Sage Journals: Discover world-class research

Abstract

This study considers politically motivated unfriending/unfollowing on Facebook in the lead up to the 2020 Presidential election. As social media has grown more central to public discourse, it has been suggested that these types of avoidance behaviors might contribute to the formation of partisan echo chambers, whereby users may limit their exposure to competing viewpoints and corrective information by eliminating the network ties that transmit it. Building on prior research, this study explores the relationship between politically motivated avoidance and user perceptions of social media’s credibility as a source of political information. A pre-election survey of voting age adults in the state of Florida is analyzed to identify both the frequency and predictors of politically motivated avoidance. The results suggest that avoidance is a function of both individual political characteristics and user perceptions of information credibility. The findings are considered in the context of previous literature, and important avenues for future research are discussed.

Keywords

unfriending selective avoidance Facebook social media information credibility

Over the past 20 years, social networking sites (SNS) such as Facebook and Twitter have evolved into a primary medium for political news, engagement, and information seeking. Data collected by the Pew Research Center show that these platforms are slowly outpacing traditional information mediums (such as print and radio), and roughly two-thirds of Americans now receive at least some portion of their news through social media (Shearer, 2018; Shearer & Matsa, 2018). With these trends has come a growing scholarly interest in the curation and maintenance of online social networks. One particular area of concern has been the practice of politically motivated unfriending/unfollowing, a form of selective avoidance whereby SNS-users remove those who share politically disagreeable content from their social networks. It has been suggested by some that politically motivated avoidance might undermine the broader goals of deliberative democracy by minimizing SNS-users’ exposure to counter-attitudinal messaging (Lu & Lee, 2019; Mutz & Martin, 2001; Sunstein, 2017). For example, Sunstein (2017)—among others—has cautioned against the “architecture of control” provided by social media platforms, arguing that it may contribute to the formation of partisan “echo-chambers,” which are believed to facilitate both polarization and the spread of misinformation (i.e., Hahn et al., 2015; Iyengar & Hahn, 2009).

Drawing on these concerns, some prior studies have conceptualized politically motivated unfriending/unfollowing as akin to partisan selective exposure, whereby SNS-users are believed to mitigate the cognitive dissonance that arises from exposure to political disagreement by severing the network ties through which it is transmitted (i.e., Yang et al., 2017; Zhu et al., 2017). While relatively nascent, this body of literature has yielded valuable insights in the areas of network curation, confirmation bias, and the increasingly critical role that social media play in the public sphere. However, when considering the motivations underlying these behaviors, prior research has often overlooked user perceptions of social media’s credibility as a medium for political news and commentary. Recent work by Metzger et al. (2020) found that selective exposure may in fact be motivated by user perceptions of informational credibility rather than a desire to mitigate cognitive dissonance. In the case of social media, this suggests that politically motivated avoidance could, in many instances, represent a form of “boundary regulation,” whereby SNS-users seek to manage and maintain information credibility in their social networks. Drawing from recent literature on uses and gratifications in digital media, as well as the broader literature on boundary regulation and network curation, there is reason to believe that these may be strong incentives for politically motivated avoidance.

This study considers these ideas by examining politically motivated avoidance (measured as unfriending/unfollowing) ahead of the 2020 presidential election in the United States. Results from a survey of voting age adults in the state of Florida are analyzed to examine the frequency and antecedents of politically motivated avoidance on Facebook in the 3 months preceding the presidential election. Building on prior studies, avoidance is considered as a potential function of both ideological motivations and user perceptions of information credibility. The findings suggest that politically motivated avoidance is indeed related to both political characteristics and perceptions of credibility, with reliance on Facebook, perceptions of its informational credibility, and ideological intensity being among the most substantial predictors of avoidance in the face of political disagreement. The results allow for a unique examination of politically motivated unfriending/unfollowing in the US context, while also contributing to our broader understanding of the factors associated with network management and selective avoidance.

Literature Review

Social media are now a central feature of American public life, serving both as a primary source of news and information for many Americans, as well as a conduit for various forms of political engagement. Data show that SNS-users increasingly rely on social media to stay up-to-date and informed on political matters (i.e., Shearer & Matsa, 2018), and as a consequence, elected officials increasingly rely on social media to communicate with the public (i.e., Evans et al., 2014; Fontaine & Gomez, 2020; Straus et al., 2013). According to data collected by the Pew Research Center, roughly two-thirds of Americans now receive at least some portion of their news through social media (Shearer & Matsa, 2018). For many SNS-users, social media also provide a convenient means of political engagement, including the ability to advocate for candidates and referenda, join political groups, sign online petitions, and participate in virtual events such as “debate watch-parties” (Neely et al., 2020).

This increased reliance on social media has led to several concerns on the part of scholars and social commentators alike. Among these concerns has been the tendency of online networks toward homophily (McPherson et al., 2001), which some argue may undermine democratic values, such as robust political debate and exposure to competing viewpoints (i.e., Sunstein, 2001, 2017). Yang et al. (2017) note three distinct mechanisms through which social media may facilitate the homogenization of social networks. These include (a) algorithmic filtration, (b) selective affiliation, and (c) selective avoidance. In the case of algorithmic filtration, social networking platforms such as Facebook employ sophisticated algorithms that direct users toward desirable content (i.e., “Friends,” topics, stories, etc.) and away from undesirable content. In contrast, selective affiliation and selective avoidance are user-driven processes. Selective affiliation refers to the pre hoc construction of one’s social network (i.e., the initial decision of who and what to “follow”), while selective avoidance refers to the post hoc removal of undesirable content, typically through the dissolution of network ties. Politically motivated content filtration represents a distinct form of selective avoidance, wherein SNS-users remove sources of undesirable content on a post hoc basis. In the context of Facebook, this can take the form of “unfriending” or “unfollowing” another user in response to disagreeable or objectional content. Yang et al. (2017) termed this “post hoc user filtration.”

Additional concerns have also been raised over the utility and reliability of social media as a political information environment. While social media provides a myriad of access points for SNS-users to learn about politics and engage in democratic processes, many remain skeptical about the information they encounter there. Data from the Pew Research Center highlight this paradox. They found that while Americans increasingly rely on social media to stay informed about politics, a majority (57%) “. . . expect the news they see on social media to be largely inaccurate” (Shearer & Matsa, 2018). There is some evidence that these concerns may be well-founded. For instance, Anspach and Carlson (2020) discovered that there is a tendency for facts to be misrepresented in the user comments which often accompany news stories on social media. They note that many SNS-users rely on these comments as “shortcuts to political information” and that those who do are often misinformed about news and current events. Even more critical concerns have been raised over the documented susceptibility of social media to both inadvertent misinformation and to large-scale, coordinated disinformation campaigns, such as those observed during the 2016 presidential election in the United States (Ferrara et al., 2016; Persily, 2017) and the subsequent COVID-19 pandemic (Kouzy et al., 2020; World Health Organization, 2020). Such high-profile incidents appear to have undermined public confidence in the reliability of social media and online news.

Politically Motivated Avoidance and Partisan Selective Exposure

Prior studies have often viewed politically motivated unfriending/unfollowing through the lens of partisan selective exposure theory (Yang et al., 2017; Zhu et al., 2017). Selective exposure (referred to in political contexts as partisan selective exposure) represents a form of confirmation bias, wherein individuals favor exposure to information that affirms rather than challenges their existing beliefs (Stroud, 2008). The psychological underpinnings of this behavior are found in Festinger’s (1957) theory of cognitive dissonance, which argues that individuals engage in “information seeking” and “information avoidance” in an effort to minimize the discomfort that arises from experiencing cognitive dissonance. Early attempts to measure selective exposure tended to conflate these seeking and avoidance mechanisms. However, recent efforts have taken a more nuanced view of the phenomenon, suggesting a multidimensional approach to information consumption animated by varied motivations and selection mechanisms (i.e., (Garrett, 2009; Garrett & Stroud, 2014; Jang, 2014; Song, 2017; Tsfati, 2020).

For example, Garrett (2009) drew a stark distinction between seeking and avoidance, arguing that a preference for attitude-consistent news and information does not equate to avoidance of attitude-challenging information or preclude such exposure. Garrett and Stroud (2014) further defined this distinction between what they termed “. . . selective approach, the tendency to seek information consistent with one’s prior beliefs, and selective avoidance, a drive to avoid contradictory information” (p. 681—italics added). Along with Garrett and Stroud (2014), a number of scholars have emphasized the importance of this distinction, noting that the latter tendency poses a more serious threat to deliberative democracy insofar as it minimizes exposure to diverse opinions and inhibits the free exchange of political ideas (Jang, 2014; John & Dvir-Gvirsman, 2015; Skoric et al., 2018). This is a significant point in the case of politically motivated unfriending/unfollowing, which represents a form of post hoc selective avoidance.

Prior research in this area suggests that Americans may be more likely to engage in this type of selective avoidance than SNS-users in other settings. In 2012, the Pew Research Center found that just under one-fifth of US based SNS-users (18%) had unfriended, blocked, or “hidden” a member of their social network over political posts (Rannie & Smith, 2012). In a subsequent survey taken 4 years later, that number had risen to nearly 40% (Duggan & Smith, 2016). A more recent study found that one-fifth of US-based Facebook users (22%) have unfriended or unfollowed someone because of something they posted about the COVID-19 pandemic (Neely, 2021). Comparatively, several studies examining politically motivated unfriending/unfollowing in non-US settings have found slightly lower levels of avoidance, even in politically charged contexts. These included Zhu et al.’s (2017) study of the Umbrella Movement protests in Hong Kong, during which just over 15% of respondents reported engaging in politically motivated unfriending. Similarly, John and Dvir-Gvirsman (2015) found that 16% of Israelis had blocked or unfriended a member of their social network over posts made during the 2014 Israel–Gaza conflict. Consistent with partisan selective exposure theory, prior research has also demonstrated a significant link between politically motivated avoidance and greater ideological intensity, wherein those with stronger ideological tendencies are more likely to sever network ties in response to political disagreement (Bode, 2016; John & Dvir-Gvirsman, 2015; Yang et al., 2017).

While partisan selective exposure has been a popular approach to the study of politically motivated unfriending/unfollowing, some important counterarguments have been presented in the literature. For instance, Dubois and Blank (2018) found evidence that these concerns may be overstated, noting that “Whatever may be happening on any single platform, when we look at the entire media environment, there is little apparent echo chamber” (p. 740). Bode (2016) offered a similar assessment, suggesting that unfriending among partisans on social media may be of little consequence, as these individuals are more likely to be politically engaged and thus exposed to counter-attitudinal ideas through a variety of mediums.

Another counterargument has suggested that avoidance over political posts may be motivated by moral disagreements rather than purely ideological ones. Viewed through this lens, politically motivated unfriending/unfollowing may represent a form of moral boundary regulation in the maintenance of social networks rather than ideological filtration (i.e., John & Gal, 2018). In a recent study, Neubaum et al. (2021) found that “. . . moral judgements of political statements are moderately related to unfriending decisions” (p. 1). A number of obvious scenarios support this argument, such as the case of political cyber-bullying, racist/xenophobic political posts, and vulgarity. However, when interpreting these studies, it is important to recognize the increasingly blurred lines between moral and ideological disagreements. For example, data collected by the Pew Research Center (2019) highlight a growing tendency among Americans to view those with different ideological perspectives as “immoral” or “closed-minded.” This tendency complicates the measurement of such distinctions.

Politically Motivated Avoidance and Perceptions of Information Credibility

Prior studies have provided some evidence that ideological confirmation bias might indeed motivate political avoidance. However, these studies have often overlooked the potential role that user perceptions of information credibility play in these behaviors. Recent work by Metzger et al. (2020) found that consumer judgments of information credibility—though often ideologically biased—are stronger predictors of selective exposure than cognitive dissonance. This is perhaps unsurprising when we consider the uses and gratifications literature, which emphasizes that information seeking and learning are among the primary motivations for use and adoption of both traditional and emerging media platforms (i.e., Papacharissi & Rubin, 2000; Sundar & Limperos, 2013). As such, it can be argued from this literature—as well as the broader literature on boundary regulation and network curation—that some SNS-users might engage in selective avoidance as a means of maintaining the credibility of their social networks as mediums for political news and information seeking. From this perspective, SNS-users may be inclined to filter out those whom they perceive as sources of unreliable or inaccurate political information to optimize information seeking and learning motivations.

While distinct from the broader concept of information utility, concerns over credibility are reflective of an information utility motivation. Wagner (2017) succinctly defines information utility as “. . . the extent to which information is useful to citizens” (p. 534). Strutzman and Hartzog (2012) identify utility as one of the four principal motivations for boundary regulation in social networks. While their analysis does not directly examine the accuracy of political information, they note that network management “. . . with an eye toward utility is indicative of an optimizing process, in which communication is effectively managed between two domains, limiting the risk of inadvertent or unwelcome disclosure. . .” (p. 775). They also note specifically that the regulation of network linkages is one option for such boundary regulating behavior.

The absence of effective quality controls in social networks such as Facebook means that these gatekeeping functions, once the province of traditional media, now fall largely to individual users. While unfriending/unfollowing may be seen as an extreme response to “uncredible” posts, the sheer volume of political content and commentary shared on social media—along with the limited processing capacity of users—may rationalize such a response on the part of those wishing to maintain the informational credibility (and utility) of their social network. Indeed, Lin et al. (2016) have pointed out the tendency of information consumers toward such “effortless information judgments” (p. 265).

There is already some evidence to support these considerations. In a more general study of content filtration, Kwak et al. (2011) found lack of “informativeness” to be a primary motivation for “unfollowing” on Twitter. In addition, while perceptions of credibility have not been a primary consideration in prior studies of politically motivated avoidance, some recent work on partisan selective exposure has suggested a link between information credibility and selective avoidance. For example, Wagner (2017) found information utility to be a significant predictor of selective approach and exposure when examining attentiveness to political debates. More recently, Metzger et al. (2020) examined perceptions of information credibility as a counter explanation for selective exposure. Building on the notion that “credibility judgements play a role during information search,” they found that “selective exposure patterns fell more in line with credibility perceptions than with reports of cognitive dissonance” (p. 21).

In an effort to extend these ideas, this study considers user perceptions of social media’s credibility as a potential predictor of politically motivated unfriending/unfollowing. Building on this logic, we might expect to see higher rates of politically motivated avoidance among those who express greater concern over the credibility of social media as a source of news and political information. However, it should be emphasized that the demarcation of these motivations is ambiguous at best. Notably, the literature on hostile media effects demonstrates a tendency on the part of consumers to interpret information credibility based on ideological preferences, such that perceptions of source or information credibility may be largely dependent on partisan predilections (Clark & Maass, 1988; Hansen & Kim, 2011; Metzger et al., 2020). Recent work by Gearhart et al. (2019) has documented the extension of this tendency to social media and the user comments accompanying news stories.

Research Questions and Hypotheses

Based on the discussion outlined above, this study considers two specific research questions and five directional hypotheses. In each case, politically motivated avoidance is measured as a function of politically motivated unfriending/unfollowing on Facebook. The first research question concerns the frequency of politically motivated avoidance in the months leading up to the 2020 election. Drawing from earlier data collected by the Pew Research Center, Bode (2016) argued that politically motivated unfriending is relatively rare in the United States, though others have suggested that Americans might be more likely, on average, to engage in such behaviors (i.e., John & Dvir-Gvirsman, 2015). The contentious nature of the 2020 election makes it an opportune timeframe in which to examine how individual SNS-users are shaping the political contours of their social networks and how likely Americans are to engage in politically motivated avoidance.

RQ1. How common was politically motivated avoidance in the 3 months prior to the 2020 presidential election?

Consistent with the literature on partisan selective exposure, political attitudes and affiliations are believed to influence avoidance. This analysis considers partisan affiliation and ideological intensity as potential predictors of politically motivated avoidance. It has been argued in the literature that conservatives may be more prone to selective approach and avoidance (i.e., Garrett & Stroud, 2014). Drawing from the social psychology literature, Nam et al. (2013) suggest that conservative tendencies toward avoidance may be motivated by “. . . stronger needs for order, structure, consistency, and closure. . .” (p. 1). However, in practice, the empirical literature has been inconclusive with regard to the effect of partisan affiliation on selective avoidance. Some studies have found that conservatives and/or Republicans are more prone to selective approach and avoidance (Garrett & Stroud, 2014; Parmelee & Roman, 2020), while others have found these behaviors to be more common among liberals and/or Democrats (i.e., Bode, 2016; Heimlich, 2012; Knobloch-Westerwick & Meng, 2009). Given the prevailing lack of consensus in the literature, the relationship between party affiliation and politically motivated avoidance is treated as an exploratory research question in this analysis, rather than a directional hypothesis.

RQ2. Did politically motivated avoidance differ based on party affiliation in the 3 months prior to the 2020 presidential election?

While the relationship between party affiliation and selective avoidance has been somewhat inconclusive, the literature has suggested a strong link between politically motivated avoidance and partisan intensity. In this case, partisan intensity refers not to an individual’s party affiliation, but rather the intensity with which they hold to and espouse their ideological beliefs. Several previous studies have suggested a strong causal link between the intensity of ideological beliefs and a preference for partisan media (i.e., Knobloch-Westerwick, 2014; Stroud, 2010), suggesting that selective avoidance might be more common among those with strong ideological tendencies. Prior studies of politically motivated unfriending have tended to reflect this relationship (i.e., Bode, 2016; John & Dvir-Gvirsman, 2015; Yang et al., 2017). It is hypothesized that a similar relationship will be seen in this analysis.

H1. The likelihood of politically motivated avoidance will be positively related to ideological intensity.

Drawing from the previous discussion outlined above, those who perceive social media as rife with inaccuracies and misinformation are expected to engage in politically motivated avoidance more frequently as a means of boundary regulation, thereby more robustly policing the credibility of their information environment. Conversely, the more confident a user is in the accuracy of the information they see on social media, the less likely they are to view unfriending/unfollowing as a necessary means of boundary regulation. For this reason, it is hypothesized that confidence in the accuracy of political information on Facebook will be inversely related to politically motivated avoidance.

H2. The likelihood of politically motivated avoidance will be negatively related to confidence in the accuracy of political information on Facebook.

Three additional hypotheses are considered based on prior studies of friendship dissolution. These include (a) the size of users’ social networks, (b) the frequency of Facebook usage, and (c) reliance on Facebook as a source of news and information. Prior studies have suggested that the size of one’s social network is likely to predict politically motivated unfriending/unfollowing (i.e., Skoric et al., 2018; Yang et al., 2017). This is premised on the underlying relationship between unfriending/unfollowing and the strength of network ties. SNS-users are more likely to break a weak network tie as opposed to a strong network tie, and as Yang et al. (2017) point out, as a general rule, “. . . larger networks contain weaker ties” (p. 24). As such, the number of Facebook “Friends” that a user has is expected to predict a greater likelihood of politically motivated avoidance (Skoric et al., 2018; Yang et al., 2017).

H3. The likelihood of politically motivated avoidance will be positively related to the size of the user’s social network.

Based on prior research, more frequent usage of Facebook is also believed to predict a greater likelihood of politically motivated avoidance. In their study of Columbian SNS-users, Yang et al. (2017) found a positive relationship between time spent on social media and politically motivated unfriending. They speculated that this relationship may have been due in part to differences in “platform literacy,” wherein users who have a greater familiarity with the platform’s functionality will be more likely to use features like unfriending to manage the boundaries of their social networks. Bode (2016) similarly suggests that the frequency with which Facebook updates its settings may result in casual/occasional users being less familiar with these tools. As such, it is expected that the amount of time spent on Facebook will be positively related to the likelihood of politically motivated avoidance.

H4. The likelihood of politically motivated avoidance will be positively related to the amount of time spent on Facebook.

Finally, it is hypothesized that the extent to which one relies on Facebook as a source of news and information will be related to politically motivated avoidance. Johnson et al. (2017) found that those who relied more heavily on social media as a source of news and information were more likely to engage in selective avoidance while using those platforms. This finding is consistent with demonstrated tendencies toward partisan media preferences among Americans (Iyengar & Hahn, 2009; Knobloch-Westerwick & Meng, 2009; Stroud, 2008), such that those who view social media as a primary source of news and information are more likely to engage in boundary regulation—whether for ideological or informational reasons—than those who view it as a purely social medium.

H5. The likelihood of politically motivated avoidance will be positively related to reliance on Facebook for news and information.

Data and Methods

A representative sample of 600 adults in the state of Florida were surveyed using a leading market research provider.1 The survey was fielded between 10 and 17 October of 2020. Survey respondents were selected using a stratified, quota sampling method, with quotas enforced (by region of the state) for gender, age, race, and ethnicity based on population estimates provided by the University of Florida’s Bureau of Economic and Business Research (BEBR). Of the 600 respondents, 508 reported having an active Facebook account. Only those with active Facebook accounts, and for whom complete responses were received, are included in the final regression model, resulting in a sample size of n = 455 for that model. Table 1 summarizes the sample in relation to the state’s demographics, including a breakdown of the original sample as well as those identified as Facebook users. While the initial sample is well representative of the state’s population, it should be noted that the specific population of Facebook users in the state is unknown.

Table 1.

Sample Respondents Compared to State Demographics (N = 600).

	Sample (%)	Facebook users (%)	State demographics (%)
Gender
Female	51	50	51
Male	49	50	49
Age
18–24	10	9	11
25–54	47	49	47
55–64	18	18	17
65+	25	23	25
Race
Black/African American	17	17	17
White/Caucasian	75	76	77
Other	8	7	6
Ethnicity
Hispanic	27	26	26
Non-Hispanic	73	74	74

Note. Sample n = 600; Facebook users n = 508.

Survey respondents were asked whether they had “unfriended or unfollowed someone because of their political posts” on Facebook in the past 3 months. Because the specific methods for forming and breaking network ties vary across social media platforms, this survey focused specifically on Facebook, as it remains the most widely utilized social media platform among American SNS-users (Perrin & Anderson, 2019). It should be noted that the survey instrument did not distinguish between unfriending and unfollowing. While the two behaviors can be seen as distinct, given the focus of this study on boundary regulation and information exposure, the distinction was deemed unimportant, as both behaviors block exposure to future posts. First, basic descriptive analyses were conducted to measure the frequency of politically motivated avoidance, as well as differences across political parties. Second, a logistic regression model was constructed to test the hypotheses outlined above. The model was estimated as follows

$\ln (\frac{\hat{π_{i}}}{1 - \hat{π_{i}}}) = α + \sum [\begin{array}{l} β (P o l i) + γ (C r e d) \\ + χ (C o n t r o l) + ρ (D e m o) + ε \end{array}]$ (1)

$\hat{π_{i}} = \frac{e x p α + \sum [\begin{array}{l} β (P o l i) + γ (C r e d) \\ + χ (C o n t r o l) + ρ (D e m o) + ε \end{array}]}{1 + e x p α + \sum [\begin{array}{l} β (P o l i) + γ (C r e d) \\ + χ (C o n t r o l) + ρ (D e m o) + ε \end{array}]}$ (2)

where $\hat{π_{i}}$ is the estimated probability that the ith case equals “yes” (i = 1); Poli is a vector of political characteristics; Cred is a measure of user confidence in the accuracy of political information on Facebook; Control is a vector of control variables; and Demo is a vector of demographic variables. The Poli vector includes variables measuring party affiliation and partisan intensity. The Control vector includes variables measuring (a) number of Facebook friends, (b) weekly Facebook usage, and (c) reliance on Facebook for political information. Table 2 summarizes each of these variables for the sample, including descriptive statistics as well as measurement and coding rules.

Table 2.

Variable Coding and Descriptive Statistics.

Variables	Coding	f	%	$\bar{X}$	σ
Facebook Friends	Continuous; range = 0–95,000 (log transformed)	–	–	781.6	5,313.7
Average weekly usage
Low (⩽2 hr)	Reference category	201	39.6	–	–
Moderate (3–8 hr)	1 = yes; 0 = no	206	40.6	–	–
High (⩾9 hr)	1 = yes; 0 = no	101	19.9	–	–
Reliance on Facebook (election)
Not at all	Reference category	214	42.1	–	–
A little	1 = yes; 0 = no	136	26.8	–	–
A lot	1 = yes; 0 = no	101	19.9	–	–
A great deal	1 = yes; 0 = no	57	11.2	–	–
Accuracy of Facebook information
Not at all confident	Reference category	152	29.9	–	–
Not very confident	1 = yes; 0 = no	166	32.7	–	–
Somewhat confident	1 = yes; 0 = no	125	24.6	–	–
Very confident	1 = yes; 0 = no	65	12.8	–	–
Party affiliation
Democrat	Reference category	255	45.8	–	–
Independent	1 = yes; 0 = no	106	19.0	–	–
Republican	1 = yes; 0 = no	196	35.2	–	–
Ideological intensity
Low/None	Reference category	235	39.2	–	–
Moderate	1 = yes; 0 = no	224	37.3	–	–
High	1 = yes; 0 = no	141	23.5	–	–
Political engagement index	continuous; range = 0–8 (log transformed)	–	–	2.03	2.23
Gender	1 = female (0 = male)	307	51.2	–	–
Age	Continuous; range = 18–91	–	–	49.3	17.1
College education	1 = college degree or higher	307	51.2	–	–

Source. 2020 Presidential Election Survey.

Along with their political party affiliation, respondents were also asked to describe their political ideology from among the following options: (a) very liberal, (b) somewhat liberal, (c) moderate, (d) somewhat conservative, and (e) very conservative. The ideological intensity variable was created by recoding very liberal and very conservative as “high intensity,” somewhat liberal and somewhat conservative as “low intensity,” and moderate as “none.”

Respondents were asked to self-report their current number of Facebook “friends.” This variable was log transformed for the purposes of the logistic regression analysis. Average weekly Facebook usage is reported as a categorical variable: “low usage” = 3 hr or less, “moderate usage” = 4–8 hr, and “high usage” = 9 or more hr. To measure reliance on Facebook, respondents were asked How much have you relied on Facebook to stay informed about the Presidential race? Response options included (a) a great deal, (b) a lot, (c) a little, and (d) not at all. To measure perceptions of information credibility, respondents were asked How confident are you in the accuracy of the information you encounter on Facebook? Response options included (a) very confident, (b) somewhat confident, (c) not very confident, and (d) not at all confident.

Demographic variables were also included for gender, age, and college education. For gender, males were omitted as the reference category. Education was recoded as a binary variable, with “less than college degree” omitted as the reference category. Additional demographic variables such as race and ethnicity were also collected, but they are excluded from the analysis due to their multicollinearity with the political attribute variables.

Findings

As noted above, the initial research question asked the following: “How common was politically motivated avoidance in the 3 months prior to the 2020 presidential election?” Table 3 shows that roughly 29% of Facebook users reported unfriending/unfollowing someone “over their political posts” during that time. While this suggests a higher rate of unfriending/unfollowing than has been observed in other politically contentious contexts (i.e., John & Dvir-Gvirsman, 2015; Yang et al., 2017), it is relatively consistent with higher rates of unfriending observed recently during the COVID-19 pandemic (Neely, 2021), as well as with the observation that politically motivated avoidance is more common among American SNS-users (John & Dvir-Gvirsman, 2015).

Table 3.

Politically Motivated Avoidance During the 2020 Presidential Election.

“Unfriended/Unfollowed” someone in the past 3 months over their political posts:	f	%
Yes	147	28.94
No	361	71.06

Source. 2020 Presidential Election Survey.

A second research question asked whether politically motivated avoidance differed based on party affiliation. Table 4 summarizes a chi-square test of politically motivated avoidance across party affiliations (χ² = 4.589, ϕ_c = 0.098; p = .10.). Republicans were slightly more likely to engage in politically motivated avoidance than Democrats (33% vs. 27%, respectively), while Independent SNS-users were notably less likely to have engaged in politically motivated avoidance (21%) during the same time. The results were only statistically significant at the 0.10 level, and the observed difference between the two major parties is relatively small. Given the general lack of consensus observed in the previous literature on this issue, these results do not appear strong enough to warrant a conclusive answer in the affirmative for Research Question 2.

Table 4.

Politically Motivated Avoidance by Party Affiliation.

“Unfriended/Unfollowed” someone in the past 3 months over their political posts:	Democrat (%)	Independent (%)	Republican (%)
Yes	27.4	20.9	33.3
No	72.6	79.1	66.7

Note. χ² = 4.589, φ_c = 0.098; p = .10.

Table 5 presents results from the logistic regression model. The data provide support for several of the hypotheses outlined above, underscoring that politically motivated avoidance is influenced by both a users’ political attributes as well as their perceptions of information credibility on Facebook. Among the most significant predictors of unfriending/unfollowing are (a) reliance on Facebook for news and information, (b) confidence in the accuracy of political information on Facebook, and (c) partisan intensity. The results are discussed below in terms of odds ratios (e^b). These are easier to interpret than traditional β coefficients, as they represent changes in the odds of a “yes” response based on a one-unit increase in the independent variable, ceteris paribus. Odds ratios are multiplicative coefficients; ratios greater than 1 indicate an increase in the odds of a given response, while ratios of less than 1 indicate a decrease in the odds. Odds ratios of less than 1 can be inverted for comparison purposes (i.e., 1/e^b), with the resultant ratio denoting the decreased odds of a “yes” response (demonstrated below).

Table 5.

Logistic Regression for Politically Motivated Avoidance (n = 455).

	e ^b	β	S.E.
Facebook friends (ln)	0.942	−0.060	0.066
Weekly Facebook usage
Low (3 or less hr)—ref. cat	–	–	–
Moderate (4–8 hr)	0.998	−0.002	0.273
High (9 or more hr)	1.117	0.111	0.388
Reliance on Facebook (election info)
Not at all—ref. cat	–	–	–
A little	2.199*	0.788*	0.806
A lot	3.723**	1.315**	1.608
A great deal	7.247**	1.980**	3.855
Confidence in Facebook accuracy
Not at all confident—ref. cat	–	–	–
Not very confident	0.427*	−0.850*	0.151
Somewhat confident	0.329**	−1.111**	0.140
Very confident	0.332*	−1.102*	0.175
Party affiliation
Democrat—ref. cat	–	–	–
Independent	0.814	−0.206	0.276
Republican	1.003	0.003	0.256
Ideological intensity
None	–	–	–
Low	1.822*	0.600*	0.506
High	2.111*	0.747*	0.650
Gender (female)	0.947	−0.055	0.506
Age	0.981*	−0.019*	0.008
College education (yes)	1.704*	0.533*	0.413
Constant	0.623	−0.474	0.428
−2LL	247.538	–	–
Pseudo R²	0.102	–	–

p ⩽ .05; **p ⩽ .01.

Hypothesis 1 posited that the likelihood of politically motivated avoidance will be positively related to partisan intensity. The data supported this hypothesis. Compared to moderates, those with high partisan intensity were 2 times more likely to have unfriended or unfollowed a member of their social network over political posts (e^b = 2.111), ceteris paribus. Even those with low ideological intensity were 1.8 times more likely to have done so (e^b = 1.822). Across the board, those who identified themselves as either liberals or conservatives were significantly more likely to engage in politically motivated unfriending than moderates.

Figure 1 depicts changes in the marginal probability of politically motivated avoidance across varying levels of partisan intensity. With all other variables held at their means/reference categories, the probability that a self-identified moderate engaged in politically motivated avoidance prior to the 2020 election was 0.19. The probability increased to 0.33 for those with high partisan intensity.

Figure 1.

Marginal effects of changes in partisan intensity (and 95% CI) with model at means/reference categories.

Hypothesis 2 posited that the likelihood of politically motivated avoidance will be negatively related to confidence in the accuracy of political information on Facebook. The results provided strong support for this hypothesis, and the measured parameter estimates were considerably large. Those who indicated that they were either “somewhat” or “very confident” in the accuracy of the political information they saw on Facebook were 3 times less likely to have unfriended or unfollowed a member of their social network over political posts (i.e., for “somewhat confident,” 1/e^b = 3.039). Those who reported being “not at all confident” in the accuracy of the political information they saw on Facebook were substantially more likely to engage in politically motivated avoidance.

Figure 2 depicts changes in the marginal probability of politically motivated avoidance based on varying confidence in the accuracy of political information on Facebook. The probability of avoidance increased dramatically as a user’s confidence in the accuracy of the political information on Facebook decreased. The probability of avoidance among those who were “very confident” was 0.19, while the probability among those who were “not at all confident” rose to 0.41.

Figure 2.

Marginal effects of changes in confidence in political information on Facebook (and 95% CI) with model at means/reference categories.

Hypothesis 5 posited that the likelihood of politically motivated avoidance will be positively related to reliance on Facebook for news and information about the election. The results show that the likelihood of politically motivated avoidance increased substantially as reliance on Facebook for political information increased. Those who indicated that they relied on Facebook “a great deal” were over 7 times more likely to have engaged in politically motivated avoidance than those who did not rely on it at all (e^b = 7.247). Even those who only relied on Facebook a little were over 2 times more likely to engage in politically motivated avoidance (e^b = 2.199).2

Figure 3 depicts changes in the marginal probability of politically motivated avoidance based on varying levels of reliance on Facebook for news and information. The probability of politically motivated avoidance increased consistently and significantly as reliance on Facebook increased. Among those who relied on Facebook “a great deal” to stay informed about the election, the probability of unfriending or unfollowing someone over political posts was 0.55. This compared to a probability of only 0.15 among those who did not rely on it at all, ceteris paribus.

Figure 3.

Marginal effects of changes in reliance on Facebook (and 95% CI) with model at means/reference categories.

Neither Hypothesis 3 nor Hypothesis 4 was supported by the data, meaning that there was not a meaningful relationship between politically motivated avoidance and either the number of Facebook “Friends” or the amount of time spent on Facebook. In each case, the measured parameter estimates were notably small, and neither was statistically significant.3 Among the demographic variables, females were slightly less likely than males to engage in politically motivated avoidance, though the difference was not statistically significant. There was a statistically significant decline in the likelihood of avoidance commensurate with age. Most notably, education was positively related to politically motivated avoidance, where those with a college degree or higher were 1.704 times more likely to have answered in the affirmative.

Discussion

This study considered politically motivated avoidance on Facebook in the 3 months leading up to the 2020 US presidential election. As noted in prior studies (i.e., Yang et al., 2017), politically motivated unfriending/unfollowing represents a unique form of selective avoidance, whereby SNS-users eliminate network ties that transmit undesirable content. It has been suggested that this type of boundary regulation might contribute to the formation of partisan echo chambers, as users may limit their exposure to competing viewpoints and corrective information (i.e., Sunstein, 2017). While prior studies have suggested that partisan selective avoidance is uncommon in digital settings (i.e., Garrett, 2009; Garrett & Stroud, 2014; Jang, 2014), the results from this survey of voting age adults showed that politically motivated avoidance was fairly common in the lead up to the election. In total, nearly a third of respondents (29%) reported unfriending or unfollowing someone over political posts during the observed timeframe. However, a deeper examination of the data suggests that this avoidance is most likely a function of varied and diverse factors and may be less reflective of confirmation bias than some have suggested. While variations in ideological intensity were predictive of avoidance, both reliance on Facebook as a source of political information and user perceptions of information credibility on Facebook were more compelling predictors of politically motivated avoidance.

The model examined in this study considered avoidance as a function of both political characteristics and user perceptions of information credibility. The results provide some support for both considerations. Among the observed political characteristics, party affiliation was not found to be a significant predictor of politically motivated avoidance. While Republicans in this sample were slightly more likely to have engaged in avoidance, the size of the observed difference was relatively minor. The broader literature on confirmation bias has been similarly inconclusive with regard to partisan differences. While some theoretical arguments have been offered to hypothesize a greater tendency toward avoidance on the part of conservatives (i.e., Garrett & Stroud, 2014; Nam et al., 2013), the empirical findings have been inconsistent, with many studies suggesting the opposite. Taken as a whole, the evidence—including this current study—suggests that partisan differences in avoidance may be contextual. For instance, in this study, a strong correlation is observed between reliance on social media for political information and the likelihood of politically motivated unfriending/unfollowing. Given the context of the 2020 election, including Donald Trump’s heavy reliance on social media to communicate with his political base, slightly higher rates of avoidance among Republicans may have been a function of their greater reliance on social media during the election. Additional research and analysis are needed to explore these considerations.

While partisan affiliation was not a significant predictor of avoidance, the observed relationship between partisan intensity and politically motivated avoidance was both statistically and practically significant. Those who identified themselves as either “very conservative” or “very liberal” were over 2 times more likely than moderates to have unfriended or unfollowed someone because of their political posts. This is consistent with broader tendencies toward confirmation bias among partisans (Knobloch-Westerwick, 2014; Stroud, 2010), as well as with current socio-political trends in the United States, including the growing tendency of some Americans to “sort” themselves—both socially and geographically—based on partisan differences (Bishop, 2009; Lukianoff & Haidt, 2018; Pew Research Center, 2019). It has been argued that these tendencies pose a threat to the foundational ideals of deliberative democracy, insofar as exposure to political differences is seen as essential to the development of political tolerance (i.e., Sunstein, 2017). To the extent that many Americans increasingly rely on social media as a primary source of news and information, this raises some concerns over the potential implications of these findings.

However, several recent studies have offered a counterargument to this interpretation, suggesting that these types of avoidance among partisans may be of little consequence, as these individuals are more likely to be politically engaged and thus exposed to counter-attitudinal ideas through a variety of mediums (i.e., Bode, 2016; Dubois & Blank, 2018). Bode (2016) 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” (p.6). While intuitively strong, this argument may be at least partially inconsistent with the strong observed correlation between avoidance and reliance on social media for political information (discussed further below). In either case, this observed tendency of partisan ideologues toward selective avoidance may suggest an unwillingness to engage meaningfully with counter-attitudinal viewpoints, regardless of where they are encountered.

While the results of this study do show some evidence of politically motivated confirmation bias on social media, they also suggest that additional factors may more heavily influence the decision to unfriend or unfollow an account over political posts. A strong relationship was observed between politically motivated avoidance and user perceptions of information credibility. Respondents were asked: How confident are you in the accuracy of the political information that you encounter on Facebook? Those who indicated a lack of confidence were substantially more likely to engage in politically motivated avoidance (measured as unfriending/unfollowing). This could indicate that many SNS-users remove others from their social networks due to concerns over the credibility of the political information being shared rather than a desire to avoid competing perspectives. This is consistent with a recent study of selective exposure conducted by Metzger et al. (2020), who argued that “Rather than seeking to reduce dissonance, it may be that individuals seek information from news sources that share their political attitudes and outlook because they perceive them as more credible . . .” (p. 5). Their results suggested that perceptions of information credibility may be a more significant driver of selective exposure than cognitive dissonance.

On one hand, this finding is consistent with the premise that information seeking and learning are among the primary motivations when adopting digital media (i.e., Papacharissi & Rubin, 2000; Sundar & Limperos, 2013), as well as the observed tendency for SNS-users to engage in boundary regulation as a means of maintaining the utility of their social networks (i.e., Strutzman & Hartzog, 2012). Viewed through this lens, politically motivated avoidance may, at least in some instances, represent a form of boundary regulation, whereby SNS-users maintain the quality of information within their social networks. However, there are also partisan considerations that may affect this interpretation, namely that individuals are prone to view ideologically congenial media sources as more credible than counter-attitudinal, or even ideologically neutral sources (Clark & Maass, 1988; Hansen & Kim, 2011; Metzger et al., 2020). In that case, the underlying causal motives may differ, but the net effect would be the same, including any potentially deleterious effects on deliberative democracy. It should be emphasized that this study cannot be taken as directly corroborating that recently conducted by Metzger et al., (2020), as confidence in the accuracy of political information on Facebook is at best an imperfect proxy for source credibility. Further research is needed to more directly measure this relationship and its potential implications.

The results also highlighted a strong relationship between politically motivated avoidance and a user’s reliance on social media for news and information. For instance, those who relied the most heavily on Facebook to stay informed about the election were far more likely to engage in politically motivated avoidance. On the high end, those who relied “a great deal” on Facebook were over 7 times more likely to engage in unfriending or unfollowing than those who did not rely on it at all. Even those who relied on Facebook just “a little” were 2 times more likely to unfriend or unfollow someone over political posts. These substantial effect sizes indicate that reliance on social media for news and information is one of the key determinants of politically motivated avoidance.

This finding can be interpreted in multiple ways, and further research is required to better understand the causes and consequences of this relationship. On one hand, this may reflect some overlap with the issue of partisan intensity discussed above. Prior studies have found a clear connection between partisan intensity and a preference for partisan news outlets (i.e., Iyengar & Hahn, 2009; Knobloch-Westerwick, 2014; Stroud, 2010). Following this logic, those who view Facebook and other social media platforms as a primary source of “news” may be more inclined to construct their social networks in a partisan manner, just as they do with other informational mediums. On the other hand, the relationship between reliance and avoidance could be an extension of the desire for greater information utility, wherein those who regard their social media feed as a source of “serious news” may be more inclined to eliminate personal political opinions/editorial comments from their social network in favor of more reliable news and information sources. Additional research is needed to better understand this relationship. In particular, our collective understanding of this phenomenon would benefit from a more nuanced level of data collection that comprehensively examines the specific sources that SNS-users seek and rely on, as well as the types of actors and accounts with which they break network ties. Without a deeper understanding of this relationship, it is difficult to know if and in what ways this type of unfriending might influence democratic deliberation at a societal level.

Finally, it should be pointed out that the rate of politically motivated unfriending/unfollowing observed in this instance is notably higher than that reported in several prior studies. For example, John and Dvir-Gvirsman studied politically motivated unfriending during the 2014 Israel–Gaza conflict and found an unfriending rate of 16%. Similarly, Zhu et al. (2017) found that 15% of SNS-users had engaged in politically motivated unfriending during the Umbrella Movement protests in Hong Kong. Even during the highly politicized COVID-19 pandemic, only one in five Facebook users reported unfriending (Neely, 2021). This difference may be due to a number of factors, including the highly contentious nature of the 2020 presidential election in the United States. Higher rates of avoidance during the presidential election may also reflect concerns over the credibility of information on platforms such as Facebook, as significant concerns were raised about the role of misinformation on social media during the 2016 presidential election (Ferrara et al., 2016; Persily, 2017). This may lead to greater public skepticism over the reliability of social media and online news outlets.

Conclusion

The findings outlined above suggest that politically motivated avoidance is more likely among those who hold strong ideological opinions. However, the results also suggest that perceptions of information credibility and reliance on social media for news and information may be more powerful predictors of such boundary regulating behaviors. This leaves the waters somewhat muddied with regard to the broader societal effects of politically motivated avoidance. Further research is needed to clarify our understanding of these relationships and mechanisms, particularly the extent to which assessments of informational accuracy are ideologically motivated. As public reliance on social media continues to grow, it will only become more important for political scientists and information scholars to understand these phenomena and their potential effects on deliberative democracy.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research,authorship,and/or publication of this article.

Funding

The author(s) received no financial support for the research,authorship,and/or publication of this article.

ORCID iD

Stephen R. Neely

Author biography

Stephen R. Neely (PhD,North Carolina State University,2013) is an Associate Professor in the School of Public Affairs at the University of South Florida. He is a coordinator of the School’s undergraduate program in Leadership and Public Service . He studies the role of social media in public life,and his broader areas of specialization include survey research,quantitative data analysis,and public policy. He is a faculty researcher on the Sunshine State Survey project and has conducted survey research on behalf of local governments in the Tampa Bay region. He can be contacted by email at srneely@usf.edu

References

Anspach

N. M.

Carlson

T. N.

(2020). What to believe? Social media commentary and belief in misinformation. Political Behavior, 42(3), 697–718. https://doi.org/10.1007/S11109-018-9515-Z

Bishop

(2009). The big sort: Why the clustering of like-minded America is tearing us apart. Mariner Books.

Bode

(2016). Pruning the news feed: Unfriending and unfollowing political content on social media. Research and Politics, 3(3), 1–8. https://doi.org/10.1177/2053168016661873

Clark

R. D.

Maass

(1988). The role of social categorization and perceived source credibility in minority influence. European Journal of Social Psychology, 18(5), 381–394. https://doi.org/10.1002/ejsp.2420180502

Dubois

Blank

(2018). The echo chamber is overstated: The moderating effect of political interest and diverse media. Information, Communication, and Society, 21(5), 729–745. https://doi.org/10.1080/1369118X.2018.1428656

Duggan

Smith

(2016, October 25). Pew Research Center: The political environment on social media. https://www.pewresearch.org/internet/2016/10/25/the-political-environment-on-social-media/

Evans

H. K.

Cordova

Sipole

(2014). Twitter style: An analysis of how house candidates used Twitter in their 2012 campaigns. PS: Political Science and Politics, 47(2), 454–462. https://doi.org/10.1017/S1049096514000389

Ferrara

Varol

Davis

Menczer

Flammini

(2016). The rise of social bots. Communications of the ACM, 59(7), 96–104. https://doi.org/10.1145/2818717

Festinger

(1957). A theory of cognitive dissonance. Stanford University Press.

10.

Fontaine

S. A.

Gomez

D. M.

(2020). Going social: A comparative analysis of presidents’ official and social media messages. Presidential Studies Quarterly, 50(3), 507–538. https://doi.org/10.1111/psq.12676

11.

Garrett

R. K.

(2009). Politically motivated reinforcement seeking: Reframing the selective exposure debate. Journal of Communication, 59(4), 676–699. https://doi.org/10.1111/j.1460-2466.2009.01452.x

12.

Garrett

R. K.

Stroud

N. J.

(2014). Partisan paths to exposure diversity: Differences in pro- and counterattitudinal news consumption. Journal of Communication, 64(4), 680–701. https://doi.org/10.1111/jcom.12105

13.

Gearhart

Moe

Zhang

(2019). Hostile media bias on social media: Testing the effect of user comments on perceptions of news bias and credibility. Human Behavior and Emerging Technology, 2(2), 140–147. https://doi.org/10.1002/hbe2.185

14.

Hahn

K. S.

Ryu

Park

(2015). Fragmentation in the Twitter following of news outlets: The representation of South Korean users’ ideological and generational cleavage. Journalism and Mass Communication Quarterly, 92(1), 56–76. https://doi.org/10.1177/1077699014559499

15.

Hansen

G. J.

Kim

(2011). Is the media biased against me? A meta-analysis of the hostile media effect research. Communication Research Reports, 28(2), 169–179. https://doi.org/10.1080/08824096.2011.565280

16.

Heimlich

(2012, April 3). Many social networking site users surprised by their friends’ political views. Pew Research Center. https://www.pewresearch.org/fact-tank/2012/04/03/many-social-networking-site-users-surprised-by-their-friends-political-views/

17.

Iyengar

Hahn

K. S.

(2009). Red media, blue media: Evidence of ideological selectivity in media use. Journal of Communication, 59(1), 19–39. https://doi.org/10.1111/j.1460-2466.2008.01402.x

18.

Jang

S. M.

(2014). Challenges to selective exposure: Selective seeking and avoidance in a multitasking media environment. Mass Communication and Society, 17(5), 665–688. https://doi.org/10.1080/15205436.2013.835425

19.

John

N. A.

Dvir-Gvirsman

(2015). ‘I don’t like you any more’: Facebook unfriending by Israelis during the Israel-Gaza conflict of 2014. Journal of Communication, 65(6), 953–974. https://doi.org/10.1111/jcom.12188

20.

John

N. A.

Gal

(2018). ‘He’s got his own sea’: Political feedback unfriending in the personal public sphere. International Journal of Communication, 12(2018), 2971–2988.

21.

Johnson

T. J.

Kaye

B. K.

Lee

A. M.

(2017). Blinded by the spite? Path model of political attitudes, selectivity, and social media. Atlantic Journal of Communication, 25(3), 181–196. https://doi.org/10.1080/15456870.2017.1324454

22.

Knobloch-Westerwick

(2014). Choice and preference in media use: Advances in selective exposure theory and research. Routledge.

23.

Knobloch-Westerwick

Meng

(2009). Looking the other way: Selective exposure to attitude-consistent and counterattitudinal political information. Communication Research, 36(3), 426–448. https://doi.org/10.1177/0093650209333030

24.

Kouzy

Jaoude

J. A.

Kraitem

El Alam

M. B.

Karam

Adib

Zarka

Traboulsi

Akl

E. W.

Baddour

(2020, March 13). Coronavirus goes viral: Quantifying the COVID-19 misinformation epidemic on Twitter. Cureus, 12(3), e7255.

25.

Kwak

Chun

Moon

(2011, May). Fragile online relationship: A first look at unfollow dynamics in Twitter [Conference session]. SIGCHI Conference on Human Factors in Computing Systems, Vancouver, BC, Canada. https://doi.org/10.1145/1978942.1979104

26.

