Microtarget acquired:
admired or undesired?
An experimental study into the possible difference in effect
between a presumed microtargeted and a non-microtargeted
message on the Dutch public’s trust in politics
Name: Alisja Winkel Student ID: 10774734
Supervisor: Dhr. T. Dobber MSc Date: 28/06/2019
Master thesis
Political Communication
Abstract
Political microtargeting is growing in popularity all across Europe. Also in the Netherlands, microtargeting is becoming unmissable in the campaigns of many political parties. However, scholars worry about the effects this could have on the functioning of democracy. This study aims to fill a void in the scientific research pertaining to these worries, by looking at the influence microtargeting may have on political trust. Specifically, at the difference in effect between a presumed microtargeted and non-microtargeted message, on political trust.
Moreover, this research also looked at the possible influence the ‘type’ of party may have on the effect of microtargeting on political trust. An online experiment with 2 x 2 between-subjects design was conducted (n =146). The results showed no significant difference in effect on political trust between a microtargeted and non-microtargeted message, also not when looking at different types of political parties. These findings suggest microtargeting does not pose an immediate threat to the political trusts of citizens. Furthermore, this study gives some suggestions for future research, as the phenomenon of microtargeting keeps evolving.
Introduction
On march 29th 2019, Facebook launched its political Ad Library for Dutch political ads. This was done in accordance with demands from the European Union, in order to create more transparency about political advertisements on social media (European Commission, 2018). Now, the public has insight into which political party posts what advertisements. However, an important aspect remains a secret to the public, as the Facebook ad library does not report why or which certain groups of people see certain ads (Wassens, 2019). However, the fact that this information is not included in the Ad Library, does not mean there is nothing to report. On the contrary: the past three national elections, starting with the House of
Representatives election in 2017, has seen a steady rise in the use of personalized,
microtargeted messages in Dutch political campaigns. (Zuiderveen Borgesius et al., 2019). In fact, the growing popularity of this campaign tool has also been visible in several other European countries (Bodó, Helberger & de Vreese, 2017).
Microtargeting can be defined as “A strategic process intended to influence voters
through the direct transmission of stimuli, which are formed based on the preferences and characteristics of an individual” (Papakyriakopoulos et al., 2018) and can be grouped under
‘political behavioral targeting’ (Dobber et al., 2017).
Microtargeting is becoming unmissable in political campaigns, which is why the question arises; what kind of effect does this have on the functioning of democracy? (Boerman & Kruikemeier, 2016). Political advertising in general can have implications on citizens’ political knowledge and political trust (Motta & Folwer, 2016). Now that a new prominent political advertising tool has come onto the playing field, it is beneficial for political campaigners as well as the government, to know how this may affect important political attitudes.
Especially in a European context, this question is left largely unanswered, as many studies regarding microtargeting and its effects, are conducted in an American context (Barocas, 2012; Motta & Fowler, 2016; Murray & Scime, 2010; Schipper & Woo, 2019). Results of such studies are not entirely applicable to the Netherlands, as it has a different electoral system than the US and adheres to the European privacy laws (Dobber et al., 2017). Moreover, a general observation is that there is a ‘greater distrust of the intrusive political marketing and campaigning techniques’ in European countries, compared to the US (Bennett, 2016). This is why it is important there be European research into the effects of
microtargeting, especially because microtargeting does have some ‘intrusive’ characteristics (Barocas, 2012).
For example, politicians use people’s public Facebook information to target specific groups of people with a certain message. In general, many Facebook users are not aware that their information can be used like this without their explicit consent (Zuiderveen Borgesius et al., 2019). It is characteristics such as these, which can give citizens a negative feeling about the use of political microtargeting. These negative feelings can eventually influence the public debate, larger democratic system and political trust. (Zuiderveen Borgesius et al., 2019).
This study will therefor focus on political trust and a possible difference in effect on political trust between a non-microtargeted political message and a microtargeted political message.
Moreover, because this study aims to fill a void in European research about the effects of microtargeting, it will look at the additional possible of effect of ‘type of party’. As many European countries have multiparty systems, it is of interest to see is some parties are less appreciated for their use of microtargeting, than others. Especially for political campaigners it can be very beneficial to know if the image a party has, can be an indication of whether or not using a microtargeted message can lead to successful results for their political party.
In order to create insight into this, the following research question was composed: RQ:
What is the difference in effect between a presumed microtargeted and a non-microtargeted message on the Dutch public’s trust in politics? And does this effect differ for the type of party that uses it? Please refer to Appendix A for a conceptual model.
Theory
Microtargeting
In essence, microtargeting works as follows: political campaigns use publically available data such as demographical information, voter registration and zip codes, and combine this with information about voters’ lifestyle, to create specific voter categories that they can target with personalized advertisements (Gorton, 2016). The lifestyle data is not publically available, but it can be bought from third party data firms. These companies track and monitor online behavior. With this information, data firms create specific categories of people. Political campaigns can use these categories to gain insight into the preferences and interests of these specific categories, in order to cater a specially tailored political
advertisement to them (Gorton, 2016).
In the Netherlands, microtargeting has less room to work than in the US, which has several reasons. For one, the Netherlands is part of the European Union, which has a stricter privacy law that prevents political parties from being able to create voter profiles such as the ones in the US (Dobber., et al, 2018). Secondly, the Netherlands has a very different electoral system from the US, in which voters do not ‘register’ to vote; they simply show up on the day of the election and cast their vote. This makes microtargeting more challenging, as the voter registration list is a very valuable asset to US political campaigns (Dobber et al.,2017).
However, this does not mean that microtargeting in the Netherlands has a lesser chance to be successful than in the US. Dutch political campaigns have access to election
results on voting booth level and they can combine this information with demographical data from CBS (Central Bureau Statistics), which can allow them to look into specific voter
categories as well (Dobber et al.,2017). Additionally; all Dutch political parties have access to Facebook, on which political parties achieve a lot when it comes to personalized
advertisements (Boerman & Kruikemeier, 2016). Through Facebook, political parties can target voters based on interests they think match their current voter base or they can target’
lookalike audiences’, by uploading the phone numbers and e-mail addresses of members of
their party to Facebook. Facebook will then automatically generate a list of people who have similar online behavior and interests as current members (Zuiderveen Borgesius et al., 2019). There is also the opportunity for ‘Dark posts’ through Facebook, which political campaigns can use to send a personalized message to a certain group of people, but which will not be shown on the Facebook page of the party itself. This can be useful when targeting a specific issue which may be controversial to other voters, who the party does not want to lose, or when targeting a specific region (Dobber et al., 2017)
The pros and cons of microtargeting
The use of personalized advertisements can be quite lucrative for many political parties: it can be done at a fairly low price and can work very effectively for parties of all sizes (Gorton, 2016; Zuiderveen Borgesius et al., 2019). Microtargeting is also a good way to reach potential voters that are difficult to reach through traditional media channels. Through microtargeting, parties can ensure they reach voters with a message that will be meaningful to them, thereby aiming to persuade the public to cast their vote based upon their personalized message (Zuiderveen Borgesius et al., 2019).
