€¦ · Web viewThis is in spite of a real word trend indicating that the ... John P, Mark R Levy, Dennis K Davis, William Gill Woodall, Michael Gurevitch, and Haluk Sahin

The impact of multiple media on public opinion towards the European Union

Leonardo Baccini, London School of Economics and Political Science

Laura Sudulich, Université Libre de Bruxelles

Today’s media environment not only is all-pervasive, but also extremely varied. Citizens gather politically relevant information from a multiplicity of sources and through different media. This paper address whether consuming online information about the European Union is responsible for variation in public opinion towards it and whether the medium effect bypasses or encompasses political knowledge. We assess the effects of online news consumption on European matters by addressing the medium effect as well as the effects of different types of online platforms. Using data from from Eurobarometer 76.3,we find that consuming information online has an effect on opinions, mediated by an increase in political knowledge The effect is positive, suggesting that online information consumption fosters more support for the EU by enhancing political knowledge. When we set apart those who trust the EU from those whom do not we find asymmetrical reinforcement effects. While for the former group we observe reinforcement effects, for the latter there is no increase in negativity, rather in increment in positive attitudes. When we explore differences among websites, we find that institutions official website as well as the online version of traditional media positively affect opinions by enhancing political awareness. Websites that by default are mixture of fact-checked and non fact-check information prove to be unable to do so.

1

Introduction The normative claim that an informed citizenry is fundamental to the well-being of democracy is rather uncontested. The information environment is vital to citizens’ capacity to learn about politics, their ability to link political preferences to parties and policies and assess the performances of institution and political actors. In most instances, people gather politically relevant information via the mass media rather than though direct interactions with political elites. The role of the mass media is therefore key to the correct functioning of society (Castells 2000; Delli Carpini and Keeter 1996). There is no shortage of evidence showing that the mass media exert a crucial influence on public opinion formation [from Van Klingeren, Boomgaarden, and De Vreese (2013); to Zaller (1992); for a review of major contribution see Bennett and Iyengar (2008)]. The mass media are particularly important when considered in relation to the formation of opinion, the gathering of political awareness and, voting behaviour in the context of the European Union, since citizens typically do not experience direct contacts with European institutions. Additionally, the widespread ignorance of the European Union (hereafter EU) politics and policies among members of the European public is notorious and persistent. Over one third of Europeans are still unable to name any EU institution; the percentage of citizens claiming to know a little or nothing at all about ‘the people who run the various EU institutions and the leaders of the EU’ is a striking 73%. A similarly high number of subjects (74%) reports to know a little or nothing at all about ‘The allocation of roles played by

2

the various institutions (who does what?)’ (Eurobarometer 77.4). This ignorance about EU institutions and mechanisms goes hand in hand with the lack of direct interactions between citizens and EU institutions. Thus, citizens forcefully rely on mass media when (in)forming their opinions about the EU. While the impact of radio, television and newspapers - the so-called traditional media - has been extensively debated in relation to electoral behaviour and attitudes formation towards the EU (De Vreese 2003; De Vreese and Boomgaarden 2006; Schuck and De Vreese 2006), the influence of the Internet_ is currently under-explored.1 This is in spite of a real word trend indicating that the Internet is rapidly becoming a fundamental source of information about European matters. Between November 2011 and December 2012, the percentage of citizens reporting that they had gathered information on the European Parliament on the World Wide Web went up by 10%, from 33% to 43% (Eurobarometer 78.2). The Internet provides a potentially unlimited amount of information. It also offers an array of heterogeneous sources, from credible news producers to users generated content, whose quality tends to be lower (Patterson 2013). These structural differences with respect to traditional media are likely to affect the opinions and behaviours of those who have integrated online news consumption into their media usage habits. The nature of online-based information may also affect the process through which individuals translate information into political knowledge. Lupia and McCubbins (1999, 25) point at a key element of the relationship between information and knowledge: ‘although you cannot have knowledge without having information, you can have information without having knowledge’. Information that does not 1 An exception is De Wilde, et al. 2013De Wilde, et al. 2013De Wilde, et al. 2013, who content analyse expressions of Euroscepticism in online media across 12 member states in the 2009 European Parliament campaign. However, their analysis does not deal with the consequences of online-based information and communication for public opinion.

3

provide knowledge is either redundant or it’s actually noise. While absorbing redundant information is a possibility when gathering information via any media, we argue that the internet maximises potential for amplification of noise by creating an environment where noise and signal are often indistinguishable. This paper takes the first step to addressing whether consuming online information about the EU is responsible for variation in public opinion towards it and whether the medium effect bypasses or encompasses political knowledge. Further to this, we explore whether - within the range of informational opportunities offered by the World Wide Web - different online loci of information affect public opinion and if the process is mediated by political knowledge. We use of data from Eurobarometer 76.3 (November 2011) posing a large battery of questions on media use for political information consumption based on a Europe wide representative sample of individuals. Importantly, this survey contains a unique array of items on online habits of political news consumption.

We find that consuming information online has an effect on opinions, mediated by an increase in political knowledge. The effect is positive, indicating that online information consumption fosters more support for the EU by enhancing political knowledge. When we set apart those who trust the EU from those whom do not we find asymmetrical reinforcement effects. While for the former group we observe reinforcement effects, for the latter there is no increase in negativity, rather in increment in positive attitudes. When we explore differences among websites, we find that institutions official website as well as the online version of traditional media positively affect opinions by enhancing political awareness. Websites that by default are mixture of fact-checked and non fact-check information prove to be unable to do so.

4

The article proceeds as follows: in the next section we outline the relationship between mediated information political knowledge and public opinion on which we base our working hypotheses. We then discuss our empirical strategy and describe the data. In the following section we present and discuss the results of our analysis. We then run a further test to account for the endogeneity intrinsically related to the relationship between media usage and public opinion. We conclude by examining the implications of our findings.

Information, knowledge and public opinion

Information is the data that allows individuals to acquire (politically) relevant knowledge and to form or redefine their beliefs. It can be gathered by direct experience –attending a candidate debate, correspondence with an MP – or by being exposed to reports. These reports can be based on the experience of people in one’s social network – e.g. a friend telling his/her experience in dealing with local government – or, can be provided by mass media. Acquiring information through direct experience of EU institutions and policies is substantially ruled out. The same applies for information obtained via personal networks as ‘very few citizens have first- or even second-hand contact with Community affairs in Brussels’ (Dalton and Duval 1986, 186). Political knowledge is the state of awareness of facts that matter to orient people’s opinions and choices. The acquisition of knowledge depends upon the availability of information, so does the redefinition of what Bartels (1993) calls ‘fund of knowledge’. Newly acquired information allows individuals to update their preexisting knowledge - for those who had some - and provides knowledge to those who had none before. The mass media are therefore key to educating the public by making the information available and accessible (Holtz-Bacha and Norris 2001). Information acquired via

5

the mass media can then translate into in-depth knowledge, superficial acquaintance or even result into no knowledge as a function of being, unclear, noisy or redundant.

Medium effects?

