MASTER THESIS POLITICAL COMMUNICATION, GRADUTE SCHOOL OF COMMUNICATION
Tweets as sources in
online and offline
news about the Dutch
E.U. elections of 2014
A study about the extent to which tweets are used
as sources in online and offline news articles by
Dutch national and regional newspapers about the
Dutch European Parliament elections of 2014.
Eva Huis in ’t Veld, 10000576, 27-06-2014, Supervisor: Sanne Kruikemeier Wordcount: 7,874
Abstract
This research examines to what extent journalists from Dutch regional and national offline and online newspapers use Twitter as a source in their news reports about the Dutch European elections of 2014. The study uses a quantitative content analysis and a sample of 121 news articles containing 203 tweets. The first results show that tweets from national political party leaders were used as a news source equally as much as tweets from their European
counterparts, indicating that journalists prefer national over European political actors. Online news articles about the Dutch European elections of 2014 source tweets the most and in general, journalists seem to prefer quoting factual tweets with a neutral tone that function as an illustration in their news reports. In general, this study extends our understanding of how journalists use Twitter as a source during European elections in terms of tweets their content, their authors and how they are displayed. More research in the field of Twitter is essential in defining the relationship between politicians, their tweets and the media.
Index
Introduction ... 1
Theoretical background ... 4
Method ... 12
Results ... 17
Conclusion and discussion ... 24
References ... 29
Appendix I: Codebook ... 33
Introduction
According to Weber Shandwick (2014), 93 percent of Dutch politicians with a seat in the Parliament uses Twitter. A large amount of politicians considers Twitter the most important social medium. Politicians mainly use Twitter to follow other politicians, experts and national news outlets, though politicians argue that they use Twitter to communicate with citizens. Also, journalists consider Twitter the most important social medium and more than half of them acknowledge using Twitter as a news source (Weber Shandwick, 2014). Yet,
traditionally, politicians and citizens communicate via traditional media. The rise of social media however might change this. Twitter empowers politicians and citizens, since they now have a platform to reach each other without the interference of journalists (Kruikemeier, 2014). It also empowers citizens, since they can easily reach out to politicians. This does not only make politics more accessible for the public, but might also give them increased
influence.
Although journalists seem, at a first glance, to be sidelined by the Internet, they can also use the opportunities social media brings. Social media can serve as a news source for journalists. This development might enrich journalism, by making the spread of news faster and
consisting of more information. It also creates a lower threshold for journalists when looking for sources. Using social media, journalists are able to reach politicians that they could not easily reach before. This easy access that journalists have to information also means lower costs for the entire process of collecting, writing and publishing news. Another possible consequence might be that it changes the role of the political journalist, since reporting is becoming more about what to amplify and what to ignore (Malik, 2012). Instead of asking politicians questions that are relevant in the eye of the journalist, Twitter only provides information about topics that politicians themselves find important. This could consequently
lead to journalists not going out anymore to look for politicians to interview. Instead they use quotes that they can find online. This means that politicians decide what journalists write about, giving them more power over journalists and news reporting (Graham, Broersma, Hazelhoff & van ‘t Haar, 2014).
When it comes to how journalists use sources in both offline and online news reporting, research has been done in the field of readers their perceived credibility of these news articles, and the factors that help increasing it (Abdulla, Garrison, Salwen, Discoll & Casey, 2002; Chung, Nam & Stefanone, 2012; Rennen, 2000; Sundar, 1998). The rise of the Internet, specifically online news, and how this changed journalistic values in terms of collecting news, has also been a popular topic of research (Ahmad, 2010; Bakker & Scholten, 2009; Carlsson, 2011; Gulyas, 2013; McManus, 1994; Westerman, Spence & Van Der Heide, 2012). Research on social media and how journalists use them to find news and collect sources, mainly focuses on general topics of news reporting and how tweets are used as sources within news coverage (Broersma & Graham, 2013; Cassidy, 2007; Kwak, Lee, Park & Moon, 2010; Singer, 2007). Only a few studies focus on how politicians use Twitter and/or how journalists source political tweets (Broersma & Graham 2012; Graham et al., 2014; Kruikemeier, 2014). The role Twitter plays during a second order election, like the European elections of 2014, has never been studied before. In addition, previous studies did not make a comparison between the use of tweets as news sources in online versus offline newspaper reporting. Such a
comparison is interesting, since it could be argued that in online news reporting, it is easier for journalists to include tweets. Twitter gives journalists the opportunity to link readers with the personal page of a Twitter user or a Twitter page that displays all tweets concerning one topic in the news article. Gaining more knowledge on how European politicians use Twitter during a European election is also interesting, since they are less known in the Netherlands compared
to national politicians. All this combined, this thesis examines to what extent journalists use Twitter as a news source in their reporting about the European Parliament election of 2014. In addition, this research examines to what extent there is a difference in the use of these tweets in offline and online news reporting. The following main research question will be answered:
‘To what extent are tweets used as sources in online and offline news articles from Dutch national and regional newspapers when reporting about the Dutch European Parliament elections of 2014?’
Answering this question will give us more insight into how journalists use Twitter as a news source during European elections. It also gives us more information on what kind of tweets are more often used and what actors are mostly selected as quoted sources. The distinction between online and offline news is also interesting, since it could be argued that tweets are more easily included by journalists in online news. Twitter enables journalists to include a gadget in their online news article that displays a certain amount of tweets from one or different users or topics. Journalists can also include a link to the personal page of a Twitter user or a Twitter page that displays all tweets concerning one topic in the news article.
This paper is divided into five parts. First the theoretical background is explained. The rise of social media and how this changed both politics and journalism is explained, and additionally, it will explain how Twitter is used as a news source in reporting by both politicians and journalists. Second, a detailed method section explains how the study is performed. It addresses the data selection, method of research and a detailed description of the codebook. Next, the research results will be described and subsequently the research question will be
answered, after which conclusions are drawn. Last, the discussion explains the theoretical and practical implications of this study.