However, there are also potential downsides to the use of microtargeting (Kruikemeier, Boerman & Segzin, 2015). First and foremost, in order to make these
information presents a threat to the privacy of voters (Zuiderveen Borgesius et al., 2019). Furthermore, this act of collecting voter data online and using it without voters’ consent for political means, may have a ‘chilling effect’ on voters. The knowledge that political
campaigns use their information unwantedly, may lead voters to refrain from visiting certain websites or engaging in political activities all together (Barocas, 2012). According to a Dutch study done in November of 2016, 83% of respondents thought using personal data for
political microtargeting was unacceptable (Zuiderveen Borgesius et al., 2019). Next to privacy concerns, there are several other negative consequences that microtargeting may have on the functioning of democracy. For instance, because
microtargeting enables political campaigns to identify potential voters, it automatically also creates groups of voters who are not fit to be targeted. If these groups of voters receive no form of personalized advertisements that are intended to motivate them to vote and other groups of people do, these voters who were already deemed ‘not worth microtargeting’ might become even more uninteresting for political campaigns during the next election and so on (Bacoras, 2012). This can then lead to a large portion of eligible voters getting systematically ignored over time. Often times, these ignored voters already belonged to marginalized groups within the democratic process (Bacoras, 2012; Zuiderveen Borgesius., et al, 2019)
Moreover, microtargeted messages usually focus on one specific issue. This message will then be targeted towards people who -according to data- feel strongly about this issue. If voters are only shown these ‘one-issue’ type advertisements, they might get a wrong idea of what voting for a particular party might mean in the grand scheme of things. This can then lead to a wrongful simplification of the political debate. (Bacoras, 2012; Zuiderveen
Borgesius et al., 2019). Because it is also difficult to pinpoint which voter saw what add about which topic that made them decide to vote for the party, there is a chance this voter will be
disappointed to find out that topic might not be the most important issue for the party once they are elected (Hillygus & Shields, 2014).
Political trust
Political trust can be quite a difficult concept to define, as it entails many components (Marien & Hooghe, 2011). Albrow & Easton (1966) gives the following definition: “Political
trust can be seen as a form of diffuse support a political system receives from its
environment”. Furthermore, Levi & Stoker (2000) state that is important to note trust is
relational as it involves a person making him/herself vulnerable to another person or party that has the capacity to do him/her harm. Trust is also not unconditional; a person can be more trusting of some aspects within a person or party, than others. It is because of this conditional trust citizens have in politics, that makes microtargeting a very plausible influencer of
political trust. For example, microtargeting can lead to lower support for politicians, lower engagement in political behavior and even skepticism towards the party that uses it (Turow et al., 2012; Boerman & Kruikemeier, 2016) which can all then, influence the amount of trust in politics citizens have.
Over the past couple years, scholars have expressed worry (Zuiderveen Borgesius., et al, 2019) about the potential negative effects microtargeting may have. Because political trust is an active and varying attitude, it may very well be influenced by new developments within the field of politics (Braun & Hitter, 2016), such as microtargeting. In order to fill a gap in scientific knowledge when it comes to the effects of microtargeting on political trust, the first hypothesis will be as follows; H1: A microtargeted message will have a more negative
influence on the trust in politics, than a non-microtargeted message Type of party
Boerman & Kruikemeier (2016) found that sponsored messages by political parties in comparison to commercial brands, lead to “reduced online behavioral intention, increased
skepticism and negatively affected source trustworthiness”. Participants regarded the brands’ sponsored messages as simply advertisements, but they did not appreciate this same technique by political parties. Furthermore, the research also looked at a potential difference between two different political senders, but found no difference in effect.
However contrastingly, research by Kreiss (2016) states that the first political party to integrate and use a new technique such as sponsored messages and microtargeting, can benefit from it tremendously. This benefit minimizes once other parties start using it, but by being the first to adapt, they are in the position to grow their knowledge and expand before other parties can follow (Kreiss, 2016). This was evident in the United States for example where, during the 2004 presidential elections, the Bush campaign used microtargeting for the first time and won the election (Bennett, 2016).
As illustrated in the US, it can be of great use to be the first party to adapt to a new technology such as microtargeting. However, the democrats were quick to follow in their footsteps because if they did not, they would risk falling irrevocably behind on their competitor and losing the next elections (Kreiss, 2016).
In a multi-party system such as the Netherlands, the loss of an election is more nuanced, as political parties in the Netherlands need to work together in order to have a majority in the House of Representatives (Mair, 2008). Perhaps this is also why the emergence of microtargeting in the Netherlands was more gradual with some clear ‘early’ adapters and ‘late comers’, as political parties could choose whether they wanted to use microtargeting; it was not an immediate necessity. (Spierings & Jacobsen, 2019). Early adapting parties in the Netherlands may have therefore enjoyed the same benefits as the republicans in 2004. One of these potential benefits, might be the fact that they are seen as more equipped to use microtargeting than other parties, because they have more experience and expertise (Kreiss, 2016).
Furthermore, because of the benefits that are assumed to be associated with early adapting parties and because microtargeting happens on social media, this study expects there will be a difference in the ‘type’ of party that uses microtargeting. More specifically, this study expects that a ‘younger’ party that was an early adapter to microtargeting and digitally advanced with good knowledge of social media, will evoke less resistance (Knowles & Linn, 2004) in the receiver of a personalized add towards the sending party, than an ‘older’ party that cannot be considered an early adapter to microtargeting. Moreover, because of the lessened resistance, this study expects that a younger party that uses microtargeting, will evoke a less negative effect on political trust, than an older party that uses microtargeting.
In order to test this assumption, this study will look at two different types of political parties. One party can be considered a ‘young, early adapting’ party and the other an ‘older, less early adapting’ party.
To define an ‘old’ and a ‘young’ party, two archetypes of political parties have been chosen. For the ‘old’ party, this study will focus on the second longest excising political party in the Netherlands, which is considered a Mass party (Bennett, 2015). It was founded in 1945 and has been through a lot of ups and downs in the recent years (Pennings & Keman, 2008). Especially since the late nineties and early two-thousands, Mass parties have had problems competing with new parties challenging the status quo of the Dutch democracy, with party supporters dwindling due to a decline in loyalty voters and low and inactive members (Pennings & Keman, 2008). Instead of actively mobilizing, members are more inclined to show support for the party by hanging a party poster in their window (Wolinetz, 2002).
The last few years, Dutch Mass parties have been incorporating digital campaigning. First on their own websites and later on social media as well (Dobber et al., 2017). Still, these online spaces are mostly frequented by younger people who are often more drawn to a direct
form of communication which Mass parties have not shown as successful in as other parties, such as post materialist political parties (Gibson & McAllister 2011).
