Effects of Social Media on Civil and Political Participation and a Field of Survey over on Facebook

Turkish youth is characterized with low level of political participation. Several surveys showed that less than 10 percent of Turkish youth are active in political parties and civil society organizations. There are cultural and behavioral reasons of this inertia. Meanwhile, Facebook, the most important social network site is widely used in Turkey and Turkish users form the 6th biggest population in Facebook but level of political participation in Facebook is unknown. By using data collected from 563 university students, this paper showed that there different dimensions of political participation in Facebook and level of activity in these dimensions is dependent on several factors. Demographic factors are important in “active participation”, meanwhile participation through “direct” or “indirect” expression are affected by Facebook usage patterns of respondent as well as “real life” political activities and perceptions.


Introduction
Turkish youth is characterized with their inertia to participate in politics. More than one survey showed that only 10 percent of young people living in Turkey are involved in direct politics such as becoming a member of a political party or directly participating in canvassing and other conventional political activities. Participation in unconventional politics is also limited, Turkish youth don"t prefer to participate in boycotts or meetings, as well as being a member of a civil society organization is not popular among them. There different explanations of this inertia, from a deterministic explanation about political culture to behavioral approach emphasizing rules and regulations designed to prevent political participation.
In the meantime, social network sites, especially Facebook is very popular among Turkish Internet users. According to official statistics, number of Facebook users is more than 30 million and 90 percent of "online" citizens have Facebook account. With these figures, Turkey composes the 6 th biggest Facebook nation in the world.
Several scholarly works showed that Internet and Facebook provides new avenues for political participation and Facebook users are becoming more active in political life. This paper targets to answer the question about to what degree new avenues of political participation may affect the inertia of Turkish youth in political life. In order to answer this question we believe that multidimensionality of political participation in Facebook has to be accepted. Hence, we propose that different factors will affect different levels of political participation in different dimensions, in addition to some common factors.
Data collected with an online survey from 563 university students is employed to discover different dimensions of political participation in Facebook. Through multivariate techniques, we tried to show determinants of political participation by using socio-demographic and political factors in addition to usage patterns of Facebook. percent for the developed world. In Turkey, there are more than 30 million Facebook users, with a penetration ratio of 40 percent for overall population. Among online users, this ratio is 90 percent, meaning that almost every online user has a Facebook account. With these figures, Turkish users form the 6 th biggest Facebook nation in the world 1 .
In addition to this scale advantage, Facebook is important for political participation due to tools provided to users. Vitak and her friends list possible political activities as follows: "For example,users can join political groups, download candidate applications,and share their political opinions through the manycommunication tools on the site. Users can view their friends"activities by scrolling through the News Feed on their homepage, and they can comment on friends" posts, thus engagingin active conversation about political issues. From a resourceperspective, these affordances also offer affordable (i.e., free)opportunities to develop civic engagement skills with little tono additional time costs for users of Facebook, while simultaneously having access to a potentially large enough ""public"" to develop civic skills" (2011, p.2) Hence, it is expected that such a tool positively contributes to political participation level of its users.
In the Turkish case, this opportunity provided by Facebook is very crucial, because Turkish society is characterized with very low level of political participation as a result of several cultural and historical characteristics (Erdoğan, 2009). Last available surveys show that only 31 percent of Turkish citizens are interested with politics, proportion of those participated in meetings is less than 5 percent and only 7 percent of Turkish citizens are members of a civil society organization (Kalaycıoğlu, 2008). This lack of civic and political participation is also observable among young people. Several studies conducted during a decade showed that 10 percent of Turkish youth is a member of a political party, less than 7 percent of them actively participated in political campaign activities, meetings and boycotts are conducted by about 10 percent of them (Erdoğan, 2009).
Reasons of this political entropy are innumerable and policies to increase this score is beyond the limit of this paper. However, considering the opportunities provided by Internet and Facebook and very high level of Facebook penetration, we believe that understanding the dynamics of political participation in Facebook may provide significant information for those are interested with young political participation.

