ArticlePDF Available

Bad News Has Wings: Dread Risk Mediates Social Amplification in Risk Communication: Social Amplification in Risk Communication

Wiley
Risk Analysis
Authors:

Abstract and Figures

Social diffusion of information amplifies risk through processes of birth, death, and distortion of message content. Dread risk—involving uncontrollable, fatal, involuntary, and catastrophic outcomes (e.g., terrorist attacks and nuclear accidents)—may be particularly susceptible to amplification because of the psychological biases inherent in dread risk avoidance. To test this, initially balanced information about high or low dread topics was given to a set of individuals who then communicated this information through diffusion chains, each person passing a message to the next. A subset of these chains were also reexposed to the original information. We measured prior knowledge, perceived risk before and after transmission, and, at each link, number of positive and negative statements. Results showed that the more a message was transmitted the more negative statements it contained. This was highest for the high dread topic. Increased perceived risk and production of negative messages was closely related to the amount of negative information that was received, with domain knowledge mitigating this effect. Reexposue to the initial information was ineffectual in reducing bias, demonstrating the enhanced danger of socially transmitted information.
Content may be subject to copyright.
Risk Analysis, Vol. xx, No. x, xxxx DOI: xxxx
Bad News Has Wings: Dread Risk Mediates Social
Amplification in Risk Communication
Robert D. Jagiello,1Thomas T. Hills2
Social diffusion of information amplifies risk through processes of birth, death, and
distortion of message content(1) . Dread risk—involving uncontrollable, fatal, involuntary,
and catastrophic outcomes (e.g., terrorist attacks and nuclear accidents)—may be
particularly susceptible to amplification because of the psychological biases inherent in
dread risk avoidance. To test this, initially balanced information about high or low dread
topics was given to a set of individuals who then communicated this information through
diffusion chains, each person passing a message to the next. A subset of these chains
were also re-exposed to the original information. We measured prior knowledge, perceived
risk before and after transmission and, at each link, number of positive and negative
statements. Results showed that the more a message was transmitted the more negative
statements it contained. This was highest for the high dread topic. Increased perceived
risk and production of negative messages was closely related to the amount of negative
information that was received, with domain knowledge mitigating this effect. Re-exposure
to the initial information was ineffectual in reducing bias, demonstrating the enhanced
danger of socially transmitted information.
KEY WORDS: risk perception, social risk amplification, social influence, public opinion,
nuclear power, food additives
1. INTRODUCTION
The development of effective risk communica-
tion methods that minimize public anxiety relies
on understanding how risk information propagates
through social communication channels(1) . This is
notoriously difficult to predict because public opinion
on topics such as climate change, nuclear energy,
disease risks, and immigration rapidly become polar-
ized(2) and at variance with scientific evidence (3,4,5).
For instance, following the first Ebola case diagnosed
in the United States, Twitter posts mentioning Ebola
jumped from 100 per minute to 6000 per minute and
1Department of Psychology, University of Warwick, Uni-
versity Road, Coventry, CV4 7AL, United Kingdom,
R.D.Jagiello@gmail.com
2Department of Psychology, University of Warwick, Uni-
versity Road, Coventry, CV4 7AL, United Kingdom,
T.T.Hills@warwick.ac.uk
quickly incorporated inaccurate claims that it could
be transmitted through the air, water, and food (6) .
This apparent unpredictability is common to socially
propagated risk information, which is often com-
municated via friends, family, online forums, blogs,
and other forms of social media. Though numerous
factors influence risk perception (7,8,9), substantial
work has shown the influence of social contagion
on behaviour and attitudes(2,10,11,12), with specific
factors including interpersonal proximity through
social networks(13,14) and group size (15,16,17,18) .
Many of these studies suggest that when scaled
to populations, information dynamics may exhibit
nonlinear amplification patterns leading to rapid
changes in attitudes and behavior. A recent study
by Moussaid, Brighton and Gaissmaier(1) examined
this directly by quantifying information evolution
around a controversial topic as a message passed
along a chain of people. This evolution led to
10272-4332/18/0100-0001$22.00/1 i
C2018 Society for Risk Analysis
2 Robert D. Jagiello, Thomas T. Hills
social risk amplification (19,20,21,22)in the form of
communication bias towards negative statements.
A significant challenge remains in understanding
how social propagation of information responds to
different kinds of risks. One form of risk capable of
inducing substantial fear is dread risk. According
to a psychological analysis of the taxonomy of
hazards(3,4,23,24), high dread risks (such as nuclear
accidents, plane crashes, and terrorism) are char-
acterized as uncontrollable, fatal, involuntary, and
catastrophic to human life. Low dread risks (such
as food additives, chlorinated water, and low levels
of pollution) are seen as controllable, non-fatal, and
voluntary. High dread leads to stricter calls for
governmental regulations (e.g. nuclear plants) and
higher levels of risk avoidance than low dread risks
associated with the same death toll (23,25). This asym-
metry in hazard perception is highly relevant to risk
judgment considering that elevated dread induces the
use of an affect heuristic(26), or ’risk as feelings’ (27) .
Specifically, within the framework of dual processing
theories(28,29), the affect heuristic is an automatic
process that prioritizes emotional information over
facts(30). This leads to the overestimation of the
probability of devastating outcomes (31,32,33). The
public outrage following disasters associated with
dread risks is closely tied to a disproportionate
perception of lethality and unfairness(34,35).
A second major challenge in opinion formation
is understanding how social risk diffusion responds
to the reintroduction of balanced information, which
is often the hallmark of expert opinion (36). Various
psychological biases influence memory and social
message transmission, such as enhanced retrieval for
personal information and anecdotes, and selection
for more easily communicated information(37,38,39) ,
all of which may limit the ability of balanced
accounts to correct social risk amplification. Higher
anxiety promotes selective processing and is therefore
associated with less risk attenuation in response
to balanced information(40,41). For example, neutral
accounts heighten public panic in relation to epi-
demic outbreaks(5) and perpetuate fears whenever
fears are already disproportionately high (4). Might
the reintroduction of balanced information have a
limited effect after social risk amplification has taken
place? Moreover, might it’s efficiency be reduced
further by dread risks? Slovic(3) states that high
dread risks ’are resistant to change because they
influence the way that subsequent information is
processed’, which is in line with the previously
described affect heuristic (30). High dread topics
may therefore be potent enough that they cause
the reintroduced information to be viewed through
the tint of ’risk goggles’. This would render a
neutral account not only ineffective in terms of
inducing bias extinction but would potentially even
be counteractive, by providing the subject with a new
source of negative information(40) .
In this article, we examine the social trans-
mission of information about high and low dread
risks (nuclear energy and food additives) and further
examine how this information responds to the
reintroduction of the initial account. Our approach
involves treating message content as subject to the
evolutionary processes of birth, death, and mutation,
and tracking the change in information over repeated
transmissions in human diffusion chains. In addition,
we also focus on the sentiment of the message, asking
how dread risk and the reintroduction of the initial
information influences the dynamics of transmitted
negativity. How is social contagion of risk mediated
by these factors and what, if anything, does trying
to provide balanced information accomplish?
2. METHOD
2.1 Participants
The study was advertised in social media
groups and in public environments affiliated with
the University of Warwick and the University of
Luxembourg. A total of 154 participants (82 male
and 72 female; Mage = 23.23; SD = 3.56 and
Mage = 22.16; SD = 1.95 respectively) took part in
exchange for the opportunity to win a £20 Amazon
voucher.
2.2 Design
The study consisted of 14 chains of 8 participants
each. Participants in the first position in the chain
(N= 14) read a set of articles and then wrote a
message for the next person in the chain. Participants
in positions 2-5 (N= 4 14 = 56) read the
message from the previous position in their chain
and then wrote their own message. Positions 6-8
(N= 3 14 2 = 84) were exposed to a ’fork
design’ which was employed to determine the effect
of reintroducing information into the chain. We split
positions 6-8 into two branches, an ’informed’ and
’uninformed’ branch (Figure 1). For the ’informed’
branch, position 6 received the same information as
Social Amplification in Risk Communication 3
Fig. 1. Fork chain, split at the 6th node into an ’informed branch’ and ’uninformed branch’.
position 1, in addition to the message from position
5. For the ’uninformed’ branch, position 6 received
only the message from position 5. Dread taxonomy
consisted of two levels: the ’high dread’ condition
(N= 77; 7 chains) with participants reading and
writing about ’nuclear energy’ and the ’low dread’
condition (N= 77; 7 chains) with participants
reading and writing about ’food additives’.
All 154 messages were coded for negative and
positive statements (see Appendix A - Coding
Criteria). All messages were coded twice, once by
the experimenter who was aware of the hypothesis
and an additional time by the combined efforts of
five confederates who were blind to the rationale of
the study. Cronbach’s Alpha indicates high inter-
rater agreement for both negative (α=.961) and
positive statements (α=.974) between the two
resulting sets. Messages were also compared with
the previous message to detect new, distorted, and
omitted statements. Prior knowledge as well as risk
and dread estimates were recorded using scales from
0 to 100 (for the questions used see Appendix B).
2.3 Materials and Stimuli
The first participant in each chain read four
documents discussing the dangers and advantages of
the topic. These documents were taken from various
sources, such as the BBC, Dailymail, Telegraph, and
National Geographic. Pictures were removed from
the articles and the texts were balanced across the
two dread conditions, controlling for the amount
of words (Nuclear: 1521; Food: 1507), reading time
(piloted; N= 5; Nuclear: 8:36 min.; Food: 8:42
min.) and density of positive and negative statements
(21 statements each). The documents were presented
online on a white background with instructions and
text in font size 11.
2.4 Procedure
A survey was created for each chain position
using Qualtrics. Participants in the first session
were allocated to the first position of each of the
14 chains. This was repeated until all first chain
positions were completed. Then the first session was
closed and session 2 (containing the messages from
session 1) began. Participants who opened one of
the session links read and agreed to the informed
consent document and, following random assignment
to either nuclear energy (high dread condition) or
food additives (low dread condition), were asked
to fill out a short questionnaire evaluating their
knowledge of the topic as well as their perceived risk.
Participants starting the chain (position 1) as
well as those who were re-informed (position 6
informed) were exposed to the initial four balanced
documents. The order of the text documents was
randomized across chains. Subsequently, subjects
were asked to compose a message for the next
participant and then to judge the amount of per-
ceived risk they associated with the topic. Sessions
2-5, 6 ’uninformed’, and 7-8 were exposed only to
the message constructed by the previous participant
in their chain. Session 6 ’informed’ saw both the
previously constructed messages and the original
articles.
4 Robert D. Jagiello, Thomas T. Hills
3. RESULTS
We first establish that there were differences
in dread risk between the two conditions and
that message content changed over repeated trans-
missions. Participants had higher concerns (dread
questionnaire; see 2.2) for nuclear energy (M=
0.48; SD = 0.2) as compared to food additives
(M= 0.36; SD = 0.19), t(152) = 4.007, p <
.001, demonstrating that nuclear power was more
dreaded than food additives (Figure 2). Message
content also changed substantially over repeated
transmissions (Figure 3). Based on Moussaid et
al.(1), we computed the probability that a statement
was created (pBirth), vanished (pDeath), or was
distorted (pDistort) during transmission. For the
High Dread condition, pBirth was 45%, pDeath
was 37%, and pDistortion was 43%; for the Low
Dread condition, pBirth was 37%, pDeath was 35%,
and pDistortion was 46%. These did not differ in
relation to high or low dread (pBirth, p=.115;
pDeath, p=.807, pDistortion, p=.561). However,
upon re-information the rate for pBirth at chain
position 6 was 78% in the informed branch, compared
with 41% in the uninformed branch. The informed
condition also saw a decrease in message distortion
compared with the uninformed branch: 20% and
61%, respectively, t(26) = 3.341, p =.003. No
effect of information manipulation was found for
pDeath (p=.598). These results suggest that
socially transmitted information is under constant
flux, with the death, birth, and distortation rates
of information continuously shaping message content
over the period of our study. The findings also
indicate that re-information reduces transmission
inaccuracy.
3.1 Chain Analysis
Did messages become more negative over time?
The proportions of negative (ω
p) and positive state-
ments (ω+
p) at each chain position were calculated as
follows:
ω
p=n
p
(n
p+n+
p)ω+
p=n+
p
(n
p+n+
p)
where n
pand n+
prepresent the number of negative
and positive statements, respectively. Figure 4 shows
a heatmap illustrating the increasing prevalence
of negative content over time. Chain position was
a strong predictor of the proportion of negative
statements ω
p,F(10,143) = 5.283, p < .001.
Fig. 2. Average scores of dread questionnaire across all
chains, with error bars representing standard error. Nuclear
energy caused significantly greater concern than food addi-
tives.
Post-hoc tests reveal that ω
pwas significantly
greater in the later positions (5, 6-8 informed and
uninformed) than in the earlier positions (1-4) (p=
.038). Hence, the proportional amount of negative
statements made by subjects increased as messages
were transmitted from node to node.
3.1.1 Dread Risk
High dread was associated with the accumu-
lation of an overall higher proportion of negative
information (comparing high (74%) vs. low (65%)
dread chains, t(152) = 2.927, p =.001). A 2x11 Fac-
torial ANOVA with dread taxonomy (high/low) and
chain position (1-5; 6-8 informed; 6-8 uninformed) as
independent factors and ω
pas dependent variable
revealed a main effect for taxonomy, F(1,153) =
10.893, p =.002, and for chain position, F(10,154) =
5.506, p =.001, while the interaction was not
significant, F(10,48) = 0.616, p =.622. This effect
is also found in informed and uninformed high dread
conditions, where the proportions of negative state-
ments increased as messages were communicated
along the chain (’High Dread Informed’, F(7,48) =
8.847, p =.001, and ’High Dread Uninformed’,
F(7,48) = 8.271, p =.001). However, the low
dread conditions were not significantly affected
by chain position in either information condition
(informed, F(7,48) = 1.386, p =.233, or uninformed,
F(7,48) = 1.426, p =.217) (Figure 5).
Social Amplification in Risk Communication 5
Fig. 3. Probability that a new statement was created (pBirth), an old statement vanished (pDeath), and that a statement was
distorted (pDistortion). Positions 1 - 5 are the same in both figures, with the information manipulation taking place at the 6th
node. The comparison of A) and B) graphically illustrates that the reintroduction of information causes a marked increase in
pBirth, as well as a significant alleviation in pDistortion at the 6th position, with fluctuations in pDeath not reaching significance.
Fig. 4. Proportional negativity plotted for each node. Upper branches are ’informed’ and lower branches are ’uninformed’. Colors
indicate changes in the level of negativity.
6 Robert D. Jagiello, Thomas T. Hills
Fig. 5. Proportions of negative statements for each node and condition: High Dread (black), Low Dread (grey), Uninformed
(line), Informed (dashed).
3.1.2 Reintroduction of balanced information
How did risk information in the diffusion chains
respond to the reintroduction of balanced informa-
tion? To test this, we compared position 1 with
position 6 (informed and uninformed). A one-way
ANOVA comparing the proportions of negativity
found a significant difference, F(2,18) = 11.818, p =
.001. Further exploration showed that position 6-
informed and 6-uninformed both had higher levels of
negativity than position 1 (p=.004 and p=.0001
respectively), but failed to differ from each other
(p=.157). This suggests that the reintroduction
of balanced information was ineffective in the high
dread condition as it did not restore negativity to its
level in position 1. In the low dread condition, there
was no difference between position 1, 6-informed, and
6-uninformed, F(2,18) = 1.594, p =.230. This was
expected given the lower overall negativity within
low dread chains. In general, these findings suggest
that despite the fact that re-information yielded
mutational effects (pBirth and pDistort), it did not
influence the amount of negativity transmitted from
one person to the next.
3.2 Individual Change
What makes individuals resistant to social risk
amplification? We measured this in two ways. First
we examined the influences on (1) negative output,
which is the amount of risk-focused information
the individual passes on to the next participant,
followed by an analysis of (2) opinion change, which
reflects the propensity towards increasing one’s risk
perception.
3.2.1 Predictors and Moderators of Negative Output
A linear multiple regression analysis was used
to investigate the predictive power of (1) re-
ceived negativity (ω
p1) (2) initial risk perception
(IR) and (3) prior knowledge (knp) in regard
to the message negativity ω
pthat the individual
transmitted. Including both high and low dread
conditions and chain positions 1-8, a significant
regression equation including the aforementioned
predictors was found F(3,136) = 39.09, p < .0001
with an R2of .463. Table I summarizes these
results. Accordingly, transmitted negativity (output)
increased with elevated received negativity as well as
high initial risk assessment, while being dampened
by prior knowledge (Figure 6). In the high dread
condition, only increases in received negativity
were predictive of heightened output negativity,
F(1,68) = 30.78, p < .0001 with an R2of .312. In
the low dread condition, both received negativity and
knowledge were significant predictors, F(2,67) =
Social Amplification in Risk Communication 7
Table I . Beta coefficients and p-values of predictors of negative output, with standard error in brackets, in high and low dread
conditions as well as overall
Received Initial Prior
Constant Negativity Risk Knowledge
(β0, p) (β1, p) (β2, p) (β3, p)
High 0.373, p =.0001 0.547, p =.0001 0.154, p=.087 -0.221, p=.099
Dread (SE =.094) (SE =.097) (S E =.088) (SE =.132)
Low 0.430, p =.0001 0.541, p =.0001 0.143, p=.109 -0.480, p =.0001
Dread (SE =.092) (SE =.095) (S E =.108) (SE =.130)
Overall 0.381, p =.0001 0.565, p =.0001 0.188, p =.005 -0.362, p =.0001
(SE =.063) (SE =.065) (S E =.066) (SE =.091)
Fig. 6. Scatterplots depicting correlations between the negativity an individual passed on and (1) the negativity of the message
they received, (2) their prior knowledge regarding the topic, and (3) their initial risk perception. Transmitted negativity increases
whenever the individual has an initially hazardous view and is confronted with negative messages. It decreases, on the other hand,
in relation to elevated expertise.
30.24, p < .001 with an R2of .474. Interactions
between knowledge and received negativity were not
significant (p=.09), suggesting that the mitigating
force of knowledge is limited when it comes to the
detrimental effects of received negativity.
3.2.2 Opinion Change
Did message content change perceived risk?
Change in perceived risk, ∆R, is the difference
between pre- (IR) and post-risk (SR) assessments
for a given topic (∆R=SR IR). The mean
of the perceived risk change across all participants
was 0.13 (SD = 0.15). We observed an increase
in perceived risk in high and low dread conditions,
t(69) = 8.281, p =.001 (high dread), and t(69) =
6.269, p =.001 (low dread) with no difference
between conditions (p=.806) (Figure 7).
Change in perceived risk should be a function
of the content individuals receive. Additionally
expertise (prior knowledge) may inhibit attitude
change. To address these assumptions, a multiple
regression analysis was performed with attitude
change ∆Ras dependent variable. Again, (1) received
negativity ω
p, (2) initial risk perception (IR) and (3)
prior knowledge (knp) served as predictors. Changes
in risk perception for positions 1-8 were predicted
by the aforementioned variables across conditions
F(3,136) = 46.506, p < .0001 with an R2of .273,
as well as for high dread, F(2,67) = 26.365, p <
.0001 with an R2of.545, and low dread, F(2,67) =
23.507, p < .001 with an R2of .517, separately.
Opinion change varied positively with the amount
8 Robert D. Jagiello, Thomas T. Hills
Fig. 7. Averages of risk change in both dread conditions
(sig. different from nil).
of negative information received and negatively with
the amount of prior knowledge. Additionally, high
initial risk perceptions are linked to less severe
opinion changes, most likely reflecting a ceiling effect.
Results are summarized in Table II .
We also found an interaction between received
negativity and prior knowledge (ω
p1knp) across
both conditions (β=0.818, SE =.406, p =.046),
indicating that high received negativity is linked
to elevated changes in perceived risk especially if
prior knowledge is sufficiently low. This suggests
a resilience effect of knowledge that dampens an
individual’s susceptibility to changing their opinion.
To visualize this effect, the data were broken
down into tertiles, forming knowledge categories
(Figure 8). In the third tertile where knowledge
scores were highest (ranging from 0.39 - 0.71) the
correlation between received negativity and opinion
change was weaker (r(47) = .343, p =.021) than in
the other two tertiles (r(46) = .405, p =.004; r(47) =
.541, p =.0001).
4. DISCUSSION
Socially transmitted information is how humans
learn about potential hazards they have not yet
experienced. This provides the foundation for opinion
formation about topics that often have long-term
consequences, such as climate change, alternative
energy sources, and health risks. Experimental
research on social transmission is only beginning to
take form, but its study is critical to addressing
problems associated with public hysteria, opinion
polarization, and the reduction of public anxiety.
To our knowledge, the present work is the first to
experimentally investigate (i) the impact of dread on
social amplification of risk and (ii) to examine the
effect of re-exposure to balanced information in the
ongoing process of social diffusion. In this regard,
the present article makes five contributions. First,
we demonstrate social risk amplification in a new
domain. Moussaid et al.(1) found social amplification
of risk in relation to an antibacterial agent. In the
present work we observe a similar effect in relation to
nuclear energy, for which messages tended to become
more negative over time. Second, we demonstrate
that social amplification of risk is sensitive to dread,
with high dread topics suffering a more rapid increase
in negativity than low dread. Third, participants
increased their risk attitudes the most when reading
messages that displayed an elevated proportion of
risk information. Fourth, greater personal knowledge
of an area prior to receiving risk information did not
counter the amplification of transmitted information,
but resulted in less change in risk perception. Finally,
the reintroduction of information from the initial set
of articles was insufficient to reinstate the negativity
levels of chain position 1. Even though re-information
led to a drop in factual distortion and an increased
amount of message content, this did not reduce the
amount of transmitted negativity, which is especially
relevant in the high dread case.
The resilience of the message content to the
reintroduction of balanced information highlights
an important element of human diffusion: socially
transmitted information may evolve to become
more learnable, as it experiences selection for
content that is more easily understood and later
reproduced (39,42). Bartlett’s(43) studies of human
memory demonstrate this at the level of individuals,
by showing that people tend to recall stories in
patterns that are consistent with their own prior
expectations. By reintroducing balanced accounts
into chains, the present results demonstrate a similar
process of message adaptation across individuals and
further show that this makes the message content
more resilient than the information on which it was
initially based. Additionally, messages from peers
are likely to have more weight and hence greater
influence in regard to selective re-transmission of
information. It is therefore likely that negative peer
interpretations induced a selective bias, which led
participants at the re-informed 6th position to dis-
regard the positive side of the balanced information
in favour of a confirmation of the negatively focused
facts. Past research has demonstrated mechanisms of
Social Amplification in Risk Communication 9
Table I I . Beta coefficients and p-values of predictors of opinion change, with standard error in brackets, in high and low dread
conditions as well as overall.
Received Initial Prior
Constant Negativity Risk Knowledge
(β0, p) (β1, p) (β2, p) (β3, p)
High 0.201, p =.002 0.306, p =.0001 -0.580, p =.0001 -0.216, p =.012
Dread (SE =.094) (SE =.097) (S E =.088) (SE =.132)
Low 0.050, p=.499 0.528, p =.0001 -0.452, p =.0001 -0.180, p =.046
Dread (SE =.092) (SE =.095) (S E =.108) (SE =.130)
Overall 0.109, p =.021 0.450, p =.0001 -0.505, p =.0001 -0.203, p =.001
(SE =.063) (SE =.065) (S E =.066) (SE =.091)
Fig. 8. In order to visualize the impact of knowledge on the relationship between received negativity and opinion change,
participants’ knowledge scores were categorized into thirds of the subject pool (N = 47). As can be seen, the relation between
received negativity and opinion change is strongest in the first tertile and weakest in the third (where knowledge scores are highest),
hence suggesting a mitigation effect of expertise.
social proof and group-think that show the capacity
of inducing such biases(44,45) .
Our results further indicate that higher levels
of prior knowledge were associated with lower
transmission negativity. Further, individual’s with
higher content relevant knowledge were more resilient
to the detrimental effects that receiving a high risk
message had on one’s risk perception. This is in line
with past research that has shown that the inte-
gration of balanced information may be facilitated
through a process of inoculation, whereby individuals
are prepared for biased information through pre-
10 Robert D. Jagiello, Thomas T. Hills
exposure to more balanced arguments(46,47,48,49).
