Social networking sites and offline political ...

The Social Science Journal xxx (2016) xxx–xxx

Contents lists available at ScienceDirect

The Social Science Journal journal homepage: www.elsevier.com/locate/soscij

Leveling or tilting the playing field: Social networking sites and offline political communication inequality Jaeho Cho a,1 , Heejo Keum b,∗ a

Department of Communication, University of California, 1 Shields Ave., Davis, CA 95616, USA Department of Journalism and Mass Communications, Sungkyunkwan University, 50505 First Faculty Hall, 25-2 Sungkyunkwan-Ro, Jongno-Gu, Seoul, South Korea b

a r t i c l e

i n f o

Article history: Received 20 August 2014 Received in revised form 20 December 2015 Accepted 19 January 2016 Available online xxx Keywords: Social networking sites Communication gap Political discussion gap Political posting Political discussion Political inequality

a b s t r a c t Building on a resource theory, this study investigates (a) how individuals’ socio-economic status is related to political communication in offline situations and on social networking sites and (b) whether political expression on SNS improves socio-economic stratification in offline political discussion. Analyses of a national survey demonstrate that the impact of individuals’ socio-economic status (SES) is much weaker on political expression via SNS than on offline political discussion. It is also found that the political use of SNS reduces the strength of the link between individuals’ SES and offline discussion. Implications of these findings for the Internet and political inequality are discussed. © 2016 Western Social Science Association. Published by Elsevier Inc. All rights reserved.

1. Introduction One common observation in American politics is that citizens from disadvantaged backgrounds are significantly less vocal in the governing process than the more affluent (Milbrath & Goel, 1977; Stockemer, 2014; Verba, Schlozman, & Brady, 1995). This inequality worries scholars because it leads to unbalanced government responsiveness and limits democratic legitimacy (Dahl, 1989; Lijphart, 1997). Verba (2003) argues, for example, “of the various ways in which US citizens can be unequal, political inequality is one of the most significant and troubling” (p. 663). Given this concern, research has sought to understand how

∗ Corresponding author. Tel.: +82 2 760 0689. E-mail addresses: [email protected] (J. Cho), [email protected] (H. Keum). 1 Tel.: +1 530 754 0975.

the inequality develops and persists and how to ameliorate it (Barber, 2001; Bonfadelli, 2002; Hargittai, 2008a; Krueger, 2002; Schlozman, Verba, & Brady, 2010). In this context, the exponential growth of Web-based political activities over the past decade raises questions about what the Internet brings to long-standing political inequality (Boulianne, 2009 for a meta analysis). Building on this research, the present study expands the context of political inequality to include citizen political communication and investigates whether and how political expression on social networking sites (SNS) ameliorates discussion deficit among those from disadvantaged backgrounds. Indeed, similar to the pattern of socio-economic stratification in participation, face-to-face political discussion is unevenly distributed in the population, with those with higher socio-economic statuses (SES) being more active than their low status counterparts (Fraser, 1992; Young, 1996). The gap in political discussion is not just a matter of communication. Given that political

http://dx.doi.org/10.1016/j.soscij.2016.01.002 0362-3319/© 2016 Western Social Science Association. Published by Elsevier Inc. All rights reserved.

Please cite this article in press as: Cho, J., & Keum, H. Leveling or tilting the playing field: Social networking sites and offline political communication inequality. The Social Science Journal (2016), http://dx.doi.org/10.1016/j.soscij.2016.01.002

2

J. Cho, H. Keum / The Social Science Journal xxx (2016) xxx–xxx

discussion is essential to citizen competence and participation (Conover, Searing, & Crewe, 2002; Mansbridge, 1999; Marques & Maia, 2010), this communicative inequality implies a socio-economic stratification in a basic and fundamental feature of democratic citizenship. Relative to the participation gap, however, the issue of communication gap has received little attention in the literature. Recognizing this, we examine whether political communication on SNS is stratified by socio-economic status and whether the SNS use ameliorates the discussion deficit among those from disadvantaged backgrounds. By shifting the focus from formal participatory activity to citizens’ communicative engagement, the present study adds another layer to the discussion of political inequality. Another contribution of this study to the literature is that it provides a direct empirical examination of the possibility that online political behavior narrows offline political inequality. Although much discussed, this possibility has rarely been tested in previous research. Indeed, past studies have largely focused on the relationship between SES and online political engagement. A positive link is considered an indication of participatory inequality online while a non-significant association is interpreted as evidence toward equality (Krueger, 2002; Schlozman et al., 2010). Although non-stratified political engagement online has the potential to improve political inequality offline, it remains untested. The present study fills this gap in the literature by testing whether citizens’ political use of SNS influences the associations between their socio-economic status and their offline political discussions. In the sections that follow, this study builds on a resource theory to discuss how individuals’ SES is related to political communication in their offline lifeworld and in their computer-mediated social space. Next, we predict that the impact of SES on political communication will differ depending on the mode of communication. We further discuss how political expression via SNS reduces the extent to which SES is related to everyday face-to-face political discussion. We then turn to the 2012 Pew Internet & American Life Project data to empirically examine our theoretical discussions. 1.1. Socio-economic status and offline political discussion Early studies find that individuals’ socio-economic status is a strong predictor of political participation (Lane, 1959; Milbrath & Goel, 1977; Verba, Nie, & Kim, 1971). Although this research successfully identifies likely voters, the question as to why structural characteristics exert such influence is never fully answered. Later work searches more carefully for explanatory mechanisms underlying the socio-economic stratification of participation, which culminates in a civic voluntarism model (Verba et al., 1995). The work by Verba and colleagues specifies three factors as accounting for participation—direct resources such as time, money, and civic skills; psychological engagement with politics; and networks of political recruitment. The model suggests that citizens’ voices are unequal because the resources that facilitate political participation are not equally distributed across the population. High SES citizens, as compared to their low SES counterparts, tend to be