Methodology, Research Question and Hypotheses
Our research question in this paper may be stated as follows: "What are the factors affecting young people"s political participation in Facebook?" This major question is followed by another research question: "What are the dimensions of political participation in Facebook?" Within this research question, our major hypotheses are listed as follows:

H1: Political Participation in Facebook is affected by demographic and socioeconomic
factors.Considering the fact that politics is a matter of resource, it is expected that one having access to critical resources such as money, socioeconomic status or time, tends more to participate. Hence, it is expected that political participation in Facebook is more often among upper segments of society.

H2: Political Participation in Facebook is affected by political participation in the real life.
Political participation may be a medium-free action. People already participating to politics may continue to their activities in Facebook. All these hypotheses will be tested by using multivariate statistical techniques to prevent spurious correlation.

Methodology
Data presented in this paper is collected between 4th and 20th of April 2012, with participation of 563 university students having a Facebook account. According to Facebook  (Bryman, 2008;Salganik and Heackathorn, 2004).
In the recruitment of participants, different strategies are preferred by the research team.
Team prepared a Facebook advertisement and this advertisement is shown to target population by using Facebook advertising campaign management tool. According to Facebook statistics, this advertisement has been shown more than 3 million times to 392946 users with an average of 8 times per user and 1657 users have clicked on it. By using our database, we calculated that 219 people clicked this advertisement completed this survey. It means that our advertisement had access to 41 percent of total target population, 4 per thousand of them has clicked on it and 13 percent of those clicked completed the survey. We don"t have information about other Facebook campaigns to compare our reach statistics, but our previous research projects showed that expected return ratio is less than 6 per thousand by using Internet databases.
Parallel to Facebook advertisement, we conducted a snowball sampling methodology, a group of students are invited to complete this survey and they are asked to invite their friends to participate to this project. 344 students participated in the project and completed the survey. Snowball or Referral sampling strategy is highly criticized of not being representative and sometimes this methodology may create a bias. However we believe that it"s the most appropriate method for collecting data from unknown populations and lists are not available to prepare a sampling framework.
When we controlled the differences between these two strategies of sampling, we didn"t observe any significant difference on behavioral questions, therefore we can assume that these two sampling strategies accessed to same profile of users.
Online Journal of Communication and Media Technologies Volume: 4 -Issue: 3 -July -2014

Profile of Participants
Target population of the project was university students between 18-27 ages having a Facebook account. Basic demographic information about participants is as follows:  Average age of participants is 22.8 with a standard deviation of 3.4 years;  55 percent of them is man, and 45 percent is woman;  15 percent of participants are working parallel to university education;  21 percent of participants are from state universities of Istanbul, 38 percent are from private universities of Istanbul and remaining 40 percent is from universities abroad.
When education of parents are analyzed as an indicator of socioeconomic status, it"s observed that 40 percent of fathers have an education of less than secondary school, 30 percent of them have a secondary school graduation and percentage of those having a tertiary degree is 30 percent. 52 percent of mothers of participants have an education less than secondary school.
Those graduated from secondary school forms 29 percent and graduates of university and higher are only 16 percent.
Median household income is between 2000 and 2500 TL, while 30 percent of participants have a household income less than 1500 TL and 20 percent of them has an income higher than 4500 TL.