Hence the findings suggest that inoculation may also
work in social diffusion chains by demonstrating that
individuals with higher knowledge are less prone to
increase their perceived risk in response to propa-
gation of negative content. In regard to the actual
content amplification in transmitted messages, no
interaction could be found and therefore expertise
may not offer the same level of resilience as it does
for attitudes. Future research should therefore aim
to address how certain factors may affect opinions
differently than their written manifestations and
how this discrepancy influences the diffusion of risk
information.
Limitations of the present design can further
serve to pave the way for future directions of inquiry.
Firstly, in regard to the questionnaire that was used,
the concept of psychological distance and the role
it plays in risk perception need to be considered.
Risk perceptions are generally higher in response
to threats that are in closer proximity as opposed
to geographically distant hazard sources(50) . The
questions that were used to measure risk perceptions
of nuclear energy (’Imagine now that near to where
you live a new nuclear power plant is in construction
[...]’) could have been interpreted by participants as
a more proximate threat than in the food additives
condition (’Imagine reading an article about how
the general amount of food additives in the country
you live in has been increased [...]’). On the other
hand, it could be argued that food is interpreted
as more proximate due to its ubiquitous day-to-day
presence. Both topics are inherently different—an
issue that is unavoidable given that phenomena
that lie on different ends of Slovic’s dread axis
differ along numerous dimensions. Furthermore, it
needs to be noted that the topics used in the
present design are well known, widely discussed
phenomena that are likely to have been subject to
social diffusion prior to this experimental induction.
Future replication attempts of the observed effects
may include artificially made-up risks that respect
the dread characteristics of Slovic’s taxonomy, in
order to limit the magnitude of effects that may
have already occurred outside of the lab. This could
provide a powerful approach to understanding the
factors that lead benign risk topics to incite public
hysteria.
To conclude, following the work of Moussaid
et al.(1), we demonstrate that processes of opinion
formation and social amplification of risk can be
investigated in an experimental setting that tracks
sentiment and content over time. The present work
extends this investigation by further demonstrating
the ability to elucidate the role of risk types and
the influence of interventions that attempt to restore
balanced attitudes. Because dread risk is unique in
its ability to pose socially contagious public health
problems(3,30), documenting its enhanced sensitivity
to social amplification of risk is a first step towards
reducing hazardous responses in the future.
REFERENCES
1. Moussad M, Brighton H, Gaissmaier W. The amplification
of risk in experimental diffusion chains. Proc Natl Acad
Sci. 2015 May 5;112(18):56316.
2. Myers DG, Lamm H. The group polarization phe-
nomenon. Psychol Bull. 1976;83(4):60227.
3. Slovic P. Perception of risk. Science. 1987 Apr
17;236(4799):2805.
4. Slovic P, Weber EU. Perception of risk posed by
extreme events. Risk Manag Strateg an Uncertain World.
2002;121.
5. Mosquera M, Melendez V, Latasa P. Handling Europes
first Ebola case: Internal hospital communication experi-
ence. Am J Infect Control. 2015;43(4):3689.
6. Luckerson V. Fear, information, and social media com-
plicate Ebola fight. Time Magazine [Internet]. 2014;
Available from: http://time.com/3479254/ebola-social-
media/
7. Hertwig R, Pachur T, Kurzenhuser S. Judgments of risk
frequencies: tests of possible cognitive mechanisms. J Exp
Psychol Learn Mem Cogn. 2005;31(4):62142.
8. Pachur T, Hertwig R, Steinmann F. How do people judge
risks: Availability heuristic, affect heuristic, or both? J
Exp Psychol Appl. 2012;18(3):31430.
9. Trumbo CW, McComas KA. The function of credibility
in information processing for risk perception. Risk Anal.
2003;23(2):34353.
10. Salganik MJ, Watts DJ. Leading the Herd Astray: An
Experimental Study of Self-fulfilling Prophecies in an
Artificial Cultural Market. Soc Psychol Q. 2008 Dec
1;71(4):33855.
11. Myers DG, Kaplan MF. Group-induced polariza-
tion in simulated juries. Personal Soc Psychol Bull.
1976;2(1):636.
12. Lorenz J, Rauhut H, Schweitzer F, Helbing D. How social
influence can undermine the wisdom of crowd effect. Proc
Natl Acad Sci. 2011 May 31;108(22):90205.
13. Christakis N a, Fowler JH. The spread of obesity in
a large social network over 32 years. N Engl J Med.
2007;357(4):3709.
14. Christakis NA, Fowler JH. The Collective Dynamics of
Smoking in a Large Social Network. N Engl J Med. 2008
May 22;358(21):224958.
15. Asch SE. Studies of independence and conformity: I. A
minority of one against a unanimous majority. Psychol
Monogr Gen Appl. 1956;70(9):170.
16. Darley JM, Latane B. Bystander intervention in emer-
gencies: Diffusion of responsibility. J Pers Soc Psychol.
1968;8(4, Pt.1):37783.
17. Latane B, Darley JM. Group inhibition of bystander
Social Amplification in Risk Communication 11
intervention in emergencies. J Pers Soc Psychol.
1968;10(3):21521.
18. Latane B, Nida S. Ten years of research on group size and
helping. Psychol Bull. 1981;89(2):30824.
19. Kasperson RE, Renn O, Slovic P, Brown HS, Emel J,
Goble R, et al. The Social Amplification of Risk: A
Conceptual Framework. Risk Anal. 1988 Jun;8(2):17787.
20. Renn O, Burns WJ, Kasperson JX, Kasperson RE,
Slovic P. The Social Amplification of Risk: Theoretical
Foundations and Empirical Applications. J Soc Issues.
1992;48(4):13760.
21. Bostrom A. The Social Amplification of Risk. Pidgeon N,
Kasperson RE, Slovic P, editors. Vol. 8, Journal of Risk
Research. Cambridge: Cambridge University Press; 2003.
277-280 p.
22. Pidgeon N, Henwood K. The Social Amplification of
Risk Framework (SARF): Theory, Critiques, and Policy
Implications. In: Risk Communication and Public Health.
Oxford University Press; 2010. p. 5368.
23. Slovic P. Perception of Risk: Reflections on the Psychome-
tric Paradigm. In: Social Theories of Risk. 1992. p. 117152.
24. Slovic P, Fischhoff B, Lichtenstein S. Rating the Risks.
Vol. 21, Environment: Science and Policy for Sustainable
Development. 1979. p. 1439.
25. Fischhoff B, Slovic P, Lichtenstein S, Read S, Combs
B. How safe is safe enough? A psychometric study of
attitudes towards technological risks and benefits. Policy
Sci. 1978;9(2):12752.
26. Slovic P, Peters E. Risk Perception and Affect. Curr Dir
Psychol Sci. 2006 Dec;15(6):3225.
27. Loewenstein GF, Weber EU, Hsee CK, Welch N. Risk as
feelings. Psychol Bull. 2001;127(2):26786.
28. Tversky A, Kahneman D. Judgment under Uncer-
tainty: Heuristics and Biases. Science (80- ). 1974 Sep
27;185(4157):112431.
29. Petty R., Cacioppo J. The elaboration likelihood model
of persuasion. Adv Exp Soc Psychol. 1986;19:123205.
30. Slovic P, Finucane ML, Peters E, MacGregor DG. Risk as
analysis and risk as feelings. Risk Anal. 2004;24(2):31122.
31. Slovic P, Monahan J, MacGregor DG. Violence risk
assessment and risk communication: The effects of
using actual cases, providing instruction, and employing
probability versus frequency formats. Law Hum Behav.
2000;24(3):27196.
32. Denes-Raj V, Epstein S. Conflict Between Intuitive and
Rational Processing: When People Behave Against Their
Better Judgment. J Pers Soc Psychol. 1994;66(5):81929.
33. Yamagishi K. When a 12.86 % Mortality is More Danger-
ous than 24.14 %: Implications for Risk Communication.
Appl Cogn Psychol. 1997;11:495506.
34. Lachlan K, Spence PR. Communicating Risks: Examining
Hazard and Outrage in Multiple Contexts. Risk Anal.
2010;30(12):187286.
35. Sandman PM, Miller PM, Johnson BB, Weinstein
ND. Agency communication, community outrage, and
perception of risk: Three simulation experiments. Risk
Anal. 1993;13(6):58598.
36. Tetlock PE. Experts All the Way Down. Natl Interes.
2010;(110):7686.
37. Mickes L, Darby RS, Hwe V, Bajic D, Warker JA, Harris
CR, et al. Major memory for microblogs. Mem Cognit.
2013 May 12;41(4):4819.
38. Kirby S. Spontaneous evolution of linguistic structure-
an iterated learning model of the emergence of regu-
larity and irregularity. IEEE Trans Evol Comput. 2001
Apr;5(2):10210.
39. Kirby S, Cornish H, Smith K. Cumulative cultural
evolution in the laboratory: An experimental approach
to the origins of structure in human language. Proc Natl
Acad Sci. 2008 Aug 5;105(31):106816.
40. Mathews A, Mackintosh B. A Cognitive Model of Selective
Processing in Anxiety. Cognit Ther Res. 1998;22(6):53960.
41. Mathews A. Why worry? The cognitive function of
anxiety. Behav Res Ther. 1990;28(6):45568.
42. Hills TT, Adelman JS. Recent evolution of learnability
in American English from 1800 to 2000. Cognition. 2015
Oct;143:8792.
43. Bartlett FC. Remembering: A Study in Experimental and
Social Psychology. Cambridge, Soc Psychol. 1932;111.
44. Kelman, H. C. (1958). Compliance, identification, and in-
ternalization: Three processes of attitude change. Journal
of Conflict Resolution, 2, 51-60.
45. Cialdini, R. B., Wosinska, W., Barrett, D. W., Butner,
J., & Gornik-Durose, M. (1999). Compliance with a
request in two cultures: The differential influence of social
proof an commitment/consistency on collectivists and
individualists. Personality and Social Psychology Bulletin,
25, 1242-1253
46. Papageorgis D, McGuire WJ. The Generality of Im-
munity to Persuasion Produced by Pre-Exposure to
Weakened Counterarguments. J Abnorm Soc Psychol.
1961;62(1):47581.
47. Compton J, Jackson B, Dimmock JA. Persuading Others
to Avoid Persuasion: Inoculation Theory and Resistant
Health Attitudes. Front Psychol. 2016 Feb 9;7(Febru-
ary):19.
48. Compton J, Pfau M. Spreading Inoculation: Inoculation,
Resistance to Influence, and Word-of-Mouth Communica-
tion. Commun Theory. 2009;19(1):928.
49. McGuire WJ. The Effectiveness of Supportive and Refu-
tational Defenses in Immunizing and Restoring Beliefs
Against Persuasion. Sociometry. 1961 Jun;24(2):184.
50. Trope, Y., & Liberman, N. (2010). Construal-Level
Theory of Psychological Distance. Psychological Review,
117(2), 440463.
5. APPENDIX A - CODING CRITERIA
The coding criteria are adapted from Moussaid
et al.(1). Criteria for negative statements in nuclear
energy chains are the following:
1. Any mention of a negative aspect of nuclear
energy, such as
a. detrimental consequences of its use
b. anecdotal evidence of danger
c. indicators of risk avoidance
Examples:
a. ’The production of nuclear energy causes ra-
dioactive waste, harming the environment.’
b. ’Remember Chernobyl?’
c. ’Germany is decreasing their nuclear energy
production over the next years.’
2. Any personal judgment that is suggestive of
the dangers of nuclear energy, such as ’I feel
like a world without nuclear power would
definitely be a bit safer!’
12 Robert D. Jagiello, Thomas T. Hills
The criteria for positive statements are the
following:
1. Any mention of
a. positive aspects of nuclear energy
b. any form of risk mitigation
c. any type of anecdotal endorsement
Examples:
a. ’Nuclear power plants drive the economy.’
b. ’Disposal of radioactive waste is regulated
by the government.’
c. ’My cousin works at a power plant. It’s
super safe.’
2. Personal judgments such as ’I don’t think
power plants are that dangerous.’
The criteria for food additives have been established
in an analogous manner.
6. APPENDIX B - QUESTIONNAIRE
The questions that were used in the course
of participation in this study can be seen below,
(1) nuclear energy and (2) food additives. All
questions were answered via the use of a slider (3).
The question regarding the general risk perception
(1.3. and 2.3.) was administered twice, once before
message transmission and once after.
1.1 How familiar are you with nuclear energy, its
production, risks, and benefits?
0% No knowledge
1% - 20% Minimal amount of knowledge (I
have read 1 article/seen 1 news report)
20% - 40% Some knowledge (I followed the
news coverage)
40% - 60% Moderate knowledge (I read
more than 5 articles/1 book)
– 60% - 100% Advanced knowledge (I study
physics or another science related to the
topic and I am regularly following up on the
newest findings)
1.2. Imagine now that near to where you live a new
nuclear power plant is in construction. How
concerned would you be about that?
1.3. Please indicate how dangerous you perceive
nuclear energy production to be.
2.1. How familiar are you with food additives, their
risks, and benefits?
0% No knowledge
1% - 20% Minimal amount of knowledge (I
have read 1 article/seen 1 news report)
20% - 40% Some knowledge (I followed the
news coverage)
40% - 60% Moderate knowledge (I read
more than 5 articles/1 book)
– 60% - 100% Advanced knowledge (I study
nutrition or another science related to the
topic and I am regularly following up on the
newest findings)
2.2. Imagine reading an article about how the
general amount of food additives in the
country you live in has been increased. How
concerned would you be about that?
2.3. Please indicate how dangerous you perceive
food additives to be.
3. Slider
Social Amplification in Risk Communication 13
7. APPENDIX C - MATERIALS
1. Nuclear Energy
Doc 1. Nuclear energy is the energy in the nucleus, or core, of an atom. Energy is what holds the nucleus
together. There is a huge amount of power in an atoms dense nucleus. In fact, the power that holds the nucleus
together is officially called the ’strong force’. Nuclear energy can be used to create electricity, but it must first
be released from the atom. In nuclear fission, atoms are split to release the energy. A nuclear reactor, or power
plant, consists of a series of machines that can control nuclear fission to produce electricity. The fuel that
nuclear reactors use to produce nuclear fission comes from pellets of the element uranium. Uranium is the fuel
most widely used because its atoms split apart relatively easily. In a nuclear reactor, atoms of uranium are
forced to break apart. As they split, the atoms release tiny particles called fission products. Fission products
cause other uranium atoms to split, starting a chain reaction. The energy released from this chain reaction
creates heat.
Nuclear energy produces electricity that can be used to power homes, schools, businesses, and hospitals.
Power plants produce renewable, clean energy, as they do not pollute the air or produce greenhouse gases. They
can be built in urban or rural areas, and do not radically alter the environment around them. The steam which
is powering the turbines and generators is ultimately recycled: It is cooled down in a separate structure called
a cooling tower. The steam turns back into water and can be used again to produce more electricity. Excess
steam is simply recycled into the atmosphere, where it does no harm as clean water vapor. About 15 percent
of the worlds electricity is generated by nuclear power plants. The United States has more than 100 reactors,
although it creates most of its electricity from fossil fuels and hydroelectric energy. Nations such as Lithuania,
France, and Slovakia create almost all of their electricity from nuclear power plants. However, the byproduct
of nuclear energy is radioactive material, which is a collection of unstable atomic nuclei. These nuclei lose their
energy and can affect many materials around them, including organisms and the environment, causing burns
and increasing the risk for cancers, blood diseases, and bone decay. Radioactive waste is what is left over from
the operation of a nuclear reactor, mostly protective clothing worn by workers, tools, and cloths that have been
in contact with radioactive dust. Radioactive waste is long-lasting and tools can stay radioactive for thousands
of years. The government regulates how these materials are disposed of so they don’t contaminate anything
else.
Used fuel and rods must be stored in special containers that look like large swimming pools. Water cools
the fuel and insulates the outside from contact with the radioactivity. Some nuclear plants store their used fuel
in dry storage tanks above ground. Critics of nuclear energy worry that the storage facilities for radioactive
waste will leak, crack, or erode. Radioactive material could then contaminate the soil and groundwater near
the facility. This could lead to serious health problems for the people and organisms in the area. This is what
happened in Chernobyl, Ukraine, in 1986. A steam explosion at one of the power plant’s four nuclear reactors
caused a fire, called a plume. This plume was highly radioactive, creating a cloud of radioactive particles that
fell to the ground, called fallout. The fallout spread over the Chernobyl facility, as well as the surrounding area.
The fallout drifted with the wind, and the particles entered the water cycle as rain. The environmental impact
of the Chernobyl disaster was immediate. For kilometers around the facility, the pine forest dried up and died.
The red color of the dead pines earned this area the nickname the Red Forest. Fish from the nearby Pripyat
River had so much radioactivity that people could no longer eat them.
More than 100,000 people were relocated after the disaster, but the number of human victims of Chernobyl
is difficult to determine. The effects of radiation poisoning only appear after many years. Cancers and other
diseases can be very difficult to trace to a single source.
Doc 2.Nuclear energy presents itself with economic benefits. Each year, the average nuclear plant generates
approximately $470 million in economic output or value, which includes more than $40 million in total labor
income. These figures include both direct output and secondary effects. The direct output reflects the plants
annual electricity salesapproximately $453 million. The secondary effects at the local levelapproximately $17
millioninclude subsequent spending by firms attributable to the presence of the plant and its employees as plant
expenditures filter through the local economy. There are also secondary effects outside the local area, at the
state and national level. For a nominal 1,000-megawatt nuclear plant, these secondary effects are $80 million
14 Robert D. Jagiello, Thomas T. Hills
and $393 million, respectively. Analysis shows that every dollar spent by the average nuclear plant results in
the creation of $1.04 in the local community, $1.18 in the state economy and $1.87 in the U.S. economy. The
average nuclear plant pays about $16 million in state and local taxes annually. These tax dollars benefit schools,
roads, and other state and local infrastructure. The average nuclear plant also pays federal taxes of $67 million
annually.
Doc 3. The Sixth Report of Fukushima Prefecture Health Management Survey, released in April, included
examinations of 38,114 children, of whom 35.3 percent - some 13,460 children - were found to have cysts or
nodules of up to 5 mm (0.197 inches) on their thyroids. ”Yes, 35.8 percent of children in the study have lumps
or cysts, but this is not the same as cancer,” said Naomi Takagi, an associate professor at Fukushima University
Medical School Hospital, which administered the tests. ”We do not know that cause of this, but it is hard to
believe that is due to the effects of radiation,” she said. ”This is an early test and we will only see the effects
of radiation exposure after four or five years.” The local authority is carrying out long-term testing of children
who were under the age of 18 on March 11 last year, the day on which the magnitude-9 Great East Japan
struck off the coast of north-east Japan, triggering the massive tsunami that crippled the Fukushima nuclear
plant. Thyroid examinations were first conducted in October last year and will be carried out every two years
up to the age of 20 and every five years for the rest of the children’s lives. A second report has been issued by
Japan’s Institute of Radiological Sciences in which it found that some children living close to the plant were
exposed to ”lifetime” doses of radiation to their thyroid glands.
Doc 4. The World Health Organization (WHO) released a report that estimates an increase in risk for
specific cancers for certain subsets of the population inside the Fukushima Prefecture. A 2013 WHO report
predicts that for populations living in the most affected areas there is a 70% higher risk of developing thyroid
cancer for girls exposed as infants (the risk has risen from a lifetime risk of 0.75% to 1.25%), a 7% higher risk
of leukemia in males exposed as infants, a 6% higher risk of breast cancer in females exposed as infants and a
4% higher risk, overall, of developing solid cancers for females. Preliminary dose-estimation reports by WHO
and the United Nations Scientific Committee on the Effects of Atomic Radiation (UNSCEAR) indicate that,
outside the geographical areas most affected by radiation, even in locations within Fukushima prefecture, the
predicted risks remain low and no observable increases in cancer above natural variation in baseline rates are
anticipated. In comparison, after the Chernobyl accident, only 0.1% of the 110,000 cleanup workers surveyed
have so far developed leukemia, although not all cases resulted from the accident. Estimated effective doses
from the accident outside of Japan are considered to be below (or far below) the dose levels regarded as very
small by the international radiological protection community. The United Nations Scientific Committee on the
Effects of Atomic Radiation is expected to release a final report on the effects of radiation exposure from the
accident by the end of 2013. A June 2012 Stanford University study estimated, using a linear no-threshold
model, that the radioactivity release from the Fukushima Daiichi nuclear plant could cause 130 deaths from
cancer globally (the lower bound for the estimate being 15 and the upper bound 1100) and 199 cancer cases in
total (the lower bound being 24 and the upper bound 1800), most of which are estimated to occur in Japan.
Radiation exposure to workers at the plant was projected to result in 2 to 12 deaths. However, a December
2012 UNSCEAR statement to the Fukushima Ministerial Conference on Nuclear Safety advised that because
of the great uncertainties in risk estimates at very low doses, UNSCEAR does not recommend multiplying
very low doses by large numbers of individuals to estimate numbers of radiation-induced health effects within
a population exposed to incremental doses at levels equivalent to or lower than natural background levels.”
2. Food Additives
Doc 1. Many substances are added to foods to prolong shelf and storage life and to enhance color,
flavor, and texture. The possible role of food additives in cancer risk is an area of great public interest.
New food additives must be cleared by the US Food and Drug Administration (FDA) before being allowed
into the food supply, and thorough testing is done in lab animals to determine any effects on cancer as
part of this process. Additives are usually present in very small quantities in food, and some are nutrients
that may have beneficial effects (for example, vitamins C and E are sometimes added to food products as a
Social Amplification in Risk Communication 15
preservative). Other compounds find their way into the food supply through agricultural use, animal farming,
or food processing, even if their use is not directly intended for human consumption. Examples include growth
hormones or antibiotics used in animal farming, small amounts of pesticides and herbicides in plant-based
foods, and compounds such as bisphenol A (BPA) or phthalates that enter food from packaging. Some of these
compounds are not known to directly cause cancer, but they may influence cancer risk in other ways for
example, by acting as hormone-like substances in the body. Unintended contamination of food may also result
in exposure to chemicals that are a cause of concern and may be related to cancer risk. Examples include heavy
metals such as cadmium or mercury. These metals may enter the food supply if they build up the food chain,
such as from fish, or they may enter through contamination or their natural presence in soil or water. For many
other compounds for which the effects on cancer risk are not clear, there may be other good reasons to limit
exposure. But at the levels that these are found in the food supply, lowering cancer risk is unlikely to be a major
reason to justify this. Food processing may also alter foods in ways that might affect cancer risk. An example
is the refining of grains, which greatly lowers the amount of fiber and other compounds that may reduce cancer
risk. The processing of meat, by adding preservatives such as salt or sodium nitrite to prevent the growth of
germs, or smoking the meat to preserve or enhance color and flavor, may add compounds that might increase
the potential of these foods to cause cancer. Studies have linked eating large amounts of processed meats with
an increased risk of colorectal cancer. Some food processing, such as freezing and canning vegetables and fruits,
can preserve vitamins and other components that may decrease cancer risk. Cooking or heat-treating (such
as when canning) vegetables breaks down the plant cell walls and may allow the helpful compounds in these
foods to be more easily digested.
Doc 2.Well, let’s start with a short explanation of what E numbers are. E stands for Europe, and the E
number code relates to a set of EU rules about which foods can contain them and how much you should be
able to consume in a day. For instance E284 boric acid can only be used in caviar, and E252 potassium nitrate
(used in bacon and salami) has an acceptable level of daily intake (ADI) of 0-3.7% mg/Kg body weight. Many
E numbers are very familiar and important to good food and nutrition: for instance E300 is vitamin C, E101
is vitamin B2, E948 is oxygen and E160c is paprika.
The rules were developed to regulate additives (rather than encourage their use), so that dangerous
substances like toxic lead tetroxide could be banned from use in children’s sweets, for instance. In the past,
food adulteration was a deadly problem.
But what about the bad E numbers? E621 monosodium glutamate is anecdotally blamed for an
extraordinary range of symptoms, but in fact if you grate parmesan on your pasta you are likely to be adding
more glutamate to your meal than you’d ever find in an MSG-laden ready meal. There’s a group of food
colours called the ’Southampton Six’ that have a small but proven association with hyperactivity in children,
and which you might want to avoid. Sulphur dioxide (E220) can exacerbate asthma, although without it wine
usually tastes foul and in any case it’s been used in pretty much every bottle of wine produced since Roman
times.