A substantial corpus of studies has unveiled differences among the mass media in their capacity of translating information into knowledge that subsequently affects political evaluations [for a review see Norris and Sanders 2003]. While several studies support the idea of print superiority (Robinson and Davis 1990; Robinson et al. 1986) others cast some doubts on it (Graber 2001; Mondak 1995), whit no ultimate consensus on what medium carries the highest learning potential for the public. Studies of traditional media share the implicit assumption that the information provided is both relevant and factual. Media publishers and regulatory authorities act as gatekeepers on what can be broadcast and printed, therefore making this assumption relatively safe when applied to traditional media. On the other hand, the Internet remains largely unregulated. The amount and type of information available online differ substantially from what is available via other media, firstly in terms of heterogeneity of content, secondly with regard to types of sources, and thirdly in relation to the amount of noise associated with online-based information. Given the radically different nature of the Internet and online-based news consumption, there are good reasons to explore whether the medium per se makes a difference. The structure of web pages – characterized by hyperlinks, menu bars, and audio-visual content – affects learning (Tewksbury and Rittenberg 2012). While many scholars consider differences in the effect of mass media as due to content more than to medium’s characteristics as such (Newton 1999; Norris and Sanders 2003) experimental studies have proved some

6

forms of medium effects (Norris and Sanders 2003; Robinson and Davis 1990; Robinson et al. 1986). In particular, Kaid and Postelnicu in a study of public perception of candidates based on TV and Internet supplied stimuli (2005: 273) found that ‘the channel of communication has a strong impact on how the audience interprets the message an builds its perception of political candidates’. Therefore we first test whether using the Internet per se has an independent effect on attitudes and whether this explained through an increase in levels of political knowledge in our first working hypothesis:

H1: Those who use the Internet as main source of political newsgathering display significantly different opinions on the EU to those who do not, and this process is mediated by political knowledge.

This will shed some light on whether the channel though which information is accessed affects political attitudes and if that happens via an increase of political knowledge. Further to this, we address whether the pull-in nature of the medium conditions the process by reinforcing pre-existing preferences. The internet is particularly well suited for preference reinforcement (Iyengar and Hahn 2009; Nie et al. 2010) facilitating selective exposure to information already in line with one’s preferences. While selectivity is clearly also happening with regard to traditional media, it has been argued that online based news consumption holds the potential to maximize selectivity by creating ‘filter bubbles’ Therefore, we explore whether predisposition towards the EU condition the process (in)forming attitudes via the web. Simply put, we explore whether reinforcement effects occur.

H2: Europhiles who use the Internet as main source of political newsgathering display significantly more positive attitudes than their counterparts who do not. Eurosceptics who use the Internet as main

7

source of political newsgathering display significantly more negative attitudes than their counterparts who do not. The reinforcement process is mediated by increases in political knowledge.

This initial aggregate analysis, while key to unveil whether medium specific effects exist, is likely to mask certain nuances of the process. The Internet, as much as any other media is not monolithic in nature. When differences between quality newspapers and tabloids are explored, the extent to which the printed press contributes to knowledge is showed to be differential (Holtz-Bacha and Norris 2001). In a similar vein, studies of the effects of television have proved that public and commercial channels differ in their capacity of transmitting knowledge and affecting political opinions and behaviours (Aarts and Semetko 2003; Norris and Sanders 2003). Equally, to gather a fuller understanding of the mechanism linking online-based information to citizens attitudes towards the EU and to unveil the effects of single platforms we need to undertake an addition analytical step.

Platform effects?

One of McLuhan’s (1964) insights, in formulating the famous the medium is the message theory, was that “the content of any medium is always another medium”. The Internet essentially contains all the other mass media that preceded it: the printed press, TV and radio. All major new outlets have now an online version. Additionally, there are many other online spaces that deliver politically relevant information: institutions official websites, social media sites - including social networking sites like Facebook and Twitter as well as sharing platforms like Wikis and Youtube - and weblogs. In the words of Patterson “the internet is at once a gold mine of solid content and a hellhole of misinformation” (2013: 79)

8

Online information can therefore take the most diverse shapes, depending on the source and hosting platform: from credible news producers to unverifiable information posted by individuals simply voicing their own opinions. Online spaces vary in the heterogeneity of their content, in the extent to which they host unverified versus verified facts and in the amount of space they give to comments and opinions. For instance, the website of a major newspaper is likely to publish content that is verified and has been provided by accredited sources. At the other end of the spectrum, platforms like blogs and forums are, by definition, aggregators of comments and opinions, often seeking to advocate more than to report (Scott 2007). Therefore, we expect the effects of information encountered on these websites to be different from the effects of information encountered on ‘more reliable’ platforms. Particularly, platforms that are likely to supply a mixture of noise and news, like Social Networking Sites, video hosting spaces and blogs - are more likely to fail translating information into knowledge. Empirically, we separately analyze the effects of platform that hold strong potential for the amplification of noise from those that are most likely to maximize reliable information, to gather a more fine-grained understanding of how the new media affect individuals’ opinions.

H3: Different online loci exert differential effects on public opinion on the EU. Websites promoting reliable fact-checked information affect users’ opinions by enhancing levels of political knowledge.

Empirical strategy

As a result of our theoretical framework, we implement a causal mediation analysis, which allows us to distinguish the direct effect of

9

online-based newsgathering from its mediated effect through an increase in the knowledge of the EU. In particular, the causal mediation analysis allows exploration of the role of an intermediate variable that lies along the causal paths between the treatment and the dependent variable (Hicks and Tingley 2011; Imai, Keele, and Yamamoto 2010). In our case, knowledge about the EU is the mediator that lies along the causal path between information gathered online and attitudes towards the EU. The path mode is illustrated in Figure 1.2

[Figure 1 about here]

To carry out the causal mediation analysis, we rely on the STATA 13 ‘mediation’ package developed by Hicks and Tingley (2011). The model for both the mediator variable and the outcome variable is an Ordinary Least Squares (OLS) regression. We include country fixed effects to account for cross-country heterogeneity and mitigate omitted variable problems. We use robust standard errors and we run 1000 simulations for the quasi-Bayesian approximation of parameter uncertainty.To balance out differences between those who used the internet for political newsgathering and those who did not, we use entropy balancing (Hainmueller 2012).3 Specifically, we balance our entire set of covariates with respect to our treatments, which capture those respondents who go online as well as the type of website they visit.4 Differences in the means between the treated and control group vanish for all the covariates after implementing entropy balancing

2 The literature on causal mediation analysis is large and fast-growing. For pioneering statistical studies, see Baron and Kenny (1986) and Robins and Greenland (1992).3 We use the STATA 13 package ‘ebalance’ (Hainmueller, 2013).4 The description of the covariates is provided in the following section.

10

(pre- and post-matching descriptive statistics for the relevant variables are reported in Tables 1, 2, and 3). We note that balancing covariates with respect to the treatment is similar to controlling for such confounding factors in a standard multivariate regression without imposing parametric functional form or distributional assumptions. Finally, we run all our parametric models using the weights obtained from the entropy balance estimation as well as the entire set of control variables to account for any residual differences between the treated and control groups.5 An advantage of entropy balancing over a matching technique is that the former technique does not drop unmatched observations (Hainmueller 2012: 2).

[Tables 1, 2, and 3 about here]

DataWe use of data from Eurobarometer 76.3, which among others, contains a large battery of items on media use for political information on EU level matters. This enables us to investigate the effects of new media and to dig deeper into the effects of different online news sources. We exclude observations from countries that were not members of the EU at time of data collection (November 2011).

Outcome variableThe outcome variable captures respondents’ attitudes towards the EU. The question on which we build our dependent variable was posed as follows:

“In general, does the EU conjure up for you a very positive, fairly positive, neutral, fairly negative or very negative image?”

