalists are generally wary of the intentions of PR professionals and, as such, approach the use of press releases
differently depending on the organizational context (Pettigrew & Reber, 2010). For example, by interviewing U.K. journalists, Lewis, Williams, and Franklin (2008) depicted that
barely fact check information coming from trusted sources and often simply adapted the content of their press releases.
Organizations, enjoying a high media reputation are portrayed with a positive sentiment by the media (Deephouse, 2000). Being confronted with less negativity
consequently means that the organization has few reasons to engage in topics surrounding them in the media landscape, since they have little need to defend themselves or to uphold their legitimacy (Patriotta, Gond, & Schlutz, 2011). A low media reputation, on the other hand, indicated by a rather negative sentiment in organizational news reporting, suggests distrust from the side of the journalists (Tucker & Melewar, 2005). A low level of trust leads journalists to be more conscious when consulting the organization itself for information (Lewis et al., 2008). Therefore, topics from the organizational agenda are less likely to be accepted if they originate from the organization holding a low media reputation.
Furthermore, journalists assert the role of a watchdog: signaling misconduct in society and critically assessing the (mis)conduct of powerful institutional actors, such as corporations (Eriksson & Östman, 2013). To counter bad press, most organizations are likely to take communicative action in some way. In doing so, an organization would generally aim to make sure their side of the story is heard and will therefore release communicative statements addressing the issue visible in the media (Tucker & Melewar, 2005). Contrarily,
organizations with a media reputational advantage (i.e. high media reputation) may encounter less critique from the media and are, as such, less likely to be forced to react to media
content. Subsequently, the agenda building interplay dynamics are arguably dependent on both: (1) the topic of discussion as well as (2) the media reputation of the organization. Hence, the following hypothesis and sub-RQ are formulated:
H2a-b: The extent to which the media agenda is influenced by the corporate agenda is contingent upon media reputation; in such a way that (a) low reputed organizations
have weaker agenda building effects towards the media than vice versa and (b) high reputed organizations have stronger agenda building effects towards the media than vice versa.
RQ2: What are the differences between high- and low media reputation regarding the interplay between the media- and the organizational agenda across different topics?
Method Data Sample
For this paper, communication data from 25 Dutch firms, in the form of press releases and articles originating from the Dutch media landscape, were gathered. The organizations included in the sample were chosen by the following selection procedure; only companies large and visible enough to have an influence on their national news media were considered as meaningful units for this research. Therefore, a convenience sample of 25 firms from the Elsevier 500, which ranks Dutch companies according to their revenue, was constructed (e.g. Schafraad et al., 2016). Starting at the top of the list, the 25 firms were selected based on who fulfilled all preliminary conditions. The conditions for a company to be included were data availability, the language in which the press releases were written, as well as the origin of the company. Companies that did not provide press releases on their web page from the
beginning of 2015 to the end of 2016, as well as firms which did not receive any media coverage in that time span, were excluded. Furthermore, companies that did not provide a Dutch version of their press releases were also removed from the sample, since translating text objects could lead to complications. Further, companies that did not originate in the Netherlands were also excluded. Dutch branches of companies with non-Dutch origins, such as Ingka Holding (IKEA), were not included in the sample. Therefore, the sample does not reflect the 25 highest ranked companies from the Elsevier 500 list, but the first 25 to fulfill the aforementioned criteria.
Press releases. The communication materials of the firms were collected in the form of press releases. All published press releases from the 1st of January 2015 to the 31st of December 2016 were collected from the 25 companies chosen for the sample. The material was manually retrieved directly from the company websites. Press releases which did not contain any text elements, consisting only of pictures or videos, were not retrieved. The collected sample consists out of 3,868 press releases. An overview of all included companies can be found in Appendix A.
News articles. For the same time span, newspaper articles about the companies included in the sample were derived from the International News Collection Amsterdam (INCA) Project. The project provides different data, collected via web scraping, such as content of news websites, to social scientists (Trilling, 2016). Only articles mentioning one of the 25 companies in the title or the lead paragraph were included in the sample. The content that is made available through the INCA project, and that is analyzed in this study, is spread over eleven different national Dutch newspapers, five of which incorporate online as well as offline content. Articles that were published both online and offline were analyzed merely once. An overview of the news outlets, from which the newspaper article sample originated, can be found in Appendix B. From the collected material (N = 14,373), all items that did not include any text were removed, which resulted in a newspaper article sample of 13,566 items. The distribution of the news articles over the 25 different firms can be found in Appendix A. Constructing the dataset
Considering that the initial dataset of press releases and news articles was comprised only of text and a few additional descriptive variables, such as date and source, useful variables had to be created with the help of sentiment analyses and Latent Dirichlet Allocation (LDA) topic modeling (e.g. Boumans & Trilling, 2006). Media reputation was
measured in order to divide the sample of different organizations into two groups – higher and lower reputation. Next, an investigation of the issues present in the sample was initiated in order to divide the press releases and newspaper articles into topic categories. Depicting the topics present in each item (N = 17,429) via LDA topic modeling, an unsupervised machine learning approach, is an appropriate instrument for content analyses, which automatically organizes large amounts of texts such as the one gathered for this research (Strycharz, Strauss, & Trilling, 2018).
