Facebook Fallout: Future Contact Avoidance After Being Unfriended on Facebook
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
2013 46th Hawaii International Conference on System Sciences
Facebook Fallout: Future Contact Avoidance After Being Unfriended on
Facebook
Christopher Sibona
University of Colorado Denver, The Business School
christopher.sibona@ucdenver.edu
Abstract and emotional significance, but there is little research to
confirm this view.
Objectives: Determine the factors that predict
The research examines potential consequences to
whether a person who is unfriended on Facebook will
relationship dissolution on Facebook. Facebook behav-
avoid future contact with the person who unfriended
ior is naturally confined within the computer-mediated
them.
setting; however, the consequences of actions on the site
Results: The research results show that six factors
extend beyond the confines of Facebook. The dissolution
can predict whether a person will avoid future contact
of the relationship may have real world consequences
with the individual who unfriended them: whether the
like future contact avoidance of the person who has
person discussed the event after it occurred, the emo-
unfriended a member of the network. This research
tional response to the unfriending, whether the person
may help identify where computer-mediated and face-to-
who was unfriended believed it was related to their
face relationships share similarities. This research may
offline behavior, the perceived geographical distance be-
help develop future lifecycle models of online behaviors
tween dyad, whether difficulties were discussed between
from the initial friend request to maintenance of the
the dyad prior to the event and how the person valued
relationship and potential dissolution.
the peak-tie strength of the relationship
1. Introduction 1.1. Friendships Formation and Dissolution
Friendships are formed and maintained because they are
Social network sites (SNS) are where Americans spend
rewarding to individuals [29]. Friendships tend to be
the largest share of their time online; Americans spend
formed by people who share certain similarities (such
approximately 25% of their time online on SNS and
as values) [14, 18]. People tend to create friendships
blogs [19]. Social network users form these connections
with those who share a similar race and ethnicity fol-
under a variety of contexts, ranging from maintaining ex-
lowed by age, religion, education, occupation and gender
isting relationships, forming new romantic connections,
and roughly in that order [18]. The largest portion of
and creating new online friendships [26], these relation-
friendships that are formed with those who are not
ships are fluid and the relationship may be dissolved
family members are through organizational structures
as easily as they are formed. Unfriending has become
[18]. Schools, work, and geographic location are major
a widely-used feature of social networking sites; Pew
factors in how relationships are formed.
Internet found that 63% of users unfriended at least one
Friendship formation in the real world has more
member of their online social network in 2011 up from
nuance than in the online world. The initiator of the
56% in 2009 [17].
friend request may communicate the desire to be friends
Facebook’s mission is to “make the world more
with varying degrees of directness [10]. Those who
open and connected”1 but these connections are not
initiate the friendship in less direct ways can avoid
immutable. The word unfriend was named the word of
embarrassment and rejection should the request not be
the year by the New Oxford American Dictionary for
accepted. Computer-mediated settings (e.g. Facebook)
2009 [8]. The dictionary defined unfriend as follows:
lack this nuance and make it very clear that one per-
“unfriend – verb – To remove someone as a ‘friend’ on
son requests the other’s friendship through the visible
a social networking site such as Facebook”2 . The word
friend request. Variations in how directly or indirectly
suggests that the dissolution of the relationship has social
friendship requests are made offline [10] are somewhat
1 http://newsroom.fb.com/content/default.aspx?NewsAreaId=22 eliminated on Facebook, where a request is sent or
2 http://blog.oup.com/2009/11/unfriend not sent. The online requests are either accepted or
1530-1605/12 $26.00 © 2012 IEEE 3270
3272
DOI 10.1109/HICSS.2013.229“ignored.” boyd [1] notes that some social network site Grave-dressing allows a member to work through a post-
(SNS) users accept friend request because it is easier to mortem of the relationship to avoid similar issues in the
say “yes” than “no.” future.
Facebook users undergo a series of steps to unfriend. Sibona and Walczak [22] found four common on-
There is no way to mass unfriend; a user must go to line reasons and two common offline reasons for un-
each individual’s page, scroll down and click the link friending on Facebook. The four online reasons were
“unfriend” to unfriend. This link is placed in a relatively frequent/unimportant posts, polarizing posts (politics and
obscure location compared to the prominently placed religion), inappropriate posts (sexist, racist remarks, etc.)
