Spiral of Cynicism 2.0:
The Effects Of Negative User Comments On Citizens’
Political Attitudes and Interpersonal Trust
Master’s Thesis: Graduate School of Political Communication Master’s programme: Communication Science
Student: Milan van Ooijen Student number: 5735769
Thesis supervisor: Magdalena Wojcieszak
Abstract
In an ever-expanding online news media landscape, the rise of negative reader comments on news items is a worrying phenomenon that could possibly harm other people’s perceptions on politics, citizenship and trust. Building on the spiral of cynicism theory, we set up an online experiment with four manipulation groups to see whether negative comments featured in a proxy news item had an effect on
participants’ cynicism, and whether it lowered their feelings of political efficacy and interpersonal trust. Political knowledge was incorporated as a moderating variable to see if it could tame these effects. Results showed no significant direct effects of comment category on any of the three dependent variables, nor was it shown that there was a significant interaction of political knowledge on these effects. Although the tests proved insignificant, we did find some small hints of evidence - but with great reserve as to any form of generalizability - that pointed towards political knowledge having some effect on the relation between our independent variable and the dependent variables.
Introduction
In the new age of digital news media the consumption of news has increasingly become a social affair, as interaction between news media and consumers has begun to play an important factor in the daily routines of news media organizations that want to stay relevant in an increasingly interconnected world, where every idea, attitude or opinion can be blogged, tweeted, or commented upon. It has helped to create online news media communities where citizens can discuss local, national and international issues of all sorts, give their opinion on news items and have the opportunity to immediately hold journalists and their practices accountable much more than before
the Internet age, and much more vigilantly as well (Chen, 2011).
Next to social media like Twitter, Facebook or Instagram, these online news media communities are also vested in the comments sections that are now present at nearly every major news website. In a study by Tsagkias, Weerkamp and de Rijke (2010) it was found that in the period between November 2008 and April 2009, more than 600.000 comments were posted by readers on NUjij, the comment channel for popular news site NU.nl, while readers of the web version of free newspaper Spits generated over 425.000 comments posted. The online edition of Telegraaf, one of the largest and longest running newspapers in the Netherlands, also accumulated more than half a million comments by its readers in this period alone. In these comments sections, readers contribute in different ways to the news piece at hand by sharing their opinion or express their agreement or disagreement with the content or context of the news item. There has not yet been enough research done to provide exact figures on the share of online users that actually read these comments, but according to the Pew Internet & American Life Project (Purcell, Raince, Mitchell, Rosenstiel & Olmstead, 2010) there are strong indicators that the audience of comments sections are substantial.
An interactive news cycle can be beneficial to democracy when constructive ideas are revolving in our web space, when culture, politics, and social questions are vigorously being discussed upon. In its most perfect form, those discussions can spark more political interest and involvement in citizens, and as such can create a more democratic society (Henrich & Homes, 2012). For instance, approximately three-quarters of Internet users in the US became involved in the 2008 presidential
campaign because they were able to give their opinion through Internet channels. Of course the big winners here were Facebook and Twitter, but news sites also saw
significant increases in user engagement by posting comments and letting their voice be heard (Smith, 2009).
But what happens when comment sections are not used to their full potential? For when people have interpersonal talks and discussions about politics, the tendency to express very strong opinions always looms, and when it comes to politics, negative attitudes and dissatisfaction with government are more frequent than positive,
approving messages (Cappella & Jamieson, 1996; Richardson & Stanyer, 2011). An extensive body of research has shown us that political cynicism has been on the rise in Western democracies over the years prior to the consolidation of the modern Internet age, with anti-government sentiment on the rise in new generations of Internet users (Cooper & Mackenzie, 1981; Sveningsson, 2014). This ‘spiral of cynicism’ will be our main theoretical framework that we will introduce into the web 2.0 sphere to make our case. Old media and their use of strategic frames, conflict and generally more negative reporting that have gone before the Internet age are pointed out as being the main culprits, as well as more negative election races (Cappella & Jamieson, 1996, 1997).
Other signs of the worrying effects of exposure to negativity in the media have been sought in the lowered sense of political efficacy in citizens, which can result in an erosion of a sense of citizenship, and a lowered conviction that individual political action can make a difference (Ansolabehere & Iyengar, 1995). And while political distrust is seen as one potential result of the influences of negativity in political news content, some authors note that interpersonal trust of citizens is also in harm’s way (Kleinnijenhuis, van Hoof & Oegema, 2006).
There are, however, some moderating effects to be found that will alter the breadth of these effects on each individual. One of these moderators that have been
thoroughly acclaimed is political knowledge (De Vreese, 2005). There is evidence that the more knowledgeable someone is about politics, the better he or she can form own ideas and opinions on political developments and politics in general, and is less inclined to take over other people’s opinion instead. In our research we will
incorporate political knowledge as our moderator for the influence of negative comments on our dependent variables.
As we should be looking into modern-day sources of cynicism and new, harmful influences on the lubricates of proper democratic citizenship like efficacy and interpersonal trust, few studies have actually incorporated negative comments as a testing tool for the above-mentioned factors. Our task at hand is to fill this scientific gap and shed more light on this possible danger. The subsequent research questions that will be answered is as follows:
RQ1: Does exposure to negative reader comments (IV) in online news media result in
a rise in cynical attitudes towards politics (DV)?
RQ2: Does exposure to negative reader comments in online news media result in a
decline in political efficacy?
RQ3: Does exposure to negative reader comments in online news media result in a
decline in interpersonal trust?
RQ4: Does political knowledge have a moderating effect on the influences of negative
comments?
Theoretical background SPIRAL OF CYNICISM
Sceptical attitudes of citizens towards politics are a recurring phenomenon that seems to have always been an integral part of a healthy democracy (Cappella & Jamieson, 1997). Political cynicism, however, is not. Political cynicism has generally been defined as a distrust of the political system (Cappella & Jamieson, 1997; Cooper & Mackenzie, 1981; De Vreese & Elenbaas, 2008). More specifically, it has been explained as “a bitter or resentful attitude about the motivations and intentions of specific candidates under certain circumstances” (Cappella and Jamieson, 1997)
(p.142). It implies that citizens who experience political cynicism are, first of all, of the opinion that politicians are not honest with media and the public. Secondly, they feel that politicians are not looking after the interests of citizens, but rather more after their own. They generally feel that politicians are out of touch with what is really important in the lives of regular citizens and are therefor frustrated by the functioning of the political system in their country.
Instead of healthy scepticism, cynicism has a more negative basis. Within the confines of our research it implies that no matter what kind of message the political news content bears, it will be interpreted into a negative context by some, and will subsequently be negatively commented upon by them. An example of this perpetually negative state of mind could be the following: when news media announce that there will be a public hearing regarding the political decision-making process of the state supporting large banks in the Netherlands after the financial crisis of 2008, a typical online commenter would post a comment that spins this story as proof of corrupt politicians, who look after the bankers’ interests, and not the interests of the common citizen. Of course there is not a real issue with a few cynics living in society; the problem here is that those cynics now have the power to transfer their cynicism to other citizens, who are more unbiased towards government and politics, as well.