5 Residual differences might come from the variance and skewness of the covariates.

11

The resulting variable ranges from 0 to 4 where 0 indicates (very negative) and 4 (very positive). As said, we rely on an OLS with country fixed effects and robust standard errors. We are unable to run an ordered logit (or probit), which is not supported by the ‘mediation’ package. However, if we run a simple ordered probit as outcome variable, the results (not reported) are very similar to the ones from the OLS regression. This makes us confident of the reliability of the estimation even when we implement the causal mediation analysis.

Treatment variablesWith regard to the part of the analysis where we test the aggregate effects of the Internet as such, we employ a variable that scores 1 if respondents use the Internet as their first source of information gathering on European affairs and 0 otherwise. In the analysis concerned with the effects of different types of informational sources online we employ the following: A variable that scores 1 if respondents use exclusively ‘institutional

and official websites (governmental websites, etc.)’ to gather information on European matters, and 0 otherwise;

A variable that scores 1 if respondents use exclusively ‘information websites (websites from newspapers, news magazines, etc.)’ to gather information on European matters, and 0 otherwise. We label this variable Traditional media websites

A variable that scores 1 if respondents use exclusively ‘online social networks’ to gather information on European matters, and 0 otherwise;

A variable that scores 1 if respondents use exclusively ‘blogs’ to gather information on European matters, and 0 otherwise.

Figures 2 and 3 show the distribution of our treatments by country. The take-home messages from these figure is that at the time of fieldwork the large majority of respondents, in every country, used

12

television as first source of newsgathering on EU matters. The Internet was widely preferred to the printed press in most Easter member states. When we look at the distribution of web information sources, we clearly see that the websites of traditional media outlets were the most visited platforms in every country.

[Figures 2 and 3 about here]

MediatorFor the mediating variable we use an index of objective political knowledge on the EU, compiled by the Eurobarometer and available in the dataset. It ranges from 0 to 2 (bad, average, good) and it is based on answer to the following standard questions:“For each of the following statements about the EU could you please tell me whether you think it is true or false:

The EU currently consists of 27 Member States The members of the European Parliament are directly

elected by the citizens of each Member State Switzerland is a member of the EU”

In line with the estimation strategy implemented with the outcome variable we run an OLS with country fixed effects and robust standard errors. Results (not reported) from a simple ordered probit are very similar to the results from a simple OLS regression with Knowledge of the Treaty as outcome variable.

Control Variables We control for Socio Economic Status of respondents (gender, age, working status) and other factors that are likely to affect both media usage choice and political opinions towards the EU. Namely, we include controls for how frequently respondents discuss both national and European political matters with friends and relatives (frequently, occasionally, never). We then control for consumption of traditional

13

media as our goal is to unveil medium effects, if any. Moreover, we include standard indicators (Gabel and Palmer 1995; Van Klingeren, Boomgaarden, and De Vreese 2013) of subjective perception of the national and European economies, as citizens tends to blame the EU if they believe the economy – both national and European – is performing poorly (Kritzinger 2003) and a measures of respondents ‘attitudes towards immigrants. Previous studies indicate that attitudes toward immigration are strongly correlated to support for the EU. Those who are negatively disposed towards out-group tend to oppose free circulation of citizens and the opening of national borders within Europe (De Vreese and Boomgaarden 2005). Finally, the variable in use to test the presence of reinforcement effects is based on a question that asks respondents whether they “tend to trust” (coded as 1) or not (coded as 0) the EU. Table 4 reports the descriptive statistics.

[Table 4 about here]

Empirical results

We start our analysis by running a model, where out treatment variable – Internet – sets apart those who use the Internet from the reference category (see Model 1 in Table 5) composed by respondents who acquire information mostly through traditional media as well as by respondents who do not consume any media. The coefficient of Internet is positive and statistically significant both in the mediator equation and in the outcome equation. Importantly, the Average Causal Mediation Effect (ACME) is positive, statistically significant, and accounts for 17 percent of the total effect. Figure 4 shows the average causal mediation effect (ACME) as a function of degree of violation of the sequential ignorability assumption. The

14

results show that the point estimate of the ACME would be 0 if the correlation between the error term of the mediation equation and the error term of the outcome equation is about 0.25. Concretely, this indicates that an omitted variable must explain a large part of the remaining variance in both the mediator and outcome equation in order for the ACME to be zero.

This finding provides a first validation of HP1. It implies that respondents who use the Internet as their first source of information gathering on European affairs have different attitudes – specifically more positive - towards the EU compared to respondents who do not use the Internet as their first source of information. Such a positive effect comes from two channels: (1) an increase in knowledge of the EU, which, in turn, impacts positively on attitudes towards it (i.e. mediated effect); (2) a direct effect of using the Internet, which increases the general level on information on the EU available to respondents.

[Table 5 and Figure 4 about here]

We then include dummy variables to control for the effects of traditional media (Model 2). The coefficient of Internet remains positive and statistically significant in both the mediation equation and the outcome equation. Importantly, the effect of the Internet on knowledge and attitudes towards the EU is larger than the effect of TV and radio and roughly the same as the one accounting for the printed press. This suggests that the medium as such not only matters but challenges also the print superiority. The ACME is again positive, statistically significant, and accounts for 12 percent of the total effect. Figure 5 shows the average causal mediation effect (ACME) as a function of degree of violation of the sequential ignorability

15

assumption. Furthermore, we re-run our analysis splitting the sample in respondents who trust the EU and respondents who do not trust the EU to test H2. With regard to Europhiles, Model 3 shows that the effect of the Internet not only is positive (and statistically significant) but also (significantly) larger than the effect of traditional media on both knowledge of and attitudes towards the EU. Regarding Euroskeptics, the effect of Internet remains positive and statistically significant, though it is roughly the same than the other media. Importantly, in both Model 3 and Model 4 ACME is positive, statistically significant, and accounts for respectively 12 percent and 4 percent of the total effect. Thus, even among the Euroskeptic group, individuals who use the internet as their first source of information gathering on European affairs tend to have a better knowledge of the EU, and in turn, to have a more positive attitudes toward the EU. This clearly runs against reinforcement effects. While for one group there is evidence of online newsgathering strengthening predispositions, for the other group the effects show the opposite pattern. The asymmetry of these results shows the complexity of capturing the effects of the Internet. (Pariser 2011). Browsing the web for information holds a strong potential for reinforcement as well as an equally strong potential for accidental exposure to unanticipated content -including preference challenging information- (Tewksbury and Rittenberg 2012) and by-product unintended learning (Chadwick 2012) that may result in lack or weakening of reinforcement effects. Respondents who use the Internet as their first source of information gathering on European affairs tend to have different attitudes towards the EU from those who do not. Such a difference in attitudes towards the EU is unidirectional: the Internet increases the probability of having a better opinion of the EU. This effect arises through the knowledge channel as well as the information channel. This may be a

16

function of traditional media devoting low coverage to European affairs while online environments provide larger volumes of information. The printed press, TV and radio dispose of limited amount of space and the ratio of national/European affairs is disproportionally in favor of the latter. These same new producers in their online versions dispose of an endless amount of space; so do other types of platforms and we now turn to the test of H3.

Table 6 reports the results for the mediation analysis in relation to each treatment capturing a specific type of platform. Note that we include dummies for traditional media in every model. Results are clear cut: only some types of online loci have a mediated positive effect on knowledge and a direct positive effect on attitudes toward the EU. Specifically, Traditional media websites and especially institutional and official websites are the ones driving our results, whereas blogs and online social networks have no impact on knowledge (mediated effect) and attitudes toward the EU (direct effect) as shown in Models 7 and 8.6 Importantly, ACME is positive and statistically significant in Models 5 and 6. The average causal mediation effect accounts for 14 percent of the total effect in Model 5 and for 15 percent of the total effect in Model 6. Interestingly, once we control for online platforms, the significance of dummies capturing traditional media fades away.