Theoretical background
The importance of source use in news reporting
Sources are crucial for journalists when it comes to reporting the news (Rennen, 2000). The Society of Professional Journalists spends a great amount of attention in their code of ethics, a voluntarily journalistic guideline, to the importance of sources for journalists in reporting the news (Society of Professional Journalists [SPJ], 1996). Rennen (2000) even states that the quality of a newspaper, and thus its articles, is determined by the quality of its sources. The source of the news article itself also seems to be very important to readers in forms of
perceived quality and credibility, especially in online news (Abdulla et al., 2002). Chung et al. (2012) conclude that when online news articles use other online sources and links to other news sites, readers perceive them as more fair, objective and unbiased. By linking to a range of other news sources, they state, supplementary information is added that increases the quality and reliability of the information presented. Sundar (1998) also confirms readers their appreciation for source attribution in online news, by stating that it increases news articles’ perceived believability and objectivity and that readers evaluated the writing caliber of these stories superior to identical stories without source attribution. Extra important for perceived credibility are quoted sources, since readers estimated the quality and credibility of a story significantly higher when quoted sources were present.
When determining the type of sources that are used in news reporting, journalists seem to prefer official sources (van Dalen, 2011). In political news there is generally more attention for actors in power, meaning that government member receive more media attention than
members of parliament (Bennet, 1990; van Aelst, Shehate & Dalen, 2010; van Dalen, 2011). Ryfe (2009) states that this is the result of journalists going to predictable places like weekly governmental press conferences, press releases and official proceedings to collect news and sources, meaning that official sources have a higher chance of being the subject of news reporting.
The rise of online news and social media
The question that now remains is whether the rise of online news in general, and social media in particular, changed the way journalists use sources in news reporting. Carlsson (2011) argues that economical, technical, political and cultural developments changed the way journalists work. Newspaper sales have been decreasing for over a decade (Mediamonitor, 2013), indicating a decline in circulation (Bakker & Scholten, 2009), but also a decrease of journalistic capacity in newsrooms. This consequently means less to less time to verify sources, relevant news and background information (Commissie Brinkman, 2009), while the same amount, or maybe even more news articles have to be written. This automatically leads to the question of whether the quality of news articles can be preserved under less time and higher pressure to perform (McManus, 1994). But these developments, for a large extent caused by the rise of the Internet and (free) online news, also present new market
opportunities. Not only for traditional news organizations but also for nontraditional news sources like social media (Chung et al., 2012). Social media can be defined as a variety of online platforms that are built on the idea of collaborative creation and dissemination of content, centered on the principle of harnessing collective intelligence (Westerman et al, 2012). Social media can be conceptualized as a platform for users that enables them to create content, discuss that content in a collaborative effort to create content and eventually come to a shared understanding of that created content (Westerman et al., 2012).
According to Gulyas (2013) social media and the Internet in general changed the traditional practices of newsgathering, verifying stories and reporting, meaning that is has a significant impact on journalism. As mentioned earlier, the rise of the Internet and online news speeded up the process of news production, increasing journalists’ need for easily accessible
information (McManus, 1994). Livingstone and Bennet (2003), on the other hand, state that the rise of social media and technological development did not significantly change the routine journalists use when collecting news and sources, since official sources still tend to dominate news reporting. But even though journalists might still prefer official sources, social media can function as a tool for journalists to collect easy accessible information that can be used in news reporting. Instead of going out to conduct interviews, journalists can stay behind their computer and use social media to find newsworthy information and collect quotes (Gulyas, 2013). Especially Twitter, which allows users to post 140 characters long messages that can be read on the authors profile page by people that ‘follow’ the author, seems to be a very useful tool for journalists (Ahmad, 2010).
Twitter as a (political) news source for journalists
Twitter is used as a collaborative research tool by editors and journalists working on stories and blogs both for ideas and to provide evidence for all branches of news (Ahmad, 2010). Politics is one of the topics in which tweets are mostly used as sources, together with human interest, sports and media (Broersma & Graham, 2013). Even though Kwak et al. (2010) found that CNN Headline news turned out to be ahead in reporting breaking news over half of the time compared to trending topics on Twitter, they state that Twitter does seem to fulfill journalists with the information they need (Kwak et al., 2010). When it comes to political news, journalists can collect information and quotes from politicians online by visiting their Twitter page, instead of going out and interview them. This makes Twitter an easy and rather
cheap tool to collect possible interesting information and it might even give journalists the opportunity to quote politicians that they cannot get in front of the camera, because they for example do not want to do interviews or talk to journalists (Graham et al., 2014). But are there any differences in the use of tweets as sources in different types of news? Broersma and Graham (2013) found that in general, popular newspapers sourced tweets more often than quality newspapers, even though the quality press discovered Twitter first. Since there is not yet any research available on the extent to which journalists use tweets as sources in online news, a focus on how journalists perceive online information as credible might give more information. When it comes to the perceived credibility of online information, research shows that online journalists rate Internet news information as significantly more credible than offline journalists. Or in other words, Internet reliance seemed to be a strong positive predictor of credibility (Cassidy, 2007). Friend and Singer (2007), on the other hand, state that even though social media are interesting for journalists, they mainly hold on to traditional sources. Since there does not seem to be a clear direction in the just presented theory, the following sub-research question is stated:
RQ1 ‘To what extent do newspapers and their online and offline news articles about the Dutch European elections of 2014 differ in using tweets as sources?’
The content of tweets used as news sources
Journalists use tweets as news sources for different purposes: (1) tapping into the private sphere of well known and newsworthy people, who can only be reached through Twitter. An example of this is Dutch politician Geert Wilders, who did not want to talk to journalists for quite a while but extensively used and still uses Twitter. Twitter can also be used as an illustration (2), meaning that a range of tweets is used that represents a different range of
people or opinions. Twitter can lastly be used to trigger the news report itself (3), meaning that an article gets published because of the content of a tweet (Broersma & Graham, 2013). In classic agenda setting theory the media decides what is news and what is not, meaning that they are in control of what the important subjects of news are and what citizens should think about (McCombs, 1972). When politicians write a tweet with the purpose of it getting picked up by a journalist, this can be perceived as politicians setting the agenda. In this scenario, it is not the media alone who decides on what is news, but also the politicians, who give
journalists the information that they want to become news.When it comes to political news though, there were almost no tweets that triggered a news story. Broersma and Graham (2012) found that in the United Kingdom and the Netherlands, political tweets were mostly used as an illustration in news articles that present news events, reportages or background stories about national elections, meaning that they do not serve as an essential part of the news article. Tweets also seemed to be used by columnists to illustrate an argument in a quote (Broersma & Graham, 2013). This leads to the following hypothesis:
H1: Political tweets presented as sources in news articles about the Dutch European elections of 2014 will be more often used as an illustration of news than as a trigger for news.
