Abstract
Social media allow users some degree of control over the content to which they are exposed, through blocking, unfriending, or hiding feeds from other users. This article considers the extent to which they do so for political reasons. Survey data from Pew Research suggests that political unfriending is relatively rare, with fewer than 10% of respondents engaging in the practice. Analysis finds support for the idea that political unfriending is most common among those who talk about politics, those strongest in ideology, those that see the most politics in social media, and those that perceive the greatest political disagreement in their social networks. This suggests that social media are not exacerbating the political information gap as political information on social media is likely still reaching the least politically engaged, whereas the most politically engaged may opt out of political information within social media but still receive it elsewhere.
Introduction
Research about social media represents a growing interdisciplinary subfield, with political science, communication, and sociology scholars all interested in the role social media may play in politics (Edgerly et al., 2013). Notably, though, as a result of network customization, the social media experience is different for each user, making it a challenge to study. This customization matters, affecting what types of information people see on social media (Bode, 2016).
Of particular importance is the degree to which users take advantage of social media customization specifically for the purpose of avoiding politics. If users engage in this sort of opting out regularly, social media may be a less important medium of political information transmission than some scholars presume (Bode et al., 2014; Boulianne, 2015; Kim et al., 2013). Equally important is what types of people are customizing their feeds to avoid political information or confrontation. For those least engaged in politics, social media may offer an incidental way of happening upon political information, when they might not otherwise seek it out. Those most engaged, on the other hand, likely see political information elsewhere (Bode and Dalrymple, 2016; Prior, 2007). Therefore, the consequences of different types of people avoiding politics on social media may vary.
For these reasons, this study investigates the extent to which users exercise control over their social media experiences, and who tends to engage in avoidance of political information by unfriending people who post about politics on social media.
Why it matters: opting out of politics
In a basic understanding of media effects, various elements, including motivations related to uses and gratifications, habits, and predispositions lead a particular user to choose a particular type of media (Ruggiero, 2000). Part of that choice is the amount of control such a user will have over the content she sees from that medium. By control, I mean the extent to which users consciously expose themselves to a particular medium and its associated content. Imagine a spectrum of control in which one pole is incidental exposure – users come across content without intending to at all – and the other pole is conscious exposure, in which users both know the type of content present in a particular medium and consciously choose to expose themselves to that content. Some media, like RSS readers and news aggregators, are almost entirely customizable, and thus a high control environment. Others, such as network television in its early days, allow users very little control once they decide to partake in using the medium, and are thus a low control environment.
The amount of control exerted by a user in a particular medium is extremely important, because it directly affects the content to which a user is exposed. In a high control environment, this will likely closely resemble the ideal content of the user, since she may customize it as much as she likes. In a low control environment, on the other hand, content may be wide-ranging in subject and scope and may diverge from the user’s personal preferences for content.
Social media 1 are one genre in which control is only partial. In social media, users often choose to use a platform for non-political (usually social but also informational) purposes (Smith, 2011). However, once they have opted into a particular platform and its corresponding network, they may be exposed to information they did not seek out or care to see, including political information (Bode, 2016). In this way, social media resemble a low control environment. However, social media allow greater customizability than do most low control environments. For instance, if another user is exposing you to information you do not care to see, you can simply remove that user from your network (generally referred to as “unfriending”).
It is unclear how this partial control plays out in practical terms. Some scholars have voiced concerns that the customizability of social media will allow people to insulate themselves from information to which they would otherwise be exposed (Pariser, 2012; Prior, 2007; Sunstein, 2007). On the other hand, it is not entirely clear that it is as easy to opt out of political information online as it may first appear. Similar concerns surfaced following the dawn of television, but incidental exposure to political information still occurred (Blumler and McQuail, 1969; Downs, 1957; Krugman and Hartley, 1970). Social media, much like television, provide an environment in which political information is interspersed with other types of content, therefore reaching even the politically uninterested (Bode, 2016). The question remains, however, as to the extent to which users can protect themselves from such information – the control that social media allow does not matter if that control is not used.
Examining the extent to which people intentionally avoid political information in social media is particularly important, given that exposure to politics and engaging in political expression on social media has been shown to be positively associated with democratic behaviors and attributes, including civic engagement, volunteerism, political knowledge, efficacy, and participation (Bode et al., 2014; Boulianne, 2015; Jang et al., 2014; Vitak et al., 2011), and the fact that people are increasingly reliant on social media for news and information (Gottfried and Shearer, 2016).
Unfriending
While unfriending has been an option since the dawn of social media in the early 2000s, there is only scant literature examining its trends, tendencies, and implications. Much of this work is theoretical in nature, speculating on the functions unfriending can serve in the broader social networking site (SNS) arena (Light and Cassidy, 2014), or the issues unfriending may cause for measurement of social influence (Noel and Nyhan, 2011).
Research on unfriending is also generally limited in that it does not discuss political unfriending specifically, rather focusing on more general unfriending motivations (Sibona, 2014). What little research has focused on political unfriending was conducted in an extremely heated setting, focusing on the Israel–Gaza conflict in 2014 (John and Dvir-Gvirsman, 2015). In this case, unfriending was most prevalent among the more ideologically extreme and the more politically active. While this study gives us great insight into political unfriending, it is limited by its context (a unique and complicated conflict), and its narrow focus on Facebook in Israel (only used by about half the population, whereas 65% of adults and 76% of online adults use at least one SNS in the United States) (Perrin, 2015).
Finally, it is worth noting that there are various constraints affecting the exercise of the control available to users. First, there may be structural constraints affecting the control users may exert over their networks – users may not know how to unfollow or unfriend within a given platform, creating a de facto lower degree of control. This is exacerbated by the fact that Facebook changes its settings quite regularly, making it hard to keep up for casual users. Additionally, there are social constraints encouraging users to resist “unfriending,” opting out of a network, or removing someone from their network. There may be an expectation of retaining someone in your network, particularly if they are close friends, colleagues, or relatives. Even if a crazy uncle annoys you with his Facebook rants, you may not feel like you can remove him from your network without offending him or other family members. This is evidenced by the fact that uses and gratifications offered for using social media are overwhelmingly social, with staying in touch with friends and family members, seeing photographs and videos, and reconnecting with old friends listed as the most powerful motivations (Ellison et al., 2011; Mitchell and Page, 2013). Social media, in general, are used for relationship maintenance, which should create aversions to unfriending, as it is explicitly anti-social. Finally, people in general are just not that absorbed by politics (DeLuca, 1995), and may find it easier to skim over content they do not enjoy, rather than engaging in the technically and socially more difficult action of unfriending. These constraints may work to produce a stickier network than would otherwise be expected. That is, people may be less likely to opt out of networks or information flows within social media as a result of these constraints than they otherwise would be, thus exerting less control over their social media environment than they actually can. As an example of this, even during the intense Israel–Gaza conflict, only 16% of respondents reported unfriending or unfollowing for political reasons (John and Dvir-Gvirsman, 2015).
Expectations
Several elements should affect whether someone chooses to unfriend another person in their network for political purposes.
First, people with stronger political associations might have stronger preferences with regard to political information, and therefore greater motivation to customize the type of political information they see from others. This could be because of a preference for ideologically congruent content, a preference for greater or less political content, or a frustration with the political content of others not meeting one’s expectations. Stronger political associations might manifest in strength of political preferences, as indicated by ideological strength (not just whether you consider yourself conservative or liberal but the extent to which you feel strongly in either direction), as seen in John and Dvir-Gvirsman (2015). Those with stronger political dispositions are more likely to engage in selective exposure, or otherwise tailor the content to which they are exposed (Stroud, 2008; Taber and Lodge, 2006). This leads us to believe:
H1A: Those highest in ideological strength will be more likely to unfriend others for political reasons.
Strong political associations might also be represented through a tendency to engage in politics more broadly, which is operationalized here as political talk. Political talk is traditionally an important predictor of political engagement, affecting policy opinions (Barabas, 2004), tolerance (Mutz, 2002), and participation in electoral politics (McClurg, 2006). Due to its effects on such a wide variety of outcomes, we would expect that political talk would likewise increase the likelihood of political unfriending – those engaging most in talk have stronger political preferences (Verba et al., 1995), which will increase the likelihood of them tailoring the political information they see online. This leads us to expect:
H1B: Those highest in political talk will be more likely to unfriend others for political reasons.
A second element that likely comes into play is the opportunity one has to be exposed to political information in the first place. If one’s friends never post about politics, there is not much reason to unfriend them for doing so 2 and, therefore, this expectation is relatively straightforward:
H2: Those with friends who post more about politics will be more likely to unfriend others for political reasons.
An additional likely reason for choosing to exert control over the content you see on social media is related to disagreement (Nir, 2011). Part of the reason people are often opposed to political content on social media (Vraga et al., 2015), and opposed to talking about politics more generally (Eliasoph, 1998), is because they perceive it to be in opposition to their views (Lee et al., 2014), and those who perceive greater disagreement with their network engage less in social media in general (Grevet et al., 2014). For this reason, we should expect that the more users perceive disagreement in their social media networks, the more they would prefer to opt out of such content altogether.
H3: Those who perceive greater disagreement in their social media networks will be more likely to unfriend others for political reasons.
Finally, as this is an exploratory project, I include a number of control variables, in order to see what correlates of unfriending emerge. These are based first on existing research on unfriending (John and Dvir-Gvirsman, 2015), which finds that age, political Facebook use, ideological strength, and perceived exposure to disagreement are all predictive of political unfriending. Additional variables, including gender, education, race and ethnicity, income, and party, are included to reflect existing research on political behaviors in the United States more broadly (Verba et al., 1995), and are detailed below in the data section. In order to explore which of these variables predict political unfriending, I pose the following research question.
RQ1: What other variables predict political unfriending?
Data
In order to test the hypotheses posed above, I use survey data from Pew Research Center’s Internet, Science, and Technology Project. In 2012, they fielded a survey focusing on use of search engines and social networks with an emphasis on political activities. The total sample consists of 2253 individuals (1352 respondents contacted by landline with a response rate of 11.1%, and 901 respondents contacted by cell phone with a response rate of 10.8%), reached via random digit dialing and surveyed with live interviewers from 20 January to 19February 2012. Data are weighted to account for number of adults in each household, and then to match national population parameters for sex, age, education, race, Hispanic origin, region (US Census definitions), population density, and telephone usage (all descriptive statistics included below are post-weighting). 3
Measures
The main outcome variable of interest is whether people report unfriending others for political reasons. Pew asks this in a series of questions: “When using social networking sites, have you ever blocked, UNfriended or hidden someone because they: (a) posted TOO FREQUENTLY about politics or political issues; (b) posted something about politics or political issues that you DISAGREED with or found OFFENSIVE; (c) ARGUED about political issues on the site with you or someone you know; (d) disagreed with something YOU posted about politics or political issues; or (e) posted something related to politics or political issues that you worried would OFFEND your other friends or people who follow you” (capitalization is all from Pew, 2012). The frequency of each of these measures is shown below, in Figure 1. In general, these five measures capture a broader tendency to unfriend someone for political reasons. They are also strongly related, and so were combined into a mean index (α= 0.77, mean (M) =0.07, standard deviation (SD) =0.19). 4
Reasons for unfriending in social media.
Due to the political nature of the outcome, independent variables reflecting political predispositions and tendencies are included. These include party (Republican, 27.2%; Independent, 37.2%; Democrat, 35.6%), strength of ideology (a scale ranging 0 to 2, where 0 is moderate, 1 is conservative or liberal, and 2 is very conservative or very liberal, M=0.78, SD=0.68), and conversation about politics, measured here by the question: “How often, if ever, do you talk about politics or current events with your family and friends?” (1 never to 4 very often, M=2.89, SD=1.01).
Additionally, social media-specific variables should also play a role. First, I include a measure of political social media motivations, a mean index of the following items: “Overall, how important are social networking sites to you personally when it comes to: (a) keeping up with political news, (b) debating or discussing political issues with others; (c) finding other people who share your views about important political issues; and (d) recruiting people to get involved with political issues that matter to you” (measured from 0 not at all important/don’t do this to 3 very important, α=0.87, M=0.88, SD=0.84) 5 . Additionally, I include the key variable of how often your friends post about politics: “How about the people you are friends with on social networking sites? How much of what THEY SHARE AND POST is related to politics, political issues or the 2012 elections?” (0 none to 4 almost all of it, M=1.28, SD=0.98). Finally, Pew measures the extent to which users perceive political disagreement in their networks: “How often do you disagree with the political opinions or political content your friends post on social networking sites?” (0 never to 3 almost always, M=1.14, SD=0.65).
Additionally, basic demographics are included, consisting of gender (51.3% female), age (M=46.4, SD=18.3), education (measured in seven categories, M=4.3, SD=1.7: 0.5% less than 8th grade, 8.7% some high school, 31.3% high school graduate or equivalency, 2.2% trade school, 24.6% some college (includes associates degree), 17.4% college graduate, 11.1% post-graduate training or degree), income (measured in nine categories, M=4.74, SD=2.46, with 56.5% making less than $50,000, 27.5% making between $50,000 and $100,000, and the remaining 15.9% making more than $100,000), race (23.8% non-White), and ethnicity (14.2% Hispanic).
Analysis and results
As seen in Figure 1, the most common cause of unfriending is people posting too frequently about politics, with disagreement and arguments less frequent reasons. Notably, less than 10% of the sample reports any given unfriending behavior, suggesting that political unfriending is a rare behavior.
Because the dependent variable is roughly continuous, an ordinary least squares regression was estimated with political unfriending as the outcome. Results can be found in Table 1.
Predicting political unfriending.
Note: dependent variable is an index of five variables indicating different types of political unfriending. n=2078; R2=0.08.
Several demographic characteristics seem to play a role in predicting political unfriending, with older, male, non-Hispanics and non-Whites more likely to unfriend or unfollow. All three political predispositions predict unfriending, with Democrats more likely than Republicans to unfriend. Recent research suggests that online networks of Democrats tend to be more homogeneous than those of Republicans (Colleoni et al., 2014), which may suggest a broader preference for ideologically congruent content among Democrats. Those who are talking about politics more are also more likely to unfriend for political reasons, supporting H1b. The effect of political talk is dramatic – in general, unfriending is happening most among those who are engaging in the most political discussion (see Figure 2).
Unfriending for political reasons by political talk.
And as expected by H1a, ideological strength is also significantly predictive of political unfriending, with the strongest ideologues most likely to unfriend for political reasons. To illustrate this further, I broke down scores on political unfriending by the five different categories of ideology. As seen in Figure 3, those most likely to unfriend for political reasons are the very conservative and the very liberal, with liberals in general somewhat more likely to unfriend than conservatives. This suggests that moderates – who are more likely to be persuaded by political information in the first place – are also less likely to opt out of it within social media (Zaller, 1992).
Unfriending for political reasons by ideology.
Interestingly, political motivations for using SNS are unrelated to unfriending for political reasons. However, the other two SNS-based measures are both significant predictors. Those who see more content from friends posting about politics are more likely to unfriend for political reasons, supporting H2. And those who perceive greater disagreement in their social networks are also more likely to unfriend for political reasons, lending support to H3.
Discussion and conclusions
In general, it is clear that the vast majority of social media users are not regularly unfriending others in their networks for political reasons. When they do so, it is most likely to be for reasons of volume. This suggests that people are unlikely to opt out of political content entirely on social media, simply because it is political. This reinforces the partial controlled nature of social media – even though users are free to opt out of political content, they do so only rarely. Because exposure to political content is associated with increased political knowledge (Bode, 2016), tolerance (Mutz, 2006), and moderation (Testa et al., 2014), this is positive news.
Ideological strength and political talk also play a major role. On the one hand, the major influence of ideological strength invites concerns about polarization, with those with the most strongly held views also most likely to avoid political posts from others (Barberá, 2014). On the other hand, the role of political talk suggests that those most likely to be getting political information elsewhere – through offline conversations – are those most likely to opt out of that information on social media. Therefore those less engaged in politics may still be exposed to it via social media (Bode, 2016), which could help to narrow the knowledge gap between the most and least engaged (Price and Zaller, 1993).
The specifics of an individual’s network also matter a great deal. In order to opt out of politics, one’s network must first post something about politics. Only about one-third of people tend to post about politics on social media, but nearly three-quarters report that their friends post about politics at least sometimes (Pew, 2012). Most people, then, have at least some exposure to politics, though if this type of content becomes too frequent, it increases the likelihood that users will begin to opt out of it.
Finally, perceptions of political disagreement in a network increase the likelihood of unfriending. This is particularly interesting given that disagreement is not the most common reason for unfriending given by respondents (though it is a close second). A social desirability bias towards tolerance of the other side may affect the accuracy of reporting on these motivations. It is also worth noting that people are notoriously bad at accurately recognizing disagreement (Wojcieszak and Price, 2012), suggesting that their perceptions of their networks may not be entirely accurate.
Finally, it is worth noting that the variance explained by the model is quite low, suggesting that there is a great deal about political unfriending that we do not yet understand. Part of this is likely due to imperfect measures of political unfriending, which are relatively blunt and offer only dichotomous answer possibilities. Political contention on social media may be a relatively idiosyncratic occurrence (Vraga et al., 2015), with nuances that go far beyond what this analysis has uncovered. It is also very likely to be platform-specific – different social media have different structures, affordances, and social norms (Gottfried and Shearer, 2016; Halpern and Gibbs, 2013), all of which are likely to change the patterns and predictors of political unfriending. Future research should continue to delve into this question to further understand when and why users choose to avoid political content in different social media. Development of better measures, with greater variance, and the use of qualitative research would likely result in a more nuanced understanding of what circumstances, characteristics, and motivations are associated with political unfriending.
In general, pruning the news feed is the exception rather than the norm. Users engage in control of the political information to which they are exposed only rarely, and those most likely to do so are likely still getting such information elsewhere. For these reasons, social media is likely contributing to the dissemination of political information to low-information voters.
Footnotes
Correction (June 2025):
Declaration of Conflicting Interest
The authors declare that there is no conflict of interest.
Funding
This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.
Notes
Carnegie Corporation of New York Grant
The open access article processing charge (APC) for this article was waived due to a grant awarded to Research & Politics from Carnegie Corporation of New York under its ‘Bridging the Gap’ initiative. The statements made and views expressed are solely the responsibility of the author.