Please (Don’t) Bug Me
News Framing and its Effects on Public Support on the Dutch Intelligence and
Security Services Act and Turnout Intention in the Referendum
Job Borggreve ID: 12026204 Master’s Thesis
Graduate School of Communication
Master’s programme Communication Science Supervisor: T. Dobber
Word count: 7476 28-6-2019
Abstract
Many studies have shown that framing affects the public support on the issue at stake and turnout intention in referendums. However, there is a gap in the scientific literature as research focusses on EU topics. This study presents a detailed experimental research on the effect of framing, specifically an opportunity and a risk frame, on the public support for the Dutch Security and Intelligence Services Act and the turnout intention in the referendum that was organized about this Act. The moderation effect of political knowledge and issue
importance is also included. Participants showed higher levels of support in the opportunity frame condition compared to participants in the risk frame condition. They did not show more turnout intention compared to participants in the risk frame condition. These findings were not significant. The moderation effect of political knowledge and issue importance was also not significant. Possible future research is discussed to improve internal validity by including a content analysis and to expand research to find a reversed mobilization effect.
Keywords
Public Support, Turnout Intention, Framing Effect, Political Knowledge, Issue Importance, Risk and Opportunity Framing, News Framing
Introduction
“It is bleak that you can be bugged. If you toy with that, the basis of society will crumble.” - Tijn de Vos (Fiola, 2017)
On the 9th of October 2017, the propose for the consultative referendum about the Intelligence
and Security Services Act in the Netherlands reached the threshold of 300,000 requests. Tijn de Vos and four other UvA students were the initiators. Worried that the new Act did not get any attention in the media, they agreed to increase public attention to make sure that
parliament reviews the Act. In their opinion the untargeted interception of online traffic with the new act is not necessary (Parool, 2018).
The Dutch Intelligence and Security Services Act referendum took place on the 21st of
March 2018, it resulted in a win for the No-vote: 49 per cent against 46 per cent (Kiesraad, 2018). With the Consultative Referendum Act from 2015 it was possible in the Netherlands to request a consultative referendum on adopted acts. Two proposes reached the threshold of 300,000 requests which resulted in the Ukraine–European Union Association Agreement referendum in 2016 and the Intelligence and Security Services Act referendum in 2018. As the Intelligence and Security Services Act referendum took place last year there is still a gap in the scientific literature for this referendum.
This is not the case for the European Constitution referendum that was held in the Netherlands. This referendum from 2005 resulted in a victory for the No-vote with a high turnout of 63%. According to Aarts and van der Kolk (2006) it caused that the gap between the elite and voters was fully exposed for the first time. The media attention was concentrated on the different interpretation of what the referendum was about, the big difference between the yes and no camp was constantly highlighted which caused that the Dutch electorate was exposed to two-sided information (Aarts & van der Kolk, 2006; Baden, 2010). Baden and de Vreese (2008) examined how the Dutch voters in the European Constitution referendum
construct meaning and on what information that is based. In their focus groups they found that despite low political knowledge and political interest, voters can still construct complex understandings about the issue at stake. They retained control over the information the media provided and selected, they weighted and framed the information in a way that it will fit their understanding. Voters with the same belief came to different evaluations, supporting the findings of Aarts and van der Kolk (2006) that the European Constitution referendum
revealed different interpretations at voters with the same beliefs (Baden & de Vreese, 2008). Jumping forward in time to the Intelligence and Security Services Act referendum, it seems that the Dutch electorate was again exposed to two-sided information. The main argument of the yes camp in the Intelligence and Security Services Act referendum was tracking down terrorists by giving the intelligence and security services more possibilities to do so. The privacy concerns about untargeted interception of internet and telephone traffic was the supporting argument of the no camp (Kokkeler, 2018). These privacy concerns may be a society-wide phenomenon in the Netherlands as Dobber, Trilling, Helberger and de Vreese (2018) found that education level is not a predictor of privacy concerns in examining the relation between attitude towards political behavioural targeting and privacy concerns.
Information about Intelligence and Security Services Act referendum becomes two-sided by using frames. Valenced frames affect the public support for the issue at stake as de Vreese and Boomgaarden (2003) found by examining frames in EU-related news. More specific, Schuck and de Vreese (2006) found that, compared to a Risk frame, an Opportunity frame causes more public support towards the EU enlargement. Making the step to
behavioural intention, framing the EU constitution in positive terms led to higher turnout intention compared to a frame with negative terms. However, it did cause a reversed
mobilization as the positive frame caused more risk perception and motivation to vote against the statement (Schuck & de Vreese, 2009). Combining these framing effects on public
support and turnout intention and to test if their applicable beyond the EU-context, this research question will be answered:
In what way does framing in news stories about the Dutch Intelligence and Security Services Act affect public support towards this Act and turnout intention in the referendum?
Understanding how framing affects support and turnout intention in this context, will help explaining how Dutch citizens construct meaning from these frames and what the behavioural intention will be. This can contribute to design and implement a more effective campaign strategy in our digital society. In targeting the Dutch electorate as potential voters it is evident to know what effect a certain frame has on support or behavioural intention. By including two moderators Political Knowledge and Issue Importance in this study, it is possible to be even more specific. Do potential voters that find the issue of high importance act differently when exposed to a frame? Is a potential voter with low political knowledge more affected by a frame? As this study examines the framing effects on the public support for an Act and the turnout intention in a referendum about that same Act, it provides a less discussed
combination of these variables in the current media effect research. Previous studies only examine separately public support or turnout intention in a EU context.
In the theoretical framework the function of referendums in a democracy and the typology of framing is discussed where after the hypotheses for the main effect and the framing effect moderators Political Knowledge and Issue Importance are included. The conceptual model is in Appendix A.
Theoretical background Referendums in a democracy
Referendums in a democracy are an instrument in political decision making and they remain a popular discussion topic. On the one side, citizens can be mobilized and may become more
politically engaged but on the other side it can show a public disinterest and ignorance in politics (de Vreese & Semetko, 2004). A referendum is not a replacement but rather an addition to party democracy, campaigning and reporting it can affect the public opinion and even be vital for a desired outcome. It is an unique chance for frustrated citizens to vote against their own party without dropping them directly as in election. Ideology also plays a different role as parties with very different ideologies may support the same side (de Vreese & Semetko, 2004).
Tierney (2012) states that clarity of the issue is central in referendums, but that there are concerns which can cause that an issue is unclear. First, the intelligibility. The question can be unclear due framing or grammar. Second, the formulation of the question causes that there are two questions running at the same time. Third, the consequences of a vote against the proposal of the agenda-setter are unclear: how will politicians react? Last, the nature of the issue can be too complex. Especially the latter two raise the question if a referendum is a suitable instrument in decision making (Tierney, 2012). For example, the consequence of a no-vote in the Dutch Ukraine–European Union Association Agreement referendum in 2016 was not clear (NRC, 2016), it led to a complex political situation (RTLnieuws, 2016) which eventually led to an attachment in the Association Agreement. However, this attachment was later criticized by the Dutch Council of State, questioning the legal significance (Raad van State, 2017).
Given that an informed citizen is part of a deliberative democracy, Tierney (2012) states that excluding the public in decision making does not work. A successful referendum is the outcome of deliberation and not a frustrating end point. This seems the opposite of the Intelligence and Security Services Act referendum where University of Amsterdam students launched a campaign to organize the referendum and eventually succeeded (Kiesraad, 2017). The question remains if people in a divided society are capable to reflect different sides of the
issue and to link it to a broader perspective. If they can, the outcome of the referendum will still cause that one side loses. In a polarized society the interest of this minority will be ignored after the referendum which can cause more tension between both sides (Tierney, 2012).
Typology of framing
Framing is ambiguous concept, there is a lack of consistency in the conceptualisation of framing in the current literature. Several scholars proposed a new conceptualisation of the different aspects of framing (Cacciatore, Scheufel & Iyengar 2016; de Vreese, 2005;
Scheufele, 2000; Vliegenthart & van Zoonen, 2011). Lecheler and de Vreese (2019) state that a news frame has a selective function by stressing out certain aspects so that implies a
suggestion for the receiver. To shape this, there is a process called ‘framing’. This process includes frame-building, the presence of frames and frame-setting (Lecheler & de Vreese, 2019). Studies often examine ‘framing’ as a cognitive media effect along with the theories of agenda-setting and priming or employ the same research design for the three different
approaches (Scheufele, 2000).
The process model of framing is discussed by several scholars (de Vreese, 2005; Lecheler & de Vreese, 2019). The model of framing first mentions frame-building, which refers to factors that can influence of news frames (de Vreese, 2005). Vliegenthart and van Zoonen (2011) add that frame-building is operationalized in content analyses and
observations in the newsroom. Frame-building leads to a news frame, a broad definition of a news frame is given in de Vreese (2005): “a central organizing idea or story line that provides meaning to an unfolding strip of events, weaving a connection among them. The frame suggests what the controversy is about, the essence of the issue.” (p. 53). He made the distinction between an issue-specific and a generic frame. Issue-specific frames are only on
specific topics, generics can be applicable on different topics which makes them suitable for generalization, comparison and theory building. Contrary to issue-specific frames where this is not possible (de Vreese, 2005).
Frame-setting refers to the interaction between media frames and individuals’ prior knowledge and predispositions according to de Vreese (2005). Frames can affect the interpretation and evaluation of issues. This part of the process model of framing is investigated extensively in the current literature to determine, like in this study, to which extent and under which circumstances a frame affects the opinion and attitude of the receiver (Lecheler & de Vreese, 2019). As in this study, in previous frame-setting research the news frame is manipulated in an experiment to examine the impact (Scheufele, 2000). The consequences of framing, framing effect, can be on individual or societal level which is a popular topic in empirical studies (de Vreese & Boomgaarden, 2003; Lecheler & de Vreese, 2011, 2012; Schuck & de Vreese, 2006).
Despite the process model of framing which helps clarifying the concept, there is still critique. Cacciatore et al. (2016) suggest that the label of ‘framing’ has to be abandoned in future research, the conceptualisation differs too much which causes that comparisons cannot be made anymore. They point out that the concept needs a refocus; different types of framing have to be distinguished. Lecheler and de Vreese (2019) share this frustration and state the focus of future framing research has to be on the process model of it.
There is consensus in the current literature that the process of framing has to be distinguished of the agenda-setting theory (McCombs & Shaw, 1972) as that deals with the salience of issues (de Vreese, 2005). Framing is focused on the presentation of these issues. The priming theory (Iyengar, Peters & Kinder, 1982) is another conceptual model that is said to overlap with framing according to Cacciatore et al. (2016). They state that the priming theory is based on a network of cognitive structures that are used in the retrieval and use of
information. The agenda-setting and priming theory rely on accessibility theory, where the media increases the salience of some issues to influence the standard by which the audience judges (Scheufele, 2000). Framing is rooted from the attribution theory in which the audience ability to detect patterns in the given information based on pre-existing cognitive schemas is central (Cacciatore et al., 2016). With subtle changes the audience might interpret the issue differently, not by making it more salient but by including schemas that influence the interpretation of the information (Scheufele, 2000).
In addition, there are two views on the process of framing (Cacciatore et al., 2016; Chong & Druckman, 2007; Lecheler & de Vreese, 2019). The psychology-rooted equivalence framing is a form where the presentation of information is manipulated (e.g., 12% single or 88% relationship) and from a sociological view emphasis framing is a form where the content of communication is manipulated. Frames from the last view are closer to real journalistic frames (Chong & Druckman, 2007) and therefore more compatible in politics and
communicative situations (Lecheler & de Vreese, 2019). The present study is an example of sociological emphasis framing, as it is central how the Dutch electorate constructs meaning on the Act and the referendum from the different content it receives.
Generic consequence frames: Opportunity and Risk
Hobolt (2009) found that evaluations of the public on a certain referendum campaign are influenced by the framing of the proposal and the reversion point. The reversion point is “the outcome to occur if the proposal offered by the agenda setter is defeated” (Hobolt, 2009: 7). The reversion point is a type of elite framing and it can be explained with the simple spatial model of vote choice. This theory is originally developed in the context of a two-party system, but it is also applicable on referenda with their binary choice option. The electorate has to choose between two alternatives: the proposal described on the ballot and the policy
entailed by a no-vote, the reversion point (Hobolt, 2009). Framing the proposal and the reversion point influences the citizens’ evaluations, this framing is considered as consequence framing where the emphasis lays on the positive or negative outcome. The opportunity and risk frame mentioned in Schuck and de Vreese (2006) are an example of the consequence frame (Hobolt, 2009).
Few studies have tested the effects of valenced frames on the individual level, de Vreese and Boomgaarden (2003) introduced it in political research. Valenced frames on political issues are “indicative of “good and bad” and carry positive and/or negative elements” (de Vreese & Boomgaarden, 2003: 363). Lecheler and de Vreese (2011) state this valence is one of the most fundamental characteristics in the political discourse, as elites try influence to support or reject the issue at stake by emphasizing the positive or negative part. De Vreese and Boomgaarden (2003) examined the effect of valenced frames on the support on the EU and the EU enlargement, the authors found that these frames can affect public opinion and support for the issue at stake with a positive or negative tone. Specifically, participants exposed to a disadvantageously framed news article showed less support for the EU and its enlargement compared to participants who were exposed to advantageously framed news article. When these frames appear in a competitive context as a campaign referendum, Hobolt (2009) adds that competing frames decrease the likeability that voters are undecided. So the opinion and support of the voters is affected by these frames and they are stirred in a direction to choose. Going from support to actual behaviour, Schuck and de Vreese (2009) found that framing the EU constitution in positive terms lead to a higher turnout compared to frame with negative terms. However, these positive terms caused that these voters voted against the constitution. The positive frame caused more risk perception and motivated the electorate to vote against and turning it into a reversed mobilization. Risk perception as a mobilizing effect can however not be generalized to other referendums because pre-existing policy preferences
and the meaning of regarding that the status quo will determine the perception of risk (Schuck & de Vreese, 2009).
Schuck and de Vreese (2006) set up an experiment without control group and found that participants who were exposed to the opportunity frame in a news article showed a higher level of support towards EU enlargement compared to participants who were exposed to the risk frame. Given that a news frame is more than an argument on a topic but a whole
construction of an issue which affects the public opinion, it is too simplistic to define the opportunity and risk frame as a positive and negative frame as they are altered to the study context (Schuck & de Vreese, 2006). In their study, the risk frame emphasizes the negative consequence of EU enlargement and it raises concerns whereas the opportunity frame
emphasizes the positive consequences and raises hope. The findings in Schuck and de Vreese (2006) are in line with the findings in de Vreese and Boomgaarden (2003) – a valenced news frame affects the public attitudes on the issue. Schuck and de Vreese (2006) even state that the risk and opportunity frame are not inherent to EU-related news, which makes them a generic frame in terms of de Vreese (2005) and generalizable which leads to these hypotheses:
H1a: Framing the Dutch Intelligence and Security Services Act as an ‘opportunity’ has
positive effects on public support towards the Act and on the turnout intention in the referendum.
H1b: Framing the Dutch Intelligence and Security Services Act as a ‘risk’ has negative effects
on public support towards the Act and on the turnout intention in the referendum.
Moderator: Political Knowledge
Including moderators in the design within news framing effects research allows a more accurate prediction on the public support and turnout intention, it determines how strong the
effect is and which direction the effect operates (Lecheler & de Vreese, 2011, 2019). These moderator variables can be divided in two groups – the first one represents the individual-level conditionalities of news framing effects, political knowledge is an example. These individual-level moderators are divided in three subgroups: trait-like (e.g. gender), malleable perceptions (e.g. political knowledge) and situational (e.g. emotions) (Lecheler & de Vreese, 2019). Lecheler and de Vreese (2019) state that the general idea is that individual-level moderators influence news framing effects a lot, because the process of framing is influenced by predispositions as political knowledge.
Political knowledge is one of the most important moderators for framing effects (de Vreese & Boomgaarden, 2003; Hobolt, 2009; Lecheler & de Vreese, 2011, 2012, 2019; Lecheler, de Vreese & Slothuus, 2009). However, the effect of political knowledge remains divided (de Vreese & Boomgaarden, 2003; de Vreese & Semetko, 2004; Schuck & de Vreese, 2006). Some studies like de Vreese and Semetko (2004) find that participants with more political knowledge are affected more by a news frame because they already have stored more information that can be framed. On the other side, Schuck and de Vreese (2006) found that participants with low levels of political knowledge were more affected by the news frames and more susceptible to risk framing. Further, Lecheler and de Vreese (2011) did not find a different framing effect immediately after the exposure for participants with low, moderate and high political knowledge.
Schuck and de Vreese (2006) found in their study on the support for EU enlargement that participants with more political knowledge responded similar in the opportunity and risk condition, less knowledgeable participants were affected by the risk condition. This study also looks into public support, so based on the categorization of news framing effects in Lecheler and de Vreese (2011, 2019) this hypothesis is formulated:
H2: Political knowledge moderates framing effects so that individuals with low levels of
political knowledge are more affected by the news frame than individuals with high political knowledge.
Moderator: Issue Importance
Besides the individual-level moderators there are also contextual moderators where the focus lays on how the context within a frame is received and changes the strength and direction of the effect. Therefore, including these kind of moderators is important, as they show if the finding in an experimental setting can be generalized (Lecheler & de Vreese, 2019). The contextual moderators are divided in three subgroups: situational (e.g. interpersonal
discussing), information environment (e.g. issue importance) and one group in between where competitive frames are an example (Lecheler & de Vreese, 2019).
Contextual information environment moderators contain conditionalities from outside the news frame that may influence how the frame is perceived. An example of this moderator is given in Lecheler et al. (2009). They found that for low-importance issues the news framing effect was stronger, for high-importance issues there were no differences found. When an issue is important, the important attitudes and pre-existing ideas of participants are stronger and more elaborate which makes them less susceptible to framing effects. It causes more engagement to require more knowledge about the issue. Individuals who do not find the issue important, will not have that same engagement level (Lecheler et al., 2009). Therefore, new information in a frame affects their opinion by giving new considerations instead of given some considerations with more relevance (Lecheler et al., 2009). This leads to the last hypothesis:
H3: Issue importance moderates framing effects so that individuals who think the Dutch
Intelligence and Security Services Act is not important are more affected by the news frame than individuals who think the Act is important.
Nadeu, Niemi and Amato (1995) examined the relationship between political knowledge and issue importance in the language issue in Quebec. They found that political knowledge and issue importance have a reciprocal relationship, where the importance-to-knowledge relationship is the strongest.
Method Sample
In the Netherlands one has to be at least 18 years old to vote. The Dutch Intelligence and
Security Services Act was central in this study and therefore Dutch citizens 18 years and older were part of the study. The link to the internet survey was send via Facebook, Twitter,
LinkedIn, WhatsApp and e-mail. Therefore, it was a convenience sample because it was available by virtue of its accessibility. This was not a random selection method, so it was a non-probability sample. This affected the external validity, which made it harder to generalize the findings to the whole Dutch electorate. 233 people agreed to participate in this study. Of these 233 participants, 33 had to be excluded because they indicated that they did not read the news article. Another 6 participants were excluded because they did not know the frame of the news article. 1 participant picked ‘Other’ as gender and was excluded. Last, 9 participants were excluded because it took them more than 60 minutes to complete the survey and 2 participants did not finish the survey. Analyses were thus conducted with a final sample of N = 182, which is 78.1% of all participants that filled in the survey. 95 participants were exposed to the Opportunity frame and 87 participants were exposed Risk frame. The average
age was 31 (SD = 11.93), most participants were 25 years old. Regarding age groups, 22-28 year olds were overrepresented with 58.5%. 54.9% of the participants was male and 45.1% was female. Most participants completed a University Master (24.7%), followed by a University Bachelor (20.9%), higher professional education (hbo) (20.3%) and upper secondary vocational education (mbo) (16.5%), high school (havo, vwo) (14.8%), lower vocational education (2.3%) and primary school (0.5%).
To check if the random assignment to both conditions was successful a randomization check was conducted. The randomization was successful; the conditions did not differ significantly from each other1. As in previous framing effect studies Political Interest (de Vreese &
Semetko, 2004; Schuck & de Vreese, 2009; Wettstein 2012), Political Self Placement (Wettstein, 2012) and Political Efficacy2 (de Vreese & Boomgaarden, 2003; de Vreese &
Semetko, 2004; Schuck & de Vreese, 2009, 2011) are included in this check because they influence the turnout intention.
Design
To test the possible effects of an Opportunity or Risk frame a survey-embedded experiment with a, post-test only, between-subject design with a random assignment to one of the two conditions was conducted. A two (Frame: Opportunity, Risk) factorial design was employed with the quasi experimental moderators Political Knowledge and Issue Importance. Within this design the first condition represented the Opportunity frame. The second condition represented the Risk frame. According to Schuck and de Vreese (2011), media effect research often employs a randomised post-test design to avoid pre-test sensitisation and detect between group differences. Therefore, this study had the same design. This design was an experiment because de Vreese and Boomgaarden (2003) state that in media effect research
experimentation is used to establish causality by ascribing differences in the findings to the manipulation; the internal validity increases.
Stimulus material
The experimental stimulus material consisted of one news article in two alternative frames
(see Appendix B). To increase external validity, the arguments that are put forward in the two articles are based on media content research for this study to ensure that the Opportunity frame and the Risk frame are reflected. Both articles are written together with a Dutch journalist to also increase the external validity, they were only produced for this study. To control for any predispositions about existing newspapers, a new newspaper Hollands
Dagblad was created. A graphic designer developed the webpage for this newspaper where
the article was displayed (see Appendix C). Powell, Boomgaarden, De Swert and de Vreese (2015) found in their study on visual framing effects that the combination of images and text, the framing effect of an image did not play a big role. However, it did had an influence on the participants’ behavioural intentions. As participants in this study filled in their turnout
intention, an image was not provided in the stimuli to gain full control over the experimental manipulation.