more interested and better versed in political issues, have more opportunities to develop organizing and communications skills, and have more access to social networks where civic and political engagement is the norm. The resources and opportunities disproportionately available to those in the higher socio-economic strata make political participation more affordable and feasible for them than for others in lower strata. The likelihood of political discussion is also largely explained by the resources necessary to participate in the political process. Individual resources like civic skills and psychological engagement with politics foster political discussion. Likewise, spare time (Nie & Hillygus, 2002; Putnam, 1995; cf. Robinson & Haan, 2006) and social networks (Huckfeldt & Sprague, 1995) are necessary conditions for political discussion to occur. Especially, given the interpersonal nature of political discussion, the availability of discussants geared for political talk is crucial (Sinclair, 2012). Because these individual-level and sociallevel resources are concentrated amongst high SES citizens, everyday political conversation, as in the case of participation, is more likely to occur among those in the higher strata (Fraser, 1992; Young, 1996). This SES-based explanation of communication gap has been recognized in the literature of political communication. The knowledge gap hypothesis, for example, posits that individuals’ SES is systematically associated with interpersonal discussion on public affairs issues, which ultimately leads to a knowledge gap (Tichenor, Donohue, & Olien, 1970). Tichenor et al. (1970) note, “education generally indicates a broader sphere of everyday activity, a greater number of reference groups, and more interpersonal contacts, which increase the likelihood of discussing public affairs topics with others” (p. 162). A communication mediation model by McLeod et al. (2001) also illustrates how individuals’ socio-economic position is related to political discussion. Citizens with higher education and income tend to be more active than others in using news media, which in turn spurs political conversation (Eveland, Hayes, Shah, & Kwak, 2005). Taken together, the resource theory and related literature suggest that socio-economic advantages translate into information and communication advantages, which leads to the following hypothesis: H1. Individuals’ socio-economic status will be positively related to the frequency of political discussion. 1.2. Socio-economic status and political expression on SNS Over the past decade, SNS have quickly gained popularity at all SES levels. According to the 2012 Pew Internet and American Life Survey, two thirds of Internet users use a social networking site of some kind and these SNS users are not stratified by levels of education or income (Hargittai, 2008b). Among Internet users, SNS adoption is almost equal between those with a “high school education or less” (66%) and those with a “college degree or above” (65%). Similarly, adoption rate does not increase with levels of household income. Rather, use of SNS is more common among Internet users with annual



household incomes of less than $30,000 (72%) than those with higher income levels (65–66%), which might be a reflection of the fact that SNS are more popular among young adults whose income is generally lower than their older counterparts. Given the diversified socio-economic makeup, it is less likely that individuals’ socio-economic status will determine their adoption of SNS. Yet, the recent observation that SNS are widely adopted among Internet users does not necessarily suggest that users at various socio-economic strata are equally active in political communication on SNS. High SES users already possess the resources necessary for the communicative actions, much of which can apply to communication on SNS. For example, individual resources such as civic skills and psychological engagement matter not only for face-to-face political discussion but also for networking and communicating with others about political issues in computer-mediated social space. When viewed from the resource perspective, a lack of these relevant resources leads to a lack of political use of SNS. Supporting this view, available research suggests that it is user motivation that accounts for the way the Internet is used and its political consequences. Prior (2007), for instance, shows that users with surveillance motivation engage in information seeking and political communication more actively than those with entertainment motivation, which ultimately results in a gap in political involvement. In sum, citizens from privileged backgrounds already possess the resources that foster communicative actions. When a new communication technology like SNS is introduced, these citizens draw upon the same resources to take advantage of the new opportunity. Although SNS attract users from different corners of the socio-economic spectrum, high SES citizens are still better positioned to make use of SNS to voice their opinions. On the other hand, relative to offline political discussion, citizen political communication via SNS is less likely to be stratified by social class. The networking tools and technologies provided by SNS make it easier for lower SES citizens to access social networks where political discussion is likely to occur. Indeed, one of the advantages granted to higher status citizens is that they are better equipped with social networks whose members have individual resources that facilitate political engagement (Tichenor et al., 1970). When surrounded by the politically sophisticated and engaged, citizens are more likely to be recruited into political life. They learn from their peers not only about political issues but also about the norms of citizenship, both of which encourage political discussion and engagement (Huckfeldt & Sprague, 1995; McLeod et al., 1999; Sinclair, 2012). Limited access to this social resource (or opportunity) is one of the deficiencies that make political discussion less likely for lower SES citizens. By lowering the social and psychological barriers to networking, however, SNS have the potential to fill the gap, allowing low SES users to make connections with people who would normally be outside of their social sphere (Conroy, Feezell, & Guerrero, 2012; Ellison, Steinfeld, & Lampe, 2007; Kim, Hsu, & Gil de Zuniga, 2013; Parks, 2011; cf. Mesch, 2012; Valenzuela, Park, & Kee, 2009). On SNS, the underprivileged users, if interested, can easily develop

3

connections with politically active users or groups; they can even become a “friend” of high-profile political candidates or the President (Peterson, 2012). Furthermore, even without political interest or motivation to seek political networks, low SES users might stumble across individuals/groups that are highly interested or directly involved in politics while surfing the profiles of SNS friends (Brundidge, 2010; Utz, 2009). This inadvertent encounter with political networks (or political messages at least) provides an opportunity for users from less advantaged backgrounds to engage in political activities at least in computer-mediated social space. In sum, it is often challenging for low SES citizens to find and become involved in relevant political networks in the offline environment, even when they are politically interested (Verba et al., 1995). However, the enhanced networking and communication tools offered by SNS help low SES users overcome their social resource deficiency and catch up with high SES users online. Thus, we expect that the overall resource advantage in favor of high SES citizens is not nearly as great on SNS as it is in offline face-to-face situations. Based on this reasoning, the following hypothesis is proposed: H2. The extent to which individuals’ socio-economic status is associated with political expression on SNS will be weaker than it is with offline political discussion. 1.3. SNS and socio-economic stratification in offline political discussion If political use of SNS is less constrained by users’ socioeconomic status as hypothesized above, it suggests that SNS have the potential to improve the socio-economic stratification in face-to-face political discussion. Some of the resources that foster political conversation are often developed through communicative actions. Political discussion helps citizens acquire and sharpen communications skills and civic norms (Gamson, 1992; Walsh, 2004). Likewise, via political discussion, citizens engage in various intrapersonal mental processes including learning (Eveland, 2004), reflective reasoning (Cho et al., 2009), cognitive integration and differentiation (McLeod et al., 2001), and composition of ideas for expression (Pingree, 2007). In the same vein, Benhabib (1996) highlights the deliberative virtue of expressive behavior, noting “when presenting their point of view and position to others, individuals must support them by articulating good reasons in a public context to their co-deliberators. This process of articulating good reasons in public forces the individual to think of what would count as a good reason for all others involved” (pp. 71–72, emphasis in original). Engaging in conversation or at least an act of expression, therefore, provides an opportunity for citizens to develop, practice, and renew the internal resources such as skills, cognitive sophistication, and motivational engagement, which in turn foster further political discussion. It is this reciprocity between resources and political discussion that suggests that SNS could provide a platform to ameliorate the existing political discussion disparities between high and low SES citizens (Schlozman et al., 2010). The enhanced and open opportunities for communication and networking on SNS have the


