Docs 3
Docs 3
Docs 3
Bradley J. Bond
University of San Diego, USA
Abstract
The COVID-19 pandemic provided an extraordinarily unique opportunity to investigate
how sudden, involuntary alterations to social routines influenced not just social rela-
tionships, but also parasocial relationships with fictional characters and celebrities.
Results from a four-wave panel survey administered during the COVID-19 pandemic
(N ¼ 166) revealed that social relationships maintained their stability during social dis-
tancing, particularly among participants who increased mediated social engagement with
friends and those low in attachment anxiety. Parasocial closeness with media personae
increased over time, suggesting that favorite media personae became more meaningful as
participants engaged in social distancing. Parasocial closeness increased with greater
intensity among participants who increased parasocial engagement, reduced face-to-face
social engagement with friends, increased mediated social engagement with friends, and
those low in attachment anxiety. Results are discussed in terms of the potential influence
of increased mediated social engagement with real-life friends on the parasocial pro-
cessing of celebrities and fictional characters.
Keywords
Friendship, media, media multiplexity, parasocial, social distance
Friendships are capable of fulfilling the instinctive human desire for social contact.
Unlike familial, marital, or vocational relationships, friendships are free from societal or
contractual obligations, making friends among the most important others with whom
Corresponding author:
Bradley J. Bond, Department of Communication Studies, University of San Diego, 5998 Alcala Park, San Diego,
CA 92110, USA.
Email: [email protected]
2 Journal of Social and Personal Relationships XX(X)
people voluntarily maintain association (Fehr, 1996; Korn & Nicotera, 1993). Human
propensity for social contact may not only explain friendships, but may also explain
individuals’ perceived connections with celebrities and fictional characters (i.e., media
personae) known to them only through screens. The enduring socio-emotional bonds that
individuals develop with media personae are defined as parasocial relationships (Horton
& Wohl, 1956). Parasocial relationships mimic the familiarity of real-life friendships.
People have close group of friends that they consider their social circle, but they may
also feel connected to television characters, YouTube influencers, and musical artists
such that they too are perceived as social companions.
Social and parasocial relationships are not interchangeable; yet, theories pre-
dicting interpersonal relationship development have successfully been applied to
parasocial relationships (Eyal & Dailey, 2012; Rosaen & Dibble, 2016), individual
differences that have traditionally predicted friendship ties have also predicted how
individuals engage with celebrities and fictional characters (Cole & Leets, 1999;
Schiappa et al., 2007), and both friends and media personae can gratify the need for
connection and belonging (Bond, 2018; Madison et al., 2016). How might social and
parasocial relationships function, however, when people are faced with a global
crisis that abruptly and involuntarily halts their abilities to sustain routine in-person
social contact, thereby increasing time spent with screens? The historic global
pandemic caused by the novel coronavirus COVID-19 afforded an unprecedented
opportunity to study the influence of social distancing on our relationships, both
social and parasocial.
The COVID-19 outbreak was first reported in December 2019. Once epidemiologists
determined that COVID-19 was an airborne virus, medical professionals began pleading
with the public to “social distance.” Social distancing referred to the need to maintain 6
feet of distance between oneself and others, wear facial coverings whenever in the
presence of others, and avoid congregating. To maintain social distancing protocol,
people began to shelter-in-place either voluntarily or as required by city and state
governments. Schools transitioned to remote learning, restaurants and movie theaters
shuttered their doors, and corporations moved meetings from the boardroom to the living
room. Social distancing by definition required a hasty loss of face-to-face (FtF) inter-
action with friends, but the practice did not obstruct people’s access to media. As long as
people had their smart phones and television screens, they had access to mediated
communication with friends and to their favorite media personae.
The objective of the present study was to explore how impediments to FtF social
engagement during a global pandemic altered the way people used media to
maintain communication with friends, and how media dependency for social
engagement may have influenced the function of favorite celebrities and fictional
characters in people’s perceived social networks. The literatures on social and
parasocial relationships are briefly reviewed as a means of introducing the potential
influence of social distancing. Hypotheses rooted in media multiplexity theory,
attachment anxiety, and parasocial compensation are then posed before detailing the
results of a longitudinal survey designed to investigate engagement with and clo-
seness to friends and favorite media personae at four intervals during the initial
months of the COVID-19 pandemic.
Bond 3
Social relationships
Friends act as confidants and companions by providing individuals with cognitive,
emotional, and behavior resources needed to thrive in their social environments. Varying
definitions and dimensions of friendships exist in the literature, though intimacy,
acceptance, and mutual assistance are often-cited attributes of close friends (Korn &
Nicotera, 1993). Consistent, affirming communication is generally seen as vital to
maintaining these social ties (Fehr, 1996). Modern history has not furnished scholars
with many crises in which to make lateral comparisons to pandemic-induced social
distancing, though the literature on long-distance friendships could prove fruitful here.
Social distancing may not have increased geographical distance between friends, but the
constraints on normative FtF interactions during the pandemic parallel the typical long-
distance friendship.
Long-distance friendships have been studied through the lens of turning points,
defined as events or occurrences that lead to significant alterations within interpersonal
relationships (Becker et al., 2009). Increased physical distance has been the most fre-
quently cited turning point negatively influencing long-distance friendships (Johnson
et al., 2003). From high school students moving away to college (Fehr, 1999) to service
members being deployed for duty (Vogt et al., 2005), the difficulty in maintaining long-
distance social relationships has been attributed to the dissatisfaction stemming from
deficits in FtF social engagement; increased investment required to sustain self-
disclosure, reciprocity, and mutual support; and increased attractiveness of geo-
graphically close alternatives.