The red line in both articles was identical, to ensure that the Opportunity and Risk frame were displayed as in Schuck and de Vreese (2006) the different argumentation in both articles was kept to a minimum to control for other explanations. In total, there were six different parts within the stimulus that varied. The deductive approach for producing news frames in de Vreese (2005) consists four criteria that a frame must meet. The parts of the
Opportunity and Risk frame in this study had identifiable conceptual characteristics, they were
observed in journalistic practice, they were designable from each other and other frames and last, they were recognized by others (see Appendix B).
Measures
Dependent variable: Public Support.
Public Support towards the Intelligence and Security Services Act was measured through five
items (in translation) on a 7-point Likert scale from 1 = strongly disagree to 7 = strongly agree. These five items build a reliable index scale (M = 4.78, SD = 1.00, = .84)3. See
Appendix D for the operationalization.
Dependent variable: Turnout Intention.
The intention to go out and vote in the referendum was measured on a 5-point Likert scale from 1 = very unlikely to 5 = very likely (M = 3.98, SD = 1.17).
Moderating variable: Political Knowledge.
Knowledge about Dutch politics was measured with five items (in translation) with three answer options. These were binary coded as either 0 (wrong answer) or 1 (correct answer). Participants were defined as having ‘low political knowledge’ when they gave less than four correct answers and as having ‘high political knowledge’ when they gave four or five correct answers. This resulted in an answer scale ranging from 1 (low political knowledge) to 2 (high political knowledge), with α = .36 (M = 1.71, SD = 0.45). According to the factor analysis these items load on two factors.4 As these items were based on Lecheler and de Vreese (2011,
2012) and Political Knowledge is a moderator of framing effects in this study, the items were computed as one factor. See Appendix D for the operationalization.
Moderating variable: Issue Importance.
Participants were asked before the manipulation if they find The Intelligence and Security
Services Act important on a 5-point Likert scale from 1 = not at all important to 5 = very important. Participants that picked 1 or 2 were placed in (1) ‘Low Importance’. With 3 they were placed in (2) ‘Neutral Importance’ and with 4 or 5 they were placed in (3) ‘High Importance’ (M = 2.69, SD = 0.52).
Manipulation check.
A manipulation check revealed a successful manipulation. The frames were recognized by both groups which allows to explain the between-group differences by the experimental manipulation. Participants were asked what they thought was the frame of the news article with the answer scale (1) ‘Opportunities with the Act’, (2) ‘Risks with the Act’ and (3) ‘Don’t know’. 33 participants picked the ‘Don’t know’ option and were excluded. 28 of the 95 participants in the Opportunity frame condition picked the wrong ‘Risks with the Act’ option. 19 of the 87 participants in the Risk frame condition picked the wrong ‘Opportunities with the Act’ option. An Independent T-test revealed that participants who were exposed to the
Opportunity frame (M = 1.29, SD = 0.46) perceived this frame significantly more as
Opportunity compared to participants who were exposed to the Risk frame (M = 1.78, SD = 0.42), t(179.98) = - 7.52, p < .001, 95% CI[-0.62, -0.36].
Procedure
The experiment was conducted with a self-completion questionnaire in Qualtrics in May 2019. To prevent survey fatigue, participants were informed about the estimated time to fill in the survey and they were asked to agree with the consent statement. Participants first
completed a pre-test questionnaire asking for political interest, political efficacy, political knowledge and issue importance of the Intelligence and Services Act. Next, participants were randomly exposed to one of the two experimental conditions. Last, participants filled in a post-test questionnaire, which asked for their public support for the law and vote intention as of today. It also included an exposure and manipulation check. After completing the last check, the participants filled in their demographic variables after which they were debriefed (see Appendix G for the full survey). The survey received ethical approval in the Department of Communication Science at the University of Amsterdam. The incentive for the participants
was five times a € 10, - voucher of the webshop Bol.com, if participants wanted to make a chance to win it there was the possibility to fill in the email address at the end of the survey.
Results
Two Multi-Way ANOVA’s were conducted to answer the hypotheses.5 Public Support and
Turnout Intention have a positive significant correlation but it is not strong enough to conduct
a MANOVA, r = 0.15, p = .049. Two PROCESS models (Hayes, 2017) were also conducted but did not reveal any significant findings.6 Each hypothesis contains both dependent
variables and therefore each hypothesis is discussed twice. First for the results on Public
Support and then on Turnout Intention.
Effects on Public support for Condition
Hypotheses 1a and 1b predicting that participants in the ‘Opportunity’ condition will show significantly more support and participants in the ‘Risk’ condition will show significantly less support towards the Dutch Intelligence and Security Services Act are rejected. The main effect of Condition on Public Support is not significant, F(1,171) = 0.00, p = .958, 2 = .00.
Table 1 displays the mean difference in support for the Act between participants in the Opportunity condition (M = 4.95, SE = 0.19) and the Risk condition (M = 4.83, SE = 0.23). People who were exposed to the Opportunity frame showed more support towards the Dutch Intelligence and Security Services Act (Mdiff = 0.13) but this difference is not significant, p =
.673.
Effects on Turnout Intention for Condition
Hypotheses 1a and 1b, predicting that participants in the ‘Opportunity’ condition will show significantly more turnout intention and participants in the ‘Risk’ condition will show
significantly less turnout intention if the Dutch Intelligence and Security Services Act
referendum was held today, are rejected. The main effect of Condition on Turnout Intention is not significant, F(1,171) = 0.50, p = .479, 2 = .00. Table 1 displays the mean difference in Turnout Intention between participants in the Opportunity condition (M = 3.68, SE = 0.21)
and the Risk condition (M = 3.95, SE = 0.25). Contrary to the hypotheses, people who were exposed to the Opportunity frame showed less turnout intention towards the Dutch
Intelligence and Security Services Act (Mdiff = - 0.27) but this difference is not significant, p =
.416.
Table 1
Support for Dutch Intelligence and Security Services Act by Condition Opportunity condition (N = 95) Risk condition (N = 87) Support 4.95 (0.19) 4.83 (0.23) Turnout Intention 3.68 (0.21) 3.95 (0.25)
Note. Cell entries are mean scores of Public Support on a 7-point scale and mean scores of
Turnout Intention on a 5-point scale, standard error in parentheses. *p < .05
Moderator Political Knowledge
Hypothesis 2 predicting that participants with low levels of political knowledge are more affected by a news frame is rejected. An interaction model was tested and there is no significant interaction effect found between Political Knowledge and Condition on Public
for the Act between the participants. People with a high level of political knowledge and in the Opportunity condition showed more support compared to people with a high level of knowledge and in the Risk condition (Mdiff = 0.41), but this difference is not significant, p =
.161. People with a low level of political knowledge and in the Opportunity condition showed less support compared to people with a low level of knowledge and in the Risk condition (Mdiff = - 0.13), but this difference is not significant, p = .782.
Figure 1
Support for high and low politically knowledgeable participants in both conditions
Note. Bars show mean scores of Public Support of participants with low and high political
knowledge on a 7-point scale in the opportunity frame and risk frame conditions.