“Add as Friend” button that is displayed prominently and everyday life posts (child, spouse, eating habits, etc.)
next to a person’s name. The process to add a friend and in that order of frequency. The two offline reasons
requires that the invited person grant permission for the were disliked behavior and changes in the relationship.
union; unfriending is unilateral and no permission is The research also showed that 55% of people unfriended
needed to unfriend. In most cases the person who was someone for their online posting behavior, 28% for their
unfriended does not receive notification that they have offline behavior and 17% unsure.
been unfriended. Many people only notice they have Relationship dissolution in computer-mediated set-
been unfriended when they go to view a member of tings may resemble dissolution in non-mediated set-
their network and that person is no longer displayed as tings although little is known about either phenomenon
a friend, or when they compare their friendship list to [20, 18]. Hatfield et al’s [11] noted that reactions to
recent memory to find additions and absences. dissolution include a range of emotions including: shock,
Research evidence is still emerging to clarify how disbelief, denial, anger and bitterness, guilt, sadness
Facebook friendships do and do not resemble offline and depression, loneliness, desperation, indifference, re-
friendships in their formation, characteristics, and dis- lief, euphoria. The reactions are moderated by gen-
solution [20]. After users join Facebook, they are asked der, culture, self-esteem and overall approach to the
to identify others in the network with whom they have relationship. Those who experience relationship disso-
an existing relationship. Thus Facebook users tend to lution often contemplate the causes of the dissolution
maintain existing social ties with people they know through attributional-search or account making [27]. The
rather than seek out new friends on the social network members tend to have different narratives regarding the
[7]. However, although these relationships may vary cause of the dissolution regarding the same event; the
from weak- to strong-ties, there is some consensus that stories members create tend to be simple, linear and
the majority of ties on Facebook are weak [7, 16]. This more logical than the actual events [4]. This process
may be because of the ease at which friendship can be of account making often involves both discussing the
maintained through low levels of commitment [16], or dissolution with the partner, and "going public" about
because strong-ties are more likely to be maintained with it with others in the immediate network. Duck [4] and
face time, thus obviating a need for social media. Weber & Harvey [27] argue that this is the key to gaining
Relationship dissolution in the real world has been a sense of finality and completion of the event, and to
defined by Duck as the permanent dismembership of the restoration of positive affect.
an existing relationship [6]. Duck [6] developed a four Relationship dissolution is less researched than
phase process model for relationship dissolution. The friendship formation partly because longitudinal datasets
intrapsychic phase is the first phase of dissolution where are required to compare the ties at different times
a person reflects on the relationship and evaluates its [18]. Friendship dissolution research in non-computer-
costs and benefits; the person tends to discuss the re- mediated settings is largely based on close relationships
lationship with third parties who not directly familiar including close friends, romantic partners and divorce
with the relationship. The dyadic phase is where the [18], emotional reactions to unfriending on Facebook
dyad discusses the difficulties in the relationship directly may differ simply due to the greater preponderance of
in order to resolve any issues. The relationship may weak-ties [7]. McPherson [18] found that network struc-
still be repaired at this stage. The social phase begins ture has a role in tie dissolution; members who are less
when the dyad begins to discuss their dissatisfaction with central to the group (those are have fewer connections to
the relationship with others in their social network (e.g. core members) are more likely to leave the group. Some
mutual friends). During the social phase members may real-world friendships end in conflict but most simply
provide support, mediation, and advice and may help fade away [24]. Facebook users might only hide the posts
maintain the relationship or encourage its dissolution. of weak ties, rather than choose to unfriend them, and
The final phase, grave-dressing, occurs when the dyad instead reserve that action when one is making a point
broadcasts to others that the relationship has dissolved. about the end of formerly meaningful relationships, not
3273
3271unlike the process of going public proposed by Duck person who unfriended the survey respondent, their per-
[5]. Quercia et al. [20] examined how online unfriend- ception of whether it was for online or offline behavior,
ing between Facebook dyads may differ from offline questions about the friendship and questions about their
unfriending and found few differences. The research offline behavior. Part two adds additional questions to
found that important factors that predicted friendship part one to determine how the survey respondent was
dissolution were whether the dyad was embedded in affected by the unfriending. Part three asks questions
the same social circle, the age difference between the about how many friends the survey respondent has, how
dyad, and whether one of the two members were neurotic many people they have unfriended, how many people
or introverted. Relationships that had a common female they regularly interact with, and questions about their
friend were more stable than those with common male online posting behavior. Part three also asks questions
friends. about satisfaction, perceived usefulness and perceived
Williams et al. [28] examined the emotional effects ease of use of Facebook. Part four asks demographic
of ostracism in computer-mediated environments (cy- questions: age, gender, education, the number of years
berostracism) through a series of experiments. Williams of social network use and whether the person lives in
et al.’s model of ostracism has two factors; (1) the the United States of America. The analysis of this study
needs-threatened which includes a sense of belonging, concentrates on part two of the survey.
