Content uploaded by Nguyen Ha
Author content
All content in this area was uploaded by Nguyen Ha on May 11, 2021
Content may be subject to copyright.
Accepted Manuscript
#StayHome #WithMe: How Do YouTubers Help with COVID-19
Loneliness?
Shuo Niu
Ava Bartolome∗
Cat Mai∗
Nguyen B. Ha∗
{shniu,abartolome,cmai,joha}@clarku.edu
Clark University
Worcester, MA, USA
ABSTRACT
Loneliness threatens public mental wellbeing during COVID-19. In
response, YouTube creators participated in the #StayHome #WithMe
movement (SHWM) and made myriad videos for people experienc-
ing loneliness or boredom at home. User-shared videos generate
parasocial attachment and virtual connectedness. However, there
is limited knowledge of how creators contributed videos during dis-
asters to provide social provisions as disaster-relief. Grounded on
Weiss’s loneliness theory, this work analyzed 1488 SHWM videos
to examine video sharing as a pathway to social provisions. Find-
ings suggested that skill and knowledge sharing, entertaining arts,
homelife activities, live chatting, and gameplay were the most pop-
ular video styles. YouTubers utilized parasocial relationships to
form a space for staying away from the disaster. SHWM YouTu-
bers provided friend-like, mentor-like, and family-like provisions
through videos in dierent styles. Family-like provisions led to the
highest overall viewer engagement. Based on the ndings, design
implications for supporting viewers’ mental wellbeing in disasters
are discussed.
CCS CONCEPTS
•Human-centered computing →Empirical studies in collab-
orative and social computing.
KEYWORDS
YouTube, video sharing, parasocial, social provisions, disaster, lone-
liness
ACM Reference Format:
Shuo Niu, Ava Bartolome, Cat Mai, and Nguyen B. Ha. 2021. #StayHome
#WithMe: How Do YouTubers Help with COVID-19 Loneliness?. In CHI
Conference on Human Factors in Computing Systems (CHI ’21), May 8–13,
2021, Yokohama, Japan. ACM, New York, NY, USA, 15 pages. https://doi.org/
10.1145/3411764.3445397
∗Student authors contributed equally to this research.
Permission to make digital or hard copies of all or part of this work for personal or
classroom use is granted without fee provided that copies are not made or distributed
for prot or commercial advantage and that copies bear this notice and the full citation
on the rst page. Copyrights for components of this work owned by others than the
author(s) must be honored. Abstracting with credit is permitted. To copy otherwise, or
republish, to post on servers or to redistribute to lists, requires prior specic permission
and/or a fee. Request permissions from permissions@acm.org.
CHI ’21, May 8–13, 2021, Yokohama, Japan
©2021 Copyright held by the owner/author(s). Publication rights licensed to ACM.
ACM ISBN 978-1-4503-8096-6/21/05. . . $15.00
https://doi.org/10.1145/3411764.3445397
1 INTRODUCTION
In 2020, the COVID-19 pandemic swept the globe and forced billions
of people to stay home for months. Social media became an alter-
native venue for people to stay connected and cope with loneliness
during this global crisis. Loneliness is a signicant public mental-
health issue during COVID-19 [
22
,
45
]. YouTube, as the largest
video sharing platform, called on YouTube creators (YouTubers)
to join the #StayHome #WithMe movement (SHWM) by creating
“content that will entertain, inform, and connect with people who
are social distancing during this pandemic”
1
. This hashtag move-
ment attracted a lot of video artists and creators, which witnessed
a 600% increase in viewership
2
. There is increasing attention in
the literature on using social media to support public safety and
wellbeing during disasters [
27
]. In HCI, researchers have explored
the aordances of Twitter [
61
], Facebook [
5
,
11
], and Reddit [
23
] in
supporting disaster awareness and communication. However, there
is limited understanding of how YouTubers and video-sharing plat-
forms can uniquely support public mental health during a long-term
crisis. #StayHome #WithMe is a YouTubers’ voluntary response
to help people who might feel lonely or bored during COVID-19.
This trending movement provides a valuable and unique lens to
examine the social connections that YouTubers oered to mitigate
loneliness during social distancing.
Video-sharing platforms such as YouTube, Twitch, TikTok, and
Facebook Videos saw a rapid increase in popularity in the past
decade. In 2019, YouTube was the second most popular social me-
dia, and 73% of adults have visited this video sharing platform [
47
].
The participatory culture on YouTube encourages grassroots to
participate in video creation. In contrast to platforms like Twitter
and Facebook, social interactions on YouTube are based on video
itself rather than personal proles and friending [
7
]. YouTubers self-
construct their value by contributing unique videos and actively
interacting with fans [
28
]. The regular postings of YouTubers make
the audience form a one-sided sense of closeness to the YouTu-
bers. Prior studies described the audience’s social and emotional
attachment to a video persona as “parasocial relationship” [
65
].
The interactions with videos to generate parasocial relationship are
called “parasocial interactions” [
24
]. When other social interactions
are diminished during social distancing, SHWM can be seen as an
1https://services.google.com/fh/les/misc/stay_home_with_me.pdf
2https://www.theverge.com/2020/3/27/21197642/youtube-with-me-style-videos-
views-coronavirus-cook-workout-study-home-beauty
arXiv:2101.03706v3 [cs.HC] 13 Jan 2021
Accepted Manuscript
CHI ’21, May 8–13, 2021, Yokohama, Japan Niu, Bartolome, Mai, and Ha
eort to supplement social connections through parasocial rela-
tionships. However, there is limited understanding of YouTubers’
roles and YouTube’s aordances in oering social connections and
mitigating disaster loneliness. Considering YouTube’s massive cre-
ator base and high popularity among young people [
47
], the HCI
community needs to address this knowledge gap to better design
applications and services to support disaster mental health. This
work explores this new social media phenomenon by collecting
and analyzing SHWM video data. Grounded on Weiss’s loneliness
theory, this work examines what loneliness-supporting videos were
created, how they sought to oer social provisions, and whether dif-
ferent social provisions aected viewer engagement. An overview
of SHWM can provide new perspectives on YouTube’s roles in dis-
asters and inform social media platforms’ socio-technical design
for tackling loneliness.
8023 SHWM videos published between March 11th and May 15th
were crawled from YouTube Data API, among which 1642 made by
creators in the United States were annotated by Amazon Mechanical
Turk (MTurk) participants. These videos came from 695 video mak-
ers and attracted 206,319,247 views. The analysis of SHWM videos
was guided by Weiss’s framework of social-emotional loneliness
[
54
]. Weiss’s theory conceptualizes six core types of social provi-
sions that people need to refrain loneliness (Table 1) – attachment,
social integration,reassurance of worth,reliable alliance,guidance,
and opportunity for nurturance. Weiss’ model incorporates the major
elements of social relationships people need from families, friends,
and mentors [
14
], which are also the connections that are likely to
be absent during social distancing. YouTubers approach viewers
by sharing original content and arousing viewers to parasocially
interact. Viewers might generate feelings of connection and inti-
macy to the creators [
39
,
50
] and interact with YouTubers through
liking and commenting. Examining the styles of SHWM videos and
analyzing what social provisions the videos sought to oer will
deepen the understanding of YouTube’s social functionalities in
disasters and the categories of virtual connectedness provided by
YouTubers. Grounded on Weiss’s framework, this paper addresses
three key research questions:
•
RQ1. What #StayHome #WithMe videos did YouTubers make
and how they relate to the COVID-19 pandemic?
•
RQ2. How did #StayHome #WithMe videos in dierent styles
and mentioning COVID-19 in dierent degrees aect the
social provisions in the video?
•
RQ3. How did videos with dierent social provisions aect
viewer engagement?
RQ1 is an initial exploration of SHWM video styles and if YouTubers
chose to communicate the ongoing COVID-19 pandemic. Rooted
in Weiss’s loneliness theory, RQ2 explores whether and how the
six social provisions were associated with videos in dierent styles
and dierent COVID-19 mentioning. Following the understanding
of video styles and social provisions, RQ3 further explores whether
oering various social provisions aected video popularity, viewers’
activeness, and viewers’ emotional expression in the comments.
These metrics reect the parasocial interactions with YouTubers.
The authors found that SHWM videos primarily themed in: how-to
videos of sharing skills and knowledge; entertaining content of
music, arts, and performance; videos of homelife activities; chat-
ting with the audience; and videos of gameplay. In contrast to
other social media platforms where disaster information is inten-
sively communicated and spread, these videos formed an online
space where the disaster is not actively mentioned. The analysis
of six social provisions in SHWM explained how parasocial re-
lationships supplement social connections to mitigate loneliness.
Most SHWM videos oered social integration by sharing interests
and recreational activities. A large number of how-to videos pri-
marily supported the guidance provision. Videos of homelife and
chatting supported the provisions of attachment, nurturance, and
alliance. Providing family-like social provisions had better overall
viewer engagement with the video, despite their smaller propor-
tions in SHWM. This work’s ndings foster new possibilities to
design video-sharing-based applications and technologies to ad-
dress disaster-related loneliness.
2 RELATED WORK
2.1 Social Media and Disasters
Much research has explored the use of dierent social media dur-
ing disasters. Houston et al. reviewed prior works and concluded
key aordances of social media include signaling and detecting
disasters, sending and receiving requests for help, informing con-
ditions, providing mental/behavioral support, etc. [
27
]. Lindsay
summarized that social media such as Twitter and Facebook are
primarily used to disseminate and receive disaster information and
serve as an emergency management tool [
37
]. Studies examined
the unique aordances of dierent social media in emergencies
and disasters. For example, Twitter is a platform for sharing short
and immediate disaster messages to raise public awareness of the
critical situation [
42
,
61
]. Facebook users seek ocial information
and organize disaster-supporting communities [
5
,
11
,
55
,
56
]. Red-
dit users perceive and speculate risks in the long run, or dierent
regions [
23
]. YouTube videos educate the public as well as spread
misinformation [
4
,
44
]. Disasters are stressful events and can cause
mental health problems for both people who are directly aected
and the community at large [
53
]. Besides spreading disaster in-
formation, social media is also an outlet for expressing emotions
[
32
,
49
] and receiving mental health service [
27
]. Therefore studies
have looked into technologies to measure public mental health
by mining social media data [
3
,
15
,
32
]. Recent studies suggested
that COVID social distancing caused loneliness and other mental
health challenges to many who stayed at home [
36
,
45
], especially
the young adults [
16
,
22
]. However, the use of video-sharing for
supporting mental health during disasters is under-explored [
47
].
There is limited understanding of how video creators contribute
to a social media movement like #StayHome #WithMe to support
disaster mental health and what aordances of the video-sharing
platform and community can reduce disaster loneliness.
2.2 YouTube and Parasocial Relationship
In HCI and CSCW, understanding the unique characteristics and
communication aordances of dierent platforms centers the re-
search on social media [
48
,
57
]. After years of growth, video-sharing
platforms like YouTube have developed their characteristics of plat-
form cultures. Dijck noted that on YouTube, user-generated videos
Accepted Manuscript
How Do YouTubers Help with CO VID-19 Loneliness? CHI ’21, May 8–13, 2021, Yokohama, Japan
boost online production and distribute diverse content [
59
]. In
contrast to platforms based on friending and networking, social
interactions on YouTube rely on the video itself rather than of-
ine relationships [
7
]. YouTubers interact with viewers through
sharing new videos to cultivate relationships with fans [
26
,
28
].
Horton and Wohl dened the one-sided intimacy generated by
the “conversational give-and-take” with performers as parasocial
relationship [
25
]. Video interactions that lead to parasocial relation-
ships are called parasocial interactions [
24
,
25
]. Gardner found that
people turn to parasocial relationships with a media gure when
they need to regulate the needs of social belongingness [
19
]. Hart-
mann concluded that parasocial relationships could provide social
support and shield against the eects of exclusion and loneliness
[
24
]. Rotman et al. suggested YouTube users mostly focused on
the parasocial interactions with the creator, rather than building a
friend and community network like other social media [
52
]. Wohn
examined live streamers and found parasocial relationships corre-
late with viewers’ emotional, instrumental, and nancial support
for the performers [
65
]. Anjani et al. studied YouTube food-eating
videos and suggested that they generate a sense of connectedness
[
2
]. The parasocial interactions on YouTube lie in that YouTubers
make original content to engage viewers, and viewers respond to
and endorse YouTubers by video-watching, liking, and comment-
ing [
7
,
26
,
28
,
33
]. #StayHome #WithMe is a movement in which
YouTubers respond to the pervasive loneliness and use parasocial
relationships to oer loneliness support. However, little is known
about how parasocial relationships are embodied during an ongoing
pandemic and their aordances to supplement social interactions.