4


potential to make the traditional resource deficiencies less of an obstacle to engaging in political expression on SNS for low SES citizens. The less stratified political communication on SNS in turn provides them with compensatory opportunities to exercise civic and communication skills and gain individual resources, which they might not otherwise have the chance to develop (Conroy et al., 2012; Tang & Lee, 2013). These SNS-acquired resources would then encourage them to participate in offline political discussion. Thus, as political use of SNS increases, socio-economic distinctions will become less salient in offline political discussion. Based on this reasoning, we propose the following hypothesis: H3. The influence of individuals’ socio-economic status on offline political discussion will diminish as political expression on SNS increases. 2. Method 2.1. Data To test the hypotheses, we use one of the 2012 Pew Internet & American Life Project national surveys, entitled Search, Social Networks, and Politics. Drawing on a combination of landline and cellular random digit dialing samples between January 20 and February 19, 2012, this survey interviews 2253 adults, age 18 and older, in English or Spanish (Appendix A for a full statistical description of the demographic profile of the sample). Cooperation rates are 19.5% and 18.5% for the landline and cellular samples, respectively (Rainie & Smith, 2012). Given that the focus of the present study is on SNS, our analyses only utilize data from respondents who use a social networking site of some kind (N = 1047). Further reductions in sample occur in the results of some statistical analyses because of missing data1 . 2.2. Measures Three groups of variables are constructed: (a) citizen political communication—offline political discussion and political expression on SNS; (b) socio-economic status—education, income; and (c) control variables.

1 As a way to handle the missing data, we imputed simulated values to replace missing values in variables. Following Myers’ hot deck procedure, we used a multiple imputation technique based on the EMis (Expectation Maximization with Importance Sampling) algorithm (see King, Honaker, Joseph, & Scheve, 2001; Myers, 2011). The multiple imputation was implemented in two steps. First, multiple (typically five) imputed values were generated for each missing item, yielding multiple imputed replicate datasets. Then, the missing values were replaced with each of the five imputed values, while the observed values remained the same. Second, the specified regression models were run for each of the imputed datasets with no missing data. For the purpose of comparison, the same models were run again using listwise deletion, with observations from any missing item being excluded from the analysis. Little difference was found between the two approaches (multiple imputation and listwise deletion). Given that the two approaches produced nearly identical results, we decided to report the original results based on the listwise deletion method.

2.2.1. Citizen political communication Offline political discussion is measured by asking respondents how often they talk about politics or current events with family and friends. Responses are recorded on a four-point scale with 1 being “Never” and 4 being “Very often.” Political expression on SNS is measured by asking respondents how much of what they have posted on social networking sites, such as status updates, comments, or links to news stories, is related to politics, political issues, or the 2012 elections. Responses are measured on a fourpoint scale with 1 being “Not at all” and 4 being “Most or almost all of it.” Both measures are rescaled to run from 0 to 1 (M = .68, SD = .31 for offline discussion; M = .21, SD = .30 for SNS expression). 2.2.2. Socio-economic status An index of SES is constructed by combining education and income. Respondents’ level of education is measured on a seven-point scale ranging from “None, or grades 1–8 (1)” to “Post-graduate education after college (7).” The sample mean is 4.94 which falls between “Technical, trade, or vocational school after high school (4)” and “Some college, no 4-year degree (5)” (SD = 1.57). Income is measured as the total family income on a nine-point scale, with 1 being “Less than $10,000” and 9 being “$150,000 or more.” The sample mean of 5.41 is between the “$40,000 to under $50,000 (5)” and “$50,000 to under $75,000 (6)” brackets (SD = 2.38). Because education and income are measured on different scales, the two measures are standardized prior to being combined (inter-item correlation = .440). 2.2.3. Intervening and control variables The main premise of the present study is that the differences in income levels and educational attainment across the citizenry create unequal psychological and social resources, which in turn translate into inequalities in communication. Yet, because of socioeconomic status’ broad impact, education and income are related, either causally or not, to other variables that might influence online/offline citizen communication. It is thus useful to control for nondemographic variables that might account for part of the income and educational impact. This allows us to eliminate alternative explanations and narrow down the analysis to the hypothesized relationship between socioeconomic status and citizen communication. In this effort, we employ a number of attitudinal and behavioral variables and demographic variables other than education and income for control purposes. First, basic demographic variables such as age (M = 42.5, SD = 15.9), sex (a dummy variable with female coded 1; 56% female), race (a dummy variable with White coded 1; 76% White), marital status (a dummy variable with married coded 1; 52% married), and employment status (a dummy variable with employed coded 1; 66% employed) are considered. Second, to hold technological proficiency and familiarity constant in the analyses, general use of SNS, technology adoption, and informational use of online search engines are considered as control variables. Frequency of SNS use is measured by asking respondents how often, on a six-point scale ranging from “Less often (1)” to “Several



times a day (6),” they visit social networking sites (M = 4.45, SD = 1.55). Technology adoption is measured by asking respondents whether they use a smartphone (57% yes) and a tablet computer (22% yes). Search engine use is measured by asking how often, on a seven-point scale ranging from “Never (1)” to “Several times a day (7),” respondents use search engines to find information online (M = 5.67, SD = 1.55). Third, we consider respondents’ political orientation (party identification, partisanship strength), which is known to covary with socio-economic status and political behaviors. Party identification is measured on a five-point scale with 1 being Democrat and 5 being Republican (M = 2.85, SD = 1.62). The partisanship strength is measured on a three-point scale by folding the party identification measure (M = 2.44, SD = .74). Lastly, life satisfaction is employed as a control variable because it is often considered a correlate of SES (Ahuvia & Friedman, 1998), SNS use (Correa, Hinsley, & Gil de Zuniga, 2010), and political engagement (Shah, Kwak, & Holbert, 2001). It is measured by asking how respondents would rate the overall quality of life for themselves and their family on that day. Answers are recorded on a fivepoint scale from “Poor (1)” to “Excellent (5)” (M = 3.50, SD = 1.07).

2.3. Analytic strategy To test the hypotheses, hierarchical multiple regression analyses are conducted, with SES placed in the initial block (Model 1), demographic controls added in the second (Model 2), life and political orientations (i.e., life satisfaction, party identification, and partisanship strength) in the third (Model 3), and technology use and adoption (i.e., overall SNS use, online search engine use, smartphone and tablet computer adoption) in the last (Model 4). The rationale for adopting the hierarchical modeling approach is to explore the possibility that some of the variables considered for the control purposes may mediate the impact of SES on the outcome variables (i.e., SNS political expression and face-to-face discussion). Specifically, measures of technology use and adoption are considered as potential mediators that may explain some of the mechanisms for the link between SES and citizen communication. Results from hierarchical regression analysis demonstrate (a) the initial relationship (or total effect) between the main IV (i.e., SES) and the DV (i.e., citizen communication) before considering mediators and (b) the partial relationships between mediators and the DV in the final model. The results also show, though not testing statistically, how much of the initial relationship between the IV and the DV is mediated by the variables entered in the later block (i.e., mediators), as indicated by the change in the effect of the IV on the DV between the earlier and later models. Based on the results of hierarchical regression analysis, a formal testing of indirect effects is performed using PROCESS, a modeling tool specialized for mediation and moderation analysis (Hayes, 2013).

5

3. Results 3.1. Face-to-face political discussion The first hypothesis is tested by a regression equation where the frequency of offline political discussion is accounted for by individuals’ socio-economic status (H1) and a variety of controls. As reported in Table 1, SES is a significant predictor of offline discussion (ˇ = 242, p < .001) even after considering demographic differences and life/political orientations (Model 3). That is, citizens with a higher SES are more likely to engage in political conversation than their low SES counterparts, which creates a gap in everyday political communication among citizens. When all of the potential mediators are considered simultaneously, the effect of SES is reduced to some degree but still remains substantial and statistically significant (ˇ = 207, p < .001) (Model 4). Of the potential mediators, online search engine use (ˇ = 144, p < .001) is a significant predictor of offline political discussion. Taken together, the results point to the possibility that an intervening variable (i.e., search engine use) might mediate the impact of SES on offline political discussion. As a second step, the mediating path from SES to offline political talk via search engine use is assessed with PROCESS. Results show a significant mediation (z = 3.65, p < .001), where SES is significantly associated with search engine use, which in turn is significantly related to offline talk. The effect of SES on offline political discussion is beyond those of other individual differences that may potentially confound the relationship between SES and political discussion. Although not the focus of this study, some findings about the influence of these control variables are worth briefly noting. First, consistent with previous literature about demographic profiles of the politically engaged (Burns, Schlozman, & Verba, 2001), this study finds that other demographic factors such as sex, age, and marital status matter for political discussion. Male citizens are more likely to have political conversations than female citizens. Likewise, the elderly are more active in discussing politics than their younger counterparts. Those who are married tend to discuss political issues more than those not married. Second, identification with either the Democratic or Republican party does not make a difference, but the strength of partisanship does. Party identifiers, whether Republicans or Democrats, are more active in political discussion than leaners or independents. This finding suggests that psychological attachment to a political party might function as a motivational force driving communicative actions. 3.2. Political expression on SNS The second hypothesis predicts that the influence of SES on political expression via SNS will be weaker than on offline political discussion. To test this hypothesis, a regression model is estimated with the dependent variable being political expression on SNS and the independent variables being individuals’ socio-economic status. The same set of control and mediating variables is employed as in the regression model testing H1. As shown in Table 2, SES is not


6


Table 1 Hierarchical multiple regression analysis: offline political discussion. DV: Offline political discussion

Block 1: socioeconomic status SES Incremental R2 (%)

Model 1

Model 2

Model 3

.315*** 9.9

.264***

.242***

Block 2: demographics Age Sex (Female = 1) Race (White = 1) Employment status Marital status Incremental R2 (%)

.119*** −.062 −.008 −.046 .062 2.7

Block 3: life and political orientations Life satisfaction Party ID (Republican = high) Partisanship strength Incremental R2 (%)

Model 4 .207***

.093** −.076* .027 −.040 .068

.128*** −.068* .019 −.037 .078*

−.018 −.063 .214*** 4.9

−.022 −.048 .206***

Block 4: technology use and adoption General SNS use Online search engine use Smartphone use Tablet computer use Incremental R2 (%)

.057 .144*** −.047 −.018 2.4

Total R2 (%)

19.9

Notes: Data in this model are from the 2012 Pew Internet & American Life Project national surveys, entitled Search, Social Networks, and Politics (http://www.pewinternet.org/datasets/february-2012-search-social-networks-and-politics/). Entries are standardized coefficients. N = 866. Significance levels are as follows: * p-value < .05; ** p-value < .01; *** p-value < .001.

Table 2 Hierarchical multiple regression analysis: political expression on SNS. DV: SNS political expression Model 1 Block 1: socioeconomic status SES Incremental R2 (%) Block 2: demographics Age Sex (Female = 1) Race (White = 1) Employment status Marital status Incremental R2 (%) Block 3: life and political orientations Life satisfaction Party ID (Republican = high) Partisanship strength Incremental R2 (%) Block 4: technology use and adoption General SNS use Online search engine use Smartphone use Tablet computer use Incremental R2 (%) Total R2 (%)

.013 .0

Model 2

Model 3

Model 4

.035

.040

.011

−.023 −.105** −.069* −.010 −.015 1.8

−.050 −.110*** −.041 −.005 −.002

.020 −.107*** −.046 .001 .002

−.101** −.051 .141*** 3.0

−.108** −.028 .130***

.222*** .093** −.053 .041 6.2 11.0

Notes: Data in this model are from the 2012 Pew Internet & American Life Project national surveys, entitled Search, Social Networks, and Politics (http://www.pewinternet.org/datasets/february-2012-search-social-networks-and-politics/). Entries are standardized coefficients. N = 857. Significance levels are as follows: * p-value < .05; ** p-value < .01; *** p-value < .001.