control, meaningful existence and self-esteem and (2)
the aversive impact index which includes lowered mood,
intensity of ostracism and lowered group cohesiveness. 2.1. Data Collection
Unfriending on a social network may resemble cyberos- Survey recruitment was conducted by sending Twitter
tracism which is a form of exclusion and ignoring users who posted about unfriending a reply asking
that occurs in computer-mediated environments [28]. them to take a survey about the topic. The respondents
Ostracism differs from other forms of rejection in that were found and recruited through Sibona and Walczak’s
it is less clearly defined and more ambiguous; those method of Twitter recruitment [23]. Twitter was used to
who are ostracized are often kept from participating with recruit survey participants for several reasons: Twitter
others without an explicit confrontation or explanation. has a large user population where the majority of users
Ostracism in online settings can be more ambiguous have publicly accessible messages; Twitter users had
than in real-life settings because there is a lack of real- a good fit with research (social network sites); it is a
time cues that occur in face-to-face settings [28]. In one simple process to contact a person on Twitter through the
experiment, participants were placed in a online chat @reply mechanism; and the tweets can be screened for
room and were intentionally left out of conversations. recruitment purposes. It is also helpful to recruit people
Subjects who experienced cyberostracism had lower to the survey who had a recent experience with the matter
moods, feelings of belonging, control, self-esteem and for two important factors [2]: (1) Those who experienced
meaningful existence [28]. People who are unfriended an event more recently may be able to provide more
may face similar psychological effects as cyberostracism accurate answers because the event occurred recently. (2)
because unfriending may be viewed as a form of social Those had recently experienced an event may be more
exclusion. willing to take a survey about the topic because they
may still be thinking about the topic. Experiences need
2. Study Design to be reported immediately after they have happened
in order to be remembered [2]. There is not a random
The research was conducted using a survey to determine sample in this research; a purposive sampling method
the survey respondents’ opinions and behaviors about was used to recruit participants. The recruitment tweet
unfriending on Facebook. The survey was conducted was sent in a single tweet of 140 characters and provided
solely on the Internet using a commercially available enough information to the Twitter user to take the
survey tool. The survey questions are a combination survey. The recruitment tweet was designed to follow
of established questions from previous studies and new the methodology of Dillman [2] as much as possible
questions to examine friendship dissolution in online within the constraints of Twitter.
settings plus demographic questions. Surveys were collected between April 17th and
Part one of the survey asked questions about the type September 15, 2010 for 151 total days. 7,327 recruitment
of person unfriended, whether it was for online or offline tweets were sent during the time period. A total of 2,865
behavior, questions about the friendship and questions surveys were started and 1,552 were completed; 54% of
about online and offline behavior. Part two mirrors part those who started the survey completed the survey. The
one of the survey and asks questions about the type of number of surveys in the analysis varies depending on
3274
3272the path the user took during the survey as not every sur- to determine its coefficient and statistical significance. vey respondent answered all four parts of the survey. The Linear regression was used in a forward stepwise fashion analysis of future contact avoidance analyzed 582 survey to include the variables that met statistical significance responses. The surveys were started by 39.6% of those criteria (p
Affect/Behavior Factors
the affect/behavior questions asked were distinct factors Question Affect Avoid Net-
based on Eigenvalues greater than 1. The factors were Contact work
rotated using the Varimax function to determine factor Vigi-
lance
loadings. Component groupings were then analyzed and
It bothered me that this .862 .246
named according to the questions in the group. Three person unfriended me
constructs were confirmed and generated for affect and I felt sad after this person .838
behavior: affect, avoid contact & network vigilance. Two unfriended me
I was surprised to be .736
constructs were confirmed based on Duck’s dissolution unfriended by this person
model: dyadic phase and grave-dressing phase. I was amused to be -.665
The overall model fit was assessed on two models - unfriended by this person