In sum, we find strong support for the HP3. Only fact-checked websites increase respondent’s knowledge of the EU and, in turn, knowledge has a positive mediated effect on attitudes toward the EU.

6 In the next section we show that these findings are not affected by reverse causality and/or selection into the treatment.

17

Moreover, fact-checked websites have a positive direct effect on attitudes toward the EU, in addition to the effects through the knowledge channel. On the contrary, websites providing a large space for less reliable information and personal opinions have effect neither on knowledge of the EU nor on attitudes towards it. Our results provide some indications that the Internet is replacing traditional media as the main instrument to acquire information on EU affairs and that traditional outlets and institutions website hold great potential for transmission not only of information but also of political knowledge. This results in public opinion turning more favorable towards Europe.

Addressing endogeneity The previous section has showed that online newsgathering has a positive effect on both knowledge of the EU and attitudes towards it. A line of objection to these findings is that the selection into our treatments is not random. In other words, it would be plausible to argue that respondents who are knowledgeable about the EU are also more likely to go online to search information on it. Similarly, it could be claimed that respondents who have a positive (negative) attitudes toward the EU are more likely to visit pro-EU (anti-EU) websites, which in turn reinforce their priors. Such concerns are sound. However, note that the Internet increases knowledge of the EU and improves the attitudes toward the EU also for the Euroskeptics sub-sample, which does not experience a reinforcing effect. Therefore, this is preliminary evidence that our findings cannot be completely explained by the selection into the treatment. To further corroborate our results, we implement an instrumental variable approach for these treatments that were statistically significant in the mediation analysis.Our instrument is a dummy that scores one if respondents have

18

access to the Internet either at home or at work (labeled Online Access). Respondents were asked whether they had access to the Internet at home or at work. The variable in use thus takes the value of 0 when respondents said they had neither access at work nor at home. The idea behind this identification strategy is that without access to the internet at home or at work respondents are less likely to go online to gather information on the EU. Indeed, Internet and Online Access are highly correlated, i.e. ρ=0.66 (p-value=0.00). Similarly, the correlation between Online Access and Information and Official websites is respectively 0.67 (p-value=0.00) and 0.54 (p-value=0.00). Clearly, respondents without online access at home or at work can still browse the web though their smartphones, and this explains why we do not find a perfect correlation between our instrument and our treatments. The problem of relying on Online Access is that it could be correlated with our dependent variables, i.e. knowledge of and attitudes toward the EU. For instance, if respondents with online access are geographically clustered, our instrument could be correlated with socio-economic variables (i.e. urban or rural place of residence), which consequently affect our outcome variables. Since we control for socio-economic status, the exclusion restriction should still hold. Moreover, acknowledging that our instrument is not randomly assigned, we balance our entire set of covariates with respect to Online Access to create a counter-factual (i.e. respondents without online access), which is as close as possible as to our treated units, i.e. respondents with online access. Table 7 shows the distribution of the covariates before and after balancing. Then we use the weights obtained from entropy balance in our instrumental variable estimations.


19

Armed with this identification strategy, we re-run our main models with both knowledge of the EU and attitudes towards it as dependent variables. We begin with knowledge of the EU. Results are reported in Table 8 and are in line with the ones showed in the previous section. Specifically, the coefficient of Internet is always positive and statistically significant in the whole sample as well as in the Europhiles sub-sample and Eurosceptics sub-sample. Similarly, Official websites and Information websites have a positive and statistically impact on knowledge of the EU. Importantly, the effect of the Internet is always larger than the effect of traditional media. Furthermore, all the diagnostics show that our instrumental approach is robust. In particular, orthogonality conditions are valid (see the Anderson-Rubin Wald test) and our instrument is sufficiently strong (see the Cragg-Donald Wald F statistics).

[Tables 8 and 9 about here]

Table 9 shows results on attitudes towards the EU and indicates that our previous results are all validated. Indeed, the coefficient of Internet is always positive and statistically significant in Models 14, 15, and 16, though the significance is weaker in the last one. Similarly, Official websites and Information websites always strengthen positive opinions on the EU (see Models 17 and 18). Even in this case the Internet has a stronger impact on the dependent variable than traditional media, except for the Eurosceptic sub-sample. Finally, we note that all the diagnostics confirm the validity of our identification strategy.

Conclusions The news media serve the purpose of supplying the information that

20

gives citizens awareness of political elites’ positions and policies and, ultimately, offers them guidance when making politically relevant choices. Traditional media dedicate relatively little attention to European matters, particularly in routine periods (Peter and De Vreese, 2004). While the emergence of a European public sphere is highly desirable, we have not yet seen that happening. A global, transnational, multilingual medium like the Internet carries the potential to supply all the information Europeans need and often do not find on traditional media. On the other hand, the Internet also contains large amounts of noise. Our study addresses the effects of the Internet as medium of political knowledge gathering and its effects on public opinion. Our findings indicate that new media exert an effect on public attitudes towards the EU. Those who search for information on European matters mainly online are more positive towards the EU, and this happens via increases in levels of knowledge of the Union. In other words, citizens who use the Internet as their main channel of political newsgathering are more knowledgeable about the EU and this affects their views, which are significantly more positive. The medium per se has therefore an effect on both levels of knowledge (marking an increase) and attitudes (more positive). This effect does not fade away when we include specific controls for usage of traditional media; therefore the Internet is effectively independently from and in addition to them. While it maximizes opportunities for self-selection of source in line with one’s preferences, our results show that reinforcement effects are not symmetric. Those who trust the EU are more positively disposed towards it as a result of consuming information online. On the other hand those who regard the EU as untrustworthy do not experience a reinforcement of their negativity towards Europe by browsing the web. While the puzzle of new media effects remains very much open to debate between studies providing support for the reinforcing

21

theory and studies challenging it, our contribution indicates that reinforcement per se does not suffice as an overarching explanation. Most of the debate about reinforcement versus conversion assumes symmetry in the effects, while our results indicates this is not necessary the case and the effects of new media are characterized by high degrees of complexity. Overall, gathering news on European matters on the Internet makes citizens more knowledgeable about the EU and this may depend on the limitless information available online – compared to low salience on traditional media - as well as on the medium capacity of translating them into knowledge.We also seek to unveil nuances of how different online platforms affect both political knowledge of and attitudes towards the EU. As the Internet enormously varies in terms of types of websites hosting information, we set apart loci that provide unbiased fact-checked information from those containing a mixture of opinions, non-fact checked information as well as reliable facts. The latter are intrinsically prone to contain high amounts of noise and proved to be, in fact, ineffective at enhancing knowledge and affect opinions. This may be due to the difficulties that users experience in separating the signal from the noise when the two are simultaneously consumed. Conversely, fact checked website – either the official website of institutions or those of newspapers or TV channels – exert a positive effect on opinions. Increased levels of knowledge mediate this process. The internet has dramatically changed information consumption patterns across Europe, and while the availability of higher volumes and more diverse information is a positive feature, additional research should address the potential for noise and confusion in the online information environment. Here we have showed that certain online environments do not enlighten preferences via increased awareness. However, this study is limited in its capacity to reveal nuance of the

22

process of learning by accessing online sources as observational data are unsuitable to such an end. Insights on how different platforms affect cognition and learning can be gathered only by means of laboratory experiments.

ReferencesAarts, Kees, and Holli A Semetko. 2003. The divided electorate: Media use and political

involvement. Journal of Politics 65 (3): 759-784.Baron, Reuben M, and David A Kenny. 1986. The moderator–mediator variable distinction in

social psychological research: Conceptual, strategic, and statistical considerations. Journal of personality and social psychology 51 (6): 1173.

Bartels, Larry M. 1993. Messages received: The political impact of media exposure. American Political Science Review: 267-285.

Bennett, W Lance, and Shanto Iyengar. 2008. A new era of minimal effects? The changing foundations of political communication. Journal of Communication 58 (4): 707-731.

Castells, Manuel. 2000. Materials for an exploratory theory of the network society1. The British journal of sociology 51 (1): 5-24.

Chadwick, Andrew. 2012. Recent shifts in the relationship between the Internet and democratic engagement in Britain and the United States: granularity, informational exuberance, and political learning. Digital Media and Political Engagement Worldwide: A Comparative Study: 39-55.

Dalton, Russell J, and Robert Duval. 1986. The political environment and foreign policy opinions: British attitudes toward European integration, 1972–1979. British Journal of Political Science 16 (01): 113-134.

De Vreese, Claes. 2003. Framing Europe: television news and European integration. Aksant Academic Pub.

De Vreese, Claes H, and Hajo G Boomgaarden. 2005. Projecting EU Referendums Fear of Immigration and Support for European Integration. European Union Politics 6 (1): 59-82.

———. 2006. Media Effects on Public Opinion about the Enlargement of the European Union*. JCMS: Journal of Common Market Studies 44 (2): 419-436.

Delli Carpini, Michael X , and Scott Keeter. 1996. What Americans know about politics and why it matters. Yale UniversityPress, New Haven, CT.

Gabel, Matthew, and Harvey D Palmer. 1995. Understanding variation in public support for European integration. European Journal of Political Research 27 (1): 3-19.

Graber, Doris A. 2001. Processing politics: Learning from television in the Internet age. University of Chicago Press.

Hainmueller, Jens. 2012. Entropy balancing for causal effects: A multivariate reweighting method to produce balanced samples in observational studies. Political Analysis 20 (1): 25-46.

23

Hicks, Raymond, and Dustin Tingley. 2011. Causal mediation analysis. Stata Journal 11 (4): 605.Holtz-Bacha, Christina , and Pippa Norris. 2001. " To Entertain, Inform, and Educate": Still the

Role of Public Television. Political communication 18 (2): 123-140.Imai, Kosuke, Luke Keele, and Teppei Yamamoto. 2010. Identification, inference and sensitivity

analysis for causal mediation effects. Statistical Science 25 (1): 51-71.Iyengar, Shanto, and Kyu S Hahn. 2009. Red media, blue media: Evidence of ideological

selectivity in media use. Journal of Communication 59 (1): 19-39.Kritzinger, Sylvia. 2003. The influence of the nation-state on individual support for the European

Union. European Union Politics 4 (2): 219-241.McLuhan, Marshall. 1964. The medium is the message. Media and cultural studies: keyworks:

129-38.Mondak, Jeffery J. 1995. Newspapers and political awareness. American Journal of Political

Science: 513-527.Newton, Kenneth. 1999. Mass media effects: mobilization or media malaise? British Journal of

Political Science 29 (04): 577-599.Nie, Norman H, Darwin W Miller III, Saar Golde, Daniel M Butler, and Kenneth Winneg. 2010.

The world wide web and the US political news market. American Journal of Political Science 54 (2): 428-439.

Norris, Pippa, and David Sanders. 2003. Message or medium? Campaign learning during the 2001 British general election. Political communication 20 (3): 233-262.

Pariser, Eli. 2011. The filter bubble: What the Internet is hiding from you. Penguin UK.Patterson, Thomas E. 2013. Informing the News. Random House LLC.Robins, James M, and Sander Greenland. 1992. Identifiability and exchangeability for direct and

indirect effects. Epidemiology 3 (2): 143-155.Robinson, John P, and Dennis K Davis. 1990. Television news and the informed public: An

information‐processing approach. Journal of Communication 40 (3): 106-119.Robinson, John P, Mark R Levy, Dennis K Davis, William Gill Woodall, Michael Gurevitch, and

Haluk Sahin. 1986. The main source: Learning from television news. Sage Publications Beverly Hills, CA.

Schuck, Andreas RT, and Claes H De Vreese. 2006. Between Risk and Opportunity News Framing and its Effects on Public Support for EU Enlargement. European Journal of Communication 21 (1): 5-32.

Scott, D Travers. 2007. Pundits in muckrakers’ clothing: political blogs and the 2004 US presidential election. Blogging, citizenship, and the future of media: 39-58.

Tewksbury, David, and Jason Rittenberg. 2012. News on the Internet: Information and Citizenship in the 21st Century. Oxford University Press.

Van Klingeren, Marijn, Hajo G Boomgaarden, and Claes H De Vreese. 2013. Going Soft or Staying Soft: Have Identity Factors Become More Important Than Economic Rationale when Explaining Euroscepticism? Journal of European Integration 35 (6): 689-704.

Zaller, John. 1992. The nature and origins of mass opinion. Cambridge university press.

24

Mediated information Public Opinion Towards the EU

Knowledge

Figure 1. Path analysis testing the mediation role of Knowledge of the EU.

25

Figure 2. Distribution of media by country.

0 .2 .4 .6 .8

RO - RomaniaBG - BulgariaSI - Slovenia

SK - SlovakiaPL - PolandMT - Malta

LT - LithuaniaLV - Latvia

HU - HungaryEE - Estonia

CZ - Czech RepublicCY - Cyprus (Republic)

AT - AustriaSE - Sweden

FI - FinlandDE-E Germany East

PT - PortugalES -Spain

GR - GreeceGB-GBN - Great Britain

IE - IrelandDK - Denmark

LU - LuxembourgIT - Italy

DE-W - Germany - WestNL - The Netherlands

BE - BelgiumFR - France

mean of internet mean of tvmean of press mean of radio

26

Figure 3. Distribution of online platforms by country.

27

Figure 4. Sensitivity analysis of Model 1 and Model 2.

28

Table 1. Balancing on Internet.

Before balancingTreat Control

MeanVarian

ceSkewne

ss MeanVarian

ceSkewn

ess

Urban 2.1060.641

2 -0.1812 1.903 0.618 0.184

Sex 1.4530.247

9 0.1887 1.5430.248

2-

0.1716

Age 35.47 190.7 0.6838 50.14 318.5

-0.0588

5

Occupation 8.669 30.2

-0.0070

58 7.981 31.66 0.453Discuss EU

politics0.929

10.421

90.0699

70.754

10.418

5 0.287Discuss national

politics 1.1450.442

2 -0.17080.985

70.451

70.0167

3Evaluation

national economy 2.96

0.6434 -0.105 3.031 0.693

-0.2124

Evaluation EU economy 3.03

0.5608 0.1036 3.153

0.6318 0.2093

Attitudes towards

immigrants0.092

170.083

7 2.820.084

40.077

28 2.99

After balancingTreat Control

MeanVarian

ceSkewne

ss MeanVarian

ceSkewn

ess

Urban 2.1060.641

2 -0.1812 2.1060.624

4-

0.1687

Sex 1.4530.247

9 0.1887 1.4530.247

8 0.1887Age 35.47 190.7 0.6838 35.47 225.1 0.7579

Occupation 8.669 30.2

-0.0070

58 8.669 35.86 0.1229Discuss EU

politics0.929

10.421

90.0699

70.929

10.430

30.0743

3Discuss national

politics 1.1450.442

2 -0.1708 1.1450.433

2-

0.1626Evaluation

national 2.96 0.643

4-0.105 2.96 0.704

3-

0.1825

29

economyEvaluation EU

economy 3.030.560

8 0.1036 3.03 0.619 0.1476Attitudes towards

immigrants0.092

170.083

7 2.820.092

170.083

68 2.82

Table 2. Balancing on Information Website.


