Context collapse and privacy management: Diversity in Facebook friends increases online news reading and sharing

Context collapse and sharing news

Facebook provides its users with an information environment curated by their contacts. In fact, three key features that differentiate an SNS like Facebook from other websites are the ability to maintain (1) a personal profile, (2) a list of contacts whose profiles and contents we can view, and (3) user-generated content that is persistent and accessible (Carr and Hayes, 2015; Ellison and boyd, 2013). SNS users are likely to accrue more contacts over time—called “friends” on Facebook—coming from different contexts of their lives, such as family members, friends, or co-workers (Child and Petronio, 2011). Facebook, like most SNS, treats all contacts as a single audience by default, which causes those various contexts to collapse when engaging in online activities. While some users might further segment contacts beyond limiting posts to “friends only,” research has shown that most users (81%) report that their friends all see the same content (Madden et al., 2013). Similarly, in a multi-method study investigating Facebook disclosure, Sleeper et al. (2013) found that users do not generally granularly manage which particular friends receive shared content. In fact, they found that even users who went through the effort of setting up custom groups usually abandoned the more granular approach in favor of sharing content to all friends. This suggests that disclosure-based rather than technological-based approaches are potentially more important to explore in order to understand how individual users manage privacy on Facebook. The makeup of a user’s online social network is influential on someone’s news exposure and willingness to read, share, and discuss news content. SNS scholars have often discussed the role of “context collapse” in users’ online routines, including information exposure and disclosure (e.g. Davis and Jurgenson, 2014; Marwick and boyd, 2011; Vitak, 2012).

When a person posts content on Facebook, he or she must imagine her audience—and the contexts (i.e. group membership) of those imagined audience members (Litt, 2012; Marwick and boyd, 2011). Research that has examined the effects of context collapse on SNS use and online disclosure has found mixed results. Some researchers have contended that a more diverse imagined audience will lead to greater tension and less willingness to share private information online (Binder et al., 2009; Child and Staracher, 2016). For example, several qualitative studies using interviews have found that users say they engage in self-censorship when they perceive their audience as more diverse, especially when discussing sensitive areas such as news, politics, and religion (Child, 2015; Marwick and boyd, 2011; Sleeper et al., 2013; Vitak et al., 2015). Furthermore, when people are more concerned about a diverse Facebook network engaging in mediated lurking of their site, they respond by managing their privacy boundary more tightly and engaging in more strategic ambiguity or vaguebooking when interacting with them on Facebook (Child and Staracher, 2016). However, other empirical work has found a link between context collapse and increased disclosing of information online. For example, Facebook researchers, Das and Kramer (2013), found that people who have contacts with greater political diversity share information more readily through Facebook. Similarly, Vitak (2012) found that having more online friends and more context collapse predicted greater online disclosure, contrary to her predictions. So, for some individuals social network diversity is a sign of cautiousness and for others it is an opportunity to engage in enhanced perspective taking.

One way to make sense of these mixed findings would be to understand that increased context collapse is likely to be intentionally utilized. In their discussion of context collapse, Davis and Jurgenson (2014) differentiate between context collisions and context collusions. Context collisions occur when people’s online behavior accidentally crosses relationship boundaries, which could lead to negative outcomes, such as unintended online disagreement and privacy management breakdowns. People who are more concerned about their privacy may be more likely to see further sharing in a diverse network of people as potentially leading to context collisions and privacy breakdowns. Collusions occur when people are utilizing the various contexts of their social network to receive social utility. When discussing online information sharing, several studies have found potential benefits, including increased social capital, individual expression, political engagement, and gains in political knowledge (Beam et al., 2016; Larsson and Moe, 2012; Vitak, 2012; Yamamoto et al., 2015). These results indicate that online sharing can be a strategic collusion of a diverse online audience to garner increased surveillance, persuasion, or engagement. People who are less concerned about their privacy may have a collusion mindset about the sharing of information with their network.

Sleeper et al. (2013) found that people were more likely to share political content they found interesting and not self-censor, despite some concerns about context collapse. Taken together, the evidence shows that people are likely to consider the risks and benefits of online disclosure, a premise of individual privacy rule considerations according to CPM theory (Petronio, 2002). Ultimately, we believe that people overall in the context of Facebook are more likely to see the sharing of news as a context collusion rather than a collision when posting information to a diverse audience. Therefore, we believe that increased context collapse will be related to increased online news sharing (H1). Furthermore, there is evidence that reading news motivates the act of sharing news content. Next, we will introduce a mediated model where the relationship between context collapse and online news sharing is mediated by reading news on Facebook.

Facebook and reading news

Facebook has become a primary platform for accessing and disseminating all types of information, including news. With nearly one-third of all US adults reporting that they get news from the site, Facebook feeds have become a central repository of news information that people rely on to make sense of the world. Furthermore, two-thirds of US adults under the age of 33 report that Facebook is their primary source of news (Mitchell et al., 2015). Therefore, one possible benefit of a diverse network is that individuals will see and read more varied types of news stories compared to those with more homogeneous networks. By sharing news online, non-elite SNS users also become gatekeepers to an audience of a size unreachable in a pre-Internet era (Dylko et al., 2012). Despite this increased clout, scholars worry that greater levels of online fragmentation will produce lower levels of news engagement.

Selective exposure has provided a theoretical perspective to understand news information exposure patterns. In the context of news, this theoretical perspective is based on the notion that people will choose to read news stories that most align with their own interests and perspectives (Stroud, 2011). As media choice has increased over the years, some scholars have argued many people will opt out of viewing news in favor of more entertaining content (Arceneaux and Johnson, 2013; Prior, 2007, 2013); other scholars have voiced concerns about automated algorithms that filter information based on users’ interests because they may also reduce diversity of news a person sees, leading to a less informed citizenry (Dylko, 2016; Pariser, 2011; Sunstein, 2007). These concerns have been exacerbated by rising levels of misinformation, decreasing levels of political knowledge, and rising partisan polarization (Hmielowski et al., 2016; Nyhan and Reifler, 2010; Prior, 2005).

However, the majority of the recent empirical work focused on selective exposure in the online context has found that while people do tend to select information that aligns with their interests, they are not likely to selectively avoid information that does not (Garrett, 2009; Garrett et al., 2013). For example, Beam and Kosicki (2014) found that, despite offering more selectivity, users relying on personalized news recommendations from news aggregators will read news more frequently and view more categories of news. They argued that personalized news recommendations lead to greater news exposure for users due to increased personal relevance (Beam and Kosicki, 2014). Similarly, Messing and Westwood’s (2014) experimental research showed that social news recommendations containing counter-attitudinal perspectives are more likely to be read when they are shared by other Facebook users.

Taken together, these studies show that diverse news recommendations can foster increased news exposure. Furthermore, we have other reasons to believe increased context collapse might be related to increased news reading. Social media works as a hybrid news environment, where users can both read and discuss content. Past research has found that increased political discussion is related to increased news exposure (Eveland, 2004; Scheufele, 2000). Eveland (2004) argues that discussion encourages additional information processing and seeking in order to prepare for future discussion. Similarly, research has linked more heterogeneous social networks with increased news use (Scheufele et al., 2006) and increased social media use (Kim et al., 2013). In sum, past research indicates that when people see news stories shared online from diverse contacts they should be more likely to explore that news and engage in online news reading behaviors. As the diversity of sources increases, the need to read additional information also increases. Therefore, we believe that people who have a more diverse set of social network contacts through increased context collapse will be more likely to engage in online news reading on Facebook (H2).

Facebook and sharing news

The act of sharing news allows users to become gatekeepers in the flow of information, filtering which stories from other sources come through their friends’ newsfeeds. Collectively, this act of “gatewatching” allows for whole online communities to set trends in news story popularity, which has resulted in some news organizations producing news that is likely to “go viral,” resulting in increased page views (Bruns, 2005; Nahon and Hemsley, 2013). Scholars have viewed the act of sharing news and engaging in political discussion as being essential to democracy (Delli Carpini and Keeter, 1996).

Fewer people actively engage in news sharing compared to those who read news online (Rau et al., 2008). Pew Research recently found less than half of Facebook users who view news on the site also post news links and less than one-third of Facebook newsreaders discuss news on the site (Mitchell et al., 2013). However, research has shown a clear link between people who read greater amounts of news online and people who share news online. For example, Weeks and Holbert (2013) found that increased online news viewing and partisan strength were strongly related to online news sharing. Beam et al. (2016) also found a strong link between more online news reading and online news sharing. They also showed a subsequent relationship between those activities with political knowledge acquisition. Thus, people who take time to read more articles seem to be more individually disposed to want to share newly acquired insights from reading news with their individual networks. Therefore, we expect to find a positive relationship between increased news reading on Facebook and sharing news on Facebook (H3).

Next, we will explore the theoretical mechanism of privacy management that will moderate this relationship between news reading and sharing and formalize a prediction for an overall mediated-moderated relationship between context collapse, news reading, privacy management, and news sharing.

Privacy management

Research on privacy management practices provides an important theoretical mechanism to help us better understand when people are more willing or reluctant to share potentially sensitive information online. People using SNSs, especially those contending with context collapse, are often concerned about issues of privacy regulation (Child and Staracher, 2016; Marwick and boyd, 2011). Furthermore, research has shown that news and politics are particular areas of concern for some social media users due to their potentially polarizing nature. For example, Sleeper et al. (2013) wrote that people who deleted previously shared political news articles did so because the content was considered “potentially controversial” (p. 796). Similarly, in their study of regret in social media sharing, Wang et al. (2011) found that some users feared damaging relationships when making political posts. Thus, some people are not interested in using their Facebook collective privacy boundary to explore controversial issues. These people are more likely to see the sharing of information online as leading to a context collision. Others, who see sharing as facilitating context collusions, may be more likely to value the benefits of hearing from a diverse network on an article than any potential concerns about privacy that may lead to a context collision.

CPM theory predicts that contextual factors affect privacy rules. A number of studies highlight the types of decisions people make about either protecting or sharing their private information in the context of Facebook. For example, research supports that people who are more inhibited mark more information as private and do not share as much through social media than people who are disinhibited (Child et al., 2012). Family members and family privacy rule orientations affect individual decisions of openness on Facebook (Child and Westermann, 2013). People who are more concerned about mediated lurking are less open in their own privacy management practices on Facebook (Child and Staracher, 2016). People with a planning or a sharing orientation disclose more private information through social media than other privacy rule orientations (Child et al., 2012). Finally, people who share more private information through social media are also more likely to engage in deleting previously posted content on their SNS sites than those individuals who enact a higher level of privacy management online (Child et al., 2012).

In addition to contextual influences on privacy rules, CPM theory also predicts that individual motivations affect how people make decisions about managing their private information (Petronio, 2002). Social media users who are more open in their privacy management practices reflect a stronger individual rule motivation to trust and share private information more freely with their social network than those who are more protective and private (Child and Petronio, 2011, 2015). Therefore, we predict that an individual motivation toward being open on Facebook and sharing private information on the site in general will moderate the relationship between reading news online and sharing news on Facebook (H4). Taking all of our hypotheses together, we propose the moderated-mediated model found in Figure 1, where context collapse is related to more news sharing mediated by reading news on Facebook. This mediation model should be moderated by openness in privacy management, where people with higher levels of openness relative to privacy management practices will be more likely to share news when they also read news on Facebook (H5).

OPEN IN VIEWER

Figure 1.

Moderated-mediated model of context collapse and Facebook news sharing.

Method

Data presented in this study were obtained from a national online panel of participants recruited by Qualtrics, an online survey software and sample provider. Data were collected between 28 October 2014 and 30 October 2014, prior to the 2014 US midterm elections (N = 771). Panel members were randomly sampled from quota groups that matched national census characteristics, including age and gender. The original sample included 1001 valid responses. However, this study is concerned with participants who are also Facebook users, which excluded 220 participants. An additional 10 participants were not included due to missing variables, leaving 771 individual responses for the analyses. Online panel samples have become more popular in recent years as the public continues to move away from landlines to cell phones, which has reduced researchers’ ability to collect data using random digit dialing to reach a probability sample (Schaffner, 2011). As a result, scholars have examined the quality of quota samples. Although some studies have shown that opt-in Internet panels using quota groups provide higher levels of error compared with traditional probability methods (Yeager et al., 2011), there is some evidence that use of these samples makes little difference in terms of response quality (Ansolabehere and Schaffner, 2014).

Endogenous measures

Context collapse

Our measurement of context collapse focuses on relationship context diversity. This approach is adapted from previous studies (e.g. Binder et al., 2009; Vitak, 2012). Participants were asked to respond to the prompt “By placing your friends into a single category, what percentage of your friends belong to.” Participants were given five categories: family, friends, classmates, co-workers, or members of religious organizations. Next, we calculated Simpson’s D, a diversity measurement that has been utilized in communication research to quantify diversity using proportions (see Eveland and Hively, 2009; McDonald and Dimmick, 2003). Simpson’s D is measured as

D = 1 − ∑ p i 2

where p_i is the proportion of friends within each context category. With five categories, this measure ranges from 0 representing no diversity, where all friends are from a single context, to 0.8, representing the most diversity, where friends are evenly distributed between categories (M = 0.52, SD = 0.19). Past measures of context collapse have included a number of other constructs outside of the scope of this study, including total number of friends (e.g. Vitak, 2012). However, we control for total friends throughout our analyses (M = 304.6, SD = 465.8).

Privacy management

Participants completed an adapted version of a validated measure of privacy management (see Child et al., 2009; Child and Staracher, 2016). The adapted measure has been used to examine issues of privacy regulation and management in several studies on Facebook (e.g. Child et al., 2012, 2015; Child and Staracher, 2016; Child and Westermann, 2013, 2015; Frampton and Child, 2013). The scale was constructed from theoretical principles and the conceptualization of privacy according to CPM theory (Petronio, 2002, 2013); based on past research using CPM theory, this measure contains questions about the three dimensions of privacy management, including boundary permeability, boundary linkages, and boundary ownership (Petronio, 2002). Participants were asked to answer five items using a 7-point Likert scale ranging from “Never true” (coded 0) to “Always true” (coded 6), using the prompt “Please consider how you handle information on Facebook.” The scale consisted of five items, including “I like my Facebook posts to be long and detailed,” “When I face challenges in my life, I feel comfortable talking about them on Facebook,” “I often tell intimate, personal things about myself on my Facebook page without hesitation,” “I comment on posts to have others check out my own Facebook page,” and “I try to use Facebook so that other users with similar interests can link to me.” The mean response was calculated to produce a final scale where lower scores represent closed privacy management practices and higher scores represent open privacy management practices (M = 1.75, SD = 1.52, α = .89).

Facebook news reading

Two items were used to assess the frequency of reading news. Using response options from “Never” (coded 0) to “Several times a day” (coded 5), participants were asked how often they “use the different sources listed below to get news information.” Sources included, “read news on Facebook without following links” and “click on links shared from Facebook.” The mean response was calculated (M = 2.92, SD = 1.46, Spearman–Brown coefficient = .79).

Facebook news sharing

Four items were adapted from prior measures assessing frequency of sharing news on Facebook (e.g. Gil de Zúñiga et al., 2014; Kushin and Yamamoto, 2010). Respondents were asked how often they engaged in a variety of information sharing acts on Facebook, including sharing your thoughts or opinions about the news, posting news links or information, liking a news story or link, and commenting on a news story or link someone else posted. All of these items used the same 6-point scale of “Never” (coded 0) to “Several times a day” (5). The mean response was calculated for our final measure (M = 2.04, SD = 1.44, α = .92).

Results

First, we proposed that context collapse would be related to sharing and reading news. Results from the OLS regression analyses predicting news sharing and news reading can be found in Table 1. The second column of Table 1 shows that context collapse is related to sharing news on Facebook, supporting H1 (B = .57, standard error [SE] = 0.21, p < .001). The first column of Table 1 shows that our second hypothesis is also supported, as context collapse is positively related to reading news on Facebook (B = .54, SE = 0.27, p < .05). Taken together, we see that having more diversity in the relationship contexts of Facebook friends is related to more news engagement, both through more sharing and reading news online.

OPEN IN VIEWER

Table 1.

Ordinary least squares regression models. ^a

	Read news on Facebook	Share news on Facebook	Share news on Facebook
Education	−.03 (.04)	−.04 (.03)	−.02 (.03)
Income	−.00 (.01)	−.03* (.01)	−.01 (.01)
Gender (F)	.04 (.11)	−.04 (.08)	.17* (.08)
Age	−.02*** (.00)	.00 (.00)	.00 (.00)
Minority	.02 (.14)	.25* (.11)	.14 (.09)
Adjusted R2	.03***	.03***	.03***
Political ideology (liberal)	−.02 (.03)	−.00 (.02)	−.00 (.02)
Political interest	.08* (.03)	.15*** (.03)	.09*** (.02)
Conflict avoidance	.11* (.05)	−.11** (.04)	−.09** (.03)
Perceived agreement	.13** (.04)	.09** (.03)	−.02 (.03)
Adjusted R2	.07***	.11***	.11***
Total friends	.00** (.00)	.00* (.00)	.00 (.00)
Context collapse	.54* (.27)	.57** (.21)	.32 † (.19)
Read news on Facebook	–	.55*** (.03)	.34*** (.04)
Privacy management practices (open)	–	–	.16* (.07)
Privacy management × read news on Facebook	–	–	.07*** (.02)
Constant	2.41*** (.37)	.21 (.30)	.37 (.27)
R 2	.091**	.428***	.554***

a

N = 771.

Unstandardized coefficients with standard error in parentheses.

R² values calculated for blocks in hierarchical regression models.

†

p < .10; *p < .05; **p < .01; ***p < .001, two-tailed.

The second column of Table 1 provides results that support H3. Specifically, the results show a positive relationship between reading news and sharing news on Facebook (B = .55, SE = 0.03, p < .001). Higher levels of news reading are associated with significantly higher levels of sharing news on Facebook.

Finally, the third column of Table 1 shows the results of our proposed moderated-mediation model. Figure 2 illustrates that people who read news on Facebook report significantly more online news sharing when they are more open in their privacy management practices (B = .07, SE = 0.02, p < .001), where open privacy management is 1 SD over the mean (PM = 3.27) and closed privacy management is 1 SD under the mean (PM = .23). Probing the interaction with the Johnson–Neyman technique shows that the interaction is significant at all values of privacy management, indicating that differences in privacy management techniques play a significant moderating role for all Facebook users. H4 is supported. A conditional process analysis using 5000 bootstraps for bias corrections shows that reading news on Facebook mediates the relationship between diversity of contexts of friends on Facebook and sharing news on Facebook. That is, increased online news sharing on Facebook due to context collapse is occurring, in part, because of the increased exposure to news on Facebook. Furthermore, this relationship is moderated by privacy management openness, where openness amplifies the indirect relationship of the diversity of contexts of friends on sharing news on Facebook through reading news on Facebook. The conditional indirect bootstrapped lower and upper confidence intervals do not contain 0, which indicates a significant indirect effect. People with more closed privacy management practices (e.g. B = .19, confidence interval (CI) = [.001, .414] for those with a score of 1 SD below the mean) show smaller indirect effects of context collapse on news sharing through news reading on Facebook than those with more open privacy management practices (e.g. B = .30, CI = [.003, .644] for those with a score of 1 SD above the mean). H5 is supported.

OPEN IN VIEWER

Figure 2.

Interaction of privacy management and reading news on Facebook.

Discussion

The results from this study build on the ongoing scholarly research focused on the role of privacy management and diversity in online news exposure and sharing. First, our results show that increased context collapse is related to sharing news on Facebook. This result shows that people are likely to view the diversity of their Facebook friends as context collusion, supporting the notion that users are mindfully using SNS when posting content (Davis and Jurgenson, 2014). In other words, more context collapse provides people to a large platform for sharing information and engaging in discussion about news information that is relevant to their lives. Instead of dampening online discourse, we find that context collapse seems to spur online disclosure. This could be due to the fact that people use context collapse mindfully to build social support (Vitak, 2012).

Next, we find that increased context collapse is related to reading more news on Facebook. Theoretically, these results may indicate that people do not selectively avoid news when they are situated in an increasingly diverse information environment. This finding adds to a growing body of literature showing that when users have increased control over their information environment, they may tend to read more news (Beam and Kosicki, 2014; Garrett et al., 2013). Furthermore, this result is consistent with the idea that people will engage with news despite being immersed in an increasingly diverse social environment (Messing and Westwood, 2014).

Our conditional process analysis reveals a mechanism for understanding the relationship between context collapse and online sharing. Our research supports the relationship found previously between online news reading and online news sharing (Beam et al., 2016; Weeks and Holbert, 2013). As a mediator, we find a significant indirect relationship between context collapse and online news sharing through news reading. In this process, people who have more types of Facebook friends and read the news those friends share on Facebook engage in more news sharing themselves. Therefore, increased diversity in online social networks could be related to more engagement with news content in general, which prepares people to share and discuss news more readily. Furthermore, we find the link between reading and sharing news is moderated by privacy management openness.

Our mediated-moderated model adds to the growing set of studies that have demonstrated support for CPM-based propositions in the context of Facebook (Child et al., 2012; Child and Petronio, 2015; Child and Staracher, 2016; Child and Westermann, 2013; Petronio, 2002). In particular, CPM theory predicts that contextual influences and individual motivational influences affect how people set up privacy rules and make decisions about further opening up or protecting privacy boundaries (Child and Petronio, 2011; Petronio, 2002, 2013). In this study, we show that in the context of Facebook individuals who were more comfortable sharing their private information in general also experienced a stronger individual privacy rule motivation to share more news stories with their Facebook friends. Privacy management practices moderate the mediated relationship between social network context collapse, news reading, and news sharing. In particular, our data support that openness in privacy management predicts increased news sharing for those who read news on Facebook, and those who are open show significantly more positive indirect effects on news sharing from diverse Facebook friends through news reading. Furthermore, when people were more concerned about issues of privacy management, they were less likely to share news online within a diverse network. When privacy was less of a concern, they viewed the enhanced perspectives gained by circulating the news in a diverse network more beneficial overall. Thus, CPM theory and issues of effective privacy management regulation help untangle the ideas about when context collisions versus collusions are more likely to occur among social media users who share news online (Child and Staracher, 2016). CPM theory is a valuable tool for predicting which people are likely to actively engage in online gatewatching and political discussion and which people do not have issues with perhaps more online commentary about news stirring controversy among their diverse online Facebook network.

We believe this research offers a contribution to the ongoing debate about echo chambers and filter bubbles. A theory-based democratic argument states that when users are placed in an information environment populated by people in their personal social networks and personalized algorithms, they are less likely to be exposed to diverse perspectives (Dylko, 2016; Pariser, 2011; Sunstein, 2007). Instead, our research indicates that when people connect with more diverse others, they are increasingly engaging in news exposure and sharing. One possibility here is that people may find news on social media more relevant than news found elsewhere because it is shared from their friends who are like themselves or hold similar views. However, we believe the finding that increased diversity drives news engagement shows that people find utility in surveillance of news information when having a more diverse social network. It may also normalize the routine of sharing news because they see friends from diverse contexts sharing news on Facebook. Therefore, we argue that social media use can lead to pro-social outcomes, including increased news engagement through reading and sharing news, especially for those with disparate contacts in their network.

However, we must acknowledge methodological limitations to this study. While our data allow us to make connections to online news sharing through context collapse and online news exposure, we do not have information concerning the type of news that our participants are viewing and sharing. We are unable to say that diversity in Facebook friends is related to diversity in the content that people are sharing. It is possible that coming across news from diverse others leads people to share more partisan information. Second, our study focuses on the affordance of social network contacts, not SNS algorithms. That means that our measure of news exposure is also unable to tease out algorithmic effects of the Facebook newsfeed. There is some evidence that Facebook’s proprietary personalized algorithm slightly favors pro-attitudinal news stories compared with opinion-challenging news stories (Bakshy et al., 2015; Tufekci, 2015). While some counter-attitudinal news headlines may be demoted in users’ newsfeeds, past research also indicates that when given the choice, people prefer to include at least some counter-attitudinal news sources in their newsfeeds (Beam, 2014). Furthermore, selective avoidance is unlikely to occur when counter-attitudinal information does pass algorithmic gatekeeping (Garrett et al., 2013; Messing and Westwood, 2014). Despite this methodological limitation, our evidence indicates that connecting with more types of people results in increased engagement with news, which runs counter to the idea of the social media echo chamber.

Next, our measure of news sharing does not tell us whether users are using Facebook’s privacy features to share to a subset of their online connections. CPM-based research can help predict people’s proclivity to using these features. Past research has shown that the vast majority of Facebook users rely on default settings of sharing stories with “friends only” or the “public” (Madden et al., 2013; Sleeper et al., 2013). We feel confident that our measure of online news sharing is likely to capture information disclosed to the entirety of most of our participants’ online Facebook “friends.” Furthermore, it is important to note that people’s imagined audience is often far smaller than the actual audience (Bernstein et al., 2013). Our measure of context collapse includes five contexts of relationships based on past research, but it is possible to be more granular with relationship context (e.g. close friends vs acquaintances). With these caveats, we believe that our findings, combined with other studies focused on the pro-social utility of using SNS for reading and sharing news and political information, offer some reasons for optimism that SNS is a growing platform for news engagement, especially at a time when many people have the freedom to opt out of news reading in favor of other choices (Arceneaux and Johnson, 2013; Beam et al., 2016; Gil de Zúñiga et al., 2012, 2014; Prior, 2005, 2007; Holt et al., 2013; Vitak et al., 2011).

This study has a few other notable limitations. We rely on a diverse opt-in Internet panel for our results that resemble demographics similar to those of the US adult population. Fortunately, as noted in method section above, prior research has indicated that these volunteer samples provide data of similar quality to random sampling methods (e.g. Ansolabehere and Schaffner, 2014). While imperfect, we believe our sample provides useful data, especially considering that we are collecting data about online activities from a diverse national sample of Internet users during a national election.

Next, our measure of online sharing includes both posting “news links and information” and “thoughts or opinions about the news” on social media. We combined these measures as they were highly correlated and provided us with confidence that respondents were active in sharing news information on SNS. However, it would be useful in future studies to investigate these as distinct behaviors. For example, simply sharing a politically charged news story with no individual frame for thinking about the story reveals less personal insight than sharing the same news story and also revealing to a Facebook network how the story meshes (or not) with personal and political values. We also are unable to differentiate the motivations and content that is shared. It is possible that people are sharing information that they believe will be useful for their social network contacts to make a quality decision or posting inflammatory or trolling posts to stymie opponents. Regardless of the content, we do believe that people are likely seeing context collapse as a “collusion” because they are mindfully engaging in sharing. Even deeper insight into the management of private information related to news story sharing practices on Facebook could be advanced by considering individual user and Facebook network commentary that occur with sharing news stories on Facebook using CPM as a theoretical perspective. Taken together, future studies may use experimental or observational approaches to better understand the exact nature of context collapse in sharing particular types of online news.

Future research should also investigate the role of context collapse on other political variables such as political participation and political knowledge. Furthermore, this study was looking at context collapse in terms of the different contexts of relationships with their online friends. While this perspective on context collapse is consistent with past context collapse research, future studies should also focus more closely on the political makeup of one’s social network. We did control for perceived political agreement in our models, which indicated some support for the idea that people are more likely to engage in news reading and news sharing when their network has more people who support their own views. However, our final models showed that perceived agreement was not a predictor of sharing news on Facebook. Instead, openness in privacy management and reading news online played a more important role in online sharing behaviors compared with perceived agreement.

Conclusion

Our study shows that online social media users who have friends that span social contexts are more likely to read and share online news. This study is able to differentiate between specific types of news engagement while demonstrating the important role of privacy management in communication in online activities. This is a step toward understanding the political consequences of increased reliance on social media for online news. As Facebook and other SNSs continue to reshape the process of journalism, we need to continue to understand how this transformation impacts public opinion.