Democratic Digital Inequalities: Threat and Opportunity in Online Citizenship From Motivation and Ability

The Internet as a Public Sphere

The public sphere and public opinion are inextricably tied to the birth of democracy (Habermas, 1991). Habermas (1991) argues that a normatively functioning democracy operates through a public discourse environment shared among citizens, the public sphere, where participants reflexively and rationally communicate. It is through the public sphere that legitimate, justified decisions on governance can be made through an emergent public opinion. Historically, public discourse is largely diffused through the mass media. Indeed, Zaret (2000) traced the infancy of democracy to the advent of the printing press. The ability to share ideas widely helped ignite a coherent public debate and fostered vibrant citizenship through a politically informed electorate.

Structural changes in public opinion and politics are tied to changes in the information landscape (Bimber, 2003). Bimber (2003) argued that innovation through information revolutions leads to changes in politics. As we evolve to an Internet- and social media–connected world, scholars are continually assessing the democratic impact of the transformation from a broadcast and mass media environment into an Internet society (see Gil de Zúñiga, Molyneaux, & Zheng, 2014; Halpern & Gibbs, 2013; Sunstein, 2007). This article focuses on two key components of ideal citizenship: seeking out news information and participating in the online public sphere through online news sharing.

Reading and Sharing News Online

In terms of access to information relevant to politics, the Internet affords citizens nearly limitless amounts of news, opinions, and discourse from global news organizations and other online users at a low cost. Indeed, the diffusion of Internet access through computers and mobile technologies has led to a shift in where people get their news. Although television is still the dominant platform for news, more than a third of the people in the United States report frequently reading news online (Mitchell, Gottfried, Barthel, & Shearer, 2016). Furthermore, there is evidence that people who use the Internet for news in the context of political elections are better informed than those who do not (Garrett, 2009). The meteoric rise of social media has meant that increasing numbers of U.S. adults are coming across political information on their social network site newsfeeds. In fact, about half of U.S. adults reported learning about the 2016 U.S. election through social media (Lu & Holcomb, 2016). Like Internet news use in general, reading news on social network sites has also been shown to be related to increased political knowledge (Beam, Hutchens, & Hmielowski, 2016), decreased levels of political cynicism and apathy (Yamamoto, Kushin, & Dalisay, 2017), and greater political participation (Gil de Zúñiga et al., 2014). Although the majority of these studies rely on self-reported survey panel data collected over time, they provide strong circumstantial evidence that there is hope that the Internet may actually do more good than harm for citizens through increased exposure to political information and public opinions.

Attending to political information is only one requirement of citizen deliberation. A second is participation in the public sphere. The Internet, as opposed to the older, one-way media technologies, also allows nonelite citizens to interact with large audiences in popular political information distribution channels like social media (e.g., Shirky, 2011), web forums (e.g., Bruns, 2005), and YouTube (e.g., Dylko et al., 2012). However, considerably fewer users engage in online political information sharing compared with those who access it. For example, Mitchell, Kiley, Gottfried, and Guskin (2013) found that only about half of the users on Facebook who read news are also likely to post news links and less than a third of those users discuss news on the site. Political expression online through sharing and commenting on news has been linked to a variety of outcomes related to citizenship, including increased structural knowledge (Beam et al., 2016) and political mobilization (Yamamoto, Kushin, & Dalisay, 2015). The remainder of this article investigates three key predictors of reading and sharing online news.

Internet Skill and Digital Inequalities

Since the early days of the Internet, researchers have provided evidence that information and communication technologies (ICTs) diffuse more rapidly among privileged demographics (e.g., Rogers, 2001). Many researchers focused on the digital divide were worried that privileged people (i.e., young white men) gained advantages through access to ICTs that caused systematic inequalities between those who are connected and those who are not (e.g., Katz & Rice, 2002). Despite the subsequent widespread diffusion of ICTs, scholars have expanded the digital inequalities literature to focus on other variables that may still contribute to unequal benefits of the Internet (e.g., Elliott & Earl, 2018; Robinson et al., 2015; Wei, 2012).

One potentially important variable is Internet skill. Internet skill has been shown to be a predictor of online activities (Hargittai, 2002; Litt, 2013). Put simply, Internet skill is “the ability to use the Internet effectively and efficiently” (Hargittai & Shaw, 2015, p. 427). Internet skill has been shown to be predictive of several online behaviors, including managing online privacy (Hargittai & Litt, 2013), contributing to Wikipedia (Hargittai & Shaw, 2015), and creating online content (Hargittai & Walejko, 2008). More closely related to this article, Hargittai and Shaw (2013) found that students with high Internet skill were more likely to engage in online political information practices like viewing, commenting in, or posting on political blogs in the aftermath of the 2008 presidential election. Similarly, Min (2010) showed that Internet skill is related to political information use by using dichotomous measures of viewing websites and discussing politics online from the 2004 General Social Survey. Although these past studies have found a link between Internet skill and online political engagement, digital inequalities are an ongoing concern that need to be assessed at different times as technologies become more commonplace in society (Wei, 2012; Yu, 2006). We believe this is especially true when discussing systematic inequalities in political engagement, as online political communication has evolved rapidly in the past decade. Therefore, we present our first hypothesis:

Motivation and Cognitive Ability

Another important set of variables that could help explain people’s online political engagement come from the OMA framework. This framework is useful for understanding which people are likely to engage in particular forms of communication activities (see Delli Carpini & Keeter, 1996; Prior, 2007; Strömbäck, Djerf-Pierre, & Shehata, 2013). In the context of online political engagement, opportunity would be access to websites and apps containing political information and offering people a platform to share and discuss political information; motivation would be people’s interest in politics; and ability would be people’s level of cognitive ability to process and understand information. In this article, we treat education as our indicator of cognitive ability.

The OMA framework has been used to explain people’s selective exposure to political information (Prior, 2007; Strömbäck et al., 2013), processing of political information (Luskin, 1990), and acquisition of political knowledge (Delli Carpini & Keeter, 1996). Taken together, these studies have found that mere opportunity—or the accessibility of political information—is a weaker predictor of political outcomes compared with motivation and cognitive ability (e.g., Luskin, 1990; Prior, 2007). Because of the decreasing importance of opportunity (e.g., Luskin, 1990; Prior, 2007), we focus on political interest (motivation) and education (cognitive ability) as predictors of engaging in online citizenship in the form of reading and sharing news.

Motivation Through Political Interest

Prior (2007) used the OMA framework to show that the combination of political interest and a fragmented media environment leads to segments of the public being disengaged with political information, whereas “news junkies” become highly engaged with political information. In a media environment rich with many information options, political interest (or motivation) is a key factor in understanding whether people choose entertainment or news media content. Indeed, political interest has been shown to be a strong predictor of reading and viewing news (Arceneaux & Johnson, 2013). According to this line of research, people who are less interested in politics will likely tune out of news and political information when they have access to other options, such as entertainment. Guided by the OMA framework, Strömbäck and his colleagues (2013) showed not only that political interest predicts news exposure across platforms but also that those who are politically interested have increased their news consumption over the past 30 years compared with others.

In an online context, research has also found that political interest correlates with online political communication. In her meta-analysis of 38 studies on Internet use for political engagement, Boulianne (2009) found that political interest was a key predictor variable for political communication effects. Min (2010) found that political interest predicted dichotomous measures of viewing online political websites and discussing politics online. Past studies (using student samples) have linked sharing news on social media with political interest, although political interest was not their focal variable (e.g., Conroy, Feezell, & Guerrero, 2012). Based on this work, we will test if political interest correlates with our indicators of online political engagement.

Ability Through Education

In addition to examining motivational predictors of online news use and online news sharing, we also examined education as an ability predictor of these two outcomes. Scholars argue that individuals with higher levels of education are more capable of integrating and critically evaluating the information that they encounter (Prior, 2007; Tichenor, Donohue, & Olien, 1970). For example, a longitudinal analysis, from 1986 to 2010, of news reading showed that education is a significant predictor of news reading (Strömbäck et al., 2013). Communication and political science scholars have spent an extensive time examining education when looking at political outcomes. Indeed, education has been shown to correlate with higher levels of knowledge and greater political engagement (Eveland & Scheufele, 2000). Scholars have also examined the extent to which education predicts the use of various online sources, and they have consistently found that more educated individuals are spending more time online. This finding holds when examining the use of partisan sources (Garrett, Carnahan, & Lynch, 2013) or general online news (Dutta-Bergman, 2004), or when individuals are explicitly asked if they use the Internet to learn things (Horrigan, 2016). To the best of our knowledge, past studies have not investigated the role of cognitive ability in online news sharing. However, research shows a strong positive relationship between online reading and online sharing (Beam et al., 2016). As outlined above, reading has been consistently linked to education in communication research. Based on this, we predict that education serves to foster both online news reading and online news sharing:

Communication Gaps

Next, we turn our attention to examining whether the relationship between Internet skill and our outcome variables varies based on political interest (motivation) and education (cognitive ability). Research studies on communication gaps step beyond focusing on the simple direct effects between media use and outcome variables to examining the conditional effects of the media on politically oriented outcome variables. A communication gap occurs when a particular group of people derive greater gains from communication on an outcome (e.g., knowledge) compared with other groups (Tichenor et al., 1970; Rogers, 1976). In essence, these studies use the theoretical lens of communication gaps to explain who is affected by the media. Proponents of communication gaps explain that scholars should evaluate the conditional effects of the media on the “attitudinal and overt behavioral effects of communication” (Rogers, 1976, p. 233). The digital divide and digital inequalities literature has also been rooted in communication gaps research (Rogers, 2001). These studies have investigated if variables such as Internet skill produce communication gaps in activities such as accessing informational websites and creating online content (e.g., Hargittai & Shaw, 2013, 2015; Hargittai & Walejko, 2008).

The variables that have been used to examine communication gaps vary. For example, studies have examined the moderating effects of structural variables, such as the ideological plurality of the community (Viswanath & Finnegan, 1996), and channel-specific variables, such as television, newspaper, and interpersonal communication (Gaziano 1997; Viswanath & Finnegan, 1996). Specific to this article, extant research focused on motivation has indicated that this variable could increase or decrease knowledge gaps (Gaziano, 1997). In terms of education, the most frequently cited literature on communication gaps stems from Tichenor et al.’s (1970) work on the knowledge gap. The theory outlines that people of higher socioeconomic status (which includes education) acquire information faster than those of lower socioeconomic status (Tichenor et al. 1970). Given the extensive research on communication gaps, we assessed whether the relationship between Internet skill and our two outcomes varies by political interest and education. Specifically, a unique contribution of this study is in assessing whether Internet skill–based communication gaps in news reading and sharing are exacerbated or diminished based on a person’s level of political interest or education. Our analysis contributes to our understanding of online citizenship by testing these variables through the theoretical lens of the OMA framework. Therefore, we propose the following hypotheses:

Survey Data and Participants

This study utilizes a national sample selected from U.S. Internet users during the 2016 U.S. general election. The data were collected via an Internet survey conducted from October 24 to 30, 2016, on a sample provided by the polling firm YouGov. YouGov utilize a two-step sample-matching procedure, wherein they initially oversample and then create a final sample that matches the national population characteristics. Panelists are recruited using a variety of methods including online advertising and automated telephone dialing. Panelists opt to take online surveys in exchange for points that they can use to redeem prizes or money. After a large number of nonrepresentative opt-in panelists of Internet users respond to a particular survey, a sampling frame to create the final sample is created using benchmarks from large, high-quality probability samples, such as government population surveys. An article by Ansolabehere and Schaffner (2014) discusses YouGov’s proprietary sample-matching procedure in more detail. Their research also empirically compared YouGov sample-matched data with random sampling procedures and found that they exhibit little to no selection bias. Indeed, the scholars empirically demonstrated that sample-matching procedures generate samples that are better than other Internet panels, and are equivalent to the random methods used to create high-quality surveys such as the American National Election Survey. The data used here consist of 500 sample-matched participants. The sample was created using a sampling frame that matched participants on gender, age, race, education, party identification, ideology, political interest, voter registration, and voter turnout. Weights are provided by YouGov based on propensity scores created via age, gender, race/ethnicity, years of education, and ideology.

Measures

Analysis Plan

This study utilized a series of hierarchical ordinary least squares regression models in STATA to test the hypotheses. Data were weighted with the sample weights described above provided by YouGov, using STATA’s SVY tools, which are appropriate for complex survey data. To examine conditional effects from the interactions, we utilized the pick-a-point approach (Hayes, 2013). This technique allows researchers to probe moderating variables at specific values in order to find regions of significance where conditional effects occur. STATA’s margins command was used to execute the pick-a-point approach. It provides a confidence interval for the interaction terms. Regions of significance occur when the confidence intervals do not include 0. One case with missing data was removed from the analysis using listwise deletion.