This work seeks to give a deeper understanding of the phenome-
non of video-sharing for mitigating disaster loneliness. Considering
video sharing and video interaction center the social activities on
YouTube, it is essential to examine what SHWM videos did YouTu-
bers provide, how dierent videos oer parasocial relationships to
supplement social connections, and whether dierent social con-
nections aect viewer interactions with the video.
2.3 Social Provisions of #StayHome #WithMe
Loneliness is prevalent among people who experienced disasters
and crises [
30
,
38
,
45
]. During COVID-19, technologies play a vi-
tal role in oering social support and helping people deal with
loneliness and isolation [
36
]. This work utilizes Weiss’s theory of
loneliness [
64
] to examine the role of SHWM videos in oering
social provisions during COVID-19. Weiss conceptualized social
and emotional loneliness and argued that people need six social
provisions to deal with loneliness [
54
]; see Table 1 for denitions,
which are elaborated in section 3.2. Cutrona and Russell examined
Weiss’s loneliness theory in psychological practice and identied
the sources of social provisions[
14
]. Attachment, reliable alliance,
and opportunity for nurturance demonstrate intimacy or trust and
are usually provided by family members [
14
]. Social integration
is usually obtained from friend relationships [
14
]. Guidance is ob-
tained from teachers, mentors, or parent gures [
14
]. Reassurance
of worth is a type of self-ecacy and self-esteem obtained by help-
ing others and receiving acknowledgment [14].
Weiss’s loneliness framework has been used to study the social
connections people can obtain from social media. The six social
Table 1: Weiss’s Framework of Social Provisions for Lone-
liness
Concept Denition Source
attachment
A relationship in which people receives
a sense of safety and security
family
social
integration
A network of relationships in which in-
dividuals share interests, concerns, and
recreational activities
friend
reassurance
of worth
A relationships in which the person’s
skills and abilities are acknowledged
self
reliable alliance
A relationship in which one can count
on assistance under any circumstances
family
guidance
A relationship with trustworthy and au-
thoritative individuals who can provide
advice and assistance
mentor
opportunity
for nurturance
A relationship in which the person feels
responsible for the wellbeing of another
family
provisions were examined in the loneliness-support functions of
Twitter [
21
] and Facebook [
62
]. It was also used as a framework to
study how parasocial relationships aect loneliness [
29
,
63
]. Weiss’s
framework described the necessary relationships people need in
the context of loneliness [
14
], which are also the social supports
that are likely to be absent during social distancing. YouTubers
perform various kinds of aective relationships and serve as micro-
celebrities to oer accessibility, authenticity, and connectedness to
the audience [
39
,
51
]. Viewers experience YouTubers’ characters as
close friends or online family members to fulll the need for social
interaction [
33
,
50
]. The parasocial relationships can be a source
of social connections for people who need to improve emotional,
cognitive, and behavioral health [
17
]. By examining Weiss’s six
provisions in SHWM videos, this work oers an overview of how
YouTubers construct the parasocial relationships to supplement
social provisions during social distancing. This understanding is
essential to guide future research on viewers’ perception of social
relationships with YouTubers and the psychological and aective
aordances of YouTube in reducing disaster loneliness.
During COVID-19, viewers may choose to watch SHWM videos
to obtain insucient social provisions in real life to avoid loneliness
[
20
]. This work uses engagement metrics to measure the viewers’
participation in parasocial interactions. Prior studies suggested
that participating in parasocial interactions can promote social
connectedness and mitigate loneliness [
19
,
24
]. Khan noted that
viewers who want to socialize on YouTube were more likely to
like/dislike and comment on the videos [
33
]. Rasmussen found that
viewer interactions such as commenting and sending messages to
the YouTubers can simulate realistic social interactions [
50
]. Wright
et al. noticed that video-based platforms outperformed message-
based media in promoting mental wellbeing [
66
]. Similarly, on other
social media, commenting and liking others’ social media posts are
positive indicators of mental wellbeing [
46
,
60
]. Studies showed
that discussing common interests on social media helped users
cope with loneliness [
18
] and promoted comfort feelings [
43
]. This
analysis provides a preliminary understanding of how six social
provisions aect engagement in parasocial interactions.
Accepted Manuscript
CHI ’21, May 8–13, 2021, Yokohama, Japan Niu, Bartolome, Mai, and Ha
3 STUDY DESIGN
Grounded in prior research, this work uses SHWM as a lens to exam-
ine the social aordances of YouTube and YouTuber communities
in supporting social connections and mitigating disaster loneliness.
Based on Weiss’s framework, the three research questions examine
how SHWM videos provide social provisions and whether they
aect viewer engagement. This section describes the structure of
the study and the factors examined in the data analysis (Figure 1).
Figure 1: The analysis framework of the three research ques-
tions
3.1 Video Styles and COVID-19 Mentioning
Research on social media interaction needs to capture user be-
haviors and participation styles [31]. Addressing RQ1 provides an
overview of how SHWM YouTubers crafted parasocial relationships
for viewers who need to avoid or mitigate loneliness. Considering
the diversity of #StayHome #WithMe participation, RQ1 examines
SHWM videos by categorizing video styles and rating the degrees
of COVID-19 mentioning. Video styles are the major themes and
production styles of SHWM videos, including the video’s topics
and the main activities YouTubers performed. The categorization of
SHWM video styles consisted of two steps. YouTube allows creators
to specify their video content using the default YouTube categories,
including “music”, “entertainment”, “gaming”, etc. As an initial step,
the category distribution was compared between SHWM videos and
the overall YouTube category distribution in 2015 [
12
]. This step
provided a general understanding of what video categories trended
in SHWM. In the second step, considering that the YouTube de-
fault categories lack coverage of topics such as homelife or chatting
about the pandemic, grounded-theory methods were performed to
derive video styles. The rating of COVID-19 mentioning pertained
to the pandemic context of SHWM videos. Examining the levels to
which COVID-19 was mentioned can unveil if YouTubers shared
disaster-related information like other social media [
27
,
37
,
56
,
61
].
These two factors allow the authors to infer how parasocial rela-
tionships were delivered during COVID-19 and whether they were
aected by the disaster.
3.2 Social Provisions
Following the categorization of SHWM videos, RQ2 measures whe-
ther and how SHWM videos provided dierent social provisions.
This work adopts Weiss’s loneliness theory as the theoretical frame-
work. Weiss’s theory depicts the categories of roles YouTubers can
play via parasocial relationships in providing needed social pro-
visions during social distancing. In Weiss’s theory, attachment is
a social provision for a sense of safety and security. In SHWM,
creators may give the viewers a sense of emotional security and
closeness by showing intimate content such as activities at home
or a face-to-face chat. Social integration is a provision to share in-
terests and concerns. By sharing videos of hobbies and interests,
YouTubers can entertain the viewers and provide social integra-
tion. Reassurance of worth emphasizes one’s skills and abilities
are acknowledged. YouTube has a culture of developing video cre-
ation skills and growing the subscriber community [
7
]. By sharing
SHWM videos, YouTubers demonstrate their unique talents and
be acknowledged by the viewers and gain reassurance. Reliable
alliance is a provision which the person can count on under any
circumstances. To help people who are staying at home, YouTubers
may express a willingness to help at any time. Guidance is a provi-
sion through which people obtain directions and advice. YouTube is
known for communities for skill sharing and learning [
10
]. SHWM
videos can provide instructions, guidelines, or advice during the
pandemic. Opportunity for nurturance is a provision in which the
person feels responsible for another’s wellbeing. During COVID-
19, YouTubers may communicate with their viewers to show their
concern for their health and wellbeing.
The annotation of social provisions in SHWM videos was through
crowd-sourcing on Mechanical Turk
3
. The six social provisions of
SHWM videos were rated by participants who watched SHWM
videos. The annotation generated six binary variables for each video
to represent whether the video provides each of the six social provi-
sions. Logistic regression models were built with the video style and
COVID-mentioning as independent variables and the six social pro-
vision variables as dependent factors. The predictive models reveal
how videos in dierent styles and mentioning COVID-19 in dier-
ent degrees aect the video’s social provisions. How dierent video
styles mention COVID-19 was also compared by the Wilcoxon test
(posthoc uses Dunn’s test with Bonferroni adjustment).
3.3 Viewer Engagement
Parasocial relationships could help people shield against loneliness
[
19
,
20
,
24
]. Interacting with the videos can simulate more realistic
social connections to the YouTubers [
50
]. Video watching, liking,
and commenting reect viewers’ participation in the parasocial
interactions on YouTube, which may mitigate the eects of loneli-
ness [
24
,
33
]. Interactions such as commenting and liking on other
social media are also considered positive indicators of mental well-
being [
46
,
60
]. RQ1 and RQ2 explore what videos were created for
SHWM and how they oered social provisions for people in social
distancing. RQ3 seeks to capture each social provision’s eects
on the interactions with SHWM videos and their potential eects
on loneliness by measuring viewer engagement metrics. User en-
gagement in online services is dened as “the quality of the user
experience” that motivates people to interact [
35
]. Lehmann et al.
identied popularity,activity, and loyalty as three key metrics to
measure users’ engagement [
35
]. On YouTube, these measurements
indicate viewers’ participation and behaviors in the parasocial in-
teractions to fulll the need for social interaction [
33
]. A video’s
popularity can be reected by the number of views, likes, and com-
ments it received. Popularity measurements reect if providing a
social provision allowed the YouTubers to reach and support more
viewers. Activity is the average viewer’s activeness in the interac-
tion with the video and the YouTuber. The frequencies of likes and
3https://www.mturk.com/
Accepted Manuscript
How Do YouTubers Help with CO VID-19 Loneliness? CHI ’21, May 8–13, 2021, Yokohama, Japan
comments a video received per 100 views were collected to measure
this dimension. Activity measurements indicate viewers’ activeness
in the parasocial interactions. Loyalty metrics are how often the
users return to the social media site. Since it is dicult to collect
viewers’ watching history from YouTube, this work leveraged user
comments’ sentiment as an alternative factor of loyalty. NRC Word-
Emotion Association Lexicon [
40
] was used to count emotional
words in viewer comments. Comment emotions evaluate viewers’
emotional connection to the YouTuber’s content. Although viewer
engagement metrics do not directly measure how much loneliness
SHWM videos can mitigate, they reect the degree to which view-
ers were attracted by the videos and willing to participate in the
social interactions on YouTube [
33
]. Ordinary least squares models
(OLS) were built to measure whether videos with dierent social
provisions resulted in dierent viewer engagement. The model used
the six provision variables (dummy variables) to predict popular-
ity, activity, and comment emotion measurements. However, it is
widely recognized that the number of followers a creator has is a
decisive factor for their content popularity [
6
,
9
]. Viewer engage-
ment is signicantly correlated with subscriber count. Therefore
the multivariate models included subscriber count to investigate
whether the six social provisions added an extra layer of eects on
viewer engagement besides creators’ popularity.
4 DATA AND ANALYSIS METHODS
The SHWM video data was collected with YouTube Data API
4
between Jun 3 and Jun 5, 2020. The data crawling included videos
with publishing dates between Mar 11 (the declaration of state
emergency) and May 15 (the rst round of reopening in most U.S.
states). The authors chose to collect the data at least 20 days after
the video was published because it left enough time for the newer
videos to get views [
9
]. Videos published in countries other than the
U.S. were not considered because of dierent COVID-19 quarantine
time-frame and diculties in analyzing non-English videos. The
keyword “#StayHome #WithMe” was used to retrieve an initial list
of videos between the start and end dates. Video metadata includes
the title, description, duration, publishing date, view count, like
and dislike count, comment count, subscriber count, and the rst
200 available comments. The initial crawling returned 6375 videos
(including videos of all countries and all languages). After raw
metadata processing, videos without channel information, shorter
than 5 seconds, or being viewed less than 100 times were eliminated.
The sanity check also removed private videos, deleted videos, and
videos with broken links. After the cleaning, 4733 videos were
excluded from the dataset. The remaining 1642 videos constituted
the dataset for data encoding. These videos were from 695 dierent
YouTubers, viewed 206,319,247 times, and attracted 3,301,412 likes
and 310,879 comments. The median duration of the videos was
10.73 minutes (𝑚𝑒𝑎𝑛 =28.70,𝑆𝐷 =65.28).
4.1 Categorizing Video Styles
Grounded theory approach [
8
] was applied to identify video styles.
In the open coding stage, one author watched 100 randomly selected
SHWM videos and generated 100 notes. The author specically
inspected what activities were done and the videos’ production style
4https://developers.google.com/youtube/v3
(e.g., livestream, animation, photos/memes). For example, in a video
of “cooking with me”, the author noted “a video to teach cooking, the
creator talked about cooking steps, and the video shared the cooking
hobby”. After that, the notes were summarized into emerging video
styles through anity diagramming. Open encoding generated nine
video styles. A discriminant analysis was conducted through closed-
encoding to validate the video styles’ accuracy in representing
SHWM videos. Four authors used the nine styles to tag another 300
videos, with each video tagged by two authors. Tagging results were
then compared to identify discrepancies. When an inconsistency
was identied, revisions to the style denitions were made. After
the discussion, the authors nalized a reliable categorization of nine
video styles for crowd-sourcing annotation. Researchers reached a
substantial agreement on the video styles during the discriminant
analysis (
𝐹𝑙 𝑒𝑖𝑠𝑠 ′𝑘𝑎𝑝 𝑝𝑎 =
0
.
725,
𝑝<
0
.
001). The resulted styles are
presented in Table 2 and example videos can be found in Figure 2.
Table 2: Nine video styles identied from the grounded the-
ory analysis.
Style Description
artistic
A video of music, art (drawing/painting/carving/etc.), performance,
or animation
challenge
A video showing exciting activities or participating in a unique
challenge to grab attention (e.g. stunts, self-challenges)
chatting
A video or livestream checking in with or interacting with the au-
dience (e.g. chatting, reading viewer comments, or doing activities)
game A gameplay or a recorded livestream game video
homelife A video or vlog showing family, homelife, or lifestyle
how-to
A how-to video or knowledge video that gives step-by-step guide-
lines to complete a task or explains a specic subject to the audience
(e.g. cooking, learn language, safety tips)
religious A religious video showing prayers or praying
review A video reviewing or recommending products or services
story
A video sharing an interesting story, pictures (i.e. photos, memes),
or video clip(s)
4.2 Annotating Video Styles, COVID-19
Mentioning, and Social Provisions
782 participants from MTurk were recruited to complete the annota-
tion task. All participants were from the U.S. and have completed at
least 5000 tasks with an accuracy of 97% or higher. Each participant
was asked to watch a video for at least three minutes. The task
consists of three questions that annotate video styles, COVID-19
mentioning levels, and whether it provides each of the six social
provisions, plus one quality check question to estimate if the par-
ticipants watched the video carefully. Q1 asked the participants
to pick one of the nine video styles (or “none of the above”) that
best describes the video content (Table 2). Q2 was a quality-control
question that asked the participants to select the YouTube video
category from three randomly generated options. Q3 asked the par-
ticipants to pick all applied social provisions, with an extra choice
of “none of the above.” The provision descriptions were rephrased
for easy understanding (see Table 3 for Q3 options). Participants
were asked to rate the video by considering someone who regularly
views this kind of video. Q4 was a question to annotate the degree
to which the video mentions COVID-19, coronavirus, or pandemic
by selecting one of “none”, “low”, “medium”, and “high”. Partici-
pants were asked to rate based on their perception of whether the
Accepted Manuscript
CHI ’21, May 8–13, 2021, Yokohama, Japan Niu, Bartolome, Mai, and Ha
(a) True Colors - Piano Covers & Chill
#StayHome #WithMe #RolandAtH-
ome
(b) INSTAGRAM DAREScritique d by
Jordan Matter!!! EPIC #dares #stay-
home #withme
(c) Blue October - #Stayhome and
hangout #withme
(d) #stayhome #withme 7 minute
Candy Crush Saga - themudan
(e) Dinner with the Gagans March
27th 2020 - Jim Gagan #stayhome
#withme
(f) Lemon Pepper Wings Recipe - ... -
#StayHome and Cook #WithMe
(g) Easter Online 2020 #StayHome
and go to church #WithMe
(h) Full Face of Makeup Brands
I’ve Never Tried Before! #stayhome
#withme
(i) Try Not To Laugh Or Smile While
Watching Couple BAT TLE Ep. # 150
Figure 2: Example videos in 9 video styles: (a) artistic, (b)
challenge, (c) chatting, (d) game, (e) homelife, (f) how-to, (g)
religious, (h) review, and (i) story.
video discussed COVID-19. No participants tagged more than 1/10
of all 1642 videos. To ensure the validity of the data annotation, the
authors ltered out and republished tasks that either not watched
enough time as required, with unanswered questions, with con-
icting answers (e.g. “none of above” and one above option are
checked at the same time), or with apparently wrong answers for
Q2. The task was performed in two rounds to ensure agreement
on the choices. In the rst round, each video was tagged by three
dierent participants. A task was considered to reach an agreement
if at least two participants picked the same video style or “none
of the above”. If a video has three dierent style answers, it was
tagged by two additional participants. Then the video style was de-
termined by the one that was selected by three participants. Videos
without agreement were also identied. The average rating of Q4
was used as the value of COVID-19 mentioning (0 for “none” and 3
for “high”). The variables for six provisions were set to 1 if more
than half of the participants picked that provision in Q3; otherwise,
they were set to 0. The agreed results of Q1, Q3, and Q4 were saved
as variable values to perform multivariate analysis (see Table 4).
Table 3: Annotations of social provisions in Q3
Social
Provisions Provision Coding
attachment
The video creator gives a sense of emotional security and
closeness to the audience.
integration
The video creator shares common or specialized hobbies and
interests, or shares entertaining content, activities, or experi-
ences with the audience.
reassurance
The video creator demonstrates special skills and abilities in
hopes to be acknowledged, or acknowledges the audience’s
thoughts and comments.
alliance
The video creator expresses a willingness to help anytime,
i.e. being available to help with the audience’s problems or
diculties.
guidance
The video creator gives step-by-step instructions, guidelines,
or advice on a subject that they are knowledgeable in.
nurturance
The video creator feels responsible for and interested in the
audience’s wellbeing.
4.3 Measuring Viewer Engagement
This work considers three aspects of viewer engagement to probe
how SHWM viewers parasocially interact with YouTubers: popular-
ity, activity, and comment emotion [
35
]. The number of views, likes,
and comments constitute the measurements of popularity. View
count (
𝑣𝑖𝑒𝑤
), like count (
𝑙𝑖𝑘 𝑒
), and comment count (
𝑐𝑜𝑚𝑚𝑒𝑛𝑡
) are
common measurements to assess to what degree the video reached
viewers [
9
]. Activity is measured by how many likes and comments
a video got for every 100 views – like rate (
𝑙𝑖𝑘 𝑒_𝑟𝑎𝑡𝑒
) and comment
rate (
𝑐𝑜𝑚𝑚𝑒𝑛𝑡 _𝑟𝑎𝑡 𝑒
). The like rate is the dierence between like
and dislike count per 100 video views, calculated using Eq. 1. The
comment rate is the number of comments a video received per
100 video views (Eq. 2). Those two factors reect the activeness of
parasocial interactions with the YouTubers. To measure comment
emotions, emotional words in NRC Word-Emotion Association
Lexicon [
40
] was used to count how many positive and negative
emotional words were used in the viewer comments. Positive (or
negative) emotion score of a comment is calculated by the total
number of positive (or negative) words in the collected comments
divided by the number of counted comments (
𝑝𝑜𝑠𝑖𝑡𝑖 𝑣𝑒 _𝑠𝑐𝑜𝑟𝑒
and
𝑛𝑒𝑔𝑎𝑡𝑖𝑣𝑒 _𝑠𝑐𝑜𝑟 𝑒
, Eq. 3) [
67
]. The frequencies of positive and nega-
tive words in comments reect whether viewers express favorable
or unfavorable sentiment towards the video, which implies the
positivity of parasocial interactions with the YouTuber.
𝑙𝑖𝑘 𝑒_𝑟𝑎𝑡𝑒 =
𝑙𝑖𝑘 𝑒𝑠 −𝑑𝑖𝑠𝑙𝑖𝑘𝑒𝑠
𝑣𝑖𝑒𝑤_𝑐𝑜𝑢𝑛𝑡 ∗100 (1)
𝑐𝑜𝑚𝑚𝑒𝑛𝑡 _𝑟𝑎𝑡 𝑒 =
𝑐𝑜𝑚𝑚𝑒𝑛𝑡 _𝑐𝑜𝑢𝑛𝑡
𝑣𝑖𝑒𝑤_𝑐𝑜𝑢𝑛𝑡 ∗100 (2)
𝑝𝑜𝑠 𝑖 (𝑛𝑒𝑔𝑎)𝑡𝑖𝑣𝑒 _𝑠𝑐𝑜𝑟 𝑒 =
𝑛𝑢𝑚𝑏𝑒𝑟 _𝑜 𝑓 _𝑝𝑜 𝑠𝑖 (𝑛𝑒𝑔𝑎)𝑡𝑖𝑣𝑒_𝑤𝑜𝑟𝑑𝑠
𝑡𝑜𝑡𝑎𝑙 _𝑐𝑜𝑢𝑛𝑡𝑒𝑑_𝑐𝑜𝑚𝑚𝑒𝑛𝑡𝑠 (3)
Accepted Manuscript
How Do YouTubers Help with CO VID-19 Loneliness? CHI ’21, May 8–13, 2021, Yokohama, Japan
Table 4: The measured factors and their respective variables used in the data analysis
Factor Variable Denition
Video
Creation
Video style 𝑠𝑡𝑦𝑙 𝑒
The video style in Table 2 that picked by more than half of the partici-
pants in Q1.
COVID-19 Mentioning 𝑐𝑜𝑣 Average rating of COVID-mentioning in Q4
Social
Provisions
Attachment 𝑎𝑡𝑡 𝑎𝑐ℎ𝑚𝑒𝑛𝑡
Value is set to 1 if more than half of the participants
checked the provision in Q3, otherwise 0.
Social Integration 𝑖𝑛𝑡𝑒𝑔𝑟 𝑎𝑡𝑖𝑜 𝑛
Reassurance of worth 𝑟𝑒𝑎𝑠 𝑠𝑢𝑟𝑎𝑛𝑐 𝑒
Reliable alliance 𝑎𝑙𝑙𝑖𝑎𝑛𝑐 𝑒
Guidance 𝑔𝑢𝑖𝑑𝑎𝑛𝑐𝑒
Opportunity for nurturance
𝑛𝑢𝑟𝑡𝑢 𝑟𝑎𝑛𝑐𝑒
Popularity
View count 𝑣𝑖𝑒𝑤 Number of times the video has been
watched/liked/commented
Comment count 𝑐𝑜𝑚𝑚𝑒𝑛𝑡
Like count 𝑙𝑖𝑘𝑒
Activity Like rate 𝑙𝑖𝑘𝑒 _𝑟𝑎𝑡 𝑒 Likes/comments a video received per 100 views
Comment rate 𝑐𝑜𝑚𝑚𝑒𝑛𝑡 _𝑟𝑎𝑡𝑒
Comment
Emotion
Positive emotion 𝑝𝑜𝑠𝑖 𝑡𝑖𝑣 𝑒_𝑠𝑐𝑜𝑟 𝑒 The frequency of positive/negative emotional words in
the comments
Negative emotion 𝑛𝑒𝑔𝑎𝑡𝑖𝑣𝑒 _𝑠𝑐𝑜𝑟 𝑒
5 FINDINGS
The participants spent an average of 5.42 minutes on the task (
𝑆𝐷 =
2
.
52). 14 were annotated as “none of the above” per consensus,
and 140 videos reached no agreement on the video style. Since
the videos that could not be categorized constituted less than 1%,
the authors decided to exclude them from the analysis. For the
140 (8.5%) no-agreement videos, the authors reviewed a portion
and posited that these videos have contents from multiple styles
(e.g., presenting ne arts and talking about steps or reviewing
products and showing family activities at the same time). Since
this study focuses on the social provisions of clearly discernible
video styles, the authors decided to exclude the 140 videos to avoid
confusion and uncertainty in the prediction models. For all included
videos, Fleiss’ kappa between the rst three crowd workers’ style
annotations was 0.566 (
𝑝<
0
.
001). To verify encoding reliability, the
authors randomly selected 50 videos annotated by the participants
and determined their styles. There was a substantial agreement
between the author annotated styles and the crowd-annotated video
styles (
𝐹𝑙 𝑒𝑖𝑠𝑠 ′𝑘𝑎𝑝 𝑝𝑎 =
0
.
727,
𝑝<
0
.
001). The lower agreement
between individual MTurk participants suggests that viewers may
be interested in dierent aspects of the same video. For example,
a yoga video at home might be annotated as homelife, but some
participants may be interested in how the YouTubers did it and
tagged it as a how-to. But annotating the video by the majority rule
allowed the authors to decide the closest style to describe the video.
5.1 RQ1: What are #StayHome #WithMe
Videos?
RQ1 probes what SHWM videos were created and how they relate
to the pandemic. The analysis consists of an initial comparison
with the overall YouTube video category distribution [
12
] and a
grounded-theory encoding of SHWM videos. The degree to which
each video mentions the pandemic is also rated to explore whether
SHWM videos were made for sharing COVID-19 information.
5.1.1 Video Styles of #StayHome #WithMe. The comparison re-
vealed that in contrast to the distribution in [
12
] (overall video
category distribution in 2015), SHWM had more videos in “En-
tertainment”, “Howto & Style”, “People & Blogs”, and “Education”
(Figure 3). There were 20.36% “Howto & Style” videos and 12.63%
“Education” videos in SHWM, in contrast to their 5.1% and 2.9%
overall proportions. There were 22.38% and 16.87% in the categories
of “Entertainment” and “People & Blogs”, higher than the 16.0% and
8.1% in the overall distribution. This result indicates that during the
COVID-19 quarantine, YouTubers used #StayHome #WithMe hash-
tags to entertain, share advice and tips with, and present personal
life and vlogs to their viewers to help them avoid or mitigate lone-
liness. The encoding of video styles reveals similar results: how-to,
artistic,homelife,chatting, and game were the ve most common
video styles (Figure 4). The style distribution suggests that the
most common ways for YouTubers to build parasocial relationships
and help with loneliness were showing step-oriented and knowl-
edge building videos, making entertaining videos by demonstrating
artistic or gaming skills, and showing a video of home activities or
chatting to the audience (See Table 5 for example videos).
Figure 3: Comparison of category distributions between
SHWM videos and 2015 YouTube video distribution (data
source: [12])
Figure 4: Number of videos in each video styles
Accepted Manuscript
CHI ’21, May 8–13, 2021, Yokohama, Japan Niu, Bartolome, Mai, and Ha
Title: Pro chefs make 9 kinds of pantry sandwiches | Test Kitchen Talks @ Home | Bon Appétit.
Content: 9 chefs show how to cook pantry sandwiches at home. The YouTubers cook while
explaining each step.
Provisions: guidance and integration
Title: Ariana Grande Performs ’I Won’t Say I’m In Love’ - The Disney Family Singalong
Content: A YouTuber sings a Disney song while acting funny postures. The song is about the
experience of falling in love.
Provisions: attachment and integration
Title: Corona Virus Do’s & Don’ts! #StayHome (FV Family Fun Facts Fundraiser to Laugh #withme)
Content: The YouTubers use exaggerated and funny at-home performance to encourage viewers
to wash hands, not to shake hands, and share foods wearing gloves. The YouTubes tell viewers to
stay safe.
Provisions: attachment, guidance, and nurturance
Title: Breaking Modern Loneliness: A Conversation on Mental Health
Content: Five speakers live a chat to share stories, talk about coping with diculties, and answer
questions.
Provisions: alliance, attachment, and nurturance
Title: #StayHome and Play Brawl Stars #WithMe | Games, Photoshop, Drones, Cooking, and other
random things
Content: A live streaming gamer plays an online multiplayer game. The gamer explains his moves
and game progress while showing his excitement to the audience.
Provisions: attachment and integration
Table 5: Videos with highest like count in the top 5 video styles. From top to bottom: how-to,artistic,homelife,chaing, and
game. The social provisions were annotated by MTurk participants.
5.1.2 Mentioning the COVID-19 Pandemic. #StayHome #WithMe
were two YouTube hashtags popularized during the COVID-19
pandemic. However, the authors noticed that most SHWM did not
intensively mention the pandemic. The average rating of COVID-
19 mentioning was 0.55, between “none” and “low” mentioning.
Only 21 videos (1.41%) were rated as highly mentioned COVID-19
pandemic, whereas 615 videos (41.33%) did not even mention the
pandemic (see distribution in Figure 5). Wilcoxon test with posthoc
pair-wise comparisons (
𝜒2(
8
)=
141
.
70,
𝑝<
0
.
0001) suggested
that homelife and chatting had signicantly higher
𝑐𝑜𝑣
than artistic,
how-to,challenge,story and game videos (all
𝑝<
0
.
0064). This result
suggests that YouTubers did not discuss or spread pandemic-related
information when they sought to mitigate viewers’ loneliness dur-
ing #StayHome #WithMe. Instead, SHWM videos had more content
for skill-learning and entertaining to redirect viewers’ attention
from COVID-19. For creators who mentioned COVID-19 more,
their videos showed homelife or chat with the audience. SHWM
videos didn’t intensively mention COVID because YouTubers felt
discussing this stressor in the video would negatively impact view-
ers’ mental states and wellbeing [16].
5.2 RQ2: Video Styles, COVID-19 Mentioning,
and Social Provisions
RQ2 probes how SHWM videos associate with the social provisions
[
14
]. Social integration and guidance were the most-supported pro-
visions (Figure 6). Attachment and reassurance of worth had a
similar amount of videos. Opportunity for nurturance and reliable
Figure 5: The distribution of COVID-19 mentioning. On X-
axis, 0 for “not mentioned” and 3 for “highly mentioned”
alliance, the two social provisions come from family-like relation-
ships, were supported by the fewest. Among 1488 videos, 77 were
tagged not to associate with any social provision. Spearman’s
𝜌
test suggested weak positive correlations between
𝑎𝑡𝑡 𝑎𝑐ℎ𝑚𝑒𝑛𝑡
and
𝑛𝑢𝑟𝑡𝑢𝑟𝑎𝑛𝑐𝑒
(
𝜌=
0
.
35,
𝑝<
0
.
0001),
𝑎𝑡𝑡 𝑎𝑐ℎ𝑚𝑒𝑛𝑡
and
𝑎𝑙𝑙𝑖 𝑎𝑛𝑐𝑒
(
𝜌=
0
.
20,
𝑝<
0
.
0001),
𝑎𝑙𝑙𝑖 𝑎𝑛𝑐𝑒
and
𝑛𝑢𝑟𝑡𝑢𝑟𝑎𝑛𝑐𝑒
(
𝜌=
0
.
19,
𝑝<
0
.
0001),
and
𝑟𝑒𝑎𝑠𝑠𝑢𝑟 𝑎𝑛𝑐𝑒
and
𝑖𝑛𝑡𝑒𝑔𝑟𝑎𝑡 𝑖𝑜𝑛
(
𝜌=
0
.
122,
𝑝<
0
.
0001). Nom-
inal logistic regression (LR) was used to predict each of the six
social provisions by
𝑠𝑡𝑦𝑙 𝑒
and
𝑐𝑜𝑣
. The alpha value to decide model
signicance was 0.0083 (0.05/6, after Bonferroni correction). One-
sided Fisher’s exact test was the posthoc to identify associations
between styles and provisions (
𝛼=
0
.
0056 after Bonferroni correc-
tion). Figure 7 shows the percentages of videos in dierent video
styles oering each social provision. Figure 8 illustrates signicant
association between video styles and social provisions.
Accepted Manuscript
How Do YouTubers Help with CO VID-19 Loneliness? CHI ’21, May 8–13, 2021, Yokohama, Japan
Figure 6: Number of videos providing each of the social pro-
visions. Vertical bars show correlations.
Figure 7: The percentages of videos providing the six social
provisions by dierent video styles
Figure 8: Signicant associations between video styles and
social provisions. Green triangles are positive associations.
Red ones are negative associations. p* < 0.05, p** < 0.01, p***
< 0.001.
5.2.1 Social integration. Social integration describes a friend-based
network in which people share common interests and concerns. In
SHWM, social integration was a pervasive social provision provided
by most video styles. Except for religious videos, the participants
rated more than 49% videos in all other styles to provide social in-
tegration (Figure 7). The LR model revealed a collective signicant
eect of
𝑠𝑡𝑦𝑙 𝑒
and
𝑐𝑜𝑣
on
𝑖𝑛𝑡𝑒𝑔𝑟𝑎𝑡 𝑖𝑜𝑛
(
𝜒2(
9
)=
98
.
45,
𝑅2=
0
.
05,
𝑝<
0
.
0001), with
𝑠𝑡𝑦𝑙 𝑒
and
𝑐𝑜𝑣
having signicant variable eects
(
𝑝𝑠𝑡 𝑦𝑙𝑒 <
0
.
0001,
𝑝𝑐𝑜𝑣 =
0
.
0017). Posthoc revealed that artistic,chal-
lenge,game, and review had signicantly more videos providing
social integration (Figure 8). While how-to and religious had signi-
cantly fewer videos related to social integration than other styles.
It was also noticed that videos with high COVID-19 mentioning
were less associated with
𝑖𝑛𝑡𝑒𝑔𝑟𝑎𝑡 𝑖𝑜𝑛
(
𝑐𝑜𝑒 =−
0
.
25,
𝜒2=
9
.
82).
These results indicate that sharing interests was the most common
way to support loneliness during SHWM. The interests covered
a wide range of topics: from artistic content to showing stunts
and self-challenges, from live gaming to recommending products
or services. YouTubers made videos of interests and hobbies that
entertain viewers and mitigate loneliness.
5.2.2 Guidance. Guidance refers to a mentor-like relationship in
which people obtain advice and assistance from a trustworthy per-
son. YouTube communities include professional creators who can
impart skills and knowledge through step-by-step instructions or
knowledge explanation. [
7
,
10
]. The LR model revealed that the
𝑠𝑡𝑦𝑙 𝑒
and
𝑐𝑜𝑣
had a signicant positive eect on
𝑔𝑢𝑖𝑑𝑎𝑛𝑐𝑒
(
𝜒2(
9
)=
903
.
18,
𝑅2=
0
.
45,
𝑝<
0
.
0001,
𝑝𝑠𝑡 𝑦𝑙𝑒 <
0
.
0001,
𝑝𝑐𝑜𝑣 =
0
.
0290). In
contrast to social integration, guidance was a provision primarily
supported by how-to videos (Figure 8). Posthoc showed that only
the how-to style had a signicantly higher proportion (88.77%) of
videos that provide guidance. While guidance videos were created
during COVID-19, they were not particularly mentioning this dis-
aster (
𝑐𝑜𝑒 =−
0
.
26,
𝜒2=
4
.
77,
𝑝𝑐𝑜𝑣 =
0
.
0290). This result indicates
that YouTubers shared general skills and knowledge to give viewers
guidance and a mentor-like relationship, but the knowledge was
not about health and safety information.
5.2.3 Aachment. Attachment describes family-like relationships
that create a sense of safety and closeness to dispel emotional lone-
liness. LR model suggested a collective signicant eect between
𝑠𝑡𝑦𝑙 𝑒
,
𝑐𝑜𝑣
, and
𝑎𝑡𝑡 𝑎𝑐ℎ𝑚𝑒𝑛𝑡
(
𝜒2(
9
)=
192
.
20,
𝑅2=
0
.
11,
𝑝<
0
.
0001,
𝑝𝑠𝑡 𝑦𝑙𝑒 <
0
.
0001,
𝑝𝑐𝑜𝑣 <
0
.
0001). Posthoc suggested that chatting,
homelife, and religious videos had signicantly higher proportions
of videos providing attachment, while challenge,game, and how-to
had signicantly fewer (Figure 8). In chatting videos, YouTubers
interacted with the audience through face-to-face talking, which
gave the audience a sense of attachment. Homelife videos were
at-home activities to show YouTuber’s intimacy. Religious videos
provided attachment by showing prayers or quotes from religious
scriptures. In contrast to social integration and guidance, videos
that oered attachment appeared to be more related to COVID-19.
The coecient for
𝑐𝑜𝑣
was 0.52 (
𝜒2=
35
.
8,
𝑝𝑐𝑜𝑣 <
0
.
0001). These
results suggest that attachment in SHWM videos was provided
by talking to the audience, showing YouTubers’ home activities,
or giving a religious prayer. YouTubers concerned about the audi-
ence’s wellbeing during the pandemic were more likely to oer the
attachment to help them mitigate loneliness.
5.2.4 Opportunity for nurturance and reliable alliance. Similar to
attachment, the opportunity for nurturance and reliable alliance
are derived from family-based relationships. The former describes
a relationship where one feels responsible for the other’s wellbe-
ing. The latter emphasizes one can count on the other under any
circumstances. However, these two provisions were supported by
the fewest SHWM videos. Since they had similar LR and posthoc
results, this section explains these two provisions together. LR
model suggested that
𝑠𝑡𝑦𝑙 𝑒
and
𝑐𝑜𝑣
had an signicant eect on
𝑛𝑢𝑟𝑡𝑢𝑟𝑎𝑛𝑐𝑒
(
𝜒2(
9
)=
149
.
89,
𝑅2=
0
.
10,
𝑝<
0
.
0001,
𝑝𝑠𝑡 𝑦𝑙𝑒 <
0
.
0001,
𝑝𝑐𝑜𝑣 <
0
.
0001) and
𝑎𝑙𝑙𝑖 𝑎𝑛𝑐𝑒
(
𝜒2(
9
)=
48
.
25,
𝑅2=
0
.
08,
𝑝<
0
.
0001,
𝑝𝑠𝑡 𝑦𝑙𝑒 <
0
.
0001,
𝑝𝑐𝑜𝑣 <
0
.
0001). Posthoc analysis suggested that
homelife,chatting, and religious videos had signicantly more videos
with
𝑛𝑢𝑟𝑡𝑢𝑟𝑎𝑛𝑐𝑒
(Figure 8). The coecient of
𝑐𝑜𝑣
in the model
was 0.81 (
𝜒2=
74
.
01,
𝑝𝑐𝑜𝑣 <
0
.
0001), which indicates that higher
COVID-19 mentioning was associated with a higher chance of
providing nurturance. For
𝑎𝑙𝑙𝑖 𝑎𝑛𝑐𝑒
, posthoc analysis showed that
chatting and religious videos were associated with
𝑎𝑙𝑙𝑖 𝑎𝑛𝑐𝑒
(Fig-
ure 8). The coecient of
𝑐𝑜𝑣
in the LR model was 0.58 (
𝜒2=
15
.
7,
Accepted Manuscript
CHI ’21, May 8–13, 2021, Yokohama, Japan Niu, Bartolome, Mai, and Ha
𝑝<
0
.
0001), indicating the reliable alliance was also correlated with
higher COVID-19 mentioning. These results indicate that similar
to attachment, the opportunity for nurturance and reliable alliance
were provided by videos of chatting and religious content. Home-
life videos also provided opportunity for nurturance. The concerns
over COVID-19 encouraged YouTubers to make videos to provide
these two family-like provisions, despite they were only provided
by fewer SHWM videos.
5.2.5 Reassurance of Worth. Reassurance of worth is a relation-
ship in which one’s skills and abilities are acknowledged by others,
which is usually achieved by oering help or understanding to
others [
14
]. The LR analysis suggested that
𝑠𝑡𝑦𝑙 𝑒
was a signicant
impact factor to this provision, but
𝑐𝑜𝑣
did not aect
𝑟𝑒𝑎𝑠𝑠𝑢𝑟 𝑎𝑛𝑐𝑒
(
𝜒2(
9
)=
139
.
61,
𝑅2=
0
.
08,
𝑝<
0
.
0001,
𝑝𝑠𝑡 𝑦𝑙𝑒 <
0
.
0001,
𝑝𝑐𝑜𝑣 =
0
.
8621). Posthoc showed that artistic is the only video style posi-
tively associated with this social provision. 163 out of 340 artistic
videos were tagged to support reassurance of worth, while homelife,
religious,review, and story videos had signicantly fewer videos
with this provision (Figure 8). Considering that artistic videos also
provided social integration, participants considered that YouTu-
bers presented artistic content or performance to entertain viewers
while hoping their skills and abilities can be acknowledged.
5.3 RQ3: Social Provisions and Viewer
Engagement
RQ3 explores how videos with dierent social provisions aected
viewer engagement to imply how dierent types of parasocial rela-
tionships aect viewers’ interactions with the video and their eects
on mitigating loneliness. For each social provision, the authors com-
pared if demonstrating a social provision increased or decreased
the three popularity measurements, two activity measurements,
and two comment emotion measurements. The comparison was
performed within each of the nine video style groups. Ordinary
Least Squares (OLS) model was used as the multivariate analysis
method. To ensure the model’s comprehensiveness, the model also
incorporated subscriber count (
𝑠𝑢𝑏
) as an additional independent
factor. Therefore, social provisions (dummy variables) and sub-
scriber count (numerical) were the independent variables to predict
the dependent variables of viewer engagement. 125 videos were ex-
cluded from the viewer engagement analysis because their creator
disabled comment and/or like functions. The alpha to determine
model signicance was 0.0071 (0.05/7, after Bonferroni correction).
The correlation between
𝑠𝑢𝑏
and each provision was tested with
Spearman’s
𝜌
test to avoid multicollinearity. No signicant correla-
tion was detected.
5.3.1 Popularity. In the OLS model which predicted video popu-
larity,
𝑣𝑖𝑒𝑤
and
𝑙𝑖𝑘 𝑒
did not have any association with any social
provision variables (only correlated with
𝑠𝑢𝑏
). But for how-to,home-
life, and review videos, besides
𝑠𝑢𝑏
, one or more social provisions
had signicant eects on the comment count (
𝑐𝑜𝑚𝑚𝑒𝑛𝑡
). For how-
to videos,
𝑐𝑜𝑚𝑚𝑒𝑛𝑡
was signicantly associated with
𝑎𝑡𝑡 𝑎𝑐ℎ𝑚𝑒𝑛𝑡
(
𝐹(
7
)=
9
.
09,
𝑅2=
0
.
11,
𝑝<
0
.
0001). OLS model showed that the co-
ecient of attachment was 56.01 (
𝑡𝑎𝑡𝑡 𝑎𝑐ℎ𝑚𝑒𝑛𝑡 =
2
.
08,
𝑝𝑎𝑡𝑡 𝑎𝑐ℎ𝑚𝑒𝑛𝑡 =
0
.
0384) in predicting how-to videos’ comment count, which sug-
gested a positive association between attachment and comments.
For homelife videos, opportunity for nurturance also had a pos-
itive coecient of 73.45 (
𝐹(
7
)=
25
.
08,
𝑅2=
0
.
50,
𝑝<
0
.
0001,
𝑡𝑛𝑢𝑟𝑡 𝑢𝑟𝑎𝑛𝑐 𝑒 =
2
.
39,
𝑝𝑛𝑢𝑟𝑡 𝑢𝑟𝑎𝑛𝑐 𝑒 =
0
.
0179) in predicting
𝑐𝑜𝑚𝑚𝑒𝑛𝑡
.
For review videos, the model suggested that reassurance of worth
had a coecient of 139.95 (
𝐹(
7
)=
7
.
11,
𝑅2=
0
.
53,
𝑝<
0
.
0001,
𝑡𝑟𝑒 𝑎𝑠𝑠𝑢𝑟 𝑎𝑛𝑐𝑒 =
2
.
26,
𝑝𝑟𝑒 𝑎𝑠𝑠𝑢𝑟 𝑎𝑛𝑐𝑒 =
0
.
0287) in predicting comments.
While social provisions did not aect the view and like amount
of SHWM videos, videos in how-to,homelife, and review styles at-
tracted more comments by providing attachment, nurturance, or
reassurance. It implies that despite social provisions didn’t help
SHWM videos to reach more audience; they had a positive eect on
encouraging more viewers to leave a comment. It is also interesting
to note that for how-to and review videos, the social provisions that
positively aected commenting were family-like provisions. How-
ever, they were not the main social provisions oered by SHWM.
5.3.2 Activity. Like rate (𝑙𝑖𝑘𝑒 _𝑟𝑎𝑡 𝑒) and comment rate (𝑐𝑜𝑚𝑚𝑒𝑛𝑡
_𝑟𝑎𝑡𝑒) were the two metrics to measure viewers’ activeness in the
interactions with SHWM videos. OLS model suggested a collective
signicant eects of social provisions and
𝑠𝑢𝑏
on
𝑙𝑖𝑘 𝑒_𝑟𝑎𝑡𝑒
and
𝑐𝑜𝑚𝑚𝑒𝑛𝑡 _𝑟𝑎𝑡 𝑒
in how-to and artistic groups. For how-to videos,
there was a signicant eect of social provisions on
𝑙𝑖𝑘 𝑒_𝑟𝑎𝑡𝑒
(
𝐹(
7
)=
3
.
18,
𝑅2=
0
.
04,
𝑝=
0
.
0027). The model suggested that
providing alliance made how-to videos attracted more likes per 100
views (
𝑐𝑜𝑒 =
1
.
05,
𝑡𝑎𝑙𝑙𝑖 𝑎𝑛𝑐𝑒 =
2
.
01,
𝑝𝑎𝑙𝑙𝑖 𝑎𝑛𝑐𝑒 =
0
.
0444). The variable
of
𝑛𝑢𝑟𝑡𝑢𝑟𝑎𝑛𝑐𝑒
had signicant positive eects on
𝑐𝑜𝑚𝑚𝑒𝑛𝑡 _𝑟𝑎𝑡 𝑒
of artistic videos (
𝐹(
7
)=
4
.
17,
𝑅2=
0
.
09,
𝑝=
0
.
0002). The coef-
cient of
𝑛𝑢𝑟𝑡𝑢𝑟𝑎𝑛𝑐𝑒
in the model was 1.46 (
𝑡𝑛𝑢𝑟𝑡 𝑢𝑟𝑎𝑛𝑐 𝑒 =
2
.
25,
𝑝𝑛𝑢𝑟𝑡 𝑢𝑟𝑎𝑛𝑐 𝑒 =
0
.
0253). These results indicated that although al-
liance and nurturance were the least provided social provisions in
SHWM, oering those two provisions helped how-to videos to gain
more likes and artistic videos to gain more comments. The result
implies that showing alliance in how-to videos and nurturance in
artistic videos encouraged viewers to more actively participate in
the parasocial interactions on YouTube.
5.3.3 Comment Emotion. The comment analysis predicts the fre-
quencies of emotional word in viewer comments by the factors
of social provisions and the subscriber count. 70245 comments
from 1135 videos (at most 200 for each video) were included in
this analysis (353 videos had no comment or disabled comment-
ing). OLS model suggested no social provisions had signicant
eects on
𝑝𝑜𝑠𝑖𝑡𝑖 𝑣𝑒 _𝑠𝑐𝑜𝑟𝑒
or
𝑛𝑒𝑔𝑎𝑡𝑖𝑣𝑒 _𝑠𝑐𝑜𝑟 𝑒
in any of the nine style
groups. Therefore the authors cannot conclude that providing
dierent social provisions had an eect on comment emotions.