Sentiment analyses. Media reputation refers to the attributes given to an issue by journalists (Deephouse, 2000). Therefore, three different analyses were performed in order to depict the overall average sentiment with which an organization and its attributes are described by the media. Using several approaches has the advantage of being able to compare the different results to reach a higher validity.
SentiStrength is a sentiment analysis opinion mining algorithm, which can extract the sentiment of texts in several languages by applying dictionaries of sentiment words
(Thelwall, Buckley, Patogolou, & Kappas, 2010). The output consists of two values, a positivity- as well as negativity measure, each indicated on a continuous scale from 1 (not at all) to 5 (extremely). The negative value was subtracted from the positive to get one number as an indication of the sentiment. The next method used was pattern.nl, which is a package for Python with different functionalities. Its natural text processing tools, including a sentiment dictionary, were used to determine the sentiment of all newspaper articles as an additional measure (Smedt & Daelemans, 2012). Lastly, a list matching approach was used. The lists of words were retrieved from Kaggle. The amount of negative words in every article was subtracted from the number of positive ones to achieve a sentiment score (Tateman, n.d.; Trilling, 2016).
Every approach calculated the sentiment for each news item separately. In a second step, a mean for each company was depicted per approach. The three lists of scores were then tested for their correlation with each other. High correlations would indicate that the different approaches evaluated the sentiment of the media reporting per organization equally.
However, since the correlations were very low - SentiStrength and list matching correlated the highest with a rather low coefficient of 0.267 - another approach to sort the organizations was performed.
To be able to divide all companies by low and high reputation, the results of every method was divided into the two groups (high and low) by a mean split. This resulted in three categorizations per company. For 11 companies the categorization was clear, with all three methods classifying them into the same group. For the rest, the results were mixed. Having three binary results meant that each of those companies were rated the same two times. Therefore, these companies were classified by whichever group they were categorized in for two out of the three sentiment analyses. An overview of the sentiment scores are summarized in Appendix C and the correlation coefficients of the sentiment analyses can be found in Appendix D.
LDA Topic Modeling. The goal of this research was to identify the distinct interactions between organizations and the media on the first level. The texts of the whole sample, press releases as well as newspaper articles, were investigated with topic modeling in order to determine the different topics present in the sample. To do so, LDA was used; this describes an unsupervised machine learning algorithm which assumes that texts consist of different topics, whereby each topic is represented by a distribution of words (Strycharz et al., 2018). A word's importance is determined by both its co-occurrences with other topic related terms and the level of relevance of the term for the topic (Blei, Ng, & Jordan, 2003). Before training the machine, it was crucial to clean the text data adequately. In the first step, all
company names and other organization specific information, such as the names of CEOs, were removed from all text items. After all the stop words were excluded, all texts were lowered and stemmed. Pre-processing is a very important step for LDA topic modeling for two reasons. First of all, the organizational indications had to be removed to prevent the machine to sort all texts according to company indications. Secondly, since LDA topic modeling works by assigning words to topics, it is important to have texts consisting only of relevant words, such as nouns, verbs and adjectives. Pronouns and conjunctions occur frequently in texts, but do not add any value to a topic and were therefore removed entirely (e.g. Boumans & Trilling, 2016). Furthermore, words that appear too rarely or too frequently add little to distinguish topics. Therefore, extreme cases were also filtered out. Every word that only appeared once in the whole sample was removed, as well as all words that appeared in more than 50% of all sample items. (e.g. Strycharz et al., 2018).
As a result, twenty topics were created. The output of the LDA topic modeling can be found in Appendix E as well as further insight into how the topics were named and merged. The output of the LDA topic model shows the five most important words for each of the twenty topics. Based on this output, a title for all of the topics was created. Since some topics seemed to greatly overlap regarding the importance of specific words, a few topics were merged resulting in a total of 12 separate topics (corporate social responsibility (CSR), investments, events, online services, revenue, international business, competitors, customers, management, product information, future business perspectives, other). The values for merged topics were multiplied with each other. For every single text item, 12 additional variables were created reflecting the created topics. The continuous value represented the intensity in which the topic is represented in one text item. Therefore, for every text item, 12 continuous topic scores were depicted, where zero means a topic is not at all present.
Analysis
Three datasets with the exact same structure were created for the analyses. One dataset includes all retrieved newspaper articles and press releases. For dataset two and three, the first dataset was split into two, whereby one consists of all high reputational- (N = 14) and the other, of all low reputational organizations (N = 11).
All datasets have 1,096 rows and 27 columns. Every row represents one day in the timespan between the 1st of January, 2014 and the 31st of December, 2016. The row further entails summarized information about all press releases and newspaper articles released on one day. Twelve columns represent the topics present in the press releases and another twelve show the same information about the newspaper articles. The topics were depicted in the LDA topic modeling, which estimated how present each of the topics is in every item. For all press releases published on the same day, intensity scores were summarized per topic. The same was done for the newspaper articles per dataset. Therefore, each topic consisted of two columns, one representing the press statements and the other representing newspaper articles.