The term negativity, as it is employed in our case involving negative
commenting, holds any form of messages by the media and online media users that inclines discredit and a breaking of trust of the political system, as the above
paragraphs exemplify. However, negativity in this case should not be confused with ‘incivility’, a term used just as often in the academic debate on political cynicism. In theoretical frameworks of previous studies, incivility has been described as “claims that are inflammatory and superfluous” (Brooks & Geer, 2007) (p. 5). Brooks and Geer continue that there needs to be a clear distinction between being civilly critical and being uncivilly critical:
“Incivility requires going an extra step; that is, adding inflammatory comments that add little in the way of substance to the discussion” (Brooks & Geer, 2007) (p. 5).
It is exactly this substance of the debate that we are interested in, more specifically the cynical beliefs that are part of comments that could have potential effects on those exposed to them.
Feelings of political cynicism by citizens can be ascribed to different factors that have been developing over the span of decades. First of all, the media has been seen as a major culprit. It first showed in the era of television news, where scholars were starting to notice a change in the way news media reported on politics. News broadcasts, from the seventies on, were starting to rely more on negative bias towards government, preferring crisis, conflict and intrigue as political topics over neutral topics like important issues and legislation. This tendency of overly sceptical journalistic practices saw the start of a trend of cynicism that would start to grow in the following decades. Research in later years started to test what the actual effects of
negative political news reporting were on those that are exposed to it on a daily basis. A theory inextricably linked to this matter is that of the spiral of cynicism, making the case that the increasing focus on strategic frames in news content has led news consumers to become disenchanted from politics and politicians, because of the underlying negative context that they are fed on a daily basis (Cappella & Jamieson, 1996, 1997; de Vreese, 2004; de Vreese & Elenbaas, 2008). Secondly, election campaigns and political advertising have also been proven to carry part of the blame for growing cynicism within the general public (Schenck-Hamlin, Procter & Rumsey, 2000). In a longitudinal analysis by Patterson (1993), it is shown that elections and political campaigns were starting to focus more on candidates’ strategies and success and failure stories of their own camp and their adversary’s, and less so on actual relevant issues. Cappella and Jamieson (1997) carried out experiments that concluded that voters exposed to these types of frames in campaigns were causing higher levels of cynicism.
EFFICACY AND INTERPERSONAL TRUST
Perhaps less of a main topic in the academic world regarding media influence, but equally important to mention in our case, are two other effects that media negativity might have on other instruments of citizenship, starting with political efficacy. Political efficacy can be described as the belief of citizens that their political
involvement and opinions matter to the political decision-making process (De Vreese, 2005). The concept is divided into two parts: internal and external efficacy. Internal efficacy has been interpreted as “the belief of one’s own competence to understand, and to participate effectively in politics” (Niemi, Craig and Mattei, 1991) (p. 1407), while external efficacy is defined as “beliefs about the responsiveness of
governmental authorities and institutions to citizen demands” (Niemi, Craig and Mattei, 1991) (pp. 1407-1408). These factors are undoubtedly of great importance to one’s political self-consciousness as a valued citizen to society. In the renowned study by Ansolabehere and Iyengar (1995), rather alarming evidence was given towards the effects of negativity, whether in the form of strategic news or attack advertising in political campaigns, as they bear the potential to have an erosive effect on one’s sense of political efficacy. It could turn people away from the abovementioned notion that their actions can actually have an effect on politics, with all its possibly dire
consequences.
Secondly, interpersonal trust will be included into this study as well.
Interpersonal trust is explained as a layer in the grander concept of trust, also known as social capital, of which political trust is also a part. These different parts of a person’s social capital are interconnected (Kleinnijenhuis, 2006). As such, there is reason to assume that specific layers of social capital can affect one another for better or worse. To illustrate, when politicians are being portrayed in the media as self-obsessed, apathetic people who do not feel any responsibility in their actions
regarding citizens, this could also have an effect on how citizens view fellow citizens. The negativity that has damaged trust in politics could have a cascading effect on other layers of social capital, like interpersonal trust. This could help explain why virtually all dimensions of social capital have been declining in the last fifty years (Norris, 2002).
With this foundation of negative news- and campaign frames in mind, we can look into our present day to see in what way web 2.0 could be suffering from the same symptoms as old media – but instead of the newsroom, it is actually the citizen himself who could potentially be instigating cynicism now.
NEGATIVE COMMENTS, CYNICISM AND DEMOCRACY
In the new age of interactive online media where citizens have become an integral part of the news landscape, we should consider their impact just as seriously as we consider the impact of news content itself. There are signals that user-generated content like reader comments are not always as constructive as media professionals hope them to be (Sood, Churchill & Antin, 2012). We argue that while prior studies have shown that exposure to televised or printed political negativity violates the social norms of face-to-face political discussion, where one honours each other’s political disagreement and can adversely affect political attitudes and beliefs as such (Mutz & Reeves, 2005), the same can be said for political negativity in user-generated
comments.
Proof of online user-generated negativity can be found in studies like the one by Diakopoulos and Naaman (2011). Here, research was done to examine the
discourse experiences of media professionals in online comments sections. The results of the interviews and surveys with editors, writers and journalists showed that they were unanimously disillusioned by the apparent gap between how they themselves envisioned discourse in comments sections to be and daily reality. Many of the respondents and interviewees claimed that, despite efforts to keep online discussions constructive, many times conversations ended in extremely uncompromising
negativity by the commenters (Diakopoulos & Naaman, 2011). This has caused for great frustration in moderators and editors of online news media.
Daily exposure of vast numbers of readers to such negative and expressions of thought could have eroding effects on the way people think about their politicians and the political system. If cynicism starts to take root in the minds of the average reader, the democratic fabric of political life could be in serious harm. Governments operate
for a large part on the trust that is lent to them by the citizens living in them. Without that trust the daily operations of politicians can become increasingly difficult, through lower voter turnouts, a lowered sense of political efficacy and even interpersonal trust – on which we will elaborate in a following section.
To find out why negative commenting is so omnipresent in the online news landscape, we have to take a look at a number of aspects that are at play here. Most importantly is the fact that there is complete anonymity for all who wish to spread their view on news websites, which clears them of any inhibition they might have when facing people in real life (Tolkin, 2013; Santana, 2012). A commenter can post whatever he likes under a fictitious name without any real consequences, except for perhaps a quasi-punishment like a ban from the website or being excluded from the comment boards (Tolkin, 2013; Santana, 2012). The absence of non-verbal cues in reader comments sections have proven to be another motivation for some people to treat the online news media as a carte blanche catharsis for any form of
dissatisfaction or frustration with the political system (Davis, 2005; Borah, 2012).
THE MODERATING EFFECT OF POLITICAL KNOWLEDGE
However, there is reason to believe that the impact of negativity is moderated by various factors. A notion that has gained wide acceptance in the academic field over the last decades is that political knowledge moderates the effect that strategic frames have on people. Political knowledge or political sophistication has been described as those individuals who “pay more attention to politics and understand what [they] have encountered” (Zaller, 1992) (p. 21). There is some consensus that the more politically knowledgeable someone is, the less likely that person is to actually be affected by negativity in certain messages (Kahn & Kenney, 1999). The assumption is
that negativity would not have much weight for a citizen who has a firm grasp of the political developments around him and is thus able to make his own assumptions and ideas, and would therefore not value such messages with the same gravity the way somebody with little political knowledge would. The basis of information on
politicians, politics and parties could create a context for interpreting these stories to a person’s own beliefs and convictions and thereby minimizing its persuasive impact (Valentino, Beckmann & Buhr, 2010). This notion can be implemented into our case as well, as political knowledge could have the same power to resist the effects of cynicism-inducing negativity on the web, where negativity is created by the users themselves and not the media. With that in mind, we expect political knowledge to negatively moderate for the effects that negative comments will have on levels of political cynicism, political efficacy and interpersonal trust.
HYPOTHESES
So, as we can understand from the body of literature discussed above, we want to take the next step and study whether the possible effects on political cynicism, efficacy and interpersonal trust are not coming from just the news media or politicians, as the spiral of cynicism proclaims, but from (interactive) citizens themselves and their online contributions that are potentially spreading distrust and an erosion of feelings
of citizenship. We can distill the following hypotheses from what we have gathered in the existing literature:
H1a: Exposure to negative comments results in more cynicism towards politics within readers than exposure to positive comments.
H1b: Exposure to negative comments results in more cynicism towards politics within readers than exposure to no comments at all.
H1c: Exposure to negative comments results in more political cynicism within readers than exposure to mixed negative and positive comments. H2: The higher someone’s political knowledge, the less pronounced the effect of negative comments on political cynicism will be.
H3a: Exposure to negative comments will have a greater negative effect on both internal and external political efficacy within readers than exposure to positive comments.
H3b: Exposure to negative comments will have a greater negative effect on both internal and external political efficacy within readers than exposure to no comments at all.
H3c: Exposure to negative comments will have a greater negative effect on both internal and external political efficacy within readers than exposure to mixed comments.
H3d: The higher someone’s political knowledge, the less pronounced the negative effect of exposure to negative comments on both internal and external political efficacy will be.
H4a: Exposure to negative comments will have a greater negative effect on interpersonal trust within readers than exposure to positive comments.
H4b: Exposure to negative comments will have a greater negative effect on interpersonal trust within readers than exposure to no comments at all. H4c: Exposure to negative comments will have a greater negative effect on interpersonal trust within readers than exposure to mixed comments. H4d: The higher someone’s political knowledge, the less pronounced the negative effect of exposure to negative comments on interpersonal trust will be.
Method
Research Design
The study was conducted through an online experiment, hosted on Qualtrics survey software, to test for possible effects of reader comments on the respondents’ political attitudes. Respondents read a short objective domestic political news story about fines on biking without lights in Amsterdam. The proxy webpage, called ‘De
Amsterdamsche Courant’, was created through copy-pasting parts from actual Dutch news sites (‘NU.nl’, ‘De Gelderlander’ and ‘Het Parool’). The news story itself was gathered from Het Parool. The ‘fake’ name is unique and is not used for any other news organization in the Dutch media landscape. This safeguarded internal validity by making sure respondents would not be influenced by reading from an existing newspaper that they might already know and have a pre-existing opinion on.
Respondents were divided into four conditions that were randomly assigned through Qualtrics. One condition carried 50 respondents. There were three conditions with comments to the right of the news item, and one without comments that
functioned as the control group. The three different conditions with comments featured twelve comments. The negative condition carried non-constructive, rude,
alienating and other types of negative comments, the second featured constructive and positive comments, while the third condition had a mix of positive and negative comments. To guarantee every respondent had the same chance of falling into one of four conditions in the randomization process, we employed the setting in Qualtrics which makes this possible – the result was a practically even division of respondents into the four manipulation categories (positive: 22.3%; negative: 25%; mixed: 28.7%; no comment 23.9%).
Selection of research units
Respondents were gathered through random bulk mail-outs to a first round of family, colleagues and acquaintances. Subsequently, they were all asked to share the
Qualtrics survey via mailout to their contacts, after which they too were asked to re-share, etcetera. This initiated a snowball effect where every new round of mail-outs gathered new respondents. All surveys were initiated and filled out online. The response rate was 34.6%. The sample frame consisted of two properties, namely that the respondent needed to be 18 years or older and a citizen of the Netherlands. The completion rate was 58.8%. After excluding the cases that quit the survey before the manipulation (so who had no reason to be taken into consideration in the rest of the analysis) the total number of valid respondents was 188 (N=188). The survey data was retrieved from Qualtrics, on which the survey ran for a total of two weeks.
Sample characteristics
The sample consisted of Dutch citizens only. The most important reason for this was that the moderating variable political knowledge was tested through a number of questions of which two were focused on Dutch politics (the name of the minister for
Foreign Affairs and the number of years in between Dutch parliamentary elections), which would have made it far harder to answer for non-Dutch citizens. Also, we needed Dutch citizens as our proxy news site will have a Dutch news item with Dutch comments.
The mean age of participants was 36 years (M=36.39, SD=14.91), while slightly more than half of all participants were female (52% female; M=1.52, SD=.50). The education level of participants was relatively high (29.2% of respondents had a college education (‘hogeschool’), and 40.6% had a University degree; M=6.07, SD=1.76). Lastly, the mean party preference was situated in the center (highest score for center-right party D66 with 39.4%; M=4.94, SD=2.70).
Manipulations
The news piece was of an appropriately short size (263 words) to ensure that
respondents had the possibility to be directly exposed to the comments section next to it. To ensure that all respondents were able to read the news item and the comments, short instructions were given on how to zoom in and out of the fabricated webpage of ‘The Amsterdamsche Courant’ for both Mac and PC users. A 15-second clock was installed on all four conditions, granting respondents enough time to read through the text and comments before being able to continue the survey.
All comments were completely fictional; none of them were copied from actual comments sections, message boards or other existing public opinion displays. Also, every detail in the comments sections in the four conditions was equated, meaning that factors like word count (an average of 13 words per comment across manipulation groups), tone of voice and context within were mirrored as optimally as possible in the positive and negative conditions. The mixed condition was built on six
negative and six positive comments. The arguments constructed for the positive comments were the opposite in the negative condition and vice versa. The four manipulations acted as our independent variables on our dependent variable ‘political cynicism’.
We ran a test pilot before the actual survey itself, consisting of a small test sample (N=34) where people were randomly assigned to one of the four conditions, and asked what their opinion and attitude was on different aspects of the proxy news item and the reader comments (i.e. clarity, readability, credibility of the comments, etc.). From these results, final alterations were made to the manipulation setup, the visualization of the manipulation and the questions that would follow them.
Measures: In the post-test questionnaire, questions were asked regarding the main dependent variable cynicism, as well as questions on the secondary dependent
variables political efficacy and interpersonal trust. All items were set up as positively formulated statements on which respondents could agree or disagree on a 7-point Likert scale (1=completely agree, 7= completely disagree).
To measure our main dependent variable political cynicism we gathered the various sets of standardized items (that were relevant to our research) adapted by the
SAGE Handbook of Political Communication (Semetko & Scammell, 2012). The items in the form of statements revolved around the following dimensions regarding politicians and parties and their perceived actions: credibility, trustworthiness, interest in electorate, competence, division of power, issue-solving capability, politics as an ‘old-boys network’, and politicians’ capability of understanding of issues deemed important by the public.1
Since the answer categories were coded in such a manner that 1 meant ‘very cynical’ and 7 meant ‘not cynical’, we decided to reverse code these items (aside from questions 7.6 and 7.7, which were already positively formulated). After a reliability analysis (a=.81) and a factor analysis, it turned out that all items loaded onto one component, except for item 7.6 and 7.7, which loaded onto a second component with a lesser reliability (a=.68). We decided these two items would be exempted from analysis. We constructed a 7-item scale ‘cynicism’ (M=4.30; SD=1.00).