the affect/behavior model and Duck’s Dissolution Model I would avoid seeing this .922
person in real life after the
and are considered acceptable. KMO measure of sam- unfriending
pling adequacy for the affect/behavior model and Duck’s I do not want to see this .910
Dissolution model are .857 and .640, respectively, and person in real life after the
unfriending
are considered acceptable by Hair [9]. The three fac-
It would be uncomfortable .904
tors in the affect/behavior model explain 75% of the for me to see this person in
variance for the factors. The two factor loadings for real life after the unfriending
Duck’s Dissolution model explain 75% of the variance I would avoid seeing this .901
person in real life after the
for the factors. Factor analysis is considered acceptable unfriending (2)
for social science research where more than 60% of the I do not want to see this .864
variance is explained [9]. Bartlett’s Test of Sphericity is person in real life after the
unfriending (2)
statistically significant for the both factor models at the It would be uncomfortable .246 .853
.001 level. for me to see this person in
real life after the unfriending
(2)
I look at the number of .893
3.2. Construct Creation & Reliability Results friends I have on Facebook
to see if someone unfriended
me
Constructs were generated based on the factor analy- I tried to find out who .867
sis results – see Table 2. . The Cronbach’s alpha for unfriended me
the constructs were calculated. The six constructs are Duck’s Dissolution Model
considered reliable: Cronbach’s alpha measures above Question Dyadic Grave-
.70 are considered acceptable [9]. Table 2. shows the Dressing
reliability of the nine constructs and number of questions Person who did the unfriending .923
in the construct. Fifteen types of friends were analyzed stated dissatisfied with relationship
Person who did the unfriending .919
in the research; these friend types are: church, college, stated friendship in trouble
common interest, family member, friend of a friend, Common friends know the .851
friend through a child, friend through a parent, friend relationship is over
through a spouse, grade school, graduate school, high Mutual friends know through .790
Facebook
school, neighbor, romantic partner, work and other. Unfriended told others .782
Cross-loadings above a .200 threshold are shown in the summary table
Table 1.
3.3. Future Contact Avoidance Histogram FACTOR ANALYSIS
Future contact avoidance is skewed to the right where
a large percentage of survey respondents who were un-
friended stated that they would not avoid future contact 3.4. Avoidance Descriptives
(13%) – see Figure 1 . This research shows that there is
Three questions were used to determine a survey respon-
a wide range in the future contact avoidance of friends
dent’s position of future contact avoidance after being
on Facebook and those who are unfriended appear to
unfriended; survey respondents were asked if they avoid,
experience a wide range of responses from avoidance
did not want to see, or would be uncomfortable after
to non-avoidance but the largest group was unlikely to
being unfriended – see Table 3. The three questions
avoid future contact.
were combined into a single construct future contact
avoidance with factor analysis and reliability shown in
3276
3274Measure Questions Cronbach’s Mean Std. Dev N
Alpha
Affect bothered, sad, surprised, amused .794 4.20 1.58 609
Avoid Contact avoid, do not want to see again, uncomfortable, .950 3.56 1.80 611
avoid (2), do not want to see again (2),
uncomfortable (2)
Network Vigilance look at friends number, find person who unfriended .750 3.19 1.83 609
Dyadic Discussion stated dissatisfaction with relationship, stated .846 2.30 1.64 611
friendship in trouble
Grave-Dressing common friends know relationship is over, mutual .748 4.20 1.66 611
friends know through Facebook, unfriended person
told others
Offline Behavior did misdeed, distrust, dislike, disliked behavior, .840 3.03 1.23 599
betray broke rule, disliked personality, new
information
Geographical Distance lives far away, do not see often, will probably not .707 4.66 1.71 598
see again
Peak tie strength very good friends, committed to relationship, spent .907 4.16 2.00 590
lots of time together
Personal Responsibility could have tried to maintain relationship, could .747 3.50 1.17 584
have eased tension in relationship
Valid (listwise) 571
All questions are based on a 1-7 Likert-type scale
Table 2.
C ONSTRUCT D ESCRIPTIVES
Question Disagree No Agree
% Opin- %
ion
%
It would be uncomfortable for me to 47.4 12.3 40.3
see this person in real life after the
unfriending
I do not want to see this person in 47.0 16.5 36.5
real life after the unfriending
I would avoid seeing this person in 53.8 14.7 31.5
real life after the unfriending
Table 3.
AVOIDANCE D ESCRIPTIVES
this person after being unfriended.
3.5. Regression Analysis for Future Contact
Avoidance
Skewness is .267 and Kurtosis is -1.064.