associated with SNS political expression (ˇ = 040, n.s.) in the initial model with basic demographics and life/political orientations as controls (Model 3). Unlike offline political discussion, the result suggests that political expression on SNS is not stratified by users’ socio-economic status. When potential mediators are added, the already weak and nonsignificant relationship between SES and SNS expression shrinks further (ˇ = 011, n.s.) (Model 4). Of the potential mediators, two variables are significant in predicting SNS expression: general SNS use (ˇ = 222, p < .001) and search engine use (ˇ = 093, p < .01)2 . Taken as a whole, the patterns of relationships speak to mediation possibilities. Even though the initial relationship (the total effect) between SES and SNS expression is not significant, mediation still might underlie the SES-SNS expression relationship, especially when positive and negative indirect effects cancel out the mediation (Hayes, 2013). The mediation possibilities are tested with PROCESS. Results show that both mediation paths are statistically significant (i.e., SES > search engine use > SNS expression, z = 2.40, p < .05; SES > general SNS use > SNS expression, z = −2.25, p < .05) but in the opposite direction. This is because SES is positively associated with online search engine use but negative ly associated with general SNS use, even though the two technology-use variables are positive predictors of SNS expression. It appears that the positive and negative mediations explain why the total effect of SES on SNS expression, which is the sum of indirect effects, is not significant. Also, noteworthy is that adoption of communication technologies such as smartphones and tablets does not mediate SES effects on citizen communication. Although SES is significantly related to adoption of both smartphones and tablets, when assessed simultaneously with all other variables considered in the model, the adoption of technologies is not a significant predictor of citizen communication, either offline or in an SNS context. Overall, the hierarchical regression and mediation analyses shed light on some mechanisms underlying the link between SES and citizen communication. Individuals with higher socioeconomic backgrounds are more likely than lower SES counterparts to use online search engines. This specific type of Internet use explains the positive link between SES and citizen communication in both the offline and SNS settings. On the other hand, data suggest that, people from lower socioeconomic backgrounds tend to use social media more often than their higher status

2 Some of the control variables – sex and partisanship strength – are significant factors in accounting for political expression both on SNS and in the offline discussion model. Interestingly, the data show a gender gap in political expression on SNS. As in face-to-face political discussion, female citizens are less likely to make use of SNS for political expression than male citizens. Party identifiers are more active in political expression via SNS than leaners or independents. Some differences between the offline and SNS models in terms of the influences of control variables are also observed. Age and marital status are significant factors for offline political discussion whereas neither of them is associated with political expression on SNS. Lastly, the degree of life contentment, although not related to offline political discussion, is negatively associated with SNS political expression. This finding is consistent with previous work on psycho-demographic profiles of SNS users (Correa et al., 2010).

7

counterparts. In turn, social media use creates opportunities for users to engage in political expression. Not surprisingly, however, general SNS use is not associated with offline political discussion. All in all, the negative indirect effect of SES on SNS expression via general SNS use suggests that social media help the socioeconomically underprivileged catch up on political communication in the virtual sphere. Along with the detailed understanding of the relationship between SES and citizen communication, the overall results provide robust evidence that SES is a positive predictor of offline political discussion but not of political expression via SNS. That is, although there exists some mechanism – i.e., Internet use driven by surveillance motivation such as search engine use – that links socioeconomic advantages to political communication advantages, expressions on the online sphere, especially on social media, are not determined (or stratified) by citizens’ socioeconomic resources. When comparing results for offline discussion and SNS expression, it is clear that the coefficients for SES are quite different between the two models. The coefficient for SES is positive and statistically significant in predicting offline political discussion (Table 1) yet are non-significant when predicting political expression on SNS (Table 2). Although this descriptive comparison of regression coefficients across the two regression models is useful in showing the pattern of difference, it does not test whether the differences are large enough to be statistically significant. To formally test the differences in regression coefficients, we estimate a regression equation in which the two regressions are forced into the same analysis by using the respondent ID to tie the paired observations to each other and including a dummy variable that indicates whether a given response is for the face-to-face or the social media setting. To statistically compare slopes, the analysis includes an interaction between the dummy variable of communication setting (offline = 0; SNS = 1) and SES and allows for the fact that the residual standard deviation may be different for the two settings. The results reveal that there is a significant, negative interaction between SES and the setting (b = −.059, SE = .007, p < .001), demonstrating that the qualitative difference in regression coefficients for SES in Tables 1 and 2 (.207 in the offline model vs. .011 in the SNS model) is statistically meaningful. Thus, lending empirical support to H2, the formal testing of this differential pattern by interaction between individuals’ SES and the communication setting further confirms that the influence of SES on citizen communication is significantly reduced on SNS as compared to face-to-face. 3.3. Political expression on SNS and the political discussion gap The last hypothesis (H3) looks at whether and how political expression on SNS moderates socio-economic stratification in offline political discussion. To test the hypothesis, a regression model is specified where individuals’ socio-economic status interacts with SNS political expression to impact offline political discussion. Consistent with H3, the estimated coefficient for the interaction


8


Table 3 Interaction between SES and SNS use on offline political discussion. Political discussion offline Baseline model Interactions b/w SES and SNS expression SES Political expression on SNS SES × political expression on SNS Control variables Age Sex (Female = 1) Race (White = 1) Employment status Marital status Life satisfaction Party ID (Republican = high) Partisanship strength General SNS use Online search engine use Smartphone use Tablet computer use R-squared (%)

Interaction model

.202*** .339*** –

.244*** .342*** −.077*

.124*** −.031 .032 −.041 .077* .016 −.037 .162*** −.019 .111*** −.028 −.029 30.0

.124*** −.032 .033 −.042 .076* .017 −.033 .160*** −.011 .108*** −.025 −.026 30.4

Notes: Data in this model are from the 2012 Pew Internet & American Life Project national surveys, entitled Search, Social Networks, and Politics (http://www.pewinternet.org/datasets/february-2012-search-socialnetworks-and-politics/). Entries are standardized coefficients. N = 863. Significance levels are as follows: * p-value < .05; *** p-value < .001.

