Content uploaded by Kate Keib
Author content
All content in this area was uploaded by Kate Keib on Oct 11, 2017
Content may be subject to copyright.
Full Terms & Conditions of access and use can be found at
http://www.tandfonline.com/action/journalInformation?journalCode=hmep20
Download by: [174.218.134.92] Date: 11 October 2017, At: 09:28
Media Psychology
ISSN: 1521-3269 (Print) 1532-785X (Online) Journal homepage: http://www.tandfonline.com/loi/hmep20
Picture This: The Influence of Emotionally
Valenced Images, On Attention, Selection, and
Sharing of Social Media News
Kate Keib, Camila Espina, Yen-I Lee, Bartosz W. Wojdynski, Dongwon Choi &
Hyejin Bang
To cite this article: Kate Keib, Camila Espina, Yen-I Lee, Bartosz W. Wojdynski, Dongwon
Choi & Hyejin Bang (2017): Picture This: The Influence of Emotionally Valenced Images,
On Attention, Selection, and Sharing of Social Media News, Media Psychology, DOI:
10.1080/15213269.2017.1378108
To link to this article: http://dx.doi.org/10.1080/15213269.2017.1378108
Published online: 05 Oct 2017.
Submit your article to this journal
Article views: 86
View related articles
View Crossmark data
Picture This: The Influence of Emotionally Valenced
Images, On Attention, Selection, and Sharing of Social
Media News
Kate Keib
a
, Camila Espina
b
, Yen-I Lee
b
, Bartosz W. Wojdynski
b
, Dongwon Choi
b
,
and Hyejin Bang
c
a
Communication Studies, Oglethorpe University, Atlanta, Georgia, USA;
b
Grady College of Journalism
and Mass Communication, University of Georgia, Athens, Georgia, USA;
c
School of Journalism and Mass
Communication, The University of Kansas, Lawrence, Kansas, USA
ABSTRACT
Social media platforms have become the primary conduits to
news for many consumers, yet little is known about how the
content in social media posts is viewed and evaluated by
consumers or how it shapes their decisions about selecting
and sharing this information. A within-subjects eye-tracking
experiment (N= 60), was conducted to examine the influence
of image presence and valence on attention to and engage-
ment with news stories on social media. Participants viewed a
series of 29 social media posts of news stories, each of which
was either paired with no image, a positively valenced image,
or a negatively valenced image. Participants attention to the
images was captured via eye tracking, and they answered
dependent measures to gauge level of emotion and arousal,
and intention to click and share. The results show that posts
containing positive images elicited a higher level of visual
attention than those with negative or no images, which led
to higher intentions to click and share posts with positive
images. The results provide a deeper understanding of the
importance of images in driving news consumption, and offer
practical implications for journalists, news organizations and
groups using social media to spread a message.
In recent years, social media have become an important conduit through which
consumers find and select online content (Anderson & Caumont, 2014).
Facebook alone accounts, on average, for 43% of traffic to many major news
sites (Ingram, 2015), and in deference to that fact, Facebook and many of the
leading news organizations have collaborated to make news consumption on
the platform itself faster, with instant publishing and Facebook Live (DeMers,
2016; Mullin, 2015). Considering that 63% of Americans report that they
consume news via Facebook (Barthel, Shearer, Gottfried, & Mitchell, 2015),
the content of the post is critical to consumers’impressions of the linked story
CONTACT Kate Keib kkeib@Oglethorpe.edu Oglethorpe University, 4484 Peachtree Road NE, Atlanta, GA
30319, USA.
Color versions of one or more of the figures in the article can be found online at www.tandfonline.com/hmep.
MEDIA PSYCHOLOGY
https://doi.org/10.1080/15213269.2017.1378108
© 2017 Taylor & Francis Group, LLC
Downloaded by [174.218.134.92] at 09:28 11 October 2017
and their selection decision. Although this transition in online news consump-
tion has been documented at the network level (Hermida, 2012; Stieglitz &
Dang-Xuan, 2013), relatively little is known about the psychological factors that
underlie why and how stories on social media platforms capture users’attention
and influence their decision to disseminate the content further.
Social media platforms such as Facebook and Twitter present users with a
continuous feed of posts from members of their social network, and many of
these posts contain links to external content (Matsa, 2016; Truong, 2016).
Users must make decisions about whether to click through to view content,
or whether to share it to their network by evaluating the limited information
that the platforms allow a post to contain: a source name and thumbnail, a
headline, a single image, and often a short description. Studying specific
elements of social media posts can lead to greater understanding of the
factors that motivate individuals to engage with a particular news story
amidst an unimaginable number of alternatives. Past work on social media
has determined that images elicit more interest than content without images
(Ulloa, Mora, Pros, & Tarrida, 2015), and high numbers of likes and shares
depicted on the post encourages news consumption on social media
(Hermida, 2012; Hille & Bakker, 2013). Additionally, research has also
shown that when sharing news articles specifically through social media,
other users’comments on the post can drive interest and involvement in
the news topic, as well as subsequent information seeking intentions
(Oeldorf-Hirsch & Sundar, 2015).
Although the informational value of socially shared information can drive
selection, attention, and sharing, consumers’emotional responses also play
an important role (Coviello et al., 2014). Emotions are internal mental states,
brought about in response to valenced external stimuli (Nabi, 2003; Ortony,
Clore, & Collins, 1990). Definitions of this concept across the literature
emphasize that emotions consist of three main components—appraisal,
intention to act, and feeling state (Tan, 2008)—or even broadly extending
to motor expressions and actions (Shuman, Clark-Polner, Meuleman, Sander,
& Scherer, 2017). Functional theories of emotion address the role of emotion
in shaping the allocation of mental and physical resources to interact with the
world around them (Lazarus, 1991; Nabi, 2003). Past work has also shown
that the valence of emotions experienced also influences consumers’evalua-
tion of content, and their resulting behavioral intentions (De Vreese &
Boomgaarden, 2003; Harber, Podolski, & Dyer, 2014).
Social media news posts play a dual role in passing on information:
They convey news information directly through the content therein, and
also typically serve as links to additional content. Both the emotion
generated by the post itself and the expected emotional gains of consum-
ing the story can influence selection and sharing. Recent work into news
consumption on social media platforms has shown that people will engage
2K. KEIB ET AL.
Downloaded by [174.218.134.92] at 09:28 11 October 2017
with content that elicits emotion, and that negative emotions may lead to
greater sharing (Harber et al., 2014). Shoemaker and Cohen (2006)pre-
sented the notion that people are biologically inclined to prefer news with
social significance and deviance in it. This type of news, they say, would
drive people to think they are informing themselves to protect themselves
from potential harm (Shoemaker & Cohen, 2006). Emotional broadcaster
theory (EBT) has demonstrated that individuals are more likely to share
content that violates their schema (Harber et al., 2014). Given their choice
of over 50 stories to read, participants chose negative news over positive
or neutral news, despite also telling researchers that they preferred posi-
tive news (Trussler & Soroka, 2014). In contrast to these studies, Berger
and Milkman (2012) analyzed articles from the New York Times over 3
months, and found that content with positive valence was more viral than
content with negative or no valence (Berger & Milkman, 2012). The
preponderance of research shows a marked preference for negative con-
tent, but it is not unequivocal.
Not all content in a social media post may be equally effective at
eliciting emotional reactions. In the increasingly mobile and social land-
scape of digital news, a news story must first catch the user’s eye before
shecanmakethedecisiontoengagewiththecontentorscrollon(Fogg,
2003). Photos, graphics, and other images in the post play a key role in
these processes; social media posts containing static images received
almost three times as many shares as text-only posts (Guerini, Staiano,
& Albanese, 2013). Similarly, large images in Facebook news posts have
been shown to positively influence readers’attraction, interest and further
information seeking (Ulloa et al., 2015). Therefore, news images in social
media posts are far more than decoration; they govern attention, percep-
tion, and engagement. We believe that greater understanding of the use of
images in social media posts is key in understanding how citizens will
select and consume online news content.
This study sought to examine how images in social media posts pro-
moting news stories affect emotional response and visual attention. In a
within-subjects eye-tracking experiment (N= 60), participants viewed a
series of social media news posts, each of which was paired with either no
image, a negatively valenced photograph, or a positively valenced photo-
graph; the source and headline text were held constant. The study also
examined the impact of image presence and valence on users’emotional
responses to the social media post, and their intention to engage with it
further by clicking or sharing. The study was undertaken in the interest of
not only building theory in the area of image processing and emotion but
also in the hopes that a greater understanding of the role of images can
help journalists and editors develop engaging and compelling content that
increases website traffic and generates revenue.
MEDIA PSYCHOLOGY 3
Downloaded by [174.218.134.92] at 09:28 11 October 2017
Literature review
Images in social media news posts
Images are important conduits of information transmission on the Web
(Kenix, 2013). The presence or absence of an image is a design cue viewers
use when making decisions about consuming and interpreting content
(Leckner, 2012). Consumers’visual attention in online news reading is first
drawn to photographs, provided a photo is present near the beginning of the
content (Bucher & Schumacher, 2006). In addition to viewing pictures first,
online news readers also look at images for a longer duration than text areas
of the same size (Ulloa et al., 2015). Both the primacy and processing
duration effects of images allow them to have significant influence on how
news content is consumed.
Because social media are increasingly prevalent paths to news, the content
of a single image within a social media news post can influence one’s
motivation to consume the story itself. News consumers are more likely to
click into a news story with a picture (Ulloa et al., 2015). Images that arouse
the emotional response of fear and sympathy have been shown to influence
behavior (Powell, Boomgaarden, De Swert, & de Vreese, 2015), and greater
cognitive processing took place after news stories featuring threatening
images, versus nonthreatening images, were viewed (Sargent, 2007). Thus,
the visual framing and presentation style of news images influence news
consumers’cognitive process and interpretation of the world as well as result
in different judgments about the news value, and motivate subsequent news
consumption.
A second reason why news photos play a significant role in news
consumption is their ability to help readers easily and quickly understand
the topic of the story (Lester, 2000;Rodgers,Kenix,&Thorson,2007)and
make sense of their world (Gross & Aday, 2003; Matthews & Reuss, 1985;
Miller & Roberts, 2010). News images serve as key determinants of the
way a news story frames its subject matter. Gamson and Modigliani
(1987) conceptually defined a news frame as the “central organizing idea
or storyline that provides meaning to an unfolding strip of events”(p.
143). Visual elements, due to their ability to draw users’attention, convey
emotion, and depict participants and setting, play a key role in shaping
meaning (Rodgers et al., 2007). The way the photographer has chosen to
take the picture, the placement of the subject, and the light, all convey a
visual effect (Cappella & Jamieson, 1997), which in turn creates meaning,
or feeling, in the consumer. Images alone were shown to have stronger
framing effects than images with text or text alone (Powell, Boomgaarden,
De Swert, & De Vreese, 2015). Thus, news images not only provide a clear
anddirectinformationcuethatbuildreader’s perception of social reality
(Messaris, 1994;Rodgersetal.,2007), but can also guide readers’
4K. KEIB ET AL.
Downloaded by [174.218.134.92] at 09:28 11 October 2017
knowledge of, and attitude toward, the events or storyline (Iyengar, 1991;
Powell et al., 2015).
Because of their likelihood of being noticed, and the speed with which they
can be processed, images within social media posts can be a significant driver
of whether participants choose to click on the post to view the story. News
stories with images influence selective exposure (Sargent, 2007). The process
of selective exposure to media content can be driven by situational emotional
needs, such as mood modification (Chang, 2006) or situational factors and
personality (Knobloch-Westerwick, Appiah, & Alter, 2008; Zillmann &
Bryant, 1985).
Thus, images in news can play multiple roles: increasing readers’attention,
interest, and motivation, as well as aiding in the decision whether or not to
further consume the news content (Arapakis, Lalmas, Cambazoglu, Marcos,
& Jose, 2014; Bucher & Schumacher, 2006).
Emotional responses to images and their role in framing content
Work by Robin Nabi has demonstrated that consumers’own emotional
responses to content can serve as an additional framing mechanism for
the content via “repeated pairing of certain emotions with particular
ideasoreventseventuallyshapesthewayinwhichoneinterpretsand
responds to those events”(Nabi, 2003, p. 227). Nabi explains that
emotions become frames themselves when they are repeatedly involved
in the collection of information, as well as the storage and recall of it,
then used to make attributions that direct behavior (Nabi, 2003). Recent
declines in news newspaper readership and local TV viewership
(Mitchell & Holcomb, 2016) could be a result of frames being rejected
by the public. If the media are framing news from a negative or fear-
driven place, for example, the public will develop habits as to how to act
based on the message. Therefore, it would be in the media’s best interest
to reframe news content in a manner that would be better received by
the public.
There is also evidence of emotional arousal playing a role in selective
attention (Brosius, 1993; Nabi, 2003). Emotional images have been shown to
enhance recall of news items (Brosius, 1993). However, the same study found
an increase in errors of judgement when participants who viewed emotional
images were asked to make evaluations about what they saw (Brosius, 1993).
Nabi (2003) similarly concluded that the evidence of emotions elicited by
messages can lead consumers to selectively process parts of intended mes-
sages, thus meeting the first requirement that emotions are frames. Other
required elements are information accessibility, information preference, and
the role of decision-making (Nabi, 2003). Support for the presence of these
MEDIA PSYCHOLOGY 5
Downloaded by [174.218.134.92] at 09:28 11 October 2017
elements was found in his study, although Nabi cautions the results could
depend on schema development.
A news frame’s valence generally casts the content as good or bad (de
Vreese & Boomgaarden, 2003). The framing of news as either positive or
negative has received much attention, due primarily to highly publicized
criticism that news is too negative (Trussler & Soroka, 2014). News, itself,
contains a valenced frame because it inherently suggests valenced outcomes,
solutions, or events, for example (de Vreese & Boomgaarden, 2003). Given
that news inherently contains emotion, the question is which emotion leads
to greater engagement with content?
RQ1: For viewers of social media posts promoting news stories, what is the
relationship between image valence (positive, negative) and visual
attention?
Emotions and social sharing
Recent work focusing on the role of emotion in online social sharing has
found that individual expressions of emotion beget other expressions of
emotion, which contribute to a ripple effect that serves to magnify the
intensity of a given emotion across online social networks (Coviello et al.,
2014). EBT contends that people have an intrinsic need to share psycho-
logically arousing (i.e., emotional, harrowing, sad) stories to other people
in their social networks (Harber & Cohen, 2005). This theory of social
sharing argues that the transfer of stories is related to functions of
emotional release and support consolidation. Further, EBT proposes that
the degree to which a particular story emotionally affects the original
teller predicts secondary and tertiary information transfer across social
networks.
One of the main tenets of EBT is that social sharing is driven by the
emotional arousal that follows an encounter with information that challenges
one’s deeply held beliefs and expectations. According to Harber and collea-
gues (2014), news is that which is arousing to the point that sharing it with
others relieves the emotional upheaval, and the receivers benefit from the
shortcut to knowledge that the broadcast has afforded. Previous research has
advanced the notion that consumers are drawn to news entailing events that
are deviant or unusual, due to the potential threat they might pose
(Shoemaker, 1996), and has shown that negative images led to longer expo-
sure times among readers of new stories (Sargent, 2007). The line of research
that has extended the empirical application of the EBT has developed this
argument further by suggesting that people’s responses to news are actually
shaped by schema violation. Specifically, Harber and colleagues (2014) used
6K. KEIB ET AL.
Downloaded by [174.218.134.92] at 09:28 11 October 2017
the discrepancy theory of emotion to explain the role of emotional arousal
that stems from exposure to news, and the subsequent social disclosure
behavior that ensues.
Research that builds on EBT lends support to the argument that
emotional content begets information dissemination. For instance, a
study conducted by Ibrahim, Ye, and Hoffner (2008)foundthatindivi-
duals that reported sadness following the Space Shuttle Columbia dis-
aster were more likely to engage in information transfer and discussion
as emotional coping mechanisms. Other research findings point to the
emotional intensity of the media content as a key driver of the amount
of social sharing that follows exposure. Specifically, Luminet and collea-
gues (2000) conducted an experiment that tested whether exposing
participants to negative emotional stimuli varying in intensity would
result in increased social sharing. The results of this study suggest that
the relationship between emotional intensity and extent of social sharing
is a step-function, rather than linear. In other words, people do not
engage in social sharing unless the information or content they encoun-
ter elicits an emotional response of required, yet undetermined, inten-
sity.Similarly,Berger(2011) showed that emotions associated with high
arousal (i.e., anxiety or amusement) lead to increased sharing in contrast
with emotions that are characterized by low arousal (i.e., sadness or
contentment). The results of Berger’s(2011) study highlight that phy-
siological arousal plays a strong role in the transmission of news,
regardless of valence.
The relationship between high-arousal emotions and social sharing can
also apply to socially-mediated information environments. Berger and
Milkman (2012) found that when online news stories elicited high-arousal
emotions, these stories were more likely to be shared than stories that elicited
lower levels of arousal. Other studies have found similar results when explor-
ing emotions and sharing behavior in social media. Much like the findings in
Luminet and colleagues’(2000) study, Botha (2014) also found that the
intensity of evoked emotions is what drives social sharing—in this case,
however, focusing on the viral success of political satire in online campaigns.
Similarly, Stieglitz and Dang-Xuan (2013) also analyzed the emotional inten-
sity and polarity of political tweets and found that emotionally charged
content was more likely to be retweeted, and more quickly than neutral
tweets.
In sum, emotions—perhaps particularly ones that are highly arousing
or negative in valence—play a role in what gets attention and engagement
on social media. The literature on the role of images in evoking such
emotions led us to propose the following hypotheses and research
question:
MEDIA PSYCHOLOGY 7
Downloaded by [174.218.134.92] at 09:28 11 October 2017
H1: Viewers of social media posts that contain an emotionally valenced
image will indicate a greater willingness to click to read the story than
those who view posts without an image.
H2: Viewers of social media posts that contain an emotionally valenced
image will indicate a greater willingness to share the story link than those
who view posts without an image.
RQ2: For viewers of social media posts that promote news stories, what is
the relationship between image valence and the level of arousal?
Method
The proposed research questions and hypotheses were tested in a single-
variable, three-level (image valence: positive vs. negative vs. no image)
within-subjects eye-tracking experiment. Each participant viewed a series of
29 social media posts promoting news stories, each of which either included
no image, a positive image, or a negative image.
1
Stimulus materials were
presented either as Facebook posts or Twitter posts to match participants’
most-used platform, as reported in a pre-experiment questionnaire.
Participants were asked to view each image for long as they wanted, and
were allowed to advance to each subsequent image at their own pace by
pressing a keyboard key.
Participants
Participants (N= 60) were undergraduate students form a large south-
eastern university. Mean participant age was 20.3 years (SD =1.3),andthe
participant sample was 73% female. The racial distribution of the sample
was as follows, 73% Caucasian or White, 17% Asian, and 7% African
American, with respondents marking other or multiple races comprising
therestofthesample.FivepercentofthesampleidentifiedasHispanic/
Latino.
Participants answered a prestudy questionnaire to determine their pre-
ferred social media platform for news consumption, Facebook or Twitter, to
reduce threats to ecological validity caused by creating artificial situations
(Leckner, 2012). The questionnaire also included questions about time spent
on social media and time spent consuming news. Mean time spent per day
on social media was 2.5 hr (SD = 1.5), mean time spent per week consuming
news was 3.3 hr (SD = 2.5). Participants signed an informed consent form
prior to participation.
8K. KEIB ET AL.
Downloaded by [174.218.134.92] at 09:28 11 October 2017
Stimulus materials
Twenty-nine news stories were selected for which a positively or negatively
valenced image would be appropriate without altering the headline. News
stories were chosen to represent a broad selection of online news outlets
(e.g., ABC News, Vox, Washington Post). For each news story, six versions
of a realistic social media post promoting the news story were created. Each
post contained text and source information held constant across all condi-
tions, along with either a positive image, negative image, or no image. For
each of the 29 stories, three versions of the post for the three image
conditions were created to resemble Facebook posts, and three versions of
the post were created to resemble Twitter posts (see Figure 1 for examples).
Each participant viewed one post each for all 29 stories, a third of which
were shown with a positive image, a third with a negative image, and a
third with no image at all. Specific story-image pairings varied as a three-
level between-subjects condition, and the presentation of the stories was
randomized for each participant. Positively and negatively valenced images
were selected from a broader set of images based on a pr-test (N= 126) to
be certain they properly elicited positive or negative valence using the Self-
Assessment Manikin (SAM) to measure the pleasure, arousal, dominance
(PAD) scale. PAD was developed by Russell and Mehrabian in 1977 and
these three elements represent the necessary and sufficient indicators of
emotion (Russell & Mehrabian, 1977). This scale has been repeatedly found
reliable (Rodgers & Thorson, 2012). Respondents used the manikin to
evaluate the valence of the images by rating the extent to which the
image made them feel pleasant or unpleasant on a nine-point Likert scale.
Four images were pretested for each of 41 stories in the pretest, and final
stories and images were chosen based on the availability of two images that
differed significantly in emotional valence. The 29 final stories were
Positive image condition, Facebook format
Negative image condition, Twitter Format
Figure 1. Examples of stimulus materials in positive and negative conditions (News story #12
shown).
MEDIA PSYCHOLOGY 9
Downloaded by [174.218.134.92] at 09:28 11 October 2017
clustered into three groups so that each group of 9 or 10 stories had similar
average positive scores and negative scores.
Procedure
Participation in the study occurred one participant at a time in a campus lab
equipped for eye-tracking research. Each participant was exposed to 29 social
media posts dealing with a diverse set of news story topics (e.g., politics,
money, health, entertainment). Each participant was randomly assigned to
one of three permutations of the stimulus materials, each of which paired a
third of the stories with a negative image, a third with a positive image, and a
third with no image. While participants viewed the images, their visual
attention was captured by a Tobii X2-60 eye-tracker, allowing researchers
to examine fixation duration and time to first fixation for each area of
interest within the post.
Participants were instructed to click a button at the bottom of each image
screen to advance to next image when they had finished viewing the current
image. Image order was varied using a Latin-square design to avoid presen-
tation order effects. After viewing all 29 images, participants completed a
questionnaire consisting of a thumbnail image and four measures for each
image: perceived emotional valence of the post, perceived arousal level of the
post, likelihood to click through to read the story, and likelihood of sharing
the story with members of their social network.
Dependent measures
Visual attention to the news post wasmeasuredasthetotalamountof
time in which the user recorded fixations in the area of the news post.
The eye tracker recorded participant’spupilfixationsatasamplingrate
of60Hz,or60samplespersecond.Eachparticipant’stimespent
fixating within rectangular borders of the social media post was com-
puted for each image. Subsequently, for each participant, three separate
mean attention scores were calculated by averaging attention to all posts
with positive images, all posts with negative images, and all posts with
no images (M=2.21, SD = 1.71). Means by valence are reported in
Table 1.
Intention to click was measured by asking “How likely would you be to
click through and read this story?”(Shamdasani, Stanaland, & Tan, 2001).
Participants answered the question using a 7-point semantic differential scale
ranging from very unlikely to very likely. For each participant, three separate
mean intention to click scores by averaging attention to posts with positive
images, posts with negative images, and posts with no images (M=4.24,
SD = .8). Means by valence are presented in Table 1.
10 K. KEIB ET AL.
Downloaded by [174.218.134.92] at 09:28 11 October 2017
Intention to share was measured by asking “How likely would you be to
share this story by clicking the ‘share’button?”(Twitter question would say
“by retweeting this post?”) (Wojdynski & Evans, 2016). Participants
answered the question using the same 7-point scale ranging from very
unlikely to very likely, but with the addition of the option, “I don’t have
sufficient information to decide.”For each participant, three separate mean
intention to click scores were calculated by averaging attention to posts with
positive images, posts with negative images, and posts with no images
(M=2.83, SD = .99). Means by valence are presented in Table 1.
Self-reported arousal level was measured using the arousal scale from the
SAM (Rodgers & Thorson, 2012). After participants saw a news post, they
were asked to rate the image using the 9-point scale. The levels for arousal
ranged from calm (1) to exciting (9). In between each of the poles of the scale,
five manikin images were used to depict different levels of arousal. For each
participant, three separate mean arousal scores were calculated by averaging
attention to posts with positive images, posts with negative images, and posts
with no images, (M=3.91, SD = 1.21). Means by valence are presented in
Table 1. Paired-samples t-tests are used as the primary statistical test, because
the variables being compared vary within subjects, not between subjects.
Emotional valence was measured using the pleasure scale from the SAM
(Rodgers & Thorson, 2012). After participants saw a news post, they were
also asked to rate the image using the 9-point scale. The levels for emotion
ranged from unpleasant (1) to pleasant (9). In between the poles of the scale,
five manikin images were used to depict different levels of emotion. For each
participant, three separate mean emotional valence scores were calculated by
averaging attention to posts with positive images, posts with negative images,
and posts with no images (M=4.89, SD = 1). Means by valence are presented
in Table 1.
Zero-order correlations between the dependent measures for posts with
positively valenced images ranged from –.061 to .452. For the no-image
posts, correlations all ranged between –.058 and .419 with one exception,
r= .610 between intent to click and intent to share. For the no-images posts:
correlations all ranged between –.085 and .420 with two exception, r= .525
between intent to click and intent to share, and r= .549 between intent to
click and arousal. The only correlations between measures pertaining to posts
Table 1. Within-subjects mean differences between post type conditions.
Variable No Image Positive Image Negative image
Visual attention (sec) 1.53 (SD = 1.21)
a
2.53 (SD = 1.93)
b
2.57 (SD = 1.99)
b
Intention to click 4.14 (SD = .756)
a
4.38 (SD = .832)
b
4.24 (SD = .897)
a,b
Intention to share 2.72 (SD = .913)
a
2.97 (SD = 1.09)
b
2.79 (SD = .869)
a,b
Level of arousal 3.70 (SD = 1.30)
a
4.29 (SD = 1.17)
b
4.29 (SD = 1.17)
b
Emotional valence 4.67 (SD = 1.08)
a
5.53 (SD = .845)
b
4.47 (SD = 1.07)
a
Means sharing the same superscript are not significantly different from each other (p < . 05).
MEDIA PSYCHOLOGY 11
Downloaded by [174.218.134.92] at 09:28 11 October 2017
of a different valence that exceeded r= .40 were between visual attention
measures, which were r= .58 for the positive/no-image and the negative/no-
image pairings, and r= .79 for the positive/negative pairing.
Results
RQ1, regarding the relationship between post image content and visual atten-
tion, was examined using paired-samples t-tests between each pair of condi-
tions. The results showed that posts with positive images (M=2.53,SD =1.93,t
(59) = 4.91, p< .001) and negative images (M=2.57,SD =1.99,t(59) = 4.97,
p< .001) received significantly greater visual attention than posts with no
images (M=1.53,SD = 1.21). The difference between visual attention to
posts with positive images and those with negative images was not significant.
Hypothesis 1, regarding the relationship between post image content and
intention to click, was examined using paired-samples t-tests between each pair
of within-subjects. The results showed that posts with positive images
(M=4.38,SD = 0.83, t(58) = 2.93, p< .01) led to higher intention to click
than posts with no images (M=4.14,SD = 0.76). The difference in intention to
click between negative (M= 4.21, SD = 0.81) and positive images and negative
and no images was not significant. Thus, H1 was partially supported.
Hypothesis 2, regarding the relationship between post image content and
intention to share, was examined using paired-samples t-tests between each
pair of conditions. These samples were of a smaller size because participants
had the option to reply that they did not have enough information to answer
the question. In total, three people did not answer the question. The results
showed that posts with positive images (M= 2.97, SD = 1.09, t(59) = 2.52,
p< .05) led to higher intention to share than posts with no images (M= 2.72,
SD = 0.91). There was no significant difference in intention to share between
no images and negative images, (M= 2.79, SD = 0.97), or positive and
negative images. Thus, H2 was partially supported.
RQ2, regarding the relationship between post image content and level of
arousal, was examined using paired-samples t-tests between each pair of
conditions. The results showed that posts with positive images (M= 4.29,
SD = 1.17, t(59) = 5.25, p< .001) and negative images (M= 4.23,
SD = 1.17, t(59) = 4.46, p< .001) led to greater levels of self-reported
arousal than posts with no images (M=3.20,SD = 1.30). There was no
difference in arousal between posts with positive images and those with
negative images.
Post-Hoc mediation analysis
In light of the significant differences shown between posts with a positive
image and those with no images on arousal, intention to share, and intention
12 K. KEIB ET AL.
Downloaded by [174.218.134.92] at 09:28 11 October 2017
to click, additional analyses were undertaken to examine whether the differ-
ences in pleasure and arousal mediated the relationship between post type
and clicking, and the relationship between post type and sharing. Although
within-subjects mediation analysis was previously only available piecemeal by
examining components of the model (Judd, Kenny, & McClelland, 2001),
recent developments have allowed the testing of within-subjects mediation
through a path-analysis based series of regression (Montoya & Hayes, 2017).
The MEMORE macro for SPSS and SAS, developed by Andrea Montoya,
allows not only for the calculation of indirect effect size, but also methods for
inference including bootstrap confidence intervals that mirror those used to
estimate indirect effect size in between-subjects designs (Hayes, 2013;
Montoya & Hayes, 2017).
To examine the potential concurrent mediation of the relationship post
type (positive vs neutral) on intention to click by pleasure (level of emotional
valence) and arousal (see Figure 2 for diagram with path weights), a within-
subjects mediation was conducted using the MEMORE macro in SPSS. The
results of the regression analyses showed a significant total effect of post type
on intention to click, b= .25 (s.e. = .09), t(58) = 2.93, p< .01. (all bs reflect
the increase from neutral to positive). Post type also had significant direct
effects on both arousal, b= .58 (s.e. = .11), t(58) = 5.14, p< .001 and pleasure,
b= .88 (s.e. = .15), t(58) = 5.84, p< .001. On the right side of the model,
although there was no significant relationship between arousal and intention
to click, the path between pleasure and intention to click was significant,
b= .24 (s.e. = .09), t(54) = 2.977, p< .01. Bias-corrected bootstrapping
(10,000 samples) of the total indirect effect of post type on intention to
Figure 2. Mediation of relationship between post type (positive vs. no image) on intention to click.
Note: N = 58, two participants were excluded from analysis because of missing intent to click values.
Path values represent unstandardized regression coefficients. * p<.05,**p<.01,***p<.001.
MEDIA PSYCHOLOGY 13
Downloaded by [174.218.134.92] at 09:28 11 October 2017
click via pleasure estimated the effect at .21, 95% CI: .02 –.40,
z= 2.51, p< .05.
To examine the potential concurrent mediation of the effect of post type
(positive vs neutral) on intention to share the post by pleasure and arousal
(see Figure 3 for diagram with path weights), a within-subjects mediation
analysis was conducted using the MEMORE macro in SPSS. The results of
the regression analyses showed a significant total effect of post type on
intention to click, b= .25 (s.e. = .10), t(59) = 2.52, p< .05. (all bs reflect
the increase from neutral to positive). Post type also had significant direct
effects on both arousal, b= .58 (s.e. = .11), t(59) = 5.14, p< .001 and pleasure,
b= .86 (s.e. = .15), t(59) = 5.84, p< .001. On the right side of the model,
neither arousal nor pleasure was a significant predictor of intent to share,
indicating no mediation.
Discussion
Social media increasingly play a role in processes of online news creation,
consumption, and engagement, as well as dissemination. Indeed, previous
research positions social media as a legitimate source of news and informa-
tion for its users (Barthel et al., 2015). In light of these findings, under-
standing editorial decisions that can potentially influence audience attention
and engagement with online news content in social media becomes
paramount.
The image seen on the social media post acts as a frame for the corre-
sponding content. This frame, in turn, leads the consumer to make a decision
about what to consume (de Vrees & Boomgaarden, 2003; Nabi, 2003). Our
results show that intent to engage with content by clicking or sharing
Figure 3. Mediation of relationship between post type (positive vs. no image) on intention to share.
Note: N = 59, one participant was excluded from analysis because of missing intent to share values.
Path values represent unstandardized regression coefficients.* p<.05,***p< .001
14 K. KEIB ET AL.
Downloaded by [174.218.134.92] at 09:28 11 October 2017
increases when the post has an image, and in particular a positive image. The
use of images versus text alone then should be standard practice for practi-
tioners wishing to get the most engagement and consumption of their
content. Through careful selection of an image, and choosing one that
conveys a positive emotional valence, news organizations and journalists
could see growing audiences.
The results suggest higher levels of emotion, higher intention to share, and
higher intention to click when participants viewed positive images, compared
to posts with no images, which stands in contrast with previous research that
advocates for the prominent role of negative emotions and events in audience
predilections for news, the ensuing emotional arousal, and the social sharing
it prompts (Harber et al., 2014; Shoemaker & Cohen, 2006; Trussler &
Soroka, 2014). Recent work continues to show that negative news is selected
more than positive news (Trussler & Soroka, 2014). Berger and Milkman
however, found that positive news was more viral, in other words, more
likely to be shared (Berger & Milkman, 2012). In a content analysis of the
New York Times, they found more positively valanced articles versus nega-
tively valenced article in the most-emailed list (Berger & Milkman, 2012). Our
results mirror this. This behavior may be due to the notion that sharing on
social media is somewhat public and consumers may consider sharing as a
reflection of their personal brand. Thus, sharing positive content may be
more appealing in regards to self-curated representations, rather than caus-
ing emotional upheaval by sharing negative stories. This study contributes to
the extant media effects literature that considers the implications of emo-
tions. Leaning on the notion of visual attention as a proxy for appraisal—one
of the three main components of emotions (Tan, 2008)—this study suggests
that the impression a user makes about an image takes place quickly.
Specifically, our results show that the mean times spent looking at social
media posts with positive, negative, and no images were 2.5 (SD = 1.93), 2.6
(SD = 1.99), and 1.53 (SD = 1.21) sec, respectively. As called for by
Vuilleumier (2005), this study also suggests extensions to cognitive research
through the incorporation of positive visual stimuli when exploring the effect
of emotional sensory events in appraisal and attention.
Our findings show that level of arousal was highest for posts with images,
but the direction of valence did not impact arousal. Images attracted longer
gaze than posts with no image, captured by eye tracking, but here again,
valence did not differ significantly in visual attention. These results support
past work that showed images attracted more attention than text only posts
on social media (Guerini et al., 2013). The content of the image is a factor
used on social media by consumers when evaluating information. Although
the direction of valence was not a factor, we do know that the valence of
image elicited emotion-based on our pretest. Emotion does play a role in how
users consume news content on social media.
MEDIA PSYCHOLOGY 15
Downloaded by [174.218.134.92] at 09:28 11 October 2017
Limitations and future work
Our study is limited in several ways. The first limitation involves the stimuli
material for the experiment. Due to the nature of data collection, we could
not use breaking news stories as part of our study. However, stories regarding
the latest events or developments are more likely to be part of the content
featured by news media organizations, as well as the posts users are more
likely to share with other people in their social media networks. A second
limitation concerns our sample. Because the latter consists of undergraduate
students from a large southeastern university, the findings of our study are
not representative of the general American population.
Although this study focused specifically on positively and negatively
valenced visual stimuli, as determined by participants’pleasure, arousal,
and dominance assessments, emotions are certainly more nuanced and
complex. As one example, a given stimulus could be rated as negatively
valenced and arousing because it triggers fear or because it triggers anger.
Although work to date (Berger, 2011; Berger & Milkman, 2012) has not
shown broad differences between these emotions, it’s possible that users may
situationally be more likely to share one type of content versus the other;
men, who in interpersonal contexts are less likely to express fear (Braungart,
Braungart, & Hoyer, 1980), might be less likely to share content that makes
them fearful. Because previous media use studies have highlighted the ben-
efits of a discrete and dimensional approach to the study of emotion (Lee &
Lang, 2009), future studies should certainly consider the role of discrete
emotion on consumption of and engagement with online news content.
Future research should apply the findings of this experiment in other
contexts and using a more representative sample to examine the effects of
image selection on audience engagement with news content on social media.
One example of such possible future applications of this study include
exploring whether the proposed relationships still apply when users engage
with social media content through mobile platforms. Because researchers
have highlighted mobile as the fastest growing platform for news and infor-
mation consumption (Mitchell, 2015), this might be an ideal opportunity to
further advance our understanding of the relationship between images,
emotion, and social sharing. Last, researchers should also explore engage-
ment with news content in additional social media platforms that are more
image or video driven (i.e., Instagram or YouTube).
Conclusion
The purpose of this study was to understand how visual elements on social
media drive audience engagement. Employing both eye-tracking and self-report
measures, this project builds on previous work on images and emotion, as well
16 K. KEIB ET AL.
Downloaded by [174.218.134.92] at 09:28 11 October 2017
as social sharing, and explores the nuances of how consumers engage with news
on social media. In line with recent insights that underscore the importance of
social media platforms in the consumption and dissemination of online news
content, this study contributes a clearer understanding of the role of positive
visual elements in attention, as well as prospective clicking and sharing beha-
vior. In effect, the findings of this study hold considerable practical implica-
tions, as news editors and journalists can now make more focused decisions in
terms of image selection, ultimately ensuring that their content is designed for
optimal audience engagement.
Note
1. Positive, negative, and no-image versions were crafted for each of the 29 stories, and
each participant viewed an even mix of the three post types. Participants were assigned
to one of three between-subjects permutations, with different third of the stories to
each (positive, negative, and neutral) condition. One condition had stories 1-10 nega-
tive, stories 11-20 as positive, and stories 21-29 as neutral, another condition would
have 1-10 as neutral, 11-20 as negative and 21-29 as positive, and the third completed
the balanced design. The actual order of presentation of the posts was algorithmically
randomized.
References
Anderson, M., & Caumont, A. (2014). How social media is reshaping news. Retrieved from
http://www.pewresearch.org/fact-tank/2014/09/24/how-social-media-is-reshaping-news/
Arapakis, I., Lalmas, M., Cambazoglu, B., Marcos, M.-C., & Jose, J. M. (2014). User engage-
ment in online news: Under the scope of sentiment, interest, affect, and gaze. Journal of the
Association for Information Science & Technology,65(10), 1988–2005. doi:10.1002/
asi.23096
Barthel, M., Shearer, E., Gottfried, J., & Mitchell, A. (2015). The evolving role of news on
twitter and facebook. Retrieved from http://www.journalism.org/2015/07/14/the-evolving-
role-of-news-on-twitter-and-facebook/
Berger, J. (2011). Arousal increases social transmission of information. Psychological Science,
22, 891–893. doi:10.1177/0956797611413294
Berger, J., & Milkman, K. L. (2012). What makes online content viral? Journal of Marketing
Research,49(2), 192–205. doi:10.1509/jmr.10.0353
Botha, E. (2014). A means to an end: Using political satire to go viral. Public Relations Review,
40, 363–374. doi:10.1016/j.pubrev.2013.11.023
Braungart, M. M., Braungart, R. G., & Hoyer, W. J. (1980). Age, sex, and social factors in fear
of crime. Sociological Focus,13(1), 55–66. doi:10.1080/00380237.1980.10570360
Brosius, H. (1993). The effects of emotional pictures in television news. Communication
Research,20(1), 105–124. doi:10.1177/009365093020001005
Bucher, H., & Schumacher, P. (2006). The relevance of attention for selecting news content.
An eye-tracking study on attention patterns in the reception of print and online media.
Communications-European Journal of Communication Research,31, 347–368.
Cappella, J. N., & Jamieson, K. H. (1997). Spiral of cynicism: The press and the public good
Oxford. Oxford, UK: Oxford University Press.
MEDIA PSYCHOLOGY 17
Downloaded by [174.218.134.92] at 09:28 11 October 2017
Chang, C. (2006). Beating the news blues: Mood repair through exposure to advertising.
Journal of Communication,56(1), 198–217. doi:10.1111/jcom.2006.56.issue-1
Coviello, L., Sohn, Y., Kramer, A. D. I., Marlow, C., Franceschetti, M., Christakis, N. A., &
Fowler, J. H. (2014). Detecting emotional contagion in massive social networks. Plos One,9
(3), e90315–e90315. doi:10.1371/journal.pone.0090315
de Vreese, C. H., & Boomgaarden, H. (2003). Valenced news frames and public support for
the EU. Paper presented at the annual meeting of the International Communication
Association, Marriott Hotel, San Diego, CA, May 26, 2009. Retrieved from http://www.
allacademic.com/meta/p111822_index.html
DeMers, J. (2016). Facebook live: Everything you need to know. Retrieved from https://www.
forbes.com/sites/jaysondemers/2016/04/26/facebook-live-everything-you-need-to-know/
#e8d1d7855f2f
Fogg, B. J. (2003, April). Prominence-interpretation theory: Explaining how people assess
credibility online. In G. Cockton, & P. Korhonen (Eds.), CHI’03 extended abstracts on
human factors in computing systems (pp. 722–723). New York, NY: ACM.
Gamson, W. A., & Modigliani, A. (1987). The changing culture of affirmative action. In R. G.
Braungart & M. M. Braungart (Eds.), Research in political sociology (Vol. 3, pp. 137–177).
Greenwich, CT: JAI Press.
Gross, K., & Aday, S. (2003). The scary world in your living room and neighborhood: Using local
broadcast news, neighborhood crime rates, and personal experience to test agenda setting and
cultivation. Journal of Communication,53,411–426. doi:10.1111/jcom.2003.53.issue-3
Guerini, M., Staiano, J., & Albanese, D. (2013). Exploring image virality in google plus. In
SOCIALCOM '13 Proceedings of the 2013 International Conference on Social Computing
(pp. 671–678). Washington, DC: IEEE Computer Society.
Harber, K. D., & Cohen, D. J. (2005). The emotional broadcaster theory of social sharing.
Journal of Language & Social Psychology,24, 382–400. doi:10.1177/0261927X05281426
Harber, K. D., Podolski, P., & Dyer, L. (2014). Hearing stories that violate expectations leads
to emotional broadcasting. Journal of Language and Social Psychology,33(1), 5–28.
doi:10.1177/0261927X13502793
Hayes, A. F. (2013). Introduction to mediation, moderation, and conditional process analysis.
NewYork, NY: Guilford.
Hermida, A. (2012). Tweets and truth: Journalism as a discipline of collaborative verification.
Journalism Practice,6, 659–668. doi:10.1080/17512786.2012.667269
Hille, S., & Bakker, P. (2013). I like news. searching for the ‘Holy grail’of social media: The
use of facebook by dutch news media and their audiences. European Journal of
Communication,28,663–680. doi:10.1177/0267323113497435
Hille, S., & Bakker, P. (2014). Engaging the social news user. Journalism Practice,8(5), 563–
572. doi:10.1080/17512786.2014.899758
Ibrahim, A., Ye, J., & Hoffner, C. (2008). Diffusion of news of the Shuttle Columbia disaster:
The role of emotional responses and motives for interpersonal communication.
Communication Research Reports,25(2), 91–101. doi:10.1080/08824090802021970
Ingram, M. (2015). Facebook has taken over from google as a traffic source for news. Retrieved
from http://fortune.com/2015/08/18/facebook-google/
Iyengar, S. (1991). Is anyone responsible? Chicago, IL: University of Chicago Press.
Judd, C. M., Kenny, D. A., & McClelland, G. H. (2001). Estimating and testing mediation and
moderation in within-subject designs. Psychological Methods,6, 115–134. doi:10.1037/
1082-989X.6.2.115
Kenix, L. J. (2013). A converging image? Commercialism and the visual identity of alternative
and mainstream news websites. Journalism Studies,14, 835–856. doi:10.1080/
1461670X.2012.754239
18 K. KEIB ET AL.
Downloaded by [174.218.134.92] at 09:28 11 October 2017
Knobloch-Westerwick, S., Appiah, O., & Alter, S. (2008). News selection patterns as a
function of race: The discerning minority and the indiscriminating majority. Media
Psychology,11, 400–417. doi:10.1080/15213260802178542
Lazarus, R. S. (1991). Progress on a cognitive-motivational-relational theory of emotion.
American Psychologist,46, 819–834. doi:10.1037/0003-066X.46.8.819
Leckner, S. (2012). Presentation factors affecting reading behaviour in readers of newspaper
media: An eye-tracking perspective. Visual Communication,11, 163–184. doi:10.1177/
1470357211434029
Lee, S., & Lang, A. (2009). Discrete emotion and motivation: Relative activation in the
appetitive and aversive motivational systems as a function of anger, sadness, fear, and
joy during televised information campaigns. Media Psychology,12, 148–170. doi:10.1080/
15213260902849927
Lester, P. (2000). Visual communication: Images with message (2nd ed.). Belmont, CA:
Wadsworth.
Luminet, IV, O., Bouts, P., Delie, F., Manstead, A. S., & Rimé, B. (2000). Social sharing of
emotion following exposure to a negatively valenced situation. Cognition & Emotion,14,
661–688. doi:10.1080/02699930050117666
Matsa, K. E. (2016). Facebook, twitter play different roles in connecting mobile readers to
news. Retrieved from http://www.pewresearch.org/fact-tank/2016/05/09/facebook-twitter-
mobile-news/
Matthews, M. L., & Reuss, C. (1985, August). The minimal image of women in time and
Newsweek, 1940–1980. Paper presented to the Magazine Division at the annual conference
of the Association for Education in Journalism and Mass Communication, Memphis, TN.
Messaris, P. (1994). Visual “literacy”: Image, mind and reality. Boulder, CO: Westview.
Miller, A., & Roberts, S. (2010). Visual agenda-setting & proximity after hurricane katrina: A
study of those closest to the event. Visual Communication Quarterly,17,31–46.
doi:10.1080/15551390903553697
Mitchell, A. (2015). State of the news media 2015. Retrieved from http://www.journalism.org/
2015/04/29/state-of-the-news-media-2015/
Mitchell,A.,&Holcomb,J.(2016).State of the News Media 2016. Pew Research Center. Retrieved
from https://assets.pewresearch.org/wp-content/uploads/sites/13/2016/06/30143308/state-of-
the-news-media-report-2016-final.pdf
Montoya, A. K., & Hayes, A. F. (2017). Two condition within-participant statistical mediation
analysis: A path-analytic framework. Psychological Methods,22(1), 6.
Mullin, B. (2015). News outlets started publishing within facebook today. Retrieved from
http://www.poynter.org/2015/news-outlets-started-publishing-within-facebook-today/
344209/
Nabi, R. L. (2003). Exploring the framing effects of emotion: Do discrete emotions differen-
tially influence information accessibility, information seeking, and policy preference?
Communication Research,30, 224–247. doi:10.1177/0093650202250881
Oeldorf-Hirsch, A., & Sundar, S. S. (2015). Posting, commenting, and tagging: Effects of
sharing news stories on Facebook. Computers in Human Behavior,44, 240–249.
doi:10.1016/j.chb.2014.11.024
Ortony, A., Clore, G. L., & Collins, A. (1990). The cognitive structure of emotions. Cambridge,
UK: Cambridge University Press.
Powell, T. E., Boomgaarden, H. G., De Swert, K., & De Vreese, C. H. (2015). A clearer picture:
The contribution of visuals and text to framing effects. Journal of Communication,65, 997–
1017. doi:10.1111/jcom.12184
Rodgers, S., Kenix, L. J., & Thorson, E. (2007). Stereotypical portrayals of emotionality in news
photos. Mass Communication & Society,10(1), 119–138. doi:10.1080/15205430709337007
MEDIA PSYCHOLOGY 19
Downloaded by [174.218.134.92] at 09:28 11 October 2017
Rodgers, S., & Thorson, E. (2012). Advertising theory. New York, NY: Routledge.
Russell, J. A., & Mehrabian, A. (1977). Evidence for a three-factor theory of emotions. Journal
of Research in Personality,11, 273–294. doi:10.1016/0092-6566(77)90037-X
Sargent, S. L. (2007). Image effects on selective exposure to computer-mediated news stories.
Computers in Human Behavior,23(1), 705–726. doi:10.1016/j.chb.2004.11.005
Shamdasani, P. N., Stanaland, A. J. S., & Tan, J. (2001). Location, location, location: Insights
for advertising placement on the web. Journal of Advertising Research,41(4), 7–21.
doi:10.2501/JAR-41-4-7-21
Shoemaker, P. J. (1996). Hardwired for news: Using biological and cultural evolution to
explain the surveillance function. Journal of Communication,46(3), 32–47. doi:10.1111/
jcom.1996.46.issue-3
Shoemaker, P. J., & Cohen, A. A. (2006). News around the world: Content, practitioners, and
the public. New York, NY: Routledge.
Shuman, V., Clark-Polner, E., Meuleman, B., Sander, D., & Scherer, K. R. (2017). Emotion
perception from a componential perspective. Cognition and Emotion,31(1), 47–56.
Stieglitz, S., & Dang-Xuan, L. (2013). Emotions and information diffusion in social media
sentiment of microblogs and sharing behavior. Journal of Management Information
Systems,29(4), 217–248. doi:10.2753/MIS0742-1222290408
Tan, E. S. H. (2008). Entertainment is emotion: The functional architecture of the entertain-
ment experience. Media Psychology,11(1), 28–51. doi:10.1080/15213260701853161
Truong, A. (2016). Facebook drives more traffic to articles, but twitter users spend more time
reading them. Retrieved from https://qz.com/676677/facebook-drives-more-traffic-to-arti
cles-but-twitter-users-spend-more-time-reading-them/
Trussler, M., & Soroka, S. (2014). Consumer demand for cynical and negative news frames.
International Journal of Press/Politics,19, 360–379. doi:10.1177/1940161214524832
Ulloa, L. C., Mora, M.-C., Pros, R. C., & Tarrida, A. C. (2015). News photography for
Facebook: Effect of images on the visual behavior of readers in three simulated newspaper
formats. Information Research,20(1), 315–333.
Vuilleumier, P. (2005). How brains beware: Neural mechanisms of emotional attention.
Trends in Cognitive Sciences,9, 585–594. doi:10.1016/j.tics.2005.10.011
Wojdynski, B. W., & Evans, N. J. (2016). Going native: Effects of disclosure position and
language on the recognition and evaluation of online native advertising. Journal of
Advertising,45, 157–168. doi:10.1080/00913367.2015.1115380
Zillmann, D., & Bryant, J. (1985). Selective exposure to communication. Hillsdale, N.J: L.
Erlbaum Associates, 1985.
20 K. KEIB ET AL.
Downloaded by [174.218.134.92] at 09:28 11 October 2017
