sh

etss

.orre

c People have stable attitudes towards nuclear power.

Article history:

Received 10 April 2012

Accepted 31 July 2012

tional research questions. In a longitudinal survey that included

tor of whether it

data, suggest that nuclear power is perceived as a rather dreadful

Contents lists available at SciVerse ScienceDirect

.el

Energy

Energy Policy 59 (2013) 112119affective evaluation of the hazard (Dohle et al., 2010). Due to theseE-mail address: msiegrist@ethz.ch (M. Siegrist).three waves, the present study examined how the accident in and unknown hazard (Fischhoff et al., 1978; Slovic, 1987). Thisnding could be replicated by analyzing non-aggregated data, inwhich individual differences were taken into account (Siegrist et al.,2005). The dreadfulness dimension seems to be associated with an

0301-4215/$ - see front matter & 2012 Elsevier Ltd. All rights reserved.

http://dx.doi.org/10.1016/j.enpol.2012.07.051

n Corresponding author. Tel.: 41 44 6326321.towards nuclear power (Eiser et al., 1989; Verplanken, 1989). Theaccident in Japan provided an opportunity for examining addi-

is used in a country for the production of electricity (Goodfellowet al., 2011). Results of the psychometric paradigm, using aggregatednot well prepared for such an event. After the accident, radio-active elements were released or leaked from three reactors intothe environment. People living in a 30 km radius around thepower plant were evacuated.

Based on the research about the accident in Chernobyl, it isknown that such accidents may result in a more negative attitude

being produced by nuclear power plants.

1.1. Lay peoples perception of nuclear power

Public perception of nuclear power is a key facIn various countries, discussions about the future of nuclearpower have been fueled by the accident in the Fukushima Daiichinuclear power plant in Japan. This accident happened March 11,2011. It was the result of an enormous earthquake and thesubsequent tsunami, combined with a power company that was

after the accident, compared to before the accident. The studydesign allowed investigating the stability of attitudes towardsnuclear power. The study further examined how attitude towardsnuclear power and the accident in Fukushima shaped the evalua-tion of scenarios with different percentages of national power1. Introduction Fukushima inuenced attitudes towards and acceptance ofnuclear power immediately after the accident and half a yearAvailable online 19 August 2012

Keywords:

Nuclear power

Risk perception

Fukushimaa b s t r a c t

Utilizing a longitudinal study design, the impact of the 2011 accident in Fukushima on acceptance of

nuclear power and the evaluation of several scenarios with different percentages of nuclear power were

examined. Mail surveys were conducted in the German-speaking part of Switzerland. The rst survey

took place before the accident in Fukushima (Autumn 2010), the second survey immediately after the

accident (March 2011), and the third survey half a year after the accident (October 2011). A sample of

463 persons participated in all three surveys. The accident had a negative impact on the acceptance of

nuclear power. The mean change was moderate, and high correlations between the measurement

points were observed. Overall, participants thus showed rather stable attitudes towards nuclear power

across the three measurement waves. Results of the present study demonstrate the importance of prior

beliefs and attitudes for the interpretation of an accident. The evaluation of the various scenarios was

strongly inuenced by participants pre-Fukushima attitudes towards nuclear power.

& 2012 Elsevier Ltd. All rights reserved.a r t i c l e i n f oAcceptance of nuclear power: The Fuku

Michael Siegrist n, Vivianne H.M. Visschers

ETH Zurich, Institute for Environmental Decisions (IED), Consumer Behavior, Universita

H I G H L I G H T S

c Longitudinal studies are important for risk perception research.c The accident in Fukushima had only a moderate impact on acceptancec Acceptance of nuclear power before and after Fukushima was highly c

journal homepage: wwwtrasse 22 CHN J 76.3, 8092 Zurich, Switzerland

lated.ima effect

sevier.com/locate/enpol

Policy

M. Siegrist, V.H.M. Visschers / Energy Policy 59 (2013) 112119 113qualitative aspects of nuclear power, lay people tend to perceivenuclear power as a risky technology.

The affect heuristic suggests that images associated with, andassociations evoked by, nuclear power determine peoples per-ception of this technology (Slovic et al., 2004; Finucane et al.,2000). The importance of affect for shaping peoples perception ofnuclear power has been shown in several studies (Dohle et al.,2010; Keller et al., 2012; Siegrist et al., 2006). Opponents andproponents not only differ in the proposed affect evoked by thistechnology, but also in the concrete images and associations.Affective images seem to determine peoples acceptance ofnuclear power (Keller et al., 2012). People opposed to thereplacement of nuclear power plants associated nuclear powerplants with images like radioactivity, nuclear accidents, risks andnegative consequences for health and the environment, or evennuclear war. These results indicated that the concepts of risk anddread are more often expressed by people who oppose thereplacement of nuclear power plants than by people who are infavor of nuclear power plants. People who supported the replace-ment of nuclear power plants often associated nuclear powerplants with neutral and positive concepts such as energy, and to asmaller extent, with necessity. The perception of benecial energysupply primarily explains their acceptance of replacing nuclearpower plants. Thus, people in favor of replacing nuclear powerplants perceived nuclear power plants as a viable and safetechnology that ensures future energy supply.

Affect evoked by nuclear power may also be an importantdeterminant of perceived benet and perceived risk (Finucaneet al., 2000). Some researchers have focused only on risk percep-tion associated with the acceptance of nuclear power, and not onbenet perception (Whiteld et al., 2009). However, it seems thatpossible economic benets are the main driver for proponents ofthe technology. Opponents of the technology value the economicbenets of this technology much less, and are, therefore, notwilling to accept the risks associated with it (Eiser and van derPligt, 1979). In a study conducted in Switzerland, it was foundthat benet perception (i.e., secure energy supply) is the mostimportant predictor for acceptance of nuclear power plants, andthat risk perception is a less important predictor (Visschers et al.,2011). Even though nuclear power does not emit CO2 whenproducing electricity, climate change concerns have only hadlimited impact on acceptance of nuclear power plants (Corneret al., 2011; Visschers et al., 2011). If nuclear power was framed asa contribution to climate change mitigation, acceptance of thetechnology increased a bit, but even then the impact was small(Pidgeon et al., 2008).

Peoples acceptance of nuclear power may be inuenced by thealternatives that are available. In one study, it was shown that USresidents wanted greater reliance on solar, wind, and hydro-electric energy sources (Greenberg, 2009). About half of therespondents were in favor of a decrease in electricity producedby nuclear power, and the other half of the respondents were infavor of the status quo or of an increase of nuclear powerproduction. However, it seems that people evaluate nuclearpower more positively when they receive some informationabout the technology, and when they have to rank various energyscenarios, and, therefore, have to make some trade-offs(Fleishman et al., 2010).

1.2. Studies examining attitudes before and after a nuclear accident

Immediately after the accident in Chernobyl, people had morenegative attitudes towards nuclear power, and they perceivedmore risks (Renn, 1990). The more a country or region wasaffected by fallout caused by the accident, the stronger was the

attitudinal change (Drottz-Sjoberg and Sjoberg, 1990; Renn,1990). Furthermore, results suggest that the accident had a shortterm effect (i.e., some time after the event, opposition towardsnuclear power decreased again) (Renn, 1990; Verplanken, 1989).It should be noted, however, that in one study the Chernobylaccident actually decreased the perception of threat (Lindell andPerry, 1990). However, this study was conducted in the US;respondents were, therefore, distant from the place of theaccident.

Most of the studies that examined the impact of a nuclearaccident on peoples attitudes and risk perceptions utilized cross-sectional data. There are only a few longitudinal studies thatallow examining of the stability of attitudes within individuals(Eiser et al., 1989; Lindell and Perry, 1990; Midden andVerplanken, 1990; Verplanken, 1989). A longitudinal study con-ducted in the UK showed that, after the Chernobyl accident,people showed greater opposition to existing nuclear powerplants and to new power plants in the UK compared with beforethe accident (Eiser et al., 1989). Regarding non-nuclear industrialdevelopments, like new chemical plants or oil wells, no changesin attitudes could be observed. The authors focused on meandifferences between the two measurement points, but they didnot report correlations across the measurement points.

Midden and Verplanken (1990) compared within-subject ana-lysis and between-subject analysis of attitudes towards nuclearpower, using longitudinal data, after the Chernobyl accident.Based on the results of this comparison, the authors concludedthat attitudes towards nuclear power are much less stable overtime than one would expect based on across-subject analysis.Longitudinal data are, therefore, needed for a better understand-ing of psychological processes related to the perception ofhazardous technologies.

The stability of the attitudinal measures before and after anuclear power accident can either be investigated by analyzingthe mean values or the correlations. Based on past research, itseems clear that nuclear accidents result in more negativeattitudes towards the technology (Eiser et al., 1989; Verplanken,1989). Even though accidents have changed the means of themeasured constructs, there can still be high correlations betweenthe pre- and the post-measures. Such a result would indicate thatall participants, independent from their perception of nuclearpower before the accident, have similarly changed their attitudesor perceptions. Only one study reports correlations between pre-Chernobyl and post-Chernobyl measures (Lindell and Perry,1990). For 10 of the 18 variables they investigated, signicanttestretest correlations were observed; all of the correlationcoefcients were below 0.5. These results suggest, therefore, thatattitudes to and perceptions of nuclear power measured bothbefore and after an accident are only moderately correlated.

1.3. Changes in attitudes towards nuclear power since previous

nuclear accidents

Since the nuclear accident in Chernobyl in 1986, the publicsperception of nuclear power seems to have changed. Polls in theUSA about the acceptance of nuclear power show that attitudestowards nuclear power have become more positive since the late1980s (Bolsen and Cook, 2008). Lately, a similar increase inacceptance of nuclear power was found in Europe (EuropeanCommission, 2010). Rising oil prices and more concern aboutclimate change have been suggested as having caused thisincreased support for nuclear power. These factors were part ofa so-called nuclear renaissance, i.e., a revival of nuclear power(Teravainen et al., 2011; World Nuclear Association, 2011). Thebenets of nuclear power for the climate appeared only slightlyrelated to the publics acceptance of this technology (Corner et al.,

2011), especially compared to the relation between perceived

Fukushima results in a decreased acceptance of nuclear power. At

M. Siegrist, V.H.M. Visschers / Energy Policy 59 (2013) 112119114economic benets and acceptance of nuclear power (Visscherset al., 2011). In other words, people accepted nuclear powerreluctantly when it was framed as a strategy to reduce climatechange (Corner et al., 2011). Another factor that inuenced thenuclear renaissance was a change in the political climate: nuclearpower was no longer associated with nuclear weaponsthe ColdWar was over (Rosa and Dunlap, 1994)but with energy (Kelleret al., 2012).

To summarize, the previous nuclear accidents increased thepublics opposition towards nuclear power. This increase inopposition directly after such an event was not very dramatic(Eiser et al., 1989). Since the nuclear accident in Chernobyl, thepublics opinion about nuclear power has become more positivein the US and in Europe. The recent nuclear accident atFukushima might therefore have had a smaller effect on thepublics opinion than did those accidents in the 1970s and 1980s.Moreover, the positive image of nuclear power resulting from therecent nuclear renaissance may have resulted in attitudestowards nuclear power that are more stable since the accident.The stability of attitudes regarding a nuclear accident has hardlybeen investigated.

1.4. Heuristics and biases inuencing the impact of Fukushima

How serious people interpreted the Chernobyl accident to bewas strongly related to peoples attitudes towards other nuclearrelated topics (Eiser et al., 1990). People may have used the affectassociated with nuclear power for interpreting the accident. Sucha result is in line with the affect heuristic (Finucane et al., 2000).The affect heuristic may also help to predict peoples responses tothe Fukushima accident. Immediately after the accident, evenproponents of nuclear power may have had more negative affectrelated to nuclear power; as a result, an energy mix with nuclearpower may have been evaluated less positively. One wouldexpect, however, that after some time without permanent mediacoverage, proponents perceive less negative affect, and, conse-quently, energy mixes with nuclear power are more positivelyassessed again.

1.5. Nuclear power in Switzerland

In Switzerland, electricity is mainly produced by hydroelectricpower stations (55%) and nuclear power stations (40%) (SwissFederal Ofce of Energy, 2010). Five nuclear power reactorsare operating in Switzerland, located at four different sites inthe German-speaking part of the country: Gosgen, Beznau (tworeactors), Muhleberg, and Leibstadt. The rst nuclear powerstation, Beznau I, was put into operation in 1969, and the lastone, at Leibstadt, was put into operation in 1984 (Swiss FederalOfce of Energy, 2009). Before the accident in Fukushima, it wasplanned to replace three nuclear power reactors because theywere to reach the end of their running time by 2025. Threeapplications to build new nuclear power plants had been sub-mitted to the Swiss Federal Ofce for Energy. After the accident inFukushima, the Swiss Government decided to back out of thenuclear energy program (March 14, 2011). However, it remainsunclear how the electricity produced by nuclear power plants willbe substituted. It is also unclear as to when the nuclear powerplants will cease operations.

Before the accident in Fukushima, the acceptance of nuclearpower among the Swiss has been comparable to that of otherEuropean populations (European Commission, 2008). A largepart of the respondents (43%) of a representative sample of theSwiss population appeared to oppose nuclear power in 2007

(Stauffacher et al., 2008). Acceptance of nuclear power is lowerleast immediately after an accident, such a decrease can beexpected. It is an open question, however, as to how stronglyacceptance of nuclear power before and after the accident inFukushima is correlated. Studying public acceptance of nuclearpower is considered to be crucial for making sound policydecisions. However, there is a lack of longitudinal studies thatexamine the stability of lay peoples attitudes. Only when atti-tudes and perceptions are stable over time can preferencespronounced by the public be taken into account by thegovernment.

Another goal of the current study was to examine peoplesacceptance of different proportions of nuclear power in the totalelectricity mix, before and after the nuclear accident. Nuclearpower is an important source of electricity for the currentelectricity mix in Switzerland. Participants were asked to evaluatefour scenarios with different percentages of nuclear power plant(NPP) production. We hypothesized that people who are stronglyopposed towards nuclear power would favor an electricity mixwith 0%, but would be relatively non-sensitive to changes in thepercentage of nuclear power, when it is larger than 0%. Such aresult would be in line with the work about protected values(Baron and Spranca, 1997) that suggests that people may not besensitive to variation in quantity when sacred values are at stake.Protected values pose serious problems for costbenet analysis,for example, because they imply that certain values cannot betraded-off.

Additionally, it was expected that the Fukushima accidentwould not have a substantial effect on the acceptance of thevarious scenarios with different percentages of electricity pro-duced by nuclear power for the opponents of nuclear power. Theopponents already had negative attitudes towards an energy mixwith nuclear power before Fukushima. Thus, there was not muchroom for changes in this group. However, for proponents ofnuclear power, it was expected that, immediately after Fukush-ima, a more negative evaluation of the current situation inSwitzerland could be observed. Furthermore, we assumed thatsuch an effect could only be observed immediately after theaccident in this group, and that, in the third survey (6 monthsafter the Fukushima accident), results similar to those in the rstsurvey could be observed.

2. Method

2.1. Procedure and sample

Data were collected in three waves. The rst survey wasconducted in autumn 2010. A questionnaire together with astamped and addressed return envelope were sent to a randomsample of households in the German-speaking part of Switzer-land. Addresses were randomly selected from the telephonedirectory. After two reminders, a response rate of 40% wasachieved (N1232). About 2 weeks after the accident in thein Switzerland than in other European countries that use nuclearpower plants for the production of electricity.

1.6. Aims of the present study

One goal of the present study was to examine the stability ofthe acceptance of nuclear power. We should emphasize, however,that the rst wave of the present study was initially planned as across-sectional study. Only after the accident in Fukushima,we decided to conduct a longitudinal study. Based on pastresearch, it seems to be clear that an accident like the one inFukushima Daiichi nuclear power plant in Japan (March, 2011),

questionnaire should also ll out the new questionnaire. Cases in

before the Fukushima accident (M4.56, SD1.79) than respon-

M. Siegrist, V.H.M. Visschers / Energy Policy 59 (2013) 112119 115which respondents birth year and gender were not identical forall three surveys were deleted. The nal dataset included 463respondents.

Based on data from the rst two waves, a model explainingacceptance of nuclear power was tested (Visschers and Siegrist, inpress). In the present paper, not only data from those two, butalso from a third wave are analyzed. In addition, the focus is ondifferent research questions and also on different variables.

2.2. Questionnaire

For the rst survey, a comprehensive questionnaire aboutpeoples perceptions of nuclear power in Switzerland was used.For the second and third surveys, much shorter questionnaireswere used. Some of the questions were identical, whereas otherswere different in each of the three questionnaires. In the presentpaper, the focus is on the questions that were identical across thethree surveys.

Acceptance of nuclear power was measured in all threesurveys utilizing the following three items: Switzerland needsa lot of electricity; people should therefore accept nuclear powerstations, Switzerland can renounce nuclear power stationswithout any problems (this item was recoded for analyses),and We need nuclear power stations in Switzerland becauserenewable energy sources alone do not produce sufcient elec-tricity. All three items were rated on a 7-point Likert scale. Onlythe two extreme points of the scales were verbally labeled (1donot agree at all; 7agree completely).

Participants evaluated four scenarios with various proportionsof nuclear power production for Switzerland in all three surveys.The following information was provided: In Switzerland 40% ofthe electricity is produced by NPPs at the moment. Assume that inthe following percentages of electricity produced by nuclearpower, possible electricity gaps are covered by imports. Howdo you assess overall the following proportions of nuclear powerin energy production? Participants then answered four state-ments. The rst statement read: In Switzerland 0% of theelectricity will be produced by NPP. The same statement wasused for the remaining questions about the percentages 20%, 40%,and 60%. Participants could rate each proportion of nuclear poweron a 7-point scale. The extreme response categories (3verynegative; 3very positive) and the middle response category(0neutral) were verbally labeled. Last, respondents were askedto ll out their gender, birth year, and education level.

3. Results

3.1. Respondents

The mean age of the nal sample of 463 respondents was58 yrs (SD14); 69% (n318) of the respondents were male.Most respondents had nished vocational school (n182, 39%) orcollege or university (n159, 34%). About 19% of the sample hadcompleted higher secondary school (n90), and 6% had nisheda shorter questionnaire was sent to the persons, who hadparticipated in the rst study. This second survey had a responserate of 77% (N929). The respondents of survey 2 who hadindicated to be willing to participate once more, received a thirdsurvey 6 months after the nuclear accident, in October, 2011. Inthe third survey, 570 questionnaires were returned (a responserate of 86%).

In the instructions of the second and third questionnaires, theauthors emphasized that the person who has answered the rstthe countrys obligatory school levels (n28). Four respondentsdents (M4.26, SD1.97), t(1111)2.24, p0.03, r0.07.Men were more likely (63.0%) to participate in all three

surveys than to dropout after the rst two surveys (37.0%),whereas 55.3% of the women completed all three surveys, and44.7% dropped out, (w2(1)4.19, p0.04). Non-respondents ofsurvey 3 were just as old (M56.24, SD15.67) as respondents ofall three surveys (M57.90, SD14.45), t(765)1.50,p0.13. Respondents of all three surveys again had a somewhathigher education level (19.4% with a higher secondary schooleducation and 34.3% with a college- or university-level education)than non-respondents (17.4% with a higher secondary schooleducation and 23.7% with a college- or university-level educa-tion), U57,751.00, p0.0001. Non-respondents (M3.58,SD1.76) and respondents (M3.52, SD2.01) had similarlevels of acceptance directly after the Fukushima accident,t(765)0.48, p0.63.

3.2. Changes in acceptance scores

The three items measuring acceptance of nuclear power werecombined into a scale. Cronbachs alpha for the rst surveywas 0.88, for the second survey the coefcient was 0.90, and forthe third survey a value of 0.90 was observed. The level ofacceptance of nuclear power was signicantly higher before theaccident in Fukushima (M4.15, SD2.02) compared with afterthe accident (M3.50, SD2.00), t(456)10.38, po0.001. Thelevel of acceptance in the second survey was not signicantlydifferent from that of the third survey (M3.51, SD1.99). Thus,results suggest that after the accident in Fukushima the respon-dents in Switzerland showed a lower level of acceptance. Thus,after the accident the acceptance remained on a relativelystable level.

The acceptance of nuclear power prior to Fukushima washighly correlated with the acceptance immediately after theaccident, r0.79, po0.001 (N457). The acceptance scale wasalso highly correlated between the second and the third surveys,r0.86, po0.001 (N459). Results of the correlation coefcientssuggest that participants had relatively stable attitudes towardsnuclear power, even when there was a general shift towardssomewhat more negative attitudes after the accident indid not report their education level. The nal sample was older,included more men, more people with a college or universitydegree, and fewer people who had only nished obligatory schoolthan was representative of the average Swiss population (SwissStatistics, 2009, 2010).

We compared the respondents who participated in surveys1 and 2 to those who only completed survey 1 (i.e., the non-respondents of the second survey, n304). Additionally, wecompared the participants who completed all three surveys tothose who dropped out after survey 2 (i.e., the non-respondents ofthe third survey, n358, of which 304 cases could be matched ongender and birth year for surveys 1 and 2).

Non-respondents of survey 2 had a gender distribution similarto that of respondents of both surveys (w2(1)0.45, p0.50).Non-respondents of survey 2 were slightly younger (M54.17,SD16.37) than respondents of surveys 1 and 2 (M57.41,SD15.03), t(1125)3.10, po0.01, r0.09. Respondents ofboth surveys had a somewhat higher education level (18.4% witha higher secondary school education and 30.5% with a college- oruniversity-level education) than non-respondents (13.2% with ahigher secondary school education and 25.3% with a college- oruniversity-level education), U106,294.50, p0.01. Additionally,non-respondents reported greater acceptance of nuclear powerFukushima.

nd

ill

0

00.71 1.53

0.34] [0.89; 0.53] [1.72; 1.35]

ry positive).

Fig. 1. The impact of acceptance of nuclear power on the evaluation of fournuclear power proportion scenarios.

M. Siegrist, V.H.M. Visschers / Energy Policy 59 (2013) 1121191163.3. How much of the electricity production should be covered by

NPPs?

Participants indicated how positively or negatively they eval-uated various scenarios about nuclear electricity production. Inthe four scenarios, the percentages covered by nuclear powerwere 0%, 20%, 40%, and 60%. Table 1 shows how participantsanswered these questions in the three surveys. Results suggestthat, before Fukushima, the 0% and 40% scenarios were lessnegatively evaluated compared with the 20% and 60% scenarios.In other words, participants preferred either no nuclear power orthe status quo (40%). After Fukushima, the less nuclear energy ascenario involved the less negatively it was evaluated.

We further examined how general acceptance of nuclearpower was related to the evaluation of the scenarios. Based onthe acceptance scale of the rst survey, participants were groupedinto the following four quartiles: low acceptance (26.1%, n120),somewhat low acceptance (23.0%, n106), somewhat high accep-tance (24.6%, n113), and high acceptance (26.3%, n121). Howthe participants of these four groups evaluated the four scenariosin the rst survey is shown in Fig. 1. Results of 4 (acceptance: 1st,2nd, 3rd, and 4th quartiles) 4 (scenarios: 0%, 20%, 40%, and 60%)split-plot ANOVA with scenario as a repeated measurement factorwas conducted on the data assessed before the Fukushima accident.The main effect for acceptance (F(3,439)7.47, po0.001), the maineffect for scenario (F(3,1317)30.79, po .001) and the interactionacceptance x scenario (F(9,1317)116.89, po .001), were signi-cant. For the low acceptance group, the largest change was observedbetween the 0% and the 20% scenarios. The most important fact forthis group was that no electricity is produced by nuclear power.Whether a low or a high percentage is produced by nuclear power isless important for their evaluation of the scenarios. This result is inline with the hypothesis that protected values are important forpeople against nuclear power. The somewhat low acceptance groupshowed a linear evaluation with almost a constant differencebetween all four scenarios. The more energy produced by nuclearpower plants, the more negative was the assessment of the scenario.

Table 1The evaluation of four nuclear power proportion scenarios before a

In Switzerland X% of the electricity w

0% 20%

1st survey 0.03 0.52(N449454) [0.26; 0.19] [0.68;2nd survey 0.93 0.26(N441450) [0.71; 1.15] [0.43;3rd survey 0.74 0.52(N428443) [0.50; 0.98] [0.69;

Note: Response categories ranged from 3 (very negative) to 3 (veThe status quo seems to strongly inuence the evaluation of thesomewhat high acceptance group and the high acceptance group.Both groups evaluated the 40% scenario as the most positive one.

We further examined how the accident in Fukushima inu-enced the evaluation of the four scenarios with different nuclearpower proportions within each acceptance group. For the lowacceptance group, the Fukushima accident did not have a sig-nicant impact on the evaluation of the four scenarios (see Fig. 2).A 3 (time: before, directly after, and half a year after Fukushima) 4 (scenarios: 0%, 20%, 40%, and 60%) ANOVA with two repeatedmeasurement factors indicated a signicant main effect forscenario (F(3,309)404.80, po0.001). Neither the main effectfor time (F(2,206)1.71, ns) nor the interaction of time scenario (F(6,618)1.05, ns) was signicant. Not surprisingly,the Fukushima accident did not have a signicant impact on theafter Fukushima.

be produced by NPP

40% 60%

0.24 0.96.36] [0.41; 0.08] [1.15; 0.77]

0.68 1.62.09] [0.86; 0.49] [1.80; 1.43]evaluation of the scenarios in the low acceptance group becauseprior to the accident, the participants of this group had alreadyevaluated the nuclear power proportions as very negative. Thus,there was not much room for this group to change its attitudes.Based on the protected value hypothesis, we expected a non-linear effect for the four scenarios. An additional analysis utilizingGLM (generalized linear model) revealed that for all threemeasurement waves, not only was the linear effect signicantbut also were the quadratic trend (F(1,103)459.2, po0.001) andthe cubic trend (F(1,103)48.8, po0.001).

Fig. 2. Low acceptance group (4th quartile): evaluation of four nuclear powerproportion scenarios.

high acceptance group, the 60% scenario was evaluated clearlyless positively both directly after and 6 months after compared to

power proportion scenarios.

Fig. 5. High acceptance group (1st quartile): evaluation of four nuclear powerproportion scenarios.

M. Siegrist, V.H.M. Visschers / Energy Policy 59 (2013) 112119 117For the somewhat low group (2nd quartile), a 3 (time: before,directly after, and half a year after Fukushima) 4 (scenarios; 0%,20%, 40%, and 60%) ANOVA with two repeated measurementfactors indicated a signicant main effect for scenario (F(3,261)136.70, po0.001) and a signicant interaction time scenario(F(6,522)12.42, po0.001). The main effect for time was notsignicant (F(2,174)2.22, ns). The somewhat low acceptancegroup evaluated the 0% scenario more positively after the acci-dent and the other scenarios were evaluated in a more negativeway compared to before the accident. The evaluations of the 20%,40%, and 60% scenarios observed in the third survey were closerto the evaluations observed in the second survey compared tothose in the rst survey. Thus, the accident in Fukushima stillseems to have inuenced responses in the third survey. However,it is remarkable how little inuence the Fukushima accident hadon the evaluation of this group, considering the 7-point scale, therespondents could have decreased their evaluation much more. Inall three surveys, this group showed almost a linear evaluation ofthe four scenarios. The more power is produced by nuclear powerplants, the more negative is the evaluation of the scenario (Fig. 3).

For the somewhat high acceptance group (3rd quartile) a 3(time: before, directly after, and half a year after Fukushima) 4(scenarios; 0%, 20%, 40%, and 60%) ANOVA with two repeated

Fig. 3. The somewhat low acceptance group (3rd quartile): evaluation of fournuclear power proportion scenarios.measurement factors indicated a signicant main effect forscenario, F(3,303)24.77, po0.001, and a signicant interactiontime scenario (F(6,606)18.56, po0.001). The main effect fortime was not signicant (F(2,202)1.59, ns). The means areshown in Fig. 4. Before the accident in Fukushima, the statusquo (40%) was the most positively evaluated scenario. AfterFukushima, there was a linear effect similar to that of the loweracceptance group (2nd quartile). However, the slope seemedmuch less steep. Immediately after Fukushima, this group shiftedtowards a more positive evaluation of the 0% and 20% scenarios.Half a year after Fukushima, these two scenarios were lesspositive evaluated than they were directly after the accident.

For the high acceptance group (4th quartile), the main effectsfor time (F(2,196)3.61, p0.03) and scenario (F(3,294)54.55,po0.001) were signicant, as was the interaction effect time scenario (F(6,588)10.73, po0.001). It is remarkable that thisgroup evaluated the 40% scenario in each survey as the mostpositive. Immediately after the accident in Fukushima, the sce-narios with low percentage of nuclear power (i.e., 0% and 20%nuclear power) were less negatively evaluated than before. Half ayear after Fukushima, a shift towards a more negative evaluationof these two scenarios could be observed again. However, for theFig. 4. Somewhat high acceptance group (2nd quartile): evaluation of four nuclearbefore Fukushima (Fig. 5).

Discussion

Longitudinal studies are scarce in the eld of risk perceptionand acceptance of technologies. As a result, it remains unknownhow stable peoples attitudes towards technologies are, and howstrongly they are inuenced by technological accidents. It hasbeen suggested that people construct their preferences whenanswering questions in a survey (Slovic, 1995). Contextual infor-mation could, therefore, have a large impact on peoplesresponses. It could even be that attitudes before and after anaccident are not correlated, because peoples responses aremainly inuenced by the salient information. The accident inFukushima provided, therefore, an excellent opportunity to llthis research gap, and to examine what impact a real accident hason peoples acceptance of nuclear power.

The accident in Fukushima had a negative impact on accep-tance of nuclear power in Switzerland. It should be noted,however, that the mean changes were moderate, given the broadcoverage of the accident in the media, and that the Swiss

M. Siegrist, V.H.M. Visschers / Energy Policy 59 (2013) 112119118government decided not to replace the current reactors and tophase out of nuclear energy. The results are in line with pastresearch (Eiser et al., 1989; Verplanken, 1989), which, similarly,showed a moderate change towards less acceptance and morenegative attitudes towards nuclear energy directly after theaccident in Chernobyl. It seems that people mainly base theirinterpretation of a technological accident on their prior beliefs(Siegrist et al., 2008) and perceptions of related risks (Visscherset al., 2007). As a result, attitudinal changes are moderate and theaccident may be interpreted in a way that existing beliefs andconvictions are not changed. The present study shows that suchan effect cannot only be observed in studies utilizing hypotheticalscenarios, but also in the case of real accidents. Our ndings arealso in line with the renowned concept of latitude of acceptance,which suggests that information has a stronger impact when itfalls within peoples scope of acceptance, compared with its effecton people for whom the information is outside their scope ofacceptance (Sherif et al., 1965).

The stability of peoples acceptance can be (and has so farbeen) tested focusing on mean differences. In addition to meandifferences, associations between measures across measurementoccasions are of interest. The relative stability of persons atti-tudes can be measured by correlations. In the present study, highcorrelations of the acceptance scores between the three waveswere observed. The high correlation coefcient between attitudesbefore and after Fukushima suggests that participants had arelatively similar position on the acceptance continuum beforeand after Fukushima. The mean values have changed, but apersons acceptance of nuclear power after the accident can berelatively well predicted based on the persons acceptance prior tothe accident. This result further supports the idea that the priorbeliefs and attitudes are crucial for the interpretation of a newevent (Visschers et al., 2007). In the present study, highercorrelations compared to prior studies were observed (Lindelland Perry, 1990). The latter study examined correlations betweensingle items, whereas in the present study a scale was used. Singleitems often have a lower reliability compared with scales, and,therefore, lower correlations are observed.

The analyses of the present study are based on a longitudinalstudy design with three measurement points. This is a bigadvantage compared with past studies, because we had thebenet of being able to follow changes in the acceptance ofnuclear power within the same sample of people. The secondsurvey took place immediately after the accident in Fukushima.People may not have been able to form a stable opinion so quicklyafter the event. Moreover, the situation at the power plants inFukushima was not stable at this point of time. Based on theevaluation of the four nuclear power proportion scenarios, we canconclude that this uncertainty has resulted in more negative andinstable attitudes by the proponents of nuclear energy. Thescenarios with nuclear power were evaluated more positivelyhalf a year after the accident compared with immediately afterthe accident. However, responses to the acceptance questions hadnot recovered 6 months after the accident. It seems plausible,however, that nuclear power will regain the previous acceptancelevels, as has been the case after the previous accidents.

People most strongly opposed to nuclear power seem to haveprotected values (Baron and Spranca, 1997). The energy mixscenario with 0% nuclear power was very positively assessed bythe low acceptance group, of course. However, there was notmuch differentiation between the three scenarios with nuclearpower among these opponents. This insensitivity to changes inquantity is a clear sign for protected values. The existence of suchprotected values makes costbenet analysis almost impossible,because these absolute values are so strongly weighted that

benet cannot compensate for the risks associated with thetechnology. However, it remains unknown which values causedthe negative evaluations of the scenarios that included nuclearpower from the results of the present study. This should beexamined in a future study. The proponents of nuclear powershowed a status quo bias (Samuelson and Zeckhauser, 1988);their decisions were most likely anchored to the current level ofenergy produced by nuclear power plants. This perseverance onthe status quo could be observed in all three surveys amongthis group.

Some limitations of the present study need to be addressed.The geographical location of a nuclear power plant accident iscertainly important. After the accident in Chernobyl, relativelyhigh levels of radioactive fallout were measured in variousEuropean countries. As a result, people changed aspects of theirlives, such as their diets (Peters et al., 1990). The accident inFukushima was much more distant from Europe, and the Eur-opean public did not suffer directly from the accident. Had theaccident been geographically closer to Switzerland, differenteffects might have been observed. Also, the cause of a givennuclear accident could have an impact on peoples perception. Forinstance, the event in Fukushima was precipitated by a tsunami.People in Switzerland may have thought that such an accidentcould not happen in Switzerland because a tsunami is an unlikelyproblem for nuclear power plants in Switzerland.

Additionally, the Swiss political situation is different from thatof other countries, and this may have inuenced our results. Threeof the ve nuclear power reactors in Switzerland ought to bereplaced by 2025 in order to maintain the current level ofelectricity production. Not only does the decision to rebuild thesereactors have to be approved by the executive government andthe legislative government, it is also very likely that the Swisspopulation will have to vote on in a national facultative refer-endum. This direct public participation process may result in abinding decision, which the government has to implement. Swisspeoples interest in the topic of nuclear power may have beeninuenced by past referenda about this issue.

The fact that 40% of the electricity in Switzerland is producedby nuclear power may have inuenced participants responses. Itseems likely that this information resulted in an anchoring effect(Plous, 1989). Particularly, respondents with somewhat high orhigh acceptance of nuclear power may have evaluated the 40%scenario as being more positive than they would have withoutthis information.

The longitudinal design is a benet to the present study,despite the dropout of respondents in each wave. Our analysesshowed that the composition of the sample of survey 2 wassomewhat affected by drop-outs and, to a much lesser extent, thatof survey 3. Participants who dropped out after the rst surveyhad, on average, a higher level of acceptance of nuclear powerprior to Fukushima compared with people who participated in allthree surveys. As a result of this non-random dropout, we mayhave underestimated peoples acceptance of nuclear power in ourstudy. Moreover, the length of our longitudinal survey onlyaffected the size of our sample and only slightly affected thecomposition of our sample. We therefore believe that dropoutsare unproblematic in longitudinal surveys as long as the samplecomposition remains stable and relatively comparable to thepopulation and if the sample size remains large enough forthe complex analyses that can be performed in longitudinalstudies.

In the present study, we have focused on how the size of thecontribution of nuclear energy to the energy mix is evaluated.It has been emphasized that a nation should develop andcommunicate a comprehensive energy strategy (Greenberg andTruelove, 2011). Without understanding the energy mix of a

society, lay people may show inconsistent preferences. The public

Baron, J., Spranca, M., 1997. Protected values. Organizational Behavior and Human

Goodfellow, M.J., Williams, H.R., Azapagic, A., 2011. Nuclear renaissance, publicperception and design criteria: an exploratory review. Energy Policy 39,

M. Siegrist, V.H.M. Visschers / Energy Policy 59 (2013) 112119 119Decission Processes 70, 116.Bolsen, T., Cook, F.L., 2008. Public opinion on energy policy: 19742006. Public

Opinion Quarterly 72, 364388.Corner, A., Venables, D., Spence, A., Poortinga, W., Demski, C., Pidgeon, N., 2011.

Nuclear power, climate change and energy security: exploring British publicattitudes. Energy Policy 39, 48234833.

Dohle, S., Keller, C., Siegrist, M., 2010. Examining the relationship between affectand implicit associations: implications for risk perception. Risk Analysis 30,11161128.

Drottz-Sjoberg, B.-M., Sjoberg, L., 1990. Risk perception and worries after theChernobyl accident. Journal of Environmental Psychology 10, 135149.

Eiser, J.R., Hannover, B., Mann, L., Morin, M., van der Pligt, J., Webley, P., 1990.Nuclear attitudes after Chernobyl: a cross-national study. Journal of Environ-mental Psychology 10, 101110.

Eiser, J.R., Spears, R., Webley, P., 1989. Nuclear attitudes before and afterChernobyl: change and judgment. Journal of Applied Social Psychology 19,689700.

European Commission, 2008. Attitudes towards Radioactive Waste. TNS Opinion &Social, Brussels.

European Commission, 2010. Europeans and Nuclear Safety. TNS Opinion & Social,Brussels.

Finucane, M.L., Alhakami, A., Slovic, P., Johnson, S.M., 2000. The affect heuristic injudgments of risks and benets. Journal of Behavioral Decision Making 13,117.

Fischhoff, B., Slovic, P., Lichtenstein, S., Read, S., Combs, B., 1978. How safe is safeenough? A psychometric study of attitudes towards technological risks andbenets. Policy Sciences 9, 127152.

Fleishman, L.A., Bruine De Bruin, W., Morgan, M.G., 2010. Informed publicpreferences for electricity portfolios with CCS and other low-carbon technol-ogies. Risk Analysis 30, 13991410.against additional emission of CO2. Forcing people to make suchtrade-offs may change their responses (Fleishman et al., 2010). Alimitation of the present study is, therefore, that only thecontribution of nuclear power to the total energy mix was varied,not explicitly stating how the same level of energy could beprovided without nuclear power. The reason for this is the factthat public discussions in Switzerland focus more on the use ornon-use of nuclear power than on how the electricity that is todayproduced by nuclear power should be substituted in the future.

In various countries, politicians reacted in very different waysto the accident in Japan. In Germany and Switzerland, politiciansdecided to phase out nuclear energy, whereas in France and theUSA, politicians did not abandon their support of nuclear power.In the present study, respondents did not drastically change theiracceptance of nuclear power or their evaluation of the Swissenergy mix. It is unknown, therefore, whether the Swiss publicwill support the phasing out of nuclear power at any price in thefuture. A discussion about the trade-offs associated with variousenergy mixes may be a better approach to designing an accep-table energy mix for Swiss society than just banishing someelectricity sources and subsidizing others. It remains a challenge,of course, whetherand, if so, howprotected values should beconsidered when designing an energy mix.

Acknowledgments

The rst survey reported in this paper (conducted in 2010) wasfunded by Swissnuclear. Swissnuclear comprises representativesof the Swiss electric supply companies (Alpiq, Axpo, BKW, CKW,and EGL), which are committed to the safe and economic opera-tion of the nuclear power plants in Switzerland.

Referencesanalysis of US national and site-specic data. Energy Policy 37, 32423249.Greenberg, M., Truelove, H.B., 2011. Energy choices and risk beliefs: is it just global

warming and fear of a nuclear power plant accident? Risk Analysis 31,

819831.Keller, C., Visschers, V., Siegrist, M., 2012. Affective imagery and acceptance of

replacing nuclear power plants. Risk Analysis 32, 464477.Lindell, M.K., Perry, R.W., 1990. Effects of the Chernobyl accident on public

perceptions of nuclear plant accident risks. Risk Analysis 10, 393399.Midden, C.J.H., Verplanken, B., 1990. The stability of nuclear attitudes after Chernobyl.

Journal of Environmental Psychology 10, 111119.Peters, H.P., Albrecht, G., Hennen, L., Stegelmann, H.U., 1990. Chernobyl and the

nuclear power issue in West-German public opinion. Journal of Environmental

Psychology 10, 121134.Pidgeon, N.F., Lorenzoni, I., Poortinga, W., 2008. Climate change or nuclear power

no thanks! A quantitative study of public perceptions and risk framing inBritain. Global Environmental Change 18, 6985.

Plous, S., 1989. Thinking the unthinkable: the effects of anchoring on likelihoodestimates of nuclear war. Journal of Applied Social Psychology 19, 6791.

Renn, O., 1990. Public responses to the Chernobyl accident. Journal of Environ-mental Psychology 10, 151167.

Rosa, E.A., Dunlap, R.E., 1994. Nuclear power: three decades of public opinion.Public Opinion Quarterly 58, 295324.

Samuelson, W., Zeckhauser, R., 1988. Status quo bias in decision making. Journal ofRisk and Uncertainty 1, 759.

Sherif, C.W., Sherif, M., Nebergall, R.E., 1965. Attitude and attitude change: the

social judgment-involvement approach. Greenwood Press, Westport, CT.Siegrist, M., Cousin, M.-E., Frei, M., 2008. Biased condence in risk assessment

studies. Human and Ecological Risk Assessment 14, 12261234.Siegrist, M., Keller, C., Cousin, M.-E., 2006. Implicit attitudes toward nuclear power

and mobile phone base stations: support for the affect heuristic. Risk Analysis26, 10211029.

Siegrist, M., Keller, C., Kiers, H.A.L., 2005. A new look at the psychometric paradigmof perception of hazards. Risk Analysis 25, 211222.

Slovic, P., 1987. Perception of risk. Science 236, 280285.Slovic, P., 1995. The construction of preference. American Psychologist 50, 364371.Slovic, P., Finucane, M.L., Peters, E., MacGregor, D.G., 2004. Risk as analysis and risk

as feelings: some thoughts about affect, reason, risk, and rationality. Risk

Analysis 24, 311322.Stauffacher, M., Krutli, P., Scholz, R.W., 2008. Gesellschaft und radioaktive Abfalle:

Ergebnisse einer schweizweiten Befragung [Society and radioactive waste:results of a survey in Switzerland]. Ruegger, Zurich, Chur.

Swiss Federal Ofce of Energy, 2009. Nuclear Energy. /http://www.bfe.admin.ch/themen/00511/ index.html?lang=enS. Accessed July 22, 2010.

Swiss Federal Ofce of Energy, 2010. Graken der Elektrizitatsstatistik 2009

[Graphs of Electricity Statistics 2009]. /http://www.bfe.admin.ch/themen/00526/00541/00542/00630/index.html?lang=de&dossier_id=00768S.Accessed January 13, 2011.

Swiss Statistics, 2009. Bevolkerungsstand und -struktur [Population size andcomposition]. /http:// www.bfs.admin.ch/bfs/portal/de/index/themen/01/02/blank/data/01.htmlS. Accessed May 31, 2011.

Swiss Statistics, 2010. Bildungsstand der Wohnbevolkerung nach Alter und

Geschlecht [Education level of the resident population according to age andgender]. /http://www.bfs.admin.ch/bfs/portal/de/index/themen/15/02/key/ind5.indicator.51131.511.html?open=9#9S. Accessed May 31, 2011.

Teravainen, T., Lehtonen, M., Martiskainen, M., 2011. Climate change, energy

security, and riskdebating nuclear new build in Finland, France and the UK.Energy Policy 39, 34343442.

Verplanken, B., 1989. Beliefs, attitudes, and intentions toward nuclear energy

before and after Chernobyl in a longitudinal within-subjects design. Environ-ment and Behavior 21, 371392.

Visschers, V.H.M., Keller, C., Siegrist, M., 2011. Climate change benets and energybenet supply benets as determinants of acceptance of nucler power

stations: investigating an explanatory model. Energy Policy 39, 36213629.Visschers, V.H.M., Meertens, R.M., Passchier, W.F., de Vries, N.K., 2007. How does

the general public evaluate risk information? The impact of associations with

other risks. Risk Analysis 27, 715727.Visschers, V.H.M., Siegrist, M. How an accident in a nuclear power plant inuences

acceptance: results of a longitudinal study before and after Fukushima. RiskAnalysis, http://dx.doi.org/10.1111/j.1539-6924.2012.01861.x, in press.

Whiteld, S.C., Rosa, E.A., Dan, A., Dietz, T., 2009. The future of nuclear power:value orientations and risk perception. Risk Analysis 29, 425437.

World Nuclear Association, 2011. The Nuclear Renaissance. /http://www.world-nuclear.org/info/inf104.htmlS. Accessed May 5, 2011.support gas power plants for the production of electricity, but be 61996210.Greenberg, M., 2009. Energy sources, public policy, and public preferences:may support expansion of wind and solar power, but may not bein favor of new transmission lines; alternatively, the public may

Acceptance of nuclear power: The Fukushima effectIntroductionLay peoples perception of nuclear powerStudies examining attitudes before and after a nuclear accidentChanges in attitudes towards nuclear power since previous nuclear accidentsHeuristics and biases influencing the impact of FukushimaNuclear power in SwitzerlandAims of the present study

MethodProcedure and sampleQuestionnaire

ResultsRespondentsChanges in acceptance scoresHow much of the electricity production should be covered by NPPs?

DiscussionAcknowledgmentsReferences