When it comes to elite sources, like politicians, Twitter can be seen as a tool to obtain more control over the public disclosure (Broersma & Graham, 2012). By using Twitter, politicians can exclude the media from the process of spreading their message and communicate directly with the public (Graham et al., 2014). Not per se because politicians and political news are not interesting to news media and journalists, but because Twitter gives politicians the
quality newspapers are interested in these tweets, since, compared to all other news
categories, tweets are used the most as sources in political news (Broersma & Graham, 2013). Journalists also almost exclusively quote political tweets in full (Broersma & Graham, 2012; 2013). This means that, to a certain extent, politicians gain control over the news discourse. Journalists seem to rely on statements without contacting the politicians and thereby partially abandon their duties by downplaying the responsibility for the information presented in the tweets (Broersma and Graham, 2013). This indicates a general shift in the power balance between journalists and politicians, in the benefit of the latter (Broersma & Graham, 2012). Concluding, it seems to be the case that both politicians and journalists prefer displaying tweets as quotes. That is the case for politicians – they get their message spread in the exact words they used – and journalists as well, since they can not only downplay the responsibility for the information that is presented in the tweets, but also because they can save time by quoting instead of paraphrasing. Based on this, the following hypothesis is formulated:
H2: Political tweets used as sources in news articles about the Dutch European elections of 2014 will be quoted more often than paraphrased.
Another interesting question is whether there is a difference in the use of tweets as quoted or paraphrased sources in online and offline news articles. As already mentioned, online news gives journalists the opportunity to add interactivity to their news articles (Chung et al., 2012). Online journalists can link to tweets or Twitter pages in their news articles. Linking to tweets by using a gadget in which different tweets can be displayed can be an effective way for journalists to quote and include tweets in their news articles. Since no further research on this topic exists, the following sub-research question is stated:
RQ2 ‘To what extent are there differences in the amount of tweets that are quoted or
paraphrased in online and offline news articles about the Dutch European elections of 2014?’
Politicians and their tweets during elections
As Kruikemeier (2014) notices correctly, Twitter is personalized per definition. As the political candidate is most of the time the owner and holder of the Twitter account, it is the perfect vehicle for self-promotion. She also emphasizes the importance of Twitter for
politicians themselves when it comes to collecting votes, especially since Twitter use among politicians increases during the course of an election. But what do politicians tweet about most of the time? Mainly personal information? Broersma and Graham (2012) found that tweets with factual statements, arguments or opinions are the most dominant type of sourced tweets in political news in both the U.K. and the Netherlands, besides other tweet-characters like conveying an acknowledgement, giving advice or giving directives. Graham et al. (2014), who examined tweets from U.K. politicians, also agreed on this. They found that slightly more than two-third of politicians’ tweets were about broadcasting particular factual information about for example the campaign, promoting themselves or party members, or critiquing opponents. There were also differences between the two countries, since U.K. newspapers seemed to cite humorous tweets, while Dutch newspapers rarely cited these. All of this leads to the following hypothesis:
H3: Political tweets presented as sources in news articles about the Dutch European
Tweets as news sources in articles about European elections
As already mentioned before, European Parliament elections have never been very popular in general (Franklin, 2006). In the Netherlands they can be considered ‘second-order’ elections (Irwin, 1995). Over the past ten years, between 30 and 40 percent of the eligible voters have actually voted during an European election, indicating how unpopular these elections are in the Netherlands (“Opkomst Europese verkiezingen…”, n.d.). As earlier mentioned, research on the extent to which journalists use political tweets as sources in news reporting proves that journalists do not only use Twitter as a collaborative research tool in general (Ahmad, 2010; Broersma & Graham, 2013), but also during national elections (Broersma & Graham, 2012). An interesting question is whether this will also count for tweets from European politicians. Will journalists prefer their tweets in news articles about the Dutch European elections of 2014, or will they still prefer tweets from national politicians, who are, unlike European politicians, well known by the public? By using more tweets from national politicians than European politicians as sources in news articles about European elections, journalists might give readers the (probably unintended) impression that national politics are of higher importance than European politics. Also, by preferring tweets from national politicians, journalists prevent readers from learning more about European politicians and the European elections. This leads to the following sub-research question:
RQ 3 ‘Are national political actors more often used as Twitter sources in news articles about the Dutch European elections of 2014 than non national political actors?’
When it comes to predicting the tone of the tweets that are selected as sources in news articles about the Dutch European elections of 2014, research results contradict. One the one hand, Harcup and O’Neill (2001) tell us that negativity is one of the most important news values,
indicating that politicians might use this to make sure their tweets get picked up by
journalists. On the other hand, this negativity can be perceived as negative campaigning by voters, which might have negative consequences for politicians in terms of demobilizing voters (Ansolabehere & Iyengar, 1995; Ansolabehere, Iyengar & Simon, 1999) or the so called ‘boomerang effect’, meaning that the negativity backlashes on the sender of the message (Lau, Sigelman & Rovner, 2007). What might also influence the tone of a tweet is the emotion it contains. Besides negative emotions like anger and fear, more positive emotions like joy or pride might also be present in tweets. Unfortunately the effect of
emotions in political tweets has never been researched before. The use of emotions in political campaigns on the other hand is a tool that politicians use to appeal to voters (Brader, 2005), what might indicate that this emotional appeal is also used in their tweets. The previous discussion leads to the following sub-research questions:
RQ 4 ‘To what extent do tweets that are used as sources in news articles about the Dutch European elections of 2014 have a positive, negative or neutral tone?’
RQ5 ‘To what extent do the tweets that are used as news sources in news articles about the Dutch European elections of 2014 contain specific emotions?’
Method
To answer the hypotheses and research questions, a quantitative content analyses is employed to determine the extent to which tweets are used as news sources in both online and offline news articles from national and regional newspapers that report about the Dutch European elections of 2014. The codebook for this study is partially based on the codebooks used by Broersma and Graham (2012; 2013).