The authors then chose to examine whether dierent video styles
and COVID-19 mentioning had an eect on comment emotions.
The OLS model indicated signicant eects of
𝑠𝑡𝑦𝑙 𝑒
and
𝑐𝑜𝑣
on
𝑝𝑜𝑠𝑖𝑡𝑖 𝑣𝑒 _𝑠𝑐𝑜𝑟𝑒
and
𝑛𝑒𝑔𝑎𝑡𝑖𝑣𝑒 _𝑠𝑐𝑜𝑟 𝑒
.
𝐶𝑜𝑣
showed a signicant eect
in the model which predicts
𝑛𝑒𝑔𝑎𝑡𝑖𝑣𝑒 _𝑠𝑐𝑜𝑟 𝑒
(
𝑐𝑜𝑒 =
0
.
05,
𝑡=
2
.
22,
𝑝=
0
.
0267).
𝑆𝑡𝑦𝑙𝑒
had signicant eects on
𝑝𝑜𝑠𝑖𝑡𝑖 𝑣𝑒 _𝑠𝑐𝑜𝑟𝑒
and
𝑛𝑒𝑔𝑎𝑡𝑖𝑣𝑒 _𝑠𝑐𝑜𝑟 𝑒
(
𝑝𝑝𝑜𝑠𝑖 𝑡𝑖 𝑣𝑒_𝑠𝑐𝑜𝑟𝑒 <
0
.
0001,
𝑝𝑛𝑒𝑔𝑎𝑡𝑖 𝑣𝑒_𝑠𝑐𝑜𝑟 𝑒 =
0
.
0004).
Pairwise comparison with Dunn method suggested religious and
homelife videos had signicantly higher positive word frequencies
than game,challenge,artistic,chatting,how-to, and story videos (all
𝑝<
0
.
0089, Figure 9). Providing specic social provisions didn’t
alter viewers’ emotional expression in the comments. But view-
ers’ comments in videos which mentioned COVID-19 more were
Accepted Manuscript
How Do YouTubers Help with CO VID-19 Loneliness? CHI ’21, May 8–13, 2021, Yokohama, Japan
more negative, which suggest that COVID-19 content may evoke
negative feelings. In contrast, viewers’ comments to religious and
homelife videos, the two styles associated with attachment and nur-
turance, were more positive than other video styles. As SHWM is a
movement for promoting positive mental attitudes and avoiding
loneliness, these results indicate that videos in these two styles led
to more positive viewer reactions and emotions to the YouTubers.
Figure 9: The frequencies of positive and negative emotional
words in videos in dierent styles
6 DISCUSSION
6.1 #StayHome #WithMe Did Not Escalate
Pandemic Attention
The analysis of SHWM suggested YouTubers sought to de-escalate
disaster attention in SHWM videos. Social media is used as a tool
to produce, spread, and access disaster information during emer-
gencies and disasters. Prior research focused on examining the
aordances of social media, including YouTube [4, 44], in dissemi-
nating and receiving disaster information and serving as an emer-
gency management tool [
27
,
37
,
57
]. For example, Twitter enhanced
public situational awareness [
61
]; Facebook was a source of com-
munity and ocial information [
5
,
11
]; and Reddit was used for risk
perception and speculation [
23
]. However, more exposure to the
overwhelming COVID-19 news via social media caused negative
mental states such as stress and depression [
16
,
22
]. Prior works
on social media and disaster mental health mainly examined me-
dia’s roles in mental health surveillance [
32
] oering mental health
service [27], and enhancing community ties [49].
The trending of #StayHome #WithMe implies an alternative af-
fordance of YouTube that contrasts prior usage of social media in
disasters. Instead of accentuating COVID-19 information and rais-
ing situational awareness, this work found that SHWM YouTubers
utilized parasocial relationships to create a virtual space to allow
people to reduce the pandemic stress. In contrast to the information
source on Twitter and Facebook who provide immediate disaster
information, creators of #StayHome #WithMe did not prioritize in-
formation on the pandemic. They provided various social provisions
to redirect or de-escalate viewers’ attention from disaster-related
stressors. The average COVID-19 mentioning was between “none”
and “low”. Only 1.41% of the videos were rated as highly related
to the pandemic, whereas 41.33% videos did not even mention the
pandemic in videos. Videos with relatively higher COVID-19 men-
tioning were not about alarming the disaster or directing disaster
management; instead, they sought to provide supportive and com-
forting provisions such as attachment, alliance, and nurturance. The
most prominent social provisions in SHWM – social integration and
guidance – were negatively associated with COVID-19, indicating
YouTubers demonstrated hobbies or taught skills as an approach to
de-escalate pandemic attention and prioritize connection.
The prolonged social distancing and overwhelming COVID in-
formation put people, especially young adults, at risk for increased
loneliness and depression [
16
,
22
]. SHWM movement indicates new
design opportunities for using video sharing to mitigate the loneli-
ness and stress caused by disaster information. Besides informing
emergencies during a disaster, it is also imperative to call on social
media platforms and services to consider providing content that is
not disaster-intensive to allow people to experience normality and
reduce negative mental states. The ndings of this work suggest
that SHWM YouTubers would not put the disaster situation at the
center of the parasocial relationships with viewers. Instead, they
tended to share entertaining, educational, and comforting content.
This aordance of video-sharing platforms is valuable during a
crisis like COVID-19 when people cannot socialize and do everyday
things they usually enjoyed. The non-disaster-focused videos that
supplement absent social provisions can support viewers who need
to relax and reduce loneliness and stress. This characteristic of
YouTube and video-sharing communities should be considered and
leveraged in future technology and service designs for loneliness
relief and mental health support during disasters.
6.2 From Parasocial Relationships to Social
Provisions in a Disaster
The analysis of social provisions in SHWM videos implied the
manners in which parasocial relationships deliver social connec-
tions. The participatory nature of YouTube makes it a platform
for ordinary people to contribute content and engage viewers to
establish parasocial relationships [
7
,
65
]. Prior understanding of
parasocial relationships in video sharing surrounded how they are
formed [
13
,
34
] and their eects on the viewers’ daily activities
[
41
,
58
,
65
]. Although studies showed that parasocial relationships
were a source of alternative companionship and can help people
shield against loneliness [
19
,
20
,
24
], there is limited knowledge of
how YouTube-based parasocial relationships are embodied during
social distancing to help with loneliness.
The examination of SHWM videos under Weiss’s typology of pro-
visions revealed the construct of parasocial relationships in SHWM
videos and the roles YouTubers played in social distancing. The en-
coding of SHWM videos suggested that YouTubers sought to oer
social connections typically obtained from friend-like, mentor-like,
and family-like relationships. The authors found that social inte-
gration was a dominant provision in most SHWM videos (provided
by 63.35% videos). Acting as a friend-like character and sharing
common interests was the most common parasocial relationship
in SHWM. This echoes prior ndings that interest sharing helped
social media users overcome loneliness and depression [
18
,
43
].
Guidance was the second most common provisions available in
SHWM videos. 42.32% of videos provided the guidance provision,
most of which were how-to videos. This nding suggested that
building informal mentorship through how-to videos [
10
] was also
a common YouTubers’ style to supplement companionship. Family-
like provisions, including attachment, nurturance, and alliance,
were supported by the fewest videos. These videos shared intimate
Accepted Manuscript
CHI ’21, May 8–13, 2021, Yokohama, Japan Niu, Bartolome, Mai, and Ha
content, such as at-home activities, live-chatting, and religious
prayers, to build intimate provisions and family-like relationships
for people who need intimacy and emotional support. Social media
is known for connecting families and friends [
27
] and sharing emo-
tional content during disasters [
32
]. YouTubers act as dierent roles
through their relationships with viewers to oer mental support
during COVID-19. SHWM suggested that YouTubers can mitigate
disaster loneliness by sharing entertaining content to provide so-
cial integration, teaching skills to provide guidance, and showing
at-home activities or chatting to provide attachment.
The correlations between video styles and social provisions sug-
gest practical ways to utilize parasocial relationships to oer lone-
liness support during a disaster. Viewers can nd YouTubers who
imitate friend-, mentor-, and family-like connections to mitigate
loneliness resulted from social distancing. Social integration was
the most prevalent social provision in SHWM videos; therefore,
designers may leverage YouTubers’ videos to design video applica-
tions and services for people who were isolated from friends. People
who miss mentor-like relationships, especially children and youth
whose schools are closed during the pandemic, can leverage the
repository of how-to videos to learn various skills and knowledge.
This aordance of YouTube can supplement the inadequate guid-
ance provision. Videos that oer social integration and guidance
can also redirect viewers’ attention from the crisis. For people who
need intimacy and closeness from a family-like relationship, such as
people separated from families due to the disasters, video-sharing
applications and services may recommend YouTubers’ content in
which they share their homelife activities or chat with the audience
to engender attachment and nurturance.
6.3 Encouraging Family-Like Social Provisions
in Loneliness-Supporting Videos
The viewer engagement analysis suggested videos with family-like
provisions had better eects on inducing viewers’ social interac-
tion participation. Prior studies suggested parasocial interactions
can engender intimacy and attachment [
13
], and help people fulll
the social interaction need [
33
] and mitigate loneliness [
24
]. Video
views, likes, and comments are metrics of viewer engagement to
evaluate parasocial interactions in SHWM [
33
,
50
]. The authors an-
alyzed how social provisions aect viewer engagement to explore
social provisions’ eects on parasocial relationships. The results
suggested that when YouTubers seek to provide parasocial relation-
ships to support mental wellbeing, oering family-like provisions
have a better overall eect on increasing viewer interactions and
mitigating loneliness.
There was no evidence that social provisions had eects on
helping SHWM videos to reach more viewers. Viewers’ emotional
expression in comments was also unaected by the expression of
social provisions. However, family-like social provisions demon-
strated an overall positive eect on viewers’ activeness in parasocial
relationships. Attachment increased the number of comments of
how-to videos, and opportunity for nurturance increased comments
of homelife videos. For viewers’ activity, alliance helped how-to
videos to receive more likes per 100 views. Providing nurturance
allowed artistic videos to gain more comments for every 100 views.
Although social provisions did not signicantly aect comment
emotions, the analysis on video styles revealed that religious and
homelife videos – the two styles bound to attachment and nurtu-
rance – had more positive comments than others. Viewers were
more positive after watching videos in those two styles. Prior stud-
ies suggested parasocial interactions with YouTubers can avoid and
mitigate loneliness [
19
,
24
,
50
]. The authors’ ndings suggested that
SHWM videos that supply attachment, nurturance, and alliance
had higher overall viewer engagement. As a result, videos with
those provisions positively aected viewers’ activeness in the so-
cial interactions on YouTube, indicating more potentials to mitigate
loneliness. However, these three family-like provisions were also
among the least provided social provisions in SHWM videos. Only
around 27% SHWM videos provided attachment and opportunity
for nurturance, and only 4.9% videos provided reliable alliance.
Experiencing close family and friend relationships was helpful
to reduce loneliness during COVID-19 [
16
]. YouTubers should con-
sider providing more family-like provisions – by showing intimacy,
caring for viewers’ wellbeing, and showing a willingness to help
– to increase viewers’ interactions and positivity in the parasocial
relationships. Platform designers may consider increasing engage-
ment on YouTube for mitigating loneliness by encouraging videos to
express family-like provisions. Promising design solutions to grow
YouTuber-viewer intimacy include recommending video styles that
enhance family-like feelings and implementing communication
methods that allow YouTubers to show intimacy and support. Many
YouTubers already established family-like proles among their fans,
such as popular homelife vloggers, family-friendly streamers, and
many ASMRtists [
1
]. Platforms can invite and encourage these
YouTubers to make loneliness-supporting videos during disasters
to support public mental wellbeing.
7 CONCLUSION AND FUTURE WORK
Social media plays an increasingly important role in supporting
people’s mental wellbeing during a dicult time like COVID-19.
