ArticlePDF Available

Easier contagion and weaker ties make anger spread faster than joy in social media

Authors:

Abstract and Figures

Similar to face-to-face communication in daily life, more and more evidence suggests that human emotion also spread in online social media through virtual interactions. However, the mechanism underlying the emotion contagion, like whether different feelings diffuse unlikely or how the spread is coupled with the social network, is rarely investigated. Indeed, because of the costly expense and spatio-temporal limitations, it is challenging for conventional questionnaires or controlled experiments. While given the instinct of collecting natural affective responses of massive connected individuals, online social media offer us an ideal proxy to tackle this issue from the perspective of computational social science. In this paper, based on the analysis of millions of tweets in Weibo, a Twitter-like service in China, we surprisingly find that anger is more contagious than joy, indicating that it can sparkle more angry follow-up tweets and anger prefers weaker ties than joy for the dissemination in social network, indicating that it can penetrate different communities and break local traps by more sharings between strangers. We conjecture that easier contagion and weaker ties could together make anger spread faster than joy, which is further justified by the diffusion of bursty events with different dominant emotions. To our best knowledge, for the first time we quantificationally disclose the difference between joy and anger in dissemination mechanism and our findings would shed lights on personal anger management in human communication and collective outrage control in cyber space.
Content may be subject to copyright.
arXiv:1608.03656v1 [cs.SI] 12 Aug 2016
Easier contagion and weaker ties make
anger spread faster than joy in social media
Rui Fan1, Ke Xu1and Jichang Zhao2,*
1State Key Lab of Software Development Environment, Beihang University
2School of Economics and Management, Beihang University
Corresponding author: jichang@buaa.edu.cn
(Dated: August 15, 2016)
Abstract
Similar to face-to-face communication in daily life, more and more evidence suggests that human
emotion also spread in online social media through virtual interactions. However, the mechanism
underlying the emotion contagion, like whether different feelings diffuse unlikely or how the spread
is coupled with the social network, is rarely investigated. Indeed, because of the costly expense
and spatio-temporal limitations, it is challenging for conventional questionnaires or controlled ex-
periments. While given the instinct of collecting natural affective responses of massive connected
individuals, online social media offer us an ideal proxy to tackle this issue from the perspective of
computational social science. In this paper, based on the analysis of millions of tweets in Weibo, a
Twitter-like service in China, we surprisingly find that anger is more contagious than joy, indicat-
ing that it can sparkle more angry follow-up tweets and anger prefers weaker ties than joy for the
dissemination in social network, indicating that it can penetrate different communities and break
local traps by more sharings between strangers. We conjecture that easier contagion and weaker
ties could together make anger spread faster than joy, which is further justified by the diffusion of
bursty events with different dominant emotions. To our best knowledge, for the first time we quan-
tificationally disclose the difference between joy and anger in dissemination mechanism and our
findings would shed lights on personal anger management in human communication and collective
outrage control in cyber space.
1
I. INTRODUCTION
Emotions have a substantial effect on human decision-makings and behaviors [22]. Emo-
tions also transfers between different individuals through their communications and interac-
tions, indicating that emotion contagion, which make others around experience the similar
emotional statuses, could possibly promote social interactions [34, 37] and synchronize col-
lective behavior, especially as the individuals are embedded in social networks [42]. From
this perspective, better understanding of the emotion contagion can essentially help up-
grade emotion management and disclose collective behavior patterns. However, the detailed
mechanism of emotion contagion in context of the social network still remains unclear.
Conventional approaches like laboratory experiments are pervasively employed to testify
the existing of emotion contagion in realistic circumstances [21, 28, 41]. However, to unravel
the detailed mechanism beyond the emotion contagion is much more challenging. It is hard
for the controlled experiments to establish a large social network, stimulate different emo-
tions simultaneously and then track the propagation of emotions in real-time. Meanwhile,
in order to study the coupled dynamics between the social structure and emotion contagion,
properties like relationship strengths should be considered, while which might further import
uncontrolled contextual factors and fundamentally undermine the reliability of the experi-
ment. We argue that it is extremely difficult for conventional approaches to investigate the
spread mechanism in the context of large social network and long-term observation, and
the footprints of natural affective responses from massive individuals in online social media
indeed offer us a new and computational perspective [23, 33, 34].
It is not easy to differentiate online interactions from face-to-face communication in terms
of emotion contagion [17]. More and more evidence from both Facebook and Twitter in
recent years consistently demonstrates the existing of emotion contagion in online social
media [10, 13, 14, 25, 26, 31, 32]. Kramer et al. for the first time provide the experimental
evidence of emotion contagion in Facebook by manipulating the amount of emotional content
in the News Feed [32]. Later Ferrara and Yang disclose the evidence of emotion contagion
in Twitter [19]. Instead of controlling the content, they measure the emotional valence of
the content and show that posting positive and negative tweets both follow significant over-
exposures, indicating the spread of different feelings. Evidence from both studies suggests
that even in the absence of non-verbal cues of in-person interactions [19], emotions like
2
positive and negative feelings can still transfer from one user to others in online social media.
In fact, through posting, reposting and other virtual interactions, users in online social media
express and expose their natural emotional states into the social network in real-time as the
context evolves. Hence being an ubiquitous sensing platform, the online social media collects
emotion expression and dissemination in the most realistic and comprehensive circumstance
and thus provides us an unprecedented proxy to obtain universal insights of the underlying
mechanism of emotion contagion in social network.
However, in the previous study, emotions in online social media are usually simplified into
the positive and the negative [5, 6, 19, 32] and many fine-grained emotional states, especially
the negative ones like anger and disgust are neglected. In fact, negative feelings like anger
on the Internet might greatly dominate online bursts of societal issues or terrorist attack
and play essential roles in driving the collective behavior during the event propagation [11,
44]. Aiming at fill this vital gap, in this paper, we categorize the human emotion into
four categories [18, 44], including joy, anger, disgust and sadness and try to disclose the
mechanism underlying their spread. Fine-grained emotional states enrich the landscape
of emotion contagion and make it possible to systematically understand the underlying
mechanisms between social structures and propagation dynamics.
In this paper, we employ over 11 million tweets posted by around 100 thousand users
of a half year in Weibo, a variant of Twitter in China to perform the computational and
quantificational investigation of emotion contagion. Surprisingly, we find that among the
four emotions we study, joy and anger demonstrate significant evidence of contagion and
anger is the most contagious mood, indicating that under the same volume of exposure,
angry messages will sparkle more follow-up retweets or tweets than joy in the near future.
Besides, from the view of coupled dynamics with the social structure, different measures
of relationship strengths consistently show that anger prefers weaker ties than joy in the
dissemination, indicating that angry tweets can break local traps and arrive the global
coverage with greater odds. We conjecture that easier contagion and weaker ties could make
anger spread faster than joy in online social media, then over 40 million tweets from 616
online bursts in China further solidly testify this argument. In these sudden events, we
find that negative ones with anger as the dominating emotion arrives their diffusion peaks
with shorter intervals and higher velocities than joyful counterparts. Our findings about the
emotion contagion will be insightful for personal anger management and collective behavior
3
understanding in realistic scenarios either online or offine.
