See	discussions,	stats,	and	author	profiles	for	this	publication	at:	https://www.researchgate.net/publication/302553545
Exploring	extreme	events	on	social	media:	A
comparison	of	user	reposting/retweeting
behaviors	on	Twitter	and...
Article		in		Computers	in	Human	Behavior	·	May	2016
DOI:	10.1016/j.chb.2016.04.032
CITATIONS
4
READS
113
3	authors:
Some	of	the	authors	of	this	publication	are	also	working	on	these	related	projects:
Verbal	Aggression	View	project
Impressions	of	Social	Media	Messages	and	Practices	View	project
Xialing	Lin
Penn	State	University,	Worthington	Scranton
15	PUBLICATIONS			163	CITATIONS			
SEE	PROFILE
Kenneth	A.	Lachlan
University	of	Connecticut
109	PUBLICATIONS			1,887	CITATIONS			
SEE	PROFILE
Patric	Spence
University	of	Kentucky
94	PUBLICATIONS			1,212	CITATIONS			
SEE	PROFILE
All	content	following	this	page	was	uploaded	by	Patric	Spence	on	24	July	2016.
The	user	has	requested	enhancement	of	the	downloaded	file.
Full length article
Exploring extreme events on social media: A comparison of user
reposting/retweeting behaviors on Twitter and Weibo
Xialing Lin a, *
, Kenneth A. Lachlan b
, Patric R. Spence c
a
Department of Corporate Communication, Penn State Worthington Scranton, USA
b
Department of Communication, University of Connecticut, USA
c
School of Information Science, University of Kentucky, USA
a r t i c l e i n f o
Article history:
Received 15 February 2016
Accepted 19 April 2016
Available online xxx
Keywords:
Crisis communication
Risk communication
Social media
Twitter
Weibo
a b s t r a c t
Although recent research suggests that Twitter and other forms of micro-blogging are becoming
increasingly relied upon by both the public and response agencies dealing with crises and disasters, little is
known about how these dynamics may play out in a non-Western context. The current study examines the
use of the Chinese Weibo service during a 2013 smog emergency, and compares user generated content to
that found in earlier data concerning a weather event in North America. The results indicate that by way of
comparison, the Weibo sample contained proportionately similar degrees of informative and affective
content, but that users were less likely to use humor and showed no increase in affective outpouring as the
crisis developed. Results are discussed in terms of implications for those designing social media campaigns
for informing and motivating those affected by large-scale weather-related crises.
© 2016 Published by Elsevier Ltd.
1. Introduction
Social media have evolved into widely used and legitimate
sources of news and information, particularly for natural disasters,
crises, and other extreme events (Pepitone, March 10, 2010). This
evolution is happening globally, within and between cultures and
across individual technological expertise. Previous extreme
weather situations, such as the 2010 Haiti cholera outbreaks, the
2011 Japanese tsunami, and Hurricane Sandy of 2012, have evident
that citizens in closest proximity to an event often act as eyewitnesses,
reporting information on social media in a timely manner
(e.g., Dredze, 2012; Lachlan, Spence, Lin & Del Greco, 2014; Sutter,
2010). As a group of interactive, collaborative, conversational, and
community-based platforms, social media collapse diverse social
contexts and media audiences into one, providing a comprehensive
outlet for the dissemination of crisis and risk information (Spence,
Lachlan, Lin, & del Greco, 2015). Despite the increasing volume of
research regarding the utilization of social media during extreme
events (Lachlan, Spence, Lin, Najarian, & Del Greco, 2016), the
extent to which different social media functions (such as reposting
or retweeting on microblogging services) are used for crisis
communication and information accessibility is understudied
(Lachlan, Spence, Lin, Najarian, & Del Greco, 2014). Further, most
research has focused on Western cultures and their social media
applications, while less is known about the differences in the information
consumption and learning across social media platforms
from different cultures. Such cross-cultural explorations of social
media are needed due to rapid technological diffusion and
increased opportunities for international travel and work.
As Twitter and other social networking sites are unavailable in
China, applications such as Sina Weibo ﬁll the void and take the
lead in the Chinese microblogging market (Tai & Liu, May 6, 2015).
To expand and solidify the understanding of social media functions
within different culture domains, the current study examines
mediated communication on Twitter and Weibo in the time leading
up to the impact of two large-scale weather-related events. In
particular, this study focuses on reposting/retweeting behaviors,
and compares the content and frequency of these behaviors
regarding the 2013 winter storm Nemo on Twitter and the 2013
Eastern China smog on Weibo. Speciﬁcally, examining retweeting/
reposting (the act of relaying a message that has been written by
another user of the platform) may reveal patterns or behaviors that
could be helpful during an extreme event. Moreover, cross-cultural
comparisons of crisis communication using social media are scarce.