Video-sharing platforms like YouTube are conducive for providing
social connectedness and reducing loneliness during social distanc-
ing. This work examines the #StayHome #WithMe movement as
a space for YouTubers to help mitigate COVID-19 loneliness. The
authors identied video styles and obtained a panoramic under-
standing of YouTubers’ creation activities and parasocial relation-
ships in SHWM. Grounded on Weiss’s theory of loneliness, six
social provisions were rated by MTurk participants to analyze how
videos with dierent styles and COVID-19 mentioning aect the
provisions YouTubers sought to oer. The authors also explored
how dierent social provisions aect video popularity, viewers’
activities, and comment emotion, as indicators of participation in
parasocial interactions. From the results, the #StayHome #WithMe
hashtags were primarily used as a space for teaching skills and
knowledge, presenting entertaining videos such as artistic presen-
tation and gameplay, and showing homelife activities or chatting
with the audience. The videos were less related to the on-going
pandemic. The analysis revealed how parasocial relationships sup-
plement social provisions during COVID-19. Social integration was
a dominant provision provided by most of the video styles. How-to
videos supported the need for guidance. Homelife,chatting, and
religious videos oered a sense of attachment and nurturance. The
Accepted Manuscript
How Do YouTubers Help with CO VID-19 Loneliness? CHI ’21, May 8–13, 2021, Yokohama, Japan
viewer engagement analysis suggested that family-like provisions
were the least oered provisions, but they positively aected viewer
engagement and parasocial interactions. Based on these ndings,
the authors suggest that SHWM videos sought to de-escalate the
mental tension caused by COVID-19. YouTubers oered friend-like
and mentor-like provisions the most, while the family-like provi-
sions are supported the least. YouTubers and platform designers
should encourage content that oers attachment, nurturance, and
alliance during the pandemic to increase parasocial interactions
and avoid or mitigate loneliness.
Moving forward, future studies will extend the ndings of the
present work to advance the knowledge of supporting disaster
mental health through video sharing. As user-generated videos
will play an increasing role, new studies and designs are needed to
understand the interplay between video sharing and mental well-
being. Follow-up research will extend the ndings of this study
and develop new design knowledge. For example, one unanswered
question in this work is to what degree parasocial interactions can
psychologically supplement various social needs during a disas-
ter. The authors do not argue that parasocial relationships with
YouTubers can or should replace realistic social interactions with
families and friends. However, YouTubers oered an alternative but
growingly popular way to let people stay socially connected during
disasters; therefore, it requires a more in-depth investigation. Social
interactions are easily aected by disasters. YouTube provides an
option to satisfy social-emotional needs through parasocial interac-
tions. This work’s ndings oer a seminal idea regarding the use of
YouTube and YouTubers’ roles in supporting disaster mental well-
being. It is necessary to examine YouTube viewers’ cognitive and
behavioral changes after interacting with YouTube videos during
and after disasters. Future work will also investigate new trending
video styles such as ASMR and live-streams in obtaining social pro-
visions. These eorts will identify new possibilities of applications
and services to utilize video sharing to support mental wellbeing
in intensive situations. It is essential to explore their options in
intervening in mental health issues of the vulnerable populations.
REFERENCES
[1]
Joceline Andersen. 2014. Now You’ve Got the Shiveries: Aect, Intimacy, and the
ASMR Whisper Community. Television & New Media 16, 8 (11 2014), 683–700.
https://doi.org/10.1177/1527476414556184
[2]
Laurensia Anjani, Terrance Mok, Anthony Tang, Lora Oehlberg, and Wooi Boon
Goh. 2020. Why Do People Watch Others Eat Food? An Empirical Study on
the Motivations and Practices of Mukbang Viewers. In Proceedings of the 2020
CHI Conference on Human Factors in Computing Systems (CHI ’20). Association
for Computing Machinery, New York, NY, USA, 1–13. https://doi.org/10.1145/
3313831.3376567
[3]
Shrey Bagroy, Ponnurangam Kumaraguru, and Munmun De Choudhury. 2017.
A Social Media Based Index of Mental Well-Being in College Campuses. In
Proceedings of the 2017 CHI Conference on Human Factors in Computing Systems
(CHI ’17). Association for Computing Machinery, New York, NY, USA, 1634–1646.
https://doi.org/10.1145/3025453.3025909
[4]
Corey H Basch, Charles E Basch, Kelly V Ruggles, and Rodney Hammond. 2015.
Coverage of the Ebola virus disease epidemic on YouTube. Disaster medicine and
public health preparedness 9, 5 (2015), 531–535. https://doi.org/10.1017/dmp.2015.
77
[5]
Deanne Bird, Megan Ling, and Katharine Haynes. 2012. Flooding Facebook–the
use of social media during the queensland and Victorian oods. INTERES 27, 1
(2012), 27.
[6]
Youmna Borghol, Sebastien Ardon, Niklas Carlsson, Derek Eager, and Anir-
ban Mahanti. 2012. The Untold Story of the Clones: Content-Agnostic Fac-
tors That Impact YouTube Video Popularity. In Proceedings of the 18th ACM
SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD
’12). Association for Computing Machinery, New York, NY, USA, 1186–1194.
https://doi.org/10.1145/2339530.2339717
[7]
Jean Burgess and Joshua Green. 2018. YouTube: Online video and participatory
culture. John Wiley & Sons, Cambridge, UK.
[8]
Kathy Charmaz and Linda Liska Belgrave. 2015. Grounded Theory. https:
//doi.org/10.1002/9781405165518.wbeosg070.pub2
[9]
G Chatzopoulou, C Sheng, and M Faloutsos. 2010. A First Step Towards Under-
standing Popularity in YouTube. In 2010 INFOCOM IEEE Conference on Com-
puter Communications Workshops. IEEE, San Diego, CA, USA, 1–6. https:
//doi.org/10.1109/INFCOMW.2010.5466701
[10]
Clement Chau. 2010. YouTube as a participatory culture. New Directions for
Youth Development 2010, 128 (2010), 65–74. https://doi.org/10.1002/yd.376
[11]
Apoorva Chauhan and Amanda L Hughes. 2017. Providing Online Crisis Informa-
tion: An Analysis of Ocial Sources during the 2014 Carlton Complex Wildre.
In Proceedings of the 2017 CHI Conference on Human Factors in Computing Systems
(CHI ’17). Association for Computing Machinery, New York, NY, USA, 3151–3162.
https://doi.org/10.1145/3025453.3025627
[12]
Xianhui Che, Barry Ip, and Ling Lin. 2015. A Survey of Current YouTube Video
Characteristics. IEEE MultiMedia 22, 2 (2015), 56–63. https://doi.org/10.1109/
MMUL.2015.34
[13]
Chih-Ping Chen. 2014. Forming digital self and parasocial relationships on
YouTube. Journal of Consumer Culture 16, 1 (2 2014), 232–254. https://doi.org/
10.1177/1469540514521081
[14]
Carolyn E Cutrona and Daniel W Russell. 1987. The provisions of social rela-
tionships and adaptation to stress. Advances in personal relationships 1, 1 (1987),
37–67.
[15]
Munmun De Choudhury, Michael Gamon, Scott Counts, and Eric Horvitz. 2013.
Predicting depression via social media. Icwsm 13 (2013), 1–10.
[16]
Wendy E Ellis, Tara M Dumas, and Lindsey M Forbes. 2020. Physically isolated
but socially connected: Psychological adjustment and stress among adolescents
during the initial COVID-19 crisis. Canadian Journal of Behavioural Science /
Revue canadienne des sciences du comportement 52, 3 (2020), 177–187. https:
//doi.org/10.1037/cbs0000215
[17]
Jordan Eschler, Eleanor R Burgess, Madhu Reddy, and David C Mohr. 2020. Emer-
gent Self-Regulation Practices in Technology and Social Media Use of Individuals
Living with Depression. In Proceedings of the 2020 CHI Conference on Human
Factors in Computing Systems (CHI ’20). Association for Computing Machinery,
New York, NY, USA, 1–13. https://doi.org/10.1145/3313831.3376773
[18]
Lauren Fratamico and Deb Roy. 2019. I’m Lonely. Who Should I Talk To?. In
Companion Proceedings of The 2019 World Wide Web Conference (WWW ’19).
Association for Computing Machinery, New York, NY, USA, 631–634. https:
//doi.org/10.1145/3308560.3317309
[19]
Wendi L Gardner, Cynthia L Pickett, and Megan Knowles. 2005. Social snacking
and shielding. In The social outcast: Ostracism, social exclusion, rejection, and
bullying. Psychology Press New York, NY, New York, NY, 227–242.
[20]
David C Giles. 2002. Parasocial Interaction: A Review of the Literature and a
Model for Future Research. Media Psychology 4, 3 (8 2002), 279–305. https:
//doi.org/10.1207/S1532785XMEP0403_04
[21]
Kimberly Glasgow, Jessica Vitak, Yla Tausczik, and Clay Fink. 2016. Grieving
in the 21st Century: Social Media’s Role in Facilitating Supportive Exchanges
Following Community-Level Traumatic Events. In Proceedings of the 7th 2016
International Conference on Social Media & Society (SMSociety ’16). Association
for Computing Machinery, New York, NY, USA, 1–10. https://doi.org/10.1145/
2930971.2930975
[22]
Jenny M Groarke, Emma Berry, Lisa Graham-Wisener, Phoebe E McKenna-
Plumley, Emily McGlinchey, and Cherie Armour. 2020. Loneliness in the
UK during the COVID-19 pandemic: Cross-sectional results from the COVID-
19 Psychological Wellbeing Study. PloS one 15, 9 (2020), e0239698. https:
//doi.org/10.1371/journal.pone.0239698
[23]
Xinning Gui, Yubo Kou, Kathleen Pine, Elisa Ladaw, Harold Kim, Eli Suzuki-Gill,
and Yunan Chen. 2018. Multidimensional Risk Communication: Public Discourse
on Risks during an Emerging Epidemic. In Proceedings of the 2018 CHI Conference
on Human Factors in Computing Systems (CHI ’18). Association for Computing
Machinery, New York, NY, USA, 1–14. https://doi.org/10.1145/3173574.3173788
[24]
Tilo Hartmann. 2017. Parasocial interaction, parasocial relationships, and well-
being. In The Routledge handbook of media use and well-being: International per-
spectives on theory and research on positive media eects. Routledge/Taylor & Fran-
cis Group, New York, NY, US, 131–144. https://doi.org/10.4324/9781315714752
[25]
Donald Horton and R Richard Wohl. 1956. Mass Communication and Para-Social
Interaction. Psychiatry 19, 3 (8 1956), 215–229. https://doi.org/10.1080/00332747.
1956.11023049
[26]
Mingyi Hou. 2018. Social media celebrity and the institutionalization of YouTube.
Convergence 25, 3 (1 2018), 534–553. https://doi.org/10.1177/1354856517750368
[27]
J Brian Houston, Joshua Hawthorne, Mildred F Perreault, Eun Hae Park, Marlo
Goldstein Hode, Michael R Halliwell, Sarah E Turner McGowen, Rachel Davis,
Shivani Vaid, Jonathan A McElderry, and Stanford A Grith. 2015. Social media
and disasters: a functional framework for social media use in disaster planning,
response, and research. Disasters 39, 1 (2015), 1–22. https://doi.org/10.1111/disa.
Accepted Manuscript
CHI ’21, May 8–13, 2021, Yokohama, Japan Niu, Bartolome, Mai, and Ha
12092
[28]
Mu Hu, Mingli Zhang, and Nuan Luo. 2016. Understanding participation on video
sharing communities: The role of self-construal and community interactivity.
Computers in Human Behavior 62 (2016), 105–115. https://doi.org/10.1016/j.chb.
2016.03.077
[29]
Mu Hu, Yueyu Zhao, Zitong Liu, Ziyu Li, and Ximin Kong. 2020. Just my
imagination: The inuence of celebrities’ romantic relationship announcements
on romance fans and friendship fans. In Psychology of Popular Media. Edu-
cational Publishing Foundation, Hu, Mu: College of Arts and Sciences, Texas
A&M University-San Antonio, One University Way, San Antonio, TX, US,
78224, mhu@tamusa.edu, No Pagination Specied–No Pagination Specied.
https://doi.org/10.1037/ppm0000324
[30]
Karin Hugelius, Mervyn Giord, Per Ortenwall, and Annsoe Adolfsson. 2016.
“To silence the deafening silence”: Survivor’s needs and experiences of the impact
of disaster radio for their recovery after a natural disaster. International Emergency
Nursing 28 (2016), 8–13. https://doi.org/10.1016/j.ienj.2015.11.009
[31]
Kawaljeet Kaur Kapoor, Kuttimani Tamilmani, Nripendra P. Rana, Pushp Patil,
Yogesh K. Dwivedi, and Sridhar Nerur. 2018. Advances in Social Media Research:
Past, Present and Future. Information Systems Frontiers 20, 3 (2018), 531–558.
https://doi.org/10.1007/s10796-017- 9810-y
[32]
Dhivya Karmegam, Thilagavathi Ramamoorthy, and Bagavandas Mappillaira-
jan. 2020. A Systematic Review of Techniques Employed for Determining
Mental Health Using Social Media in Psychological Surveillance During Dis-
asters. Disaster Medicine and Public Health Preparedness 14, 2 (2020), 265–272.
https://doi.org/10.1017/dmp.2019.40
[33]
M Laeeq Khan. 2017. Social media engagement: What motivates user participation
and consumption on YouTube? Computers in Human Behavior 66 (2017), 236–247.
https://doi.org/10.1016/j.chb.2016.09.024
[34]
Kate Szer Kurtin, Nina O’Brien, Deya Roy,and Linda Dam. 2018. The Development
of Parasocial Interaction Relationships on YouTube. The Journal of Social Media
in Society; Vol 7, No 1 (2018): The Journal of Social Media in Society 7 (5 2018),
233–252. https://www.thejsms.org/index.php/TSMRI/article/view/304
[35]
Janette Lehmann, Mounia Lalmas, Elad Yom-Tov, and Georges Dupret. 2012. Mod-
els of User Engagement. In International conference on user modeling, adaptation,
and personalization, Judith Mastho, Bamshad Mobasher, Michel C Desmarais,
and Roger Nkambou (Eds.). Springer Berlin Heidelberg, Berlin, Heidelberg, 164–
175. https://doi.org/10.1007/978-3- 642-31454- 4_14
[36]
Y Saltzman Leia, Cross Hansel Tonya, and S Bordnick Patrick. 2020. Loneliness,
Isolation, and Social Support Factors in post-COVID-19 Mental Health. , 55 –
57 pages. https://doi.org/10.1037/tra0000703
[37]
Bruce R Lindsay. 2011. Social media and disasters: Current uses, future options,
and policy considerations. Technical Report. Congressional Research Service.
[38]
Ivan Wing Chit Mak, Chung Ming Chu, Pey Chyou Pan, Michael Gar Chung Yiu,
and Veronica Lee Chan. 2009. Long-term psychiatric morbidities among SARS
survivors. General Hospital Psychiatry 31, 4 (2009), 318–326. https://doi.org/10.
1016/j.genhosppsych.2009.03.001
[39]
Alice E Marwick. 2015. You May Know Me from YouTube: (Micro-)Celebrity in
Social Media. , 333–350 pages. https://doi.org/10.1002/9781118475089.ch18
[40]
Saif M Mohammad and Peter D Turney. 2013. Crowdsourcing a Word-Emotion
Association Lexicon. Computational Intelligence 29, 3 (2013), 436–465.
[41]
Juha Munnukka, Devdeep Maity, Hanna Reinikainen, and Vilma Luoma-aho.
2019. “Thanks for watching”. The eectiveness of YouTube vlogendorsements.
Computers in Human Behavior 93 (2019), 226–234. https://doi.org/10.1016/j.chb.
2018.12.014
[42]
Mor Naaman, Jerey Boase, and Chih-Hui Lai. 2010. Is It Really about Me?
Message Content in Social Awareness Streams. In Proceedings of the 2010 ACM
Conference on Computer Supported Cooperative Work (CSCW ’10). Association for
Computing Machinery, New York, NY, USA, 189–192. https://doi.org/10.1145/
1718918.1718953
[43]
Kathleen O’Leary, Stephen M Schueller, Jacob O Wobbrock, and Wanda Pratt.
2018. “Suddenly, We Got to Become Therapists for Each Other”: Designing
Peer Support Chats for Mental Health. In Proceedings of the 2018 CHI Conference
on Human Factors in Computing Systems (CHI ’18). Association for Computing
Machinery, New York, NY, USA, 1–14. https://doi.org/10.1145/3173574.3173905
[44]
Jacqueline K Owens, Andrea Warner Stidham, and Elizabeth L Owens. 2013.
Disaster evacuation for persons with special needs: A content analysis of in-
formation on YouTube. Applied Nursing Research 26, 4 (2013), 273–275. https:
//doi.org/10.1016/j.apnr.2013.10.001
[45]
Yuval Palgi, Amit Shrira, Lia Ring, Ehud Bodner, Sharon Avidor, Yoav Bergman,
Sara Cohen-Fridel, Shoshi Keisari, and Yaakov Homan. 2020. The loneliness
pandemic: Loneliness and other concomitants of depression, anxiety and their
comorbidity during the COVID-19 outbreak. Journal of aective disorders 275 (10
2020), 109–111. https://doi.org/10.1016/j.jad.2020.06.036
[46]
Sungkyu Park, Inyeop Kim, Sang Won Lee, Jaehyun Yoo, Bumseok Jeong, and
Meeyoung Cha. 2015. Manifestation of Depression and Loneliness on Social
Networks: A Case Study of Young Adults on Facebook. In Proceedings of the 18th
ACM Conference on Computer Supported Cooperative Work & Social Computing
(CSCW ’15). Association for Computing Machinery, New York, NY, USA, 557–570.
https://doi.org/10.1145/2675133.2675139
[47]
Andrew Perrin and Monica Anderson. 2019. Share of U.S. adults us-
ing social media, including Facebook, is mostly unchanged since 2018.
https://www.pewresearch.org/fact-tank/2019/04/10/share-of- u-s- adults-using-
social-media- including-facebook-is- mostly-unchanged- since-2018/
[48]
Jenny Preece and Diane Maloney-Krichmar. 2002. Online Communities: Focus-
ing on Sociability and Usability. In The Human-Computer Interaction Handbook:
Fundamentals, Evolving Technologies and Emerging Applications. L. Erlbaum As-
sociates Inc., USA, 596–620.
[49]
Claire H Procopio and Steven T Procopio. 2007. Do You Know What It Means
to Miss New Orleans? Internet Communication, Geographic Community, and
Social Capital in Crisis. Journal of Applied Communication Research 35, 1 (2 2007),
67–87. https://doi.org/10.1080/00909880601065722
[50]
Leslie Rasmussen. 2018. Parasocial Interaction in the Digital Age: An Examination
of Relationship Building and the Eectiveness of YouTube Celebrities. The Journal
of Social Media in Society 7, 1 (2018), 280–294.
[51]
Tobias Raun. 2018. Capitalizing intimacy: New subcultural forms of micro-
celebrity strategies and aective labour on YouTube. Convergence 24, 1 (1 2018),
99–113. https://doi.org/10.1177/1354856517736983
[52]
Dana Rotman, Jennifer Golbeck, and Jennifer Preece. 2009. The Community is
Where the Rapport is – on Sense and Structure in the Youtube Community. In
Proceedings of the Fourth International Conference on Communities and Technolo-
gies (C&T ’09). Association for Computing Machinery, New York,N Y, USA, 41–50.
https://doi.org/10.1145/1556460.1556467
[53] Juliet Roudini, Hamid Reza Khankeh, and Evelin Witruk. 2017. Disaster mental
health preparedness in the community: A systematic review study. Health
Psychology Open 4, 1 (1 2017), 2055102917711307. https://doi.org/10.1177/
2055102917711307
[54]
Dan Russell, Carolyn E Cutrona, Jayne Rose, and Karen Yurko. 1984. Social
and emotional loneliness: An examination of Weiss’s typology of loneliness.
Journal of Personality and Social Psychology 46, 6 (1984), 1313–1321. https:
//doi.org/10.1037/0022-3514.46.6.1313
[55]
Irina Shklovski, Leysia Palen, and Jeannette Sutton. 2008. Finding Community
through Information and Communication Technology in Disaster Response. In
Proceedings of the 2008 ACM Conference on Computer Supported Cooperative Work
(CSCW ’08). Association for Computing Machinery, New York, NY, USA, 127–136.
https://doi.org/10.1145/1460563.1460584
[56]
Amber Silver and Lindsay Matthews. 2017. The use of Facebook for information
seeking, decision support, and self-organization following a signicant disaster.
Information, Communication & Society 20, 11 (2017), 1680–1697. https://doi.org/
10.1080/1369118X.2016.1253762
[57]
Robert Soden and Leysia Palen. 2018. Informating Crisis: Expanding Critical
Perspectives in Crisis Informatics. Proc. ACM Hum.-Comput. Interact. 2, CSCW
(11 2018), 1 – 22. https://doi.org/10.1145/3274431
[58]
Karina Sokolova and Hajer Ke. 2020. Instagram and YouTube bloggers promote it,
why should I buy? How credibility and parasocial interaction inuence purchase
intentions. Journal of Retailing and Consumer Services 53 (2020), 1 – 9. https:
//doi.org/10.1016/j.jretconser.2019.01.011
[59]
José Van Dijck. 2013. The culture of connectivity: A critical history of social media.
Oxford University Press, New York, NY. 110 – 132 pages.
[60]
Philippe Verduyn, David Seungjae Lee, Jiyoung Park, Holly Shablack, Ariana
Orvell, Joseph Bayer, Oscar Ybarra, John Jonides, and Ethan Kross. 2015. Passive
Facebook usage undermines aective well-being: Experimental and longitudinal
evidence. , 480–488 pages. https://doi.org/10.1037/xge0000057
[61]
Sarah Vieweg, Amanda L Hughes, Kate Starbird, and Leysia Palen. 2010. Mi-
croblogging during Two Natural Hazards Events: What Twitter May Contribute
to Situational Awareness. In Proceedings of the SIGCHI Conference on Human
Factors in Computing Systems (CHI ’10). Association for Computing Machinery,
New York, NY, USA, 1079–1088. https://doi.org/10.1145/1753326.1753486
[62]
Jessica Vitak, Nicole Ellison, and Charles Steineld. 2011. The Ties That Bond: Re-
Examining the Relationship between Facebook Use and Bonding Social Capital.
In 2011 44th Hawaii International Conference on System Sciences. IEEE, Kauai, HI,
USA, 1–10. https://doi.org/10.1109/HICSS.2011.435
[63]
Qi Wang, Edward L Fink, and Deborah A Cai. 2008. Loneliness, Gender, and
Parasocial Interaction: A Uses and Gratications Approach. Communication
Quarterly 56, 1 (2 2008), 87–109. https://doi.org/10.1080/01463370701839057
[64]
Robert S Weiss. 1973. Loneliness: The experience of emotional and social isolation.
The MIT Press, Cambridge, MA, US. xxii, 236–xxii, 236 pages.
[65]
Donghee Yvette Wohn, Guo Freeman, and Caitlin McLaughlin. 2018. Explain-
ing Viewers’ Emotional, Instrumental, and Financial Support Provision for Live
Streamers. In Proceedings of the 2018 CHI Conference on Human Factors in Com-
puting Systems (CHI ’18). Association for Computing Machinery, New York, NY,
USA, 1–13. https://doi.org/10.1145/3173574.3174048
[66]
Robert R Wright, Chad Schaeer, Rhett Mullins, and Austin Evans. 2020. Com-
parison of Student Health and Well-Being Proles and Social Media Use. Psi Chi
Journal of Psychological Research 25 (2020), 14 – 21. https://doi.org/10.24839/2325-
7342.JN25.1.14
Accepted Manuscript
How Do YouTubers Help with CO VID-19 Loneliness? CHI ’21, May 8–13, 2021, Yokohama, Japan
[67]
Yang Yu and Xiao Wang. 2015. World Cup 2014 in the Twitter World: A big data
analysis of sentiments in U.S. sports fans’ tweets. Computers in Human Behavior
48 (2015), 392–400. https://doi.org/10.1016/j.chb.2015.01.075
