FACEBOOK AND POLITICAL INFORMATION IN ITALY AND THE UK: AN ANTIDOTE AGAINST POLITICAL FRAGMENTATION AND POLARISATION

Due to the rise of digital technologies citizens can today count on innumerable and diverse sources of political information. Arguably such a proliferation of media choices in conjunction with a structural aspect of the internet, namely the presence of a pro-active and self-selecting audience, offers the conditions most conducive to selective exposure (Bimber & Davis 2003). The tendency for selectivity of the internet audience has raised serious concerns as it may lead to a more polarised and less informed electorate (Sunstein 2001, Polat 2005, Bennett & Iyengar 2008). However, despite many theoretical speculations, the relationship between the internet and the exposure to politically diverse information is still unclear. The present paper aims to contribute to this debate. Through a sequential explanatory mixed methods strategy, it examines the impact of Facebook on the consumption of political information in Italy and the United Kingdom and argues that this social networking website could reduce the risks of selective exposure and operate as an antidote against political fragmentation and polarization.


Introduction
The informative impact of the internet on today's society is certainly manifold.Digital technologies have not only multiplied the number of information sources and the amount of available information, but they have also transformed the nature of such information by providing increasingly interactive and networked content (Hardy, Jamieson & Winneg, 2009).Two main areas of the information environment have been particularly affected by the internet.The first relates to the flow of information which has been heavily accelerated, while the second individuals to focus on certain topics or the disposition of sites to tailor their content to specific audiences.In their view, the specialization of news exposure has generated the segmentation of audiences and, ultimately, fragmentation which is defined as "the lack of widespread public exposure to some content of interest " (p. 196).They argue that a fragmented information environment is prone to polarization which occurs when audience groups consume idiosyncratic content.Tewksbury and Rittenberg developed an interesting argument and claim that by providing "too much freedom", by enabling users to focus on content and activities which are relevant to them, the internet can lead to selective exposure and the exclusion of counter-attitudinal political information.The second structural aspect is the presence of a pro-active and self-selecting audience.The active role of the audience in content selection and consumption is, in fact, a necessary condition to the occurrence of selective exposure (Dutta-Bergman & Wonjun, 2005).This control could induce segmentation as individuals may select only the information strengthening their existing position (Anduiza, Cantijoch and Gallego, 2009).The tendency for selectivity of the internet audience is an area of concern for certain academics who believe that this phenomenon will progressively lead to a more polarised and less-informed electorate (Sunstein 2001, Polat 2005, Bennett & Iyengar 2008).One of most prominent advocates of this position is Sunstein (2001) who suggests that in the online environment individuals tend to operate in eco-chambers interacting with like-minded users.
Despite many theoretical speculations the relationship between the internet and the exposure to politically-diverse information is, today, still unclear.Some research supports the argument that the internet increasingly exposes individuals to pro-attitudinal perspectives.For instance, Bimber and Davis (2003) analysed the audiences of campaign websites during the 2000 U.S. presidential election and conclude that when compared with television and newspapers the internet provides the conditions most conducive to selective exposure.Adamic and Glance (2005) study the links among the posts of a series of blogs and show that Liberal blogs linked primarily to other Liberal blogs while the opposite happens for Conservative blogs.Similarly, Stroud (2008) establishes that people's political beliefs relate to their media exposure, both online and offline.These findings are confirmed by Nie, Miller, Golde, Butler and Winneg (2010) who demonstrate that online consumers expose themselves to news content in line with their own political views.On the other hand, there is a strand of research asserting that internet users can brake away from the dynamics of selective exposure through accidental exposure to information.Brundidge (2010) speaks of inadvertency and argues that in the online environment individuals are exposed to more political difference than they would be otherwise, even if only inadvertently.The inadvertency thesis is supported by several studies.As highlighted by Brundidge (2010, p. 685), Wojcieszak and Mutz (2009) observe that the exposure to counter-attitudinal political messages is more likely to happen in non-political chat rooms.Accordingly, Cornfield (2005) finds that 36 percent of internet users claim to get campaign news and information not through a directed search but accidentally, while surfing the web for different purposes.According to Tewksbury, Weaver and Maddex (2001), the traditional mass-media models of news dissemination may not be applicable to the internet because online audiences acquire information even when they are not deliberately looking for it.Traditional models, in fact, conceptualize news exposure as an intentional process whereas internet users can obtain information in an incidental fashion, as "a byproduct of their other online activities" (Tewksbury et al., 2001, p. 533).
Among the various internet tools, social networking websites (SNSs) appear to be particularly prone to accidental exposure (Baresch, Knight, Harp & Yaschur, 2011).SNSs can be defined as "web-based services that allow individuals to (1) construct a public or semi-public profile within a bounded system, (2) articulate a list of other users with whom they share a connection, and (3) view and traverse their list of connections and those made by others within the system" (Boyd & Ellison, 2007, p. 211).Because of the increasing possibilities of sharing content offered by SNSs many internet users organize and obtain their news via their social networks (Baresch et al. 2011).Research has attested to the political relevance of SNSs and found that one of the main motivations behind their usage is to gather political information (Rainie & Smith, 2011).Lerman and Ghosh (2010) analyse the mechanisms of news diffusion on SNSs and speak of information contagion.By means of sharing, liking, retweeting, accidental exposure to information can occur as individuals do not always choose what to consume and information is often presented to them.SNSs' users could, therefore, evade selective exposure which relies on the active role of the audience and encounter political difference.An, Cha, Gummadi and Crowcroft (2001) confirm this theory and find that on Twitter there is a non-negligible amount of indirect media exposure which expands the diversity of news users are exposed to.Similar results emerge from the investigations of Vickery (2009) and Kim (2011) who establish that SNSs can enable exposure to crosscutting opinions.On the contrary, in a qualitative content analysis of Facebook posts, Meyer (2012) observes that in the U.S. Republicans and Democrats have built highly partisan social media communities and that selective exposure can take place also on SNSs.Taking into account the academic literature presented so far, a mixed picture emerges with regards to the tendency of selective exposure of internet users.This state of affairs calls for further research particularly in relation to SNSs, an environment where accidental dynamics strongly intervene upon the flow of information.