To test for internal and external efficacy, we also used standardized items from the SAGE Handbook of Political Communication (Semetko & Scammell, 2012).
Regarding internal efficacy, the following statements were adapted: “I know more about politics than most of the people my age”; “When there are political issues being discussed, I normally want to engage in those discussions” and “I am easily capable of comprehending political issues”. Just like with the cynicism items, we reverse coded these positively formulated items so that 1 meant low political efficacy and 7 meant high political efficacy. After a reliability analysis (a=.77) and a factor analysis that showed the items loaded on one component, a three-item scale was created named ‘internal efficacy’ (M=4.44; SD=1.17). External efficacy was tested by introducing respondents to the following two statements: “The government cares about citizens’ opinions when regarding new legislative bills” and “Government does its best to seek out what ordinary citizens want”. After checking the inter-item
correlation (r=.66), a two-item scale ‘external efficacy’ was computed (M=3.73; SD=1.16).
Interpersonal trust was tested through two standardized questions from the SAGE Handbook of Political Communication (Semetko & Scammell, 2012), one asked whether respondents agreed with the statement: “Generally, most people can be
trusted”, while the second statement asked respondents to agree or disagree on
whether “Most people would try to take advantage of me if they have the chance”. As the first statement was positively formulated and the second negatively, the first was reverse coded in order to let value 7 stand for most interpersonal trust. After checking for inter-item correlation (r=.52) we constructed a two-item scale ‘interpersonal trust’ (M=4.87; SD= 1.14).
Political knowledge was constructed on a three-item scale consisting of dichotomized (0=wrong answer, 1=right answer) items relating to politics. As mentioned earlier, two questions, regarded Dutch politics: “What is the amount of years in between parliamentary elections in the Netherlands?” and “Who is the current Dutch minister of Foreign Affairs?”, and one question regarding the EU: “Who has been the president of the EU parliament in the last two years?” (2012-2014). All three items were derived from the Dutch Parliamentary Election Study (Aarts, van der Kolk, & Kamp. 1999). The outcome of the three questions was constructed into a scale where 0 to 1 questions rightly answered was coded as ‘low political knowledge’, 2 questions right was coded as ‘medium political knowledge’ and 3 out of 3 right was coded as ‘high political knowledge’. This scale was then dichotomized on the mean (M=1.88; SD=.85), where scores under the mean were coded as 0 (low-medium political knowledge) and scores above the mean were coded as 1 (medium-high political knowledge).
Results
To test our first three hypotheses, we ran a univariate analysis of variance
(UNIANOVA) on comment category by our first dependent variable cynicism F(3, 184) = 1.77 p =.154, as shown in table 2. From the results in table 1 we can see that
levels of cynicism are higher for the negative manipulation group (M=4.41; SD=.97) than for the positive manipulation group (M=4.30; SD=1.00) and the mixed
manipulation group (M=4.08; SD=.85).2 However the control group showed higher levels of cynicism (M=4.50; SD=1.15) than the negative manipulation participants. Against predictions, the test showed no direct effect of comment categories on cynicism, and as such we cannot get into post-hoc testing to find significant between-group effects. We have to reject H1a to H1c.
Table 1 here Table 2 here
Our second analysis of variance tested the effect of comment category on cynicism while moderated by political knowledge. As shown in table 4, the direct effect of comment category on cynicism slightly changed, F(3, 184) = 2.07, p=.106, the interaction effect of comment category by political knowledge did not prove significant, F(3,184) = .73, p = .534, however. Looking at table 3 and figure 1, we do see that participants with higher levels of political knowledge score slightly lower on cynicism (M=4.34; SD=1.05) than participants with average to low levels of political knowledge (M=4.59; SD=.73). Even though there is no basis for these results to be deemed statistically significant, these findings do hint that there might be some dependence on political knowledge by the effects of negative comments on cynical attitudes. Nonetheless, we need to reject H2.
Table 3 here
Table 4 here Figure 1 here
To test for our next hypothesis, we will run an UNIANOVA for comment category by political efficacy in two parts, starting with internal efficacy. As table 6 shows, the test did not prove a significant effect of comment categories on internal political efficacy, F(3, 182) = 1.00, p = .395. The mean values as shown in table 5 also proved to be opposite of what we expected, as participants in the negative manipulation actually scored higher on internal efficacy (M=4.59; SD=1.31) than participants in the positive manipulation (M=4.49; 1.21), the mixed manipulation (M=4.48; SD=1.05) and the control group (M=4.19; SD=1.11). We then replaced internal efficacy with external efficacy in our next UNIANOVA. Again, there was no statistically significant relation between comment category and the dependent
variable, external efficacy, F(3, 182) = 1.36, p = .256, as is shown in table 8. The mean scores in table 7 fall slightly more into our line of expectations, as participants in the negative manipulation scored lower on external efficacy (M=3.80; SD=1.33) than participants in the positive manipulation (M=3.82; SD=1.22), while the mixed manipulation (M=3.86; SD=.91) and the control group (M=3.43; SD=1.16) scored higher than the negative manipulation. We must reject H3a to H3c, as there is no statistically significant direct effect of comment category on both internal and external efficacy.
Table 5 here Table 6 here Table 7 here Table 8 here
To test for H3d we introduced political knowledge into the equation as our moderator for the direct effect of comment categories for political efficacy. Our UNIANOVA test for comment category by internal efficacy while moderated by political knowledge did not produce significant results either, which can be seen in table 10, as the moderator effect fell just outside of the confidence level, F(3, 178) = 2.38, p = .071. The mean table 9 shows us that there is quite a difference in internal efficacy levels between the knowledge groups in the negative manipulation, however. The low-medium knowledge participants scored lower on internal efficacy after exposure to negative comments (M=3.77; SD=1.34) than the medium-high
knowledge participants in the same manipulation group (M=4.91; SD=1.17). In figure 2 we can also see a noticeable difference in levels of internal efficacy for the low to medium knowledge group in the positive condition (M=4.72;SD=.77) as compared to the average score on internal efficacy by the other manipulation conditions in the low to medium knowledge group (M=3.79; SD=1.12). What is remarkable is that the mean score for the low-medium knowledge participants across all manipulation groups on internal efficacy is higher than that for the medium-high knowledge group across all manipulation groups (M=4.45; SD=1.27).
Table 9 here Table 10 here Figure 2 here
When internal efficacy was replaced by external efficacy in the moderator analysis, we got more or less the same outcome, as there was not a statistically
significant moderator effect to be found here either F(3,178) = .08, p = .970, as can be seen from table 12. The mean table 11, however, shows us that participants belonging to the higher-politically knowledgeable group score higher on external political efficacy (M=3.94; SD=1.27) after begin exposed to negative comments than participants in the low to medium knowledge group (M=3.46; SD=1.48). As both tests failed to produce any statistically significant moderator effects for political knowledge for the effect of comments category on political efficacy, we need to reject H3d.
