Energy Policy 122 (2018) 727–735

Contents lists available at ScienceDirect

Energy Policy
journal homepage: www.elsevier.com/locate/enpol

Differentiated effects of risk perception dimensions on nuclear power T

acceptance in South Korea
⁎
Seungkook Roha, Jin Won Leeb,
a
Policy Research Division, Korea Atomic Energy Research Institute (KAERI), 989-111 Daedeok-daero, Yuseong-gu, Daejeon 34057, Republic of Korea
b
Marketing Department, School of Business Administration, Jimei University, 185 Yinjiang Rd., Jimei District, Xiamen 361021, China

A R T I C LE I N FO A B S T R A C T

Keywords: Perceptions of risk from nuclear power generation have received considerable attention as a significant de-
Risk perceptions from nuclear plants terminant of public acceptance of nuclear power—a requirement to draw social and political support for nuclear
Nuclear power acceptance power generation. However, the distinction between the dimensions of risk perception and the differentiated
Intrinsic risk effects of such dimensions has been less explored. The present study demonstrates that two distinct dimensions
Extrinsic risk
of risk perception from nuclear plants—perceived intrinsic risk and extrinsic risk—reveal contrasting patterns of
Public policy
influence on the acceptance of nuclear power. Our results, using a sample from South Korea, show that the
relative effect of perceived extrinsic risk from nuclear plants on the acceptance of nuclear power (compared to
that of intrinsic risk) is stronger at higher levels of such acceptance than at lower levels. This finding provides
implications for public policy to encourage the acceptance of nuclear power. Such policy should selectively
choose between the two approaches—reducing the public's perceived level of intrinsic risk and that of extrinsic
risk—depending on whether the policy's aims is to soften opposing voices or to promote positive voices for
nuclear power.

1. Introduction incidents or catastrophes, such as natural disasters or military attacks.

Nuclear power is an electric power generation source of which the
development requires social and political sanctioning, given that the
double-edged nature of nuclear power—carrying great risk while of-
fering great benefits (Rogner, 2013)—causes sharp conflicts with regard
to nuclear power policies among stakeholders. Thus, the public accep-
tance of nuclear power exerts a significant influence on the nuclear and
energy policy of the country (Glaser, 2012; Visschers et al., 2011).
As a determinant of this acceptance, perceptions of the risks stem-
ming from nuclear power generation have received considerable at-
tention; the extant literature shows that such perceptions by the public
negatively influence their acceptance of nuclear power. Thus, reducing
the risks of nuclear power generation perceived by people has been one
of the main foci of public communications with regard to the devel-
opment and deployment of nuclear power policies (Sjöberg, 2009;
Stoutenborough et al., 2013).
Regarding such risks, the following research opportunity arises.
Although the risk consequences regarding nuclear plants vary, we posit
that one method by which to classify such risk consequences is to ca-
tegorize the risks into two types: (a) risks that arise from the normal
operation of a nuclear plant and (b) those that are caused by abnormal
We term the former and the latter intrinsic risk and extrinsic risk, re-
spectively. Given that these two types of risk conceptually differ, it may
be that the public perceives and reacts to them differently. However,
investigations of this possibility are difficult to find in the existing lit-
erature, which currently places potentially distinct types of negative
consequences from nuclear power generation into a single dimension of
risk.
Using a sample from South Korea, one of the leading countries in the
area of nuclear power generation (Choi et al., 2009), the present study
fills this research gap. In particular, our results demonstrate that a
decrease in perceived intrinsic risk from nuclear plants contributes only
to preventing individuals from the strongly negative acceptance of
nuclear power. In contrast, a decrease in perceived extrinsic risk is
found to contribute to turning moderately negative acceptance into
moderately positive acceptance and even promoting moderately posi-
tive acceptance to strongly positive acceptance. These findings imply
that public policy to encourage nuclear power acceptance must vary the
foci according to the goal (i.e., whether to soften strongly opposing
voices or to promote strongly positive agreement).

⁎
Corresponding author.
E-mail addresses: skroh@kaeri.re.kr (S. Roh), nevermean@empas.com (J.W. Lee).

https://doi.org/10.1016/j.enpol.2018.05.018
Received 14 December 2017; Received in revised form 6 May 2018; Accepted 8 May 2018
0301-4215/ © 2018 Elsevier Ltd. All rights reserved.
S. Roh, J.W. Lee
2. Theoretical and practical background
2.1 Risk perception measures in the extant empirical literature on nuclear
power acceptance
The risk perception approach toward nuclear power acceptance
refers to a model in which indicators of people's perceived risk from
nuclear technology or facilities (e.g., nuclear power generation, nuclear
plants, nuclear waste management, or nuclear waste repositories) are
important predictors of their acceptance of such technology or facilities
(Chung et al., 2008; Kunreuther et al., 1990; Sjöberg and Drottz-
Sjöberg, 2001). The measures for risk perception used in this approach
can largely be categorized in two groups: (1) measures that have
adopted the psychometric paradigm of risk perception, and (2) nuclear-
specialized measures for risk perception.
First, a group of researchers applied the risk attributes established in
the psychometric paradigm of risk perception (Fischhoff et al., 1978;
Slovic et al., 1980) to the measurement of perceived risk from nuclear
technology or facilities. The psychometric paradigm is a universal
theory about risk perception itself, rather than one focusing on risk
perception from nuclear technology or facilities. This paradigm assumes
several attributes/characteristics underlying the composition of risk
perception (e.g., dread, control, familiarity, numbers affected, cata-
strophic potential, etc.), and early studies such as Fischhoff et al. (1978)
used 18 attributes. The psychometric paradigm began as the re-
presentation and comparison of people's risk perceptions of multiple
targets. In a typical psychometric study of risk, respondents rate a set of
multiple risky targets (e.g., hazards) on a number of risk attributes.
Then using psychometric scaling and multivariate analysis techniques,
the researcher draws quantitative representations of the respondents’
risk perceptions of the targets in the form of a ‘cognitive map’ (Slovic
et al., 1986). Because these risk attributes had been found to be useful
in the understanding of underlying mechanisms of risk perception
(Sjöberg, 2000), researchers came to apply such attributes to the
measurement of people's risk perception from nuclear-related tech-
nology or facilities in the risk perception approach toward nuclear
power acceptance. Table 1A shows that measurement items or dimen-
sions from the psychometric paradigm have a substantial level of con-
sistency across studies. For example, the risk characteristics of newness
and dread have been commonly used as items or dimensions for risk
perception. Other items or dimensions than these two are also based on
the comprehensive sets of risk characteristics developed by Fischhoff
et al. (1978) and refined in other studies.
Second, the other category of indicators of risk perception is more
specialized to the contexts/cases of nuclear technology or facilities. As
aforementioned, the risk attributes and their associated factors (i.e.,
grouped sets of these attributes) from the psychometric paradigm are
universal to various kinds of risky targets. In contrast, several studies on
public acceptance of nuclear power have invented their own measures
for risk perception from nuclear technology or facilities, as summarized
in Table 1B. These nuclear-specialized measures have an advantage
from a practical viewpoint: they can be customized to measure people's
risk perceptions from specific risk sources (i.e., components or stages)
accompanying nuclear technology or facilities (e.g., see the measures
from Flynn et al. (1992) and Greenberg and Truelove (2011) in
Table 1B). Analyses of the responses to these measures can provide
understanding about the perception of which risk sources need to be
improved. However, in spite of this advantage, the nuclear-specialized
measures currently have room for improvement: they vary greatly
across studies, unlike the measures from the psychometric paradigm.
Moreover, there lacks a set of dimensions (factors) into which nuclear-
specialized items for risk perception could be categorized in a more
structural manner.
728
Energy Policy 122 (2018) 727–735
2.2 Subdividing risk perceptions from nuclear plants
As seen in Table 1B, nuclear-specialized measures provide descrip-
tions of various sources of risks that might occur during the operation of
nuclear technology or facilities. For example, the risk index items from
Flynn et al. (1992) and the risk belief items from Greenberg and
Truelove (2011) include risks that arises from abnormal events coming
from outside the nuclear technology or facility (i.e., earthquake, vol-
canic activity, sabotage, or terrorist attack). In contrast, some items
measure the respondents’ perception of whether a given nuclear-related
technology or facility is naturally (or chronically) safe or risky (e.g.,
“the buried waste will be contained in the repository so that con-
tamination of underground water supplies cannot occur” from Flynn
et al. (1992); “uranium mining degrades animals, plants, land, and
water” from Greenberg and Truelove (2011)). Noticing that risks from
nuclear plants also can be distinguished in a similar manner, we posit
that the perceptions of risk from nuclear plants can be largely cate-
gorized according to the sources of risks, as in the following paragraphs.
First, we define intrinsic risk from a nuclear plant as the risk of
negative, harmful consequences that arise chronically from the normal
operation of a nuclear plant. For example, the operation of a nuclear
plant can cause (or can be perceived by the public to cause) increases in
morbidity or radioactive contamination of the surrounding areas, even
when the plant is operating normally. We define perceived intrinsic risk
as an individual's perception of the degree to which nuclear plants are
prone to this intrinsic risk.
Second, we define extrinsic risk as the risk of negative consequences
stemming from abnormal external catastrophes such as natural dis-
asters or military attacks. For example, along with operational acci-
dents, natural factors such as earthquakes and tsunamis and artificial
factors such as war and terrorist events could cause major leakages of
radiation or radioactive pollutants from a nuclear plant. We define
perceived extrinsic risk as an individual's perception of how vulnerable
nuclear plants are to this extrinsic risk.
2.3 Relative effects of the perceptions of intrinsic and extrinsic nuclear risks
on nuclear power acceptance
Given that these two types of risk—intrinsic risk and extrinsic
risk—conceptually differ, there is a possibility that the public perceives
and reacts to them differently, as follows. In a nuclear plant context,
intrinsic risk refers to the risk that such plants chronically cause ne-
gative, harmful consequences (e.g., morbidity or radioactive con-
tamination) even if they are operating normally. In this respect, high
levels of intrinsic risk are a condition that leads to guaranteed failure
regarding the safety of nuclear plants. However, a low level or absence
of such a risk does not guarantee safety. This is because intrinsic risk is
not the only type of risk for nuclear plants; its absence does not ne-
cessarily connote overall safety if another risk type (i.e., extrinsic risk)
exists or is high. In these lights, low levels of intrinsic risk from nuclear
plants correspond to a must-be factor (Kano, 1984; Matzler and
Hinterhuber, 1998): when not fulfilled (i.e., high levels), they result in
guaranteed failure (i.e., very dangerous), but even when fulfilled (i.e.,
low levels), they only result in ambiguous success/failure (i.e., either
dangerous or safe).
Unlike intrinsic risk, high levels of extrinsic risk do not necessarily
lead to guaranteed failure regarding the safety of nuclear plants. Even if
extrinsic risk is high (i.e., nuclear plants are highly vulnerable to ex-
ternal disasters, such as earthquakes or military attacks), nuclear cat-
astrophes, such as major leakages of radiation or radioactive pollutants,
will not be actualized unless those external disasters actually occur. In
this respect, low levels of extrinsic risk are not a must-be factor for
nuclear safety: their un-fulfillment (i.e., high levels) does not
S. Roh, J.W. Lee Energy Policy 122 (2018) 727–735

Table 1
Empirical studies on the relationship between risk perception and acceptance of nuclear technology or facility.
Study Target of risk perception Measurement of risk perception Factor structure of risk perception

A. Studies that adopted risk characteristic variables from the psychometric paradigm of risk perception
Kunreuther et al. High-level nuclear waste Risk to self (“perceived seriousness of the risk to oneself Each item was used as an independent variable.
(1990) repository in local area from a repository”); risk to future (“a repository poses
serious risks to future generations”); controllability of risk
(“persons living near a repository would be able to control
the risks”); dread (“a repository would be dreaded by
nearby residents”).
Sjöberg and Drottz- Nuclear waste repository in Personal risk of nuclear waste repository; risk to children Each item was used as an independent variable.
Sjöberg (2001) local area and future generations; the risk being hard to avoid; risk
giving rise to anxiety and dread.
Sjöberg (2004) High-level nuclear waste New risk (three items on familiarity with the risk of Each of the three factors was measured using multiple
repository in local area radioactive material); dread (seven items on perceived items.
dread of accidents related to radioactive waste disposal
facility); interfering with nature (seven items on risk
created by tampering with nature).
Chung, Kim, and Radioactive waste disposal Perceived overall risk of radioactive waste disposal facility; Each item was used as an independent variable.
Rho (2008) facility in local area newness (familiarity with the risk of radioactive material);
dread (perceived dread of the accidents related with
radioactive waste disposal facility; private (trust in private
sector); central (trust in central government); local (trust in
local government).
Sjöberg and Drottz- Nuclear waste 21 characteristics of perceived risk: larger disasters, All the characteristic items were pooled to form one index
Sjöberg (2000) irreversible effects, injuries to animal life, injuries to plant variable for perceived risk.
life, injuries to future generation, causes cancer, hard to
avoid, certain to be deadly, warning of worse things,
immediate consequences, dreaded risk, against nature,
increases over time, hard to understand, cannot be sensed,
involuntary, unjust and immoral, hubris, kills immediately,
new and unknown, unknown by science.
Chung and Kim Radioactive waste disposal Overall risk (perceived overall risk of radioactive waste The items except newness were used as indicators of a
(2009) facility disposal facility); probability (perceived probability of an factor (risk perception) in structural equation model.
accident); seriousness (perceived seriousness of an
accident); newness (familiarity with the risk of radioactive
material); dread (perceived dread of accidents related to
radioactive waste disposal facility).

B. Studies that used nuclear-specialized measures

Flynn et al. (1992) High-level radioactive waste Image score: the respondent was asked to think about an Image score and risk index were used as indicators of a
repository underground high-level radioactive waste repository for a factor (perceived risk) in a structural equation model.
minute and write the first thought or image that comes to
mind; then the respondent evaluated this word or image on
a scale ranging from very negative to very positive.
Risk index: summed score for six questions—“highway and
rail accidents will occur in transporting the wastes to the
repository site”; “the repository can be made safe so that
future generations will not accidentally dig into the site
looking for resources”; “a future earthquake or volcanic
activity may cause release of the nuclear wastes”; “the
buried waste will be contained in the repository so that
contamination of underground water supplies cannot
occur”; “accidents will occur in handling the materials
during the burial operations and result in contamination of
workers or radioactive releases into the air and ground”;
“shipments of nuclear wastes can be made safe from
sabotage or attack by terrorists.”
Tanaka (2004) Nuclear facilities Risk from nuclear plants: “Do you think nuclear plants are Each item was used as an independent variable.
safe or dangerous?”; risk from high-level radioactive waste
repositories: “Do you think high-level radioactive waste
repositories are safe or dangerous?”
Whitfield et al. Nuclear technology and its Perceived level of risk about nuclear electric power plants; The three items were used as indicators of a factor (nuclear
(2009) by-product nuclear waste; nuclear weapons. risk) in structural equation model.
Greenberg and Nuclear energy Risk beliefs that “nuclear plants could have a serious Each risk belief item was used as an independent variable.
Truelove (2011) accident”; “nuclear waste is not safely managed”; “nuclear
facilities are vulnerable to terrorist attack”; “nuclear
material transport is dangerous to those who live along the
transport path”; “uranium mining degrades animals, plants,
land, and water.”
Visschers et al. Nuclear plant “The accident risks in Swiss nuclear power stations are The two items were used as indicators of a factor (risk
(2011) small”; “Swiss nuclear power stations are safe.” perception) in a structural equation model.
Visschers and Nuclear plant “I am very concerned about the dangers of nuclear power The three items were used as indicators of a factor
Siegrist (2013) stations in Switzerland”; “The nuclear power stations that (perceived risks) in a structural equation model.
are built now or in the near future are safe”; “The
catastrophe risks in new nuclear power stations are very
small.”
(continued on next page)

729
S. Roh, J.W. Lee
Table 1 (continued)
Study Target of risk perception Measurement of risk perception
Roh and Lee (2017a, Nuclear plant and radioactive Risk from nuclear plants: “Do you think nuclear plants in
2017b) waste our country are safe or not?”; risk from radioactive waste:
“Do you think radioactive waste in our country is managed
safely or not?”
Note: For economy of space, the table only includes studies that adopted the risk perception as a predictor of the acceptance of nuclear technology or facilities. For
example, studies that adopted risk characteristics from the psychometric paradigm as predictors of overall risk perception of nuclear technology/facility are not
included in the table.
necessarily bring guaranteed failure. Instead, such un-fulfillment only
results in ambiguous success/failure since overall safety is also depen-
dent on whether the external disasters actually occur and another risk
type (i.e., intrinsic risk) is high. Low levels of extrinsic risk also do not
necessarily connote overall safety since extrinsic risk is not the only risk
type for nuclear plants. Thus, low levels of extrinsic risk are con-
ceptually far from being an attractive factor—a factor whose un-ful-
fillment does not necessarily bring failure but whose fulfillment brings
success (Kano, 1984; Matzler and Hinterhuber, 1998). However, at
least, the safety consequence from low levels of extrinsic risk is closer to
success than that from high levels of extrinsic risk (i.e., ambiguous
success/failure) since the less vulnerable nuclear plants are to external
disasters, the more likely the plants are to be safe, with other things
being equal. In these lights, high and low levels of extrinsic risk result in
ambiguous success/failure and relatively greater success, respectively.
Overall, intrinsic and extrinsic risks from nuclear plants are likely to
lead to different ranges of safety of nuclear plants. As stated above, high
and low levels of intrinsic risk result in guaranteed failure and ambig-
uous success/failure, respectively. On the other hand, high and low
levels of extrinsic risk result in ambiguous success/failure and relatively
greater success, respectively. Since intrinsic risk is closer to being a
determinant between guaranteed failure and ambiguous success/failure
regarding safety, its perception is likely to influence whether the per-
ceivers’ responses to nuclear power generation are very negative or
neutral. Since extrinsic risk is closer to being a determinant between
ambiguous success/failure and relatively greater success, its perception
is likely to mainly influence whether the responses are neutral or po-
sitive. As such, there is a possibility that the relative effect of perception
of extrinsic risk from nuclear plants on the acceptance of nuclear power
will be stronger (compared to that of intrinsic risk) at high levels of the
acceptance than at lower levels. We will investigate this possibility
through empirical analyses in Section 4.2.
2.4 The effect of perceived nuclear risk on nuclear power acceptance in
South Korean society
The present study not only distinguishes between the dimensions of
risk perception (as stated in Section 2.2) but also examines the differ-
entiated effects of such dimensions (as stated in Section 2.3) on the
acceptance of nuclear power by using South Korea as the research
setting. The reasons for using a South Korean sample are twofold. One is
that acceptance of nuclear power is currently an important issue in
South Korean society. Due to insufficient domestic resources for fossil
fuel, about 98% of the fossil fuel South Korea consumes is imported.
Thus, nuclear power generation has been a strategic priority for the
country, and as of August 2017, 24 nuclear reactors provide about one-
third of the country's electricity (World Nuclear Association, 2017).
However, particularly in the aftermath of the Fukushima accident in
2011 (Siegrist and Visschers, 2013), there have been sharp conflicts of
opinion on whether to phase out nuclear power. These conflicts have
led recently to a deliberative opinion poll on whether to permanently
abandon the construction of the Shin-Gori No. 5 and 6 nuclear reactors.
730
Energy Policy 122 (2018) 727–735
Factor structure of risk perception
In Roh and Lee (2017a), the two items were pooled to form
one index variable for perceived risk; in Roh and Lee
(2017b), each item was used as an independent variable,
according to the research goals.
Although the poll resulted in the support of construction resumption,
the recent controversy over the Shin-Gori reactors seems to be only the
beginning of a long journey toward a social consensus on the country's
energy policy (The Diplomat, 2017). Using this deliberative democratic
process as a precedent, the public opinion is expected to exert a more
significant influence as a source of domestic legitimacy for the nuclear
and energy policy of the country.
The other factor we considered is that in South Korean society, risk
perception of nuclear power generation has been found to be a sig-
nificant determinant of the acceptance of nuclear power (e.g., Chung
and Kim, 2009; Chung et al., 2008; Roh and Lee, 2017a, b). Particu-
larly, Roh and Kim (2017) suggested that to improve the public ac-
ceptance of nuclear power, communication efforts should focus on
enhancing the perception of the safety of nuclear power (i.e., reducing
the risk perception) rather than emphasizing its necessity. Overall,
South Korea is a society where the public acceptance of nuclear power
is expected to exert a significant influence on national policies per-
taining to nuclear power generation, and risk perceptions are important
to this acceptance.
3. Methods
3.1 Sample and data collection
Our study dataset was compiled by the Korea Nuclear Energy
Agency—a Korean government-affiliated organization—in 2015. The
data collection, targeting a population of South Korean adult residents
(i.e., aged 19 and over), was subcontracted to an opinion research firm.
The collection adopted a quota sampling method so that the sample
could well represent the population. The subjects were selected based
on demographic proportioning (i.e., gender and age) and geographic
(i.e., residential area) sub-populations. Among the original dataset with
1009 respondents, we used responses from 868 respondents that were
complete (no missing responses) regarding our key measurement items.
Table 2 presents the characteristics of the study sample.
3.2 Measures
3.2.1 Perceived risk from nuclear plants
The respondents’ perception of intrinsic risk from nuclear plants was
measured using two four-point Likert-type items (1 =strongly agree;
2 =generally agree; 3 =generally disagree; and 4 =strongly disagree).
The responses were reverse-coded.
• Intrinsic 1: “Living around a nuclear plant increases the probability
of being sick.”
• Intrinsic 2: “A nuclear plant causes radioactive contamination in the
environment, even when no accident occurs.”
The perception of extrinsic risk was measured using three four-point
Likert-type items (1 =strongly agree; 2 =generally agree; 3 =generally
disagree; and 4 =strongly disagree).
S. Roh, J.W. Lee Energy Policy 122 (2018) 727–735