To define a ‘young’ political party this study will thus focus on characteristics of the post materialist parties. According to Gibson & McAllister (2011), the focus on technological innovation and progress is common and most fitting for post materialist parties. The ideology of these types of parties are built upon the importance of valuing approachability. So, it comes naturally to post materialist parties to engage in social media activity as they enjoy being interactive with their voters, especially because their voters are often digital natives (Spierings & Jacobs, 2019). Additionally, this tech-savvy following means that post materialist parties are surrounded by social media expertise, be that in their staff or volunteers.
This allows them to stay ahead when it comes to technological developments, even if they do not have the financial means (Spierings & Jacobs, 2019).
This study chose to focus on two political parties that significantly differ on grounds of origin and digital nativity (Bennett, 2015), in order to see if this type of differentiation between parties makes a difference on attitude towards microtargeting and subsequently the trust people have in politics. And more specifically, if a young political party evokes a less negative effect on political trust then an old political party, due to their early adaptiveness and post materialist social media nativity. H2: A microtargeted message by a ‘young’ party, has a
less negative effect on political trust, than a microtargeted message by an ‘old’ political party
Persuasion knowledge & Third Person effect
The way consumers cope with advertising messages, depends on whether the
consumers recognize the persuasive nature of the message. The Persuasion Knowledge Model (Friestad & Wright, 1994) explains how persuasion knowledge is developed through life, as a person gains more and more experience with persuasion attempts, such as through
advertisements and commercials. Persuasion knowledge is expected to be well-developed once a person arrives at adulthood (Friestad & Wright, 1994). However, with the new types of social media advertisements that have emerged over the years, consumers have to adapt to new forms of advertisements such as ‘promoted’ or ‘sponsored’ messages on Facebook. Therefore, consumers might have a high amount of traditional persuasion knowledge, but may not be familiar with these news forms of advertisements online because they are different and less easy to recognize (Boerman & Kruikemeier, 2016).
Promoted or sponsored messages on Facebook can generally be distinguished by a mention of the word ‘sponsored’ in the top left corner, right under the name of the organization that is the sender of the advertisement. This cue is put there in order to help the receiver of the message recognize they are in fact looking at an advertisement and seeing it is expected to activate persuasion knowledge (Boerman., et al 2012). Microtargeted Facebook posts, also have these same clues that sponsored message have, which is why this research expects seeing a
microtargeted message will activate persuasion knowledge. This is then expected to lead to less political trust, the reasons for which are twofold.
As mentioned above, recognizing political sponsored messages, can lead to a reduced intention to interact with the content online (share, like, etc.), increase in skepticism and a negative effect on the perception of trustworthiness of the source (Boerman & Kruikemeier, 2016). Therefore, this study assumes that when persuasion knowledge is activated, negative attitudes towards the sender might be triggered which leads to a decline in political trust.
Furthermore, even if the persuasion knowledge leads to understanding the persuasive nature of the sponsored message and does not constitute negative attitudes, there is still a possibility this might lead to a decline in political trust. This can be explained by the Third Person effect. This theory is based upon the idea that, when it comes to mass communication, people generally feel that this has more influence on other people than on themselves
(Davison, 1983). This could be applicable in the current situation: if a person has a high persuasion knowledge, they might not feel mislead or skeptical because they are familiar with- and understand the purpose of the persuasive message. However, the Third Person effect might be activated, as the person can still see it as a misleading political technique, because others might be more susceptible towards this type of persuasive message than they themselves are. This will be tested with the following hypothesis; H3: The negative effect of a
microtargeted message on the trust in politics will be stronger for people with a high persuasion knowledge, than people with little persuasion knowledge
Methods
Sample and design
This study was conducted with a 2 (with microtargeting vs. without microtargeting) x 2 (young party vs. old party) between subject design. Both variables measure between-groups, as they compare participants in the ‘with micro-targeting’ condition for example, to see if there is a difference in the effect of microtargeting when it comes to the ‘type’ of party.
The convenience sample (Emerson, 2015) for this study existed of 174 respondents, of which a 147 (84.48%) completed the questionnaire (67.1 % female, mean age 32). All
respondents were eligible to vote in the Netherlands and were recruited through Facebook, WhatsApp and E-mail. In total, 70% of respondents indicated their education level was ‘HBO or WO Bachelor’ or higher. The language of the survey and stimuli were both Dutch.
Respondents were randomly assigned to one of four conditions (young party with microtargeting (n= 38), old party with microtargeting (n= 36), young party without
microtargeting (n= 34) and old party without microtargeting (n=38). After being exposed to one of four stimuli, respondents filled out several questions pertaining to the central concepts. Afterwards, respondents were thanked and debriefed.
Ethical approval was granted by the University of Amsterdam in April 2019. Data for this study was collected from may 9th through may 22nd 2019. Respondents took an average amount of 11 minutes to complete the questionnaire.
Stimulus material
The survey started with four demographical questions (Please refer to Appendix B for an overview of the survey questions). After this, respondents were introduced to a case about Robin (35), who is very passionate about preventing workplace discrimination and shares this passion through Facebook, where Robin posts, likes and shares messages pertaining to the prevention of workplace discrimination. Robin is not particularly politically engaged.
In the case about Robin, no he/she pronouns were used, in order not to create bias through gender. After the short introduction to Robin, respondents were informed they were about the see a Facebook message that came across Robin’s timeline.
The stimulus material consisted of four Facebook posts, that were made to be identical in regards to lay out, but were different on several substantive key points (Please refer to Appendix C for an overview of the stimulus material). The posts differentiated in the sense that two posts were made with a microtargeted message and two without. The microtargeted message read “We fight against discrimination in the workplace. You too? Join ~insert
political party name~ now!”. The stimuli without microtargeting read “We fight for an equal society. You too? Join ~insert political party name~ now!”. The posts with microtargeting
had a label on the upper left side that said ‘Sponsored’. Post without the microtargeted message did not have this label.
Furthermore, there were differences in the posts regarding the type of party. Two were made in the name of the young-, two for the old political party. The only thing that
differentiated the posts were the party name and picture of the political leader. For the young party, GroenLinks was chosen.
GroenLinks is a left-leaning party with a focus on green issues, such as climate change and can be categorized as a post materialist party (Spierings & Jacobsen). It was founded in 1990 and has been gaining popularity over the last few years. During the 2010 national elections, GroenLinks was one of three parties that were active on social media, making them ‘a head of the pack’ compared to other Dutch political parties (Spierings & Jacobsen, 2019). Campaign managers for GroenLinks have stated that, one of the main reasons they were able to be present and active on social media so quickly, was because they had tech-savvy
volunteers that could help them build websites, apps and were willing to function as ‘social media teams’, that could spread the party’s messages by sharing and retweeting
(Dobber et al., 2017; Spierings & Jacobsen, 2019). During the 2017 elections, GroenLinks won 10 seats, which made their total 14 (“Officiële uitslag Tweede
Kamerverkiezingen”, 2017). The party was quite successful in using microtargeting to their best advantage. For example, through the network of their campaign leader, the party was able to build a special canvassing app, that allows volunteers to gather more specific
information about the people they speak with. It is technical initiatives such as these, which have helped GroenLinks stick out as a digitally innovative party.