The predicted negative association between distance and social relationships could be
considered antiquated in the age of emojis, social media platforms, and video con-
ferencing apps. As Johnson and colleagues (2009) argue, the reward-cost ratio tradi-
tionally characterizing long-distance friendships has been amended by accessible and
affordable mediated communication. Among individuals who do not see their friends
FtF, mediated social engagement is significantly correlated with friendship satisfaction
(Johnson et al., 2009). Participants consistently report utilizing social media for rela-
tionship maintenance purposes (Baym, 2015), and specific features such as video chat
have been cited as effective tools for social engagement because they overcome prox-
imal constraints to maintaining long-distance friendships (Eden & Veksler, 2016). The
use of mediated channels of communication to maintain social relationships can be
explained by the assumptions and predictions of media multiplexity theory.
Media multiplexity theory assumes that individuals with strong social ties to one
another will utilize all available communication channels to maintain their relationships
(Haythornthwaite, 2002). Individuals with strong ties to one another are motivated by
their interdependence to use more diversified media and to use those respective media
with greater frequency than individuals with weak ties to one another. Individuals with
strong ties can also supplement the loss of one form of engagement with another; the
more varied the communication channels used between relational partners, the less
relational stress when one of those channels is altered (Haythornthwaite & Wellman,
1998). Though the theory was originally predicated on task-based relationships, research
suggests that it can also be applied to friendships (Baym & Ledbetter, 2009; Ledbetter,
4 Journal of Social and Personal Relationships XX(X)
2009). In the context of the COVID-19 pandemic, mediated engagement with friends
may have played a particularly salient role in maintaining social closeness as opportu-
nities for FtF engagement waned or dissolved entirely.
Parasocial relationships
Positive parasocial relationships are defined as feelings of intimacy and kinship that
individuals develop with celebrities and fictional characters transcending episodic media
experiences. Unlike social relationships, parasocial relationships lack reciprocity. Indi-
viduals may feel closely connected to fictional characters, but the fictional characters do
not reciprocate the intimacy. Parasocial relationships develop, maintain, and dissolve
similarly to real-life friendships because the cognitive and affective mechanisms that
humans employ to process interpersonal experiences map onto parasocial engagements
(Reeves & Nass, 1996). For example, studies show that parasocial relationships are most
often formed with media personae who depict traits or engage in behaviors that audi-
ences deem attractive and likeable, similar to friendships (see Hartmann, 2017; Schiappa
et al., 2007). Though interpersonal theories have been applied to parasocial relationships
(Eyal & Dailey, 2012), media scholars have largely rooted the concept in the uses and
gratifications perspective (Rubin et al., 1985). Given that uses and gratifications assumes
people are motivated to seek out media experiences that gratify needs, parasocial rela-
tionships have been regarded as mediated interpersonal connections that, based on social
and psychological circumstances, can satisfy individuals’ desires for social contact
(Hartmann, 2017).
Parasocial relationships can theoretically serve as functional alternatives to social
relationships, a prediction often referred to as the parasocial compensation hypothesis
(Madison et al., 2016). Contradictory to the hypothesis, studies suggest that media
personae do not generally compensate for friendship deficiencies. Rather, parasocial
relationships complement social relationships because the same psychological tools
needed to flourish interpersonally are needed to do so parasocially (Rosaen & Dibble,
2016; Rubin et al., 1985). In reviewing studies on parasocial relationships, Hartmann
(2017) concluded, “people with greater—not weaker—interpersonal skills develop more
intense parasocial relationships” (p. 135). The inability of parasocial compensation to
predict parasocial relationships holds true even among individuals who experience
relational maintenance difficulties (Madison et al., 2016) and those who experience
loneliness induced by specific negative events (Wang et al., 2008)—situations that might
be considered akin to pandemic-induced social distancing. There are notable exceptions
that support the compensation function of parasocial relationships and may also speak to
the value of parasocial relationships during the COVID-19 pandemic. Bond (2018)
found that young people who lacked real-life friends in a specific identity domain
compensated for the relational shortcoming by developing strong parasocial relation-
ships. Several studies also suggest that priming individuals to think about their most
liked celebrities and fictional characters can buffer negative emotions in times of distress
(Derrick et al., 2009; Twenge et al., 2007). Given the alterations to FtF social engage-
ment and the considerable amount of stress caused by the COVID-19 pandemic (Taylor
Bond 5
et al., 2020), favorite media personae may have played a compensatory role during social
distancing.
Attachment anxiety
Social and parasocial relationships have been studied using multiple methods and
designs that have included myriad of predictor variables. Attachment style is highlighted in
the current study because one’s anxious attachment tendencies may have particularly robust
predictive power in the context of a pandemic. Relational attachments are developed from
childhood experiences with caregivers and are defined by anxiety and avoidance (Hazan &
Shaver, 1987). Individuals low in both dimensions tend to hold positive relationship
expectations, confront conflict effectively, and commit to relational partners with ease.
Individuals low in anxiety and high in avoidance place less value in their relationships, keep
to themselves during times of distress, and are more satisfied to be alone. Of most interest
here are individuals high in anxiety who characteristically seek frequent contact with their
relational partners, experience greater relational uncertainty, and exhibit more extreme
highs and lows in their relationships. Attachment anxiety has been negatively related to
variables that are predictive of healthy social relationships, such as perceived social support
and social self-efficacy (Mallinckrodt & Wei, 2005), but has been positively correlated with
parasocial relationships (Rosaen & Dibble, 2016), potentially because forming connections
with media personae is one means of overcoming relational anxiety (Cole & Leets, 1999). A
nationally representative survey study found that individuals high in anxiety sensitivity and
those most fearful of uncertainty (e.g., classic attributes of the anxious attachment style)
experienced the most COVID-related stress (Taylor et al., 2020). As such, anxious
attachment was measured in the present study to probe the potential effects of attachment
style on relational perceptions during pandemic-induced social distancing.
Cross-sectional hypotheses
Research suggests that FtF interactions are preferred methods of communicating with
close friends (Caughlin & Sharabi, 2013; Ramirez & Wang, 2008), but that mediated
communication often eases the costs of maintaining social relationships that geo-
graphically transition from close to distant (Eden & Veksler, 2016; Johnson et al., 2009).
Media multiplexity theory also predicts that frequency of social engagement across a
variety of communication channels is related to relationship strength (Ledbetter, 2009).
Parasocial relationships mimic real-life friendships in their maintenance strategies such
that more parasocial engagement, or time spent with favorite media personae, increases
6 Journal of Social and Personal Relationships XX(X)
perceived closeness (Schiappa et al., 2007). Recent research also suggests that media
multiplexity theory can predict parasocial closeness with media personae (Ledbetter &
Meisner, 2021). As such, the following hypotheses are proposed regarding the rela-
tionships between engagement with friends and media personae and closeness:
H1: A positive relationship will exist between social closeness and both FtF social
engagement and mediated social engagement during the COVID-19 pandemic.
H2: A positive relationship will exist between parasocial closeness and parasocial
engagement during the COVID-19 pandemic.
Attachment anxiety may explain the effects of social distancing on social and para-
social relationships. Those low in anxiety may perceive their relationships as durable
beyond any social distancing effects. Among anxious individuals, however, the stress of
social distancing is likely to cause greater relational uncertainty among friends (Taylor
et al., 2020). Conversely, studies suggest that those high in anxious attachment may
develop stronger parasocial relationships as a means of coping (Cole & Leets, 1999),
reflected in the following hypothesis:
H3: Attachment anxiety will be negatively related to social closeness and posi-
tively related to parasocial closeness during the COVID-19 pandemic.
Longitudinal hypotheses
Though media multiplexity theory suggest that contemporary media affordances can
overcome hurdles to relational maintenance among those who are geographically distant,
significant disruptions to routine communication behaviors have been described as
relational turning points that can negatively affect social closeness (Johnson et al., 2003).
Moreover, individuals who experience difficulty transitioning between FtF social
engagement and mediated forms of engagement also report relational distress (Caughlin
& Sharabi, 2013). The pandemic-induced changes to social routines were abrupt and
unconventional, increasing the likelihood that individuals experienced at least some
obstacles in their transition to a socially distanced environment. As such, the following
hypotheses predicting longitudinal change in social closeness are posited:
insufficient (Bond, 2018) or during times of distress (Derrick et al., 2009; Twenge
et al., 2007), which likely characterizes the initial period of the COVID-19 pandemic
under investigation in the present study. Moreover, media multiplexity theory has been
predictive of parasocial closeness, suggesting that more frequent exposure to favorite
media personae would increase closeness (Ledbetter & Meisner, 2021). Attachment
anxiety has also been positively correlated with parasocial relationships (Cole & Leets,
1999). As such, the following hypotheses predicting longitudinal change in parasocial
closeness are posited:
Method
Procedure
Participants were recruited through paid advertisements and posts on social media
platforms. Data were collected in four intervals (T1, T2, T3, and T4). Participants were
contacted 2 weeks after the date they completed the first questionnaire with a request to
complete the second, a process that repeated through T4. The first participant in T1
completed the questionnaire on April 7, 2020. By this date, the Institute for Health
Metrics and Evaluation (2020) reported mobility in the United States to be at its lowest
(down 53% from pre-pandemic levels) and people were self-reporting adherence to
social distancing protocols (Taylor et al., 2020). The final participant in T4 completed
the questionnaire on June 7, 2020. Participants received a $10 gift card as incentive.
Participants
The T1 questionnaire garnered 323 valid responses; the T4 questionnaire was completed
by 166 participants, equating to an attrition rate of 49% over the 8-week study duration.1
Participants (N ¼ 166) ranged in age from 18 to 73 (M ¼ 26.87, SD ¼ 8.44). The sample
was 61% female (n ¼ 101), 39% male (n ¼ 65). The racial composition of the sample
was as follows: 81% White (n ¼ 134), 7% Latinx (n ¼ 11), 5% Asian American/Pacific
Islander (n ¼ 9), 4% Black/African American (n ¼ 7), and 3% mixed race (n ¼ 5).
Participants represented 25 U.S. states and the District of Columbia. Though most
participants were working full-time (59%, n ¼ 98), 18% (n ¼ 29) reported being
8 Journal of Social and Personal Relationships XX(X)
unemployed due to COVID-19. Over 90% (n ¼ 150) reported cohabitating with at least
one other adult during social distancing.
Measures
People are capable of sustaining multiple social and parasocial relationships simulta-
neously. As such, the first questionnaire asked participants to list the names of four close
friends, defined as “people you would typically socialize with who you feel connected to
and, under normal circumstances, you would see or talk to at least somewhat regularly,”
and to list the names of four fictional characters or celebrities who “you feel connected
to, almost like they are your friends even though you’ve never met them in real life.”
Participants subsequently reported engagement and closeness for each friend and each
media persona during each wave of data collection. Designing the questionnaire to
measure multiple social and parasocial relationships has the potential to advance the
parasocial literature given that previous studies investigating parasocial relationships are
almost exclusively designed to measure one parasocial relationship between the parti-
cipant and a media persona (see Bond, 2018; Eyal & Dailey, 2012; Madison et al., 2016)
even though people likely manage multiple parasocial relationships at any given time.
The alteration to the one parasocial relationship measurement common in previous work
reflects this reality (see supplementary materials for examples of participant-reported
favorite media personae). Participants reported social engagement, social closeness,
parasocial engagement, parasocial closeness, and media exposure during each cycle of
data collection. Participants reported demographics (i.e., age, race, and gender, living
conditions) and anxious attachment once.
Social closeness. Five modified items from the psychological closeness scale (Vangelisti
& Caughlin, 1997) were repeated for each of four self-reported close friends to mea-
sure social closeness. The names of participants’ friends were piped into the items.
Sample statements include, “I enjoy spending time with [name],” and “I can talk about
personal things with [name].” Participants responded on a 7-point scale from 1
Bond 9
Parasocial engagement. Participants reported how often they were exposed to each of the
four self-reported favorite media personae yesterday, responding from 1 (Not at All) to 5
(Very Often). A mean of the 4 exposure items was created for each cycle of data col-
lection (a ¼ .79–.81). For growth curve analyses, change scores were created for
parasocial engagement by subtracting T1 parasocial engagement from T4 parasocial
engagement; higher scores indicated greater increases in exposure to favorite media
personae.
Parasocial closeness. Six items from the parasocial interaction scale (Rubin et al., 1985)
were repeated for each of four self-reported favorite media personae as a measure of
parasocial closeness. The names of participants’ favorite celebrities and characters were
piped into the scales. Sample statements include, “[Name] makes me feel comfortable, as
if I am with a friend,” and “[Name] seems to understand the kind of things I want to
know.” Participants responded on a 7-point scale from 1 (Strongly Disagree) to 7
(Strongly Agree). A mean of all 24 items (4 media personae 6 items) was created for
each cycle of data collection (a ¼ .92–.95). Higher scores indicated stronger parasocial
closeness.
Anxious attachment. Participants were presented with the 5 items measuring anxious
attachment from the attachment styles scale (Feeney & Noller, 1992), modified to
address social relationships. Example items include, “I find that others are reluctant to
get as close as I would like,” and “I often worry that my friends won’t want to stay
around me.” Participants responded from 1 (Strongly Disagree) to 5 (Strongly Agree).
Higher scores indicated more anxious attachment (a ¼ .86), M ¼ 2.60 (.94).
Results
Preliminary analyses
Participants reported spending less time engaged in FtF communication with friends and
more time engaged in mediated communication with friends than pre-pandemic research
would suggest is typical (Eden & Veksler, 2016). ANOVAs revealed that participants
10 Journal of Social and Personal Relationships XX(X)
Table 1. Social and parasocial engagement means at each wave of data collection.
FtF Social Engagement Mediated Social Engagement Parasocial Engagement F (2, 495) ¼
T1 2.81 (1.39)a 3.56 (.94)b 3.30 (.93)b 19.63***
T2 2.40 (1.26)a 3.46 (.97)b 3.43 (.98)b 52.13***
T3 2.53 (1.14)a 3.49 (1.03)b 3.31 (.89)b 40.99***
T4 2.69 (1.10)a 3.52 (1.06)b 3.50 (.92)b 35.05***
Note. All engagement items were measured from 1 (Not at All) to 5 (Very Often). ANOVAs were employed to
test differences in engagement scores at each wave of data collection. Superscripts that differ across rows
indicate statistically significant differences between means using Scheffe post-hoc comparisons. ***p < .001.
Table 2. Regression analyses predicting social closeness at each wave of data collection.
T1 T2 T3 T4
Social Closeness .71*** .68*** .67***
Attachment Anxiety .06 .04 .07 .14*
FtF Social Engagement .29** .07 .06 .02
Mediated Social Engagement .25* .16* .33*** .21**
Media Exposure .13 .04 .02 .09
Parasocial Engagement .03 .01 .12 .03
F T1 (5, 160) T2–T4 (6, 159) 2.71* 35.97*** 31.09*** 37.70***
R2 .08 .55 .46 .51
DR2 .03 .08 .08
Note. All coefficients are standardized. Social closeness scores from the previous wave of data collection were
entered into the T2–T4 models prior to other predictor variables (i.e., social closeness in T2 model was
measured at T1). T2–T4 R2 represents variance explained by prior social closeness score; T2–T4 DR2 repre-
sents variance explained by the addition of the predictor variables. *p < .05, **p < .01, ***p < .001.
reported significantly less FtF social engagement than mediated social engagement or
parasocial engagement at each data collection interval, evidence that participants were
adhering to some degree of social distancing protocol during the panel study (Table 1).
Participants spent an average of 9 hours/day with media (M ¼ 8.92, SD ¼ 6.26),
slightly more than pre-pandemic national estimates of 7.5 hours/day (Nielsen, 2020).2
Watching television and streaming services garnered the most time (M ¼ 3.08,
SD ¼ 1.80), followed by browsing social media (M ¼ 2.04, SD ¼ 1.55), and surfing the
web (M ¼ 1.84, SD ¼ 1.50). Participants spent less time playing video games (M ¼ .97,
SD ¼ 1.35) or reading (M ¼ 1.00, SD ¼ 1.27). Media consumption by platform mirrored
national patterns (Nielsen, 2020), suggesting that participants’ media use habits did not
deviate from pre-pandemic norms.
Table 3. Regression analyses predicting parasocial closeness at each wave of data collection.
T1 T2 T3 T4
Parasocial Closeness .58*** .60*** .57***
Attachment Anxiety .10 .07 .19*** .17**
FtF Social Engagement .14 .19** .01 .01
Mediated Social Engagement .00 .10 .12* .11*
Media Exposure .13 .16** .03 .01
Parasocial Engagement .44*** .24*** .22*** .24***
F T1 (5, 160) T2–T4 (6, 159) 15.43*** 40.08*** 61.17*** 65.68***
R2 .33 .51 .58 .62
DR2 .10 .12 .09
Note. All coefficients are standardized. Parasocial closeness scores from the previous wave of data collection
were entered into the T2–T4 models prior to other predictor variables (i.e., parasocial closeness in T4 model
was measured at T3). T2–T4 R2 represents variance explained by prior parasocial closeness score; T2–T4 DR2
represents variance explained by the addition of the predictor variables. *p < .05, **p < .01, ***p < .001.
For all subsequent regression models, the prior closeness score was entered into the first
block, and the remaining predictor variables were entered into the second block. For
example, the model testing predictors of T3 social closeness included T2 social closeness
on the first block. Controlling for prior closeness scores increased the rigor of the
regression analyses examining closeness as an outcome variable and allowed for stronger
causal claims regarding predictor variables. Participant age, gender, and race were initially
entered as control variables in all multiple regression analyses. Two-way interactions were
also tested. None significantly contributed to social or parasocial closeness in any
meaningful way and were subsequently removed from analyses to preserve power.
H1 predicted that FtF social engagement and mediated social engagement would
positively contribute to social closeness. FtF social engagement was negatively related to
social closeness at T1, though the correlation was moved to non-significance at T2 and
remained non-significant at T3 and T4. Mediated social engagement was positively
related to social closeness at each wave of data collection, even when controlling for
prior social closeness (Table 2). H1 was partially supported.
H2 predicted that parasocial engagement would positively contribute to parasocial
closeness. This hypothesis was supported: parasocial engagement was positively cor-
related with parasocial closeness at each wave of data collection, even when controlling
for previous parasocial closeness (Table 3).
H3 predicted that attachment anxiety would be negatively related to social closeness
and positively related to parasocial closeness. The negative relationship between
attachment anxiety and social closeness only reached significance at T4 (Table 2). A
negative relationship between attachment anxiety and parasocial closeness was signif-
icant at both T3 and T4 (Table 3). Little support was found for H3.
increase the interpretability of the parameters, a standard approach to latent growth curve
modeling in the social sciences (Duncan et al., 2006). When model fit indices suggested
significant change over time, a nested model with fixed slopes (T1 ¼ .00, T2 ¼ .33,
T3 ¼ .66, T4 ¼ 1.00) was run to determine linearity. After testing simple growth curve
models predicting changes in social closeness and parasocial closeness, moderators were
entered into the models. Moderation was indicated by significant slope values (Duncan
et al., 2006). The moderation models initially included participant age, gender, race, and
general media exposure to control for variance explained by demographic differences.
However, age, gender, race, and general media exposure weakened the model fit indices
and none significantly moderated change in social closeness or parasocial closeness. As
such, they were excluded from the models detailed below.
Social closeness. Social closeness was predicted to weaken during social distancing (H4).
Social closeness means suggested little change in the variable over the duration of the
study: T1 ¼ 6.22 (.55), T2 ¼ 6.11 (.69), T3 ¼ 6.22 (.54), T4 ¼ 6.23 (.55). The growth
curve indicated moderately poor fit, confirming the stability of social closeness over
time, w2 (3) ¼ 17.49, p < .001, NFI ¼ .95, CFI ¼ .95, RMSEA ¼ .09 (.05, .13). H4 was
not supported.
Over time changes in social closeness were predicted to differ by social engagement
(H5) and anxious attachment (H6). FtF social engagement, mediated social engage-
ment, and anxious attachment were entered as moderators into the social closeness
model. The moderation model was also a moderately poor fit, w2 (12) ¼ 38.75, p < .001,
NFI ¼ .92, CFI ¼ .94, RMSEA ¼ .06 (.04, .08), though the slopes for attachment
anxiety (.10, SE ¼ .03, p < .001) and mediated social engagement (.49, SE ¼ .06,
p < .001) were significant. The slope for FtF social engagement was not significant
(.02, SE ¼ .02, p ¼ .21). Slope analyses suggest that participants lower in attachment
anxiety maintained social closeness (Figure 1A), and those who increased their
mediated social engagement with friends also increased their social closeness during
social distancing (Figure 1B). H5 was partially supported; H6 was supported.
Figure 1. Visualization of social closeness growth curve moderators. Note. Panel A displays
attachment anxiety as a moderator. Panel B displays mediated social engagement with friends as a
moderator. Moderators were recoded into ordinal measures for the purpose of visualizing the
growth curve analyses; levels were created using means and standard deviations (e.g., low
attachment anxiety ¼ >1 SD below the mean, moderate ¼ þ/1 SD, high ¼ <1 SD above the
mean).
Figure 2. Visualization of parasocial closeness growth curve moderators. Note. Panel A displays
parasocial engagement with media personae as a moderator. Panel B displays FtF social engage-
ment with friends as a moderator. Panel C displays mediated social engagement with friends as a
moderator. Panel D displays attachment anxiety as a moderator. Moderators were recoded into
ordinal measures for the purpose of visualizing the growth curve analyses; levels were created
using means and standard deviations (e.g., low attachment anxiety ¼ >1 SD below the mean,
moderate ¼ þ/1 SD, high ¼ <1 SD above the mean).
14 Journal of Social and Personal Relationships XX(X)
30.19, p ¼ < .05, NFI ¼ .95, CFI ¼ .98, RMSEA ¼ .03 (CI: .01, .05). The slopes for all
proposed moderators were significant: parasocial engagement (.15, SE ¼ .04, p < .01),
FtF social engagement (.06, SE ¼ .03, p < .05), mediated social engagement
(.42, SE ¼ .08, p < .001), and attachment anxiety (.17, SE ¼ .04, p < .001). Parasocial
closeness increased over the course of the study with greater intensity among those
who increased their parasocial engagement (Figure 2A), decreased their FtF social
engagement (Figure 2B), increased their mediated social engagement (Figure 2C), and
reported less anxious attachment (Figure 2D). H8 was supported, H9 was partially
supported, and H10 was not supported.
Discussion
The objective of the present study was to investigate social and parasocial relationship
maintenance during the COVID-19 pandemic. Results from a four-wave panel survey
suggest that social closeness among friends remained stable during the pandemic, par-
ticularly among individuals who increased their mediated communication with friends
and those with less attachment anxiety. Parasocial closeness to favorite celebrities and
fictional characters grew stronger during the pandemic. Parasocial closeness growth was
stronger among individuals low in attachment anxiety, those who spent more time with
their favorite media personae, more time communicating with friends through mediated
channels, and less time communicating with friends FtF.
Social relationships
Increased geographical distance is a turning point that can negatively influence
friendships (Johnson et al., 2003, 2009), but pandemic-induced social distancing did
not have an adverse effect on social closeness in the present study. The stability of
social closeness may be explained by the context through which social distancing is
translated as a turning point. In a large-scale survey study, participants reported turning
to friends for reassurance as a means of coping with COVID-related stress (Taylor
et al., 2020). Reflections on the COVID-19 pandemic as a turning point may focus on
the ways in which friends provided social support during this rather tumultuous time
rather than on the social distance, thereby alleviating negative influences on social
closeness (Johnson et al., 2003).
Social closeness holding constant over the course of the study may also be explained
by media multiplexity theory (Haythornthwaite, 2002). Modern technological advances
in mediated communication that have been argued to mitigate perceived costs of
maintaining long-distance friendships (Johnson et al., 2009) have similar capabilities
during the abrupt alterations to social engagement caused by the COVID-19 pandemic.
Media multiplexity theory assumes that individuals with strong ties to one another would
replace FtF communication with other communication modalities to maintain their
relationships (Haythornthwaite & Wellman, 1998). Mediated social engagement not
only occurred with greater frequency than FtF social engagement during the pandemic,
but was more frequent in the present study than in pre-pandemic research using the same
measures (Eden & Veksler, 2016). Mediated social engagement was also positively
Bond 15
correlated with social closeness at every stage of data collection even when controlling
for prior social closeness. Given that close friends fulfill human desires for social contact
in meaningful ways, individuals seem to actively invest in varying channels of mediated
communication to sustain strong social ties (Caughlin & Sharabi, 2013). Indeed, social
gatherings at bars and restaurants were replaced with social gatherings on video chat
apps, some of which experienced nearly 300% increases in downloads during the weeks
immediately prior to data collection for the present study (O’Brien, 2020). During
pandemic-induced social distancing, people seem to replace FtF social engagement with
mediated social engagement, an interpersonal transition that would be in line with
previous research on long-distance friendships (Johnson et al., 2009) and media multi-
plexity theory (Haythornthwaite, 2002).
One notable correlation in the regression analyses investigating social closeness
suggested that the normalization of social distancing was still evolving at the time of
the study: the negative relationship between FtF social engagement and social close-
ness at T1. Though seemingly counterintuitive, this finding may be explained by
expectancy violation theory (Burgoon & Jones, 1976). If participants were following
social distancing protocol as the data suggest, then FtF encounters with friends in the
early stages of the pandemic likely failed to meet relational expectations. When facial
coverings obstructed nonverbal cues and 6 feet of separation prevented physical touch,
social norms for interpersonal interactions were arguably violated. Those violations
had subsequent effects on perceptions of the interpersonal relationship. The negative
relationship between FtF social engagement and social closeness was moved to non-
significance by T2, potentially because individuals began to assimilate to social dis-
tancing and, in turn, altered their expectations for FtF interactions during the
pandemic. Alternatively, participants may have overcome the violation by recalling the
reward potential of their close friends in an environment with few quality alternatives.