This study aims to investigate public interactions of risk awareness
and concerns as well as to compare emergency information
* Corresponding author.
E-mail addresses: xialinglin.psws@gmail.com (X. Lin), Patric.spence@uky.edu
(P.R. Spence).
Contents lists available at ScienceDirect
Computers in Human Behavior
journal homepage: www.elsevier.com/locate/comphumbeh
http://dx.doi.org/10.1016/j.chb.2016.04.032
0747-5632/© 2016 Published by Elsevier Ltd.
Computers in Human Behavior xxx (2016) 1e6
Please cite this article in press as: Lin, X., et al., Exploring extreme events on social media: A comparison of user reposting/retweeting behaviors
on Twitter and Weibo, Computers in Human Behavior (2016), http://dx.doi.org/10.1016/j.chb.2016.04.032
management within different cultures and platforms.
1.1. Twitter and Weibo
Twitter launched in March of 2006, having accumulated around
302 million monthly active users with 500 million tweets being
posted per day as of 2016 (twitter.com). By initially providing users
the opportunity to post, read, and respond to text-based messages
limited to 140-characters in length, Twitter creates a multi-media
platform with constantly updated timelines for wide-open content.
These messages, called tweets, range from mundane life
chores to critical breaking news. Twitter users connect with each
other by following or being followed without reciprocal
requirement.
However, Twitter and other non-China-based networking sites
have been blocked in China since 2009, whereas some local social
networking services, such as Weibo, emerged to lead the Chinese
microblogging market (“Sina Weibo”, n.d.). Weibo was launched in
July 2010 and has occupied over 56.5% of microblogging market in
China since 2011, which can be thought of as a hybrid of Twitter and
Facebook. This social networking service currently has over 500
million users with more than 100 million posts being sent per day
(“Sina Weibo”, n.d.). Weibo also limits the lengths of each post to
140 characters; it also provides a follower-followee social network.
Weibo users can organize their social networking, to update and
follow the posts in unidirectional relationships as well.
A previous study indicates that in 2010, approximately 11% of all
tweets from a random sample were retweets, and users tend to
retweet messages containing hashtags and URLs (Suh, Hong, Pirolli,
& Chi, 2010). By using the retweet, Twitter users can repost a
message from another user and share it with one's own followers,
whereas, for Weibo users, the function is called “repost.” The
retweets or reposts are shown to the public on a senders' proﬁle
page; depending on these repost and reply mechanisms, users
choose to disseminate the information on Twitter and Weibo which
could go beyond the original messages (Kwak, Lee, Park, & Moon,
2010). Given the increased portions of retweets and reposts in
online daily updates, it is necessary to investigate the roles
these particular report functions have upon speciﬁc online
communication.
Holding great promise as a social medium for crisis communication,
these two networking sites provide publicly available online
information, reﬂecting immediate accessibility and harnessing
collective intelligence (O'Reilly & Battelle, 2009; Wunsch-Vincent &
Vickery, 2006). The public has gradually used Twitter or Weibo as a
source throughout a crisis lifecycle to make sense of an event and to
make well-informed decisions, especially for natural disasters
which have the potential to cause varying degrees of destruction or
harm to vulnerable populations (Coombs & Holladay, 2002; Qu,
Huang, Zhang, & Zhang, 2011; Zhang, Zhou, & Nunamaker Jr,
2002). For instance, Spence et al., (2015) analyzed the content
and the frequency of over 1500 tweets being sent during the prodromal
stage of Hurricane Sandy in 2012 and found that Twitter
served as a medium for information sharing and an outlet for affective
expressions. The results revealed several insights into
audience responses on Twitter that differed from previous selfreport
studies. People sought out tweets that more closely resembled
interpersonal messages than did those stemming from a
centralized, mass media entity (Spence et al., 2015). Qu et al. (2011)
examined the use of Weibo in response to 2010 Yushu earthquake
and reported a large portion of posts were opinion-related. However,
limited research compares the use of these two microblogging
platforms from different cultures in risk and crisis domains, thus,
cross-cultural research is needed (Burke & Zhou, 2010). Two
extreme weather-related events, the 2013 winter storm Nemo and
the 2013 Eastern China smog, are examined in the current study.