Linear regression was used to determine whether the
Figure 1 . Future Contact Avoidance Histogram
eight constructs (affect, network vigilance, dyadic discus-
sion about unfriending, grave-dressing, offline behavior,
geographic distance, peak relationship and personal
Table 1. and Table 2. The most common reaction to being responsibility) are significant factors in whether a person
unfriended on Facebook is discomfort; 40.3% of survey who is unfriended will avoid contact with that person.
respondents agreed that would be uncomfortable seeing There are nine discrete variables included in the regres-
the person who unfriended them. The next most common sion equation: (1) the number of friends a person has
reaction was that the person did not want to see the on Facebook, (2) the number of friends with whom the
person again - a minority of people (36.5%) expressed person interacts, (3) number of times the person has
that they do not want to see this person again after being unfriended others, (4) frequency that the person looks
unfriended. The lowest level of agreement (31.5%) were at Facebook for updates, (5) frequency that the person
survey respondents who agreed that they would avoid updates their own Facebook account, (6) the number of
3277
3275friends in common between the dyad, (7) the type of Construct B Beta Sig.
friend, (8) frequency that the person was seen in the Continuous Variables
last year, and (9) the length of the friendship. There Constant -.803 .039
are five demographic variables included in the regression
Grave Dressing .430 .396 .001
equation: age, gender, education, years of social network
site use, and whether the person lives in the US. The Affect .230 .202 .001
number survey responses analyzed for this regression Offline Behavior .232 .160 .001
analysis is 582. Geographical Distance .120 .114 .002
The analysis uses avoid contact as the dependent Discussed unfriending in dyad .137 .126 .001
variable to model how the independent continuous and Peak -.091 -.101 .001
discrete variables may predict whether a Facebook user
Discrete Variables
will avoid future contact with the person who unfriended
them. All of the constructs were measured with a 1-7 Yrs soc networking (6) -.525 -.069 .052
Likert-type scale; avoid contact was measured from 1 Type: Family Member -.622 -.084 .019
(will not avoid future contact) to 7 (will avoid future Gender 2.74 .071 .048
contact). The coefficient of determination was measured
Yrs Soc Networking (6) >10 year of social network site use, Gender:
using the adjusted R2 measures. The adjusted R2 for the Male (0), Female (1).
12 variable equation is .283; which means that approxi-
Table 4.
mately 28% of the variance in the construct avoid contact AVOID C ONTACT R EGRESSION R ESULTS
may be explained by the factors in the equation. The
residuals of the regression equation are heteroscedastic
and show no signs of bias in the equation. Six of the
constructs were statistically significant: grave dressing, avoid future contact. The fourth construct, geographical
offline behavior, affect, geographical distance, dyadic distance, indicates that the larger the perceived geo-
discussion and how the person valued peak of the graphical distance between the dyad the more likely
relationship; three discrete variables were statistically that the survey respondent would avoid future contact.
significant. Personal Responsibility was omitted from the The fifth construct, dyadic discussion, indicates that the
regression analysis because it had missing correlations more the survey respondent perceived that the person
with the dependent variable avoid contact. Positive coef- who unfriended them discussed the unfriending prior to
ficients indicate that the person was more likely to avoid its occurrence the more likely that they would avoid
future contact and negative coefficients indicate that the future contact. The last construct, peak, indicates that
person was less likely to avoid future contact. See Table the higher the survey respondent valued the peak of the
4. for details. relationship with the person who unfriended them the
The regression equation coefficients for the continu- less likely they were to avoid future contact. Two of
ous and discrete variables may be interpreted as follows. the discrete variables are negative in their direction and
Six of the constructs were statistically significant and an one, gender, is positive. Negative coefficients indicate
examination of the coefficients indicates the magnitude that the person is less likely to avoid future contact. An
and direction of the effect. Grave-dressing was the most example from the Table 4. is that if the friend type is a
important factor in whether the survey respondent would family member then the survey respondent is less likely
avoid future contact with the person who unfriended to avoid future contact with the person who unfriended
them. The more the survey respondent perceived that them compared to all other friend types. The discrete
they discussed the unfriending with someone (not the variable, gender, is positive and indicates that women
person who unfriended them) the more likely that survey are more likely to avoid future contact with the person
respondent will avoid future contact with the person who unfriended them compared to men.