between SES and political expression on SNS is negative (ˇ = −.077) (Table 3). As seen in the left panel of Fig. 1, the negative interaction suggests that the impact of SES on offline political discussion diminishes as political expression on SNS increases. The significance test further confirms that the reduction in the impact of SES on offline discussion per one unit increase in SNS expression is statistically greater than zero (t = −2.20, p < .01). To better understand the pattern of interaction in the offline discussion gap between highand low-SES users by levels of SNS expression, we flip SES and SNS expression in a graphical illustration. As seen in

the right panel of Fig. 1, the significant, negative interaction suggests that the offline discussion gap between high and low SES groups narrows when SNS expression is high. That is, using SNS for political expression ameliorates socio-economic stratification in everyday political communication. 4. Discussion Whether the Internet reduces or reproduces the existing political inequality in society is one of the central questions in the study of the Internet and democracy (DiMaggio, Hargittai, Neuman, & Robinson, 2001). Current scholarship provides mixed conclusions to this question. On one side, some suggest that political inequality will grow, as the Internet becomes a primary tool or domain of political engagement (Bonfadelli, 2002; Hargittai, 2008a; Schlozman et al., 2010). This is because the structural causes of the inequality still exist, shaping the way the Internet is adopted and used. By contrast, other efforts to clarify the role of the Internet in political inequality suggest that the Internet contributes to a narrowing of the participation gap (Barber, 2001; Delli Carpini, 2000; Krueger, 2002). Internet-based communication and participation are less constrained by class-related resources, a fact that helps the disadvantaged catch up with the privileged in the online political sphere. Within this context, the current study parses the role social networking sites play in political inequality. As hypothesized, the impact of individuals’ socio-economic status is much weaker on political expression via SNS than on face-to-face political discussion. Indeed, the data in this study reveal that individuals’ SES, a predictor of face-toface political discussion, is not associated with political expression on SNS. The finding that socio-economic distinction is not salient in political use of SNS meshes well with past research suggesting that SNS help users build social capital and form new connections, whether

Fig. 1. Interaction between education and SNS expression. Notes: Data in this model are from the 2012 Pew Internet & American Life Project national surveys, entitled Search, Social Networks, and Politics (http://www.pewinternet.org/datasets/february-2012-search-social-networks-and-politics/). The interaction between SES and SNS expression is statistically significant (ˇ = −.077, t = −2.20, p = .028).



intentionally or inadvertently (Ellison et al., 2007; Parks, 2011; cf. Valenzuela et al., 2009). If SNS use mirrored offline social relationships, offline inequality in citizen communication would likely be reproduced on SNS. Our study, however, shows that this is not the case. Instead, our analyses demonstrate that political expression on SNS reduces the strength of the link between individuals’ SES and faceto-face political discussion. That is, using SNS for political expression helps narrow the SES-based gap in offline political discussion. Although discussed and implicated in much past research, this possibility has not been the subject of much empirical testing. The findings of the current study fill this gap in the literature by providing empirical evidence that online political activities lead to changes in socio-economic stratification in offline political activities. To summarize, low SES citizens use SNS for political expression nearly as much as their high SES counterparts. This in turn functions as an opportunity to compensate for offline resource deficits among underprivileged citizens. It might still be overly optimistic to claim that the Internet will bring equality in every aspect of political life. However, the findings of the current study add to the idea that the Internet – specifically certain uses of SNS – narrows political inequality at least in the context of citizen political communication. Another aspect of the current study is to expand the context of political inequality to everyday citizen communication. Previous literature on political inequality has predominantly focused on participatory activities through which citizens’ voices are formally expressed (Verba et al., 1995; cf. Cho & McLeod, 2007). This research on higherorder political behaviors has shown that some segments of the population are more able to get their opinions across than others. However, although much is known about whose voices carry the farthest in the governing process, the question of whether there is a systematic disparity in how citizens communicate with fellow citizens on politics in their lifeworld has received little attention. This everyday citizen-to-citizen communicative interaction is a place where opinion is formed and participatory norms are brewed. By shifting the focus to a more fundamental and basic feature of democratic citizenship, the results of this study broaden the discussion of political inequality and enrich the understanding of the role the Internet plays in political inequality. As Dewey (1927) noted, “the improvement of the methods and conditions” of citizen communication is one of the key problems the public faces (p. 208). If, as shown in this study, political expression on SNS ameliorates the inequality in citizens’ political conversation, SNS use fills an essential need of democracy. To conclude, the boundaries and limitations of this study are discussed to provide some suggestions for future research. First, drawing on the literature of resource mobilization, this study examines whether, as in offline political discussion, citizen expression on SNS is structured by individuals’ socio-economic status (SES) and whether SNS use ameliorates the socio-economic stratification of offline political communication. Yet, the resource-based mechanism linking socio-economic status to communicative actions is only discussed, not empirically tested. Future study should measure SES-associated resources (civic skills, psychological engagement, social networks, peer

9

pressure and recruitment, etc.) and formally investigate how the psychological and social resources mediate the impact of education and income on communication behaviors and how the mediation process differs online and offline. Second, this study considers self-expression on SNS as communication. Yet, although expression is a basic unit of discussion and a prevalent form of communication, especially on SNS, expression and discussion are different communication acts. Indeed, expression itself, whether online or offline, requires fewer resources than discussion does. Perhaps, this difference in the nature of communication, discussion vs. expression, may also factor into the results for H2—expression is relatively less affected by socioeconomic status. It would be useful for future research in this line to investigate whether online activity reduces offline inequality within the same course of action (discussion, expression, or participation). This would then allow us to focus on the difference in context (online vs. offline), holding the difference in the nature of the action constant. Last, the findings of this study shed light on the positive side of political expression on SNS. Indeed, routine acts of political expression provide users with an opportunity to better understand their political dispositions and organize their thoughts on current issues (Pingree, 2007). However, not all citizens have equal access to everyday communication networks where political expression is possible. Citizens from advantaged backgrounds have a greater number of reference groups and more politically interested interpersonal contacts, which increases the likelihood of political expression (Tichenor et al., 1970). Given this offline communication inequality, our findings speak to an important, unique role of social media. By granting low-cost, easy-access opportunities to political expression, social media open up the benefits of political expression to a broader socio-economic range. Further, our data suggest that SNS expression works toward closing the offline communication gap. On the other hand, it is also possible that expression on SNS, unlike that in the offline setting, might result in a wide circulation of personal views and biased messages. Online expression driven by partisan motivation may add to the flame of divisive, nondeliberative, or even uncivil discourse, especially when combined with selective information seeking and communication. Despite the positive function of SNS expression found in this study, the end result might be partisan reinforcement and eventually political polarization and mutual cynicism. Because these negative consequences of SNS expression fall beyond the boundaries of this study, they should be addressed in future research.