Regardless, at no time during data collection was FtF social engagement positively
correlated with social closeness. Investigating FtF interactions in the age of social
distance rules and expectations would be an informative venue for future research on
COVID-19 effects on social relationships.
Parasocial relationships
Parasocial relationships experienced significant growth over the course of the study.
Though participants spent more time with media during the first months of the
COVID-19 pandemic than national pre-pandemic estimates, media exposure alone did
not consistently predict parasocial closeness; parasocial engagement was required to
increase parasocial closeness, mirroring social relationship development in real-life and
suggesting that media multiplexity theory may be validly applied to parasocial rela-
tionships (Ledbetter & Meisner, 2021). Hartmann (2017) argues that our parasocial
relationships benefit from reliability. A liked news anchor will appear on one’s television
at the same time each day. A favorite fictional character on a streaming service situation
comedy is available at the click of a button. During the COVID-19 pandemic, social
distancing required modifications to communication patterns with friends, but media
personae were as reliably accessible as they had been pre-pandemic. In fact, many media
16 Journal of Social and Personal Relationships XX(X)
attention among scholars dissecting the social and psychological functions of parasocial
relationships.
Limitations
Though this study provides insight into social and parasocial relationships during an
extraordinarily unique period of social distancing that would be implausible to create in a
lab setting, it is not without limitations. The recruitment procedures and high attrition
rate (resulting in a smaller than expected sample size) limit the statistical power and
generalizability of the study findings. Moreover, data collection began after social dis-
tancing protocol were already commonplace, and ended prior to the pandemic subsiding,
limiting abilities to make pre- or post-social distancing comparisons. In terms of mea-
surement, social engagement was operationalized using a series of items that measured
various means of communication (e.g., phone call, text), but the operationalization of
parasocial engagement was unidimensional, preventing analyses that would test the
primary prediction of media multiplexity theory: variance in modality of communication
is related to tie strength. Measuring parasocial engagement across platforms (e.g.,
television, film, Twitter, Instagram) would allow researchers to more closely dissect the
way that individuals engaged with parasocial others during the COVID-19 pandemic and
to more validly test the media multiplexity model as it applies to parasocial relationships.
Studies suggest that social and parasocial relationships are capable of providing social
support that alleviates stress (Korn & Nicotera, 1993; Derrick et al., 2009, respectively).
Given the highly stressful nature of a health crisis like COVID-19 (Taylor et al., 2020),
measuring emotional well-being outcomes would have been warranted. Future research
should take up this missed opportunity.
Conclusions
The COVID-19 pandemic was a global health crisis that stymied FtF interactions with
close friends. Social distancing protocol crafted to prevent the spread of the virus may
have had an initial negative effect on social closeness among those who attempted to
interact FtF, but people were capable of utilizing mediated channels to maintain strong
ties with close friends. Individuals experiencing high attachment anxiety during the
pandemic found it more difficult to sustain social closeness. Parasocial relationships with
favorite media personae grew in intensity over time, particularly for those who may have
been using media experiences to compensate for social deficiencies brought on by the
pandemic. Parasocial engagement did not require audiences to wear facial coverings
while flipping through magazines, to keep at least 6 feet away from their televisions, or
to avoid viewing film scenes depicting large gatherings. Parasocial relationships were
originally described as “intimacy at a distance” (Horton & Wohl, 1956). During the
COVID-19 pandemic, social relationships were also intimate at a distance. As FtF
interactions gave way to smart phones and social media, engagement with friends began
to mirror engagement with media personae. Ultimately the lack of FtF interactions and
reliance on screens may have increased the similarities between friends and media
Bond 19
Author’s note
This manuscript was presented at the 2021 International Communication Association conference.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/
or publication of this article: This work was funded by an internal faculty research grant from the
College of Arts & Sciences at the University of San Diego.
ORCID iD
Bradley J. Bond https://orcid.org/0000-0001-5707-2089
Supplemental material
Supplemental material for this article is available online.
Notes
1. There were no significant differences in mean scores of the predictor or criterion variables
between participants who completed the study and those who did not. Comparisons were made
after each wave of data collection.
2. Nielsen’s (2020) quarter four data were recalculated to remove time spent on social media.
Doing so increased the validity of the comparison between media exposure in the present study
and Nielsen measurement.
References
Baym, N. K. (2015). Personal connections in the digital age (2nd ed.). Polity Press.
Baym, N. K., & Ledbetter, A. M. (2009). Tunes that bind? Information, Communication & Society,
12, 408–427. https://doi.org/10.1080/13691180802635430
Becker, J. A. H., Johnson, A. J., Craig, E. A., Gilchrist, E. S., Haigh, M. M., & Lane, L. T. (2009).
Friendships are flexible, not fragile: Turning points in geographically-close and long-distance
friendships. Journal of Social and Personal Relationships, 26, 347–369. https://doi.org/10.