Data collection
The Dutch European Parliament elections were held at the 22nd of May 2014. Since the official results were not released until the 27th of May, data collection ended at the 28th of May, so that all relevant news articles about the election could be included. Data collection started one month before, at the 24th of April. All Dutch newspapers (i.e. national and regional) are included1. Besides offline news articles that appeared in the printed versions of national newspapers, articles from national newspapers their websites are also included. Since not all national newspapers have structured online archives, online articles are only included from national newspapers that own a structured online archive. Besides online and offline articles from national newspapers, all regional newspapers are selected. The Netherlands has a broad selection of regional newspapers (Bakker & Scholten, 2009). Some are dedicated to an entire province, some only to a certain region. The offline news articles are collected through the Lexis Nexis database. For all articles the same selection procedure is used, meaning that they are only selected if they contain the words ‘tweet’, ‘twitter’ or ‘twittered’ and ‘Europe’ and ‘election’. The articles are selected by doing a power search with the search string ‘(tweet OR twitter OR twitterde) AND (europ! AND verkiezing!)’. As mentioned earlier, the online news articles are selected through the online archives of the newspapers. The same search terms are used for this. Since most of these online archives show all published online articles per day, the control + F function on a computer is used to select articles containing one of the five required words. All articles, offline and online, were read before they were included in the sample, meaning that articles that did not meet the requirements were deleted from the original sample. Articles that contained the word ‘Twitter’ or ‘tweet’, but did not in fact quote of paraphrase a specific tweet, or articles about European elections in another European
country were not selected. For offline newspaper articles, this meant that 27 out of 83 news articles were selected. Since online articles are selected by hand instead of a database, the amount of deleted articles cannot be determined.
Table 1
Number of offline and online articles selected from national newspapers
Newspaper Offline articles2 Online articles3
Volkskrant 4 15 Algemeen Dagblad 1 5 NRC Handelsblad 7 - NRC next 4 - NRC.nl4 - 13 Telegraaf 2 - Trouw 3 5 Parool 1 - Financieel Dagblad 0 3 Reformatorisch Dagblad 4 2 Nederlands Dagblad 1 3 Metro 0 - Spits 0 2
As described in Table 1, a total amount of 27 offline (or printed) news articles from national newspapers are selected. When it comes to online news, a total amount of 48 online news articles are selected from national newspapers their websites. Even though not all newspapers were used to collect online news articles, a higher amount of articles that contained tweets could be selected.
2
Some newspapers did not use tweets as sources in any of their news articles about the Dutch European elections of 2014, meaning that they score ‘0’.
3 Not all national newspapers are used when collecting online news articles, meaning that they score a ‘-’. 4
Since NRC Handelsblad and NRC next share one online webpage for their online news articles, the total amount of online news articles is displayed here under NRc.nl.
Table 2
Number of articles selected from regional newspapers
Newspaper Articles Newspaper Articles
Amersfoortse Courant* 2 De Gelderlander 2
De Dordtenaar* 2 Apeldoornse Courant** 1
Haagsche Courant* 3 Dagblad Flevoland** 1
Rivierenland* 2 Deventer Dagblad** 1
Rotterdams Dagblad* 2 Gelders Dagblad** 1
Utrechts Nieuwsblad* 2 Nieuw Kamper Dagblad** 1
Groene Hart* 3 Veluws Dagblad** 1
Almere Vandaag 0 Zwolse Courant** 1
BN/Destem 2 De Twentsche Courant Tubantia 2
Brabants Dagblad 1 Einhovens Dagblad 1
Dagblad de Limburger 3 Leeuwarder Courant 2
Limburgs Dagblad 3 Provinciale Zeeuwse Courant 2
Dagblad van het Noorden 2 Leidsch Dagblad 2
De Gooi- en Eemlander 1
* All newspapers are part of Algemeen Dagblad but add own news ** All newspapers are part of de Stentor but add own news
As described in Table 2, a total amount of 46 offline (or printed) news articles from regional newspapers are selected. Taking this all together, the entire sample (N) of online and offline news articles from national and regional newspapers consists out of 121 news articles, from which 27 are printed articles from national newspapers, 48 are online articles from national newspapers their websites and 46 are printed articles from regional newspapers.
Table 3
Number of tweets per newspaper category
Newspaper category News articles Tweets Average amount of tweets per news article
National offline 27 34 1.26
National online 48 105 2.19
Besides this, the number of collected tweets is also important. The total number of collected tweets is 203 (N). As Table 3 shows, the 27 national offline newspaper articles contain 34 tweets, an average of 1.26 tweets per article. The 48 national online newspaper articles contain 105 tweets, an average of 2.19 tweets per article, and the 46 regional offline newspaper articles contain 64 tweets, an average of 1.40 tweets per article.
Codebook
As earlier mentioned, the codebook used for this study is partially based on two codebooks from Broersma and Graham (2012; 2013), who conducted a similar study. Besides some general information like the date that the article is published, the headline, the newspaper in which it is published and whether the article is published in the print version of the newspaper or online, other variables are included to measure information about both the news article and the tweet(s) they contain. When it comes the news article, the genre and topic was measured. The genre of the news article can differ between a news report: a compact article in which news, being an unexpected event or fact, is reported, a reportage/feature: a longer newsworthy story in which individual experiences reflect reality for a large amount of people, and, a background story/ news analysis: an overview and interpretation of the most important news facts that are presented in the newspaper, resulting in a conclusion that is open for other interpretations (“NRC Stijlboek”, n.d.). Besides this the option live-blog – an online news article in the style of a blog on which journalists update small pieces of information about a certain event – is also available. These live-blogs are published as an online news article after the news event has finished. For the articles that do not fit into one of these categories, the ‘other’ option is available. The question concerning the topic of the news article measures fourteen standard news topics differing from politics and government to lifestyle and sports.
In the second part of the codebook, the unit of analysis shifts from the news article in general to the tweet in particular. Besides the identification of the author of the tweet by last name, the occupation of the author and in some cases the political party of which he or she is a member is measured. After this, a question about how the tweet is reported – quoted or paraphrased – has to be answered. The function of the tweet in the news article is also measured. The style of the tweet, formal or informal, is also measured, just like the character of the tweet. Tone of voice of the tweet determines whether a tweet is positive, negative or natural. Last, a wide range of emotions can be picked to define whether the tweet contains a specific emotion. All variables in this second part of the codebook are measured on a nominal level. For more information, the full codebook is included in Appendix I.
Inter-coder reliability
To preserve the viability of the coding, a team of two coders coded 8.3 percent of the entire sample. This means that seventeen out of 203 randomly selected tweets were coded twice. Krippendorff’s Alpha was used to determine the level of agreement between the two coders, since almost all variables are measured on a nominal level. All variables but one had a score higher than .90, meaning that they could be used for the official coding process without making further adjustments. The variable that scored lower, but still higher than .70, is the variable that measures the tone of the tweet. Before this variable was included into the final codebook, the two coders redefined the selection criteria per item, so that they were made exclusive. Appendix II gives a full overview of the results.
Results
Since the amount of articles that newspapers published about the Dutch European elections of 2014 strongly differs per newspaper, it is not possible to make credible comparisons between
groups of newspapers like for example, quality and popular newspapers or national and regional newspapers. Instead, the focus will be on the tweets that are sourced in the news articles. The tweets are categorized by the newspaper and article type in which they are used as sources, being a national offline news article, a national online news article or a regional offline news article.
Figure 1 Amount of tweets used as sources per newspaper type
First we take a look at the amount of tweets that are published per newspaper and per news article type (RQ1). This amount significantly differs, chi square (2) = 37.55, p < 0.001. As Figure 1 shows, tweets are used more in online news articles from national newspapers than expected (n = 105), and less in offline news articles from national (n = 34) and regional (n = 64) newspapers than expected. The total amount of tweets used as sources in news articles about the Dutch European elections of 2014 is 203 (N). To answer sub-research question 1: newspapers strongly differ in the amount of tweets they use as sources in their news articles. Online news articles from national newspapers use tweets significantly more as sources than
105 34 64 0 20 40 60 80 100 120
Online national news articles Offline national news articles Offline regional news articles Tweets
other article types. Offline news articles from both regional but especially national newspapers use tweets significantly less as sources.
When looking at the function of tweets used as sources in news articles about the Dutch European elections of 2014, it becomes clear that Dutch newspapers never made use of the so-called ‘standalone’ tweet, a tweet that is published without extra information, or conducted an interview over Twitter. For testing hypothesis 1, these categories will therefore not be used, since otherwise using a chi square test is not allowed. There is a significant difference in the frequencies of the two other functions of tweets, chi square (1) = 134.11, p < 0.001. Tweets are more often used as an illustration in news articles (n = 184) and less as trigger of news articles (n = 19). When including newspaper categories, this significant difference in frequencies still exists, chi square (2) = 8.86, p < 0.05. But this significant difference only counts for one category: tweets trigger online news articles from national newspapers slightly more than expected (n = 16). Hypothesis 1 can be accepted.
When it comes to the amount of times that tweets are quoted or paraphrased, there is a significant difference in frequencies, chi square (1) = 115.35, p < 0.001. Surprisingly, tweets are more often quoted (n = 178) in news articles and less often paraphrased (n = 25) than was expected.
Figure 2 Display of tweets in news articles per newspaper type and published place When newspaper and article type are taken into account (RQ2), there is a significant
difference between news article type and how the tweet is displayed, chi square (2) = 17.65, p < 0.001. As Figure 2 shows, tweets displayed in online news articles from national
newspapers are less paraphrased than expected and tweets displayed in offline news articles from national newspapers are more paraphrased than expected. Hypothesis 2 can thus be accepted. When only the distinction between online and offline news is taken into account, like stated in research question 2, this significant difference in frequencies of sourced tweets that are coded or paraphrased is still there, chi square (1) = 14.57, p < 0.001. As Figure 2 shows, paraphrased tweets are used more in offline news articles than expected (n = 21) and less in online news (n = 4) than expected. But, as can be seen, an overwhelming amount of tweets is quoted in both online and offline news (178 out of 203).
101 24 53 4 10 11 0 20 40 60 80 100 120 National online news articles National offline news articles Regional offline news articles Quoted Paraphrased 101 77 4 21 0 20 40 60 80 100 120
Online news articles Offline news articles Quoted Paraphrased
Looking at the different character of the tweets that are used as sources in news articles about the Dutch European elections of 2014, a significant difference can be seen in frequencies, chi square (4) = 557.76, p < 0.001. Tweets with a factual character are displayed extensively more than expected (n = 174), and all other characters are displayed less than expected.
Figure 3 Character of the tweets that are used as sources
As Figure 3 shows, tweets with a factual character are used the most, being 175 times, nine tweets convey an acknowledgement, one gives advice, six give directives and twelve have an ‘other’ character that could not be specified. Hypothesis 3 can thus also be accepted.
Now that we know more about the type of newspapers and news articles that use tweets as sources, and about the content of these tweets, it is interesting to look into whose tweets are selected by journalist to function as sources in news articles about the Dutch European elections of 2014 (RQ3). There is a significant difference in the frequencies in which tweets from different sources are used, chi square (6) = 128.76, p < 0,001. European parliament
175 9 1 6 12 0 20 40 60 80 100 120 140 160 180 200 Factual Conveying an acknowledgement
Adive giving Directives Other
candidates’ tweets are used more than expected (n = 68), followed by Dutch national politicians (n = 50) and journalists (n = 41). Prominent E.U. members’ tweets are used less than expected (n = 2), just like tweets from organizations linked to the E.U. (n = 2) and citizens (n = 18). Tweets from authors with another occupation than the ones just mentioned were also used less frequent then expected (n = 22). The authors of the sourced tweets were most of the time also a member of a Dutch political party. Out of a total of 203 tweets (N), 121 are coming from authors that are somehow connected to a Dutch political party.
Figure 4 Dutch political parties from which some of the authors of the tweets are a member
There is a significant difference in the use of tweets as sources from authors that are also a member of a Dutch political party, chi square (9) = 207.02, p < 0.001. As Figure 4 shows, tweets written by somebody that is a PVV member are sourced more than expected (n = 56) and are also used the most in news articles about the Dutch European elections of 2014. Tweets written by D66 members are also used more frequently than expected (n = 22). Tweets from Christenunie/SGP members (n = 8), VVD members (n = 8), CDA members (n =
1 13 1 22 5 8 56 4 8 3 0 0 0 10 20 30 40 50 60 Number of tweets
5), Partij van de Dieren members (n = 4), 50 Plus members (n = 3), GroenLinks members (n = 1) and SP members (n = 1) were used less frequent than expected. The Piratenpartij and the rest of the small and fairly unknown Dutch parties that participated were not used as sources at all (n = 0). But who are the authors of the tweets from the PVV and D66, who are used the most as sources in news articles about the Dutch European elections of 2014? For the PVV the tweets from European Parliament first candidate Marcel de Graaff are used the most (n = 37), followed by party leader Geert Wilders (n = 18) and Lucas Hartong (n = 1), who recently left the PVV. For D66, tweets from party leader Alexander Pechtold are used the most (n = 10), followed by European Parliament candidate Pauline Kastermans (n = 8), followed by four members from who one tweet each was used, including Sophie in ‘t Veld, their first candidate for the European Parliament (n = 4).
Turning to the tone of the tweets that are used as sources in news articles about the Dutch European elections of 2014 (RQ4): the frequencies between positive, negative and neutral news significantly differ, chi square (2) = 19.46, p < 0.001. Neutral tweets are reported the most and more frequent than expected (n = 96), while positive (n = 46) and negative (n = 61) tweets are used less frequent than expected. Besides the tone of the tweets that are sourced, some also contain a specific emotion (RQ5). Out of the total of 203 tweets, 74 contained a specific emotion, meaning that significantly more tweets (n = 129) did not contain a specific emotion, chi square (1) = 14.90, p < 0.00. The frequencies from these tweets that do contain a specific emotion significantly differ, chi square (6) = 100.43, p < 0.001.
Figure 5 Emotions displayed in sourced tweets
As displayed in Figure 5, from the 74 tweets that do contain an emotion, joy is the one most frequently used, and also used more than expected (n = 39). Amazement is also used more frequent than expected (n = 13). The emotions disgust (n = 4), anger (n = 3), admiration (n = 2), and love (n = 2) are all less frequently used than expected. All other emotions, being grief, fear, jealousy, hope, reproach, pride, regret, hate, shame and guild were not used at all.
Conclusion and discussion
This paper investigated the extent to which online and offline journalists use Twitter in news articles about the Dutch European elections of 2014, how they are reported, who wrote them and what the content of these tweets is. Twitter has over the years become a research tool for journalists that provides them with evidence for all branches of news (Ahman, 2010), and political news has become the category in which tweets are used the most as sources by journalists (Broersma & Graham, 2013).
39 3 13 4 2 11 2 0 5 10 15 20 25 30 35 40 45
Joy Anger Amazement Disgust Admiration Reproach Love Tweets
The first thing standing out is that the amount of articles written about the Dutch European elections of 2014 in general is a lot lower than during national elections, but also that not a lot of articles contain tweets as sources. This is in line with what Franklin (2006) and Irwin (1995) stated about the unpopularity of European elections in the Netherlands. Online national newspapers sourced tweets the most in their news articles, while offline articles from regional newspapers and offline articles from national newspapers sourced tweets less. Especially the latter did not often use tweets as sources.
When tweets are used as sources in news articles about the Dutch European elections of 2014, they are generally quoted more often than paraphrased. This is in line with earlier research by Broersma and Graham (2012; 2013). The lowest amount of paraphrased tweets is used in online news, while in offline news articles tweets are paraphrased more than expected. The greatest amount of tweets in general are quoted. Combining this with the fact that tweets overall mainly serve as an illustration in news articles, which is in line with what Broersma and Graham (2012; 2013) found, it seems that quoting tweets can be the result of
convenience. By quoting, journalists downplay the responsibility for the information
presented in the tweets (Broersma & Graham, 2013) and by including a Twitter gadget they can easily display a large amount of sources without spending time on it. This indicates that social media and especially Twitter changed journalistic values (Carlsson, 2011). What is also interesting is that in online news, tweets sometimes also triggered news reports. This could be a result of politicians writing a tweet with the purpose of it getting picked up by the media and/or journalists looking for interesting tweets that they can instantly report. To a certain extent, politicians thus have an increased influence in setting the media agenda when using Twitter. Character wise, tweets that are used as sources mostly have a factual character, which is also in line with former research (Broersma and Graham, 2012; 2013). This indicates that
journalists prefer using tweets that contain an argument or opinion, and is a possible indication for the new journalistic routine that Gulyas (2013) describes as collecting statements through Twitter instead of going out and interviewing politicians. The overall neutral tone of the tweets that are used as sources probably confirms politicians their fear of negative publicity that might be the result of making negative comments on Twitter. Even though politicians might use emotions in political campaigns to appeal to voters (Brader, 2005), only one third of the tweets contained one. Joy was most frequently displayed, followed by the way less frequent used emotion amazement. It could be the case though that in neutral tweets, no specific emotions could be detected, while in positive tweets, emotions like joy or amazement were easily detectable.
When reporting about the Dutch European elections of 2014, journalists seem to prefer tweets from official sources, which is in line with what van Dalen (2011) found. Tweets from European Parliament candidates and tweets from Dutch National politicians were used as sources almost equally as much, indicating that journalists seem to prefer Dutch national party leaders over their European counterparts. This could again be the result of the unpopularity of European elections in the Netherlands (Franklin, 2006; Irwin, 1995), or a result of national politicians helping their European counterparts campaigning, since Dutch citizens probably know them better. Tweets from PVV authors are sourced the most, followed by D66 authors. The PVV is known as a party that does not talk to journalists quite often, which makes their tweets an adequate alternative for journalists. D66 leader Alexander Pechtold is a fervent Twitter user and at the moment the party is doing well in election polls, which might make these tweets more interesting for journalists. Since both parties had a news article written about them that was massively copied by almost every regional newspaper in the sample, it is
fair to say that journalists preferred tweets from the national party leaders over tweets from their European counterparts.