Mean Variance Skewness Mean Variance Skewness

Urban 2.026 0.6249 -0.04571 1.913 0.623 0.1689Sex 1.483 0.2498 0.07003 1.539 0.2485 -0.1574Age 38.91 202.4 0.4705 49.73 327.6 -0.04631

Occupation 9.269 30.14 -0.1306 7.908 31.52 0.4706

Discuss EU

politics0.890

4 0.3649 0.052920.758

7 0.4267 0.293Discuss national politics 1.142 0.3956 -0.1181 0.986 0.4574 0.01691

Evaluation national economy 2.922 0.6237 -0.03325 3.036 0.6945 -0.222Evaluatio

n EU economy 3.021 0.5024 0.1376 3.154 0.6386 0.1988Attitudes towards

immigrants

0.08625 0.07884 2.948

0.08511 0.07787 2.974



30

Urban 2.026 0.6249 -0.04571 2.026 0.6304 -0.0295Sex 1.483 0.2498 0.07003 1.483 0.2497 0.06979Age 38.91 202.4 0.4705 38.94 256.5 0.5569

Occupation 9.269 30.14 -0.1306 9.265 34.79 -0.0228

Discuss EU

politics0.890

4 0.3649 0.052920.890

2 0.4318 0.1189Discuss national politics 1.142 0.3956 -0.1181 1.142 0.4366 -0.1623



immigrants

0.08625 0.07884 2.948

0.08626 0.07882 2.947

Table 3. Balancing on Official Website.



Urban 2.033 0.6303 -0.05838 1.923 0.624 0.1501Sex 1.499 0.2505 0.003643 1.534 0.2489 -0.1358Age 40.24 227.8 0.3349 48.75 325.9 0.02268

Occupation 8.794 30.53 0.03324 8.038 31.56 0.4124

Discuss EU

politics0.939

9 0.3632 0.02510.769

2 0.4224 0.2694Discuss national politics 1.122 0.3993 -0.1022 1 0.454 -0.0001831


n EU economy 3.029 0.5283 0.01271 3.142 0.6275 0.2057Attitudestowards

immigrants

0.07468 0.06923 3.236

0.08546 0.07816 2.966

31



Urban 2.033 0.6303 -0.05838 2.033 0.6331 -0.04432Sex 1.499 0.2505 0.003643 1.499 0.25 0.003488Age 40.24 227.8 0.3349 40.26 273 0.4929

Occupation 8.794 30.53 0.03324 8.792 33.77 0.1058

Discuss EU

politics0.939

9 0.3632 0.02510.939

7 0.4317 0.06353Discuss national politics 1.122 0.3993 -0.1022 1.122 0.4346 -0.1355



immigrants

0.07468 0.06923 3.236

0.0747 0.06913 3.235

Table 4. Descriptive statistics.

Variable ObsMean

Std. Dev.

Min

Max

Attitudes towards the EU

26,208

2.04 0.91 0 4

Political Knowledge

26,594

1.32 0.59 0 2

Internet26,594

0.11 0.31 0 1

TV26,594

0.63 0.48 0 1

32

Printed Press

26,594

0.10 0.31 0 1

Radio26,594

0.07 0.25 0 1

Traditional media websites

26,594

0.11 0.31 0 1

Official websites

26,594

0.02 0.14 0 1

Online Social Networks

26,594

0.02 0.12 0 1

Blogs26,594

0.01 0.07 0 1

Urban26,594

1.93 0.79 1 4

Sex26,594

1.53 0.50 1 2

Age26,594

48.57 18.04 15 97

Occupation

26,594

8.06 5.62 1 18

Discuss EU politics

26,435

0.77 0.65 0 2

Discuss national politics

26,473

1.00 0.67 0 2

Evaluation national economy

26,594

3.03 0.83 1 5

Evaluation EU economy

26,594

3.14 0.79 1 5

Attitudes towards immigrants

26,594

0.08 0.28 0 1

Trust EU23,956

0.43 0.50 0 1

Access26,594

0.89 0.32 0 1

33

Table 5. Mediation Analysis estimating medium’s effects on Attitudes towards the EU with robust standard errors and country fixed-effects.

(1) (2) (3) (4)Whole Whole Europhiles Eurosceptics

VARIABLESKnowled

geAttitud

esKnowled

geAttitud

esKnowled

geAttitud

esKnowled

geAttitud

es

Internet0.079**

*0.053

***0.202**

*0.187

***0.138**

*0.126

***0.183**

*0.166

***

(0.012)(0.01

9) (0.022)(0.03

2) (0.036)(0.04

6) (0.030)(0.04

3)

Political Knowledge0.137

***0.134

***0.122

***0.038

*(0.01

6)(0.01

6)(0.02

0)(0.02

2)

TV 0.126***0.149*

** 0.049 0.047 0.115***0.164*

**

(0.020)(0.029

) (0.034)(0.044

) (0.027)(0.038

)

Printed Press 0.189***0.175*

** 0.128*** 0.047 0.156***0.180*

**

(0.025)(0.040

) (0.041)(0.056

) (0.036)(0.053

)

Radio 0.153***0.105*

* 0.080* 0.029 0.162***0.162*

**

(0.027)(0.042

) (0.043)(0.056

) (0.038)(0.056

)

Urban 0.036***0.028*

* 0.036***0.029*

* 0.043*** 0.018 0.036***0.033*

*

(0.007)(0.012

) (0.007)(0.012

) (0.011)(0.014

) (0.011)(0.016

)

Sex-

0.101*** -0.017-

0.100*** -0.017-

0.089***

-0.082*

**-

0.114*** 0.016(0.012) (0.018 (0.012) (0.018 (0.018) (0.022 (0.017) (0.025

34

) ) ) )

Age 0.004***

-0.004*

** 0.004***

-0.005*

** 0.004*** -0.001 0.004***

-0.005*

**

(0.000)(0.001

) (0.000)(0.001

) (0.001)(0.001

) (0.001)(0.001

)

Occupation 0.000-

0.003* 0.000-

0.003* 0.001

-0.004*

* -0.000 0.002

(0.001)(0.002

) (0.001)(0.002

) (0.002)(0.002

) (0.001)(0.002

)

Discuss EU politics 0.066*** 0.005 0.062*** 0.002 0.046**0.049*

* 0.068*** -0.038

(0.013)(0.020

) (0.013)(0.020

) (0.019)(0.024

) (0.018)(0.028

)

Discuss national politics 0.073*** 0.037* 0.068*** 0.033* 0.063***0.072*

** 0.060*** -0.013

(0.013)(0.020

) (0.013)(0.020

) (0.020)(0.023

) (0.017)(0.027

)