But the leading causes of food allergies and intolerances are entirely natural: milk, wheat, eggs, nuts, fish,
soya, celery... And of course every single food or drink on the planet, whether it contains E numbers or not,
is toxic at some level - apples contain cyanide, people have died from water intoxication, cabbage contains
goitrogens, potatoes contain toxic solanine and broccoli contains carcinogens. But, as with E numbers, the
amounts of these toxic substances are minute, and the benefits of consuming these foods and drinks invariably
far outweigh the risks. The difference with E numbers is that they have been extensively tested and analysed
to ascertain safe levels.
The reality is that all foods are a combination of chemicals, whether added by man or not, and just
because a food is organic doesn’t necessarily make it better for you. The worst nutritional problems are caused
by substances that come in purely organic form: salt, fat and sugar, none of which are E numbers.
The argument in favour of Es is that they make food healthier, safer, cheaper, better tasting and more
attractive. Of course, many horrible and unhealthy foods also contain E numbers, but invariably it’s not the
Es that make them unhealthy - it’s the salt, fat and sugar.
Doc 3.Artificial food additives and preservatives could be causing children to be disruptive, a growing
16 Robert D. Jagiello, Thomas T. Hills
bank of evidence is proving. These chemicals - or E-numbers as they are known - are added to enhance the
flavour and colour of food, and to prolong its shelf life. Gordon Walker, the headmaster of a primary school in
Cornwall, noticed how his own son’s behaviour improved after he stopped eating food containing E-numbers.
As a result, he conducted an experiment at his school to see how other children were affected by the additives.
His concerns are backed up by scientific evidence. Dr Neil Ward, a senior lecturer in analytical chemistry at
the University of Surrey, has carried out four independent studies evaluating the impact of food additives on
hyperactive children, in particular the colourings E102, E110, E123. ’All of our studies have confirmed that
additives do have a detrimental effect on the behaviour of hyperactive children,’ says Dr Ward. ’We have also
found that a lot of so called ”ordinary children are very sensitive to additives and artificial chemical in their
diet, so it’s a very widespread problem. ’And we have discovered links between additives and an increased
incidence of eczema, asthma and allergies in selected groups of children who consume high levels of additives
and artificial chemicals in their diet.’ Food preservatives have been used by mankind for centuries. Salt, sugar
and vinegar, for example, were among the first, and were used to preserve foods for longer periods. However
in the past 30 years, with the advent of processed foods, there has been a massive explosion in the chemical
corruption of foods using additives to completely change natural flavours and colours to make them last longer.
Under today’s law on European food standards, every additive or preservative put into food must be identified
and given a number, which is its E-number. All E-numbers present in UK food and drinks are regulated by the
Food Standards Agency. There are hundreds of registered E-numbers, but as new ones are added, discontinued
or even banned in some countries every week, it is impossible to put an exact figure on how many there are at
any one time. However, it is estimated there are 270 in use, numbered between E100 and E1520.
Doc 4.E621 monosodium glutamate, otherwise known as MSG, Monosodium Glutamate, E621 is a flavour
enhancer thats commonly used to pep up food products and make them taste better. Unfortunately, it is known
to cause problems for some people and certain people seem to more sensitive to its effects than others. Amongst
the known side effects, MSG can cause symptoms such as headaches, nausea, dizziness, muscle pain, palpitations
and even pain.
Aspartame, E951 is an artificial sweetener thats commonly used as a sweetening ingredient. In particular,
its often found in products aimed at dieters or diabetics, such as desserts, low-fat foods, low sugar drinks,
snacks and sweets. Its well known to be linked to problems in people who suffer from the condition PKU,
and they are well advised to avoid it completely. But aspartame has become a concern to other people too
and side effects, such as headaches, have often been reported. E211, sodium benzoate, is an E number thats
used as a preservative and is found in products such as margarine, salad dressing, soy sauce, sweets and soft
drinks. Studies have found that its linked to hyperactivity in children, plus it may cause reactions in people
have allergic conditions or asthma.
... Studies employing the "Social Amplification of Risk Framework" (SARF) have notably drawn attention to the way the transmission and circulation of risk information is intensified and dampened by the media among other key actors within society (Kasperson et al., 1988;Kasperson & Kasperson, 1996). Researchers have also extended these considerations to include more fine-grained analyses of the role of key variables such as language, culture, emotion, and politics in the amplification and attenuation of risk before, during, and after major events (Chung & Yun, 2013;Duckett & Busby, 2013;Flynn, 2003;Glik, 2007;Jagiello & Hills, 2018;Masuda & Garvin, 2006;Wardman & Löfstedt, 2018). ...
... As people do not typically encounter exposure to many risks immediately or directly, news outlets have traditionally acted as a primary "amplification station" by way of their central role in gathering, filtering, and circulating risk signals through public channels (Binder et al., 2014;Kasperson et al., 1988;Rossmann et al., 2018;Vasterman, 2005;Vasterman et al., 2005). The importance of the media is underscored by findings showing that a more rapid increase in media messaging is associated with "high dread" risks, which are known to arouse substantial fear and anxiety (Jagiello & Hills, 2018). Research also shows that people's media use is associated with the perceived severity of health hazards (Berger & Milkman, 2012), and that those risk stories which highly resonate with the concerns of news consumers are also more likely to be shared with others, especially if they contain elements which create strong emotional arousal, such as anger, anxiety, and fear (Frewer et al., 2002;Young et al., 2008). ...
... We suggest that the marked differences observed in our study regarding these two cases could be related to the magnitude of ripple effects associated with the compound nature of the risks in question. In this way, both crises presented editors and journalists, the "gatekeepers" of news, with an enduring newsworthy story that pervaded across business, social, policy, and entertainment arenas (Binder et al., 2014;Frewer et al., 2002;Jagiello & Hills, 2018;McInerney et al., 2004;Rossmann et al., 2018;Wong & Yang, 2022;Young et al., 2008). This is clearly displayed in Figure 4, for example, where the issue of COVID-19 dominates all other news in 2020. ...
Article
Full-text available
This paper addresses the question of identifying and distinguishing risk amplification incidents and patterns in the news media. To meet this objective, our study incorporates a novel “floodlight” approach utilizing the Society for Risk Analysis Glossary in conjunction with topic modeling and time‐series analysis, to investigate risk‐focused stories within a corpus of 271,854 US news articles over the past two decades. We find that risk amplification in the US news media is concentrated around seven core risk news categories—business, domestic affairs, entertainment, environment, geopolitics, health, and technology—which also vary in the risk‐related terms that they predominantly employ. We also identify 14 signal events that can be distinguished relative to general risk news within their categories. Across these events, the “War on Terror” and COVID‐19 are seen to display uniquely dynamic media reporting patterns, including a systemic influence between risk news categories and the attenuation of other risk news. We discuss possible explanations for these findings along with their wider research and policy implications.
... Public health system reputation largely hinges on their ability to provide timely and accurate information during crises [51]. To prevent vaccine-related panics and ensure public trust, to prevent vaccine-related panics and ensure public trust, healthcare systems must proactively meet the public's emotional and informational needs by offering reliable, transparent information before negative narratives take hold [52]. This approach is essential for maintaining public satisfaction and encouraging proactive protective behaviors. ...
Article
Full-text available
Objectives During the COVID-19 pandemic, online health information search has been shown to influence the public’s health beliefs, risk attitudes, and vaccination behavior. This study constructs a conditional process model to explore how online health information search impacts public vaccination behavior, considering critical factors like healthcare system satisfaction, vaccine risk perception, and the perceived usefulness of information. Methods Data from the 2021 Chinese General Social Survey (N = 2,547) were analysed. The study utilized logistic regression, path analysis, and the Bootstrap method to test the conditional process model. Results Increased online health information search promotes vaccination behavior, while increased vaccine risk perception hinders vaccination behavior. Higher satisfaction with the healthcare system encourages vaccination behavior, but online health information search reduces healthcare system satisfaction. Satisfaction with the healthcare system and vaccine risk perception play a chain mediating role between online health information search and vaccination behavior. Additionally, the perceived usefulness of information has a negative moderating effect on online health information search and healthcare system satisfaction. Conclusion The research findings provide new insights for health information dissemination and vaccination interventions.
... Of particular note is the fact that both risk tolerance and risk intensity predicted likelihood of risk communication. Given that in the absence of in-depth knowledge or direct experience, we are reliant on mediated information from others 24,54 , this has the potential to lead to further risk dissemination and thus potentially risk amplification 55 . Estimates were derived from the final models: communication -personal/impersonal, responsibility -user/external ~ ((product + place of purchase + harm cause + harm experience)^2) + (dread + benefits) + (age + gender + children + risk propensity + grid worldview + group worldview) * (place of purchase + harm cause + harm experience)). ...
Article
Full-text available
We vary greatly in our perception of risk, not just because of differences between risks themselves, but also because of individual, contextual and cultural differences too. To better understand and predict responses to risk, we need to (a) integrate these components, combining approaches from different psychological disciplines and (b) also consider risk tolerance – how individuals trade-off between risks and benefits. We therefore developed an ICONS (individual, contextual, cognitive, social) framework; using it across two empirical studies (n = 4228) to examine how individuals perceive and respond to the quotidian risks associated with consumer products. Three dimensions underlined risk perceptions: benefits, dread and individual responsibility. Risk tolerance was typically predicted by interactions between individual (demographic, cultural worldview, personality) and contextual (product type/category, harm information) factors. In turn, perceived dread, benefits and individual differences shaped how likely participants were to communicate risk information. Our results demonstrate for the first time how the interaction between individual, cognitive (risk tolerance, intensity), contextual, and social (risk communication) factors is key to understanding and predicting risk perceptions. Together, our findings help explain why societal responses to risks are often difficult to predict and have implications for the spread, and amplification, of risk information.
Article
Background: Perceptions of “dread risk,” information source beliefs, and trust in these sources are important variables towards understanding vaccine attitudes. The main objective of this research was to learn the associations among these phenomena and, particularly, how these variables may impact our understanding of vaccine attitudes. Methods: In the summer of 2020, a total of 336 (Mage = 53.55; SDage = 18.58) participants answered an online Qualtrics survey about their “dread risk” perceptions of vaccines, their information source beliefs (e.g., credible, easy-to-access, etc.), trust in information sources, and vaccine attitudes. Results: Results suggest participants had generally low levels of negative vaccine attitudes (M = 3.27; SD = 1.40). Regression analyses to predict negative vaccine attitudes found “dread risk” perceptions to be the most significant predictor (beta = .61, p < .001), over and above information source beliefs and trust in information sources (F (9, 320) = 75.07, p < .001; adjusted R2 =.67). Age was also significantly and negatively correlated with “dread risk” (r = -.22, p <.001). Conclusions: Results suggest that vaccine communication efforts should continue to identify individuals’ “dread risk” perceptions about vaccines and emphasize this content in messaging. Future research should aim to understand more about the relationships between negative vaccine attitudes, “dread risk” perceptions, information source beliefs, and trust in information sources to better predict vaccine-related decisions, (in)actions, and to support effective vaccine communication.
Article
Full-text available
When offered an opportunity to win 1oneverywintrialinwhichtheydrewaredjellybean,subjectsfrequentlyelectedtodrawfromabowlthatcontainedagreaterabsolutenumber,butasmallerproportion,ofredbeans(e.g.,7in100)thanfromabowlwithfewerredbeansbutbetterodds(e.g.,1in10).Subjectsreportedthatalthoughtheyknewtheprobabilitieswereagainstthem,theyfelttheyhadabetterchancewhenthereweremoreredbeans.Similar,butlessextremeresultswereobtainedonlosetrials,wheredrawingaredbeanmeantlosing1 on every “win” trial in which they drew a red jelly bean, subjects frequently elected to draw from a bowl that contained a greater absolute number, but a smaller proportion, of red beans (e.g., 7 in 100) than from a bowl with fewer red beans but better odds (e.g., 1 in 10). Subjects reported that although they knew the probabilities were against them, they felt they had a better chance when there were more red beans. Similar, but less extreme results were obtained on “lose” trials, where drawing a red bean meant losing 1. These results were predicted from the concretive and experiential principles of cognitive-experiential self-theory. Nonoptimal choices in the laboratory were significantly correlated with heuristic responses to relevant vignettes and with self-reported gambling in real life.
Article
Full-text available
Inoculation theory, a theory of conferring resistance to persuasive influence, has established efficacy as a messaging strategy in the health domain. In fact, the earliest research on the theory in the 1960s involved health issues to build empirical support for tenets in the inoculation framework. Over the ensuing decades, scholars have further examined the effectiveness of inoculation-based messages at creating robust positive health attitudes. We overview these efforts, highlight the structure of typical inoculation-based health messages, and describe the similarities and differences between this method of counter-persuasion and other preparatory techniques commonly employed by health researchers and practitioners. Finally, we consider contexts in which inoculation-oriented health messages could be most useful, and describe how the health domain could offer a useful scaffold to study conceptual issues of the theory.
Article
Full-text available
Concreteness-the psycholinguistic property of referring to a perceptible entity-enhances processing speed, comprehension, and memory. These represent selective filters for cognition likely to influence language evolution in competitive language environments. Taking a culturomics approach, we use multiple language corpora representing more than 350 billion words combined with concreteness norms for over 40,000 English words and demonstrate a systematic rise in concrete language in American English over the last 200years, both within and across word classes (nouns, verbs, and prepositions). Comparisons between new and old concreteness norms indicate this is not explained by semantic bleaching, but we find some evidence that the rise is related to changes in population demographics and may be associated with increasing numbers of second language learners or attention economics in response to crowding in the language market. We also examine the influence of gender and literacy. In sum, we demonstrate evolution in the psycholinguistic structure of American English, with a well-established impact on cognitive processing, which is likely to permeate modern language use. Copyright © 2015 Elsevier B.V. All rights reserved.
Article
Full-text available
Significance Public risk perception of hazardous events such as contagious outbreaks, terrorist attacks, and climate change are difficult-to-anticipate social phenomena. It is unclear how risk information will spread through social networks, how laypeople influence each other, and what social dynamics generate public opinion. We examine how messages detailing risks are transmitted from one person to another in experimental diffusion chains and how people influence each other as they propagate this information. Although the content of a message is gradually lost over repeated social transmissions, subjective perceptions of risk propagate and amplify due to social influence. These results provide quantitative insights into the public response to risk and the formation of often unnecessary fears and anxieties.
Article
Interest in risk communication has become increasingly widespread in recent years, attracting a global constituency of attention within government and industry alike. Starting, as it did, as an attempt to drive home the perceived wisdom of scientific risk assessments to unconvinced lay audiences, it has now diversified to encompass a number of wider objectives. As Baruch Fischhoff (1995, p. 138), in his recent review, concluded, whilst the starting-point of this shift may be parodied as “all we have to do is get the numbers right,” the current position amounts to “all we have to do is make (lay publics) partners.” Indeed, the emphasis on process issues: decision making, fairness, trust and “meaningful participation,” and ideas about democracy, have come to play a central role in both the research and policy development agendas (e.g. Chess et al. 1995; Renn, Webler, and Wiedemann 1995; Stern and Fineberg, 1996; Shell United Kingdom 1998; Department of Health 1998; United Kingdom Inter-departmental Liaison Group on Risk Assessment 1998b; Bennett and Calman 1999; House of Lords 2000). Underlying these changes, a number of interconnected social and political developments looms large. We are witnessing times of rapid change, economic re-structuring, and globalization, featuring a huge expansion of media and electronic communication, and significant cultural fragmentation. According to current wisdom, these changes, in combination, have produced a politicization of risk issues, the emergence of the active consumer and a new propensity for “moral panics” (Beck 1992; Giddens 1991; Thompson 1998).
Chapter
This chapter focuses on the social amplification of risk framework (SARF) proposed by Kasperson et al. (1988). Their thesis is that certain aspects of hazard events and their portrayal in mediated and other sources interact with psychological, social, institutional, and cultural processes in ways that might attenuate (decrease) or amplify (increase) perceptions of risk and, through this, shape behaviour. There is considerable evidence now that risk attenuation and intensification phenomena have occurred in America and Europe in recent years, raising the question of whether the SARF might inform our understanding of basic risk communication processes and lead to improved practice in health communication and in related areas such as environmental policy.