II. RESULTS
In this study, 11,753,609 tweets posted by 92,176 users from Sept., 2014 to Mar., 2015
are collected, including the following network of these users. Through a Bayesian classifier
trained in [44], each emotional tweet in our data set can be automatically labeled to joy,
anger, disgust or sadness. Then the diffusive probability and structure preferences are
accordingly investigated in the social network of around 100 thousand subjects. Finally the
realistic bursts in Weibo will be employed to testify the conjecture that anger spreads faster
than joy, in which 40,005,242 tweets of 616 different events are employed to perform the
computational analysis [36].
A. Anger is more contagious than joy
Rather than manipulating the content users receive [31], we measure the emotion con-
tagion through the extent to which a user will be influenced emotionally by the tweets she
received in Weibo during a certain period, i.e., an observation window. Similar to [19], it
is assumed that the emotion that users are exposed to in the recent past will stimulate
similar emotional status and which can be reflected by users’ newly posted tweets. Each
individual in Weibo can be embedded into a online social network with nodes being users
and directed links being following relationships, for example, if a user u(follower) follows
a user w(followee), a tie will be established from uto w, indicating that tweets posted by
wwill be flooded to uin real-time. Supposing the observation window is ∆ hours, a vector
v(tu) = (panger , pdisgust, pj oy, psadness) represents the emotion distribution of tweets exposed
to uwithin ∆ before posting the tweet tu. Note that we omit the tweets with total exposure
less than 20 tweets to ensure the reliability of the measurement. By averaging over each
dimension of all tweets’ exposure vectors, a vector vall can be obtained to reflect the baseline
of emotion distribution before any tweet is posted in Weibo. While for a certain emotion
i(i= 1,2,3 or 4), by averaging over each dimension of all tweets with emotion i, a vector
viis similarly obtained to stand for the average emotion distribution before any tweet with
emotion iis posted in Weibo. Then for emotion i, the difference (denoted as di) on dimen-
4
0.5
1.0
1.5
2.0
1 2 3 4 5 6 7 8 9 10 11 12
Difference (%)
a
47
48
49
50
51
52
1 2 3 4 5 6 7 8 9 10 11 12
Percentage (%)
b
Anger
Disgust
Joy
Sadness
Figure 1. Contagion of different emotions. (a) The contagion tendency of different
emotions declines with the length of observation window (in hours). (b) The percentage of
influenced tweets for both anger and joy rises with the growth of observation window and
the fraction of influenced tweets in anger even exceeds 50% when ∆ >4 hours.
sion ibetween vall and vican intuitively reflect the contagion tendency, i.e., the significance
of contagion, of iin Weibo. Higher diimplies that emotion iwill stimulate more tweets with
the same emotion in the near future than the average level, suggesting it is more contagious
from the perspective of emotional influence.
As can be seen in Fig. 1(a), dang er and djoy are significantly higher than that of other
emotions, which indicates that both emotions can spread in the social network, especially
for anger which possesses the highest significance as compared to the baseline. For example,
danger is about 2.1% when ∆ = 1, which means that users received 2.1% more angry tweets
than ordinary time before they post an angry message. However, the significance of contagion
is relatively low for disgust and sadness, especially for disgust, suggesting the possibility of
contagion is trivial. Because of this, in the following content we only focus on anger and joy.
Note that dialso decrease as ∆ grows, because the influence of the emotion decays with time
and thus users are easier to be influenced by messages they received in more recent hours,
which is consistent with our intuition. While even ∆ grows, anger persistently shows the
most significant difference, indicating that it is more contagious than joy and is a stronger
stimuli in emotion expression of human in the social network.
Emotion influence is closely entangled with emotion contagion. When emotion spread in
5
the social network, connected individuals could influence each other emotionally and later
might lead to the clustering of emotion, i.e., networked individuals demonstrate homophily
in sentimental statuses. For example, Bollen et. al find that happiness of users is assorta-
tive across Twitter [7] and Bliss et. al reveal that average happiness scores are positively
correlated between friends within three hops [5]. Particularly, Fan et. al even show that in
Weibo anger possesses stronger correlation than joy for friends within three hops, suggesting
that anger is more influential than joy [18]. Inspired by this, similar to the metric presented
in [19], here we also evaluate the contagion tendency of anger and joy from the perspective
of emotion influence. For each tweet tuwith emotion iposted by u, we calculate the Eu-
clidean distance ej(j= 1,2,3 or 4) between v(tu) and vj(j= 1,2,3 or 4). We define tuis
an emotionally influenced tweet, i.e., tuis stimulated by emotion icarried by tweets that u
received within ∆ if the emotion with the smallest distance is i. Then for anger and joy,
the one with higher percentage of influenced tweets will be more influential and contagious.
As shown in Fig. 1(b), anger possesses the higher percentage of influenced tweets than joy,
indicating that as compared to joy, it is easier to be stimulated by tweets that users are
exposed to. Meanwhile, as ∆ grows both emotions’ influenced percentages increase because
users are exposed to more tweets when ∆ is large, while anger persistently possesses the
higher percentage than joy and the gap is even enlarged for greater ∆.
We further investigate the emotion contagion from the view of emontionally influenced
users. Specifically, all users are sorted in the descending order of the percentage of their
influenced tweets, i.e., the faction of influenced tweets in their tweeting timelines. The
top and bottom 15% users are then defined as the high susceptible and the low susceptible,
respectively. As can be seen in Fig. 2, low susceptible users are easier to be affected by anger
while high susceptible users are more inclined to be stimulated by joy. Different from [19],
in which it is claimed that both high and low susceptible users are more susceptible to
positive feelings than the negative, however, our results demonstrate that in fact anger, one
of the negative emotions, is more susceptible than joy for low susceptible users. It means
that a fine-grained emotion category can lead to new insights in discussing the emotion
contagion and anger, out of expectation, is more contagious than joy. Meanwhile, relatively
lower number of followers for the low susceptible user indicates its higher occupation than
the high susceptible in the soical network and given the power-law like distribuiton of the
number of followers, it is further suggested that anger is more contagious for the majority
6
0
20
40
60
80
Low susceptible High susceptible
Percentage (%)
Anger
Joy
Figure 2. Influenced percentage of anger and joy in both high and low susceptible users,
respectively.
of the social network that joy.
Being more contagious and more susecptible to the majority indicate that anger could be
transfered faster than joy, because it will inspire more follow-up tweets under the same in-
tensity of stimuli. However, the mechanism underlying emotion contagion also relies heavily
on structure of the social network and thus how structure functions on contagion of joy and
anger indeed deserves further explorations.
B. Anger prefers weaker ties than joy
The dynamics of emotion contagion is essentially coupled with the underlying social
network that providing channels for disseminating the sentiment from one individual to
others. Except the willingness of posting an angry or joyful tweet after exposed to anger
or joy from followees, another key factor that determines the user’s action is that on which
relationship the contagion will happen with higher likelihood. Specifically, the prediction of
which friend will retweet the emotional post in the future will be insightful for the contagion
modeling and control.