Methodology
This paper has been drawn from a larger mixed-methods study investigating the impact of Facebook on political participation in Italy and the United Kingdom.Among the various mixed-methods strategies, this investigation has employed a sequential-explanatory approach characterised by a first quantitative phase and a subsequent qualitative phase.In the first phase, a web-based survey explored the links between citizens' Facebook political participation, internet political participation (excluding Facebook activity) and offline political participation.Information from the first phase has been further examined in the second qualitative stage where a series of semi-structured interviews were held.The purpose of the qualitative phase is to build upon the initial findings and to examine the mechanisms responsible for producing the identified associations.

Sampling
A pragmatic approach has been applied to sampling.This investigation's target populations are the British and Italian Facebook populations minus the 13-17, and 65+ years old age groups.The 13-17 years old age group has been excluded from the study in order to avoid ethical issues.The 65+ age group has not been taken into consideration as, in both countries, it represents a very small fraction of the total Facebook population and finding participants fitting within this category could have proved particularly problematic.
A stratified-snowball sampling strategy has been adopted to recruit respondents for the online surveys.This strategy entails an initial chain-referral recruitment stage and the random selection of the final sample through the application of a post-stratifying criterion (i.e.age).Age has been chosen as post-stratifying criterion because of its relevance to the political participation phenomenon (see Quintelier, 2007 for a detailed account).From the initial 483 participants 196 participants for the British sample (BS) and 196 participants for the Italian sample (IS) were selected.For the qualitative phase a total of 26 interviews (13 for the BS and 13 for the IS) were held.A subset of the survey participants was chosen for the qualitative stage.The selection of the sub-sample occurred according to several criteria such as the identification of particularly significant quantitative results, unexpected non-significant quantitative results, extreme cases, demographics, characteristics, and self-interest.Because the sub-sample did not provide participants with certain characteristics (e.g.political activists between the age of 18-24 willing to be interviewed) a number of qualitative participants (i.e. 4 for the BS and 3 for the IS) were selected through a further snowballing procedure.