who unfriended them. The second construct, affect, had
the second largest effect and indicates that the more 4. Discussion
negatively affected (e.g. sad, bothered) the survey re-
spondent is about the unfriending the more likely the The research results show that six factors can predict
survey respondent will avoid future contact. The third whether a person will avoid future contact with the
construct, offline behavior had the third largest effect individual who unfriended them: (1) whether the person
and indicates that the more the survey respondent per- discussed the event after it occurred, (2) the emotional
ceived the unfriending was related to their own offline response to the unfriending (affect), (3) whether the per-
behavior the more likely the survey respondent would son who was unfriended believed it was related to their
3278
3276offline behavior, (4) whether difficulties were discussed unfriended them because the one member knew that
between the dyad prior to the event, (5) the perceived there was an existing issue that caused friction in the
geographical distance between dyad, and (6) how the dyad, that the issue remained unresolved, and that the
person valued the peak-tie strength of the relationship. eventual unfriending was a signal that relationship is
Facebook users who discussed the unfriending with over. One of the technical affordances that Facebook
someone (not the person who unfriended them) after has that face-to-face relationships do not is that there
it occurred, what Duck [6] terms grave dressing, had is a clear signal regarding the relationship through the
the strongest predictive power, overall, for how strongly symbolic link that connects users on social network sites
a person may avoid or be uncomfortable around the online that does not exist offline. Often in face-to-face
person who unfriended them. Duck [6] has hypothesized settings one member does not know who dissolved the
that relationship dissolutions pass through at least four relationship because most relationships fade away [24].
stages before achieving closure. These are an intrapsy- Users who experience a larger negative emotional
chic stage, a dyadic stage, a social stage, and a grave- response to being unfriended (affect) were more likely
dressing stage. Each of these refers to a specific means to avoid future contact with the person who unfriended
of coming to terms with problems in the relationship, them compared to those who experienced a less nega-
depending on whom a dissatisfied partner discusses their tive emotional response. Affect is the second strongest
problems with, and whether this discussion occurs before predictor for future contact avoidance. It is notable that
or after the breakup. Both the dyadic and grave-dressing this construct is the second strongest predictor since it
stages were examined in the present research in order to most directly measures the psychological state of the
test whether Facebook unfriending is at all related to the person who was unfriended. Facebook users who have
current understanding of relationship breakups. processed their emotions by discussing the unfriending
Grave-dressing occurs because people feel a need with someone else (grave-dressing) is a stronger predic-
to publish a record of the relationship’s demise and tor for future contact avoidance compared to the negative
work through a post-mortem of the relationship so that emotions regarding the particular unfriending. This result
future relationships may not develop the same issues supports cognitive-behavioral therapy perspective where
[6]. Facebook users who perceived they went through how a person processes an event can be have a larger
a grave-dressing process by discussing the unfriending role than the event itself.
event with others were more likely to avoid future Users who felt more strongly that their own offline
contact with the person who unfriended them and may behavior was negative said they would avoid future
indicate that the person discussed the event and is contact with the person who unfriended them. The offline
ready to move forward. It is unlikely that those who behaviors in this survey are strongly negative in tone,
discussed the unfriending with others simply had weak- e.g., “I betrayed this person in real life,” and, “This
tie relationships with the Facebook user and thus may be person unfriended me after I violated a rule in real life,”
more likely to discuss the event. The peak-tie strength so those who believe they did these types of misdeeds
construct remains in the analysis (although is the weakest appear to avoid future contact with the person who
predictor in the model) and is a distinct construct from unfriended them more than those who did not engage
grave-dressing and the effects are disambiguated through in these behaviors. The person who was unfriended may
the step-wise regression analysis. recognize that their own behavior was unacceptable, and,
Dyadic discussions between two Facebook friends after the unfriending, may believe that the person is
occasionally occur prior to unfriending where a dissatis- sent a signal that they no longer want to see the survey
fied member addresses concerns regarding the relation- respondent. The negative behavior was based on self-
ship in attempt to fix the issues before dissolution may report survey items so the survey respondent realized
occur. This research does not differentiate who initiated that their behavior was negative; this is not a case
the conversation about the trouble in the relationship; where the person did not realize their actions could be
it could be the person who did the unfriending or the interpreted negatively. Negative offline behavior was the
person who was unfriended. Facebook users who felt third strongest predictor for future contact avoidance.
more strongly that they discussed trouble in the relation- Facebook users who perceived larger geographical
ship prior to its dissolution were more likely to avoid distances between the dyad said they were more likely
future contact with the person who unfriended them. to avoid future contact with the person who unfriended
Dyadic discussions are the fourth strongest predictor them. Geographical distance was the fifth strongest
for future contact avoidance and are also part of the predictor for future contact avoidance. Having a larger
Duck model of friendship dissolution. Facebook users physical distance between the dyad may make it easier to
may want to avoid future contact with the person who avoid contact with the person who unfriended the survey
3279
3277respondent because they simply may have less likelihood real world simply fade away [24], some relationships end
of being in the same organizational contexts as those who as a consequence of a particular event and unfriending
are close in geography. someone may feel like the appropriate action for a user
Peak-tie strength was the weakest predictor in the e.g. after romantic relationship dissolution one member
model of the constructs, overall, where users who had may unfriend the other on the social networking site to
closer relationships at the height of their friendship, the signal that this relationship is over.