The social relationship, or the tie in social network, can be measured by its strength, which
7
is instrumental for spreading both online and real-world human behaviors [8]. The strength
of a tie in online social media can be intuitively measured through online interactions, i.e.,
retweets, between its two ends. Here three metrics are presented to depict tie strengths
quantificationally. The first one is the proportion of common friends [38, 46], which is
defined as cij /(ki1 + kj1cij ) for the tie between users iand j, where cij is the
number of common friends, kiand kjrepresent degrees of iand j, respectively. Note that
for the metric of common friends, the social network of Weibo is converted to undirected
with each link representing the possible interaction between both ends. The second metric
is inspired by the reciprocity in Twitter-like services and higher ratio of reciprocity indicates
more trust and more significant homophily [47]. Hence for a pair of users, the proportion
of reciprocal retweets in the total flux between them is defined as the tie strength. The
third metric is the number of reweets between two ends of a tie in Weibo and greater values
stand for more frequent interactions. Note that different from the previous two metrics,
the retweet strength is time-evolving and we only count the retweets happening before the
corresponding emotional retweet. And in order to smooth the comparison between anger
and joy, the retweet strength, denoted as S, is also normalized by (SSmin)/(Smax Smin),
in which Smin and Smax separately represent the minimum and maximum values in all
obervations.
By investigating each emotional retweet, i.e., reposting an angry or joyful tweet posted
by the followee, we correlate emotion contagion with tie strengths. As can be seen in
Fig. 3, all the metrics consistently demonstrate that anger prefers weaker ties than joy in
contagion, suggesting that angry tweets spread through weak ties with greater odds than
the joyful ones. It is well known that weak ties play essential roles in diffusion of social
networks [24], especially in break the local trap caused by communities in terms of bridging
different clusters together [16, 38, 45]. For example, a typical snapshot of emotion contagion
with four communities are demonstrated in Fig. 4, in which anger disseminates through
more weak ties of inter-communities than joy. Because of this, as compared to joy, anger
has more chances to penetrate different communities through its preference on weak ties in
emotion contagion. And more communities lead to more global coverage, indicating that
anger could reach broader dissemination than joy over the same time period.
Both evidence from contagion tendency and relationship strength suggests that anger
could spread faster than joy, because it will stimulate more follow-up tweets and penetrate
8
0.00
0.01
0.02
0.03
Anger Joy
Common Friends
a
0.00
0.05
0.10
0.15
Anger Joy
Reciprocity (%)
b
0.000
0.005
0.010
0.015
Anger Joy
Retweets
c
Figure 3. Anger prefers more weak ties than joy. Three different metrics are averaged
over all emotional retweets in the dataset. Lower metrics for anger suggests that in
contagion anger disseminates through weaker ties than joy. Error bars represent standard
errors in (a) and (c), while in (b) there is no stand error beause the reciprocity is just a
ratio obtained from all emotional retweets.
more communities in the social network of Weibo. Next direct evidence from online bursts
in Weibo will be demonstrated to testify this conjecture further.
C. Anger spreads faster than joy
No moods are created equally online [15], the difference in contagion of positive and
negative feelings is a trending but controversial topic for years. Berger and Milkman reveal
that more positive content is more viral than negative content in study of NY Times [4]. Tadi
and ˇ
Suvakov find that for human-like bots in online social network, positive emotion bots are
more effective than the negative ones [42]. Wu et al. even claim that bad news containing
more negative words fades more rapidly in Twitter [43]. While Chmiel et. al point out that
negative sentiments boost user activity in BBC forum [11]. Pfitzner et. al state that users
tend to retweet tweets with high emotional diversity [40]. Ferrara and Yang demonstrate
that though people are more likely to share positive content in Twitter, negative messages
9
Figure 4. Emotion contagion in a sampled snapshot of the Weibo network with four
communities and links with more angry retweets will be orange and otherwise will be
green. It can be seen that anger prefers more weak ties bridging different communities
than joy. Besides, the left two communities dominated by anger and thus a large percent
of messages disseminate inter communities. While for the two right joy-dominated
communities, emotion tends to diffuse inside the community.
spread faster than positive ones at the level of contents [20]. Thus Hansen et al. conclude
that the relation between emotion and virality is much complicated [27]. Here we argue that
fine-grained division of human emotion, especially the negative, will enrich the background
of investigating the contagion difference. And in the meantime, explicit definitions of fast
spread will further facilitate the elimination of debates on this issue.
Over 600 bursty events are extracted from Weibo and for each event, the emotion occu-
pying more than 60% emotional tweets will be defined as the dominant emotion of the event.
In total we get 37 anger-dominated events and 200 joy-dominated events, as can be seen
Fig. 5(a), in which the volume of emotional tweets fluctuates with time and anger takes over
the majority. For an online burst, the fast spread means the volume of emotional tweets
grow quickly in the period from the awakening instant to the diffusion peak. Through a
parameterless approach presented in [30], we can precisely locate the instant of awakening
10
0
200
400
600
800
0 10 20 30 40 50
τ
Volume
Anger
Disgusting
Joy
Sadness
315 party
0
500
1000
1500
0 10 20 30 40
τ
Volume
0
2000
4000
6000
0 10 20 30 40
τ
Volume
a
b c
Figure 5. Illustration of the online bursts. (a) An example of an anger-dominated event
and the newly posted emotional tweets varies as time τevolves. (b) Examples of locating
the awakening and the peak for both anger and joy dominated events. xAdenotes the
instant of awakening, xPdenotes the peaking time and yAand yPstand for the volume of
emotional tweets at the instants of awakening and peak, respectively. The slope of the line
(blue) between points (xA, yA) and (xP, yP) can reflect the averaged velocity.
and peak. Then how fast the spread is can be reflected by the interval between the awak-
ening and the peak (shorter intervals suggest faster spread) or the averaged velocity of the
growth in diffusion from awakening to the peak (higher velocities indicate faster spread).
Specifically, as demonstrated in Fig. 5(b), regarding to interval, 1/(xPxA) can reflect how
fast the spread is, while with respect to averaged velocity, it can be defined as the slope,
i.e., (yPyA)/(xPxA) and for both measures higher values suggest faster spread. The
averaged results over all anger and joy dominated events are listed in Table I and it can be
seen that both metrics testify significantly that anger-dominated events incline to arrive the
peak from the awakening more quickly than joy. We can conclude that anger indeed spreads
faster than the joy in social media like Weibo.
11
Measures Anger Joy
Time-difference 0.179 0.102
Slope 0.329 0.238
TABLE I. Averaged values over all anger and joy dominated events.
III. DISCUSSION
Bad is always stronger than good [3]. From the previous studies, it is profoundly disclosed
that anger is more influential than joy in Weibo [18] and the negative contents in Twitter
spread faster than positive ones [20]. However, in this study, by finely splitting the negative
feelings into anger, disgust and sadness, we for the first time offer the empirical evidence
that anger spreads faster than joy in social media, implying that anger on the Internet
should be considered primarily in either personal emotion management or collective mood
understanding.
The way how anger is expressed and experienced on the Internet is arousing attention.
It has been found from self-reports of rant-site visitors that posting anger produces relaxed
feelings immediately [35], which makes posting anger an effective way of self-regulation
in mood. However, considering the easy contagion of anger, more angry expressions on the
Internet might arouse negative shifts in mood of the crowd that connected to posters [35, 39].
Moreover, anger’s preference of weak ties will make it likely to spread to “strangers” in the
Internet. The users want to ease anger by posting on the Internet should be suggested to
underhand the possible impact to their online social network. Even in offline scenario, like
the “road rage” [1] at rush hours of China, in which anger spreads quickly between strangers
and might cause aggressive driving or even accidents. We suggest that in personal anger
management, the unexpected contagion to strangers should be seriously considered.
The online social media has already been the most ubiquitous platform for collective
intelligence, in which various signals generated from massive connected individuals provide
the foundation of collective behavior understanding. However, how emotion contagion affects
the aggregation of the individual behavior, particularly the individual intelligence, is rarely
considered. In fact from our findings, emotion spread, especially the anger’s fast contagion,
might imply a lot to the collective behavior in the cyber space, especially the crowdsourcing
12
results. It is even stated that emotion, such as anger, can be a threat to reasoning and
arguing [34]. For example, the outrages of massive emotional individuals, which would badly
bias the public opinion, might just come from the fast spread of anger, but not because the
event itself. Meanwhile, fast contagion of anger also offers a new perspective to picture
the emotional behavior of the crowd on the Internet. We suggest that for scenarios like
crowdsourcing or collective behavior understanding, angry users should be carefully treated
to omit their possible impact on the fair judgment of the observations in the epoch of big
data. Besides, diminishing weak ties will function effectively in controlling the diffusion of
the Internet outrages and make the crowd rational and smart.
Contrarily, happiness is believed to unify and clustering the community [12] and our
finding about the joy’s preference on less weak ties also supports this (see Fig.4), implying
that strong ties inside the community disseminate more joyful content in online social media.
Meanwhile, self-reports from Facebook users also testify that communication with strong ties
is associated with improvements in well-being [9], which further testifies that our findings
from the computational view is solid.
IV. CONCLUSION
Instead of self-reports in controlled experiments, the natural and emotional postings
in Weibo social network are collected to investigate the detailed mechanism of emotion
contagion from a new view of computational social science. For the first time, we offer the
solid evidence for the fact that anger spreads faster than joy in social media because of
being more contagious and more preferential on weak ties. Our findings shed lights on both
personal anger management and collective behavior understanding.
This study has inevitable limitations. It is generally accepted that emotion expression is
culture dependent and demographics like gender also matter [2, 29], suggesting that exploring
how anger spreads in Twitter and how the male and the female response differently to
emotion contagion will be of great significance in the future work.
13
ACKNOWLEDGMENT
This work was supported by the National Natural Science Foundation of China (Grant
Nos. 71501005, 71531001 and 61421003) and the fund of the State Key Lab of Software
Development Environment (Grant Nos. SKLSDE-2015ZX-05 and SKLSDE-2015ZX-28).
[1] Road rage. https://en.wikipedia.org/wiki/Road_rage, 2016.
[2] Richard P. Bagozzi, Nancy Wong, and Youjae Yi. The role of culture and gender in the
relationship between positive and negative affect. COGNITION AND EMOTION, 13(6):641–
672, 1983.
[3] Roy F. Baumeister, Ellen Bratslavsky, Catrin Finkenauer, and Kathleen D. Vohs. Bad is
stronger than good. Review of General Psychology, 5(4):323–370, 2001.
[4] Jonah Berger and Katherine L. Milkman. Social transmission, emotion, and the virality of
online content. Wharton Research Paper, 2010.
[5] Catherine A Bliss, Isabel M Kloumann, Kameron Decker Harris, Christopher M Danforth, and
Peter Sheridan Dodds. Twitter reciprocal reply networks exhibit assortativity with respect to
happiness. Journal of Computational Science, 3(5):388–397, 2012.
[6] Johan Bollen, Bruno Gon¸calves, Guangchen Ruan, and Huina Mao. Happiness is assortative
in online social networks. Artificial life, 17(3):237–251, 2011.
[7] Johan Bollen, Huina Mao, and Xiaojun Zeng. Twitter mood predicts the stock market. Journal
of Computational Science, 2(1):1–8, 2011.
[8] Robert M. Bond, Christopher J. Fariss, Jason J. Jones, Adam D. I. Kramer, Cameron Marlow,
Jaime E. Settle, and James H. Fowler. A 61-million-person experiment in social influence and
political mobilization. Nature, 489(7415):295–298, 2012.
[9] Moira Burke and Robert E. Kraut. The relationship between facebook use and well-being
depends on communication type and tie strength. Journal of Computer-Mediated Communi-
cation, 21:265281, 2016.
[10] Anna Chmiel, Julian Sienkiewicz, Mike Thelwall, Georgios Paltoglou, Kevan Buckley, Arvid
Kappas, and Janusz A Ho lyst. Collective emotions online and their influence on community
life. PloS ONE, 6(7):e22207, 2011.
14
[11] Anna Chmiel, Pawel Sobkowicz, Julian Sienkiewicz, Georgios Paltoglou, Kevan Buckley, Mike
Thelwall, and Janusz A Ho lyst. Negative emotions boost user activity at BBC forum. Physica
A: statistical mechanics and its applications, 390(16):2936–2944, 2011.
[12] Nicholas A. Christakis and James H. Fowler. Connected: The Surprising Power of Our Social
Networks and How They Shape Our Lives. Back Bay Books, 2011.
[13] Lorenzo Coviello, Yunkyu Sohn, Adam DI Kramer, Cameron Marlow, Massimo Franceschetti,
Nicholas A Christakis, and James H Fowler. Detecting emotional contagion in massive social
networks. PloS one, 9(3):e90315, 2014.
[14] Linh Dang-Xuan and Stefan Stieglitz. Impact and diffusion of sentiment in political
communication-an empirical analysis of political weblogs. In ICWSM, 2012.
[15] Munmun De Choudhury, Scott Counts, and Michael Gamon. Not all moods are created equal!
exploring human emotional states in social media. In Proceedings of the Sixth International
Conference on Weblogs and Social Media, ICWSM ’12, 2012.
[16] Pasquale De Meo, Emilio Ferrara, Giacomo Fiumara, and Alessandro Provetti. On facebook,
most ties are weak. Communications of the ACM, 57:78–84, 2014.
[17] Daantje Derks, Agneta H. Fischer, and Arjan E.R. Bos. The role of emotion in computer-
mediated communication: A review. Computers in Human Behavior, 24:766–785, 2008.
[18] Rui Fan, Jichang Zhao, Yan Chen, and Ke Xu. Anger is more influential than joy: sentiment
correlation in weibo. PLoS ONE, 9:e110184, 2014.
[19] Emilio Ferrara and Zeyao Yang. Measuring emotional contagion in social media. PloS ONE,
10(11):e0142390, 2015.
[20] Emilio Ferrara and Zeyao Yang. Quantifying the effect of sentiment on information diffusion
in social media. PeerJ Computer Science, 1:e26, 2015.
[21] James H Fowler and Nicholas A Christakis. Dynamic spread of happiness in a large social
network: longitudinal analysis over 20 years in the framingham heart study. BMJ, 337:a2338,
2008.