Dimensions of Political Activities in Facebook
There are alternative ways of political participation in Facebook, and it changes from being a passive spectator of political discussions, such as following political activities, being a "fan" of a political candidate; to share political information such as link, news column; or to take a part in deliberation through discussion or chatting. There is a room for further political involvement: user can form a political group or join a group formed by other people. In order to understand level of political involvement, we prepared a list of possible activities and asked participants how frequently they are involved in those activities, with a four point scale., based on a previous work conducted by (Vitak et. al., 2010). Below table presents our findings. Although people are open to state their own position; they don"t prefer direct discussion through Facebook chat (19 percent) or posting something about politics to another person"s wall (17 percent). This separation of activities gives some clues about "netiquette" of politics in Facebook. Participants can state their position, they can discuss openly; however direct discussion is not welcomed.
Apart this differentiation of activities, above table shows that political activities are not popular among Facebook users, the most popular activity has been performed by only one third of respondents. Our multivariate analyses target to show which factors affect this preference of participating in Facebook. Factor analysis of different political activities in Facebook produced three major dimensions of political participation explaining 70 percent of total variation. First dimension is highly loaded by posting (a status update, a photo, and a video), sharing (a link or a news column) and following. It explains 32 percent of total variation. Common characteristic of all these activities is indirect expression of political identity of respondent. By posting something or sharing a link or a news column; she expresses her political identity through reproduction/replication of available information. Her political identity becomes visible with her collection of these items. Since these items don"t include a direct reference to political identity of user; it is a way of indirect expression of who she is as a political person. Hence, this dimension is labeled as "indirect expression".
Second dimension explains almost 20 percent of total variation and it is correlated with becoming a fan of a political candidate, "liking" a page of a political candidate or adding/deleting an application with politics or political information. This kind of Facebook activities are more direct way of expressing political identities. If a user "likes" a page of a politician or political party, her political identity becomes apparent for other users. Similarly, adding some political information in her profile, directly inform other users about her political identity. Consequently, we labeled this dimension as "direct expression".
Final dimension explaining almost 20 percent of total variation is labeled as "political activity", because it is more correlated with actions related with political activism. For example, this dimension is correlated with political discussion (through Facebook chat or writing on other"s wall). It is more open to discussion and interaction with others. Difference Online Journal of Communication and Media Technologies Volume: 4 -Issue: 3 -July -2014 between first and third dimensions seems to be initiative taken by user. In the first dimension, user only puts some information on her page, and waits for intervention of other people. In the third dimension, user is intervening through writing on somebody else"s wall, chatting or forming a Facebook group.
These dimensions will be used as dependent variables in our multivariate analyses. analyzed, it is observed that, these kind of political activities were not popular. The most popular activities, "liking Facebook page of a candidate or a party" and "liking Facebook group of a candidate or a party" performed by only one third of participants. One quarter of respondents said they participated in Facebook group of a candidate or a party.
Recommendation, a relatively more active behavior only performed by 10 percent, indicating that our respondents preferred more passive ways of political participation or abstained from political participation.
We composed an index of political activity by simply counting activities performed by users.
This index changes between 0 and 4, meaning that if a respondent did all these activities, her score would be 4. The opposite pole shows that respondent never did them. Cronbach"s Alpha measuring reliability of this index is 0.81, showing a very high level of reliability.
Another possible factor pushing people to participate in Facebook is one"s perception about effectiveness of this tool. If she perceives Facebook as an effective environment for political activities, she will be tended to further political participation.

Political Values
In addition to basic demographic variables, a number of political and behavioral factors are included to the analysis. First set of independent variables target to measure political values of respondents. These variables are the self-positioning of the respondent on nationalism, leftright spectrum and conservatism scales. In this stage of analysis, a specific index to measure anomie is calculated.
"Anomie", a concept developed by Durkheim (1897), which means lack of social norms "normlessness". It describes a situation when social bonds between an individual and their community ties are broken and social identity of individuals is fragmented. This concept has been operationalized and measured by Sroles (1957)  is not dead to same degree. Newspapers and TV are still popular as political information sources, while radio or journals are not preferred.
In order to facilitate data analysis, number of variables is reduced by using factor analysis.
Factor analysis is generally used to reduce number of variables by using communalities between variables. Our analysis of political information sources produced two dimensions.