Overall, using tweets as sources in news articles about the Dutch European Parliament elections of 2014 did not seem to be very popular among journalists, just like the European elections themselves. Tweets are mostly used as sources by online news articles. And overall, journalists seem to prefer quoting factual tweets that are used as an illustration in their news articles and have a neutral character. By quoting tweets instead of paraphrasing tweets and mostly displaying tweets from politicians that are not available elsewhere, journalists set a certain tone: politicians have a higher chance of getting their message spread when they are not available for journalists and instead write easy to quote tweets. This gives politicians a certain amount of power over the content of news reporting. Thus, instead of Twitter making politics more available to citizens, it can also, or maybe even instead, become a tool that politicians can use to influence news reporting to an extent that is not desirable. Journalists are supposed to take on a critical stance against politicians and be a watchdog instead of blindly and unquestioningly copying their tweets when they lack time or other sources. An important question then is: who carries the responsibility for this? Journalists, who decide on how they collect information, what stories they write and who to quote. Or politicians, who provide the information that journalists need.
Unfortunately, this study has a few shortcomings. Not all online news from all national newspapers could be used, since not all newspapers have well-structured online archives. Including more outlets might have made the results presented above more convincing, even though the current sample does give a good impression about the fact that tweets are sourced by online journalists more often. Besides this, it might also be interesting to include several
online news outlets in further research, to see if there are any differences between news outlets in using tweets as sources. Another interesting and also worrying finding is the amount of times where regional newspapers especially, but sometimes also national newspapers, publish the same news articles. This may have influenced the outcome of the results. Even though this is not the main focus of this paper, further research should pay attention to the extent to which journalists rely on the same sources when collecting news, like for example material from press agencies5. In general, this study can be regarded as a beginning in understanding how Dutch journalists use Twitter during European elections. It provides a good impression on the content of the sourced tweets, who wrote them and how journalists publish them in their news articles. More research in the field of Twitter is essential in defining the relationship between politicians, their tweets and the media.
5
For more information on this subject, see Huis in ‘t Veld, E.N. (2013). Knip – plak – publiceer. (Bachelor’s thesis). Universiteit van Amsterdam.
References
Abdulla, R. A., Garrison, B., Salwen, M., Driscoll, P., & Casey, D. (2002). The credibility of newspapers, television news, and online news. In Education in Journalism Annual Convention, Florida USA.
Ahmad, A. N. (2010). Is Twitter a useful tool for journalists?. Journal of Media Practice, 11(2), 145-155.
Bakker, P., & Scholten, O. (2009). Communicatiekaart van Nederland. Overzicht van Media
en Communicatie. Amsterdam: Kluwer.
Bennett, W. L. (1990). Toward a theory of press-state relations in the United States.
Journal of Communication, 40(3), 103-127.
Brader, T. (2005). Striking a responsive chord: How political ads motivate and persuade voters by appealing to emotions. American Journal of Political Science, 49(2), 388-405.
Broersma, M., & Graham, T. (2012). Social media as beat: tweets as a news source during the 2010 British and Dutch elections. Journalism Practice, 6(3), 403-419.
Broersma, M., & Graham, T. (2013). TWITTER AS A NEWS SOURCE: How Dutch and British newspapers used tweets in their news coverage, 2007–2011.Journalism Practice, 7(4), 446-464.
Cassidy, W. P. (2007). Online news credibility: An examination of the perceptions of newspaper journalists. Journal of Computer‐Mediated Communication, 12(2), 478-498.
Chung, C. J., Nam, Y., & Stefanone, M. A. (2012). Exploring online news credibility: The relative influence of traditional and technological factors. Journal of
Computer‐Mediated Communication, 17(2), 171-186.
2013. Hilversum. Commissariaat voor de Media.
Franklin, M. (2006). European elections and the European voter. European Union: power and
policy-making, 3, 227-246.
Graham, T., Broersma, M., Hazelhoff, K., & van't Haar, G. (2013). BETWEEN BROADCASTING POLITICAL MESSAGES AND INTERACTING WITH VOTERS: The use of Twitter during the 2010 UK general election campaign.
Information, Communication & Society, 16(5), 692-716.
Gulyas, A. (2013). The influence of professional variables on journalists ‘uses and views of social media: a comparative study of Finland, Germany, Sweden and the United Kingdom. Digital Journalism, 1(2), 270-285.
Harcup, T., & O'Neill, D. (2001). What is news? Galtung and Ruge revisited. Journalism
studies, 2(2), 261-280.
Irwin, G. (1995). Second-order or third-rate?: Issues in the campaign for the elections for the European Parliament 1994. Electoral Studies, 14(2), 183-199.
Koppen, C. (2012, October 24). De veranderende rol van de media: sociale media als bron. Howard’s home, from: http://w3.howardshome.com/2012/10/24/de-veranderende-rol-van-de-media-social-media-als-bron/
Kruikemeier, S. (2014). How political candidates use Twitter and the impact on votes. Computers in Human Behavior, 34, 131-139.
Kwak, H., Lee, C., Park, H., & Moon, S. (2010, April). What is Twitter, a social network or a news media?. In Proceedings of the 19th international conference on World wide
web (pp. 591-600). ACM.
Lau, R. R., Sigelman, L., & Rovner, I. B. (2007). The effects of negative political campaigns: a meta‐analytic reassessment. Journal of Politics, 69(4), 1176-1209.
technology altering the construction of news?. Political Communication,, 20(4), 363-380.
Malik, O. (2012, October 13). Amplification & the changing role of media. Gigaom, from: http://gigaom.com/2012/10/13/amplification-the-changing-role-of-media/#
McCombs, M. E., & Shaw, D. L. (1972). The agenda-setting function of mass media. Public
opinion quarterly, 36(2), 176-187.
McManus, J.H. (1994). Market-Driven Journalism: Let the Citizen Beware? Thousand Oaks: Sage.
NRC Stijlboek. (n.d.). Retrieved May 8, 2014, from
http://apps.nrc.nl/stijlboek/stijlgids/journalistieke-genres
Opkomst Europese verkiezingen in Nederland (n.d.). Retrieved May 21, 2014, from
http://www.parlement.com/id/vhx1b38yhhwz/opkomst_europese_verkiezingen_in Rennen, T. (2000). Journalistiek als kwestie van bronnen. Delft: Eburon.
Ryfe, D.M. (2009). Broader and deeper: A study of newsroom culture in a time of change. Journalism, 10(2), 197-216.
Society of Professional Journalists (1996). SPJ code of ethics. Retrieved May 12, 2014 from http://www.spj.org/ethicscode.asp
Sundar, S. S. (1998). Effect of source attribution on perception of online news stories. Journalism & Mass Communication Quarterly, 75(1), 55-68. Van Aelst, P., Shehata, 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.
Van Dalen, A. (2011). Structural bias in cross-national perspective: How political systems and journalism cultures influence government dominance in the news.
The International Journal of Press/Politics 17(1), 32-55.
Verweij, p. (2010). Twitter als een nieuwsbron voor journalisten.
Weber Shandwick. (2014). Twitter en de Tweede Kamer. Den Haag: Weber Shandwick NL. Westerman, D., Spence, P. R., & Van Der Heide, B. (2012). A social network as information:
The effect of system generated reports of connectedness on credibility on Twitter. Computers in Human Behavior, 28(1), 199-206.
Appendix I: Codebook
Introduction
This codebook is used for coding offline and online newspaper articles about the Dutch European Parliament elections of 2014. The articles that are selected for coding are all using tweets as sources. The questions that need to be answered about these news articles are
divided into two parts. In part A general questions about the news article need to be answered, in part B the questions specifically focus on the tweet(s) used in the news article. When an article contains more than one tweet, this form needs to be filled in just as many times as the amount of tweets presented in the news article. The answers to part A will then not differ, since this part only asks questions about the news article and not the tweet.
A: general information
1a. Is the newspaper a national or a regional newspaper? 1. National answer question 1b
2. Regional answer question 1c
1b. What national newspaper? (only when question 1a is answered with ‘1. National’):
1. Volkskrant 8. Financieel Dagblad
2. Algemeen Dagblad 9. Reformatorisch Dagblad
3. NRC Handelsblad 10. Nederlands Dagblad
4. NRC next 11. Metro
5. Telegraaf 12. Spits
6. Trouw 13. NRC.nl (only for online NRC articles!)
7. Parool
1c. What regional newspaper? (only when question 1a is answered with ‘2. Regional’): 1. Amersfoortse Courant 15. Apeldoornse Courant
2. De Dordtenaar 16. Dagblad Flevoland
3. Haagsche Courant 17. Deventer Dagblad
4. Rivierenland 18. Gelders Dagblad
5. Rotterdams Dagblad 19. Nieuw Kamper Dagblad 6. Utrechts Nieuwsblad 20. Veluws Dagblad
7. Groene Hart 21. Zwolse Courant
8. Almere vandaag 22. De Twentsche Courant Tubantia
9. BN/Destem 23. Einhovens Dagblad
10. Brabants Dagblad 24. Leeuwarder Courant 11. Dagblad de Limburger 25. Limburgs Dagblad
12. Dagblad van het Noorden 26. Provinciale Zeeuwse Courant 13. De Gelderlander 27. Rotterdams Dagblad
14. De Gooi- en Eemlander 28. Leidsch Dagblad
2. Is the article published in print or online?
1. In the printed version of the newspaper, so offline. 2. On the newspapers website, so online.
3. Date on which the news article is published: dd.mm.yyyy
4. Title of the news article:
5a. Genre of the news article (Broersma & Graham, 2012): 1. News report 2. Reportage/feature 3. Background/news analysis 4. Liveblog 5. Other
5b. Topic of the news article (Broersma & Graham, 2013): 1. Politics and government
2. International relations 3. Social welfare
4. Business and economy 5. Accidents and disaster 6. Crime
7. Sports
8. Nature, environment, science or technology
9. Education 10. Healthcare
11. Religion and beliefs
12. Arts, culture or multimedia 13. Human interest
14. Lifestyle 15. Other
B: information about the tweet
6. Number of tweets reported in this news article: …. Tweets are reported in this news article.
7. Author of the tweet:
8. Occupation of the author of the tweet:
1. European parliament candidate/politician. 5. Organization linked to E.U. 2. Dutch national politician 6. Journalist
3. Political party, Dutch or E.U. based 7. Citizen 4. Prominent E.U. member that is not a
member of parliament
8. Other
9a. Is the author a (member of a) Dutch political party? 1. Yes answer question 9b
2. No go to question 10
9b. From which Dutch political party is the author of the tweet a member? (only when question 9a is answered with ‘1. Yes’):
1. CDA 2. PVV 3. PvdA 4. VVD 5. D66 6. GroenLinks 7. SP
8. Partij voor de Dieren 9. Christenunie/SGP 10. 50Plus
11. Piratenpartij
12. All other small Dutch parties
10. How is the tweet reported in the news article (Broersma & Graham, 2012)? 1. Quoted
11. Function of the tweet (Broersma & Graham, 2013):
1. Trigger news report (political candidate posted an offensive tweet)
2. Tweet is used as an illustration (a non-essential element of the news article) 3. Stand alone
4. Q&A: Twitter interview
12. Style of the tweet (Broersma & Graham, 2012): 1. Formal
2. Informal
13. Character of the tweet (Broersma & Graham, 2012): 1. Factual: tweet represents an opinion/argument
2. Conveying an acknowledgement: thanking, complementing, congratulating 3. Advise giving.
4. Directives: call for action. 5. Other
14. Tone of voice of the tweet: 1. Positive
2. Negative 3. Neutral
15a. Does the tweet display a specific emotion? 1. Yes answer question 15b
15b. Emotion displayed in the tweet (only when question 15a is answered with ‘1. Yes’): 1. Joy 2. Grief 3. Fear 4. Anger 5. Amazement 6. Disgust 7. Admiration 8. Jealousy 9. Hope 10. Reproach 11. Love 12. Pride 13. Regret 14. Hate 15. Shame 16. Guilt
Appendix II: Inter-coder reliability analysis
Table 4
Results of inter-coder reliability analysis per variable
Variable measures: Kippendorf’s Alpha
National/Regional newspaper 1
Name of national newspaper 1
Name of regional newspaper 0.923
Print/Online article 1
Genre of news article 0.926
Topic of news article 0.927
Number of tweets in news article 1
Occupation author tweet 1
Author member Dutch political party 1
Political party 1
How tweet is reported 1
Function of tweet 1
Style of tweet 1
Character of tweet 1
Tone of tweet 0.757
Specific emotion displayed in tweet 1