[22] Uri Gneezy and Alex Imas. Materazzi effect and the strategic use of anger in competitive
interactions. Proceedings of the National Academy of Sciences, 111(4):1334–1337, 2014.
[23] Scott A. Golder and Michael W. Macy. Diurnal and seasonal mood vary with work, sleep,
and daylength across diverse cultures. Science, 333(6051):1878–1881, 2011.
15
[24] Mark S Granovetter. The strength of weak ties. American journal of sociology, pages 1360–
1380, 1973.
[25] Anatoliy Gruzd, Sophie Doiron, and Philip Mai. Is happiness contagious online? a case
of twitter and the 2010 Winter Olympics. In System Sciences (HICSS), 2011 44th Hawaii
International Conference on, pages 1–9. IEEE, 2011.
[26] Jamie Guillory, Jason Spiegel, Molly Drislane, Benjamin Weiss, Walter Donner, and Jeffrey
Hancock. Upset now?: Emotion contagion in distributed groups. In Proceedings of the SIGCHI
Conference on Human Factors in Computing Systems, CHI ’11, pages 745–748, New York,
NY, USA, 2011. ACM.
[27] Lars Kai Hansen, Adam Arvidsson, Finn Aarup Nielsen, Elanor Colleoni, and Michael Et-
ter. Good Friends, Bad News - Affect and Virality in Twitter, pages 34–43. Springer Berlin
Heidelberg, Berlin, Heidelberg, 2011.
[28] Elaine Hatfield, John T Cacioppo, and Richard L Rapson. Emotional contagion. Curr Dir
Psych Sci, 2(3):96–100, 1993.
[29] Yue Hu, Jichang Zhao, and Junjie Wu. Emoticon-based ambivalent expression: A hidden
indicator for unusual behaviors in weibo. PLoS ONE, 11:e0147079, 2016.
[30] Qing Ke, Emilio Ferrara, Filippo Radicchi, and Alessandro Flammini. Defining and identifying
sleeping beauties in science. Proceedings of the National Academy of Sciences, 112(24):7426–
7431, 2015.
[31] Adam D.I. Kramer. The spread of emotion via facebook. In Proceedings of the SIGCHI
Conference on Human Factors in Computing Systems, CHI ’12, pages 767–770, New York,
NY, USA, 2012. ACM.
[32] Adam DI Kramer, Jamie E Guillory, and Jeffrey T Hancock. Experimental evidence of
massive-scale emotional contagion through social networks. Proceedings of the National
Academy of Sciences, 111(24):8788–8790, 2014.
[33] David Lazer, Alex Pentland, Lada Adamic, Sinan Aral, Albert-L´aszl´o Barab´asi, Devon Brewer,
Nicholas Christakis, Noshir Contractor, James Fowler, Myron Gutmann, Tony Jebara, Gary
King, Michael Macy, Deb Roy, and Marshall Van Alstyne. Computational social science.
Science, 323(5915):721–723, 2009.
[34] Stacy Marsella and Jonathan Gratch. Computationally modeling human emotion. Commu-
nications of The ACM, 57(12):56–67, 2014.
16
[35] Ryan C Martin, Kelsey Ryan Coyier, Leah M Vansistine, and Kelly Schroeder. Anger on the
internet: The perceived value of rant-sites. Cyberpsychology, Behavior, and Social Networking,
16(2):119, 2013.
[36] All the data sets will be publicly available soon. Any information please contact
jichang@buaa.edu.cn.
[37] Lauri Nummenmaa, Enrico Glerean, Mikko Viinikainen, Iiro P. Jskelinen, Riitta Hari, and
Mikko Sams. Emotions promote social interaction by synchronizing brain activity across
individuals. Proceedings of the National Academy of Sciences, 109(24):9599–9604, 2012.
[38] J-P Onnela, Jari Saram¨aki, Jorkki Hyv¨onen, Gy¨orgy Szao, David Lazer, Kimmo Kaski, J´anos
Kert´esz, and A-L Barab´asi. Structure and tie strengths in mobile communication networks.
Proceedings of the National Academy of Sciences, 104(18):7332–7336, 2007.
[39] Minsu Park, Chiyoung Cha, and Meeyoung Cha. Depressive moods of users portrayed in
twitter. In Proceedings of the ACM SIGKDD Workshop on Healthcare Informatics (HI-KDD),
2012.
[40] Ren´e Pfitzner, Antonios Garas, and Frank Schweitzer. Emotional divergence influences infor-
mation spreading in twitter. In Proceedings of the Sixth International AAAI Conference on
Weblogs and Social Media, pages 2–5, 2012.
[41] J Niels Rosenquist, James H Fowler, and Nicholas A Christakis. Social network determinants
of depression. Molecular psychiatry, 16(3):273–281, 2011.
[42] Bosiljka Tadi and Milovan uvakov. Can human-like bots control collective mood: agent-
based simulations of online chats. Journal of Statistical Mechanics: Theory and Experiment,
2013(10):P10014, 2013.
[43] Shaowei Wu, Chenhao Tan, Jon Kleinberg, and Michael Macy. Does bad news go away faster?
In Proceedings of the Fifth International Conference on Weblogs and Social Media, ICWSM
’11, 2011.
[44] Jichang Zhao, Li Dong, Junjie Wu, and Ke Xu. Moodlens: An emoticon-based sentiment
analysis system for Chinese tweets. In Proceedings of the 18th ACM SIGKDD International
Conference on Knowledge Discovery and Data Mining, KDD ’12, pages 1528–1531, New York,
NY, USA, 2012. ACM.
[45] Jichang Zhao, Junjie Wu, Xu Feng, Hui Xiong, and Ke Xu. Information propagation in online
social networks: A tie strength perspective. Knowledge And Information System(KAIS),
17
32:589–608, 2012.
[46] Jichang Zhao, Junjie Wu, and Ke Xu. Weak ties: Subtle role of information diffusion in online
social networks. Phys. Rev. E, 82:016105, Jul 2010.
[47] Yu-Xiao Zhu, Xiao-Guang Zhang, Gui-Quan Sun, Ming Tang, Tao Zhou, and Zi-Ke Zhang.
Influence of reciprocal links in social networks. PloS ONE, 9(7):e103007, 2014.
18
... One approachreferred to as the window of interest approachis 13 to look at the content perceivers could have been exposed to and assume that it was perceived more or less equally across users. For example, in a recent study, researchers [49] estimated the content observed by Twitter users by looking at the average emotional content of tweets produced by a certain perceivers' followees during the hour preceding their posting [7,48,50]. A second approach is the overall emotional variance approach. ...
... One approachreferred to as the window of interest approachis to look at the content perceivers could have been exposed to and assume that it was perceived more or less equally across users. For example, in a recent study, researchers [49] estimated the content observed by Twitter users by looking at the average emotional content of tweets produced by a certain perceivers' followees during the hour preceding their posting [7,48,50]. A second approach is the overall emotional variance approach. ...
... We currently know of one study that found that negative emotions, and particularly anger, lead to stronger contagious on digital media [50]. Interestingly, the methods used in this paper were similar to another research project that found stronger contagion for positive emotions [49]. ...