Table 11 here Table 12 here Figure 3 here
Moving on towards our final dependent variable and hypotheses, we ran a UNIANOVA to test for the effect of comment category on interpersonal trust. Here we find a significant direct effect of comment category on levels of interpersonal trust, F(3,179) = 3.445, p<.05, which can be seen in table 14. As table 13 tells us, we can see that participants in the negative manipulation actually scored remarkably higher on interpersonal trust (M=5.21; SD=.96) than participants in the mixed condition (M=4.66; SD=1.26) and the control group (M=4.58;SD=1.28), while the positive manipulation participants scored relatively high on interpersonal trust as well (M=5.07; SD=.86). Even though Levene’s test tells us that there is a slight inequality of variance, F(3,179) = 4.65, p<.05, we decided to look at the post-hoc results
nonetheless. From the post-hoc results table 15 we see that there are statistically significant between group differences, however they are not in line with our
expectations. Participants in the negative comment manipulation scored slightly higher than participants in the positive manipulation on interpersonal trust, but the difference is insignificant (p=.578). We reject H4a. Participants in the negative manipulation also scored significantly higher on interpersonal trust than the mixed manipulation sample (p<.05). We reject H4c. The same goes for the differences in interpersonal trust between the negative manipulation and the control group, where participants scored significantly higher than participants in the control group (p<.01). We reject h4b. It is also shown that participants in the control group scored
significantly lower than participants in the positive manipulation (<.05).
Table 13 here Table 14 here Table 15 here
Our final hypothesis expects that political knowledge will moderate negatively for the effects of negative comments on interpersonal trust. Our last UNIANOVA shows an insignificant interaction effect for political knowledge on the relation between comment category and interpersonal trust F(3, 175) = .757, p=.520, as is shown in table 17. Table 16 shows us that participants in the medium to high political knowledge group scored slightly higher on interpersonal trust (M=5.23; SD=.94) after being exposed to negative comments than the low to medium knowledge group (M=5.15; SD=1.03). Figure 4 also shows us that participants in the medium-high political knowledge group scored radically higher on interpersonal trust after being exposed to positive comments than the participants with low-medium political knowledge in the same manipulation group. As the interaction of political knowledge
on the effect of comment category on interpersonal is insignificant, we must reject H4d. Table 16 here Table 17 here Figure 4 here Conclusion
The goal of this paper, while drawing on the theory of the spiral of cynicism, was to find evidence for the notion that negative comments in online news articles would have an effect on cynicism (higher feelings of political cynicism after exposure), political efficacy (lower feelings of efficacy after exposure) and interpersonal trust (lower feelings of interpersonal trust). Our second goal was to see if there were any moderating effects for political knowledge, predicting that higher knowledge would grant a safeguard against the effects of negative comments. An experiment was conducted in which we would present participants with a proxy online news article that featured a dozen comments ranging from positive to negative and mixed-substance comments.
The results made it clear that for the most part, our expectation were not met, due to statistically insignificant test results on all tests that were run and because a lot of the scores that were gathered proved to be the opposite of what we anticipated in the first place. Therefor we cannot draw any
generalizable conclusions at this point. That is not to say that the entire study proved meaningless, as there were some hints that pointed in the right
direction, of course all with great reserve, as the effect did not show any significance. The first noteworthy observation we made was that mean scores on cynicism were lower for those with higher knowledge of politics than for those participants with lower political knowledge. Secondly, the mean scores for internal political efficacy of participants in the negative manipulation were actually higher than the scores of those in the other manipulation
categories, which is odd, as political knowledge was supposed to lower levels of efficacy, not heighten them as opposed to the positive manipulation and the control group.
When political knowledge was included as the moderator, the findings did point towards our expectations again, however, as we saw sharp
differences in mean values for both internal and external efficacy between the knowledge groups, as the higher knowledgeable participants scored higher on efficacy when shown negative comments, as compared to the less
knowledgeable participants. The same goes for interpersonal trust, where it was hinted that the higher knowledgeable participants showed to be more trustful of other people after exposure to the negative manipulation than participants with lower political knowledge.
But, as mentioned before, none of these outcomes can be trusted and generalized onto the population, as the direct and interaction effects proved to be insignificant. In our discussion we will set out for an explanation for this disappointing outcome.
There are numerous factors that could have contributed to the outcome of our experiment. First and foremost, there is something to be said regarding the sample size. A total of just fewer than 200 participants is barely a robust sample to distill many significant results from. The relatively small size of our sample can partly be blamed on the disappointing response rate and an even more saddening rate of completion, as nearly half of all respondents quit the survey when they arrived at the manipulation stage. The reason for this was that participants who made the survey on a smartphone or tablet could not zoom in on the proxy news item. Even though there was a disclaimer in every survey requisition, stating clearly that the survey did not work on these devices, these respondents ignored or did not see the warning and continued to start the survey regardless of this cautionary statement.
Our second notion of critique is that the comment sections as used in the experiment did not result in their respectful effects because the news item was rather tame in nature. A piece about biking fines in Amsterdam does not evoke an emotional reaction in people, and so the comments, whether they’d be negative, positive, or mixed, perhaps seemed rather misplaced by some. Especially the negative manipulation could have struck the wrong chord with some participants, as you would see such negatively substantiated messages resulting from news items that could actually cause distress in people, like tax raises for ordinary income citizens, or delicate issues like abortion or
immigration. In our case, the news piece might have caused the comments to lose some of their power and possible impact on the negative manipulation sample. It could be for that reason that in some cases the negative
groups and lower on cynicism when compared to the other groups. Focusing on the reason why our moderator did not produce any significant interaction effects could be due to the fact that the education level in our sample was relatively high. That being the case, the division of
political knowledge was rather eschew, in favour of participants with high knowledge. This resulted in unequal groupings when running our tests and as such could not bring forth any valid results.
On a broader scale, the research design itself was far from perfect, as we tested for effects of comments in one single measure, which of course limits the chances of actually generating the expected effect in our sample. For future studies regarding our topic, it is advised to adapt a longitudinal design with repeated exposure and measures over a longitudinal period of time. Only then can there be substantial proof for the claims we made in this study.
On a final note, this study aimed at shedding light on a development in modern news media use which is ever-growing: the posting of biased messages on news items and the silent majority that is reading those
comments. As we are moving more and more towards an online news realm where the influence of traditional mass media is starting to decline, it is pivotal to expand knowledge in the academic field regarding the new ways in which citizens interact with each other regarding political information, and engage in online discussion. It is of the utmost importance to come to a clear understanding of the effects it has on the perception of government,
parliament, politicians and politics in general, and people’s feeling of actually being a part of democracy and trust. For it is well possible that we are
entering, or have already entered, a new age of cynicism, where not the media are causing this cynicism, but citizens amongst each other in the practically unregulated online discussion boards of online news media.
References
- Aarts, K., Van der Kolk, H., & Kamp, M. (1999). Dutch Parliamentary Election
Study, 1998. Inter-university Consortium for Political and Social Research.
- Ansolabehere, S., & Iyengar, S. (1996). Going negative: How political ads shrink and polarize the electorate. New York: Free Press.
- Blais, A., & Rubenson, D. (2013). The Source of Turnout Decline New Values or New Contexts? Comparative Political Studies, 46(1), 95-117.
- Borah, P. (2012). Does it matter where you read the news story? Interaction of incivility and news frames in the political blogosphere. Communication Research,
20(10), 1-19.