Information Activities
The study this paper has been drawn from conceptualises political participation as a multidimensional phenomenon which encompasses under its umbrella three different typologies of participatory activity: campaign activities (e.g.soliciting others to support or oppose a particular political party, candidate, and initiative); contact activities (e.g.contacting a political party, candidate, government department and/or local council); and communication activities (e.g.consumption of political news).This paper focuses on communication activities which Christy (1987) describes as a form of political participation not channelled through political institutions and reflecting a more individual interest and psychological involvement in politics.Three activities related to the consumption of political information have been considered in this paper: learning about a political initiative, meeting, rally and/or protest; learning about a group or an organisation developed around politics; and consumption of political news.Each of the activities has been assessed through a five-point frequency scale (1=Never, 2=Rarely, 3=Sometimes, 4=Often and 5=Very Often).Participants were asked how often in the last six months they engaged in the first two activities through Facebook, through the internet (excluding Facebook activity) and watch/listen/read political news on Facebook, on the internet, on TV, on the radio and on the press.In addition, three summated-rating scales: Facebook Political Information, Internet Political Information and Offline Political Information, were generated adding the scores of the various survey items, with each item bearing equal weight.

Data Analysis
Considerations on the nature of the samples and of the data have guided the selection of the statistical tools for the quantitative data analysis.In a political participation study Calenda and Mosca (2007) explain that samples characterized by strong non-probabilistic components cannot produce strong inferences and descriptive statistical tools are to be preferred.Accordingly, given the purposive and pragmatic nature of this study's samples, descriptive statistics have been employed to analyze the quantitative data.The most appropriate statistical tools have been chosen taking into consideration the distribution of the data.The Kolmogorov-Smirnov test with Lilliefors correction, which is best used for sample sizes of more than 50, with unknown population mean and variance (Lilliefors 1967), was run to assess the normality of the distributions of scores.The test established that the data is not normally distributed (p.< 0.05) and, consequently, medians rather than means have been employed as measures of central tendency.The qualitative data have been examined through a thematic analysis.A number of thematic categories have been developed in order to examine how selective exposure operates on Facebook (Table II).The themes were established taking into consideration the academic literature (i.e.prior research code development) and the questionnaire's results.Finally, the themes were reviewed and revised during the analysis of the qualitative content (i.e.data-driven code development).This three-step process can be described as theory-driven coding (Boyatzis, 1998).

Results
The current paper aims to assess whether Facebook could operate as an antidote against political fragmentation and polarization.In order to do so both the relevance of Facebook as a political information source and its capability to facilitate the exposure to counter-attitudinal information have to be addressed.The first issue has been examined through the quantitative data while the qualitative results have been employed to explore the second matter.

Quantitative Results
From the quantitative results it emerges that in the BS the most used channel for each of the considered information activities is the internet while Facebook and offline related activities display identical scores.In the IS, Facebook is the most used political information channel, followed by the internet and the offline world.In terms of intensity of activity, the BS displays low scores in relation to the consumption of political information whereas the IS shows a moderate consumption.These results are also confirmed when the Facebook, Internet and Offline Political Information scales are taken into consideration.In summary, the data indicate that in the BS Facebook is employed to obtain political information as much as traditional media but less than the internet, whereas in the IS Facebook is the major political information channel.These results demonstrate the relevance of Facebook as a political information source and are in line with the several studies investigating the impact of Facebook on the diffusion of political information.Hermida, Fletcher, Korrell and Logan (2011) find that two-fifths of SNSs users get the news from people they follow on websites such as Facebook, while a fifth obtain news from news organizations and individual journalists they follow.They conclude that SNSs are a significant source of news.Similarly, in Vickery's (2009) qualitative study all participants cite Facebook as a major source of political information and many participants claim to have discovered new sources of news through the links their friends post on Facebook.According to Vickery (2009) these results are indicative of a larger trend in which an increasing number of people are using SNSs as news aggregators.However, Facebook's relevance as political information source is not unanimously recognized in academia.For instance, in a recent study, Lampe, Vitak, Gray, and Ellison (2012) observe that Facebook users were not likely to use this SNS to seek information.

II. International Conference on
According to them, this suggests that users may still perceive Facebook mainly as a social tool.To evaluate the potential of Facebook to counteract the fragmentation and polarization trends the penetration of this SNS has also to be considered.In relation to the two countries this paper focuses on, Facebook has 51.61 per cent penetration of the total British population and 61.02 per cent of the British online population.In Italy Facebook's penetration is even lower with regards to the total population, 38.16 per cent, while is slightly higher for the Italian online population, 70.85 per cent (Socialbakers, 2013).This data shows that Facebook is not yet universal.In this sense, it would be possible to talk of a Facebook divide which may limit this SNS' contributions and relevance to society.

Qualitative Results
Facebook's ability to reduce selective exposure has been assessed through the qualitative data.In particular, this paper explores the thesis that Facebook users may be inadvertently exposed to counter-attitudinal political information.As these statements illustrate, from the interviews it appears that in both samples it is common for Facebook users to acquire politically-diverse information: 1. "you will find people who will post the same article you would come across on a website or TV or whatever, but you would also find a larger variation.So you would find news from websites you may not know they exist or you may had never come across.So the participation of so many people means that you get a larger net cast […].I read stuff from sites that I wouldn't have known even that existed if a friend or another person, a friend of friend, hadn't found it and posted it" (BS)

"(On Facebook) I've crossed people with completely opposing views" (BS)
3. "Facebook has no filters.I have more than 1000 contacts […] I know the political views of 20% of those contacts while I have no idea for the rest.Sometimes I see links and click on them.Therefore, (on Facebook) you have a surplus of news which comes from everywhere" (IS) 4. "I have two, three (Facebook) contacts that are located right on the opposite (political) side.Clearly they post their content virally" (IS) Tewksbury and Rittenberg (2012) describe the online consumption of news and political information as a multistep process.Online audiences can purposively select political information but also be accidentally exposed to it.As mentioned in the Introduction section, active selection is a necessary condition for the occurrence of selective exposure.On Facebook, people who are interested in politics can follow political pages and news organizations or even be members of political groups.In this case, Facebook users pro-actively select informative sources and, as a consequence, selective exposure may take place.In both samples the interviews confirm that the Facebook environment is not immune to the dynamics of selective exposure.However, the qualitative data also show that active selection of information does not lead automatically to selective exposure.In accordance with Frey's (1986) argument that ideological homogeneity is not psychologically desirable, individuals interested in politics may, in fact, also search for politically-diverse information: 1. "80%, 90% of the time I look (for political information) in (politically) contiguous settings.However, I understand that on certain issues I need also to see how different political forces think and act" (IS) Facebook users can also access political information through an accidental route.Facebook's News Feed plays a crucial role in this accidental exposure.The News Feed, which appears on each user's homepage, performs an information based function.It simplifies and accelerates the sharing of information by showing a constantly updated list of friends and pages' activities (Vitak, Zube, Carr, Ellison & Lampe, 2009).Among the various Facebook's features, arguably, the News Feed provides users with the greatest opportunities in terms of political information.This could be possibly related to the lack of active selection in the consumption of political information.The News Feed enables users to passively view activities of their network (Lampe et al. 2012).Individuals are drawn to the activities within their networks and no longer have to seek out information, but rather the information is presented to them whenever they access the site (Vickery 2009).
The thesis of information contagion developed by Lerman and Ghosh (2010)  Not all Facebook users are, however, able to gain access to counter-attitudinal political information through the accidental route.To bypass selective exposure a politically interested and politically-heterogeneous network of contacts is required.The relevance of the Facebook network for the information gathering process has been stressed by Vickery (2009) and Lampe et al. (2012).The first states that the level of political involvement of the network strongly influences the degree of exposure to political information while

Conclusions
A model explaining the dynamics of selective exposure in the Facebook environment has been developed.According to this model, called the Dual Routes of Exposure Model, Facebook's consumption of political information can occur through two routes: the direct route and the accidental route.In the direct route users have control over the flow of information and actively select information according to personal preferences, interests and habits.Due to this active selection of content selective exposure may take place.The direct route, however, does not lead necessarily to selective exposure as users may purposively consume politically diverse information.
In the accidental route users are passively and inadvertently exposed to information.Like the previous route, this can lead to the exposure to both reinforcing and counter-attitudinal content.The first can occur if users interact excursively with not politically interested or  This paper contributes to the strand of research investigating the relationship between the internet and the phenomenon of selective exposure.Focusing on the case of Facebook, it provides evidence in support of Brundidge's (2010) inadvertency thesis.In the Facebook environment accidental dynamics strongly intervene in the diffusion of political information and users may be inadvertently exposed to political difference.Active selection is a necessary condition for the occurrence of selective exposure but on SNSs the diffusion of information is both an intentional and incidental process.Through what has been labelled by Lerman and Ghosh (2010) as information contagion, users are presented with new opportunities to heterogeneously expand their social and information networks.As a result, the formation of eco-chambers, where users will interact only with like-minded individuals (Sunstein, 2001), appears less probable on Facebook.Hermida et al. (2011) report that SNSs are particularly valued information sources as they facilitate the exposure to a wider range of news and information.Accordingly, this study finds that Facebook enables the access to politically diverse content.This, however, may not be enough to counter the polarisation and fragmentation trends characterising the internet.Facebook is, in fact, not immune to selective exposure and accidental exposure to counter-attitudinal perspectives is only a component of the information consumption process.Academics celebrate Facebook's ability to provide a much larger sphere of potential influence for opinion leaders and opinion followers than other media environments (Zube, Lampe & Lin, 2009).In line with several studies addressing SNSs' political relevance (Rainie & Smith, 2012;Rainie, Smith, Lehman-Schlozman, Brady & Verba, 2012;Vickery, 2009), this paper confirms that Facebook is a relevant source of political information.At the same time, it recognises that Facebook is not a universal tool.The existence of a usage gap which could be described as Facebook divide limits the impact of this SNS on society.In conclusion, Facebook is a potential antidote against political fragmentation and polarisation but its contribution to the formation of a more widely-informed electorate is yet to be established.
II.International Conference on Communication, Media, Technology and Design 02-04 May 2013 Famagusta -North Cyprus politically contiguous contacts.Conversely, in the presence of a politically interested and politically heterogeneous network, Facebook users may bypass selective exposure and acquire counter-attitudinal political information.

Figure I -
Figure I -Facebook and Selective Exposure: The Dual Routes of Exposure Model

Table III -Information activities
1= Never, 2= Rarely, 3 = Sometimes, 4= Often, 5= Very often Please note that offline consumption of political news is an average measure generated from TV, radio and press consumption of political news

Table IV -Facebook, internet and offline political information activity
is supported by the findings of this study.Both British and Italian participants, in fact, talk of the extended, viral nature of the Facebook information environment: 2. "people are subjected to more articles, a wide variety, instead of flicking pass the news, on Facebook a headline comes up…" (BS) 3.4."sometimesyou get shared stories or shared posts that circulate among friends and you get thousand of people liking them."(BS) 5. 6. "It's like when you go fishing.Instead of using one fishing rod you use many as there are the news I get and the news my contacts get.
Lampe et al. (2012)claim that users with larger and more diverse networks should obtain more non-redundant information.The presence of politically heterogeneous contacts is not enough to be exposed to political difference.Interaction with such contacts is also needed as The News Feed displays only activities relevant to the user.The relevance of an activity is established through an algorithm (i.e.EdgeRank) based on various parameters, among which users' previous behaviours.This means that if a user does not interact with a contact over a certain period of time, this contact's activities will stop appearing in the News Feed.Consequently, as highlighted by a British participant, due to the EdgeRank algorithm lack of interaction could lead to selective exposure: