Abstract

The belongingness hypothesis suggests that humans have a fundamental need to form and maintain meaningful social bonds. Yet two contradictory impulses seem to guide associative behaviors: the need for inclusion and the tendency for in-group preference. The phenomenon of invited unfriending—posting message on social media petitioning those who differ from the poster on some stance to sever the relationship—exemplifies this tension. Two studies examined the types of messages users post when petitioning disconnection as well as the characteristics and behaviors of posters and recipients. First, a thematic analysis of 515 invited unfriending posts revealed that having different likes or dislikes, being unable or willing to do something the poster deems important, and being too politically conservative were the top three reasons for inviting unfriending. Subsequently, a survey of 445 Facebook users found that nearly 10% had invited unfriending and nearly 75% had received such an invitation. Posters did not differ substantially from non-posters, and both posters and recipients identified themes that were largely similar with those identified in the thematic analysis.

1. Introduction

Evolutionary theories implicate survival and reproduction as two superordinate behavioral motivations ². Because these goals are likely best accomplished cooperatively ³⁰, Homo sapiens evolved into an innately social species with internal mechanisms that prompt the formation of affiliative groups ¹⁰. With this in mind, Baumeister and Leary’s need to belong hypothesis states that humans have a pervasive and fundamental need to form and maintain social bonds that offer meaningful and ongoing contact ⁵. As powerful as it is, however, the need for social inclusion is countered by what is perhaps an equally pervasive in-group preference that prompts sorting into groups based on similarity. These paradoxical motivations perhaps create a dialectic tension between the desire to seek or maintain relationships and the desire to avoid or end relationships with those representing one’s out-group.

One arena in which this tension is operative is in the practice of invited unfriending. Unfriending is the “conscious act by one person to end [a] dyadic relationship” ⁴². Although such dissolution occurs in both mediated and non-mediated contexts, this term is most widely applied to the disaffiliative behavior enacted by users on social networking sites (SNS) ^{9, 42, 43} Active unfriending occurs when the connection is severed by the individual who wants to disaffiliate, such as when one Facebook user intentionally ends his or her Facebook friendship with another ¹⁹. In contrast, we define invited unfriending as a message petitioning those who differ from the poster on some outcome to sever the relationship themselves. The practice of invited unfriending shifts the burden for severing the relationship from the poster to others. The issues prompting such invitations can be as significant as adherence to political or religious ideals (e.g., “If you didn’t vote, then unfriend me now”) or as mundane as musical or food preferences (e.g., “If you don’t like jazz music, then unfriend me now).

Regarding the practice of unfriending in the SNS context, empirical research has focused only on active unfriending, documenting its frequency, the characteristics of the ties most likely to be severed, and the cognitive and emotional responses to being unfriended ^{8, 9, 19, 34, 42}. For instance, Sibona and Walczak identified four common types of online behavior causing users to actively unfriend someone: too frequent or unimportant posts, discussion of polarizing topics (e.g., politics or religion), discussion of inappropriate topics (e.g., racism, sexism), or mundane posts about everyday life (e.g., about one’s children, spouse, eating habits) ⁴³. A dearth of research addresses the phenomenon of invited unfriending, however. The purpose of this two-part study is to fill that gap by identifying the types of messages users post when inviting others to unfriend them and by examining the frequency and rationale for such behavior.

2. Literature Review

Decades of relationship science have focused on the processes of forming and maintaining friendships, both on- and offline ^{1, 13, 15, 21, 25, 33, 44}. Hallinan described the process of friendship initiation as a series of steps involving affiliative desire, expression, and reciprocation ²¹. Duck pointed out, however, that relational initiation and dissolution are not mere inverses of the same process; rather, they differ in two important ways ¹⁴. First, whereas the beginning of a friendship involves clearly identifiable behaviors, this need not be true at the end. Relational ties can be severed either abruptly and dramatically or through a protracted process of simply growing apart ⁴⁴. Second, whereas initiation is bilateral, requiring the consent and involvement of both parties, dissolution can be enacted unilaterally, even without the permission of the one being unfriended ^{6, 19}.

In the context of Baumeister and Leary’s belongingness hypothesis, the role of online connections is unclear ⁵. On one hand, the simplicity of making connections, combined with the low level of commitment required to maintain them, might encourage individuals to continuously expand, rather than prune, their networks, given that these connections can be valuable sources of social capital ^{17, 18, 29, 31, 37}. On the other hand, Gashi and Knautz suggested that unfriending online connections is growing increasingly common ¹⁹. This may attributable to the fact that the vast majority of these affiliations are what Granovetter referred to as weak ties ²⁰. Such connections are characterized by minimal face-to-face contact, emotional intensity, intimacy, commitment, and reciprocity. Indeed, the median number of Facebook friends per user is 342, yet users perceive that fewer than 7% of their Facebook friends are “true friendships” ¹¹. Research has demonstrated that users are especially willing to cut these weaker ties ^{19, 28, 42}.

Building on evolutionary principles, Tajfel and Turner discussed the psychological mechanism of in-group predilections, which motivate people to prioritize relationships with those they deem as similar to themselves ⁴⁶. This principle, known as homophily, simply states that “contact between similar people occurs at a higher rate than among dissimilar people” ³². Cooperative affiliations are typically formed and maintained with those who share values, religion, political ideals, ethnic identity, geographic location, and/or organizational membership ⁴⁸. Social congruency comes with behavioral norms that increase a sense of security by allowing for the prediction of another’s behavior ²². Although differences between individuals are to be expected in a pluralistic society, a lack of similarity may increase uncertainty and present a threat to the long-term viability of the relationship.

In contrast to the insecurity stemming from exposure to difference, selective avoidance—shielding oneself from dissonant views by breaking ties with those who hold them—can bolster one’s sense of self ^{26, 27}. This seems especially strategic for individuals engaged in intergroup conflict. Survey data suggest that just over 15% of university students in Hong Kong unfriended someone during the umbrella uprising (a three-month series of street protests against decisions made by the National People’s Congress) ⁴⁹, whereas 18% of Americans have reported unfriending others over political differences ³⁸. Moreover, in a series of interviews reflecting on participant experiences during the 2014 Israeli-Gaza conflict, John and Gal found that by unfriending someone, users make statements about what they consider unacceptable talk ²³. All participants stated they unfriended someone because they “couldn’t deal” with seeing any more of that person’s posts. Interestingly, participants used metaphors of cleanliness and purification when discussing unfriending. This suggests that the act of unfriending can be conceptualized as a type of boundary maintenance that filters the kind of information allowed in by strategically clearing one’s environment of dissimilar others. Finally, Schwarz and Shani analyzed Facebook interactions during the Israeli-Gaza conflict and noticed this boundary work occurring through both active and invited unfriending: “Our data include dozens of public posts, in which users publicly mark their red lines, stating that they would defriend anyone who crossed them, or ask these populations…to defriend themselves” ⁴¹.

In sum, although the ease of online affiliation offers users access to a significant source of social capital and utility for addressing the need to belong, social network management through unfriending is increasingly common. Walther appealed to scholars to examine the way people terminate relationships online ⁴⁷. With the exception of Schwarz and Shani ⁴¹, however, researchers have focused entirely on active unfriending, with virtually no attention paid to invited unfriending. Invited unfriending is a phenomenon worthy of study, however, insofar as it manages the tension between maintaining existing relationships and prioritizing in-group relationships by shifting that decision to others. To understand this phenomenon better, therefore, we pose the following research questions:

RQ1: What kinds of messages do users post when inviting others to unfriend them?

RQ2: How common is the practice of invited unfriending?

RQ3: What are the characteristics of those who engage in invited unfriending?

RQ4: What is the role of invited unfriending in the management of social connections?

Two studies address these questions. The first study presents a thematic analysis of randomly identified Facebook posts in which posters invited others to unfriend them (designed to address RQ1). The second study analyzes data collected from current Facebook users who have either posted an invited unfriending message or have been the recipient of one (designed to address the remaining RQs). Although invited unfriending occurs in other social networking contexts (such as unfollowing someone on Twitter), we selected Facebook for these initial exploratory studies, given that Facebook is the largest social networking site in the world ⁴⁵ and the practice of invited unfriending on that site has garnered media (even if not yet scholarly) attention ¹².

3. Study One

The two senior authors identified 525 posts from Facebook by searching the phrase “unfriend me now” under Public Posts. A random number list was used to identify posts to be selected and copied. To be selected, a post had to be written in English and include only text or static media (i.e., no video or interactive media). The two senior authors worked independently to identify and copy 525 posts, and then after eliminating duplicates and posts that did not meet the selection criteria, a total of 515 posts remained for analysis. Working independently, the two senior authors then coded each post for the gender of the poster (Cohen’s kappa = .90 for 100 posts that were duplicate-coded for gender), and disagreements or ambiguities as to the gender of the poster were resolved by consulting the poster’s Facebook page.

Each post was then cropped to remove the name and image of the poster as well as the number of comments and “likes,” leaving only the content of the post for analysis. The two senior authors independently reviewed the first 100 posts to generate a draft codebook, based on posters’ reasons for inviting users to unfriend them. The senior authors then worked together to combine their categories into a coding scheme that comprised 13 substantive categories as well as an additional category for posts that were unintelligible, unqualified calls to unfriend, and/or could not be otherwise coded. A description of the categories appears in the results section below.

Three coders, working independently, then coded the first 150 posts to establish reliability. Coders were PhD students in interpersonal communication who had not been involved in the identification of posts or the development of the coding scheme. Coders received approximately three hours of individual and group training, which included discussing the parameters for each category and conducting practice coding. Intercoder reliabilities, based on Cohen’s kappa, appear in Table 1.

Coders first evaluated each post to determine whether it indicated a specific reason for inviting others to unfriend. Posts that did not indicate a specific reason for unfriending were coded as unintelligible if the reason for unfriending was not interpretable and blanket if the message indicated an unqualified call to unfriend, such as “all friends…unfriend me now.” When a specific reason for the unfriending request could be determined, coders categorized the post into one or more of 13 substantive categories or an “other” category. The categories are described below along with quoted examples (spelling and grammatical errors have been retained).

The category don’t want to be bothered included calls to unfriend if others did not want to be annoyed by incessant pictures, posts, or self-promotion. Examples included “Last day in Rome. Gonna make it a good one. If your sick of my pics I feel bad for you. Just wait until I get home! Might want to unfriend me now!” and “Just letting everyone know…I’m gonna be posting about March Madness. So if you don’t want to read about basketball, unfriend me now.” This category included 9.8% of the posts.

The dislike me/disrespect me category asked people to unfriend if they did something to hurt the poster previously or did not communicate with the poster, and it comprised 11% of the posts. For example, “if you don’t like me, unfriend me now!” fell into this category. This category also included Facebook-related “disrespect” of the poster, including failure to comment on or share his or her posts, not adding the poster to a special friend list, or other actions related to Facebook interaction. For instance, “Pls if u dnt reply my msg, like or comment on my post, just unfriend me now…instead of being a Statue nd trash on my frd list” was part of this category.

If the invited unfriending was attributable to having different specific tastes, it fell into the have different likes/dislikes category, which included caring about and liking specific things or failing to like or care about something. This category comprised preferences toward specific athletes, teams, musical acts, and celebrities. Examples from this category include “If you aren’t a diehard Georgia Fan, WIN or LOSE you can unfriend me now. Still proud of my boys” and “If you don’t like Kanye West, I demand you to unfriend me now!!!!!” This category did not include unfriending requests due to different religious or political beliefs, and it contained 15.2% of the posts.

The different views category included invitations to unfriend from those who did not share the poster’s political or policy view on some specific issue, including supporting something he or she does not like for political reasons. This category made up 13.1% of the posts. For a post to fall into this category, the focus was on an issue, specifically, not on supporting or not supporting a group of people. For example, “If you believe the electoral college should stay, unfriend me now” fell into this category.

Too politically right included unfriending invitations because someone voted for or supported a specific Republican/conservative candidate or cause, or if the friend did not vote for or support a specific Democratic/liberal candidate or cause. For example “If you voted for trump, unfriend me now!” fell into this category. Likewise, too politically left contained unfriending invitations because the friend voted for or supported a specific Democratic/liberal candidate or cause or did not vote for or support a specific Republican/conservative candidate or cause. For instance, “If you are a Demoncrat, you can unfriend me now. I don’t want associations with people who hate God and America. Yes I’m serious” fell into this category. These categories comprised 13.5% and 4.0% of the posts, respectively.

The category don’t share religious beliefs included unfriending invitations based on both beliefs (e.g., if you don’t believe in God) and religious practices (e.g., if you don’t read the Bible or observe Lent). Of the posts, 4.6% fell into this category, such as “If you don’t believe the KJB [King James Bible] is God’s word…unfriend me now, and that will save me from doing it once I find out you don’t.”

Posts were included in the category don’t know X piece of information when the poster asked to be unfriended by those who did not know what some specific phrase meant or who a particular person was, such as a picture of “Tom” from Myspace accompanied by the caption: “If you too young to know his name unfriend me now” and “If u don’t know the DIFFERENCE between CUM & COME then pls unfriend me now now.” This category comprised 1.7% of the sample.

Posts were coded as won’t do X when the poster asked people to unfriend him or her if they did not do (or were not willing to do) something that the poster deems important, other than for explicitly religious reasons or explicitly political reasons. For instance, “If you can’t show basic respect for your fellow humans then do us a favor #UnFriendMeNOW” was coded in this category, which included 12.3% of the sample.

The category are a bad person included posts in which the unfriending invitation was based on some personal characteristic, such as being a “faker,” a “whiner,” a “hater,” or a “snowflake.” The objection had to be to an aspect of the person’s identity, rather than something specific he or she did. For instance, “FAKE PEOPLE UNFRIEND ME NOW” and “UNFRIEND ME NOW IF YOU ARE ‘PETTY’ PLEASE” were both coded in this category, which comprised 6.9% of the posts.

Posts were coded as don’t share humor when unfriending requests were based on thinking a picture or post is funny or not funny, and these made up 4.6% of the posts. Examples of this category include “If you don’t laugh when you read these, unfriend me now! (BTW…some a bit raunchy…but it is auto corrects fault” and “Honestly, if any of you think this shit is funny unfriend me now!”

The have X physical characteristic category included posts in which the unfriending reason was due to the friend’s ethnicity, height, hair color or style, weight, or other physical characteristics. This category captured posts such as “If you 16 & under, #UnFriendMeNow.” These made up 1.2% of the posts. To be included in this category, the characteristic named did not have to be visible.

The category don’t use X product included instances when people were asked to defriend if they do or do not own or use a specific brand of clothing, car, beauty product, or other type of product. This category included posts such as “If you don’t have Supreme or Adidas Breast, Unfriend me now.” and “If you’re a bloke and own a convertible Holden Astra please unfriend me now” and made up 1.7% of the sample.

Finally, the other category was used only if a posting seems not to be described by any of the other categories, and it comprised 0.6% of the posts.

Of the 515 “unfriend me now” Facebook posts categorized, the most common reasons for invited unfriending were have different likes/dislikes (15.2%), too politically right (13.5%), have different views (13.1%), won’t do X (12.3%), and dislike me/disrespect me (11.0%). These results suggest invited unfriending is strongly tied to a tendency for in-group preference. This is accomplished through soliciting disaffiliation from Facebook friends who do not share an interest in the poster’s self (dislike/disrespect me), poster’s tastes (have different likes/dislikes), or poster’s values (won’t do X, too politically right). Although still related to in-group preference, the least common messages were those framed around others more than their relation to the self: have X physical characteristic (1.2%), don’t use X product (1.7%), and don’t know X piece of information (1.7%). Interestingly, despite a significant number of posts regarding too politically right (13.5%) and different views (13.1%), the category too politically left (4.0%) represented one of the least common reasons for posters to invite unfriending.

In response to RQ1, this initial study identified the kinds of messages Facebook users post when inviting others to unfriend them. The most common themes do reflect an in-group preference; nonetheless, important questions remain. First, does the practice of invited unfriending affect the average Facebook user? Second, what are the characteristics of Facebook users who invite unfriending and those who are invited to unfriend? Finally, how does invited unfriending factor into the management of social connections? Missing from the posts analyzed in the first study are reactions from those invited to unfriend. When individuals receive a request to “unfriend me now,” how do they interpret the message, and what actions, if any, do they take?

Study Two examined these issues by addressing how common the practice of invited unfriending is (RQ2), what characteristics are common to those who engage in invited unfriending (RQ3), and how invited unfriending is implicated in the management of social connections (RQ4). With respect to RQ3, we were interested specifically in how those who had and had not invited unfriending (as well as those who had and had not been invited to unfriend) would differ in terms of their personalities and inclinations. According to Study One, invitations to unfriend are commonly based on perceived differences in values or opinions (along the lines of “if you disagree with me, unfriend me now”), so we reasoned that posters of such invitations are less likely than non-posters to endorse the principle of inclusiveness (i.e., the idea that one should accept others for who they are, even if they disagree) and the principle of free speech (i.e., the idea that people have a right to express whatever opinions they choose; H1). Because inviting others to unfriend oneself can reasonably be expected to result in some of those others actually unfriending oneself, we also predicted that posters of such invitations are more likely than non-posters to have either a secure or dismissive attachment style, on the argument that people with such styles would be less distressed than those with other attachment styles at the prospect of losing a friend (H2). As described below, we also explored whether posters and non-posters differed in their personality dispositions, their other attachment styles, and their level of loneliness.

4. Study Two

Participants (N = 445) were 226 men, 215 women, 2 transgender adults, and 2 adults declining to indicate their sex, who ranged in age from 21 to 74 years (M = 38.39 years, SD = 10.25). Most (80.2%) identified as white/Caucasian, whereas 8.8% identified as Asian/Pacific Islander, 8.8% as black/African American, 4.5% as Hispanic, 0.7% as Native American or Aleut, 0.4% as Latino/a, and 0.7% as having other racial or ethnic backgrounds (these percentages sum to >100 because some participants identified more than one racial or ethnic category). At the time of the study, 22.6% of participants had a high school diploma or less, 7.4% had a vocational or trade school diploma, 15.1% had an associate’s degree, 42.9% had a bachelor’s degree, 9.7% had a master’s degree, and 2.3% had completed a professional or academic doctoral degree. Nearly half (47.4%) were single and never married, whereas 44.7% were married, 6.3% were divorced, and 1.6% were widowed. Participants represented 48 U.S. states and the District of Columbia. Participants reported having between 0 and 5,000 current Facebook friends (M = 348.18, SD = 584.52, median = 205, mode = 150).

Participants were recruited via the Amazon.com crowdsourcing marketplace Mechanical Turk (MTurk). To be eligible for the study, participants had to be at least 18 years old, be a current user of Facebook, be located in a U.S. state or territory, be able to read and write English, and be a “master worker” (a designation indicating consistently high quality in submitted work) who had completed at least 100 previous jobs with an average approval rate equaling or exceeding 95%. Eligible participants completed and submitted an online questionnaire in exchange for $1.50US. Research has found that samples recruited on MTurk for academic research are typically more representative of the U.S. population than are in-person convenience samples ^{7, 35}.

Loneliness was assessed by the 20-item UCLA Loneliness Scale (Version 3) ⁴⁰. Extraversion, agreeableness, conscientiousness, neuroticism, and openness were measured with the Big Five Inventory ²⁴. Endorsement of free speech was measured with a two-item Likert-type scale developed for this study. (Items were “I believe strongly in the freedom of speech” and “The principle of free speech is important to me.”) Endorsement of inclusiveness was measured with a three-item Likert-type scale developed for this study. (Items were “I believe it is important to accept others for who they are, even if I disagree with them,” “I don’t mind being around people who think or believe differently than I do,” and “I prefer to surround myself with people who think or believe the same way I do” [reverse-coded].) Secure, avoidant, preoccupied, and dismissive attachment were measured by asking participants to read each of the four attachment style descriptions developed by Bartholomew and Horowitz ⁴ and to indicate the extent to which they identified with each style. All measures were assessed on a nine-point scale (1 = strongly disagree, 9 = strongly agree). Each participant saw and responded to the items for each measure in a randomized order. Reliability estimates, means, standard deviations, and intercorrelations appear in Table 2.

Preliminary Analyses. Prior to testing the hypotheses, the integrity of the data was carefully examined. Every MTurk worker has a unique respondent ID number, so to ensure that no worker performed the hit more than once, the frequencies for respondent ID number were examined and no numbers were duplicated, indicating that each respondent was unique. Time to completion (which averaged 19 minutes, 49 seconds) was also examined, and three questionnaires whose time to completion was more than two standard deviations below the mean were eliminated. In addition, responses to an attention check embedded in the questionnaire were examined, and any participants failing the attention check were deleted, resulting in the current sample size of 445.

Who Invites Unfriending? When asked whether they had ever posted an invited unfriending message, 43 participants (9.7%) responded that they had, and the remainder 402 (90.3%) responded that they had not. Women and men were equally likely to have reported posting an invited unfriending message, ² (3) = 3.87, p = .28. Posters were younger, on average (M = 34.93 years, SD = 6.92) than non-posters (M = 38.76 years, SD = 10.49), Welch’s t (64.85) = -3.25, p = .002, Cohen’s d = .43. When asked to indicate their current political orientation on a 9-point scale (1 = Extremely conservative, 9 = Extremely liberal), there was no significant difference between posters (M = 6.14, SD = 2.20) and non-posters (M = 5.63, SD = 2.33), Welch’s t (52.60) = 1.45, p = .15. No significant differences emerged as a function of race/ethnicity, marital status, or education level.

We hypothesized that posters are less likely than non-posters to endorse the principles of inclusiveness and free speech. Consistent with the prediction, posters were less likely (M = 5.92, SD = 1.70) to endorse the principle of inclusiveness than were non-posters (M = 6.86, SD = 1.51), Welch’s t (49.33) = -3.49, p < .001, d = .58. Posters were also less likely to endorse the principle of free speech (M = 7.83, SD = 1.20) than were non-posters (M = 8.06, SD = 1.25), although this difference was nonsignificant, Welch’s t (52.32) = -1.20, p = .12, d = .18.

We also hypothesized that posters have more secure and more dismissive attachment styles than non-posters, and we asked whether fearful avoidant and preoccupied styles also differ. Consistent with the prediction, posters scored higher on secure attachment (M = 6.32, SD = 2.25) than did non-posters (M = 5.21, SD = 2.56), Welch’s t (54.30) = 3.04, p = .002, d = .46. Contrary to the hypothesis, posters scored slightly lower on dismissive attachment (M = 4.91, SD = 2.49) than did non-posters (M = 5.36, SD = 2.42), although the difference was nonsignificant, Welch’s t (50.92) = -1.13, p = .13, d = .18.

Posters also scored lower on fearful avoidant attachment (M = 3.05, SD = 2.06) than did non-posters (M = 3.97, SD = 2.49), Welch’s t (54.40) = -2.70, p = .009, d = .40. Preoccupied attachment was equal among posters (M = 3.09, SD = 2.08) and non-posters (M = 2.92, SD = 2.06), Welch’s t (51.32) = .50, p = .62, d = .08.

Regarding individual differences, there were no differences between posters and non-posters with respect to the five personality dimensions (extraversion, agreeableness, conscientiousness, neuroticism, and openness) or loneliness.

Who Has Been Invited to Unfriend? Among the 402 participants who had never posted an invited unfriending message themselves, 293 (72.9%) had received an invited unfriending post from a Facebook friend, whereas the remaining 109 (27.1%) had not. Women and men were equally likely to have reported receiving an invited unfriending message, ² (3) = 3.44, p = .33. Recipients were younger (M = 37.98 years, SD = 9.64) than non-recipients (M = 40.83, SD = 12.29), Welch’s t (160.11) = -2.18, p = .03, d = .25, and were more politically liberal (M = 5.79, SD = 2.29) than non-recipients (M = 5.17, SD = 2.38), Welch’s t (186.66) = 2.34, p = .02, d = .27.

Regarding personality traits, recipients reported higher openness (M = 6.38, SD = 1.55) than did non-recipients (M = 5.95, SD = 1.57), Welch’s t (192.04) = 2.41, p = .02, d = .28; no other traits differed between the two groups. Recipients and non-recipients did not differ in their loneliness or their endorsement of inclusiveness or free speech.

Posters’ Experiences of Invited Unfriending. On 9-point scales (1 = Not at all, 9 = Very), posters were asked how closely the topic of their post was related to their personal values and how serious they were that they actually wanted people who felt differently about that topic to unfriend them. Posters reported that their topics were moderately related to their personal values, M = 5.15, SD = 2.74, and that they were moderately serious about others unfriending them, M = 4.52, SD = 2.93. These variables were positively correlated, r (44) = .58, p < .001, indicating that the more relevant posters’ topics were to their personal values, the more serious posters were about others unfriending them if they disagreed.

Posters were asked to recall their most recent invited unfriending request and to indicate the reasons behind their request, using the typology developed in Study One. The most commonly cited call to unfriend was for friends who did not want to be bothered by the poster’s current or upcoming posts. Friends with different viewpoints, and friends with different likes and dislikes, tied for the second most common call, and the third most common call was for friends who disliked or had disrespected the poster. Full results appear in Table 3.

Posters reported that an average of 5.22 Facebook friends replied to their post with message of support for what they had said (SD = 11.22), whereas an average of 2.11 friends had replied with unsupportive messages (SD = 4.68). Posters reported that, on average, 1.35 Facebook friends told them that they had unfriend them as a result of their post (SD = 1.12), and that they later discovered on their own that another 1.76 friends, on average, had unfriended them in the wake of their post (SD = 1.70).

Recipients’ Experiences of Invited Unfriending. Recipients were asked to recall the most recent invited unfriending post they had seen from one of their own Facebook friends, and to indicate (using the same typology as posters) the reasons behind the invited unfriending request. The most commonly cited call was for friends whose viewpoints were different from those of posters. The second most common call was for friends who did not want to be bothered by the poster’s current or future postings, and the third was for friends who were too politically conservative. Full results appear in Table 3.

When asked whether they actually unfriended the person, 32 (8.0%) said yes and 261 (64.9%) said no, whereas 109 (27.1%) declined to answer. Those who did unfriend were significantly less likely to agree with the position described in the post (M = 2.56, SD = 2.26) than were those who did not unfriend (M = 4.03, SD = 2.50), Welch’s t (40.90) = -3.43, p = .001, d = .62. Among those who did not unfriend, when asked how seriously they considered unfriending, their average response on a 9-point scale (1 = Not at all seriously, 9 = Very seriously) was well below the theoretic midpoint at 2.58 (SD = 2.46).

To answer the questions that emerged after conducting Study One, we recruited current adult Facebook users to complete an online questionnaire for Study Two. Knowing the variety of invited unfriending messages that emerged from RQ1, RQ2 sought to capture the prevalence of invited unfriending on Facebook. Of the 445 participants who completed the questionnaire, 43 (9.7%) had posted an invited unfriending message, whereas the remaining 402 (90.3%) had not.

RQ3 questioned how these posters and non-posters differed from each other. We hypothesized and found that posters are less likely than non-posters to endorse the principle of inclusiveness, but contrary to our hypothesis, posters did not differ from non-posters in their endorsement of free speech. As predicted, posters had a significantly higher secure attachment style than non-posters, but the groups did not differ in their dismissive style, as we had anticipated.

RQ4 sought to understand the role invited unfriending plays in these online relationships. Of the 402 participants who had not posted an invited unfriending message, 293 (72.9%) had been the recipients of an invited unfriending post. The most prevalent reasons why they were invited to unfriend was if they had different viewpoints (46.5%), if they did not want to be bothered from future posts (28.9%), and if they were too politically conservativeness (24.1%; see Table 3). In terms of the posters, there was a significant positive correlation between the invited unfriending posts being related to the posters’ personal views and their self-reported seriousness of actually wanting to be unfriended, suggesting that the more pertinent the topic was to the poster, the more serious they were about being unfriended. We now turn to a general discussion of the two studies.

5. General Discussion

Humans appear to have somewhat contradictory motives for relationship management: the motive to maintain friendships to fulfill one’s inclusion needs and the motive to eschew friendships that are not considered part of one’s in-group. In social media platforms such as Facebook, the practice of invited unfriending may represent a strategy for balancing these opposing motives, insofar as it 1) connotes the poster’s willingness to allow a Facebook friendship to end if there is too great a perceived difference between poster and friend, but 2) maintains the friendship pending recipients’ decisions. Little is known, however, about who chooses to issue an unfriending invitation (as opposed simply to unfriending others proactively) and why, which prompted the two exploratory studies reported herein.

Regarding overt reasons for inviting unfriending, the most common reason invoked in our thematic analysis was a difference in likes or dislikes. This reason manifested primarily in terms of fandom, wherein the poster indicated that those who either were or were not a fan of a particular sports team, musician, or celebrity were invited to unfriend. The second-most common reason invoked in unfriending invitations was if the recipient was too politically conservative (e.g., “if you voted for Trump, unfriend me now”), and the third-most common reason was if the recipient had different view or opinions.

To some extent, these most common reasons for inviting unfriending all invoke either acceptance or rejection of the poster him/herself. Certainly, both fandom ³⁶ and political ideology ³⁹ can be considered integral components of one’s personal identity, and we would claim the same for one’s behavior patterns (especially those related to civic engagement or equality). Consequently, this type of invitation perhaps connotes the idea that “if you don’t like who I like, do what I do, or support the candidates I support, then you don’t like me.” Friends whose behaviors or ideologies make them appear to be threats to the self may be perceived as liabilities, and may therefore be considered expendable, prompting the invitation to unfriend.

Indexing existing invitations to unfriend provided us a typology of the reasons motivating that behavior, but we also sought to explore the experiences of Facebook users who had actually issued or received such an invitation. We reasoned that because posters were willing to invite others to unfriend them (and especially over differences in values or ideologies, as Study One indicated was common), posters would be less likely than non-posters to endorse the principles of inclusiveness and free speech, and would also score higher on measures of both secure and dismissive attachment styles. We were also interested to know how posters and non-posters would differ in their demographic and personality traits.

Surprisingly few differences emerged between posters and non-posters. The former group was younger, on average; was less likely to endorse the principle of inclusiveness; and had more secure and less fearful avoidant attachment styles. No other demographic or personality characteristics differentiated the two groups, however, suggesting that those who do invite unfriending on Facebook are not substantially different than those who do not. When asked what motivated their most recent unfriending invitation, posters were most likely to cite the desire not to bother their friends with their posts (e.g., “If you don’t want to see incessant photos of my new kitten, unfriend me now”). This most common reason deviates from the most commonly identified reasons in the thematic analysis, insofar as it does not overtly invoke a difference between the poster and friend and could even be considered a prosocial act on the part of the poster. Nonetheless, the second- and third-most commonly cited reasons for inviting unfriending decidedly invoked differences that could be seen as threats: having different viewpoints and different likes/dislikes (tie for second) and having disliked or disrespected the poster him/herself.

When asked about their most recent unfriending invitation, posters reported that the more strongly their reason for inviting unfriending was related to their personal values, the more serious they were about others unfriending them. Posters reported that only three Facebook friends, on average, unfriended them as a result of their most recent unfriending invitation.

Like posters, those who had received an unfriending invitation (but had not posted one themselves) were highly similar to those who had not received such an invitation. On average, receivers were younger, more liberal, and had a more open personality, but they did not differ from non-receivers on any other demographic or personality characteristics. When asked about their most recent unfriending invitation, receivers cited different viewpoints, not wanting to bother friends, and being too politically conservative as the three most common reasons invoked in the invitation, respectively. As a result of their most recent unfriending invitation, 8% of recipients actually did unfriend the poster.

6. Conclusions and Implications

Considered collectively, these results warrant a few tentative conclusions about the practice of invited unfriending. First, it is relatively common. Although approximately one in ten Facebook users in our (admittedly non-random) sample reported having issued an unfriending invitation, nearly three in four users had received one. Because this is the first study (to our knowledge) to identify these frequencies, it is impossible to know whether the behavior is becoming more common over time, but at this point in time, we can conclude that it is not infrequent.

Second, across three samples of posts (Study One, posters in Study Two, and receivers in Study Two), the reasons for inviting unfriending predominantly invoked differences between the poster and his or her friends, often relating to identity-relevant issues such as preferences and political values. This is noteworthy to the extent that the in-group motivation prompts individuals to favor the company of like others and to be wary of those with different values, beliefs, and behavioral patterns, even at the expense of maintaining friendships for the purpose of meeting inclusion needs ¹⁶.

Third, we would surmise (although it is impossible to determine without comparative data) that the frequency of calls to unfriend for being too politically conservative is somewhat artifactual, given the political climate in the United States at the time these data were collected. To the extent that the current administration is politically polarizing, it is perhaps unsurprising to see calls to unfriend that invoke support for that administration or even for its political party be among the top three most common calls in both Study One and for recipients in Study Two. Notably, posters (in Study Two) did not differ from non-posters in their political orientation, so the prevalence of anti-conservative unfriending invitations cannot be attributed to an overly liberal or progressive population of posters. As these frequencies are replicated over time, we will be able to tell whether the anti-conservative sentiment reflected in these data is stable or transient.

A final tentative conclusion is that although invitations for unfriending are relatively common, actual unfriending as a result of those invitations is less so. Whereas approximately 10% of users in Study Two had posted an unfriending invitation, they reported losing an average of only three friends as a result of their most recent post. And whereas nearly 75% of users in Study Two had received an unfriending invitation (an estimate that is likely low because we purposely excluded those who had posted such an invitation themselves), only 8% reported actually unfriending the poster.

This observation raises the possibility that the motivation behind posting an unfriending invitation is related more to image management than to relationship management. In other words, posters may be more interested in simply taking a stand on some issue and implying that their stand is so important that they are willing to risk losing friends because of it than in actually culling friends who disagree. Consistent with that claim is the finding that posters in Study Two were only moderately serious about friends unfriending them after their most recent invitation. If this explanation is true, then invited unfriending may represent an act of virtue signaling, wherein individuals signal some socially valued virtue (e.g., honesty, tolerance, altruism) for which they incur (or are willing to incur) some type of cost, such as the cost of potentially losing friends ³. For instance, a poster may invite unfriending from those who “support the deportation of dreamers” or “don’t treat minorities with respect,” but such a behavior may be more about signaling to others the poster’s own virtue as a compassionate person dedicated to inclusion and diversity than about actually ending connections with friends who think differently. By displaying these virtues in a semi-public manner (as a Facebook post), and by implying that the poster is willing to incur the cost of losing friendships as a result of those virtues, the poster gains credibility, standing, and social capital even without doing anything virtuous.

7. Strengths and Limitations

The two studies reported herein had both strengths and limitations. Study One benefited from a sizeable and randomly selected sample of Facebook posts for analysis, and although the sample of Facebook users in Study Two was not random, it was more demographically and geographically diverse than most samples drawn from student populations. Nonetheless, the latter sample was limited to master users of Mechanical Turk who were located in the United States, so it cannot be considered representative of Facebook users writ large. Both studies were largely exploratory, so conclusions must be considered tentative until replicated.

We do see opportunities for both replication and extension, however. As the frequencies of both invited unfriending tendencies and the reasons for inviting unfriending are replicated over time, we will be able to determine both how accurate those frequencies are and how much they are temporally bound. Extensions to the study could include examining the phenomenon in social media platforms other than Facebook, as well as experimentally manipulating the content of unfriending invitations to examine how they are interpreted.

References