Table 2
Sample profile.
Variables Description Proportion

Gender Respondent's gender Male 49.77%

Female 50.23%
Age Respondent's age (measured in specific age) 19–29 17.28%
30–39 20.05%
40–49 22.92%
50–59 19.01%
60–69 12.68%
70 and older 8.06%
Educational level Respondent's education degree Middle school diploma or lower 13.94%
High school diploma 36.75%
College student or graduated 45.51%
Master's or higher 3.69%
No response 0.12%
Income level Respondent's monthly household income Below 1.0 million Korean Won 5.53%
1.00–1.99 million Korean Won 6.91%
2.00–2.99 million Korean Won 12.21%
3.00–3.99 million Korean Won 21.89%
4.00–4.99 million Korean Won 21.31%
5.00–5.99 million Korean Won 16.82%
6.00–6.99 million Korean Won 6.45%
7.00–7.99 million Korean Won 3.00%
8.00–8.99 million Korean Won 1.50%
9.00–9.99 million Korean Won 0.58%
10.00 million Korean Won and over 1.61%
No response 2.19%
Electricity consumption level Respondent's monthly household electricity consumption Below 10.0 thousand Korean Won 0.69%
10.00–19.99 thousand Korean Won 5.41%
20.00–29.99 thousand Korean Won 9.79%
30.00–49.99 thousand Korean Won 32.83%
50.00–69.99 thousand Korean Won 26.15%
70.00–99.99 thousand Korean Won 12.21%
100.00–199.99 thousand Korean Won 5.99%
200.00 thousand Korean Won and over 1.73%
No response 5.18%

Note: N = 868.

• Extrinsic 1: “Nuclear plants in our country are designed to endure
even a strong earthquake or a tsunami.”
• Extrinsic 2: “Nuclear plants in our country are unlikely to explode
given that they are protected by safety equipment, in contrast to the
Fukushima Nuclear Plant.”
• Extrinsic 3: “Nuclear plants in our country are well prepared for a
war or a terrorist event.”
3.2.3 Control variables
As general socio-demographic control variables, the gender, age,
educational level, and household income level of each respondent were
recorded. As another control variable that might be related specifically
to the respondents’ acceptance of a power generation source, their
monthly household electricity consumption levels were measured.

3.3 Analytical strategy

3.2.2 Nuclear power acceptance
Acceptance of nuclear power was measured using a four-point scale
with the following question: “Considering the situation in our country,
do you agree with nuclear power generation in our country or oppose
it?” (1 =strongly agree; 2 =generally agree; 3 =generally oppose; and
4 =strongly oppose). The responses were reverse-coded. Fig. 1 presents
the distribution of nuclear power acceptance in the study sample.
Fig. 1. Response distribution (percent) to the question about acceptance of
nuclear power. Note: The summed percent may not equal 100.0 because of
rounding.
As we noted in the previous section, we investigated the possibility
that the relative effects of perceived nuclear plant risk dimensions on
nuclear power acceptance differ across the levels of such acceptance.
Thus, there were two foci of our analysis. First, we investigated whether
the aforementioned items for perceived nuclear plant risk included our
two proposed dimensions—perceived intrinsic and extrinsic risks. For
this investigation, we utilized methods for construct validity—particu-
larly, factor analysis. Second, in estimating the effects of the two di-
mensions of perceived risk on nuclear power acceptance, we needed a
model that could deal with non-parallel effects of an independent
variable (i.e., the effect of an independent variable on the dependent
variable is not necessarily assumed to be identical across different levels
of the dependent variable). These effects can be investigated when the
dependent variable is treated as a multinomial categorical variable
(e.g., as in a multinomial logit or probit model). This is because such a
model separately estimates the effect of an independent variable with
regard to each pair of two levels of the dependent variable. Specifically,
we adopted a multinomial probit approach (McFadden, 1989). When
there are nA levels of nuclear power acceptance (i.e., n = 1, 2, …, nA),
subject i's acceptance level Ai is determined by the latent dependent
variables, Ain * s :

731
S. Roh, J.W. Lee Energy Policy 122 (2018) 727–735

Table 3
Exploratory factor analysis results.

Items Factor loadingsa Communalitiesa

Factor I Factor II
(perceived extrinsic risk) (perceived intrinsic risk)

Intrinsic 1 0.074 0.867 0.756

Intrinsic 2 0.004 0.874 0.765
Extrinsic 1 0.844 0.000 0.712
Extrinsic 2 0.880 0.048 0.777
Extrinsic 3 0.843 0.069 0.715

Eigenvalue (unrotated) 2.241 1.484

Cumulativeb
Variance explained 44.81% 29.67% 74.48%
Kaiser-Meyer-Olkin measure of sampling adequacyc 0.657
Bartlett’s test of sphericityd
Chi-square 1209.273
Degree of freedom 10
p-value 0.000

Note: N = 868. Factor extraction method: principal components method; eigenvalue threshold: 1.0; factor rotation method: varimax.
a
Common thresholds are 0.50.
b
Common threshold is 60% (Hair et al., 2006).
c
Values of 0.50 or higher are considered suitable.
d
Significant values are suitable (Tabachnick et al., 2006).