less strongly the Facebook user said that they would
avoid future contact following the unfriending. Friends 5. Limitations
who had strong-ties at the peak of their relationship may
be more likely to have bonding social capital (strong) and Participants in the present study were not recruited
less likely to have bridging social capital (weak). Face- randomly. Respondents were recruited via Twitter by
book use has been shown to have a stronger relationship approaching users who had used the terms “unfriend,”
with bridging social capital (weak-tie relationships) than “defriend,” or “unfriending.” The goal this sampling
bonding social capital [7, 25]. Bonding social capital method was to reach people from whom Facebook’s
may allow the dyad to maintain an offline relationship unfriending tool was meaningful and relevant, but it may
where future contact avoidance is less likely even after also have led to the over-representation of those who
being unfriended compared to weak-tie relationships that had been strongly affected by a recent experience. Future
mostly exist on Facebook [7, 16]. Unfriending dissolves studies will work to reach a broader range of participants
the social network site dyadic tie but does not always in order to widen the generalizability of our results.
signal the end of the offline relationship or online rela- The recruitment method may be biased toward survey
tionships in other contexts. respondents who had higher levels of grave-dressing
The technical affordances of Facebook have changed as they were tweeting about unfriending on Twitter.
over time; the site has attempted to present users with It remains unclear how many users feel that tweeting
more relevant information through news feed improve- about unfriending is the same as publicly declaring the
ments and customizations, generates smart lists based end of the relationship. Approximately 37% of survey
on organizational settings (e.g. education and work) and respondents agreed that they told others that they are
geographic proximity to more narrowly cast a message,3 no longer friends with the person who unfriended them
and allows users to hide all posts from a specific member despite their tweeting about unfriending.
from the news feed. The news feed is the center of Finally, the coefficient of determination (R2 ) may
the Facebook home page where there is a constantly be considered relatively low and indicates that approx-
updating list of posts from the member’s social network imately 28% of the variance in the construct future
and the organizations that a person follows.4 The news contact avoidance may be explained by the factors in the
feed attempts to show the most relevant content based equation. The low coefficient of determination indicates
on proprietary algorithms and includes factors like how that there are factors that may explain whether someone
many friends are commenting on a post, who posted the will avoid future contact with someone who unfriended
content, and type of content (link, photo, etc.). them that are not measured in the model. The coefficient
Despite the new technical capabilities on Facebook, of determination is likely to be acceptable given the
unfriending still occurs and has increased to 63% in 2011 difficulty in measuring the psychological response to
from 56% in 2009 [17]. Certain topics, like politics, unfriending at this stage of research.
appear to be more troubling in online conversations.
Rainie & Smith [21] found that 10% of Facebook users
have blocked, unfriended or hidden someone for posting
6. Conclusion
too frequently about political subjects. Facebook users
The goal of the present research was to investigate
could simply hide the members’ posts so they will
factors that predict whether a Facebook user will avoid
no longer be displayed in the news feed but some
contact with a person after being unfriended on Face-
members do take the step to publicly dissolve the tie.
book. The research results show that six factors can
One reason that unfriending may still occur is that social
predict whether a person will avoid future contact with
network site users have a sense of integrity, i.e. when the
the individual who unfriended them: (1) whether the
relationship is over they make a public declaration online
person discussed the event after it occurred, (2) the
by unfriending the user - this is similar to grave-dressing
emotional response to the unfriending, (3) whether the
as described by Duck [6]. While most relationships in the
person who was unfriended believed it was related to
3 http://www.facebook.com/help/friends/lists their offline behavior, (4) the perceived geographical
4 http://www.facebook.com/help/newsfeed distance between dyad, (5) whether difficulties were
3280
3278discussed between the dyad prior to the event and (6) [15] Lee, Y., Kozar, K. A., and Larsen, K. R. (2003). The
how the person valued the peak-tie strength of the technology acceptance model: Past, present, and future.
relationship. Communications of the Association for Information
These findings help place unfriending within the Systems, 12(50):752–780.
greater context of relationship dissolution. The extent [16] Lewis, J. and West, A. (2009). ’friending’:
to which social network sites are different from offline London-based undergraduates’ experience of facebook.
relationship dissolutions will need to be examined further new media and society, 11(7):1209–1229.
in subsequent research, but the present study makes [17] Madden, M. and Smith, A. (2012). Privacy management
clear that unfriending is meaningful and has important on social media sites. Technical report, Pew Research
Center’s Internet and American Life Project.
psychological consequences for those to whom it occurs.
[18] McPherson, M., Smith-Lovin, L., and Cook, J. M.
(2001). Birds of a feather: Homophily in social networks.
Annaul Review of Sociology, 27:415–444.
References [19] Nielsen (2011). State of the media: The social media
report - q3 2011. Technical report, Nielsen.
[1] boyd, D. M. (2006). Friends, friendsters and top 8: [20] Quercia, D., Bodaghi, M., and Crowcroft, J. (2012).
Writing community into being on social network sites. Losing "friends" on facebook. In Proceedings of WebSci
First Monday, 11(12):Online. ’12, Evansons, Illinois.
[2] Dillman, D. A., Smyth, J. D., and Christian, L. M. [21] Rainie, L. and Smith, A. (2012). Social networking
(2008). Internet, Mail, and Mixed-Mode Surveys: The sites and politics. Technical report, Pew Research Center’s
Tailored Design Method. Wiley, 3rd edition. Internet and American Life Project.
[3] Duck, S. (2007). Human Relationships. Sage [22] Sibona, C. and Walczak, S. (2011). Unfriending on
Publications, 4th edition. facebook: Friend request and online/offline behavior
analysis. In Proceedings of the 2011 44th Hawaii
[4] Duck, S. and McMahan, D. T. (2009). The Basics of
International Conference on System Sciences, volume 44,
Communication - A Relational Perspective. Sage
pages 1–10.
Publications, 1st edition.
[23] Sibona, C. and Walczak, S. (2012). Purposive sampling
[5] Duck, S. and McMahan, D. T. (2011). The Basics of
on twitter: A case study. In Proceedings of the 2012 45th
Communication - A Relational Perspective. Sage
Hawaii International Conference on System Sciences,
Publications, 2nd edition.
pages 3510 –3519.
[6] Duck, S. W. (1982). Personal Relationships and Personal
[24] Sprecher, S. and Fehr, B. (1998). The dissolution of
Constructs: A Study of Friendship Formation. John Wiley.
close relationships. Edwards Brother.
[7] Ellison, N. B., Steinfield, C., and Lampe, C. (2007). The [25] Vitak, J., Ellison, N. B., and Steinfield, C. (2010). The
benefits of facebook "friends:" social capital and college ties that bond: Re-examining the relationship between
students’ use of online social network sites. Journal of facebook use and bonding social capital. In Proceedings
Computer-Mediated Communication, 12:1143–1168. of the 2011 44th Hawaii International Conference on
[8] Goldsmith, B. (2009). "unfriend" named word of 2009. System Sciences, volume 44, pages 1–10.
Reuters. [26] Wang, S. S., Moon, S.-I., Kwon, K. H., Evans, C. A.,
[9] Hair, J. J. F., Black, W. C., Babin, B. J., Anderson, R. E., and Stefanone, M. A. (2010). Face off: Implications of
and Tatham, R. L. (2006). Multivariate Data Analysis. visual cues on initiating friendship on facebook.
Pearson Education, Inc., 6th edition. Computers in Human Behavior, 26:226–234.
[10] Hallinan, M. T. (1979). The process of friendship [27] Weber, A. L. and Harvey, J. H. (1994). Perspectives on
formation. Social Networks, 1(2):192–210. Close Relationships. Allyn and Bacon.
[11] Hatfield, E., Rapson, R. L., and Rapson, R. (1995). Love [28] Williams, K. D., Cheung, C. K. T., and Choi, W.
and Sex: Cross-Cultural Perspectives. Allyn & Bacon. (2000). Cyberostracism: Effects of being ignored over the
[12] Joinson, A. N. (2008). Looking at, looking up or internet. Journal of Personality and Social Psychology,
keeping up with people?: motives and use of facebook. In 79(5):748–762.
twenty-sixth annual SIGCHI conference on Human factors [29] Wright, P. H. (1984). Self-referent motiviation and the
in computing systems. intrisic quality of friendship. Journal of Social and
[13] King, W. R. and He, J. (2006). A meta-analysis of the Personal Relationships, 1:115–30.
technology acceptance model. Information & [30] Yoder, C. and Stutzman, F. (2011). Identifying social
Management, 43(6):740–755. capital in the facebook interface. In annual conference on
Human factors in computing systems. ACM.
[14] Lea, M. and Duck, S. (1982). A model for the role of
similarity of values in frienship development. British
Journal of Social Psyschology, 21:301–310.
3281
3279You can also read