To test the hypotheses, a bi-directional time series approach, namely Vector Autoregression (VAR) analysis, was used. This method is adequate to research interplays since it takes interdependencies between variables into account (Strycharz et al, 2018). Therefore, neither of the agendas were considered exogenous in this investigation. A VAR analysis was performed to measure the influence over time (Brandt & Williams, 2007; Vliegenthart, 2014). With the help of two equations, the agendas of interest were looked at as both independent and dependent variables. Two equations were calculated for each of the topic items overall, as well as per organization group (high reputation vs low reputation). For reasons of transparency it is important to mention that a VAR analysis assumes no missing values in the dataset. Therefore, every day on which no item was released for a certain agenda, all corresponding topic variables were filled in with zero values.
Time series analyses compare adjoining time intervals with each other; the length of the time intervals needs to be decided. It is a general assumption that journalists try to report news as fast as possible. The swiftness of an organization in publicizing statements, however, is less researched. For this paper, it is assumed that four weeks is a realistic response time. This rather long interval minimizes the chance to possibly oversee a causality due to a late response. Further, it also prevents categorizing several successive reactions as belonging to different conversations. Within one time-interval, all topic intensity variables were
summarized for each agenda separately. This resulted in one value per topic for every agenda in each of the three separate investigations. The last step was to carry out the Granger
causality test for each topic relating both to the organization as a whole, as well as for the two groupings based on media reputation. This causality test is performed between two time points and evaluates whether or not one predicts the other (Strycharz et al., 2018). This was carried out twice for all VAR models since both directions of causality between two series were of interest for this paper.
Results
To answer the two hypotheses as well as the two research questions, 36 VAR time series models were estimated and two Granger causality tests were performed for each of them. VAR analyses require all variable series to be stationarity, which can be tested with the Augmented Dickey-Fuller test. The test showed proof of stationarity for all variables except for the topic variable future business in each of the three samples, which had to be
differenced in order to achieve stationarity.
Preliminary to the analyses regarding topic salience, the interplay of the visibility of released items was also calculated. The count of released articles about an organization represents the degree of media visibility, whereby, the amount of press statements an organization releases reflects its self-initiated visibility. Interestingly, in all three estimated VAR time series
models regarding visibility, some significant and marginally significant agenda building effects were found. In the overall sample, the organization’s visibility in the news did significantly Granger-cause the amount of press releases. The vice versa effect is marginally significant. Regarding low reputed organizations, a significant effect of the media visibility Granger causing the self-initiated organizational visibility. Marginal effects in both directions were found for high reputed organizations.
Table 1
Results of Granger analysis of the entire sample
PR à News News à PR
Issue Vector F-Value p-Value F-Value p-Value
Visibility 1.123 0.088* 1.365 0.042** Topics CSR 1.592 0.015** 0.998 0.473 Investments 1.598 0.014** 0.695 0.914 Events 1.291 0.117 0.833 0.747 Online Service 0.637 0.954 0.664 0.937 Revenue 1.1 0.317 1.221 0.174 International Business 0.984 0.497 0.7 0.908 Competitors 1.193 0.201 1.004 0.398 Customers 1.226 0.169 0.905 0.632 Management 2.005 < 0.001*** 0.617 0.964 Product Information 1.056 0.380 0.360 0.958
Other 0.846 0.727 1.006 0.459 Note. df1 = 36, df2 = 1417, *p < .1, **p < .05, ***p < .01, Newspaper articles N = 13,566, PR statements N = 3,863.
No support was found for hypothesis 1, which proposed that the media agenda would be influenced by the corporate agenda as much as the corporate agenda would be influenced by the media agenda. Instead, agenda building effects of the organization were found
regarding three topics (CSR, investments, management). The media agenda did not Granger-cause any topics for the organizational agenda.
Regarding RQ 1, it appears that the agenda building interplay indeed differs between topics. For three topics, an agenda building effect on the first level from the organizational- to the media agenda was found. For the other nine topics, no Granger-causality was depicted in any direction. An overview of all the Granger causality tests for all VAR models relevant for hypothesis 1 and RQ 1 can be found in Table 1.
Hypothesis 2a predicted that low reputed organizations have weaker agenda building effects towards the media, than the media has on them. An overview of all relevant results concerning organizations with a low media reputation can be found in Table 2. The topic online services showed an indication of an interaction between organizations with a low media reputation and the media. In four cases, there was a significant indication that the organizational agenda built the media agenda (online services, international business, customers, competitors), with one other topic approaching significance (revenue). The reverse effect was observed only once (product information). For five other topics, no interaction or one-way agenda building processes were found. Therefore, part A of the second hypothesis is rejected, since the opposite of what was predicted was found.
Organizations with a low media reputation have a stronger agenda-building effect towards the media than the other way around.
Table 2
Results of Granger analysis of organizations with low media reputation
PR à News News à PR
Issue Vector F-Value p-Value F-Value p-Value
Visibility 1.201 0.105 1.582 0.013** Topics CSR 0.991 0.486 1.099 0.318 Investments 0.738 0.871 0.913 0.617 Events 1.172 0.226 1.02 0.438 Online Service 1.627 0.012** 1.508 0.028** Revenue 1.374 0.071* 1.22 0.175 International Business 1.992 < 0.001*** 1.071 0.358 Competitors 1.503 0.029** 1.002 0.466 Customers 1.477 0.035** 0.643 0.951 Management 0.952 0.551 0.888 0.660 Product Information 0.993 0.482 1.641 0.010**
Future Business Perspectives 1.035 0.414 1.238 0.160
Other 0.961 0.536 0.932 0.585
Note. df1 = 36, df2 = 1180, *p < .1, **p < .05, ***p < .01, Newspaper articles N = 7,570, PR statements N = 1,718.