Article
Full-text available
People spend considerable time on digital media, and are thus often exposed to expressions of emotion by other people. This exposure can lead their own emotion expressions becoming more similar to those of others, a process we refer to as 'digital emotion contagion'. This article reviews the growing literature on digital emotion contagion. After defining emotion contagion, we suggest that one unique feature of digital emotion contagion is that it is mediated by digital media platforms that are motivated to upregulate user emotions. We then turn to measurement, and consider the challenges of demonstrating that digital emotion contagion has occurred, and how these challenges have been addressed. Finally, we call for a greater focus on understanding when emotion contagion effects are likely to be strong versus weak or nonexistent.
... One approachreferred to as the window of interest approachis to look at the content perceivers could have been exposed to and assume that it was perceived more or less equally across users. For example, in a recent study, researchers [49] estimated the content observed by Twitter users by looking at the average emotional content of tweets produced by a certain perceivers' followees during the hour preceding their posting [7,48,50]. A second approach is the overall emotional variance approach. ...
... We currently know of one study that found that negative emotions, and particularly anger, lead to stronger contagious on digital media [50]. Interestingly, the methods used in this paper were similar to another research project that found stronger contagion for positive emotions [49]. ...
... Yet the relationship between the strength of network connection and contagion seems to depend on the type of expressed emotion [73]. In the first study that tested this question, [50] researchers compared how contagion of anger and joy were influenced by the strength of network connection. They found that anger contagion was stronger in weaker ties compared to joy. ...
Preprint
Full-text available
People spend considerable time on digital media during which they are frequently exposed to numerous expressions of emotions. These emotions can lead to emotion contagion, in which the exposure to others’ emotions makes one’s own emotions become more similar to others’ emotions. This paper reviews the growing literature on digital emotion contagion. After defining emotion contagion, we consider how unique aspects of digital media may influence digital emotion contagion. We then turn to measurement, and consider the challenges of demonstrating that digital emotion contagion has actually occurred, and how these challenges have been addressed. Finally, we call for a shift in research focus towards predicting the factors that determine when emotion contagion effects will be strong versus weak or non-existent.
... Moreover, researchers found various other forms of emotional contagion. For example, Facebook users who received fewer positive expressions produced fewer positive posts and more negative posts (Fan et al., 2016;Kramer et al., 2014). Therefore, emotional contagion can influence individuals' current mood. ...
... First, we examined the relation between emotional contagion and mood state among children. A typical limitation of previous studies has been that most of the research about emotional contagion has focused on adults (e.g., Kramer et al., 2014;Wu et al., 2014;Fan et al., 2016;Volante et al., 2016). According to computer searches of PsycInfo, 417 of 690 studies focused on adults, 53 studies focused on adolescents, and 54 studies focused on children. ...
Article
Full-text available
To investigate the universality and cultural specificity of emotion processing in children from three different ethnic groups (Han, Jingpo, and Dai), we administered three questionnaires, including the emotional contagion scale, emotion regulation scale, and the Chinese mood adjective check list, to 1,362 ethnic Han, Dai, and Jingpo participants (Mage = 13.78 years). Structural equation modeling was used to examine the universality and cultural specificity in the relations among emotional contagion, emotion regulation, and mood state. The results revealed that emotion regulation mediated the relation between emotional contagion and mood state: positive emotional contagion increased positive mood state and decreased negative mood state by the mediated role of reappraisal, negative emotional contagion decreased positive contagion and increased negative mood state by the inconsistent mediated role of reappraisal; negative contagion increased negative mood state by the mediated role of suppression. We found both universality and cultural specificity in the relations among emotional contagion, emotion regulation, and mood state. Regarding cultural specificity, among Dai and Jingpo participants, negative contagion positively predicted reappraisal, while for Han participants, it did not; Jingpo participants demonstrated a weaker negative relation between reappraisal and negative mood state, and a stronger positive relation between negative contagion and suppression; and Dai participants were the only ethnic group that showed a negative connection between negative contagion and positive mood state. Regarding emotion universality, the three ethnic groups all showed positive relations between negative contagion and negative mood, and between suppression and negative mood; additionally, positive contagion positively predicted positive mood state, mediated by reappraisal. Thus, some emotion processes are universal and others more specific. In this paper, we discuss universal emotion processes and ethnic cultural differences in these emotion processes.
... Nonetheless, only a few studies have done a comprehensive analysis to investigate which kind of emotion is more contagious in the digital environment. Fan et al. (2016) recently acknowledged this limitation and used four emotional categories (i.e., anger, disgust, joy, and sadness) to analyze Weibo posts; they concluded that anger was the most-spread emotion, especially in weak ties. To partly replicate but further advance their study, we adopt a Chinese-context emotion categorization which contains seven categories: good, joy, anger, sadness, fear, surprise, and disgust (Xu et al., 2008). ...
Article
Full-text available
This study collected 2 million posts and reposts regarding the early stage of COVID-19 in China on Weibo from 26 December 2019 to 29 February 2020. Emotion analysis and social network analysis were used to examine the flow of emotional messages (emotion flow) by comparing them with the flow of general messages (information flow). Results indicated that both emotional messages and general messages present a multilayer diffusion pattern and follow network step flow models. In our dataset, emotion network has a higher transmission efficiency than information network; officially verified accounts were more likely to become super-spreaders of emotional messages; good emotions were predominant but isolated from other six emotions (joy, sadness, fear, disgust, surprise, anger) in online discussions; finally, government played a vital role in spreading good emotions.
... The existence of emotional contagion on social media has also been demonstrated in previous studies, especially the contagion of negative emotions, such as hatred and anger (Del Vicario et al., 2016). A negative emotion like stigma can thus be spread and internalized more rapidly than it can as a positive emotion (Fan, Xu, & Zhao, 2016). The negative aspect of feelings can result in withdrawal from social support and higher feelings of loneliness (Gray, 2002). ...
Article
Full-text available
Gay dating apps have become increasingly popular and been intensively used. However, studies that closely examine the effects of gay dating app usage on mental health outcomes are few. This study tackles an unsolved question on whether gay dating apps alleviate or exacerbate loneliness through perceptions of sexuality stigma. A cross-sectional online survey from 371 Chinese gay or bisexual men found that the intensity of gay dating app usage is positively associated with loneliness. The association between gay dating app usage intensity and loneliness is sequentially mediated by perceived sexuality stigma and internalized sexuality stigma. That is, increased intensity of gay dating app usage strengthens perceived sexuality stigma, which in turn reinforces internalized sexuality stigma, further exacerbating loneliness. This study echoes the "alone together" phenomena and uncovers the feasibility of using stigma perceptions as a mechanism to understand the mental health of gay dating app users.
... Public outrage is easily expressed and rapidly transmitted in the digital age Crockett, 2017;Fan et al., 2016). Here, we find that, when conveyed in a believable fashion (e.g., via information suggesting growing, rather than uniformly high public anger), ...