Evaluation national economy

-0.030***

-0.172*

**-

0.029***

-0.172*

** 0.006

-0.050*

**-

0.043***

-0.153*

**

(0.010)(0.015

) (0.010)(0.015

) (0.015)(0.018

) (0.014)(0.022

)

Evaluation EU economy -0.015*

-0.241*

** -0.012

-0.238*

** 0.015

-0.102*

** -0.006

-0.191*

**

(0.009)(0.015

) (0.009)(0.015

) (0.013)(0.017

) (0.013)(0.022

)

Attitudes towards immigrants -0.014

-0.091*

** -0.013

-0.089*

** -0.017 0.014 0.003-

0.083*

(0.021)(0.032

) (0.021)(0.032

) (0.030)(0.038

) (0.031)(0.046

)

Constant 1.354***3.766*

** 1.238***3.639*

** 1.177***3.155*

** 1.202***3.085*

**(0.049) (0.074 (0.052) (0.077 (0.076) (0.097 (0.080) (0.115

35

) ) ) )

Internet≠TV38.58**

* 3.89**38.58**

*12.20

***15.03**

* 0.01Internet≠Printed Press 0.47 0.14 0.47 3.98** 0.94 0.09Internet≠Radio 5.16** 5.57** 5.16** 5.70** 0.47 0.01

ACME 0.01* 0.03* 0.02* 0.01*Direct Effect 0.05* 0.19* 0.12* 0.16*Total Effect 0.06* 0.21* 0.14* 0.17*% of Total Effect Mediated 0.17* 0.12* 0.12* 0.04*

Country fixed-effects yes yes yes yes yes yes yes yesObservations 26,038 26,038 10,299 13,340R-squared 0.149 0.150 0.110 0.127rmse 0.839 0.839 0.663 0.812Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

Table 6. Mediation Analysis estimating platforms effects on Attitudes towards the EU with robust standard errors and country fixed-effects.

(5) (6) (7) (8)

36

Mediator=OfficialWebsites

Mediator=Traditional Media Websites

Mediator=Online Social Networks Mediator=Blogs

VARIABLESKnowled

ge AttitudesKnowledg

e AttitudesKnowled

ge AttitudesKnowled

ge Attitudes

Official websites0.125**

* 0.072* 0.135*** 0.099**0.130**

* 0.0730.156**

* 0.051(0.024) (0.037) (0.027) (0.040) (0.032) (0.046) (0.034) (0.048)


0.058*** 0.041* 0.088*** 0.054***

0.058*** 0.044*

0.087*** 0.071**

(0.016) (0.024) (0.011) (0.018) (0.017) (0.025) (0.021) (0.028)Online social networks 0.058* 0.021 0.052* 0.066 0.035 0.033

0.130*** 0.002

(0.033) (0.048) (0.031) (0.046) (0.027) (0.043) (0.042) (0.058)Blogs -0.008 -0.055 0.014 -0.040 -0.002 -0.036 0.050 -0.083

(0.054) (0.077) (0.051) (0.080) (0.064) (0.089) (0.044) (0.067)Political Knowledge 0.107*** 0.120*** 0.145*** 0.124**

(0.032) (0.015) (0.035) (0.054)TV 0.011 0.034 0.003 0.020 0.021 -0.021 -0.019 0.068

(0.029) (0.043) (0.014) (0.021) (0.032) (0.044) (0.051) (0.073)Printed Press 0.130*** 0.149* 0.081*** 0.043 0.092* 0.065 -0.009 0.086

(0.041) (0.078) (0.021) (0.034) (0.048) (0.095) (0.072) (0.138)Radio 0.020 -0.016 0.049** 0.015 0.030 0.021 -0.068 0.171

(0.048) (0.066) (0.024) (0.037) (0.056) (0.092) (0.069) (0.112)Urban 0.079*** 0.021 0.040*** 0.016 0.076*** 0.042* -0.009 -0.007

(0.014) (0.022) (0.007) (0.011) (0.017) (0.024) (0.027) (0.037)

Sex-

0.113*** 0.120*** -0.105*** -0.023-

0.135*** -0.001 -0.095** 0.068(0.023) (0.035) (0.011) (0.017) (0.027) (0.038) (0.039) (0.060)

Age 0.002** -0.005*** 0.003*** -0.003*** 0.002*** -0.007*** 0.004*** -0.006***(0.001) (0.001) (0.000) (0.001) (0.001) (0.001) (0.001) (0.002)

Occupation -0.001 0.001 0.001 -0.005*** -0.001 -0.000 0.002 -0.006(0.002) (0.003) (0.001) (0.001) (0.002) (0.003) (0.003) (0.005)

Discuss EU politics 0.040 0.019 0.052*** 0.041** 0.107*** 0.093** 0.072* 0.012

37

(0.026) (0.037) (0.012) (0.018) (0.029) (0.043) (0.038) (0.072)Discuss national politics 0.092*** 0.010 0.085*** 0.012 0.025 -0.011 0.040 0.026

(0.025) (0.037) (0.012) (0.018) (0.031) (0.043) (0.039) (0.072)Evaluation national economy -0.013 -0.214*** -0.032*** -0.177*** 0.069*** -0.121***

-0.092*** -0.164***

(0.018) (0.032) (0.009) (0.014) (0.022) (0.032) (0.030) (0.048)Evaluation EU economy -0.035* -0.233*** -0.008 -0.217***

-0.072*** -0.215*** 0.010 -0.241***

(0.018) (0.030) (0.008) (0.013) (0.023) (0.030) (0.036) (0.052)Attitudes towards immigrants -0.011 -0.031 0.019 -0.041 -0.013 0.018 0.093 -0.005

(0.046) (0.056) (0.020) (0.029) (0.045) (0.062) (0.071) (0.107)Constant 1.395*** 3.614*** 1.333*** 3.714*** 1.218*** 3.482*** 1.565*** 3.896***

(0.097) (0.150) (0.048) (0.070) (0.107) (0.146) (0.157) (0.232)

ACME 0.01* 0.01* 0.01 0.01Direct Effect 0.08* 0.06* 0.04 -0.08Total Effect 0.09* 0.07* 0.04 -0.07% of Total Effect Mediated 0.14* 0.15* 0.09 -0.06

Country fixed-effects yes yes yes yes yes yes yes yesObservations 26,038 26,038 26,038 26,038R-squared 0.177 0.138 0.152 0.211rmse 0.821 0.808 0.779 0.782Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

38

Table 7. Balancing on Online Access.

before balancingTreat Control


Urban 1.935 0.6261 0.1284 1.845 0.6038 0.2805Sex 1.528 0.2492 -0.1112 1.575 0.2444 -0.3044Age 46.82 309.7 0.1233 62.25 236.1 -0.8369Occupation 8.384 32.01 0.2721 5.477 20.5 1.881Discuss EU politics

0.7902 0.42 0.2342

0.6366 0.415 0.5121

Discuss national politics 1.017 0.4479 -0.01913

0.8936 0.4803 0.1445

Evaluation national economy 2.998 0.6891 -0.1689 3.223 0.6361 -0.4491Evaluation EU economy 3.125 0.6108 0.1855 3.254 0.7271 0.2516Attitudes towards immigrants

0.08495 0.07774 2.977

0.08742 0.0798 2.921

after balancingTreat Control


Urban 1.935 0.6261 0.1284 1.935 0.6112 0.1147Sex 1.528 0.2492 -0.1112 1.528 0.2493 -0.1117Age 46.82 309.7 0.1233 46.84 316 0.06066Occupation 8.384 32.01 0.2721 8.379 41.38 0.455Discuss EU politics

0.7902 0.42 0.2342

0.7898 0.444 0.2671

Discuss national politics 1.017 0.4479 -0.01913 1.016 0.4649 -0.02029Evaluation national economy 2.998 0.6891 -0.1689 2.998 0.7081 -0.3104Evaluation EU economy 3.125 0.6108 0.1855 3.125 0.7077 0.2018Attitudes towards

0.08495

0.07774 2.977 0.0849

0.07772 2.978

39

immigrants

Table 8. Effects of online news consumption on political knowledge. Instrumental variables approach with robust standard errors and country fixed-effects.

(9) (10) (11) (12) (13)Whole Europhiles Eurosceptics Whole Whole

VARIABLESKnowle

dge Knowledge KnowledgeKnowle

dgeKnowle

dge

Internet0.703*

** 0.778*** 0.661**(0.185) (0.279) (0.275)

Official websites3.268*

**(0.876)


0.660***

(0.172)

TV0.253**

* 0.301** 0.245***0.079**

* 0.053**(0.068) (0.136) (0.091) (0.030) (0.026)

Printed Press0.375**

* 0.415*** 0.352***0.183**

*0.171**

*(0.077) (0.148) (0.103) (0.037) (0.035)

Radio0.349**

* 0.399*** 0.337***0.176**

*0.151**

*(0.078) (0.148) (0.104) (0.045) (0.042)

Urban 0.027** 0.013 0.032** 0.027** 0.026**(0.011) (0.016) (0.015) (0.011) (0.011)

Sex

-0.074**

* -0.089*** -0.055**

-0.077**

*

-0.081**

*(0.017) (0.027) (0.023) (0.017) (0.017)

Age0.002**

* 0.003*** 0.003***0.002**

*0.003**

*(0.001) (0.001) (0.001) (0.001) (0.001)

Occupation 0.003* 0.007** 0.002 0.003** 0.003*(0.002) (0.003) (0.002) (0.002) (0.002)

Discuss EU politics0.060**

* 0.050* 0.054**0.069**

*0.072**

*(0.020) (0.030) (0.027) (0.020) (0.019)

Discuss national politics

0.057*** 0.044* 0.054**

0.066***

0.063***

(0.017) (0.026) (0.023) (0.017) (0.017)

40

Evaluation national economy -0.025* 0.007 -0.027

-0.029** -0.021*

(0.013) (0.017) (0.019) (0.013) (0.012)

Evaluation EU economy

-0.045**

* -0.008 -0.040**

-0.054**

*

-0.058**

*(0.013) (0.017) (0.018) (0.012) (0.012)

Attitudes towards immigrants 0.022 -0.017 0.047 0.031 0.022

(0.026) (0.036) (0.039) (0.027) (0.026)

Constant1.062**

* 0.800*** 1.037***1.257**

*1.220**

*(0.113) (0.193) (0.160) (0.081) (0.083)

Internet≠TV13.16*

** 9.42*** 4.80**13.78**

*13.80*

**Internet≠Printed Press 7.55*** 6.10** 2.76*

12.94*** 9.44***

Internet≠Radio 8.08*** 5.32** 3.14*13.04**

* 9.77***

Access 0.10*** 0.10*** 0.09*** 0.02*** 0.11***(0.004) (0.01) (0.006) (0.001) (0.005)

Anderson-Rubin Wald test

14.34*** 8.64*** 5.60-**

14.34***

14.34***

Kleibergen-Paap rk LM statistic

305.26*** 89.81*** 166.49

228.74***

395.02***

Cragg-Donald Wald F statistic

1493.9*** 640.72*** 654.11***

210.69***

1147.94***

Observations 26,398 10,343 13,476 26,398 26,398R-squared 0.087 0.052 0.099 -0.226 0.066rmse 0.555 0.538 0.551 0.643 0.562Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

41

Table 9. Effects of online news consumption on attitudes towards the EU. Instrumental variables with robust standard errors and country fixed-effects.

(14) (15) (16) (17) (18)Whole Europhiles Eurosceptics Whole Whole

VARIABLES Attitudes Attitudes Attitudes Attitudes Attitudes

Internet 1.221*** 1.016** 0.964*(0.361) (0.446) (0.508)

Official websites 4.553***(1.385)

Traditional media websites 1.045***

(0.304)Political Knowledge 0.073*** 0.068*** 0.012 0.080*** 0.091***

(0.015) (0.016) (0.020) (0.015) (0.012)TV 0.716*** 0.663** 0.550** 0.172*** 0.146***

(0.195) (0.307) (0.245) (0.039) (0.029)Printed Press 0.782*** 0.716** 0.574** 0.214*** 0.217***

(0.204) (0.315) (0.259) (0.045) (0.042)Radio 0.738*** 0.679** 0.574** 0.200*** 0.170***

(0.203) (0.314) (0.259) (0.052) (0.041)Urban -0.001 -0.004 -0.005 0.000 -0.001

42

(0.008) (0.009) (0.012) (0.009) (0.008)Sex 0.026** -0.035** 0.059*** 0.011 0.020*

(0.012) (0.014) (0.017) (0.013) (0.012)Age -0.001 0.001 -0.001 -0.001* -0.000

(0.001) (0.001) (0.001) (0.001) (0.001)Occupation -0.003*** -0.004*** 0.000 -0.003** -0.004***

(0.001) (0.001) (0.001) (0.001) (0.001)Discuss EU politics 0.006 0.020 -0.030 0.011 0.035***

(0.017) (0.018) (0.025) (0.018) (0.013)Discuss national politics -0.025 0.028 -0.038* 0.012 -0.009

(0.017) (0.019) (0.022) (0.014) (0.014)Evaluation national economy -0.185*** -0.043*** -0.182*** -0.197*** -0.188***

(0.010) (0.012) (0.013) (0.011) (0.010)Evaluation EU economy -0.184*** -0.078*** -0.151*** -0.196*** -0.197***

(0.010) (0.011) (0.015) (0.010) (0.009)Attitudes towards immigrants -0.059*** 0.028 -0.053** -0.043* -0.054***

(0.019) (0.024) (0.026) (0.023) (0.020)Constant 2.466*** 2.223*** 2.285*** 3.034*** 2.870***

(0.221) (0.321) (0.289) (0.078) (0.108)

Internet≠TV 9.03*** 6.30** 2.45 10.51*** 10.33***Internet≠Printed Press 7.55*** 4.97** 2.38 10.33*** 9.41***Internet≠Radio 8.87*** 6.11** 2.37 10.51*** 10.28***

Access 0.05*** 0.06*** 0.05*** 0.01*** 0.06***(0.004) (0.007) (0.006) (0.001) (0.003)

Anderson-Rubin Wald test 11.97*** 5.50** 3.74* 11.97*** 11.97***Kleibergen-Paap rk LM statistic

145.02*** 54.51*** 70.91*** 99.77***

353.60***

Cragg-Donald Wald F statistic

131.74*** 68.05*** 56.41*** 20.50*** 88.45***

Observations 26,038 10,299 13,340 26,038 26,038R-squared 0.105 0.041 0.093 -0.337 0.048rmse 0.857 0.674 0.815 1.048 0.884Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

43

44

Documents

€¦ · Web viewThis is in spite of a real word trend indicating that the ... John P, Mark R Levy, Dennis K Davis, William Gill Woodall, Michael Gurevitch, and Haluk Sahin