For the old party the Dutch labor party, the PvdA, was chosen. The party was founded in 1946 and has historically been a Mass Party in the Netherlands, but over the last year the party has lost a myriad of members and voters, which has led them to have a less favorable image (Pennings & Keman, 2008). This is evident from the biggest loss in the party’s history during the 2017 national elections (“Officiële uitslag Tweede Kamerverkiezingen”, 2017).
The PvdA is also active on social media, but was not until 2012, which is two years after GroenLinks (Spierings & Jacobsen, 2019). Still they are less successful on social media, evident from the differences in the parties’ social media following. 1
The research by Dobber et al. (2017) also conducted interviews with two campaign leaders from the PvdA. They mentioned the PvdA also uses Facebook microtargeting and tracking website visits in order to gain insight in what potential voters find important.
However, they do not have an app such as the one GroenLinks used and did not invest in any third-party consultancy, because they did not have the means to do so.
Because of these differences, the two parties have cultivated different images from the publics perspective. However, they are both on the left side of political spectrum and often agree with each other and share many similar ideologies. So much so, that in May 2019, there have been news articles suggesting the two parties should merge into one big national party.
Smaller, regional parties have already done this in their municipalities (“60 keer hetzelfde standpunt”, 2019). Because of this, potential bias under respondents due to political ideology is expected to be limited, as the parties are similar in this sense.
Measurements
Persuasion knowledge can be defined as “Lay people’s common knowledge about how
persuasion works and aimed to initiate research looking at target consumers’ knowledge about persuasion” (Ham, Nelson & Das, 2015). Persuasion knowledge will be measured
across all four conditions, but in the microtargeted condition the ‘sponsored’ label will be included, which is expected to trigger the participants’ persuasion knowledge (Boerman & Kruikemeier, 2016). The condition treatment was shown right above the questions aimed to trigger persuasion knowledge, to make sure respondents would pay close attention to whether the post actually was sponsored or not, as this can sometimes go unnoticed (Kruikemeier, Sezgin & Boerman, 2016). The variable will be measured with five items, which were based in full on the research by Kruikemeier, Sezgin & Boerman, 2016 (Please refer to Appendix B). All items were measured on a 7 - point Likert scale ranging from 1 = strongly disagree, 7 = strongly agree.
Five questions were asked to measure awareness of microtargeting, in order to be able to potentially measure the impact this might have on political trust. However, the items were not included in the hypothesis, nor as a controlling variable. The items will therefore not be discussed in the main text of this study.2
Furthermore, to measure the potential Third Person effect, two questions were used, based on several studies. This was done by asking the same question twice in order to measure Third Person effect, namely; once about oneself, once about ‘others’ (Pan, Meng & Zhou, 2012; Jang & Kim, 2018). This was adapted in the current study in the following way;
1. After seeing the Facebook message, Robin will be inclined to become a member of the political party, 2. If I were Robin, after the Facebook message I would feel inclined to become a member of the political party.
Lastly, political trust was measured based upon a questionnaire by Craig, Niemi and Silver (1990) and was used in full. Items included: You can generally trust the people who
run our government to do what is right and Most government officials try to serve the public interest, even if it is against their personal. There was a total of 9 items, which are included in
Appendix B. All items were measured on a 7 - point Likert scale ranging from 1 = strongly disagree, 7 = strongly agree.
Manipulation checks
In order to check if the designed manipulations were successful, four manipulation checks were done. To ensure respondents remembered which party they saw a Facebook post from, the following question was asked: From which party was the advertisement you just
saw? 1. CDA, 2. GroenLinks 3. D66, 4. PvdA, 5. VVD. As to not create bias, several other
parties from different ideological sides are also included. According to a Chi Square test, the manipulation was effective (X2 (4) = 130.54, p= <.001). 98.6% of respondents who saw a
GroenLinks post remembered the party correctly and 90.5% in the PvdA condition remembered the party correctly. Thus, this manipulation was successful.
To see if respondents would agree with the classification of the young and old party, two statements were given which respondents could answer on a 7 -point Likert scale ranging from 1 = strongly disagree, 7 = strongly agree. 1. I would describe the political party I just
saw a Facebook post from, as a young and innovative political party, 2. I would describe the political party I just saw a Facebook post from, as a digitally progressive political party.
Both manipulations turned out to be effective, when looking at two independent t-tests. On the first question, respondents in the GroenLinks conditions scored significantly
(t (144) = 8.93, p <.001) higher on the scale from 1 to 7 (M = 4.75, SD = 1.44) than the PvdA conditions (M = 2.82, SD = 115). The same can be said of the second question, which also showed a significant difference (t (144) = 4.52, p <.001) between GroenLinks (M = 4.43, SD = 1.28) and PvdA (M = 3.45, SD = 1.37). So, respondents seem to effectively agree with the categorizations of GroenLinks as young and innovative and PvdA as less young and innovative.
Lastly, in order to check whether respondents recognized a personalized message in their treatment condition, the following statement was posed: 1. The Facebook message I just
saw, contains a personalized message which could speak to Robin. Respondents could answer
on a 7- point Likert scale ranging from 1 = strongly disagree, to 7 = strongly agree. An independent t-test proved significant t (144) = 5.23, p <0.01. The microtargeted condition (M= 5.49, SD = 1.21) scored significantly higher on this scale than the non-microtargeted condition (M = 4.10, SD = 1.93), which means the personalized undertone of the
Results
Randomization checks
In order to ensure respondents were not significantly different across the conditions, several randomization checks were done on gender (χ2 (6) = 6.82, p= .34) and education level (χ2 (15) = 9.33, p= .86). As p is >.05, it can be concluded that gender and education did not differ significantly per condition. Lastly age was also tested F (41,105) = 1.75, p = .06. as Levene’s test proved insignificant, equal variances across conditions could be assumed.3
Scale constructions
Before any tests could be run, some new scales had to be created, starting with a scale for ‘Trust in politics’. First, questions two, five and eight had to be recoded, since they were posed negatively compared to the other questions. The factor analysis showed two
components with an eigenvalue higher than one (3.34, 37.06 % explained variance and 1,26, 14.02 % explained variance, respectively). Six items loaded on the first component, three on the other. From the reliability analysis, only the first component proved reliable consisting of items 1, 2recoded, 3, 4, 6 and 7 (= .82). Thus, the scale was composed (M= 4.26, SD= 1.02)
The same process was repeated for ‘Persuasion knowledge’. Moreover, to create the scale for persuasion knowledge, all items needed to be combined for questions 1, 2, 3, 4 and 5, as they were measured separately in all conditions. A factor analysis was done and the five items (> .45) all loaded on one component with an Eigenvalue higher than one (2.74, 54.89% explained variance). The scale proved reliable (=.78). Persuasion knowledge under
respondents was relatively high (M= 5.82, SD= 1.17) on a scale from 1 to 7.