Media
Results of factor analysis are presented above with factor loadings. Factor loadings are coefficients varying between -1 and 1, when absolute value of coefficient approaches to 1, it indicated a stronger relationship, if it approaches to 0, and it indicates a weak or non-existent relationship. Factors are labeled according to factor loadings of variables. First factor, explaining 28 percent of total variance is highly correlated with Blogs, Twitter, Facebook and Websites. Consequently we"re labeling it as "unconventional media". Second dimension, explained 23 percent of variance is labeled as "conventional media" and it"s strongly correlated with TV, newspapers, radio and journals. This analysis shows that information sources per se segregated empirically. We will use these dimensions as explanatory variables in following sections. Political participation patterns of participants are presented above. Findings of above cited survey on political activities of Turkish youth is also included. In order to make a relatively meaningful comparison, we also included scores of a subset of the Turkish youth survey by filtering with employment status and age 2 Voting is the most popular political activity of survey participants. Almost all of them voted with a majority of 90 percent. This activity also was the most frequent political activity according to the 2008 Survey; however proportion of those voted was less than half of respondents.Membership in a political party -in youth branches or not-is not popular among participants. Less than 10 percent of Turkish youth is a member of a political party, and this ratio is 13 percent of our participants and equivalent subset. Active participation to a campaign is also not among popular activities, and our sample seems less active, compared to the equivalent sample.
On the other hand, above table shows that unconventional political participation is very high among respondents of our survey. Although only 8 percent of Turkish youth signed a petition about local problems, this score is 5 times higher in our sample. This statistics is 2 times higher than the subset"s average of 18 percent. The picture is same for the case of signing petition about a personal problem.Other dimensions of unconventional participation also expose such a difference. Percentage of those participated in a boycott is 32 percent (7 percent among Turkish youth), in a political demonstration is 48 percent (11 percent among Turkish youth).
According to these statistics, half of respondents participated in a protest activity conducted over the Internet; it was only 7 percent for Turkish youth and 18 percent of the equivalent sample. This difference may be attributed to increased penetration of the Internet. In 2008, Internet penetration was lower than 50 percent among the equivalent subset, and only one penetration is about 60 percent in Turkish households. Consequently, increased Internet penetration may lead to increased protest activity over the Internet. Below sections of the paper will provide further data to discuss this relationship. Factor analysis of political participation of respondents produced three different factors, parallel to previous studies (Erdoğan, 2001, Çarkoğlu, 2001and Erdoğan, 2003. First factor Online Journal of Communication and Media Technologies Volume: 4 -Issue: 3 -July -2014 explaining 25 percent of total variance is correlated with being a member of a political party (youth or not), canvassing and distributing brochures in a political campaign which are classified as "conventional methods of political participation". Consequently, we labeled this dimension as "conventional political participation".Second dimension is dominated by unconventional methods of political participation, such as participating in a political demonstration or a boycott. Participating in a protest activity over the Internet has also a higher factor loading with this dimension. This dimension is labeled as "unconventional political participation".
Final dimension explaining 18 percent of total variance is correlated with petitioning, about local or personal problems. This dimension is called as "petitioning".
The last general elections held on June 2011 give us an important opportunity to test political activities of participants. According to our statistics, 30 percent of participants tried to convince people for a candidate or a party. Same percent of them participated in a meeting of a candidate or a party. Another way of passive political activity, participating to a house meeting of a candidate or party is not popular: only 14 percent of participants had such an experience. It is clear that participants don"t prefer active involvement in political campaigns: only 8 percent of participants distributed brochures and only 5 percent participated in doorto-door canvassing activities.
By using answers given to this question, an additive index of political activity in 2011 general elections is calculated. If a respondent participated in all activities, her score of political activity is 5. If she didn"t participate in any of them, her score would be calculated as 0.
Reliability of this scale is measured as 0.71 with Cronbach Alpha coefficient. Average political activity score is 0.79 with a standard deviation of 1.2.

Facebook Usage and Attitudes
We believe that political activities in In order to discover alternative usage patterns and their effect on their political activities on the Facebook, we developed a set of questions to measure their usage preferences of different tools. Participating to social activities is relatively less popular, similar to updating profiles and photos and participating to social groups. One third of participants read their friends" walls but only 22 percent write on these walls. Forming social groups and inviting their friends and arranging social activities are least popular activities.

Arranging Updating Communicating
In order to discover common dimensions of Facebook usage patterns, a factor analysis has been conducted and three factors explaining 65 percent of total variation are derived. First factor has higher factor loadings from more active patterns such as forming a social group and inviting friends, arranging and participating in social activities. This dimension of Facebook usage explains 28 percent of variation, is labeled as "arranging" to refer its initiative taking nature.Second dimension is characterized with higher loadings from reading and writing her and her friends wall. It explains 21 percent of total variation and it"s labeled as "updating".Final dimension is related with sending messages, that"s why it"s labeled as "communicating".