Lin

Spence

P. R.

Lachlan

K. A.

(2016). Social media and credibility indicators: The effect of influence cues. Computers in Human Behavior, 63, 264–271. https://doi.org/10.1016/j.chb.2016.05.002

27.

Lee

J. K.

(2019). Stumbling upon the other side: Incidental learning of counter- attitudinal political information on Facebook. New Media and Society, 21(1), 248–265. https://doi.org/10.1177/1461444818793421

28.

Lukianoff

Haidt

(2018). The coddling of the American mind: How good intentions and bad ideas are setting up a generation for failure. Penguin Books.

29.

McPherson

Smith-Lovin

Cook

J. M.

(2001). Birds of a feather: Homophily in social networks. Annual Review of Sociology, 27, 415–444. https://doi.org/10.1146/annurev.soc.27.1.415

30.

Metzger

M. J.

Hartsell

E. H.

Flanagin

A. J.

(2020). Cognitive dissonance or credibility? A comparison of two theoretical explanations for selective exposure to partisan news. Communication Research, 47(1), 3–28. https://doi.org/10.1177/0093650215613136

31.

Mutz

D. C.

Martin

P. S.

(2001). Facilitating communication across lines of political difference: The role of mass media. American Political Science Review, 95(1), 97–114. https://doi.org/10.1017/S0003055401000223

32.

Nam

H. H.

Jost

J. T.

Van Bavel

J. J.

(2013). Not for all the tea in China! Political ideology and the avoidance of dissonance-arousing situations. PLOS ONE, 8(4), 1–8. https://doi.org/10.1371/journal.pone.0059837

33.

Neely

S. R.

(2021, January 25). 2021 COVID-19: Social media, politicization, and misinformation. https://www.usf.edu/arts-sciences/departments/public-affairs/documents/news-items/spa-fb-covid-survey-results-2021.pdf

34.

Neely

S. R.

Scacco

J. M.

Strom

(2020, October 27). 2020 Presidential election survey results, Facebook. https://www.usf.edu/arts-sciences/departments/public-affairs/documents/news-items/spa-fb-election-survey-results-2020.pdf

35.

Neubaum

Cargnino

Winter

Dvir-Gvirsman

(2021). ‘You’re still worth it’: The moral and relational context of politically motivated unfriending decisions in online networks. PLOS ONE, 16(1), Article e0243049. https://doi.org/10.1371/journal.pone.0243049

36.

Papacharissi

Rubin

A. M.

(2000). Predictors of internet use. Journal of Broadcasting and Electronic Media, 44(2), 175–196. https://doi.org/10.1207/s15506878jobem4402_2

37.

Parmelee

J. H.

Roman

(2020). Insta-echoes: Selective exposure and selective avoidance on Instagram. Telematics and Informatics, 52, 1–10. https://doi.org/10.1016/j.tele.2020.101432

38.

Perrin

Anderson

(2019). Share of U.S. adults using social media, including Facebook, is mostly unchanged since 2018. Pew Research Center. https://www.pewresearch.org/fact-tank/2019/04/10/share-of-u-s-adults-using-social-media-including-facebook-is-mostly-unchanged-since-2018/

39.

Persily

(2017). The 2016 U.S. election: Can democracy survive the internet. Journal of Democracy, 28(2), 63–76. https://doi.org/10.1353/jod.2017.0025

40.

Pew Research Center. (2019, October 10). How partisans view each other. https://www.pewresearch.org/politics/2019/10/10/how-partisans-view-each-other/

41.

Rannie

Smith

(2012, September 4). Politics on social networking sites. Pew Research Center. https://www.pewresearch.org/internet/2012/09/04/politics-on-social-networking-sites/

42.

Shearer

(2018, December 10). Social media outpaces print newspapers in the US as a news source. Pew Research Center. https://www.pewresearch.org/fact-tank/2018/12/10/social-media-outpaces-print-newspapers-in-the-u-s-as-a-news-source/

43.

Shearer

Matsa

K. E.

(2018, September 10). News use across social media platforms. Pew Research Center. https://www.pewresearch.org/journalism/2018/09/10/news-use-across-social-media-platforms-2018/

44.

Skoric

M. M.

Zhu

Jih-Hsuan

T. L.

(2018). What predicts selective avoidance on social media? A study of political unfriending in Hong Kong and Taiwan. American Behavioral Scientist, 62(8), 1097–1115. https://doi.org/10.1177/0002764218764251

45.

Song

(2017). Why do people (sometimes) become selective about news? The role of emotions and partisan differences in selective approach and avoidance. Mass Communication and Society, 20(1), 47–67. https://doi.org/10.1080/15205436.2016.1187755

46.

Straus

J. R.

Glassman

M. E.

Shogan

C. T.

Smelcer

S. N.

(2013). Communicating in 140 characters or less: Congressional adoption of Twitter in the 111th congress. PS: Political Science and Politics, 46(1), 60–66. https://doi.org/10.1017/S1049096512001242

47.

Stroud

N. J.

(2008). Media use and political predispositions: Revisiting the concept of selective exposure. Political Behavior, 30(3), 341–366. https://doi.org/10.1007/S11109-007-9050-9

48.

Stroud

N. J.

(2010). Polarization and partisan selective exposure. Journal of Communication, 60(3), 556–576. https://doi.org/10.1111/j.1460-2466.2010.01497.x

49.

Strutzman

Hartzog

(2012, February). Boundary regulation in social media [Conference session]. ACM Conference on Computer Supported Cooperative Work 2012, Seattle, WA, United States. https://doi.org/10.1145/2145204.2145320

50.

Sundar

S. S.

Limperos

A. M.

(2013). Uses and grats 2.0: New gratifications for new media. Journal of Broadcasting and Electronic Media, 57(4), 504–525. https://doi.org/10.1080/08838151.2013.845827

51.

Sunstein

C. R.

(2001). Republic.com. Princeton University Press.

52.

Sunstein

C. R.

(2017). #Republic: Divided democracy in the age of social media. Princeton University Press.

53.

Tsfati

(2020). Personality factors differentiating selective approach, selective avoidance, and the belief in the importance of silencing others: Further evidence for discriminant validity. International Journal of Public Opinion Research, 32(3), 488–509. https://doi.org/10.1093/ijpor/edz031

54.

Wagner

(2017). Selective exposure, information utility, and the decision to watch televised debates. International Journal of Public Opinion Research, 29(4), 533–553. https://doi.org/10.1093/ijpor/edw016

55.

World Health Organization. (2020, September 23). Managing the COVID-19 infodemic: Promoting healthy behaviours and mitigating the harm from misinformation and disinformation. https://www.who.int/news/item/23-09-2020-managing-the-covid-19-infodemic-promoting-healthy-behaviours-and-mitigating-the-harm-from-misinformation-and-disinformation

56.

Yang

J.-H.

Barnidge

Rojas

(2017). The politics of ‘unfriending’: User filtration in response to political disagreement on social media. Computers in Human Behavior, 70(1), 22–29. https://doi.org/10.1016/j.chb.2016.12.079

57.

Zhu

Skoric

Shen

(2017). ‘I shield myself from thee’: Selective avoidance on social media during political protests. Political Communication, 34(1), 112–131. https://doi.org/10.1080/10584609.2016.1222471