Hypotheses
The first hypothesis, H1: A microtargeted message will have a more negative
influence on the trust in politics, than a non-microtargeted message, was tested through an
Before this however, the two microtargeted groups were coded together (n = 75) and the two non-microtargeted groups were coded together (n = 72) in order two have to groups for the independent samples t-test.
Levene’s test indicated an insignificant result, so equal variances could be assumed. Furthermore, the t-test proved marginally significant, however the effect was not in the predicted direction. Seeing a microtargeted message (M = 4.30, SD = .98), did not lead to a lower on the trust in politics scale than a non-microtargeted message (M = 4.11, SD = 1.04). with t (145) = 1.69, p = .09.
When looking to see if there were any potential effects when checking for each parties’ conditions specifically, results were similar. The difference between the GroenLinks microtargeted condition (M = 4.38, SD = 1.03) and the non-microtargeted condition
(M = 4.06, SD = 1.15) showed no significance t (70) = 1.23, p = .23. The same goes for the PvdA microtargeted condition (M = 4.40, SD = .92) and the PvdA non-microtargeted condition (M = 4.15, SD = .95) which also did not show a significant effect t (73) 1.17, p = .25. H1 is therefore not accepted.
The second hypothesis, H2: A microtargeted message by a ‘young’ party, has a less
negative effect on political trust, than a microtargeted message by an ‘old’ political party,
was also tested through an independent t-test. As Levene’s indicates an insignificant; equal variances may be assumed. Looking further at the t-test results, no significant effect of a microtargeted message by a young (n = 38) political party (M= 4.38, SD = 1.05) versus old (n = 36) political party (M = 4.41, SD = .92) could be found on trust in politics t (73) = -.09,
p=.93. This means that the second hypothesis also could not be accepted.
Finally, the third hypothesis, the negative effect of a microtargeted message on the
trust in politics will be stronger for people with a high persuasion knowledge, than people with little persuasion knowledge, was tested through a two-way ANCOVA.
Before conducting this test however, the newly made mean scale for persuasion knowledge needed to be adjusted so that participants who scored high in the microtargeted conditions and participants who scored low in the non-microtargeted conditions, would be grouped into one category with a ‘high’ persuasion knowledge. The same had to be done for participants in the microtargeted condition who scored low and participants in the non-microtargeted condition who scored high, in order to have the category ‘low’ persuasion knowledge. This was done by recoding the five items which were used to measure persuasion knowledge for the two conditions that saw the not microtargeted message, from 1 to 7, 2 to 6, 3 to 5, etc. Moreover, for the two microtargeted conditions, the items remained the same for persuasion knowledge. With the new composition of the scale, two categories were made, namely; high and low persuasion knowledge. The cut off point for these categories was the median of 4.9.
After this was done, the two-way ANCOVA was conducted, with ‘political trust’ as dependent variable, the categories for persuasion knowledge and conditions as independent variables and the two questions asked to measure a Third Person effect as covariates.
The two-way ANCOVA did not find a significant effect on the interaction between persuasion knowledge and condition, when controlling for the Third Person effect F (3,137) = 0.56, p = 0.64. Furthermore, no significant results were found for the main effects, as neither persuasion knowledge F (1,137) = .04, p = .84, nor the condition a respondent was in F (3,137) = .96 p = .41, was able to show a trend in the predicted direction. The covariates however, did both show a significant result. The first question, ‘After seeing the Facebook message, Robin will be inclined to become a member of the political party’, F (1,137) = 7.42,
p = .01 and the question ‘If I were Robin, after the Facebook message I would feel inclined to
become a member of the political party’ F (1,137) = 5.80, p = .02 showed a significant effect on the scale for trust in politics. However, H3 could still not be accepted.
Conclusion and Discussion
The goal of this study was to determine a possible difference in effect between a presumed microtargeted and a non-microtargeted message on the Dutch public’s trust in politics and if this effect differs for the type of party that uses it.
Similar studies to the current one have been conducted over the past years, that also look at the effect of microtargeting on the psychology of voter behavior, such as Kruikemeier, Sezgin & Boerman (2016) who looked at the influence of microtargeting on political engagement and Boerman & Kruikemeier (2016), who looked at the public’s response on promoted political tweets. The current study however, has elaborated on these previous studies, by basing the hypotheses on this previous work and identifying a large, underlying consequence of microtargeting; a possible decline in political trust by the public. By doing so, this study appears to be one of the first to dive into this, aiming to contribute to the general knowledge about microtargeting. Moreover, this study was innovative in the microtargeting field, as it looked at microtargeting in combination with a possible Third Person effect and by categorizing Dutch political parties into an ‘old’ and ‘young’ political party.
Three hypotheses were composed in order to find a possible difference in effect between a microtargeted message and a non-microtargeted message on political trust. Based upon the evasive characteristics that come with microtargeting, this study expected that seeing a microtargeted message would lead to less trust in politics. This was tested in H1, but no significant result was found in this proposed direction. So, contrary to the expectations, seeing a presumed microtargeted message did not lead to a negative effect on political trust.
A possible explanation for this could be that in this experiment respondents were not actually microtargeted themselves. Respondents were introduced to Robin and asked
questions in regards to how Robin would feel about seeing the message they saw. Then later on, respondents were asked to answer questions in order to measure their personal political
trust. Perhaps it is because the microtargeted message did not affect respondents as if it was their own Facebook information that was used to target them, this did not negatively influence their political trust. Future research might focus on finding a way to personally microtarget respondents individually, in order to measure whether this may cause any chilling effects, for instance, which can in turn influence their political trust.
With H2, this study intended to explore the effect a ‘type’ of party could have on the negative effect of microtargeting on trust in politics. Again, no significant result could be found, meaning that H2 was not accepted either: the type of party that sends the microtargeted message, does not have an effect on the trust in politics. Possible explanations for this could be that only two political parties were used in this experiment. Specifically, two left leaning political parties. This was done in order to ensure a limited effect of respondents’ personal political bias influencing the results. However, perhaps political bias plays an important role in whether a microtargeted message has a negative effect on political trust. This study was successful in categorizing two different types of political parties, a revelation that seems to be new in the field of research on Dutch politics. By doing so the study has contributed to the notion that voters might categorize parties in certain ‘types’. Even though the current categorization did not proof to be of significance in regards to microtargeting and trust in politics, the evidence provided in this research might be interesting for future research to elaborate on. A way this could be done is by including a wider range of political parties as ‘young’ and ‘old’ types of parties into the study, such as ‘christen democrats’ or ‘right winged’ parties. This way, the possible categorization of ‘old’ and ‘young’ parties might be expanded and with that making it possible to capture a potential effect of personal political bias on microtargeting.