Article
Media coverage of climate protests within the United States and internationally has shown growing public frustration about governmental responses to climate change. But what are the effects of conveying that people are angry? And how do they contrast with more traditional norm messages about climate policy support? Here, we investigate whether social norm messaging about collective anger can impact perceptions of consensus and public support for climate mitigation. In a pilot study and two survey experiments (total N=1529), we find that relative to control messages, normative appeals that convey growing public anger about U.S. inaction on climate change (i.e., dynamic anger consensus messaging) can enhance Americans’ consensus estimates of other Americans’ climate-related beliefs and support for mitigation policy, and expectations for future climate-mitigating collective action. Moreover, exposure to a dynamic anger consensus message led to similar estimates of Americans’ policy support and belief in climate change as messages that explicitly conveyed public support for climate mitigation (Study 1) or consensus belief in anthropogenic climate change (Study 2). When tailored for a cross-partisan audience, anger consensus messaging was also effective in bolstering personal support for climate mitigation (Study 2). Notably, similar effects were observed across partisan groups. These findings suggest that, by signaling shared motivational states, emotion consensus appeals can enhance expectations for social change, with the potential to mobilize bipartisan support for climate mitigation.
... Some social media users are motivated to express strong emotions and attitudes, a strategy that attracts attention on a platform in which users compete for attention (Goldenberg & Gross, 2020). Likewise, misogynistic attitudes are characterized by anger (Parrott & Zeichner, 2003) and aggressive humor (Siebler, Sabelus, & Bohner, 2008), and anger spreads faster than joy among social media users (Fan, Xu, & Zhao, 2016). ...
Article
How online social behavior covaries with real-world outcomes remains poorly understood. We examined the relationship between the frequency of misogynistic attitudes expressed on Twitter and incidents of domestic and family violence that were reported to the Federal Bureau of Investigation. We tracked misogynistic tweets in more than 400 areas across 47 American states from 2013 to 2014. Correlation and regression analyses found that misogynistic tweets were related to domestic- and family-violence incidents in those areas. A cross-lagged model showed that misogynistic tweets positively predicted domestic and family violence 1 year later; however, this effect was small. Results were robust to several known predictors of domestic violence. Our findings identify geolocated online misogyny as co-occurring with domestic and family violence. Because the longitudinal relationship between misogynistic tweets and domestic and family violence was small and conducted at the societal level, more research with multilevel data might be useful in the prediction of future violence.
... And these confessions have been echoed with criticism and studies from others. Social media enables negative messages to be distributed farther and faster (Vosoughi et al., 2018) and its affordances enable anger to spread contagiously (Fan et al., 2016). The "incentive structures and social cues of algorithm-driven social media sites" amplify the anger of users over time until they "arrive at hate speech" (Fisher and Taub, 2018). ...
Article
Full-text available
Hate speech and toxic communication online is on the rise. Responses to this issue tend to offer technical (automated) or non-technical (human content moderation) solutions, or see hate speech as a natural product of hateful people. In contrast, this article begins by recognizing platforms as designed environments that support particular practices while discouraging others. In what ways might these design architectures be contributing to polarizing, impulsive, or antagonistic behaviors? Two platforms are examined: Facebook and YouTube. Based on engagement, Facebook’s Feed drives views but also privileges incendiary content, setting up a stimulus–response loop that promotes outrage expression. YouTube’s recommendation system is a key interface for content consumption, yet this same design has been criticized for leading users towards more extreme content. Across both platforms, design is central and influential, proving to be a productive lens for understanding toxic communication.
... Negative and active emotions not always can lead to dangerous behaviour; the expression of reaction is also influenced by a person's emotional education, temperament, social status, etc. [22]. Negative emotions spread in the crowd a lot faster than positive; thus, fast negative emotion detection is vital for threat prevention [23], [24]. ...
Article
Full-text available
To distinguish individuals with dangerous abnormal behaviours from the crowd, human characteristics (e.g., speed and direction of motion, interaction with other people), crowd characteristics (such as flow and density), space available to individuals, etc. must be considered. The paper proposes an approach that considers individual and crowd metrics to determine anomaly. An individual’s abnormal behaviour alone cannot indicate behaviour, which can be threatening toward other individuals, as this behaviour can also be triggered by positive emotions or events. To avoid individuals whose abnormal behaviour is potentially unrelated to aggression and is not environmentally dangerous, it is suggested to use emotional state of individuals. The aim of the proposed approach is to automate video surveillance systems by enabling them to automatically detect potentially dangerous situations.
... However, despite much recent work on mimetic theory across diverse areas, imitation on social media remains largely seen as secondary block of human behavior. Many studies largely focused on the contagion aspects on social media based on emotions such as anger versus joy [11], and users' personality on content engagement [12] and ...
Article
Full-text available
Social media are used as main discussion channels by millions of individuals every day. The content individuals produce in daily social-media-based micro-communications, and the emotions therein expressed, may impact the emotional states of others. A recent experiment performed on Facebook hypothesized that emotions spread online, even in absence of non-verbal cues typical of in-person interactions, and that individuals are more likely to adopt positive or negative emotions if these are over-expressed in their social network. Experiments of this type, however, raise ethical concerns, as they require massive-scale content manipulation with unknown consequences for the individuals therein involved. Here, we study the dynamics of emotional contagion using Twitter. Rather than manipulating content, we devise a null model that discounts some confounding factors (including the effect of emotional contagion). We measure the emotional valence of content the users are exposed to before posting their own tweets. We determine that on average a negative post follows an over-exposure to 4.34% more negative content than baseline, while positive posts occur after an average over-exposure to 4.50% more positive contents. We highlight the presence of a linear relationship between the average emotional valence of the stimuli users are exposed to, and that of the responses they produce. We also identify two different classes of individuals: highly and scarcely susceptible to emotional contagion. Highly susceptible users are significantly less inclined to adopt negative emotions than the scarcely susceptible ones, but equally likely to adopt positive emotions. In general, the likelihood of adopting positive emotions is much greater than that of negative emotions.
Article
Full-text available
Social media has become the main vehicle of information production and consumption online. Millions of users every day log on their Facebook or Twitter accounts to get updates and news, read about their topics of interest, and become exposed to new opportunities and interactions. Although recent studies suggest that the contents users produce will affect the emotions of their readers, we still lack a rigorous understanding of the role and effects of contents sentiment on the dynamics of information diffusion. This work aims at quantifying the effect of sentiment on information diffusion, to understand: (i) whether positive conversations spread faster and/or broader than negative ones (or vice-versa); (ii) what kind of emotions are more typical of popular conversations on social media; and, (iii) what type of sentiment is expressed in conversations characterized by different temporal dynamics. Our findings show that, at the level of contents, negative messages spread faster than positive ones, but positive ones reach larger audiences, suggesting that people are more inclined to share and favorite positive contents, the so-called positive bias . As for the entire conversations, we highlight how different temporal dynamics exhibit different sentiment patterns: for example, positive sentiment builds up for highly-anticipated events, while unexpected events are mainly characterized by negative sentiment. Our contribution represents a step forward to understand how the emotions expressed in short texts correlate with their spreading in online social ecosystems, and may help to craft effective policies and strategies for content generation and diffusion.