Descriptive Statistics of Independent Variables
Below table presents basic descriptive statistics of independent variables to be employed in our models.

Determinants of Political Participation in Facebook: Multivariate Analyses
In order to discover determinants of political participation in Facebook, a series of multiple regression analyses are conducted. Multiple regression analysis is a tool allowing us to see direct effects of each independent on dependent variable, by controlling effects of other independent variables. We preferred to use Ordinary Least Squares regression method, assuming that correlations between independent variables are not too strong to violate basic assumptions of this method. Our analysis showed that correlation coefficients between independent variables are not greater than 0.2 with two exceptions: anomie"s correlation between political positioning on "Other vs. AKP" is 0.39 and correlation between Facebook usage intensity and Facebook usage pattern of updating is 0.35. Since these correlations don"t directly affect overall structure of equations, we didn"t exclude them from equations.
In our analysis, we had four different equations. First one is about determinants of political activity in Facebook during June 2011 general elections. We hypothesized that political activity in Facebook is a function of one"s general political activities in Facebook, her perception about political power of Facebook and her real life political activities during these general elections.    When Participation dimension is taken as the dependent variable, it"s observed that demographic and socioeconomic factors continue to be statistically significant. Male users tend to participate more than female ones. As education of parents of participants increases, their scores in this dimension of political participation also increases. These differences may be understood as evidence for validity of resources hypothesis.
Political preferences and values are not effective factors, meaning that level of anomie or positioning in political spectra are not statistically significant. On the other hand, users of unconventional media have higher scores than other users. As new media becomes the major source for information; they tend to participate more.Among different dimensions of political participation; conventional political participation showing participants" current political Online Journal of Communication and Media Technologies Volume: 4 -Issue: 3 -July -2014 activities is statistically significant, leading us to conclude that this kind of political participation in Facebook is a continuum of political activities in the real life.
As Facebook usage intensity increases, people"s tendency to participate in these terms increases. Our major finding from this equation is that different styles of Facebook usage have different effects on political participation. Users having higher scores in Arranging dimension have a higher tendency to participate. And, users spending more time on updating also have higher participation scores.

Discussion
Our paper targets to discover dimensions of political participation in Facebook and their determinants. Data collected from university students using Facebook showed that political participation in Facebook is multidimensional and each dimensions are affected by different factors.
Our analysis showed that it is possible to define three different dimensions of political participation in Facebook (H4). A Facebook user may prefer to indirectly express her political opinions by posting, sharing and following; without declaring her political preferences. Or, she may prefer to openly state her political position by becoming a fan of a political candidate or liking him. Third option is more active compared to previous ones. She can discuss politics with her friend through messaging or chatting. Forming a political group or joining to a political group are grouped under this dimension.
Determinants of different dimensions of political participation in Facebook are also different.
Our multivariate analyses showed thatIndirect Expression is affected by political preferences, media usage, and political participation in real life (H2). Different styles of Facebook usage have different effects. As respondents tend to use Facebook for updating and communicating, they tend to prefer this kind of political participation (H3).There is a significant difference between male and female respondents in regards to their Direct Expression of their political identities (H1). Political preferences also affect this dimension. However, conventional political participation is the biggest factor, showing that this kind of political activity is the continuum of real life political activities of the respondent. Those tend to be more active in conventional politics tend more to use these activities (H2). Facebook usage intensity and style are also affecting factors.
Demographic and socioeconomic variables significantly affect active participation dimension. Parental education showing a higher socioeconomic status is statistically significant (H1). Usage of unconventional media leads to more active involvement in politics, similar to conventional political participation (H2). As Facebook usage intensity increases, tendency to active participation also increases. Meanwhile, people that arrange activities in Facebook are more open to participate in this dimension, meaning that dimensions of political participation in Facebook matters (H3).