Lastly, the interactions between the media and organizations with a high media reputation were analyzed. Hypothesis 2b predicted high reputed organizations would have stronger agenda building effects towards the media than vice versa. The interactions
topic (management) on the media agenda and additionally two topics (investments, events) approaching a significant effect. Conversely, no agenda building power of the media was found. Therefore, the organization with high media reputation has more agenda building power towards the media than vice versa, which indicates support for hypothesis 2b. Table 3
Results of Granger analysis of organizations with high media reputation
PR à News News à PR
Issue Vector F-Value p-Value F-Value p-Value
Visibility 1.401 0.078* 1.342 0.098* Topics CSR 0.706 0.904 1.247 0.151 Investments 1.36 0.077* 1.044 0.399 Events 1.331 0.093* 0.806 0.787 Online Service 1.052 0.386 0.915 0.614 Revenue 0.651 0.945 0.925 0.598 International Business 1.018 0.440 1.156 0.243 Competitors 0.712 0.898 0.887 0.662 Customers 0.633 0.956 0.643 0.951 Management 1.569 0.018** 0.749 0.860 Product Information 1.056 0.380 0.975 0.512
Future Business Perspectives 0.867 0.695 0.739 0.870
Other 0.787 0.813 1.144 0.258
PR statements N = 2,145.
To answer RQ 2, the results presented in Tables 2 and 3 need to be considered. Comparatively, the sample of organizations representing a low media reputation showed significant agenda building power not only in more topics (N = 3) than high reputed
organizations (N = 1) but also in different ones (low reputed organizations: online services, international business, customers, competitors; high reputed organizations: management).
Further, an influence of the media agenda was only found on low reputed
organizations (N = 1). Therefore, the agenda building interdependencies differ according to topics and depending on the level of media reputation.
Discussion
The aim of this study was to investigate the interplay between the media- and the organizational agenda regarding different topics and levels of media reputation. This paper assumed that the media and organizations exist in a reciprocal, interdependent interplay with one another, as found by prior research (e.g. Kroon & van der Meer, 2018). In the present study, three major findings emerged regarding the agenda building interplay, as well as the role of topics and media reputation as contingencies upon the relationship.
Interplay
No support was found for an interplay, the notion that the media agenda influences the corporate agenda to an equal extent as vice versa. Overall, it was found that the organization has more first level agenda building effects on the media agenda than vice versa. This finding offers support to the scholarly critique on modern journalism, which often claims that the media are increasingly under control of PR agenda building (Kroon & Van Der Meer, 2018). Journalists’ resources are increasingly limited due to restricted time as well as high pressure to generate business returns in this age of the Internet. According to Strömbäck and Nord, (2006) this development leads to a decreasing focus on investigative journalism, and
therefore a shift to a more frequent adoption of press releases. Further supporting their proposition, the authors also found that the agenda-building interplay is merely led by the source and that the media had less power in news reporting in the case of the 2002 Swedish national elections.
These findings might have been somewhat unexpected, given the evidence speaking for an agenda interplay between the organization and the media (e.g. Kroon & Van der Meer, 2018; Verhoeven, 2009). However, this study investigated the interplay on the topic level. Kim et al. (2015) found that the lack of effects on the first level of agenda research does not exclude effects on the attribute level. In their study of U.S. companies, the authors failed to identify agenda building on the first level from the organizational- to the media agenda but did, however, find effects on the second level. Therefore, it is still plausible that the
investigated organization exists in an interplay with the media, although beyond the first level.
Contingency of topics
Agenda-building dynamics differ between topics. In this study, it was found that the organization has agenda-building power in three topics (CSR, investments, management), which all reflect organizational values. The social values associated with an organization through their activities and their fit with the prevailing norms of acceptable behavior forms the organizational legitimacy (Dowling & Pfeffer, 1975). Legitimacy to exist is crucial for an organization, since it reflects the societal acceptance of the organization’s operations (Patriotta et al., 2011). Managerial decisions, such as CSR activities, are the organizational behavior that the public reflects upon (Crane & Glotzer, 2016). Disruption of legitimacy may have a serious impact on business returns, therefore topics reflecting organizational behavior are highly sensitive (Jonsson, Greve, & Fujiwara-Greve, 2009).
In the theoretical framework of this paper, it was predicted that sensitive topics lead journalists to consult third party sources rather than press releases (Lee & Riffe, 2017). Therefore, organizations should have lower agenda-building effects on the media regarding sensitive topics compared to non-sensitive topics. However, the exact opposite was found in the present study. The organization’s agenda-building power was found to be strongest regarding sensitive topics.
Journalists are aware of the importance of those topics for a harmonious society (e.g. Eilders, 2006). Arguably, the high relevance of those topics leads journalists to consult many possible sources thoroughly. This includes organizational ones as well as third parties in order to guarantee full disclosure to the public. This notion is supported by prior research by Lee and Riffe (2017) who found that journalists are more likely to consult additional sources when reporting on sensitive topics. This may be of less importance for non-sensible topics, since their news reporting are not likely to result in the same possible negative consequences (Schafraad et al., 2016). Since this study does not research the difference of agenda-building effects regarding different journalistic sources, the effects of various sources cannot be compared.
Further, this study did not measure the attributes attached to text items. It is possible that the organization truly was able to transfer sensitive topics to the media agenda, but that the media reported on them critically.
Contingency of media reputation
When comparing relationships between the media and the organization depending on high- and low media reputation, it became apparent that the media reputation has an influence on the agenda-building interaction. The lead, the organization takes in the relationship, was already evident before differentiating between high- and low reputed organizations. This trend was even more prominent when investigating organizations with a low media
reputation. The results showed that a low media reputation is more frequently connected to agenda-building effects towards the media agenda than a high media reputation. The identification of this organization group via the prevailing negative sentiment in the news coverage about them implies that they are merely negatively portrayed by the media. It is rather unlikely that an organization issues press releases with a negative sentiment towards themselves. Subsequently, the low reputational organizations were able to build the media on the first level, however, failed to do so on the attribute level. Firstly, this finding refers back to the idea that PR practitioners take a lead in the relationship, due to the increasing pressure journalists are faced with (Kroon & Van der Meer, 2008). Additionally, it can be interpreted, that while low reputed organization were able to build the media agenda on the first level, the journalists added negative sentiments to them. This supports critical voices, which currently claim that news reporting is tinted with a negativity bias. Negative stories get more attention from the audience than positive stories (Niven, 2001). Due to the competitiveness of the European media market, journalists try to maximize the size of their audience by focusing increasingly on conflict (Vliegenthart, Boomgarden, & Boumans, 2011). Subsequently, it is suggested that journalists more often consult press releases from low reputed- than from high reputed organizations because journalists are looking for negatively tinted stories to write.
Nonetheless, the organizational agenda-building effects are not the only difference found when examining low- and high reputed organizations separately. It also became apparent that the media solely had agenda-building effects on the low reputed organizational agenda. This is in line with the theoretical claims proposing that a high media reputation leads to a decreased need to pick up topics regarding their own organization from the media agenda. The news reporting is positive and therefore, commenting on the news articles is not necessary, since they present fewer consequences (Deephouse, 2000).
Next to comparing the interplay regarding media reputation, another aim of this study was to compare the differences between high- and low media reputation interdependencies across different topics. Two of the three significance approaching/ significant agenda-building effects from high reputed organizations on the media agenda were found in topics earlier identified as sensible. Low reputed organizations, on the other hand, did not show any agenda-building effects on the identified sensitive topics. These findings span back to the concept of media reputation, reflecting the amount of trust journalists have in the
organization (Deephouse, 2000). The organization is more likely to have agenda-building effects on the media regarding sensitive topics if the journalists trust the organization. Additional findings regarding visibility
Interestingly, evidence was found that the organization and the media both Granger-cause visibility of the other in the investigated relationships. These results indicate an interplay regarding the amount of released text material (e.g. Stryker, 2002). The aforementioned development in the news sector of journalists having fewer resources available (Strömbäck & Nord, 2006) offers a possible explanation as to how the amount of press releases Granger-causes the amount of news coverage. Journalists may not have enough time to research which companies are newsworthy and therefore, report on those that present themselves to the media more frequent than others. This is in line with the support found for the organizational agenda having agenda-building power towards the media. The vice versa effect of the media visibility Granger-causing the amount of published press releases could possibly be explained by the negativity bias, which is supported by the findings of this research. The Granger-cause of media visibility on PR visibility for low reputational organizations is the strongest. This could indicate that companies’ self-initiated visibility often is a reaction to counter negative news coverage with the intention of averting damage for financial returns. Nevertheless, since only little support for agenda-building power of the
media agenda was found, the organization seems to counter possible negative visibility in the news with press releases about other topics. Effectively, as a reaction, organizations do not counter argue the bad press, but simply introduce new issues to try and shift the salient topics.
Limitations
In the context of interpreting the results of this research, a few limitations should be considered. To answer hypothesis 2 and research question 2, the sample of organizations was divided into two groups. Rather than reflecting distinctly high- and low reputational
organizations, the reputational indexes of the two groups were very close. Therefore, the groupings represented one group with a higher or lower mean reputational index than the other. For further investigations in the agenda-building interplay and the media reputation as a predictor, extreme cases could be researched, which would possibly lead to more
differentiated results. For the creation of these two organizational groupings, three sentiment analyses were performed on each newspaper article collected. The correlation coefficients of the sentiment measures of the three methods were very low, which further demonstrates the ambiguity of the concept of media reputation.
Regarding the analyses, the VAR models have been conducted with 36 lags over three years. Therefore, every lag was four weeks long, which is a rather lengthy time interval. Interplays could also happen in shorter time spans. To represent the agenda of the organization, only official press releases from the organization’s website were collected. However, organizations utilize more than one channel through which they strategically communicate, such as e-mails and blogs (e.g. Miller, 2010). Additionally, the investigation only included profit organizations, and the most profitable companies of the Netherlands. Furthermore, no difference was made between different kinds of media channels. Therefore, generalizations from this study should be drawn with caution.
Future Research
This study found support for media reputation as well as topic characteristics to be contingencies in agenda-building interdependencies. For future research, scholars are therefore advised to further investigate the agenda-building interplay separately for specific types of media channels, organizations, media systems, and also message characteristics. Further, the results showed that it would be beneficial to research the first and second level of agenda building within the same study, since the results of this study had to be interpreted with assumptions about the attribute level.
Contribution
Considering the findings of this paper, it can be postulated that the agenda-building interplay differs depending on contingencies of media reputation and topic characteristics. This study contributes to existing agenda-building research by offering a starting foundation for future investigations of the broad agenda-building theory. This can eventually be
dissected further, regarding characteristics of the involved players as well as message characteristics. Despite only the first level of agenda building was evaluated, the results of the study indicated that such contingencies should also be considered when further
References
Aelst, P. V., Sehata, A., & Dalen, A. V. (2010). Members of parliament: Equal competitors for media attention? An analysis of personal contacts between MPs and political journalists in five European countries. Political Communication, 27(3), 310-325. Berkowitz, D. (1992). Who sets the media agenda? The ability of policymakers to determine
news decisions. Public Opinion, The Press, and Public Policy, 2, 81-102. Blei, D. M., Ng, A. Y., & Jordan, M. I. (2003). Latent dirichlet allocation. Journal of
Machine Learning Research, 3(Jan), 993-1022.
Boumans, J. W., & Trilling, D. (2016). Taking stock of the toolkit: An overview of relevant automated content analysis approaches and techniques for digital journalism scholars. Digital Journalism, 4(1), 8-23.
Carroll, C. E., & Deephouse, D. L. (2014). 6 The foundations of a theory explaining organizational news. Organizations and the Media: Organizing in a Mediatized World, 30, 81.
Carroll, C. E., & McCombs, M. (2003). Agenda-setting effects of business news on the public’s images and opinions about major corporations. Corporate Reputation Review, 6(1), 36–46. http://doi.org/10.1057/palgrave.crr.1540188
Cornelissen, J. P. (2010). Making sense of a crucial interface: Corporate communication and the news media. In Media, Organizations and Identity (pp. 129-145). Palgrave
Macmillan, London.
Curtin, P. A. (1999). Reevaluating public relations information subsidies: Market driven journalism and agenda-building theory and practice. Journal of Public Relations Research, 11(1), 53-90.
Deephouse, D. L. (2000). Media reputation as a strategic resource: An integration of mass communication and resource-based theories. Journal of Management, 26(6), 1091 1112.
Denham, B. E. (2010). Toward conceptual consistency in studies of agenda-building processes: A scholarly review. The Review of Communication, 10(4), 306-323. Dowling, J., & Pfeffer, J. (1975). Organizational legitimacy: Social values and organizational
behavior. Pacific sociological review, 18(1), 122-136.
Eilders, C. (2006). News factors and news decisions. Theoretical and methodological advances in Germany. Communications, 31(1), 5-24.
Eriksson, G., & Östman, J. (2013). Cooperative or adversarial? Journalists’ enactment of the watchdog function in political news production. The International Journal of
Press/Politics, 18(3), 304-324.
Haselmayer, M., Wagner, M., & Meyer, T. M. (2017). Partisan bias in message selection: Media gatekeeping of party press releases. Political communication, 34(3), 367-384. Imhof, K. (1999). Die Privatisierung des Öffentlichen: Zum Siegeszug der
Primärgruppenkommunikation in den Medien [The privatization of the public: The triumphal march of the primary group communication]. In Grenzenlose Gesellschaft? (pp. 717-732). VS Verlag für Sozialwissenschaften, Wiesbaden.
Jonsson, S., Greve, H. R., & Fujiwara-Greve, T. (2009). Undeserved loss: The spread of legitimacy loss to innocent organizations in response to reported corporate deviance. Administrative Science Quarterly, 54(2), 195-228.
Kim, J. Y., Kiousis, S., & Xiang, Z. (2015). Agenda building and agenda setting in business: Corporate reputation attributes. Corporate Reputation Review, 18(1), 25-36.
Kiousis, S., Popescu, C., & Mitrook, M. (2007). Understanding influence on corporate eputation: An examination of public relations efforts, media coverage, public opinion, and financial performance from an agenda-building and agenda-setting perspective. Journal of Public Relations Research, 19(2), 147-165.
Ledingham, J. A., Bruning, S., Thomlison, T. D., & Lesko, C. (1997). The applicability of interpersonal relationship dimensions to an organizational context: Toward a theory of relational loyalty a qualitative approach. Journal of Organizational Culture,
Communications and Conflict, 1(1), 23.
Lee, S. Y., & Riffe, D. (2017). Who sets the corporate social responsibility agenda in the news media? Unveiling the agenda-building process of corporations and a monitoring group. Public Relations Review, 43(2), 293-305.
Lewis, J., Williams, A., & Franklin, B. (2008). Four rumours and an explanation: A political economic account of journalists’ changing newsgathering and reporting
practices. Journalism Practice, 2(1), 27-45.
Maat, H. P. (2007). How promotional language in press releases is dealt with by journalists: genre mixing or genre conflict?. The Journal of Business Communication
(1973), 44(1), 59-95.
Matsaganis, M. D., & Payne, J. G. (2005). Agenda setting in a culture of fear: The lasting effects of September 11 on American politics and journalism. American Behavioral Scientist, 49(3), 379-392.
McCombs, M. E., & Shaw, D. L. (1972). The agenda-setting function of mass media. Public Opinion Quarterly, 36(2), 176-187.
Meraz, S. (2011). Using time series analysis to measure intermedia agenda-setting influence in traditional media and political blog networks. Journalism & Mass Communication Quarterly, 88(1), 176-194.
Miller, B. M. (2010). Community stakeholders and marketplace advocacy: A model of advocacy, agenda building, and industry approval. Journal of Public Relations Research, 22(1), 85-112.
Niven, D. (2001). Bias in the news: Partisanship and negativity in media coverage of presidents George Bush and Bill Clinton. Harvard International Journal of Press/Politics, 6(3), 31-46.
Patriotta, G., Gond, J. P., & Schultz, F. (2011). Maintaining legitimacy: Controversies, orders of worth, and public justifications. Journal of Management Studies, 48(8), 1804-1836. Pettigrew, J. E., & Reber, B. H. (2010). The new dynamic in corporate media relations: How
Fortune 500 companies are using virtual press rooms to engage the press. Journal of Public Relations Research, 22(4), 404-428.
Ragas, M. W., Kim, J., & Kiousis, S. (2011). Agenda-building in the corporate sphere: Analyzing influence in the 2008 Yahoo!–Icahn proxy contest. Public Relations Review, 37(3), 257-265.
Rindova, V. P., Williamson, I. O., & Petkova, A. P. (2010). Reputation as an intangible asset: Reflections on theory and methods in two empirical studies of business school
reputations. Journal of Management, 36(3), 610-619.
Schafraad, P., van Zoonen, W., & Verhoeven, P. (2016). The news value of Dutch corporate press releases as a predictor of corporate agenda building power. Public Relations Review, 42(3), 451-458.
Sleurs, K., Jacobs, G., & Van Waes, L. (2003). Constructing press releases, constructing quotations: A case study. Journal of Sociolinguistics, 7(2), 192-212.
Smedt, T. D., & Daelemans, W. (2012). Pattern for python. Journal of Machine Learning Research, 13(Jun), 2063-2067.
Smith, R. D. (2013). Strategic planning for public relations. Abingdon: Routledge.
Strömbäck, J. (2008). Four phases of mediatization: An analysis of the mediatization of politics. The International Journal of Press/Politics, 13(3), 228-246.
Strömbäck, J., & Nord, L. W. (2006). Do politicians lead the tango? A study of the
relationship between Swedish journalists and their political sources in the context of election campaigns. European Journal of Communication, 21(2), 147-164.
Strycharz, J., Strauss, N., & Trilling, D. (2018). The role of media coverage in explaining stock market fluctuations: Insights for strategic financial communication.
International Journal of Strategic Communication, 12(1), 67-85.
Stryker, J. E. (2002). Reporting medical information: Effects of press releases and
newsworthiness on medical journal articles' visibility in the news media. Preventive Medicine, 35(5), 519-530.
Stoker, K. L., & Tusinski, K. A. (2006). Reconsidering public relations' infatuation with dialogue: Why engagement and reconciliation can be more ethical than symmetry and reciprocity. Journal of Mass Media Ethics, 21(2-3), 156-176.
1Tateman, R., Sentiment Lexicons for 81 Languages (n.d.), Kaggle repository, https://www.kaggle.com/rtatman/sentiment-lexicons-for-81-languages
Tewksbury, D., & Scheufele, D. A. (2007). Special issue on framing, agenda setting, & priming: Agendas for theory and research. Journal of Communication, 57(1), 8-8. Thelwall, M., Buckley, K., Paltoglou, G., Cai, D., & Kappas, A. (2010). Sentiment strength
detection in short informal text. Journal of the Association for Information Science and Technology, 61(12), 2544-2558.
2Trilling, D., INCA, (2016), GitHub repository, https://github.com/uvacw/inca
Tucker, L., & Melewar, T. C. (2005). Corporate reputation and crisis management: The threat and manageability of anti-corporatism. Corporate Reputation Review, 7(4), 377-387.
1Kaggle is an online platform for predictive statistical modelling and data analytics on which organizations upload datasets and data experts compete to find the best solutions, which are all open sourced.
Tushman, M. L., & Katz, R. (1980). External communication and project performance: An investigation into the role of gatekeepers. Management Science, 26(11), 1071-1085. Verhoeven, P. (2009). Corporate actors in Western European television news. Public
Relations Review, 35(3), 297-300.
Van Aelst, P., & Vliegenthart, R. (2014). Studying the tango: An analysis of parliamentary questions and press coverage in the Netherlands. Journalism Studies, 15(4), 392-410. Vliegenthart, R. (2014). Moving up. Applying aggregate level time series analysis in the
study of media coverage. Quality & quantity, 48(5), 2427-2445.
Vliegenthart, R., Boomgaarden, H. G., & Boumans, J. W. (2011). Changes in political news coverage: Personalization, conflict and negativity in British and Dutch newspapers. In Political Communication in Postmodern Democracy (pp. 92-110). Palgrave Macmillan, London.
Vliegenthart, R., & Walgrave, S. (2008). The contingency of intermedia agenda setting: A longitudinal study in Belgium. Journalism & Mass Communication Quarterly, 85(4), 860-877.
Walgrave, S., & Van Aelst, P. (2006). The contingency of the mass media's political agenda setting power: Toward a preliminary theory. Journal of Communication, 56(1), 88 109.
Xiang, Z., Du, Q., Ma, Y., & Fan, W. (2017). A comparative analysis of major online review platforms: Implications for social media analytics in hospitality and tourism. Tourism
Appendix A Table 1A
Sample Overview
Organizations N press releases N newspaper articles
Abn Amro 726 2094 Aegon 58 441 Akzonobel 161 502 Asr 70 259 Bcd 42 9 Blokker 45 134 Boskalis 41 455 Delta 226 148 Douwe 39 1085 Eneco 113 44 For Farmers 87 441 Campina-Friesland 247 50 Heineken 121 1216 Heus 138 182 Nationale Nederlanden 33 38 Vandrie Group 23 37 NS 488 833 Post NL 232 1531 Rabobank 349 622 Shell 112 2052 Sligrofood Group 84 48 Tennet 239 141 Vion 63 150 Volkerswessels 65 46 Ziggo 84 1008 Total 3863 13566 Mean 309.04 542.64
Note. The amounts represent the sample after the removal of items which did not contain any words.
Appendix B
Table 1B
Overview of the news outlets from which the newspaper article sample was retrieved
News outlet N
AD (print) 836
AD (online) 247
ANP (online) 2266
Financieele Dagblad (print) 1625
Metro (online) 192
NOS (online) 485
NRC (print) 836
NRC (online) 674
Het Parool (online) 231
De Telegraaf (print) 1205 De Telegraaf (online) 2250 Trouw (print) 438 Trouw (online) 135 Volkskrant (print) 688 Volkskrant (online) 726 Total 13566 Mean 847.875
Note. The amounts represent the sample after the removal of articles which did not contain any words.
Appendix C Table 1C
Results sentiment analyses of Dutch newspaper articles
N SentiStrength Pattern.nl List Matching Verdict
Abn Amro 2094 -1.276 0.031 0.006 High
Aegon 441 -2.08 0.026 0.004 Low Akzonobel 465 -2.372 0.014 0 Low Asr 259 -2.413 0.018 0.005 High Bcd 9 -2.424 -0.005 0.008 High Blokker 134 -2.556 0.052 0 Low Boskalis 455 -2.572 0.042 0.004 High Campina-Friesland 148 -2.66 0.05 0.005 High Delta 1085 -2.581 0.026 0.0002 Low Douwe 44 -2.593 0.020 0.004 High Eneco 441 -2.608 0.093 0 High
For Farmers 50 -2.63 0.052 0.004 High
Heineken 1216 -2.724 0.055 0.004 High
Heus 182 -2.821 0.071 0.002 Low
Nationale Nederlanden 38 -2.838 0.021 0 Low
Vandrie Group 37 -2.856 0.093 0.004 High
NS 833 -2.875 0.021 0.002 Low
Post NL 1531 -2.882 0.024 0 Low
Rabobank 622 -2.887 0.023 0.003 Low
Shell 2052 -2.907 0.019 0.003 Low
Sligrofood Group 48 -3.013 0.053 0.00003 High
Tennet 141 -3.047 0.069 0.004 High
Vion 150 -3.133 0.028 0.002 Low
Volkerswessels 46 -3.220 0.050 0.005 High
Ziggo 1008 -1.00 0.065 0.003 High
Mean 542.64 -2.654 0.035 0.003 -
Appendix D Table D1
Correlations of the sentiment analyses
SentiStrength Pattern.nl List matching
SentiStrength - 0.151 0.267
Pattern.nl 0.151 - -0.164
List matching 0.267 -0.164 -
Appendix E Table E1
Topic scores LDA topic modeling Topic 1. CSR Topic 2. CSR Topic 3. Investments Topic 4. Events
Stupid 0.015 Energy 0.019 Netherlands 0.024 Festival 0.009 Responsible 0.008 Ocean 0.013 Finance 0.009 Concert 0.008
ACM 0.008 Wind farm 0.013 Bank 0.008 Nigerian 0.007
Food 0.005 Electricity 0.012 Takeover 0.006 New 0.006 Good 0.005 Responsible 0.008 Import 0.006 Netherlands 0.006 Topic 5. Online products/ services Topic 6. Sport/ Events Topic 7. Omzet Topic 8. CSR
Television 0.01 Soccer 0.01 Year 0.034 Glas 0.011
WWW 0.008 Sport 0.008 Million 0.031 Environment 0.009
New 0.008 New 0.007 Euro 0.015 Responsable 0.009
Application 0.007 Large 0.007 New 0.012 Possible 0.007 Online 0.006 Netherlands 0.006 Total 0.007 Citizens 0.005 Topic 9. Omzet Topic 10. Omzet Topic 11. International Business Topic 12. Competitiors
Netherlands 0.021 Million 0.039 Procent 0.024 Group 0.019
Year 0.013 Euro 0.029 Liberty 0.023 Share 0.015
Growth 0.008 Percent 0.012 Dollar 0.018 Market 0.009 Industry 0.007 Finance 0.009 Share 0.016 Product 0.007 Export 0.004 Importance 0.008 Global 0.013 New 0.007 Topic 13. Consumer Topic 14. Other Topic 15. Management Topic 16. International Business
Customer 0.006 Train 0.017 Board 0.037 Global 0.011
Netherlands 0.012 Bank 0.012 Commissioner 0.019 Industry 0.009
Age 0.01 Year 0.01 Minister 0.016 Bank 0.007