⎧ 1 if Ai1 * > Ai2 * , Ai3 * , …, AinA * Table 4

⎪ Confirmatory factor analysis results.
Ai = 2 if Ai2 * > Ai1 * , Ai3 * , …, AinA * where A n *
i
⎨ ⋮ ⋮ Constructs and measurement items
⎪
⎩ nA otherwise
= βn Xi + εn and ε ~ N (0, Σ).
Here, Ain * is influenced by Xi, the vectors of the two risk perception
dimensions (our main independent variables) and socio-demographic
features (control variables) of subject i. Moreover, ε , the vector of re-
siduals ε s, follows a multivariate normal distribution N (0, Σ), in which
the εs are correlated with each other.
4. Results
4.1 Investigation of perceived nuclear plant risk dimensions
Regarding our proposed constructs of two dimensions of perceived
risk from nuclear plants, we assessed construct validity as follows. First,
we conducted an exploratory factor analysis (EFA) as in Table 3. For
factor extraction, the principal components method and the 1.0 ei-
genvalue threshold were used; for factor rotation, varimax rotation was
used. A Kaiser-Meyer-Olkin measure of 0.657 (> the 0.50 criterion)
and a significant value from Bartlett's test of sphericity (χ2 (10)
= 1209.273, p < 0.05) show that the data is suitable for an EFA. The
cumulative variance explained of 74.48%, far exceeding the common
threshold of 60% (Hair et al., 2006), shows that the variance among the
measurement items is sufficiently well explained by the extracted fac-
tors. Thus, the information from the lower half of Table 3 shows that
the EFA solution is reliable enough to adopt.
As presented in the upper half of Table 3, the EFA generated the
expected scale dimensions: every measurement item is loaded on its
intended factor (dimension). All the communalities (≥ 0.71) far ex-
ceeding the threshold of 0.50 (Hair et al., 2006) show that the mea-
surement items are sufficiently well explained by the two extracted
factors. Moreover, the factor loadings on the intended factors (≥ 0.84)
far exceeded the 0.50 threshold (Hair et al., 2006). The factor loadings
on the non-intended factors (< 0.08) far below the 0.32 limit show
that cross-loading is unlikely (Tabachnick et al., 2006). Overall, the
factor loadings show that each measurement item sufficiently well
Standardized loadings
Perceived intrinsic risk (AVE = 0.629, CR = 0.835)
(1) Intrinsic 1 0.992
Intrinsic 2 0.523
Perceived extrinsic risk (AVE = 0.604, CR = 0.877)
Extrinsic 1 0.733
Extrinsic 2 0.849
Extrinsic 3 0.745
Note: N = 868. All fit indices are of sufficiently satisfactory levels according to
the standard presented in Hair et al. (2006). AVE=average variance extracted;
CR=composite reliability.
reflects its own intended factor, and only its own intended factor. Thus,
the information from the upper half of Table 3 overall indicates that our
proposed two dimensions provide a valid structure for perceived risk
from nuclear plants.
Second, among the extracted factors, perceived intrinsic risk had a
Cronbach's α value of 0.683, slightly below the 0.70 cutoff (Lance et al.,
2006), whereas perceived extrinsic risk (α = 0.818) exceeded such a
cutoff. However, Cronbach's α is reported to be biased downward in
cases with a small number of items (Cortina, 1993). Instead, the values
of the factor loadings (≥ 0.84) and communalities (≥ 0.71) from the
EFA, both of which far exceeded the thresholds of 0.50 (Hair et al.,
2006), show that the items within a construct have internal consistency.
Third, based on the risk dimensions from the EFA, we conducted a
confirmatory factor analysis (CFA) as presented in Table 4. The CFA
solution did not yield any abnormal estimate such as in the Heywood
case. The factor loadings of all items were significant and substantial
(ps < 0.001, loadings > 0.50), and the model fit was sufficiently sa-
tisfactory (χ2 (4)= 6.292, normed χ2 = 1.573, p < 0.001, goodness of
fit index= 0.997, confirmatory fit index= 0.998, Tucker-Lewis
index= 0.995, standardized root mean square residual= 0.013, root
mean square error of approximation= 0.026) (Hair et al., 2006).
The risk dimensions far exceeded the threshold of 0.50 for the
average variance extracted (AVE) and the 0.70 for composite reliability
(CR) (Fornell and Larcker, 1981; Hair et al., 2006) (see Table 4), in-
dicating good convergent validity. The AVE of each dimension was

732
S. Roh, J.W. Lee Energy Policy 122 (2018) 727–735

Table 5
Multinomial probit results.
“Generally oppose” compared to “Generally agree” compared to “Strongly agree” compared to
“strongly oppose” “generally oppose” “generally agree”
β (Std. Err.) β (Std. Err.) β (Std. Err.)

Control variables
Gendera − 0.093 (0.158) − 0.083 (0.082) − 0.211 (0.105)*
Ageb − 0.282 (0.192) 0.074 (0.095) 0.239 (0.116)*
Educational levelc − 0.156 (0.192) − 0.201 (0.096)* 0.056 (0.119)
Income levelb − 0.373 (0.164)* 0.088 (0.094) 0.134 (0.112)
Electricity consumption levelb 0.099 (0.168) 0.016 (0.083) − 0.137 (0.108)
Risk perception variables
Perceived intrinsic riskb − 0.440 (0.169)** 0.016 (0.083) 0.099 (0.104)
Perceived extrinsic riskb − 0.010 (0.174) − 0.383 (0.089)*** − 0.517 (0.103)***

Log likelihood= −948.573. Wald χ2 (21)= 113.47. Model p-value= 0.0000.

Note: N = 868.
* p < 0.05.
** p < 0.01.
*** p < 0.001.
a
Contrast coded: male= −1, female= 1.
b
The variable has been standardized.
c
Contrast coded: high school diploma or lower= −1, current college student or higher degree= 1.
The highest variance inflation factor (VIF) was 1.38, far below the threshold of 10.0 for multicollinearity (O’Brien, 2007).

larger than the square of the correlations between the dimensions does not contribute to turning a “generally oppose” opinion to “gen-
(r = 0.115), indicating good discriminant validity (Fornell and Larcker, erally agree” (βrev, intrinsic=−0.016, z = −0.20, p > 0.10) and does
1981). Overall, the two risk dimensions were both measured by in- not strengthen “generally agree” to “strongly agree” (βrev, intrinsic
dicators that sufficiently represent the corresponding dimension; it was =−0.099, z =− 0.96, p > 0.10). In contrast, a decrease in perceived
found that these two dimensions are distinct from each other. extrinsic risk does not significantly help to soften a “strongly oppose”
opinion to “generally oppose” (βrev, extrinsic = 0.010, z = 0.06,
p > 0.10). However, such a decrease contributes to turning a “gen-
4.2 Multinomial probit results
erally oppose” opinion into “generally agree” (βrev, extrinsic = 0.383,
z = 4.30, p < 0.001) and strengthening a “generally agree” opinion to
Table 5 summarizes the results of the multinomial probit regression.
“strongly agree” (βrev, extrinsic = 0.517, z = 5.00, p < 0.001). Overall,
A positive/negative coefficient indicates that individuals higher/lower
intrinsic and extrinsic risk perceptions from nuclear plants are found to
in the variable are more likely to have a given level of nuclear power
exert contrasting patterns with regard to how they influence nuclear
acceptance compared to the reference level. For example, on the left-
power acceptance levels.
side set of estimates, those who perceive intrinsic risk from nuclear
We compared the difference in the effects of decreases in the in-
plants as lower are likely to respond with “generally oppose” rather than
trinsic and extrinsic risk perceptions (i.e., the difference in the reversed
“strongly oppose” with regard to nuclear power generation
coefficients: Δβrev = βrev, extrinsic−βrev, intrinsic) across the acceptance
(β = −0.440, z = −2.60, p < 0.01).
levels, as in Table 6. The coefficient difference test of Clogg et al. (1995)
Based on these estimates, Fig. 2 compares the effects of a decrease in
showed that, regarding whether an individual “generally opposes”
perceived risk on the nuclear power acceptance levels. A reversed
nuclear power generation compared to “strongly opposing” it, a de-
coefficient, βrev, in Fig. 2 has the same absolute value as the original
crease in perceived intrinsic risk is found to exert a marginally more
coefficient of its corresponding risk perception variable in Table 5, but
positive effect than that in perceived extrinsic risk (Δβrev,1 = −0.430,
the sign is reversed (e.g., βrev, intrinsic=−1 × βintrinsic, where βintrinsic is
z = − 1.65, p < 0.10). In contrast, a decrease in perceived extrinsic
the coefficient of intrinsic risk perception). Thus, a positive coefficient
risk is found to exert a more positive effect than that in perceived in-
in Fig. 2 means that a decrease in that risk perception increases the
trinsic risk regarding whether an individual “generally agrees” to nu-
likelihood of a higher level of nuclear power acceptance. A decrease in
clear power generation compared to “generally opposing” (Δβrev,2 =
perceived intrinsic risk is helpful to soften a “strongly oppose” opinion
0.400, z = 3.12, p < 0.01) and whether he/she “strongly agrees”
to “generally oppose” (βrev, intrinsic = 0.440, z = 2.60, p < 0.01), but it

Fig. 2. The effects of decreases in perceived risk levels on

nuclear power acceptance. Note: ** p < 0.01;
*** p < 0.001. † The two coefficients are different at the
0.10 level; †† at the 0.01 level; ††† at the 0.001 level. a
Reversed β, of which the sign is reversed compared to its
corresponding coefficient in Table 5, representing the ef-
fect of a decrease in the risk perception on the likelihood
of a higher level of nuclear power acceptance (e.g., the
likelihood of “generally oppose” rather than “strongly
oppose”). The perceived risk variables have been stan-
dardized. The variance of a latent dependent variable in
the multinomial probit model is 2.

733
S. Roh, J.W. Lee
Table 6
Difference in the reversed coefficients of perceived intrinsic and extrinsic risks from nuclear plants.
“Generally oppose” compared to
“strongly oppose”
βrev, intrinsic = −βintrinsica 0.440 (0.169)**
βrev, extrinsic = −βextrinsicb 0.010 (0.174)
Δβrev, 1
Δβrev = βrev, extrinsic − βrev, intrinsic −0.430 (0.261) ‡, c, d
Note: *p < 0.05;
‡
p < 0.10;
** p < 0.01;
*** p < 0.001.
a
Denotes the effect of a decrease in the perception of intrinsic risk from nuclear plants on the latent variable of a higher level of nuclear power acceptance.
b
Such effect of a decrease in the perception of extrinsic risk.
c
The two estimates are different at the 0.01 level.
d,e
The significance of the difference between the two estimates is not testable through the current methodology of coefficient difference test (Clogg et al., 1995).
compared to “generally agreeing” (Δβrev,3 =0.616, z = 4.10,
p < 0.001). As shown in the last row of Table 6, the reversed coeffi-
cient difference regarding “generally oppose”–“generally agree” levels
of nuclear power acceptance (Δβrev,2 = 0.400) is between that re-
garding “strongly oppose”–“generally oppose” levels (Δβrev,1 = −
0.430), and that regarding “generally agree”–“strongly agree” levels
(Δβrev,3 = 0.616).
The investigation of whether this pattern is statistically significant
(i.e., Δβrev significantly differs across the acceptance levels) requires the
following consideration. To investigate whether the difference between
two given coefficients are significant, the estimates for the coefficients
and their standard errors are not sufficient: either (1) the covariance of
the two coefficients should be provided, or (2) the coefficients should
be independent (i.e., estimated from separate sub-samples) (Clogg
et al., 1995). Regarding prerequisite (1), the current regression meth-
odology can provide the estimate for the covariance between the two
coefficients (Cov (βA, βB)), but not that between two coefficient dif-
ferences (Cov (ΔβA, ΔβB)). Regarding prerequisite (2), the pair of
“generally oppose”–“generally agree” partially coincides that of
“strongly oppose”–“generally oppose” (i.e., both pairs include the
identical respondent group of “generally oppose”) and that of “gen-
erally agree–”strongly agree” (i.e., both include the identical group of
“generally agree”); only the pair of “strongly oppose”–“generally op-
pose” and that of “generally agree–“strongly agree” are mutually ex-
clusive. As a result, only between these two pairs, can we investigate
whether the reversed coefficient difference (Δβrev) differs.
A coefficient difference test (Clogg et al., 1995), in which we set the
two Δβrev to be independent, showed that the reversed coefficient dif-
ference is more positive regarding “generally agree”–“strongly agree”
levels of nuclear power acceptance than “strongly oppose”–“generally
oppose” levels (Δβrev,3 − Δβrev,1 =1.046, z = 3.47, p < 0.01). This
means that a decrease in the extrinsic risk perception exerts a more
positive effect than that in the intrinsic risk perception in strengthening
a “generally agree” opinion to “strongly agree” than in softening a
“strongly oppose” opinion to “generally oppose.” Thus, at least, we can
state that a decrease in the perception of extrinsic risk from nuclear
plants exerts a more positive effect on nuclear power acceptance than
that in the perception of intrinsic risk in promoting moderately positive
acceptance to strongly positive acceptance compared to preventing
strongly negative acceptance levels of such generation, consistent with
our proposed prediction.
5. Conclusions
5.1 Implications
We categorized perceived risk from nuclear plants into two
734
Energy Policy 122 (2018) 727–735
“Generally agree” compared to “Strongly agree” compared to
“generally oppose” “generally agree”
−0.016 (0.083) −0.099 (0.104)
0.383 (0.089)*** 0.517 (0.103)***
Δβrev, 2 Δβrev, 3
0.400 (0.128)***, d, e
0.616 (0.150)***, c, e
dimensions—perceived intrinsic risk and extrinsic risk—according to
the sources of such risk. A series of factor analyses confirmed the va-
lidity of this factor structure. Our multinomial probit results show that
the relative effect of perception of extrinsic risk from nuclear plants on
the acceptance of nuclear power (compared to that of intrinsic risk) will
be stronger at high levels of the acceptance than at lower levels. These
overall suggest the importance of the distinction between perceived
intrinsic risk and extrinsic risk from nuclear plants, for they not only are
distinctive constructs but also reveal differentiated effects on nuclear
power acceptance. These findings contribute to the knowledge of di-
mensions composing nuclear-related risk perception, which has been
scarce in the case of nuclear-specialized measures of risk perceptions
(see Table 1B).
The relative importance of the risk type provides practical im-
plications regarding the risk governance of nuclear power. In risk
governance—the institutional and social processes that guide the col-
lective activities and decisions of a society regarding a risky object
(Kheifets et al., 2010; Renn et al., 2011)—, the public is an important
actor group (van Asselt and Renn, 2011). In particular, in the risk
governance of nuclear power, the public's acceptance of nuclear power
is a significant influencer on the country's nuclear power policy (Allen,
2017; Choi and Lee, 1995; Cohen, 1996). Regarding this acceptance,
perceptions of risk from nuclear power technology or facilities nega-
tively influence, which the extant literature shows (e.g., Visschers et al.,
2011; Whitfield et al., 2009) and the present study confirms.
Perceived levels of risk from a certain object are formed based on
the actual levels of such risk, although they may not perfectly coincide
(Jasanoff, 1998). Thus, to enhance the public's nuclear power accep-
tance through risk governance, resources and efforts should be invested
to reduce both actual and perceived risks of nuclear power generation.
According to the findings of the present study, investment priorities
should vary across the risk types, depending on the policy's aims. For
example, in the recent survey of public deliberation on the Shin-Gori
nuclear reactors No. 5 and 6, 53.2% of respondents were found to favor
reduced usage of nuclear power, far outnumbering the 35.5% favoring
continued reliance on nuclear power and the 9.7% supporting its ex-
pansion (The Public Deliberation Committee on Shin-Gori Nuclear
Reactors No. 5 and 6, 2017). In such a situation, the nuclear industry
professionals and policy-makers may consider focusing on softening
public opposition to nuclear power generation by allocating resources
and efforts preferentially to the improvement of both actual/objective
and perceived/subjective levels of intrinsic risk from nuclear plants.
5.2 Limitations and future extension
The perceived intrinsic risk and extrinsic risk in the present study
are about nuclear plants. However, although risk related to nuclear
S. Roh, J.W. Lee
plants is the primary component in risk from nuclear power generation,
it is not the only, exclusive component. That risk related to radioactive
waste facilities also is a significant component in risk from nuclear
power generation and, thus, can affect the public acceptance of nuclear
power (Roh and Lee, 2017b). In this light, the investigation of the factor
structure of other components of nuclear power risk might be neces-
sary. For example, will the dimensions of intrinsic risk vs. extrinsic risk
also be valid regarding the case of risk from radioactive waste facilities?
If not, what alternative dimensions could be possible? Will distinct di-
mensions reveal differentiated effects on nuclear power acceptance?
Another issue is the degree to which the findings of the present
study can be generalized. For example, based on the characteristics of
intrinsic and extrinsic risks from nuclear plants, we have demonstrated
that the relative effect of the perception of extrinsic risk on nuclear
power acceptance is stronger (compared to that of intrinsic risk) at
higher levels of acceptance than at lower levels. However, one may
question whether this finding is universal or country-specific. For ex-
ample, in a country where extrinsic risk actualized in the past (e.g.,
Japan, which experienced the Fukushima accident in 2011), will high
levels of extrinsic risk be regarded as a must-be factor, unlike in the
results of the present study, since the catastrophes from this risk have
been imprinted in the public's mind (Poortinga et al., 2013)? Future
research under various research settings (e.g., different countries) is
necessary.
A research opportunity may not lie only in the risk perceptions.
Extending the approach used here to the dimensions of other determi-
nants of nuclear power acceptance (e.g., perceived benefits) may pro-
vide future research opportunities. For example, the benefits from nu-
clear power generation could be categorized into distinct dimensions,
such as economic and environmental benefits (e.g., energy supply
benefits and climate change benefits in Visschers et al. (2011)). An
investigation into whether the perceptions of these dimensions exert
differentiated effects on nuclear power acceptance may deepen our
understanding of the effects of the determinants of energy acceptance
overall.
References
Allen, F., 2017. The Management of Risk to Society from Potential Accidents: the Main
Report of the UKAEA Working Group on the Risks to Society from Potential Major
Accidents. Chapman and Hall/CRC.
Choi, S., Jun, E., Hwang, I., Starz, A., Mazour, T., Chang, S., Burkart, A.R., 2009. Fourteen
lessons learned from the successful nuclear power program of the Republic of Korea.
Energy Policy 37, 5494–5508.
Choi, Y.S., Lee, B.W., 1995. Analysis of policy alternatives on the public acceptance of
nuclear power plant in Korea. J. Nucl. Sci. Technol. 32, 218–225.
Chung, J.B., Kim, H.-K., 2009. Competition, economic benefits, trust, and risk perception
in siting a potentially hazardous facility. Landsc. Urban Plan. 91, 8–16.
Chung, J.B., Kim, H.K., Rho, S.K., 2008. Analysis of local acceptance of a radioactive
waste disposal facility. Risk Anal. 28, 1021–1032.
Clogg, C.C., Petkova, E., Haritou, A., 1995. Statistical methods for comparing regression
coefficients between models. Am. J. Sociol. 100, 1261–1293.
Cohen, B.L., 1996. Problems in public acceptance of nuclear power in U.S. J. Nucl. Sci.
Technol. 33, 1–6.
Cortina, J.M., 1993. What is coefficient alpha? An examination of theory and applica-
tions. J. Appl. Psychol. 78, 98–104.
Fischhoff, B., Slovic, P., Lichtenstein, S., Read, S., Combs, B., 1978. How safe is safe
enough? A psychometric study of attitudes towards technological risks and benefits.
Policy Sci. 9, 127–152.
Flynn, J., Burns, W., Mertz, C., Slovic, P., 1992. Trust as a determinant of opposition to a
high-level radioactive waste repository: analysis of a structural model. Risk Anal. 12,
417–429.
Fornell, C., Larcker, D.F., 1981. Evaluating structural equation models with unobservable
variables and measurement error. J. Mark. Res. 18, 39–50.
Glaser, A., 2012. From Brokdorf to Fukushima: the long journey to nuclear phase-out.
Bull. At. Sci. 68, 10–21.
Greenberg, M., Truelove, H.B., 2011. Energy choices and risk beliefs: is it just global
735
Energy Policy 122 (2018) 727–735
warming and fear of a nuclear power plant accident? Risk Anal. 31, 819–831.
Hair, J.F., Black, W.C., Babin, B.J., Anderson, R.E., Tatham, R.L., 2006. Multivariate Data
Analysis, 6th ed. Pearson Prentice Hall, Upper Saddle River, NJ.
Jasanoff, S., 1998. The political science of risk perception. Reliab. Eng. Syst. Saf. 59,
91–99.
Kano, N., 1984. Attractive quality and must-be quality. J. Jpn. Soc. Qual. Control (Jpn.)
14, 39–48.
Kheifets, L., Swanson, J., Kandel, S., Malloy, T.F., 2010. Risk governance for mobile
phones, power lines, and other EMF technologies. Risk Anal. 30, 1481–1494.
Kunreuther, H., Easterling, D., Desvousges, W., Slovic, P., 1990. Public attitudes toward
siting a high-level nuclear waste repository in Nevada. Risk Anal. 10, 469–484.
Lance, C.E., Butts, M.M., Michels, L.C., 2006. The sources of four commonly reported
cutoff criteria: what did they really say? Organ. Res. Methods 9, 202–220.
Matzler, K., Hinterhuber, H.H., 1998. How to make product development projects more
successful by integrating Kano's model of customer satisfaction into quality function
deployment. Technovation 18, 25–38.
McFadden, D., 1989. A method of simulated moments for estimation of discrete response
models without numerical integration. Econometrica 57, 995–1026.
O’Brien, R.M., 2007. A caution regarding rules of thumb for variance inflation factors.
Qual. Quant. 41, 673–690.
Poortinga, W., Aoyagi, M., Pidgeon, N.F., 2013. Public perceptions of climate change and
energy futures before and after the Fukushima accident: a comparison between
Britain and Japan. Energy Policy 62, 1204–1211.
Renn, O., Klinke, A., van Asselt, M., 2011. Coping with complexity, uncertainty and
ambiguity in risk governance: a synthesis. AMBIO 40, 231–246.
Rogner, H.-H., 2013. World outlook for nuclear power. Energy Strategy Rev. 1, 291–295.
Roh, S., Kim, D., 2017. The factors of nuclear energy public acceptance and relative
importance (public acceptance factors and relative importance). Energy Sources, Part
B: Econ. Plan. Policy 12, 559–564.
Roh, S., Lee, J.W., 2017a. Differentiated influences of benefit and risk perceptions on
nuclear power acceptance according to acceptance levels: evidence from Korea. J.
Nucl. Sci. Technol. 54, 830–836.
Roh, S., Lee, J.W., 2017b. Differentiated influences of risk perceptions on nuclear power
acceptance according to acceptance targets: evidence from Korea. Nucl. Eng.
Technol. 49, 1090–1094.
Siegrist, M., Visschers, V.H., 2013. Acceptance of nuclear power: the Fukushima effect.
Energy Policy 59, 112–119.
Sjöberg, L., 2000. Factors in risk perception. Risk Anal. 20, 1–12.
Sjöberg, L., 2004. Local acceptance of a high-level nuclear waste repository. Risk Anal.
24, 737–749.
Sjöberg, L., 2009. Precautionary attitudes and the acceptance of a local nuclear waste
repository. Saf. Sci. 47, 542–546.
Sjöberg, L., Drottz-Sjöberg, B.-M., 2000. Public risk perception of nuclear waste. Int. J.
Risk Assess. Manag. 11, 248–280.
Sjöberg, L., Drottz-Sjöberg, B.-M., 2001. Fairness, risk and risk tolerance in the siting of a
nuclear waste repository. J. Risk Res. 4, 75–101.
Slovic, P., Fischhoff, B., Lichtenstein, S., 1980. Facts and fears: understanding perceived
risk. in: Schwing, R., Albers, J. (Eds.), Societal Risk Assessment: How Safe Is Safe
Enough. Plenum, New York, pp. 181–214.
Slovic, P., Fischhoff, B., Lichtenstein, S., 1986. The psychometric study of risk perception.
in: Menkes, J., Covello, V., Mumpower, J. (Eds.), Risk Evaluation and Management.
Springer, New York, pp. 3–24.
Stoutenborough, J.W., Sturgess, S.G., Vedlitz, A., 2013. Knowledge, risk, and policy
support: public perceptions of nuclear power. Energy Policy 62, 176–184.
Tabachnick, B.G., Fidell, L.S., Osterlind, S.J., 2006. Using Multivariate Statistics, 5th ed.
Allyn and Bacon, New York.
Tanaka, Y., 2004. Major psychological factors determining public acceptance of the siting
of nuclear facilities. J. Appl. Social. Psychol. 34, 1147–1165.
The Diplomat, 2017. South Korea's Nuclear Energy Debate. 〈https://thediplomat.com/
2017/10/south-koreas-nuclear-energy-debate/〉 (Accessed 2 April 2018).
The Public Deliberation Committee on Shin-Gori Nuclear Reactors No. 5 & 6, 2017.
Results of Participatory Surveys for Public Deliberation on Shin-Gori Nuclear
Reactors No. 5 & 6. 〈http://cdd.stanford.edu/2017/results-of-participatory-surveys-
for-public-deliberation-on-shin-gori-nuclear-reactors-no-5-6/〉 (Accessed 2 April
2018).
van Asselt, M.B., Renn, O., 2011. Risk governance. J. Risk Res. 14, 431–449.
Visschers, V.H., Keller, C., Siegrist, M., 2011. Climate change benefits and energy supply
benefits as determinants of acceptance of nuclear power stations: investigating an
explanatory model. Energy Policy 39, 3621–3629.
Visschers, V.H., Siegrist, M., 2013. How a nuclear power plant accident influences accep-
tance of nuclear power: results of a longitudinal study before and after the Fukushima
disaster. Risk Anal. 33, 333–347.
Whitfield, S.C., Rosa, E.A., Dan, A., Dietz, T., 2009. The future of nuclear power: value
orientations and risk perception. Risk Anal. 29, 425–437.
World Nuclear Association, 2017. Nuclear Power in South Korea. 〈http://www.world-
nuclear.org/information-library/country-profiles/countries-o-s/south-korea.aspx〉
(Accessed 2 April 2018).