Hypothesis 2 predicting that participants with low levels of political knowledge are more affected by a news frame is rejected. An interaction model was tested and there is no significant interaction effect found between Political Knowledge and Condition on Turnout
Intention, F(1,171) = 0.01, p = .906, 2 = .00. Figure 2 displays the mean difference between
the participants in turnout intention for the referendum. People with a low level of political knowledge and in the Risk condition showed more turnout intention compared to people with
1 2 3 4 5 6 7
Opportunity Condition Risk Condition
Pu bli c S up po rt (m ea n sc ore )
also a low level of knowledge and in the Opportunity condition (Mdiff = 0.43), but this
difference is not significant, p = .408. People with a high level of political knowledge and in the Risk condition showed more turnout intention compared to people with also a high level of knowledge and in the Opportunity condition (Mdiff = 0.05), but this difference is not
significant, p = .888. Table 2 displays the mean scores of Public Support and Turnout
Intention for the subgroups (see Appendix E).
Figure 2
Turnout Intention for high and low politically knowledgeable participants in both conditions
Note. Bars show mean scores of Turnout Intention of participants with low and high political
knowledge on a 5-point scale in the opportunity frame and risk frame conditions.
Moderator Issue Importance
The size of the subgroups Low Importance in the Opportunity Condition (n = 4) and the Risk Condition (n = 1) is too small and these were therefore excluded. Hypothesis 3 predicting that participants that find the Dutch Intelligence and Security Services Act less important are more affected by a news frame is rejected. An interaction model was tested and there is no
significant interaction effect found between Issue Importance and Condition on Public 1 1,5 2 2,5 3 3,5 4 4,5 5
Opportunity Condition Risk Condition
Tu rn ou t In ten tio n
Support, F(2,171) = 0.38, p = .684, 2 = .00. Figure 3 displays the mean difference in support
for the Act between the participants. People who where neutral about the importance of the Act and exposed to the Risk frame showed less support compared to people who where also neutral and were exposed to the Opportunity frame (Mdiff = - 0.05), but this difference was not
significant, p = .867. People who find the Act of high importance and were exposed to the Risk frame showed less support compared to people who also find the Act of high importance and were exposed to the Opportunity frame (Mdiff = - 0.29), but this difference is not
significant, p = .156.
Figure 3
Support for participants which assigned neutral or high importance in both conditions
Note. Bars show mean scores of Public Support of participants that assigned a neutral or high
importance to the Act on a 7-point scale in the opportunity frame and risk frame conditions.
Hypothesis 3 predicting that participants that find the Dutch Intelligence and Security Services Act less important are more affected by a news frame is rejected. An interaction model was tested and there is no significant interaction effect found between Issue
1 2 3 4 5 6 7
Opportunity Condition Risk Condition
Pu bli c S up po rt (m ea n sc ore )
Importance and Condition on Turnout Intention, F(2,171) = 1.37, p = .258, 2 = .00. Figure 4
displays the mean difference in turnout intention for the referendum between the participants. People who were neutral about the importance of the Act and exposed to the Risk frame showed less turnout intention compared to people who where also neutral and were exposed to the Opportunity frame (Mdiff = - 0.39), but this difference was not significant, p = .279.
People who find the Act of high importance and were exposed to the Risk frame showed less turnout intention compared to people who also find the Act of high importance and were exposed to the Opportunity frame (Mdiff = - .28), but this difference is not significant, p =
.226. Table 3 displays the mean scores of Public Support and Turnout Intention for the subgroups (see Appendix F).
Figure 4
Turnout Intention for participants which assigned neutral or high importance in the conditions
Note. Bars show mean scores of Turnout Intention of participants that assigned a neutral or
high importance to the Act on a 5-point scale in the opportunity frame and risk frame conditions. 1 1,5 2 2,5 3 3,5 4 4,5 5
Opportunity Condition Risk Condition
Tu rn ou t In ten tio n
Discussion & Conclusion
Based on previous research on framing effects on public support (Schuck & de Vreese, 2006) and turnout intention (Schuck & de Vreese, 2009), it was expected that an opportunity frame in a news article causes more support towards the Act and a higher turnout in the referendum. This was not confirmed in the present study. The opportunity frame caused more support for the Act compared to the Risk frame. This difference is not significant, contrary to the findings in Schuck and de Vreese (2006). Hobolt (2009) states that competing frames
decrease the likeability that voters are undecided as voters are stirred in a direction to choose. The findings in this study show a high turnout intention for both frames, which seems in line with the findings in Hobolt (2009). Contrary to the findings in Schuck and de Vreese (2009), participants who were exposed to the Risk frame have more turnout intention compared to participants who were exposed to the Opportunity frame but this difference was not
significant. The effect of a positive frame causing more risk perception and therefore a higher turnout intention was not found in this study (Schuck & de Vreese, 2009).
Political knowledge is one of the most important moderators for framing effects but at the same time the perceived effect of it is divided in previous studies (Lecheler & de Vreese, 2019). Lecheler and de Vreese (2011) found no differences in framing effects for participants with low, moderate or high political knowledge, the findings in the present study are line with this as no significant differences were found. Participants with low political knowledge did not show more support and turnout intention compared to participants with high political knowledge after being exposed to the opportunity frame. They did show more public support and turnout intention compared to participants with high political knowledge after being exposed to the risk frame in line with Schuck and de Vreese (2006), but this difference is not significant.
Lecheler et al. (2009) found that for low-importance issues the framing effects were stronger. When an individual finds the issue of high importance, then there are important attitudes and pre-existing ideas which makes them less susceptible to framing effects. Contrary to this study, the findings in the present study suggest that the framing effect is stronger for participants that thought the Act was of high importance compared to participants who were neutral about the importance of the Act. However, this difference is not significant.
There are certain caveats in the design of this study that have to be acknowledged. The operationalization of Political Knowledge is based on Lecheler and de Vreese (2011, 2012). The items in these studies covered the EU and Dutch politics. As the present study deals with a Dutch act and referendum, the EU items were adjusted to Dutch politics creating five items only measuring the Dutch politics knowledge level. Lecheler and de Vreese (2011) divided participants in low, moderate and high political knowledge. In the present study, there were not enough participants to create a ‘low political knowledge’ group that gave 0-1 correct answers. This led to a distinction were participants with less than 4 correct answer had low political knowledge and with more correct answers high political knowledge. This still gave overrepresentation of high knowledgeable participants (n = 131) compared to low
knowledgeable participants (n = 52). For this same reason, it was not possible include
participants that thought the Act was of low importance (n = 5). Thus, they were excluded and it became harder to reflect the findings with Lecheler et al. (2009) as the subgroup ‘low importance’ misses. Measuring Turnout Intention in this present study is complex as the referendum already took place in 2018, participants know that the turnout question is about a referendum that is not real. The Act is not in the news anymore which causes that participants are less aware of it. Both can lead to a high turnout intention that was found in this study. There is no referendum fatigue because the referendum is not in the news and it is also possible that participants gave a higher turnout intention because it is socially desirable to
have the intention to vote. That mentioned, it is difficult to change the pre-existing ideas of participants with only one framed article. The opportunity and risk frame were both very present in the articles, which could lead to more suspiciousness among the participants. The arguments that were put forward in both articles were not the result of a content analysis but rather a content research on several websites of newspapers and magazines. In the present study it was not measured what the opinion about the Act was, besides the importance of it. It is possible to suggest that pre-existing ideas of participants were that strong that the framing effect weakened.
This study investigated if framing a news article about the Dutch Intelligence and Security Services Act as an opportunity or a risk affects public support towards this Act and the turnout intention in the referendum. Examining the framing effect of these generic frames on a non EU topic places this study in the frame-setting in the process model of framing as it is investigated to which extent and under which circumstances the frame affects the opinion and intentions of the receiver (de Vreese, 2005; Lecheler & de Vreese, 2019). Understanding how these frames affect the public support and turnout intention can help campaigns targeting their potential voters. If a potential voter finds an issue more important, then it is interesting to know which frame is then suitable to steer that voter in the right direction. If a voter has less political knowledge, how will they then react on a certain frame? This study examined framing in a less discussed combination of Public Support and Turnout Intention beyond the EU context. The frames caused a non-significant effect in the hypothesized direction on public support and a stronger effect for low political knowledgeable participants in the risk condition. This concludes that for this Act and its referendum the support and the turnout intention of Dutch electorate is not significantly affected by an opportunity or risk frame. However, the results do have implications. With more participants it can be possible to find a significant difference in effects between the opportunity and risk frame and to investigate the
moderation effect of the low importance issue and low political knowledge. To increase the internal validity, the stimulus needs to be based on content analysis creating a
multi-methodological study as Schuck and de Vreese (2006). A reversed mobilization effect (Schuck & de Vreese, 2009) can also be part of future research by asking participants what they will vote in the referendum if they decided to turn out. The limitations and insignificant findings notwithstanding, this study gives a first research brick in the Dutch Intelligence and Security Services Act (referendum) where future framing effect research can build on.
Notes
1. Randomisation check, the conditions did not differ significantly from each other in terms of Gender, X2(1, N = 182) = 3.00, p = .084, Age, F(1,180) = 0.43, p = .515, 2 =
.00, Education Level, F(1,180) = .36, p = .548, 2 = .00, Political Interest, F(1,180) =
0.92, p = .340, 2 = .00, Political Self Placement, F(1,180) = 1.41, p = .237, 2 = .01,
Political System Efficacy, F(1,180) = 0.38, p = .537, 2 = .00, Political Personal
Efficacy, F(1,180) = 0.62, p = .431, 2 = .00.
2. For the six items measuring Political Efficacy a Principal Axis Factoring (PAF) is conducted with Oblimin Rotation. There are enough participants, KMO = .65. The items also correlate sufficiently according to Bartlett’s Test of Sphericity, p < .001. Kaiser-criterion is used to determine the factors with Eigenvalues > 1. According to this criterion and the scree plot there are two components. Together, these factors explain 56.8% of the variance in the original items. After a direct Oblimin Rotation the question about the influence on the policy of the government has the strongest association with Factor 1 (factor loading = .75) and the question about politics being too complex has the strongest association with Factor 2 (factor loading = .72). Looking at the pattern matrix, Factor 1 can be labelled as Political System Efficacy
with Cronbach’s = .70. Factor 2 can be labelled as Political Personal Efficacy with Cronbach’s = .44. Thus, Political System Efficacy has a reliable scale and Political
Personal Efficacy does not have a reliable scale.
3. For the five items measuring Public Support a Principal Axis Factoring (PAF) is conducted with Oblimin Rotation. There are enough participants, KMO = .74. The items also correlate sufficiently according to Bartlett’s Test of Sphericity, p < .001. Kaiser-criterion is used to determine the factors with Eigenvalues > 1. According to this criterion and the scree plot there is one component. This factor explains 61% of the variance in the original items. The statement that the Dutch Intelligence and
Security Services Act makes the Netherlanders safer has the strongest association with the Factor (factor loading = .81). Looking at the factor matrix, the factor can be
labelled as Public Support with Cronbach’s = .84. Thus, Public Support has a reliable scale.
4. For the five items measuring Political Knowledge a Principal Axis Factoring (PAF) is conducted with Oblimin Rotation. There are enough participants, KMO = .56. The items also correlate sufficiently according to Bartlett’s Test of Sphericity, p < .001. Kaiser-criterion is used to determine the factors with Eigenvalues > 1. According to this criterion and the scree plot there are two components. Together, these factors explain 52.6% of the variance in the original items. As these items are based on Lecheler and de Vreese (2011, 2012) and Political Knowledge is a moderator of framing effects in this study the items are computed into one variable, Cronbach’s = .56.
5. The first Multi-way ANOVA is conducted with Public Support as the dependent variable. The size of the largest and smallest subgroup differs more than 10%, but Levene’s Test is not significant, F (10,171) = 1.19, p = .300. So equal variances can
be assumed. The second Multi-way ANOVA is conducted with Turnout Intention as the dependent variable. The size of the largest and smallest subgroup differs more than 10% and Levene’s Test is significant, F (10,172) = 1.94, p = .042. So equal variances cannot be assumed, which has to be taken in mind while interpreting the results. 6. The PROCESS model (number 2) with Public Support as dependent variable revealed
no significant main effect of Conditions (b = 0.14, 95% CI [-1.69,1.97], t = 0.15, p = .882), no significant interaction effect of Political Knowledge (b = -0.09, 95% CI [-0.74,0.56], t = -0.27, p = .788) and no significant interaction effect of Issue
Importance (b = -0.08, 95% CI [-0.66,0.49], t = -0.29, p = .776). The PROCESS
model (number 2) with Turnout Intention as dependent variable revealed no
significant main effect of Conditions (b = -1.07, 95% CI [-3.17,1.02], t = -1.01, p = .313), no significant interaction effect of Political Knowledge (b = 0.34, 95% CI [-0.40,1.09], t = 0.98, p = .365) and no significant interaction effect of Issue Importance (b = 0.12, 95% CI [-0.53,0.78], t = -0.37, p = .710).
References
Aarts, K., & Van Der Kolk, H. (2006). Understanding the Dutch “no”: The euro, the east, and the elite. PS - Political Science and Politics, 39(2), 243–246.
https://doi.org/10.1017/S1049096506060434
Baden, C. (2010). Contextualizing frames in political discourse: using semantic network analysis to investigate political parties’ framing strategies in the Dutch EU referendum campaign. Conference Papers: International Communication Association: Annual
Meeting, 1–42.
Baden, C., & De Vreese, C. H. (2008). Making sense: a reconstruction of people’s understandings of the European constitutional referendum in the Netherlands.
Communications, 33(2), 117–145. https://doi.org/10.1515/COMMUN.2008.008
Cacciatore, M. A., Scheufele, D. A., & Iyengar, S. (2016). The End of Framing as we Know it … and the Future of Media Effects. Mass Communication and Society, 19(1), 7–23. https://doi.org/10.1080/15205436.2015.1068811
Chong, D., & Druckman, J. N. (2007). Framing theory. Annual Review of Political Science, 10, 103– 126. https://doi.org/10.1146/annurev.polisci.10.072805.103054
de Vreese, C. H. (2005). News framing: Theory and typology. Information Design Journal +
de Vreese, C. H., & Boomgaarden, H. (2003). Valenced news frames and public support for the EU. The European Journal of Communication Research, 28(4), 361–381.
https://doi.org/https://doi.org/10.1515/comm.2003.024
de Vreese, C. H. & Semetko, H. (2004). Political campaigning in referendums: framing the
referendum issue (224). Routledge.
Dobber, T., Trilling, D., Helberger, N., & de Vreese, C. H. (2018). Spiraling downward: The reciprocal relation between attitude toward political behavioral targeting and privacy concerns. New Media and Society, 1–20. https://doi.org/10.1177/1461444818813372 Fiola (2017, October 17). Zo sleepten vijf UvA-studenten het referendum binnen. Retrieved
from:https://www.folia.nl/actueel/115383/zo-sleepten-vijf-uva-studenten-het-referendum-binnen
Hayes, A.F. (2017). Introduction to Mediation, Moderation, and Conditional Process
Analysis. Guilford Publications, New York.
Hobolt, S. B. (2009). Framing Effects in Referendums: Experimental Evidence. In Europe in
Question, 110–131. https://doi.org/10.1093/acprof:oso/9780199549948.003.0005
Iyengar, S., Peters, M.D., Kinder, D.R. (1982). Experimental Demonstrations of the “Not-So Minimal” Consequences of Television News Programs. The American Political
Science Review 76(4), 848-858.
Kiesraad (2017, November 1). Referendum over Wiv gaat door. Retrieved from:
https://www.kiesraad.nl/actueel/nieuws/2017/11/01/referendum-over-wiv-gaat-door Kiesraad (2018, March 29). Uitslag over referendum Wiv: meerderheid tegen. Retrieved
from:
https://www.kiesraad.nl/actueel/nieuws/2018/03/29/uitslag-referendum-over-wiv-meerderheid-tegen
Kokkeler, B. (2018). Sleepwet sleept achter de feiten aan. BN DeStem, p. 1.
Lecheler, S., & De Vreese, C. H. (2011). Getting Real: The Duration of Framing Effects.
Journal of Communication, 61(5), 959–983.
https://doi.org/10.1111/j.1460-2466.2011.01580.x
Lecheler, S., & De Vreese, C. H. (2012). News Framing and Public Opinion: A Mediation Analysis of Framing Effects on Political Attitudes. Journalism & Mass Communication
Quarterly, 89(2), 185–204. https://doi.org/10.1177/1077699011430064
Lecheler, S., & De Vreese, C. H. (2019). News framing effects (1st ed.). Routledge.
Lecheler, S., De Vreese, C. H., & Slothuus, R. (2009). Issue Importance as a Moderator of Framing Effects. Communication Research, 36(3), 400–425.
https://doi.org/10.1177/0093650209333028
McCombs, M. E., & Shaw, D. L. (1972). The agenda-setting function of mass media. Public
Opinion Quarterly, 36(2), 176-187.
Nadeau, R., Niemi, R.G. & Amato, T. (1995). Emotions, Issue Importance and Political Learning. American Journal of Political Science 39(3), 558-574.
NRC (2016, March 30). Oekraïne-referendum: de belangrijkste vragen en antwoorden. Retrieved from:
https://www.nrc.nl/nieuws/2016/03/30/oekraine-referendum-de-belangrijkste-vragen-en-a-1606002-a226615
Parool (2018, March 21). UvA-studenten: ‘Wij hebben het referendum al gewonnen’. Retrieved from: https://www.parool.nl/nieuws/uva-studenten-wij-hebben het-referendum-al-gewonnen~b67cdc643/
Powell, T. E., Boomgaarden, H. G., De Swert, K., & de Vreese, C. H. (2015). A Clearer Picture: The Contribution of Visuals and Text to Framing Effects. Journal of
Raad van State (2017, January 18). Voorstel van wet tot regeling van de inwerkingtreding van de wet van 8 juli 2015, houdende goedkeuring van de op 27 juni 2014 te Brussel tot stand gekomen Associatieovereenkomst tussen de Europese Unie en de Europese Gemeenschap voor Atoomenergie en haar lidstaten, enerzijds, en Oekraïne, anderzijds (Stb. 2015, 315), met memorie van toelichting. Retrieved from:
https://www.raadvanstate.nl/adviezen/@64301/w02-16-0426ii/#highlight=referendum RTLnieuws (2016, November 4). Reconstructie Oekraïne-referendum: Hoe de nee-stem een
ja wordt. Retrieved from:
https://www.rtlnieuws.nl/nederland/politiek/artikel/346556/reconstructie-oekraine-referendum-hoe-de-nee-stem-een-ja-wordt
Scheufele, D. A. (2000). Agenda Setting, Priming and Framing Revisited. Mass
Communication & Society, 5436(August 2014), 37–41.
https://doi.org/10.1207/S15327825MCS0323
Schuck, A. R. T., & De Vreese, C. H. (2006). Between Risk and Opportunity News Framing and its Effects on Public Support for EU Enlargement. European Journal of
Communication, 21(1), 5–32. https://doi.org/10.1177/0267323106060987
Schuck, A. R. T., & de Vreese, C. H. (2009). Reversed Mobilization in Referendum Campaigns. The International Journal of Press/Politics, 14(1), 40–66.
https://doi.org/10.1177/1940161208326926
Schuck, A. & de Vreese, C.H. (2011). Public Support for Referendums: The Role of the Media. West European Politics 34(2). 181-207.
Tierney, S. (2012). Framing the Substantive Issue in Constitutional Referendums. In
Constitutional Referendums (226–259).
https://doi.org/10.1093/acprof:oso/9780199592791.003.0008
Vliegenthart, R., & van Zoonen, L. (2011). Power to the frame: Bringing sociology back to frame analysis. European Journal of Communication, 26(2), 101–115.
https://doi.org/10.1177/0267323111404838
Wettstein, M. (2012). Frame Adoption in Referendum Campaigns. American Behavioral
Appendix A Conceptual Model
Appendix B
Stimulus Material Opportunity Frame (Risk Frame Italicized and in Parentheses)
Het Wiv (Sleepwet)-referendum: wat heeft het ons opgeleverd?
Op woensdag 21 maart 2018 ging Nederland naar de stembus voor de
Gemeenteraadsverkiezingen. Ook werd er gestemd over de Wet op inlichtingen- en
veiligheidsdiensten (Wiv) (, ook wel ‘Sleepwet’ genoemd). Waar staat deze wet ook alweer voor? En heeft de uitslag van het referendum - een meerderheid stemde tegen - iets
veranderd?
De Algemene Inlichtingen- en Veiligheidsdienst (AIVD) houdt zaken in de gaten die de Nederlandse staat kunnen schaden. Voor de Wiv maakte de AIVD gebruik van een wet die stamde uit 2002. Bestuurskundige Sjoerd Fransen stelt dat deze wet aan modernisering toe is, aangezien in 2002 de smartphone nog maar net bestond en er sindsdien veel meer berichten via diverse platforms worden verstuurd. “Door deze nieuwe wet kan de AIVD op hetzelfde kwaliteitsniveau komen als de buitenlandse diensten, we gaan met onze tijd mee”.
(Bestuurskundige Sjoerd Fransen stelt dat deze wet nog niet aan modernisering toe is, aangezien de AIVD al computers en telefoons mocht hacken met de oude wetten. Met de Sleepwet mag de AIVD ook hacken via de computer van een derde persoon die bijvoorbeeld toevallig dezelfde provider heeft.) De Wiv zorgt ervoor dat bulkinterceptie mogelijk wordt,
waarbij dataverkeer in een bulk van de kabel wordt opgepikt. “Deze interceptie is
voornamelijk op het buitenland gerichte communicatie, (“Het gebruik van de kwetsbaarheden
in apparaten door de overheid kan echter ook nieuwe kwetsbaarheden veroorzaken, waardoor het voor hackers weer makkelijker wordt) de privacy van Nederlandse burgers
wordt dus nauwelijks meer geschonden” (geschonden), aldus Sjoerd Fransen.
“Het beeld dat er met de invoering van de Wiv een complete wijk kan worden afgetapt, klopt dan ook niet (kan kloppen)”, aldus privacy-consultant en onderzoeker Michelle Prins.
Volgens haar is dit technisch gezien onmogelijk omdat er veel verschillende providers voor mobiel en internet actief zijn in een wijk (wordt er veel data van onschuldige burgers
verzameld en opgeslagen). Wat wel mogelijk is met de Wiv, is meer democratische controle
op de onderwerpen waarmee de diensten zich bezighouden. Dat komt doordat de premier en de ministers van Justitie, Binnenlandse Zaken en Defensie straks samen beleid maken (is het
delen van niet-geanalyseerde gegevens met buitenlandse inlichtingendiensten. Daarmee geeft Nederland privacygevoelige informatie van burgers aan het buitenland).
De duidelijke ‘nee’ in het referendum heeft ten slotte ook geleid tot een aantal aanpassingen aan de Wiv, zo geeft de AIVD aan. Ze moeten specifieker te werk gaan, gegevens mogen maar drie jaar worden bewaard en pas gedeeld worden met buitenlandse diensten nadat de AIVD een check heeft gedaan. De Wiv biedt de AIVD hierom nieuwe mogelijkheden om Nederland veilig te houden. (De data wordt na controle bijvoorbeeld gedeeld met
buitenlandse diensten. Dit zijn echter waarschijnlijk de Britten en de Amerikanen en beiden hebben een rijke geschiedenis als het gaat om datalekken bij de overheid. De Wiv leidt dus eerder tot zorgen om de privacy van burgers dan meer veiligheid.)
Appendix B (continued) Different Parts in the Stimulus
There were six different parts in the news article that varied according to the Opportunity and Risk frame. These parts cover the frame dimensions mentioned in Schuck and de Vreese (2006). The first part consists the Risk framing of the Act as the ‘Sleepwet’ in the title of the article and the first paragraph, without giving a rational justification. ‘Sleepwet’ is the nickname the no-camp gave to the Act, with reference to the dragging (‘slepen’) process of data. Thus, this part covers the frame dimension: (1) emotional pro/contra.
The second part referred to modernization of the current Act, which is necessary because the current Act is from 2002 so a new Act is needed to get to same level as foreign Intelligence Services (Opportunity frame) or is not necessary because hacking with the current Act was already possible and with the new Act it becomes possible to hack via the computer of a third person with the same provider (Risk frame). These arguments are put forward by public administrator Sjoerd Fransen, so this part covers the frame dimensions: (2) positive/negative quote and rational pro/contra.
The third part is also given in the second paragraph and given in a quote of public
administrator Sjoerd Fransen. It covers the bulk interception of data with the new Act, this interception is most of time only used on traffic to other countries which means no privacy violation for Dutch citizens (Opportunity frame) or this interception makes use vulnerabilities of devices which causes new vulnerabilities which means that the privacy of Dutch citizens is violated (Risk frame). Hence, this part covers the frame dimensions: (3) positive/negative quote and rational pro/contra.
The fourth part evaluates if it is possible to tap a whole neighbourhood with the new Act. Quoting privacy consultant and researcher Michelle Prins who says that this is technically not possible because of the different providers for internet and mobile subscriptions (Opportunity
frame) or it is possible which means that data of innocent citizens is collected and stored (Risk frame). This part covers the frame dimensions: (4) positive/negative quote and rational pro/contra.
The fifth part covers the changes of the new Act compared to the old Act, that is more democratic control as the Prime Minister and the Ministers of Justice, Home Affairs and Defence have to make policy together (Opportunity frame) or sharing non analysed data with foreign Intelligence Services, which means giving away privacy sensitive information of Dutch citizens (Risk frame). This part covers the frame dimension: (5) future benefits/costs. The last part judges the adjustments in the Act after the referendum. The Dutch Intelligence Services AIVD has to work more precisely, data can be stored only for three years and it can only be shared with foreign Intelligence Services after a check. So the new Act will give the AIVD new opportunities to keep the Netherlands safe (Opportunity frame). The data will be shared with foreign Intelligence Services after a check, but most of the time these will be the British of the Americans and both of them have a great history in data breaches. So the new Act leads to more concerns about the privacy of Dutch citizens than more safety (Risk frame). This part covers the last frame dimensions: (6) negative/positive evaluation.
Appendix C Layout Stimuli
Opportunity frame
Appendix D
Operationalization variables
Public Support: Five items based on the items that measure public support on the EU integration in Schuck and de Vreese (2006) and on EU enlargement (α = .75) in de Vreese and Boomgaarden (2003): (1) ‘The Intelligence and Security Services Act is a good thing’, (2) ‘The Intelligence and Security Services Act makes the Netherlands safer’, (3) ‘The Intelligence and Security Services Act violates the privacy’ (reverse coding), (4) ‘The Netherlands will profit from The Intelligence and Security Services Act’, (5) ‘The
Intelligence and Security Services Act will bring more advantages than disadvantages’ (M = 4.77, SD = 1.00, = .84).
Political Knowledge: Five items based on Lecheler and de Vreese (2011, 2012) and adjusted to Dutch politics: (1) ‘Which party holds the most seats in the Tweede Kamer?’, (2) ‘André Rouvoet belongs to which party?’, (3) ‘Who is the current president of the Tweede Kamer?’, (4) ‘How many seats are there in the Eerste Kamer?’, (5) ‘Emile Roemer belongs to which party?’ These were binary coded as either 0 (wrong answer) or 1 (correct answer).
Participants were defined as having ‘low political knowledge’ when they gave less than four correct answers and as having ‘high political knowledge’ when they gave four or five correct answers. Ranging from 1 (low political knowledge) to 2 (high political knowledge): (M = 1.72, SD = 0.45, = .36).
Appendix E Table Hypothesis 2
Table 2
Support and Turnout Intention for Dutch Intelligence and Security Services Act by Condition and Political Knowledge
Support Turnout Intention
Opportunity condition Low Political Knowledge (N = 25)
4.77 (0.28)
3.65 (0.32) Opportunity condition High Political Knowledge
(N = 70)
5.13 (0.26)
3.70 (0.29) Risk condition Low Political Knowledge
(N = 27)
4.90 (0.36)
4.10 (0.41) Risk condition High Political Knowledge
(N = 60)
4.72 (0.14)
3.75 (0.16)
Note. Cell entries are mean scores of Public Support on a 7-point Likert scale and mean
Appendix F Table Hypothesis 3
Table 3
Support and Turnout Intention for Dutch Intelligence and Security Services Act by Condition and Political Knowledge
Support Turnout Intention
Opportunity condition Neutral (N = 22)
4.62 (0.23)
3.79 (0.26) Opportunity condition High Importance
(N = 69)
4.89 (0.14)
4.24 (0.16) Risk condition Neutral
(N = 25)
4.56 (0.22)
3.40 (0.24) Risk condition High Importance
(N = 61)
4.61 (0.14)
3.97 (0.16)
Note. Cell entries are mean scores of Public Support on a 7-point Likert scale and mean