Finally, this study looked at the possible effect of persuasion knowledge on the influence a microtargeted message has on trust in politics in H3. More specifically, based on the Third
Person effect, this study proposed that the negative effect of microtargeting on trust in politics, would be stronger for people with high persuasion knowledge, as they themselves would be able to understand they were being microtargeted, but expected other would be more susceptible to the influence of the microtargeted message and therefore would not notice. H3 could not be accepted, but interestingly the covariates ‘Third Person effect’ did show a significant effect on political trust.
A possible explanation for the lack of significant effects of persuasion knowledge on political trust, could be that persuasion knowledge in the non-microtargeted condition was very low. Even though the non-microtargeted condition did not have a sponsored label, many respondents still answered the five items to measure persuasion knowledge, as if the post had been sponsored. An explanation for this, could be that persuasion knowledge is assumed to be established at adulthood (Friedstad & Wright, 1994). However, with new advertising formats being introduced, through Facebook for example, persuasion knowledge needs to be adjusted, as a person needs to learn to recognize the new form and become familiar with it (Boerman, Willemsen & Van Der Aa, 2017). Sponsored posts and microtargeting can be seen as new advertising formats, which many people are not familiar with yet and thus do not have persuasion knowledge over just yet. Even though research by Kruikemeier, Sezgin & Boerman (2016) found that explaining the persuasive nature of a Facebook message before being exposed to a sponsored add did not significantly affect persuasion knowledge.
However, their study consisted of a convenience sample of university students, who arguably may not have needed this explanation as they are more digitally native then older generations. For future research, this study would advise to include a short explanation on sponsored Facebook posts.
The study has some limitations, such as the sample, which was a convenience sample. Using a convenience sample, can lead to biases, as it is unsure why respondents agreed to
participate for instance (Emerson, 2015). Moreover, convenience samples are often not fit for generalizing results. Combined with the relatively small size of the sample (n = 146), the high percentage of highly educated respondents (70%) and the fact that only two political parties of the thirteen political parties that are currently in the House of Representatives were included in this experiment, it important to note that this research has several limitations which can lead to a low generalization of its finding to the broader public.
However still, this study has contributed to the field of microtargeting research, as there seems to be a gap in knowledge on understanding the psychology of voter behavior when it comes to microtargeting, especially in a European frame (Kruikemeier, Sezgin & Boerman, 2016). Researching this phenomenon is important, because the last few years scholars have expressed worry over the potential negative effects of microtargeting (Bacoras, 2012; Zuiderveen Borgesius., et al, 2019). The current study did not find any evidence to support the potential dangers of political microtargeting. The outcome of this study, confirms earlier statements from research such as Dobber., et al (2018) and Zuiderveen Borgesius (2019) that the potential dangers of microtargeting should not be overestimated; especially in a multi-party democracy such as the Netherlands. Moreover, this study has interesting practical implications as well, as it can provide political campaign teams with evidence that they can use microtargeting without having to worry about negatively affecting the political trust of potential voters.
Future research should continue looking into possible effects of microtargeting, also on other political attitudes such as political efficacy for example. With the growing popularity of microtargeting, it is important that scientific literature actively keeps tracking its potential influence, in order to make sure political microtargeting remains a nifty campaign tool, instead of a threat to the functioning of democracy.
Notes
1 The Instagram page of Jesse Klaver has 46,4.000 followers (“Jesse Klaver | Instagram”, n.d)
and the Facebook pages of the GroenLinks party and Jesse Klaver have a combined total of 194,000 followers (“GroenLinks | Facebook”, n.d; Jesse Klaver | Facebook, n.d). This is in stark contrast to the PvdA’s following: Their party leader, Lodewijk Asscher, has 9111 followers (“Lodewijk Asscher | Instagram”, n.d) on Instagram and his Facebook page combined with that of the PvdA party itself, has a following of 84.000 (“PvdA | Facebook”, n.d; Lodewijk Asscher | Facebook”, n.d). That is less than half of what GroenLinks has.
2
Awareness of microtargeting was measured through five items, on a 7 - point Likert scale ranging from 1 = strongly disagree, 7 = strongly agree. 1. I think this message was tailor
made for Robin, 2. I think Robin will feel specifically addressed after seeing this message
(both questions were adapted and slightly altered to fit the current study according to research by Baek & Morimoto (2012)). 3. I think Robin will notice this message, 4. I think Robin will
feel comfortable seeing this message, 5. I think Robin will be interested in seeing this message
(all questions were adapted and slightly altered to fit the current study according to research by Malheiros et al. (2012)).
3
Levene’s test indicated an insignificant result of .06, as this is >.05. However, .06 can be viewed as marginally significant, as it is <.10. Because of this, the Two-Way ANCOVA test for H3 was also measured with ‘age’ as an added covariate, to be sure the marginal
significance did not significantly influence the test for H3. Adding age however, did not lead to any significant findings, which is why ‘age’ is not further discussed as a covariate.
References
Albrow, M., & Easton, D. (1966). A Systems Analysis of Political Life. The British Journal
of Sociology, 17(2). https://doi.org/10.2307/589063
Barocas, S. (2012). The price of precision: voter microtargeting and its potential harms to the democratic process. In Proceedings of the first edition workshop on politics,
elections and data(pp. 31–36). ACM. https://doi.org/10.1145/2389661.2389671
Baek, T., & Morimoto, M. (2012). Stay Away From Me. Journal of Advertising, 41(1), 59– 76. https://doi.org/10.2753/JOA0091-3367410105
Bennett, C. (2015). Trends in Voter Surveillance in Western Societies: Privacy Intrusions and Democratic Implications. Surveillance & Society, 13(3/4), 370–384.
https://doi.org/10.24908/ss.v13i3/4.5373
Bennett, C. (2016). Is Your Neighbor and Democrat or a Republican? Lateral Voter
Surveillance and the Political Culture of Modern Election Campaigns. Lateral Voter
Surveillance and the Political Culture of Modern Election Campaigns (April 20, 2016) http://dx.doi.org/10.2139/ssrn.2776308
Bimber, B. (2014). Digital Media in the Obama Campaigns of 2008 and 2012: Adaptation to the Personalized Political Communication Environment. Journal of Information
Technology & Politics, 11(2), 130–150.
https://doi.org/10.1080/19331681.2014.895691
Boerman, S. C., & Kruikemeier, S. (2016). Consumer responses to promoted tweets sent by brands and political parties. Computers in human behavior, 65, 285-294.
https://doi.org/10.1016/j.chb.2016.08.033
Boerman, S. C., Willemsen, L. M., & Van Der Aa, E. P. (2017). “This post is sponsored”: Effects of sponsorship disclosure on persuasion knowledge and electronic word of
mouth in the context of Facebook. Journal of Interactive Marketing, 38, 82-92. https://doi.org/10.1016/j.intmar.2016.12.002
Braun, D., & Hutter, S. (2016). Political trust, extra-representational participation and the openness of political systems. International Political Science Review, 37(2), 151– 165. https://doi.org/10.1177/0192512114559108
Craig, S., Niemi, R., & Silver, G. (1990). Political efficacy and trust: A report on the NES pilot study items. Political Behavior, 12(3), 289–314.