- Cappella, J. N., & Jamieson, K. H. (1996). News frames, political cynicism, and media cynicism. The Annals of the American Academy of Political and Social
Science, 71-84.
- Cappella, J. N., & Jamieson, K. H. (1997). Spiral of cynicism: The press and the public good. New York: Oxford University Press.
- Chen, K. J. (2011). A test of the spiral of silence theory on young adults' use of social networking sites for political purposes. Graduate Theses and Dissertations: Paper 12214.
- Chung, D. S. (2008). Interactive features of online newspapers: Identifying patterns and predicting use of engaged readers. Journal of Computer‐Mediated
- Cooper, J. & Mackenzie, G., C. (1981). Can Congress Ever Be A Popular Institution. In The House At Work. Austin: University of Texas Press.
- Craig, S. C., Niemi, R. G., & Silver, G. E. (1990). Political efficacy and trust: A report on the NES pilot study items. Political Behavior, 12(3), 289-314.
- Davis, R. (2005). Politics online: Blogs, chatrooms, and discussion groups in
American democracy. New York: Routledge.
- De Vreese, C. H. (2004). The effects of strategic news on political cynicism, issue evaluations, and policy support: A two-wave experiment. Mass Communication &
Society, 7(2), 191-214.
- De Vreese, C. H., & Elenbaas, M. (2008). Media in the game of politics: Effects of strategic metacoverage on political cynicism. The International Journal of Press/Politics, 13(3), 285-309. - Dennis, J. & Webster, C. (1975). Children’s images of the president and of government in 1962 and 1974. American Politics Quarterly, 3, 386–405.
- Diakopoulos, N. & Naaman, M. (2011, March). Towards quality discourse in online news comments. In Proceedings of the ACM 2011 conference on Computer supported
cooperative work (pp. 133-142).
- Hanson, G., Haridakis, P. M., Cunningham, A. W., Sharma, R., & Ponder, J. D. (2010). The 2008 presidential campaign: Political cynicism in the age of Facebook, MySpace, and YouTube. Mass Communication and Society, 13(5), 584-607. - Henrich, N., & Holmes, B. (2013). Web news readers comments: Towards
developing a methodology for using on-line comments in social inquiry. Journal of
Media and Communication Studies, 5(1), 1-4.
- Kahn, K. F., & Kenney, P. J. (1999). Do negative campaigns mobilize or suppress turnout? Clarifying the relationship between negativity and participation. American
- Kleinnijenhuis, J., van Hoof, A. M., & Oegema, D. (2006). Negative news and the sleeper effect of distrust. The Harvard International Journal of Press/Politics, 11(2), 86-104.
- Liu, X., & Fahmy, S. (2011). Exploring the spiral of silence in the virtual world: lndividuals willingness to express personal opinions in online versus offline settings.
Journal of Media and Communication Studies, 3(2), 45-57.
- Mutz, D. C., & Reeves, B. (2005). The new videomalaise: Effects of televised incivility on political trust. American Political Science Review, 99(1), 1-15. - Nekmat, E., & Gonzenbach, W. J. (2013). Multiple Opinion Climates in Online Forums Role of Website Source Reference and Within-Forum Opinion Congruency.
Journalism & Mass Communication Quarterly, 20(10), 1-21.
- Norris, P. (2002). Democratic phoenix: Reinventing political activism. Cambridge University Press.
- Patterson, T. E. (1993). Out Of Order. New York: Knopf.
- Purcell, K., Raine, L., Mitchell, A., Rosenstiel, T., & Olmstead, K. (2010).
Understanding the participatory news consumer: How Internet and cell phone users have turned news into a social experience. Pew Internet & American Life Project. - Richardson, J., & Stanyer, J. (2011). Reader opinion in the digital age: Tabloid and broadsheet newspaper websites and the exercise of political voice. Journalism, 12(8), 1-21.
- Santana, A. D. (2012). Civility, anonymity and the breakdown of a new public
sphere (Doctoral dissertation, UNIVERSITY OF OREGON).
- Schenck‐Hamlin, W. J., Procter, D. E., & Rumsey, D. J. (2000). The influence of negative advertising frames on political cynicism and politician accountability.
- Semetko, H. A., & Scammell, M. (Eds.). (2012). The SAGE handbook of political
communication. New York: SAGE Publications.
- Smith, A. (2009). The Internet’s role in campaign 2008. Pew Internet & American
Life Project, 15.
- Sood, S. O., Churchill, E. F., & Antin, J. (2012). Automatic identification of
personal insults on social news sites. Journal of the American Society for Information
Science and Technology, 63(2), 270-285.
- Sveningsson, M. (2014). “I don’t like it and I think it’s useless, people discussing politics on Facebook”: Young Swedes’ understandings of social media use for political discussion. Cyberpsychology, 8(3).
- Tolkin, A. (2013). Anonymity, Polarization, and the Online Sphere. Slideshow. St. Louis: Washington University.
- Tsagkias, M., Weerkamp, W., & De Rijke, M. (2010). News comments: Exploring, modeling, and online prediction. In Advances in Information Retrieval, 191-203. Springer Berlin Heidelberg.
- Valentino, N. A., Beckmann, M. N., & Buhr, T. A. (2001). A spiral of cynicism for some: The contingent effects of campaign news frames on participation and
confidence in government. Political Communication, 18(4), 347-367.
- VVD wil dat controleurs Stadstoezicht ook fietsboetes gaan uitdelen (2014, November 25). Het Parool. Retrieved from
http://www.parool.nl/parool/nl/124/VERKEER/article/detail/3798038/2014/11/25/VV D-wil-dat-controleurs-Stadstoezicht-ook-fietsboetes-gaan-uitdelen.dhtml
- Zaller, J. R. (1992). The nature and origins of mass opinion. Cambridge: Cambridge University Press.
Appendix A Tables
Table 1: Mean values of respondents’ cynicism by comment
category
Dependent Variable: cynicism
comment category Mean Std. Deviation N
mixed 4.0794 .84614 54
negative 4.4103 .97102 47
no comments 4.5048 1.15280 45
positive 4.2313 1.02820 42
Total 4.2979 1.00379 188
Table 2: UNIANOVA Model for cynicism by comment category
Dependent Variable: cynicism
Source Type III
Sum of Squares df Mean Square F Sig. Corrected Model 5.285a 3 1.762 1.770 .154 Intercept 3457.270 1 3457.270 3473.577 .000 Comment_categories 5.285 3 1.762 1.770 .154 Error 183.136 184 .995 Total 3661.102 188
Corrected Total 188.421 187 a. R Squared = .028 (Adjusted R Squared = .012)
Table 3: Mean values of respondents’ cynicism by experimental
group moderated by political knowledge Dependent Variable: cynicism
comment category
political knowledge Mean Std. Deviation N mixed low-medium pol_know 4.2707 .65965 19 medium-high pol know 3.9755 .92419 35 Total 4.0794 .84614 54 negative low-medium pol_know 4.5934 .73273 13 medium-high pol know 4.3403 1.04920 34 Total 4.4103 .97102 47 no comments low-medium pol_know 5.0000 1.05108 16 medium-high pol know 4.2315 1.13094 29 Total 4.5048 1.15280 45 positive low-medium pol_know 4.3333 1.81528 6
medium-high pol know 4.2143 .87498 36 Total 4.2313 1.02820 42 Total low-medium pol_know 4.5714 .99363 54 medium-high pol know 4.1876 .99027 134 Total 4.2979 1.00379 188
Table 4: UNIANOVA model for cynicism by comment category moderated by
political knowledge
Dependent Variable: cynicism
Source Type III Sum
of Squares df Mean Square F Sig. Corrected Model 13.122a 7 1.875 1.925 .068 Intercept 2551.620 1 2551.620 2620.050 .000 Comment_categories 6.037 3 2.012 2.066 .106 political_knowledge 4.304 1 4.304 4.419 .037 Comment_categories * political_knowledge 2.142 3 .714 .733 .534 Error 175.299 180 .974 Total 3661.102 188 Corrected Total 188.421 187 a. R Squared = .070 (Adjusted R Squared = .033)
Figure 1: Mean interaction plot of comment category by cynicism moderated by
political knowledge groups.