1.2. 2013 Winter Storm Nemo
Winter storm Nemo primarily struck the northeastern United
States and parts of Canada in early February 2013 (“February 2013
nor'easter”, n.d.). The storm delivered rapidly accumulating
snowfall combining with hurricane-force winds and destructive
tides, which caused at least eighteen storm-related fatalities, over
600,000 customer power outages, and economic damages
throughout New England (Ariosto et al., February 9, 2013). In the
city of Boston, Nemo brought 24.9 inches of snowfall from February
8 to 10, 2013, ranking as one of the ﬁve worst snowstorms in Boston
since 1935 (Klepper & Salsberg, February 10, 2013). On February 8,
2013, Boston reported a storm surge that peaked at 4.21 feet,
marking it as the fourth-highest storm surge since 1921 (Masters,
2013). The blizzard caused a buzz on the social media during that
time. Bostonians and local organizations spontaneously used
#Bosnow on Twitter to track blizzard updates, send out emergency
alerts, and offer relief assistance in real time (e.g., City of Boston.
Gov, October 2013).
Previous studies have investigated how people used Twitter in
the early stage of winter storm Nemo. For instance, Lachlan, Spence,
Lin, Najarian, & et al. (2014) retrieved the tweets containing
localized and national hashtags during this weather event, and
suggest that Twitter might be used more for affective display than
for information seeking, which is consistent with Twitter research
on Hurricane Sandy (Spence et al., 2015). However, it is unclear
whether retweets during Nemo reﬂect such ﬁndings. Thus, as an
extended investigation of Lachlan, Spence, Lin, Najarian, & et al's.
(2014) data, the ﬁrst research question follows:
RQ1 What were the frequency and speciﬁc information users
retweeted concerning Winter Strom Nemo on Twitter?
1.3. 2013 Eastern China smog
To compare the repost/retweet behaviors between Twitter and
Weibo during weather-related disasters, the current study also
examines how online users communicated during the 2013 Eastern
China smog. This weather-related event was chosen because it
occurred in the same year as Winter Storm Nemo, thus allowing for
as close of a chronemic examination as possible. Given that the time
and type of weather related crises are not predictable; these two
events provide the best comparison available.
From December 2nd to 14th, 2013, with a combination of
weather conditions and an increase in the burning of coal for
homes and municipal heating systems, heavy smog blanketed
Eastern China, affecting more than seven provinces and nine major
cities. High levels of tiny and harmful PM 2.5 particulates were
detected, with over 150 mg per cubic meter on average (“2013
Eastern China smog,” n.d.). In Shanghai, air quality index (AQI) reported
the highest level air pollution among all the cities on
December 6th; PM 2.5 particulates also reached 602.5 mg per cubic
meter on that day (Tang & Hoshiko, December 7, 2013). It reached
the most hazardous level since the city began recording such data
and soared to a record over 24 times higher than the World Health
Origination's (WHO) safety standard (Tang & Hoshiko, December 7,
2013). The severe air pollution resulted in limited visibility, reduced
outdoor activity, increased respiratory symptoms, causing major
disruption in transportation and daily routines (“2013 Eastern
China smog,” n.d.).
This severe weather event caused discussions on Weibo. Researchers
have investigated how people use microblogging for
X. Lin et al. / Computers in Human Behavior xxx (2016) 1e62
Please cite this article in press as: Lin, X., et al., Exploring extreme events on social media: A comparison of user reposting/retweeting behaviors
on Twitter and Weibo, Computers in Human Behavior (2016), http://dx.doi.org/10.1016/j.chb.2016.04.032
online and off-line activisms (e.g., Xu, 2014). However, limited
research has focused on certain functions of microblogging platforms,
such as retweets or reposts, to investigate how users fulﬁll
their needs during extreme events. Thus, the following research
question is raised:
RQ2 What were the frequency and speciﬁc information users
reposted concerning Eastern China Smog on Sina Weibo?
It is also critical to compare the repost and retweet behavior
between users on Twitter and Weibo during these events to
determine similarities and differences based on platform and culture.
Thus, the ﬁnal research question is offered:
RQ3 Are there differences in the frequency of types of account
proﬁles from which people post on Twitter and Weibo?
2. Methods
2.1. BoSnow retweets
TweetArchivist (www.tweetarchivist.com) was used to collect
tweets at 4 hour intervals during the noon and evening of February
8, 2013. The search terms #nemo and #bosnow were used, as
#nemo was a widely used hashtag that was promoted by federallevel
relief agencies such as the National Oceanic and Atmospheric
Administration (NOAA), and #bosnow was used and promoted
by the Boston Globe and other media outlets in southern
New England (Lachlan, Spence, Lin, Najarian, & et al., 2014). Users'
information and live links were also collected by TweetArchivist. A
total of 799 tweets were collected. Of these tweets, 283 were
retweets.
Content attributes and the time of the tweets were identiﬁed in
the coding process. Two undergraduate coders were trained on the
coding scheme over the course of one week. The coders were then
recruited for coding process to make judgments concerning the
content attributes of each tweet. Reliability above the minimum
standard of 0.70 was conﬁrmed for all variables in the dataset (see
Krippendorff, 1980; Spence & Lachlan, 2005), with the reliability of
0.88 for the message type.
Both the type of content present in the tweet (information about
the storm, expressions of affect, spam, humor, or insult) and the
proﬁle characteristics were identiﬁed. The operational deﬁnition
for informative tweets was those whose primary intent was to
provide information concerning the technical aspects of the storm
or speciﬁc mitigation efforts. Expressions of affect were deﬁned as
tweets whose primary purpose was the expression of emotion
(such as fear, anger, worry, or dread). If an advertisement or a link to
a Web page advertising a consumer product was present, the tweet
was categorized as spam. Humorous tweets contained a primary
intention to entertain and included all tweets that could be identiﬁed
as digitally manipulated photographs, celebrity commentary,
and individual messages designed to induce amusement. Insults
were tweets deﬁned as those containing derogatory comments
directed at individuals or organizations (See Lachlan, Spence, Lin, &
Del Greco, 2014).
2.2. Eastern China smog reposts
Weibo Search (s.weibo.com) was used to collect posts during
December 2nd to December 14th, 2013. The search term 雾霾
(Smog) and 上海雾霾 (Shanghai smog) were used, as Shanghai was
the major city and most affected by the smog. Weibo Search produces
exact replications of 1000 random posts at a given collection
point, including links to users proﬁles and images, videos, and
active URLs embedded in each post. A total of 23,222,244 posts
were shown in the search results, of which 1000 posts were
randomly selected for the current study. Of these posts, 205 were
reposts.
Two coders were recruited for the coding process to make
judgments concerning the time, user account identity, the number
of reposts, and the content attributes for each post. For the content
of each post, the number of the images, videos, active URLs, and @
other users were identiﬁed. Acceptable inter-colder reliability were
reported, with the minimum reliability of 0.74 for message type.
Both the type and content present in the coding scheme used in
the Twitter sample were used in the Weibo sample. However, an
additional category titled reﬂection was added due to representation
in the sample. Reﬂective reposts were those where the proﬁle
owner reposted speciﬁc posts debating the causes of the smog.
These were not informative but rather focused on debates between
users on the attribution of the weather-related event.
Methods and measures for data collection through social media
in crisis situations are unstructured, untested, and there is little
agreement on the standard conventions (Nelson, Spence, &
Lachlan, 2009; Spence, Lachlan & Rainear, 2016). Due to differences
across platforms and crisis types, an exact replication of the
sampling procedures was not possible; however, given the affordances
of the available tweet/post collection programs, and the
length of each crisis event, the two sampling procedures have
complementary features. However, the randomized selection of
retweets/reposts have advantages that are represented in the
literature (see Spence & Lachlan, 2010; Spence, Lachlan, & Rainear,
2016).
3. Results
The ﬁrst research question investigates the frequency and content
of retweets on Twitter during winter storm Nemo. A total of
283 retweets were identiﬁed at exact time points February 8th,
2013, this is 35.4% of all the tweets collected. This date was chosen
as it represents the day on which the storm inﬂicted the greatest
degree of impact on the region. Among these retweets, 23.3% were
sent at noon, 31.1% retweets were sent at 4:00 p.m., 25.1% were sent
at 8:00 p.m., and 20.5% were sent at midnight. For the content of
these retweets, 41.0% of these retweets were categorized as information,
34.9% were categorized as humor, 23.0% were categorized
as affect displays, 0.7% were categorized as spam, and 0.4% were
categorized as insults.
Examining the time-series analysis of storm retweets, results
indicated the retweet content with information and affect shifted
between noon, afternoon, evening, and midnight (see Table 1). For
the informative retweets, 41.4% were sent at noon, which dropped
to 18.1% by 4:00 p.m. and 16.4% were sent at 8:00 p.m., then the
number of informative retweets increased to 24.1% by midnight.
Whereas for affective display, 6.1% of the retweets were sent at
noon with 38.5% being sent at 4:00 p.m. and 8:00 p.m.; and 16.9%
were sent at midnight. For the retweets with humor, 14.1% were
sent at noon, with 42.4% of humorous retweets being sent at 4:00
Table 1
2013 BoSnow Nemo: Retweet content type by time.
Time Retweet type
Information Affect Spam Humor Insult
12:00 41.4% 6.1% 0% 14.1% 0%
16:00 18.1% 38.5% 0% 42.4% 0%
20:00 16.4% 38.5% 100% 24.2% 100%
24:00 24.1% 16.9% 0% 19.3% 0%
X. Lin et al. / Computers in Human Behavior xxx (2016) 1e6 3
Please cite this article in press as: Lin, X., et al., Exploring extreme events on social media: A comparison of user reposting/retweeting behaviors
on Twitter and Weibo, Computers in Human Behavior (2016), http://dx.doi.org/10.1016/j.chb.2016.04.032
p.m., followed by 24.2% at 8:00 p.m., and 19.3% by midnight. The
retweets categorized as insult and spam were all sent in the evening
(8:00 p.m.).
Research question two examined how people used reposts on
Weibo during the Eastern China smog. Of the 1000 posts collected,
205 were reposts (20.5%), the repost rates range from 120,613 to 61.
Over 71% (145) of all retweets occurred over the four day period
when the smog was at its highest levels from December 5th e 8th
which will be the focus of the subsequent analysis. The two most
active days were December 7th where 33.7% of the messages were
reposted and 21.5% were on December 6, 2013, during the time
when smog reached the extremely hazardous levels in Shanghai.
For the content attributes of all the reposts on Weibo during the
weather-related event, 47.3% were categorized as information,
19.5% were categorized as affect displays, 14.6% were categorized as
humor, 13.2% were categorized as reﬂection or attribution, and 5.4%
were categorized as spam.
Examining the time-series analysis of smog reposts for
December 5th e 8th, results indicated the repost content of information,
and affect, humor, and reﬂection were consistent as the
weather event developed. To make the comparisons across the two
weather-related events similar, only the four days of the highest
smog levels are reported. The percentage of informative reposts
began to increase on December 5th with 54.6%, and continued on
December 6th with 38.6%, and expanded to 49.4% on December 7th
before reaching a peak of 70% on December 8th. Consistently, for
the affective posts, an increase in reposts began on December 5th
continuing through December 8, 2013 (see Table 2). On December
5th, 18.2% of all reposts were affective, with 20.5% sent on
December 6, remaining nearly equal to 18.8% on the 7th and
dropping to 10% on the 8th. A unique ﬁnding for affective reposts
was that there was an additional spike in affective reposts numbers
on the 9th, with 46.7% of all the reposts being affective.
For the reposts with humor, 4.5% of reposts on December 5th
were humorous, increasing to 29.5% on December 6th and 13% on
December 7th, and declining to 10% on December 8th. For the
reposts with reﬂection or critiques, 18.2% of all reposts on
December 5th were reﬂections, followed by 2.3% on December 6th,
17.4% on December 7th, and then reposts with reﬂections or critiques
were absent December 8th. Finally, reposts classiﬁed as
spam were low on December 5th (4.5%), rising to 9.1% on the 6th
then dropping to 1.4% on the 7th and 1.0% on the 8th.
The third research question asked about differences in account
type and proﬁle between Twitter and Weibo. On Twitter, for user
account identity, 87.3% of the accounts were standard user accounts
with 4.6% of all retweets coming from for-proﬁt organizations and
1.4% being retweeted from government agencies. In the Weibo
sample for the user account identity, 44.9% were organizational
Weibo accounts with blue veriﬁcation identities attached to their
usernames, including news media, business company, and local
private sectors; 44.4% were individual accounts with orange veriﬁcation
identity attached to their usernames, such as public health
practitioners, celebrities, lawyers, journalists, scholars, and business
people. A total of 9.8% were individual accounts without any
identiﬁcation veriﬁcation, and 1.0% were popular individual accounts
with start veriﬁcations.
4. Discussion
It is well documented that individuals use social media for information
seeking and subsequent learning when faced with the
challenges and threats posed by crises (Lachlan, Spence, & Lin,
2014; Lachlan et al., 2016). This study aims to present how the
public use social media during extreme weather-related events as
well as to compare the retweet and repost behaviors within
different microblogging cultures. The 2013 winter storm Nemo and
the Eastern China smog were examined as extreme weatherrelated
events in this study. The ﬁrst area of note is that the number
of retweets/reposts appears to double during a crisis related
event, compared with more routine events. The 35.4% of retweets
with winter storm Nemo and 20.5% associated with the smog event
is well above the 11% of retweets found in previous research. This
indicates that during times of crisis, people potentially attempt to
share information. However, Twitter users were more likely to
repost/retweet than users of Weibo, but this may be a cultural
difference which emerged in the data. These types of retweets
increased the percentage of total retweets in the sample, but were
not as largely represented in the Weibo sample.
The results suggest that most of the messages retweeted/
reposted were informative both in winter storm Nemo and the
Eastern China smog events. However, the ﬁndings indicate a balance
between informative and affective retweets during the time
series of the storm Nemo, whereas a consistency among informative,
affective, and humorous reposts during the time series of the
China smog. The similarities and the differences between these two
repost behaviors are discussed as below.
First, most of the messages on Twitter retweets and Weibo
reposts contained informative messages, which reﬂects the levels
of uncertainties during these extreme weather events (Seeger,
2006). These informative messages included weather updates,
general risk, as well as hazard preventions and protections. More
speciﬁcally, informative retweets with the hashtag of #bosnow
tended to provide more actionable information such as evacuation
efforts, the whereabouts of food/shelter, the whereabouts of others,
and how to how to care for the sick and elderly. Reposts with the
keywords of Shanghai smog included more information about the
use of smog masks, the suggestions for preventions, and protections
for the children and elderly. These ﬁndings suggest that
users in both cultures may tend to offer more messages with efﬁcacy
and actionable suggestions (Spence et al., 2015). For the
informative retweet ﬁndings for the winter storm Nemo, results are
also consistent with Lachlan, Spence, Lin, Najarian, & et al.'s (2014)
analyses on the original tweets. Although not speciﬁcally
measuring learning and uncertainty, people turn to social media to
learn and reduce uncertainty; and the data suggests that much of
the messages which were retweeted/reposted were informative in
nature. Examining retweets/reposts explicitly provides different
results than what emerged in previous studies (Lachlan, Spence,
Lin, Najarian, & et al., 2014; Lachlan et al., 2016; Lachlan, Spence,
Lin, & Del Greco, 2014; Spence et al., 2015); it suggests that to
ﬁnd informative information during an information glut on social
media, users may want to focus on retweeted/reposted messages.
Second, different patterns between retweet behaviors and
repost behaviors were found. For Nemo retweets, the results suggest
that, at the beginning of the storm, individuals were
retweeting informative information. This indicates that informative,
usable information is more available at the onset of the event
and then begins to dissipate; this ﬁnding is similar to what was
reported following Tropical Strom Sandy (Spence et al., 2015).
Table 2
2013 Eastern China smog: Reposts content type by date.
Date Repost type
Information Affect Spam Humor Insult Reﬂection
December 5 54.6% 18.2% 4.5% 4.5% 0.0% 18.2%
December 6 38.6% 20.5% 9.1% 29.5% 0.0% 2.3%
December 7 49.4% 18.8% 1.4% 13.0% 0.0% 17.4%
December 8 70.0% 10.0% 10.0% 10.0% 0.0% 0.0%
X. Lin et al. / Computers in Human Behavior xxx (2016) 1e64
Please cite this article in press as: Lin, X., et al., Exploring extreme events on social media: A comparison of user reposting/retweeting behaviors
on Twitter and Weibo, Computers in Human Behavior (2016), http://dx.doi.org/10.1016/j.chb.2016.04.032
However, affective retweets during the day also began to dissipate.
In a study by Lachlan, Spence, Lin, Najarian, & et al. (2014) affective
content became more prevalent as the weather event continued,
thus rendering tangible, actionable information more difﬁcult to
locate in the morass of available information. The decrease in affective
retweets in the current data may simply be related to the
increase in affective tweets at the same time. That is, if all the
tweets surrounding a hashtag appear to show displays related to
fear, dread and anxiety, individuals may see little need in
retweeting those displays. However, for the Eastern China smog
reposts, the results suggest consistency in the relative proportions
of informative and affective messages. This might be a result of the
different time lengths for collecting the data between the samples;
however, it is important to note that the differences in length of
data collection were a result of the duration of the weather-related
events. More research is needed on this issue.
Moreover, a large number of smog reposts contained emoticons,
which indirectly indicated affective displays. In particular, not only
individual accounts, but also ofﬁcial accounts, such as news media
or celebrities, were widely embedded emoticons in the posts for
connotation meanings. This is different from Nemo retweets, where
most affects were displayed by verbal denotations.
Third, lower rates of humorous reposts are seen in the Weibo
sample. This may be attributed to cultural differences indicating
less of a wiliness to engage in humor during a crisis event. Another
difference between the samples was that there was more organizational
participation in reposting on Weibo. There was almost
equal participation between organizational accounts and individual
accounts on Weibo, whereas the majority of retweets in the
Twitter sample were performed by individual (civilian) accounts.
These differences indicate larger participation from organizations
during weather-related events for the Weibo sample, a ﬁnding that
may be explained by cultural and socio/political factors.
Taken together, the ﬁndings offer a ﬁrst glimpse into the ways in
which microblogging platforms, like Twitter and Weibo, may vary
in terms of their use across different cultures in the context of an
impending crisis or disaster. While a substantive amount of
actionable information can be found in content generated by users
on both platforms, the trailing off of actionable information
detected during North American crises was not found in China. At
the same time, during this speciﬁc incident in China, the glut of
humorous tweets was not detected, presumably posing fewer difﬁculties
in users obtaining information. First responders designing
campaigns using Weibo to inform and motivate may be able to rely
upon the medium well into the evolution of the crisis, knowing that
users are going to the platform for informational needs. They may
also have more conﬁdence in the likelihood that their content
would be reposted by individuals, as information from organizations
and government sources was more likely to be reposted than
that originally generated by individuals; this ﬁnding stands in stark
contrast to earlier ﬁndings concerning western audiences.
4.1. Limitations
Several limitations were identiﬁed in this study. First, the time
series for the storm Nemo and the China smog were collected
differently, with one day for the storm retweets and two weeks for
the smog reposts. Although this was a feature of the length of the
crisis, it does create a limitation in the ﬁndings. Second, due to the
technical limitations on Weibo Search, it is impossible to retrieve
the entire posts and reposts. Thus, future research could use other
Weibo API tools to gain a more comprehensive understanding on
both reposts and original posts. Third, the small sample size of
Weibo reposts was utilized in the study. Future research should
collect more data for analysis to gain more generalizability and
validity power.
4.2. Conclusion
This study examined the content of communication and frequency
of communication on Twitter and Weibo during extreme
weather events. A comparison was made between Twitter retweets
and Weibo reposts. Results indicate the similarities and the differences
between these two networking platforms in light of reposting
behavior patterns and post content attributes. The
understanding the information processing of the public during
these weather-related events begin to provide useful insights into
crisis learning in disaster responses, recovery, and evaluations for
future disaster management. As the public becomes more reliant on
social media for information seeking and affect displays, it is
necessary for risk and crisis researchers as well as public health
practitioners to continue learning about this process, and learning
about how and why people create and consume crisis-related information
through social media.
References
Ariosto, D., Botelho, G., Boyette, C., Cherneff, E., Snow, M., Shaughnessy, L., et al.
(2013, February 9). 1 dead, 600,000 without power from ongoing blizzard. CNN.
Retrieved from http://www.cnn.com/2013/02/08/us/northeast-blizzard/index.
html.
Burke, J. A., & Zhou, J. (2010). Wenchuan earthquake preparation and information
seeking: lessons from the ﬁeld. Northwest Journal of Communication, 39,
109e124.
City of Boston Gov. (2013, October). Climate ready Boston. Retrieved from http://
www.cityofboston.gov/news/uploads/30044_50_29_58.pdf.
Coombs, W. T., & Holladay, S. J. (2002). Helping crisis managers protect reputational
assets initial tests of the Situational Crisis Communication Theory. Management
Communication Quarterly, 16(2), 165e186.
Dredze, M. (2012). How social media will change public health. Intelligent Systems,
IEEE, 27(4), 81e84.
February 2013 nor'easter. (n.d.). In Wikepedia. Retrieved from Retrieved from
http://en.wikipedia.org/wiki/February_2013_nor'easter.
Klepper, D., & Salsberg, B. (2013, February 10). Boston blizzard: Northeast snowed in
as ‘‘Nemo’’ barrels through. Hufﬁngton Post. Retrieved from http://www.
hufﬁngtonpost.com/2013/02/09/boston-blizzard-nemo_n_2650592.html.
Krippendorff, K. (1980). Content analysis: An introduction to its methodology. Newbury
Park, CA: Sage.
Kwak, H., Lee, C., Park, H., & Moon, S. (2010). What is Twitter: a social network or a
news media. In Proceedings of the 19th international conference on the World
Wide Web (pp. 591e600). New York, NY: ACM.
Lachlan, K. A., Spence, P. R., & Lin, X. (2014). Expressions of risk awareness and
concern through Twitter: on the utility of using the medium as an indication of
audience needs. Computers in Human Behavior, 35, 554e559.
Lachlan, K. A., Spence, P. R., Lin, X., & Del Greco, M. (2014). Screaming into the wind:
examining the volume and content of Tweets associated with Hurricane Sandy.
Communication Studies, 65(5), 500e518.
Lachlan, K. A., Spence, P. R., Lin, X., Najarian, K. M., & Del Greco, M. (2014). Twitter
use during a weather event: comparing content associated with localized and
nonlocalized hashtags. Communication Studies, 65(5), 519e534. http://
dx.doi.org/10.1080/10510974.2014.956940.
Lachlan, K. A., Spence, P. R., Lin, X., Najarian, K., & Del Greco, M. (2016). Social media
and crisis management: CERC, search strategies, and twitter content. Computers
in Human Behavior, 54, 647e652.
Masters, J. (2013, February 9). Snow measured in feet, not inches: Historic Nor'easter
pounds New England. Retrieved from http://www.wunderground.com/blog/
JeffMasters/comment.html?entrynum¼2346.
Nelson, L. D., Spence, P. R., & Lachlan, K. A. (2009). Learning from the media in the
aftermath of a crisis: ﬁndings from the Minneapolis bridge collapse. Electronic
News, 3(4), 176e192. http://dx.doi.org/10.1080/19312430903300046.
O'Reilly, T., & Battelle, J. (2009). Web squared: Web 2.0 ﬁve years on. Retrieved
February 1, 2013 from http://assets.en.oreilly.com/1/event/28/web2009_
websquared-whitepaper.pdf.
Pepitone, J. (2010, March 10). Twitter users not so social after all. CNNMoney.com..
Retrieved from http://money.cnn.com/2010/03/10/technology/twitter_users_
active/index.htm?hpt¼Mid.
Qu, Y., Huang, C., Zhang, P., & Zhang, J. (2011). Microblogging after a major disaster
in China: a case study of the 2010 Yushu earthquake. In Paper presented at the
Proceedings of the ACM 2011 conference on Computer supported cooperative work,
Hangzhou, China.
Seeger, M. W. (2006). Best practices in crisis communication: an expert panel
process. Journal of Applied Communication Research, 34(3), 232e244. http://
dx.doi.org/10.1080/00909880600769944.
X. Lin et al. / Computers in Human Behavior xxx (2016) 1e6 5
Please cite this article in press as: Lin, X., et al., Exploring extreme events on social media: A comparison of user reposting/retweeting behaviors
on Twitter and Weibo, Computers in Human Behavior (2016), http://dx.doi.org/10.1016/j.chb.2016.04.032
Sina Weibo (n.d.). In Wikipedia. Retrieved May 4, 2015, from http://en.wikipedia.
org/wiki/Sina_Weibo.
Spence, P. R., & Lachlan, K. A. (2005). Teaching intercoder reliability: a gentle
introduction to content analytic methods for graduate students. Texas Speech
Communication Journal, 30(1), 71e76.
Spence, P. R., & Lachlan, K. A. (2010). Disasters, crises, and unique populations:
suggestions for survey research. New Directions for Evaluation, 126, 95e106.
Spence, P. R., Lachlan, K. A., Edwards, A., & Edwards, C. (2016). Tweeting Fast
matters, but only if think about it: information updates on social media.
Communication Quarterly, 64, 55e71. http://dx.doi.org/10.1080/
01463373.2015.1100644.
Spence, P. R., Lachlan, K. A., Lin, X., & del Greco, M. (2015). Variability in twitter
content across the stages of a natural disaster: implications for crisis communication.
Communication Quarterly, 63(2), 171e186. http://dx.doi.org/10.1080/
01463373.2015.1012219.
Spence, P. R., Lachlan, K. A., & Rainear, A. M. (2016). Social media and crisis research:
data collection and directions. Computers in Human Behavior, 54, 667e672.
Suh, B., Hong, L., Pirolli, P., & Chi, E. H. (2010). Want to be retweeted? Large scale
analytics on factors impacting retweet in Twitter network. In IEEE intl conference
on social computing (pp. 177e184). IEEE.
Sutter, J. D. (2010, October, 29). Texts, maps battle Haiti cholera outbreak. Retrieved
March 13, 2013 from http://www.cnn.com/2010/TECH/innovation/10/29/haiti.
cholera.tech/index.html?hpt¼Sbin.
Tai, Z., & Liu, X. (2015, May 6). The Chinese government hops on the WeChat bandwagon.
University of Nottingham. Retrieved from http://blogs.nottingham.ac.
uk/chinapolicyinstitute/2015/05/06/chinese-government-hops-on-the-wechat-
bandwagon/.
Tang, D., & Hoshiko, E. (2013, December 7). Shanghai smog hits extremely hazardous
levels. Retrieved from the Hufﬁngton Post http://www.hufﬁngtonpost.com/
2013/12/07/shanghai-smog_n_4402477.html.
Wunsch-Vincent, S., & Vickery, G. (2006). OECD study on the participative web: User
generated content. Technical Report. Retrieved from http://www.oecd.org/
dataoecd/57/14/38393115.pdf.
Xu, J. H. (2014). Communicating the right to know: social media in the do-ityourself
air quality testing campaign in Chinese cities. International Journal of
Communication, 8, 1374e1393.
Zhang, D., Zhou, L., & Nunamaker, J. F., Jr. (2002). A knowledge management
framework for the support of decision making in humanitarian assistance/
disaster relief. Knowledge and Information Systems, 4, 370e385.
2013 Eastern China smog (n.d.). In Wikipedia. Retrieved from http://en.wikipedia.
org/wiki/2013_Eastern_China_smog.
X. Lin et al. / Computers in Human Behavior xxx (2016) 1e66
Please cite this article in press as: Lin, X., et al., Exploring extreme events on social media: A comparison of user reposting/retweeting behaviors
on Twitter and Weibo, Computers in Human Behavior (2016), http://dx.doi.org/10.1016/j.chb.2016.04.032
View publication statsView publication stats