https://doi.org/10.1007/BF00992337
Davison, W. (1983). The Third-person effect in communication. Public opinion quarterly :
journal of the American Association for Public Opinion Research, 47(1), 1–15.
https://doi.org/10.1086/268763
Dobber, T., Trilling, D., Helberger, N., & de Vreese, C. H. (2017). Two crates of beer and 40 pizzas: the adoption of innovative political behavioural targeting techniques. Internet
Policy Review, 6(4). https://doi.org/10.14763/2017.4.777
Dobber, T., Trilling, D., Helberger, N., & de Vreese, C. (2019). Spiraling downward: The reciprocal relation between attitude toward political behavioral targeting and privacy concerns. New Media & Society, 21(6), 1212–1231.
https://doi.org/10.1177/1461444818813372
Emerson, R. (2015). Convenience sampling, random sampling, and snowball sampling: How does sampling affect the validity of research? Journal of Visual Impairment and
Blindness, 109(2), 164–168. https://doi.org/10.1177/0145482X1510900215
European Commission. (2018, September 12). Free and Fair European Elections [European Commission State of the Union 2018]. Retrieved June 25, 2019, from
https://ec.europa.eu/commission/sites/beta-political/files/soteu2018-factsheet-free-fair-elections_en.pdf
Friestad, M., & Wright, P. (1994). The persuasion knowledge model: How people cope with persuasion attempts. Journal of consumer research, 21(1), 1-31.
https://doi.org/10.1086/209380
Gibson, R. K., & McAllister, I. (2011). Do online election campaigns win votes? The 2007 Australian “YouTube” election. Political Communication, 28(2), 227-244.
https://doi.org/10.1080/10584609.2011.568042
Gorton, W. (2016). Manipulating Citizens: How Political Campaigns’ Use of Behavioral Social Science Harms Democracy. New Political Science, 38(1), 61–80.
https://doi.org/10.1080/07393148.2015.1125119
GroenLinks | Facebook [Facebook fan page]. (n.d.). Retrieved June 25, 2019, from https://www.facebook.com/unsupportedbrowser
Ham, C., Nelson, M., & Das, S. (2015). How to Measure Persuasion Knowledge.
International Journal of Advertising, 34(1), 17–55,
https://doi.org/10.1080/02650487.2014.994730
Hillygus, D. S., & Shields, T. G. (2014). The persuadable voter: Wedge issues in presidential
campaigns. Princeton University Press. https://doi.org/10.1080/10584601003711445
Jang, S., & Kim, J. (2018). Third person effects of fake news: Fake news regulation and media literacy interventions. Computers in Human Behavior, 80, 295–302. https://doi.org/10.1016/j.chb.2017.11.034
Jesse Klaver | Facebook [Facebook fan page]. (n.d.). Retrieved June 25, 2019, from https://www.facebook.com/unsupportedbrowser
Jesse Klaver (@jesseklaver) • Instagram-foto's en -video's [Mobile App]. (n.d.). Retrieved June 25, 2019, from https://www.instagram.com/jesseklaver/?hl=nl
Knowles, E. S., & Linn, J. A. (2004). Resistance and Persuasion (pp. 1–337). Taylor and Francis. https://doi.org/10.4324/9781410609816
Kreiss, D. (2016). Prototype politics: Technology-intensive campaigning and the data of
democracy. Oxford University Press.
Kiesraad. (2017b, April 18). Officiële uitslag Tweede Kamerverkiezing 15 maart 2017. Retrieved June 4, 2019
Kruikemeier, S., Sezgin, M., & Boerman, S. (2016). Political Microtargeting: Relationship Between Personalized Advertising on Facebook and Voters’
Responses. Cyberpsychology, Behavior, and Social Networking, 19(6), 367–372. https://doi.org/10.1089/cyber.2015.0652
Levi, M., & Stoker, L. (2000). Political trust and trustworthiness. Annual review of political
science, 3(1), 475-507. https://doi.org/10.1146/annurev.polisci.3.1.475
Lodewijk Asscher | Facebook [Facebook fan page]. (n.d.). Retrieved June 25, 2019, from https://www.facebook.com/unsupportedbrowser
Lodewijk Asscher (@lodewijkasscher) • Instagram-foto's en -video's [Mobile App]. (n.d.). Retrieved June 25, 2019, from https://www.instagram.com/lodewijkasscher/?hl=nl Marien, S., Hooghe, M., & Marien, S. (2011). Does political trust matter? An empirical
investigation into the relation between political trust and support for law compliance. European Journal of Political Research, 50(2), 267–291. https://doi.org/10.1111/j.1475-6765.2010.01930.x
Mair, P. (2008). Electoral Volatility and the Dutch Party System: A Comparative
Perspective. Acta Politica, 43(2-3), 235–253. https://doi.org/10.1057/ap.2008.1 Malheiros, M., Jennett, C., Patel, S., Brostoff, S., & Sasse, M. A. (2012, May). Too close for
comfort: A study of the effectiveness and acceptability of rich-media personalized advertising. In Proceedings of the SIGCHI conference on human factors in
Motta, M. P., & Fowler, E. F. (2016). The content and effect of political advertising in US campaigns. In Oxford Research Encyclopedia of Politics. https:// DOI:
10.1093/acrefore/9780190228637.013.217
Murray, G., & Scime, A. (2010). Microtargeting and Electorate Segmentation: Data Mining the American National Election Studies. Journal of Political Marketing, 9(3), 143– 166. https://doi.org/10.1080/15377857.2010.497732
Pan, P., Meng, J., & Zhou, S. (2012). Examining Third-Person Perceptions in the Context of Sexually Oriented Advertising. Journal of Promotion Management, 18(2), 189–208. https://doi.org/10.1080/10496491.2012.668431
Papakyriakopoulos, O., Hegelich, S., Shahrezaye, M., & Medina Serrano, J. C. (2018). Social media and microtargeting: Political data processing and the consequences for
Germany. Big Data & Society. https://doi.org/10.1177/2053951718811844
Pennings, P., & Keman, H. (2008). The changing landscape of Dutch politics since the 1970s: A comparative exploration. Acta Politica, 43(2-3), 154-179.
https://doi.org/10.1057/ap.2008.
PvdA | Facebook [Facebook fan page]. (n.d.). Retrieved June 25, 2019, from https://www.facebook.com/unsupportedbrowser
Schipper, B., & Woo, H. (2019). Political awareness, microtargeting of voters, and negative electoral campaigning. Quarterly Journal of Political Science, 14(1), 41–88. https://doi.org/10.1561/100.00016066
Spierings, N., & Jacobs, K. (2019). Political parties and social media campaigning: A
qualitative comparative analysis of parties’ professional Facebook and Twitter use in the 2010 and 2012 Dutch elections. Acta Politica, 54(1), 145-173.
Trouw Redactie. (2019, March 22). Bekijk hier de uitslagen van de Provinciale Statenverkiezingen 2019. Retrieved April 23, 2019, from
https://www.trouw.nl/home/bekijk-hier-de-uitslagen-van-de-provinciale-statenverkiezingen-2019%7Ea1ac6c0b/
Turow, J., Delli Carpini, M. X., Draper, N. A., & Howard-Williams, R. (2012). Americans Roundly Reject Tailored Political Advertising—At a Time When Political
Campaigns Are Embracing It.
Wassens, R. (May 20, 2019). Dit geven de politieke partijen uit aan advertenties op Facebook. Retrieved on May 26, 2019, from
https://www.nrc.nl/nieuws/2019/05/20/sp-adverteert-het-meest-op-facebook-a3960892
Wolinetz, S. B. (2002). Beyond the catch-all party: approaches to the study of parties and party organization in contemporary democracies. Political parties: Old concepts and
new challenges, 136-165.
Zuiderveen Borgesius, F., Möller, J., Dobber, T., Kruikemeier, S., Irion, K., Stapel, S., … IViR. (2019). Online politieke microtargeting: Een zegen of een vloek voor de democratie? Nederlands Juristenblad, 94(10), 662–669.
Appendix A: Conceptual model
Trust in politics Persuasion knowledge Presumed Microtargeted message Type of party
Appendix B: survey questions
1. Wat is uw leeftijd? Vul uw leeftijd in getallen in (bijvoorbeeld '25'). 2. Wat is uw geslacht?
Man
Vrouw
Anders
3. Wat is uw hoogst afgeronde opleiding?
Geen onderwijs/ basisonderwijs
LBO/ VBO/ VMBO (kader- en beroepsgerichte leerweg
Eerste 3 jaar HAVO en VWO / MAVO / VMBO (theoretische en gemengde leerweg)
MBO
HAVO en VWO bovenbouw / WO en HBO propedeus
HBO of WO- bachelor of kandidaats
WO-doctoraal of master
4. Bent u stemgerechtigd in Nederland?
Ja
Nee
Weet ik niet
5. Ik wil u vragen de volgende tekst aandachtig te lezen:
Robin (35) is marketeer bij een start up in Amsterdam. Naast werken, houdt Robin van fietsen, series kijken en lekker uitgebreid koken. Bovendien is Robin zeer betrokken als het gaat om de bestrijding van racisme en discriminatie op de werkvloer. De marketeer strijdt al jaren tegen discriminatie door middel van betrokkenheid bij verschillende instanties zoals bij de "De Vreedzame Werkvloer". Dit wordt ook duidelijk op Facebook, waar Robin eigen ervaringen van DVW-acties post, inspirerende berichten en pagina's over racismebestrijding van anderen liket en nieuwsberichten over het thema racisme in Nederland deelt. Robin is in het algemeen niet extreem politiek betrokken, maar probeert zo af en toe het politieke nieuws in de gaten te houden.
Wat volgt, is een een politiek bericht dat op de Facebooktijdlijn van Robin voorbij kwam.
Het Facebookbericht is gesponsord door GroenLinks
Het Facebookbericht van GroenLinks is een advertentie
Het Facebookbericht van GroenLinks voelt als een advertentie
GroenLinks heeft betaald om dit Facebookbericht op de tijdlijn van Robin te krijge
Het Facebookbericht promoot GroenLinks
Antwoordopties: Helemaal mee oneens, Mee oneens, Beetje mee oneens, Neutraal, Beetje mee eens, Mee eens, Helemaal mee eens
7. Geef aan in hoeverre u het eens bent met de volgende statements:
Ik denk dat Robin geïnteresseerd is in het zien van dit Facebookbericht
Ik denk dat dit Facebookbericht Robin zal opvallen
Ik denk dat Robin zich in het speciaal aangesproken voelt na het zien van dit Facebookbericht
Ik denk dat Robin zich comfortabel zal voelen bij het zien van dit Facebookbericht
Ik denk dat dit Facebookbericht op maat gemaakt is voor Robin
Antwoordopties: Helemaal mee oneens, Mee oneens, Beetje mee oneens, Neutraal, Beetje mee eens, Mee eens, Helemaal mee eens
8. Geef aan in hoeverre u het eens bent met het volgende statement:
Na het zien van het Facebookbericht, zal Robin geneigd zijn om lid te worden van de politieke partij in kwestie
Antwoordopties: Helemaal mee oneens, Mee oneens, Beetje mee oneens, Neutraal, Beetje mee eens, Mee eens, Helemaal mee eens
9. Geef aan in hoeverre u het eens bent met het volgende statement:
Als ik Robin zou zijn, zou ik na het zien van het Facebookbericht geneigd zijn om lid te worden van de politieke partij in kwestie
10. Geef aan in hoeverre u het eens bent met de volgende statements:
Overheidsvertegenwoordigers proberen over het algemeen altijd het juiste doen
Ik heb het gevoel dat overheidsvertegenwoordigers vooral handelen uit eigen belang, in plaats van het publieke belang
De meeste overheidsvertegenwoordigers kiezen ervoor om te doen wat het beste is voor iedereen, zelfs als ze er zelf geen baat bij hebben
Wanneer overheidsvertegenwoordigers via traditionele media met het publiek communiceren, spreken ze meestal de waarheid
We moeten onze overheidsvertegenwoordigers goed in de gaten houden, zodat ze in het belang van alle mensen handelen
Democratisch verkozen overheidvertegenwoordigers proberen meestal hun verkiezingsbeloften na te komen
De meeste overheidsvertegenwoordigers kunnen adequaat om gaan met de problemen in ons land
Sommige democratisch verkozen overheidsvertegenwoordigers zijn niet zo eerlijk als het volk van ze mag verwachten
Van de meeste overheidsvertegenwoordigers kun je verwachten dat ze het juiste doen, zonder dat dit constant gecontroleerd hoeft te worden
Antwoordopties: Helemaal mee oneens, Mee oneens, Beetje mee oneens, Neutraal, Beetje mee eens, Mee eens, Helemaal mee eens
11. Van welke partij heeft u zojuist een Facebook bericht gezien?
CDA
GroenLinks
D66
PvdA
VVD
12. Geef aan in hoeverre u het eens bent met het volgende statement:
Ik zou de politieke partij van het Facebookbericht omschrijven als een digitaal vooruitstrevende partij
Antwoordopties: Helemaal mee oneens, Mee oneens, Beetje mee oneens, Neutraal, Beetje mee eens, Mee eens, Helemaal mee eens
13. Geef aan in hoeverre u het eens bent met het volgende statement
Het Facebookbericht dat ik zojuist heb gezien, bevat een gepersonaliseerde boodschap die Robin zou kunnen aanspreken
Antwoordopties: Helemaal mee oneens, Mee oneens, Beetje mee oneens, Neutraal, Beetje mee eens, Mee eens, Helemaal mee eens
Appendix C: Stimulus material
2. Non-microtargeted condition GroenLinks