1177/0265407509344310
Bond, B. J. (2018). Parasocial relationships with media personae: Why they matter and how they
differ among heterosexual, lesbian, gay, and bisexual adolescents. Media Psychology, 21,
457–485. https://doi.org/10.1080/15213269.2017.1416295
Burgoon, J. K., & Jones, S. B. (1976). Toward a theory of personal space expectations and their
violations. Human Communication Research, 2, 131–146. https://doi.org/10.1111/j.1468-2958.
1976.tb00706.x
20 Journal of Social and Personal Relationships XX(X)
Johnson, A. J., Becker, J. A. H., Craig, E. A., Gilchrist, E. S., & Haigh, M. M. (2009). Changes in
friendship commitment: Comparing geographically close and long-distance young-adult
friendships. Communication Quarterly, 57, 395–415. https://doi.org/10.1080/014633709033
13430
Johnson, A. J., Wittenberg, E., Villagran, M. M., Mazur, M., & Villagran, P. (2003). Relational
progression as a dialectic: Examining turning points in communication among friends. Com-
munication Monographs, 70, 230–249. https://doi.org/10.1080/0363775032000167415
Klimmt, C., Hartmann, T., & Schramm, H. (2006). Parasocial interactions and relationships. In J.
Bryant & P. Vorderer (Eds.), Psychology of entertainment (pp. 291–313). Lawrence Erlbaum.
Korn, C. J., & Nicotera, A. M. (1993). Friend and mate relationship literature, empirical proposi-
tions, and methodology. In A. M. Nicotera (Ed.), Interpersonal communication in friend and
mate relationships (pp. 13–42). SUNY Press.
Ledbetter, A. M. (2009). Patterns of media use and multiplexity: Associations with sex, geographic
distance and friendship interdependence. New Media & Society, 11, 1187–1208. https://doi.org/
10.1177/1461444809342057
Ledbetter, A. M., & Meisner, C. (2021). Extending the personal branding affordances typology to
parasocial interaction with public figures on social media: Social presence and media multi-
plexity as mediators. Computers in Human Behavior, 115, 106610. https://doi.org/10.1016/j.
chb.2020.106610
Madison, T. P., Porter, L. V., & Greule, A. (2016). Parasocial compensation hypothesis: Predictors
of using parasocial relationships to compensate for real-life interactions. Imagination, Cogni-
tion, and Personality, 35, 258–279. https://doi.org/10.1177/0276236615595232
Mallinckrodt, B., & Wei, M. (2005). Attachment, social competencies, social support, and psy-
chological distress. Journal of Counseling Psychology, 52, 358–367. https://doi.org/10.1037/
0022-0167.52.3.358
Nielsen (2020). Total audience report: Q4 2019 [white paper]. Author.
O’Brien, C. (2020, March 26). Houseparty, discore, zoom, and Marco polo downloads explode
amid coronavirus pandemic. https://venturebeat.com/2020/03/26/
Ramirez, A., & Wang, Z. (2008). When online meets offline: An expectancy violations theory
perspective on modality switching. Journal of Communication, 58, 20–39. https://doi.org/10.
1111/j.1460-2466.2007.00372.x
Reeves, B., & Nass, C. (1996). The media equation: How people treat computers, television, and
new media like real people and places. Cambridge University Press.
Rosaen, S. F., & Dibble, J. L. (2016). Clarifying the role of attachment and social compensation on
parasocial relationships with television characters. Communication Studies, 67, 147–162.
https://doi.org/10.1080/10510974.2015.1121898
Rubin, A. M., Perse, E. M., & Powell, R. A. (1985). Loneliness, parasocial interaction, and local
television news viewing. Human Communication Research, 12, 155–180. https://doi.org/10.
1111/j.1468-2958.1985.tb00071.x
Schiappa, E., Allen, M., & Gregg, P. B. (2007). Parasocial relationships and television: A
meta-analysis of the effects. In R. W. Preiss, N. Burrell, M. Allen, & B. M. Gayle (Eds.), Mass
media effects research: Advances through meta-analysis (pp. 301–314). Lawrence Erlbaum.
Taylor, S., Landry, C. A., Paluszek, M. M., Fergus, T. A., McKay, D., & Asmundson, G. J. (2020).
COVID stress syndrome: Concept, structure, and correlates. Depression & Anxiety, 37,
706–714. https://doi.org/10.1002/da.23071
22 Journal of Social and Personal Relationships XX(X)
Twenge, J. M., Zhang, L., Catanese, K. R., Dolan-Pascoe, B., Lyche, L. F., & Baumeister, R. F.
(2007). Replenishing connectedness: Reminders of social activity reduce aggression after
social exclusion. British Journal of Social Psychology, 46, 205–224. https://doi.org/10.1348/
014466605X90793
Vangelisti, A. L., & Caughlin, J. P. (1997). Revealing family secrets: The influence of topic,
function, and relationships. Journal of Social and Personal Relationships, 14, 679–705.
https://doi.org/10.1177/0265407597145006
Vogt, D. S., Pless, A. P., King, L. A., & King, D. W. (2005). Deployment stressors, gender, and
mental health outcomes among Gulf War I veterans. Journal of Traumatic Stress, 18, 272–284.
https://doi.org/10.1002/jts.20044
Wang, Q., Fink, E. L., & Cai, D. A. (2008). Loneliness, gender, and parasocial interaction: A uses
and gratifications approach. Communication Quarterly, 56, 87–109. https://doi.org/10.1080/
01463370701839057