Appendix A. Demographic characteristics of the sample Variables

Descriptive statistics

% Female % White % Employed % Married

55.9 76.4 66.1 51.5


10

Education Income Age

J. Cho, H. Keum / The Social Science Journal xxx (2016) xxx–xxx M

SD

Minimum

Maximum

4.94 5.41 42.50

1.57 2.38 15.90

1 1 18

7 9 90

Notes: Data in this table are from the 2012 Pew Internet & American Life Project national surveys, entitled Search, Social Networks, and Politics (http://www.pewinternet.org/datasets/february-2012-search-socialnetworks-and-politics/). Education is measured on a 7-point ordinal scale, with 1 = Grade 8 or lower and 7 = Post-graduate education after college. The sample mean of 4.94 is between “Technical, trade, or vocational school after high school (4)” and “Some college, no 4-year degree (5).” Income is measured on a 9-point ordinal scale, with 1 = Less than $10,000 and 9 = $150,000 or more. The sample mean of 5.41 is between the “$40,000 to under $50,000 (5)” and “$50,000 to under $75,000 (6)” brackets.

References Ahuvia, A. C., & Friedman, D. C. (1998). Income, consumption, and subjective well-being: Toward a composite macromarketing model. Journal of Macromarketing, 18, 153–168. http://dx.doi.org/ 10.1177/027614679801800207 Barber, B. (2001). The uncertainty of digital politics: Democracy’s uneasy relationship with information technology. Harvard International Review, 23, 42–48. Benhabib, S. (1996). Democracy and difference: Contesting the boundaries of the political. Princeton, NJ: Princeton University Press. Bonfadelli, H. (2002). The Internet and knowledge gaps: A theoretical and empirical investigation. European Journal of Communication, 17, 65–84. http://dx.doi.org/10.1177/0267323102017001607 Boulianne, S. (2009). Does Internet use affect engagement? A metaanalysis of research. Political Communication, 26, 193–211. http://dx.doi.org/10.1080/10584600902854363 Brundidge, J. (2010). Encountering difference in the contemporary public sphere: The contribution of the Internet to the heterogeneity of political discussion networks. Journal of Communication, 60, 680–700. http://dx.doi.org/10.1111/j. 1460-2466.2010.01509.x Burns, N., Schlozman, K. L., & Verba, S. (2001). The private roots of public action: Gender, equality, and political participation. Cambridge, MA: Harvard University Press. Cho, J., & McLeod, D. M. (2007). Structural antecedents to knowledge and participation: Extending the knowledge gap concept to participation. Journal of Communication, 57, 205–228. http://dx.doi.org/10.1111/j. 1460-2466.2007.00340.x Cho, J., Shah, D. V., McLeod, J. M., McLeod, D. M., Scholl, R. M., & Gotlieb, M. R. (2009). Campaigns, reflection, and deliberation: Advancing an O-S-R-O-R model of communication effects. Communication Theory, 19, 66–88. http://dx.doi.org/10.1111/j. 1468-2885.2008.01333.x Conover, P. J., Searing, D. D., & Crewe, I. M. (2002). The deliberative potential of political discussion. British Journal of Political Science, 32, 21–62. http://dx.doi.org/10.1017/S0007123402000029 Conroy, M., Feezell, J. T., & Guerrero, M. (2012). Facebook and political engagement: A study of online political group membership and offline political engagement. Computers in Human Behavior, 28, 1535–1546. http://dx.doi.org/10.1016/j.chb.2012.03.012 Correa, T., Hinsley, A. W., & Gil de Zuniga, H. (2010). Who interacts on the Web? The intersection of users’ personality and social media use. Computers in Human Behavior, 26, 247–253. http://dx.doi.org/10.1016/j.chb.2009.09.003 Dahl, R. A. (1989). Democracy and its critics. New Haven, CT: Yale University Press. Delli Carpini, M. X. (2000). In search of the informed citizen: What Americans know about politics and why it matters. Communication Review, 4, 129–164. http://dx.doi.org/10.1080/10714420009359466 Dewey, J. (1927). The public and its problems. New York, NY: Henry Holt. DiMaggio, P., Hargittai, E., Neuman, W. R., & Robinson, J. P. (2001). Social implications of the Internet. Annual Review of Sociology, 27, 307–336. Ellison, N. B., Steinfeld, C., & Lampe, C. (2007). The benefits of Facebook friends: Social capital and college students’ use of online social networking sites. Journal of Computer-Mediated Communication, 12, 1143–1168. http://dx.doi.org/10.1111/j. 1083-6101.2007.00367.x Eveland, W. P., Jr. (2004). The effect of political discussion in producing informed citizens: The roles of information, motivation, and elaboration. Political Communication, 21, 177–193. http://dx.doi.org/10.1080/10584600490443877 Eveland, W. P., Jr., Hayes, A. F., Shah, D. V., & Kwak, N. (2005). Understanding the relationship between communication and political

knowledge: A model comparison approach using panel data. Political Communication, 22(4), 423–446. Fraser, N. (1992). Rethinking the public sphere: A contribution to the critique of actually existing democracy. In C. Calhoun (Ed.), Habermas and the public sphere (pp. 109–142). Cambridge, MA: MIT Press. Gamson, W. A. (1992). Talking politics. Cambridge, MA: Cambridge University Press. Hargittai, E. (2008a). The digital reproduction of inequality. In D. Grusky (Ed.), Social stratification (pp. 936–944). Boulder, CO: Westview Press. Hargittai, E. (2008b). Whose space? Differences among users and non-users of social network sites. Journal of ComputerMediated Communication, 13, 276–297. http://dx.doi.org/10.1111/ j. 1083-6101.2007.00396.x Hayes, A. F. (2013). Introduction to mediation, moderation, and conditional process analysis: A regression-based approach. New York, NY: Guilford Press. Huckfeldt, R., & Sprague, J. (1995). Citizens, politics and social communication. New York, NY: Cambridge University Press. Kim, Y., Hsu, S. H., & Gil de Zuniga, H. (2013). Influence of social media use on discussion network heterogeneity and civic engagement: The moderating role of personality traits. Journal of Communication, 63, 498–516. http://dx.doi.org/10.1111/jcom.12034 King, G., Honaker, J., Joseph, A., & Scheve, K. (2001). Analyzing incomplete political science data: An alternative algorithm for multiple imputation. American Political Science Review, 95, 49–69. Krueger, B. S. (2002). Assessing the potential of Internet political participation in the United States: A resource approach. American Politics Research, 30, 476–498. http://dx.doi.org/10.1177/ 1532673X02030005002 Lane, R. E. (1959). Political life: Why people get involved in politics. New York, NY: Free Press. Lijphart, A. (1997). Unequal participation: Democracy’s unresolved dilemma. American Political Science Review, 91, 1–14. Mansbridge, J. (1999). Everyday talk in deliberative system. In S. Macedo (Ed.), Deliberative politics: Essays on democracy and disagreement (pp. 211–239). Oxford, England: Oxford University Press. Marques, A. C. S., & Maia, R. C. M. (2010). Everyday conversation in the deliberative process: An analysis of communicative exchanges in discussion groups and their contributions to civic and political socialization. Journal of Communication, 60, 611–635. http://dx.doi.org/10.1111/j. 1460-2466.2010.01506.x McLeod, J. M., Scheufele, D. A., Moy, P., Horowitz, E. M., Holbert, R. L., Zhang, W., et al. (1999). Understanding deliberation: The effects of discussion networks on participation in a public forum. Communication Research, 26, 743–774. http://dx.doi.org/10.1177/009365099026006005 McLeod, J. M., Zubric, J., Keum, H., Deshpande, S., Cho, J., Stein, S., et al. (2001). Reflecting and connecting: Testing a communication mediation model of civic participation. In Paper presented to the annual conference of the Association for Education in Journalism and Mass Communication, Washington DC (August). Mesch, G. S. (2012). Minority status and the use of computer-mediated communication: A test of the social diversification hypothesis. Communication Research, 39, 317–337. http://dx.doi.org/10.1177/ 0093650211398865 Milbrath, L., & Goel, M. (1977). Political participation: How and why do people get involved in politics? Boston, MA: Rand McNally. Myers, T. A. (2011). Goodbye, listwise deletion: Presenting hot deck imputation as an easy and effective tool for handling missing data. Communication Methods and Measures, 5(4), 297–310. Nie, N. H., & Hillygus, D. S. (2002). The impact of Internet use on sociability: Time-diary findings. IT & Society, 1(1), 1–20. Parks, M. R. (2011). Boundary conditions for the application of three theories of computer mediated communication to MySpace. Journal of Communication, 61, 557–574. http://dx.doi.org/10.1111/j.1460-2466.2011.01569.x Peterson, R. D. (2012). To tweet or not to tweet: Exploring the determinants of early adoption of Twitter by House members in the 111th Congress. Social Science Journal, 49, 430–438. http://dx.doi.org/10.1016/j.soscij.2012.07.002 Pingree, R. J. (2007). How messages affect their senders: A more general model of message effects and implications for deliberation. Communication Theory, 17, 439–461. http://dx.doi.org/10.1111/ j.1468-2885.2007.00306.x Prior, M. (2007). Post-broadcast democracy: How media choice increases inequality in political involvement and polarizes elections. Cambridge, MA: Cambridge University Press. Putnam, R. (1995). Bowling alone: America’s declining social capital. Journal of Democracy, 6, 65–78.


J. Cho, H. Keum / The Social Science Journal xxx (2016) xxx–xxx Rainie, L., & Smith, A. (2012). Social networking sites and politics. Washington, DC: Pew Internet & American Life Project. Robinson, J. P., & Haan, J. D. (2006). Information technology and family time displacement. In R. Kraut, M. Brynin, & S. Kiesler (Eds.), Computers, phones, and the Internet: Domesticating information technology (pp. 51–69). New York, NY: Oxford University Press. Schlozman, K. L., Verba, S., & Brady, H. E. (2010). Weapon of the strong? Participatory inequality and the Internet. Perspectives on Politics, 8, 487–509. http://dx.doi.org/10.1017/S1537592710001210 Shah, D. V., Kwak, N., & Holbert, R. L. (2001). Connecting and disconnecting with civic life: Patterns of Internet use and the production of social capital. Political Communication, 18, 141–162. http://dx.doi.org/10.1080/105846001750322952 Sinclair, B. (2012). The social citizen: Peer networks and political behavior. Chicago, IL: University of Chicago Press. Stockemer, D. (2014). What drives unconventional political participation? A two level study. Social Science Journal, 51, 201–211. http://dx.doi.org/10.1016/j.soscij.2013.10.012 Tang, G., & Lee, F. L. F. (2013). Facebook use and political participation: The impact of exposure to shared political information, connections with public political actors, and network structural heterogeneity. Social Science Computer Review, 31, 763–773. http://dx.doi.org/10.1177/0894439313490625

11

Tichenor, P. J., Donohue, G. A., & Olien, C. N. (1970). Mass media flow and differential growth in knowledge. Public Opinion Quarterly, 34, 159–170. http://dx.doi.org/10.1086/267786 Utz, S. (2009). The (potential) benefits of campaigning via social network sites. Journal of Computer-Mediated Communication, 14, 221–243. http://dx.doi.org/10.1111/j.10836101.2009.01438.x Valenzuela, S., Park, N., & Kee, K. F. (2009). Is there social capital in a social network site?: Facebook use and college students’ life satisfaction, trust, and participation. Journal of Computer-Mediated Communication, 14, 875–901. http://dx.doi.org/10.1111/j.1083-6101.2009.01474.x Verba, S., Nie, N., & Kim, J. (1971). The modes of democratic participation: A cross national comparison. Beverly Hills, CA: Sage. Verba, S., Schlozman, K. L., & Brady, H. E. (1995). Voice and equality: Civic volunteerism in American politics. Cambridge, MA: Harvard University Press. Verba, S. (2003). Would the dream of political equality turn out to be a nightmare? Perspectives on Politics, 1, 663–679. http://dx.doi.org/10.1017/S1537592703000458 Walsh, K. C. (2004). Talking about politics: Informal groups and social identity in American life. Chicago, IL: University of Chicago Press. Young, I. (1996). Communication and the other: Beyond deliberative democracy. In S. Benhabib (Ed.), Democracy and difference: Contesting the boundaries of the political (pp. 120–136). Princeton, NJ: Princeton University Press.