Article
Full-text available
There has been rapid growth of cross-disciplinary research employing computational methods to understand human behavior, as well as facilitate interaction between people and machines, with work on computational models of emotion becoming an important component. Modeling appraisal theory in an agent provides an interesting perspective on the relation between emotion and rationality. Appraisal theory argues that emotion serves to generalize stimulus response by providing more general ways to characterize types of stimuli in terms of classes of viable organism responses. Assessments such as desirability, coping potential, unexpectedness, and causal attribution are clearly relevant to any social agent, whether deemed emotional or not. Having been characterized in a uniform fashion, the appraisal results coordinate system-wide coping responses that serve to guide the agent's specific responses to the eliciting event, essentially helping the agent find its ecological niche. Models designed to inform intelligent systems might avoid some of the seemingly irrational application of coping humans adopt.
Article
Full-text available
Recent decades have witnessed online social media being a big-data window for quantificationally testifying conventional social theories and exploring much detailed human behavioral patterns. In this paper, by tracing the emoticon use in Weibo, a group of hidden "ambivalent users" are disclosed for frequently posting ambivalent tweets containing both positive and negative emotions. Further investigation reveals that this ambivalent expression could be a novel indicator of many unusual social behaviors. For instance, ambivalent users with the female as the majority like to make a sound in midnights or at weekends. They mention their close friends frequently in ambivalent tweets, which attract more replies and thus serve as a more private communication way. Ambivalent users also respond differently to public affairs from others and demonstrate more interests in entertainment and sports events. Moreover, the sentiment shift of words adopted in ambivalent tweets is more evident than usual and exhibits a clear "negative to positive" pattern. The above observations, though being promiscuous seemingly, actually point to the self regulation of negative mood in Weibo, which could find its base from the emotion management theories in sociology but makes an interesting extension to the online environment. Finally, as an interesting corollary, ambivalent users are found connected with compulsive buyers and turn out to be perfect targets for online marketing.
Article
Full-text available
Emotional states can be transferred to others via emotional contagion, leading people to experience the same emotions without their awareness. Emotional contagion is well established in laboratory experiments, with people transferring positive and negative emotions to others. Data from a large real-world social network, collected over a 20-y period suggests that longer-lasting moods (e.g., depression, happiness) can be transferred through networks [Fowler JH, Christakis NA (2008) BMJ 337:a2338], although the results are controversial. In an experiment with people who use Facebook, we test whether emotional contagion occurs outside of in-person interaction between individuals by reducing the amount of emotional content in the News Feed. When positive expressions were reduced, people produced fewer positive posts and more negative posts; when negative expressions were reduced, the opposite pattern occurred. These results indicate that emotions expressed by others on Facebook influence our own emotions, constituting experimental evidence for massive-scale contagion via social networks. This work also suggests that, in contrast to prevailing assumptions, in-person interaction and nonverbal cues are not strictly necessary for emotional contagion, and that the observation of others' positive experiences constitutes a positive experience for people.
Article
PERVASIVE SOCIO-TECHNICAL NETWORKS bring new conceptual and technological challenges to developers and users alike. A central research theme is evaluation of the intensity of relations linking users and how they facilitate communication and the spread of information. These aspects of human relationships have been studied extensively in the social sciences under the framework of the “strength of weak ties” theory proposed by Mark Granovetter. Some research has considered whether that theory can be extended to online social networks like Facebook, suggesting interaction data can be used to predict the strength of ties. The approaches being used require handling user-generated data that is often not publicly available due to privacy concerns. Here, we propose an alternative defnition of weak and strong ties that requires knowledge of only the topology of the social network (such as who is a friend of whom on Facebook), relying on the fact that online key insights.
Article
An extensive literature shows that social relationships influence psychological well-being, but the underlying mechanisms remain unclear. We test predictions about online interactions and well-being made by theories of belongingness, relationship maintenance, relational investment, social support, and social comparison. An opt-in panel study of 1,910 Facebook users linked self-reported measures of well-being to counts of respondents' Facebook activities from server logs. Specific uses of the site were associated with improvements in well-being: Receiving targeted, composed communication from strong ties was associated with improvements in well-being while viewing friends' wide-audience broadcasts and receiving one-click feedback were not. These results suggest that people derive benefits from online communication, as long it comes from people they care about and has been tailored for them.
Article
A Sleeping Beauty (SB) in science refers to a paper whose importance is not recognized for several years after publication. Its citation history exhibits a long hibernation period followed by a sudden spike of popularity. Previous studies suggest a relative scarcity of SBs. The reliability of this conclusion is, however, heavily dependent on identification methods based on arbitrary threshold parameters for sleeping time and number of citations, applied to small or monodisciplinary bibliographic datasets. Here we present a systematic, large-scale, and multidisciplinary analysis of the SB phenomenon in science. We introduce a parameter-free measure that quantifies the extent to which a specific paper can be considered an SB. We apply our method to 22 million scientific papers published in all disciplines of natural and social sciences over a time span longer than a century. Our results reveal that the SB phenomenon is not exceptional. There is a continuous spectrum of delayed recognition where both the hibernation period and the awakening intensity are taken into account. Although many cases of SBs can be identified by looking at monodisciplinary bibliographic data, the SB phenomenon becomes much more apparent with the analysis of multidisciplinary datasets, where we can observe many examples of papers achieving delayed yet exceptional importance in disciplines different from those where they were originally published. Our analysis emphasizes a complex feature of citation dynamics that so far has received little attention, and also provides empirical evidence against the use of short-term citation metrics in the quantification of scientific impact.
Article
Emotions have ubiquitous effects in human affairs. Vivian Gornick, in Fierce Attachments,^ recounts a typ-ical exchange with her mother. Gor-nick always begins these encounters with high hopes. "Somehow," de-spite her best intentions, the conver-sations always spiral downward: Today is promising, tremendously prom-ising I go to meet my mother. I'm flying. Flying! I want to give her some of this shiningness bursting in me, siphon into her my immense happiness at being alive. Just because she is my oldest inti-mate and at this moment I love every-body, even her. "Oh, Ma! What a day I've had," I say. "Tell me," she says. "Do you have the rent this month?" "Ma, listen ..." I say. "That review you wrote for the Times," she says. "Ifs for sure they'll pay you?" "Ma, stop it. Let me tell you what I've been feeling," I say. "Why aren't you wearing something warmer?" she cries. "It's nearly winter." The space inside begins to shimmer. The walls collapse inward. I feel breath-less. Swallow slowly, I say to myself, slowly. To my mother I say, "You do know how to say the right thing at the right time. Ifs remarkable, this gift of yours. It quite takes my breath away." But she doesn't get it. She doesn't loiow I'm being ironic. Nor does she Elaine Hatfield is a Professor of Psychology and Richard L. Rapson is a Professor of History at the Uni-versity of Hawaii. John T. Ca-cioppo is a Professor of Psychology at the Ohio State University. Ad-dress correspondence to Elaine Hatfield, 2430 Campus Rd., Honolulu, HI 96822; BITNET: psych@uhunix; INTERNET: psych@uhunix.uhcc.hawaii.edu. know she's wiping me out. She doesn't know I take her anxiety personally, feel annihilated by her depression. How can she know this? She doesn't even know I'm there. Were I to tell her that it's death to me, her not knowing I'm there, she would stare at me out of her eyes crowd-ing up with puzzled desolation, this young girl of seventy-seven, and she would cry angrily, "You don't under-stand! You have never understood!" (pp. 103-104)