Table 5: Mean values of respondents’ internal
efficacy by experimental group
Dependent Variable: internal political efficacy comment category Mean Std. Deviation N mixed 4.4780 1.04903 53 negative 4.5870 1.31315 46 no comments 4.1852 1.11136 45 positive 4.4921 1.21025 42 Total 4.4373 1.17018 186
Table 6: UNIANOVA model respondents’ internal efficacy by experimental group Dependent Variable: internal political efficacy
Source Type III Sum
of Squares df Mean Square F Sig. Corrected Model 4.104a 3 1.368 .999 .395 Intercept 3633.307 1 3633.307 2653.331 .000 Comment_categories 4.104 3 1.368 .999 .395 Error 249.220 182 1.369 Total 3915.556 186 Corrected Total 253.324 185 a. R Squared = .016 (Adjusted R Squared = .000)
Table 7: Mean values of respondents’ external
efficacy by experimental group
Dependent Variable: external political efficacy comment category Mean Std. Deviation N mixed 3.8585 .90606 53 negative 3.8043 1.33116 46 no comments 3.4333 1.15601 45 positive 3.8214 1.21886 42 Total 3.7339 1.15683 186
Table 8: UNIANOVA model respondents’ external efficacy by experimental group Dependent Variable: external political efficacy
Source Type III Sum of Squares df Mean Square F Sig. Corrected Model 5.438a 3 1.813 1.362 .256 Intercept 2568.535 1 2568.535 1930.603 .000 Comment_categori es 5.438 3 1.813 1.362 .256 Error 242.139 182 1.330 Total 2840.750 186 Corrected Total 247.577 185 a. R Squared = .022 (Adjusted R Squared = .006)
Table 9: Mean values of respondents’ internal efficacy by experimental group moderated by political knowledge Dependent Variable: internal political efficacy
comment category political knowledge Mean Std. Deviation N mixed low-medium pol_know 3.7222 1.02422 18 medium-high pol know 4.8667 .83705 35 Total 4.4780 1.04903 53 negative low-medium pol_know 3.7692 1.33600 13 medium-high pol know 4.9091 1.17341 33 Total 4.5870 1.31315 46 no comments low-medium pol_know 3.5208 1.06088 16 medium-high pol know 4.5517 .97295 29 Total 4.1852 1.11136 45 positive low-medium pol_know 4.7222 .77220 6 medium-high pol know 4.4537 1.27280 36 Total 4.4921 1.21025 42 Total low-medium pol_know 3.7862 1.12298 53 medium-high pol know 4.6967 1.08824 133 Total 4.4373 1.17018 186
Table 10: UNIANOVA model for internal efficacy by comment category
moderated by political knowledge
Dependent Variable: internal political efficacy
Source Type III
Sum of Squares df Mean Square F Sig. Corrected Model 43.119a 7 6.160 5.216 .000 Intercept 2467.651 1 2467.651 2089.590 .000 Comment_categories 4.537 3 1.512 1.281 .283 political_knowledge 19.227 1 19.227 16.281 .000
Comment_categories *
political_knowledge 8.439 3 2.813 2.382 .071
Error 210.205 178 1.181
Total 3915.556 186
Corrected Total 253.324 185 a. R Squared = .170 (Adjusted R Squared = .138)
Figure 2: Mean interaction plot of comment category by internal efficacy moderated
by political knowledge
Table 11: Mean values of respondents’ external efficacy by experimental group moderated by political knowledge Dependent Variable: external political efficacy
comment category
political knowledge Mean Std. Deviation
N
mixed low-medium
medium-high pol know 3.9571 .89419 35 Total 3.8585 .90606 53 negative low-medium pol_know 3.4615 1.47848 13 medium-high pol know 3.9394 1.26712 33 Total 3.8043 1.33116 46 no comments low-medium pol_know 3.2813 1.18278 16 medium-high pol know 3.5172 1.15328 29 Total 3.4333 1.15601 45 positive low-medium pol_know 3.5833 1.46344 6 medium-high pol know 3.8611 1.19290 36 Total 3.8214 1.21886 42 Total low-medium pol_know 3.4906 1.19087 53 medium-high pol know 3.8308 1.13299 133 Total 3.7339 1.15683 186
Table 12: UNIANOVA model for external efficacy by comment category
moderated by political knowledge
Dependent Variable: external political efficacy
Source Type III
Sum of Squares df Mean Square F Sig. Corrected Model 9.542a 7 1.363 1.019 .419 Intercept 1774.300 1 1774.300 1326.803 .000 Comment_categories 4.038 3 1.346 1.007 .391 political_knowledge 3.405 1 3.405 2.546 .112 Comment_categories * political_knowledge .325 3 .108 .081 .970 Error 238.035 178 1.337 Total 2840.750 186 Corrected Total 247.577 185 a. R Squared = .039 (Adjusted R Squared = .001)
Figere 3: Mean interaction plot of comment category by external efficacy moderated
by political knowledge.
Table 13: Mean values of respondents’ interpersonal trust by experimental group Dependent Variable: interpersonal trust
comment category Mean Std. Deviation N mixed 4.6604 1.25871 53 negative 5.2065 .95787 46 no comments 4.5814 1.28142 43 positive 5.0732 .85558 41 Total 4.8716 1.13567 183
Table 14: UNIANOVA model for interpersonal trust by comment category
Dependent Variable: interpersonal trust
Source Type III Sum
of Squares df Mean Square F Sig. Corrected Model 12.812a 3 4.271 3.445 .018 Intercept 4318.121 1 4318.121 3482.976 .000 Comment_categories 12.812 3 4.271 3.445 .018 Error 221.920 179 1.240 Total 4577.750 183 Corrected Total 234.732 182 a. R Squared = .055 (Adjusted R Squared = .039)
Table 15: Post-hoc test for direct effect of comment category on
interpersonal trust
Dependent Variable: interpersonal trust LSD (I) comment category (J) comment category Mean Differen ce (I-J) Std. Error Sig. Upper Bound mixed negative -.5461* .22437 .016 -.1034 no comments .0790 .22853 .730 .5299 positive -.4128 .23158 .076 .0442 negative mixed .5461* .22437 .016 .9889 no comments .6251* .23619 .009 1.0912 positive .1334 .23914 .578 .6053 no comments mixed -.0790 .22853 .730 .3720 negative -.6251* .23619 .009 -.1591 positive -.4918* .24304 .045 -.0122 positive mixed .4128 .23158 .076 .8698 negative -.1334 .23914 .578 .3386 no comments .4918* .24304 .045 .9714
Based on observed means.
The error term is Mean Square(Error) = 1.240. *. The mean difference is significant at the .05 level
Table 16: Mean values of respondents’ interpersonal trust by experimental group moderated by political knowledge
Dependent Variable: interpersonal trust comment category political knowledge Mean Std. Deviation N mixed low-medium pol_know 4.3056 1.38414 18 medium-high pol know 4.8429 1.16803 35 Total 4.6604 1.25871 53 negative low-medium pol_know 5.1538 1.02844 13 medium-high pol know 5.2273 .94448 33 Total 5.2065 .95787 46 no comments low-medium pol_know 4.4688 1.35976 16 medium-high pol know 4.6481 1.25434 27 Total 4.5814 1.28142 43 positive low-medium pol_know 4.3333 .81650 6 medium-high pol know 5.2000 .80623 35 Total 5.0732 .85558 41 Total low-medium pol_know 4.5660 1.25972 53 medium-high pol know 4.9962 1.06111 130 Total 4.8716 1.13567 183
Table 17: UNIANOVA model for interpersonal trust by comment category
moderated by political knowledge
Dependent Variable: interpersonal trust Source Type III Sum of
Squares df Mean Square F Sig. Corrected Model 20.464a 7 2.923 2.388 .023 Intercept 2998.580 1 2998.580 2449.042 .000 Comment_categori es 10.186 3 3.395 2.773 .043
political_knowledg e 5.647 1 5.647 4.612 .033 Comment_categori es * political_knowledg e 2.781 3 .927 .757 .520 Error 214.268 175 1.224 Total 4577.750 183 Corrected Total 234.732 182 a. R Squared = .087 (Adjusted R Squared = .051)
Figure 4: Mean interaction plot of comment category by interpersonal trust
Appendix B Survey questions
Note: including only those questions that were used in our analysis
1 Wat is uw geslacht?
Man / Woman
2 Wat is uw leeftijd in jaren? (Gelieve hier alleen cijfers in te vullen) 3 Wat is uw hoogste afgeronde opleiding?
1 Basisschool 2 VMBO/MAVO/LBO/VBO 3 HAVO 4 Atheneum/Gymnasium/VWO 5 MBO/MTS 6 HBO/HTS/HEAO 7 WO (Bachelor) 8 WO (Master) 96 Anders, namelijk
4 Wat is de naam van de huidige Nederlandse minister voor Buitenlandse Zaken? 1 Maxime Verhagen
2 Uri Rosenthal 3 Frans Timmermans 4 Jaap de Hoop Scheffer 5 Bert Koenders
6 Weet niet
- Hoe lang is normaal gesproken de periode tussen (twee opvolgende) Tweede Kamerverkiezingen in Nederland?
1. 3 jaar 2. 4 jaar 3. 5 jaar 4. 6 jaar 5. 8 jaar 6. weet niet
5 Wie was de afgelopen 2 jaar voorzitter van het Europees Parlement? 1. Javier Solana
2. Martin Schulz 3. Nicolas Sarkozy 4. Hans-Gert Pöttering 5. José Manuel Barroso 6. Weet niet
6.4 Controle group (no comments):
7 Cynicsme (all coded from 1 to 7 on Likert scale: 1 = totally agree 7 totalle disagree) 7.1 Politici beloven bewust meer dan dat ze waar kunnen maken
7.2 Ministers en staatssecretarissen zijn voornamelijk in zichzelf geïnteresseerd. 7.3 Politieke partijen zijn alleen maar geïnteresseerd in mijn stem tijdens verkiezingen, niet in mijn mening.
7.4 Om lid te worden van het Nederlands parlement, zijn vrienden van een grotere invloed dan bekwaamheid.
7.5 Politici begrijpen niet wat er als belangrijk geacht wordt door de samenleving 7.6 Politici zijn in staat om belangrijke kwesties op te lossen
7.7 De meeste politici zijn competente personen die weten waar ze mee bezig zijn. 7.8 In dit land hebben een paar individuen een grote politieke macht, terwijl de rest van de mensen heel weinig macht hebben.
7.9 Politici vergeten snel de wensen van de burgers die op ze gestemd hebben.
8 Political efficacy (all coded from 1 to 7 on Likert scale: 1 = totally agree 7 totalle disagree)
8 Ik weet meer over de politiek dan de meeste mensen van mijn leeftijd. 9 Wanneer er politieke kwesties worden besproken, dan wil ik daar normaal
gesproken over meepraten.
10 Ik ben in staat om de meeste politieke kwesties makkelijk te begrijpen. 11 De overheid geeft om de mening van burgers betreffende nieuwe wetten 12 De overheid doet haar best om uit te zoeken wat gewone mensen willen.
9 Interpersonal trust (all coded from 1 to 7 on Likert scale: 1 = totally agree 7 totalle disagree)
13 Over het algemeen denk ik dat de meeste mensen te vertrouwen zijn
14 De meeste mensen zouden proberen misbruik van mij te maken als ze daartoe de kans kregen.
Appendix C
Checklist for ethical permission for research conducted by students in the framework of education
1) Title of the research project:
Spiral of Cynicism 2.0: The Effects Of Negative User Comments on Citizens’ Political Attitudes
2) Component of the programme (Bachelor’s or Master’s/Name of module): Political Communication (Master Track)
3) Student(s) who will conduct the research: Milan van Ooijen
4) Teacher(s) who will supervise the research: Magdalena Wojcieszak
5) Brief description of the research (max. 200 words) A web-based survey-experiment
6) Research method (max. 100 words):
After a pre-test questionnaire, participants were exposed to one of four manipulation groups, after which participants were shown a proxy online news item with either negative, positive, mixed, or no comments next to it. After observing the comments and news item participants took part in the post-test questionnaire to test for results on the primary dependent variable political cynicism and two secondary DV’s: political efficacy and interpersonal trust. In the pre-test questionnaire respondents were asked three questions which tested their knowledge of the moderator ‘political knowledge’
7) Where will the research be conducted (e.g. online, location, through organisation): online
8) Duration of the research (from the start of recruitment until the close of data collection):
About 10 minutes
9) Who are the participants? How will they be recruited?
A snowball method was initiated, i.e. after initially sending the survey link to
acquaintances and colleagues, multiple layers of friends and acquaintances of this first round of respondents were reached by resharing the survey link. Amounting to a total of 306 respondents
10) Are all participants adult (18 years or older), mentally competent individuals? O no
X yes
If no, explain how active or passive permission will be obtained from the parents.
11) Number of participants to be recruited: The goal was 200 respondents
12) How will the anonymity and privacy of the participants be guaranteed? Explain. By using qualtrics as our survey tool, we safeguarded the privacy of all the
participants involved
13) Will participants receive compensation for participating in the research? X no
O yes If yes, explain.
14) Will any misleading occur? X no
O yes
If yes, explain how and why. Also explain how and when participants will be debriefed.
15) Is there a possibility that some participants/test subjects might consider the
research unpleasant or troublesome for any reason, or that they may be exposed to information, materials or questions to which they would prefer not to be exposed? X no
O yes If yes, explain.
Signature of student(s) conducting the research: Signature of teacher(s) supervising the research:
Date: Date:
