how a nuclear power plant accident influences acceptance of nuclear power: results of a longitudinal...

Risk Analysis DOI: 10.1111/j.1539-6924.2012.01861.x

How a Nuclear Power Plant Accident Influences Acceptanceof Nuclear Power: Results of a Longitudinal Study Beforeand After the Fukushima Disaster

Vivianne H. M. Visschers∗ and Michael Siegrist

Major nuclear accidents, such as the recent accident in Fukushima, Japan, have been shownto decrease the public’s acceptance of nuclear power. However, little is known about howa serious accident affects people’s acceptance of nuclear power and the determinants of ac-ceptance. We conducted a longitudinal study (N = 790) in Switzerland: one survey was donefive months before and one directly after the accident in Fukushima. We assessed accep-tance, perceived risks, perceived benefits, and trust related to nuclear power stations. In ourmodel, we assumed that both benefit and risk perceptions determine acceptance of nuclearpower. We further hypothesized that trust influences benefit and risk perceptions and thattrust before a disaster relates to trust after a disaster. Results showed that the acceptance andperceptions of nuclear power as well as its trust were more negative after the accident. Inour model, perceived benefits and risks determined the acceptance of nuclear power stationsboth before and after Fukushima. Trust had strong effects on perceived benefits and risks, atboth times. People’s trust before Fukushima strongly influenced their trust after the accident.In addition, perceived benefits before Fukushima correlated with perceived benefits after theaccident. Thus, the nuclear accident did not seem to have changed the relations between thedeterminants of acceptance. Even after a severe accident, the public may still consider thebenefits as relevant, and trust remains important for determining their risk and benefit per-ceptions. A discussion of the benefits of nuclear power seems most likely to affect the public’sacceptance of nuclear power, even after a nuclear accident.

KEY WORDS: Acceptance; Fukushima; longitudinal study; nuclear accident; social trust

1. INTRODUCTION

On March 11, 2011, an enormous earthquakeand its subsequent tsunami devastated the coolingsystems of the Fukushima Daiichi nuclear powerplant in Japan. As a result, vapor, cooling water, andexplosive particles with radioactive elements were

ETH Zurich, Institute for Environmental Decisions (IED), Con-sumer Behavior, Zurich, Switzerland.

∗Address correspondence to Vivianne Visschers, ETH Zurich, In-stitute for Environmental Decisions, Consumer Behavior, Uni-versitaetsstrasse 22 CHN J75.2, CH-8092 Zurich, Switzerland;tel: +41 44 6326149; fax: +41 44 6321029; [email protected].

released or leaked from three reactors into the envi-ronment. The authorities decided to evacuate peopleliving in a 30-km radius around the power plant, anda 20-km zone was declared a no-go area. The Inter-national Atomic Energy Association calculated thenuclear accident to be at level 7 on the InternationalNuclear and Radiological Event Scale (INES), thisbeing the highest level.(1) Soon after the accident, dis-cussions about the safety and future of nuclear powerplants began in many countries around the world.

Previous research shows that a major accidentat a nuclear reactor strongly affects laypeople’sattitudes toward nuclear power.(2−4) The accident in

1 0272-4332/12/0100-0001$22.00/1 C© 2012 Society for Risk Analysis

2 Visschers and Siegrist

Japan provided an opportunity to test additional hy-potheses that have not been addressed in previousresearch regarding the effect of a nuclear accidenton public perception. In a longitudinal study that in-cluded two waves, we examined how the accident atthe nuclear power plant in Fukushima changed peo-ple’s trust, their perceptions of benefits and risks, andtheir acceptance of nuclear power in the German-speaking part of Switzerland.

1.1. Public Perception of Nuclear Power Aftera Nuclear Accident

After the accidents at the nuclear power plantsin Chernobyl in 1986 and on Three Mile Island(TMI) in 1979, a large number of studies investi-gated people’s attitudes toward nuclear power. Somestudies compared laypeople’s opinions of nuclearpower before and after the disasters by comparingthe findings of different samples both over time(5−7)

and between countries.(8,9) In addition, a few lon-gitudinal studies investigated attitudes assessed be-fore and after the catastrophic events, thus usingthe same sample of respondents.(2−4) These stud-ies will be discussed in more detail later. Also, afew polls in the United States and Europe moni-tored the public’s opinion of nuclear power.(10,12) Weare, of course, aware of the fact that a large num-ber of studies have been conducted to investigatepeople’s attitudes and acceptance of nuclear powerstations and related facilities, as well as the deter-minants of these opinions.(13−18) This introduction,however, concentrates on studies that intended to ex-amine the effect of a nuclear accident on the public’sopinion of this energy technology.

Even though opinion polls revealed that thepublic’s support for nuclear power had been de-creasing since the mid-1970s in Europe and theUnited States,(5,10,12) the two previous nuclearaccidents seemed to have had additional nega-tive impacts on people’s attitudes toward nuclearpower.(5,7,10,11,19−21) Hence, these nuclear accidentsaccelerated the decline in public support of nuclearpower. Several studies investigated possible determi-nants of the more negative attitudes resulting fromthe nuclear accidents.(8,9,20,21) In countries that wereexposed to higher radiological doses after Cher-nobyl, people’s attitudes toward nuclear power be-came more negative than those of people in countriesin which the exposure levels were lower.(9) Thus, theextent to which a nuclear accident challenged previ-

ous beliefs about nuclear power appeared to affectpeople’s attitudes.

A study in Sweden examining various spe-cific groups (e.g., farmers, parents, and adolescents)showed that perceived risks of nuclear power wereimportant predictors of people’s nuclear power eval-uation.(20) However, only negative consequenceswere included in this study; positive issues, such asperceived benefits of nuclear power, were not con-sidered. A study conducted in the United States afterTMI included an assessment of the expected bene-fits of a new, local nuclear power plant and an as-sessment of its expected risks to explain the accep-tance of this new plant.(21) In addition, respondentswere asked to estimate the influence of the accidentat TMI on their acceptance of the new plant. Both ex-pected risks and expected benefits influenced accep-tance of the local plant. The self-reported influenceof TMI affected expected risks to a larger extent thandid expected benefits. However, a self-reported mea-sure of TMI influence is prone to biases, which mayhave indirectly resulted in perceived risks having alarger influence than perceived benefits.

In a study that compared perceptions of nuclearpower before and after Chernobyl between two dif-ferent samples, indications of slightly higher risk per-ceptions were found after Chernobyl (though onlyamong students) compared to before.(6) More inter-estingly, after the accident, people reported higherlevels of dread, more knowledge about the risks ofnuclear power, and a smaller degree of perceivedseverity of a nuclear accident. The latter was prob-ably a result of increased knowledge about the num-ber of casualties.

1.2. Previous Longitudinal Studies Regardinga Nuclear Accident

To the best of our knowledge, only three stud-ies have examined the impact of a nuclear accidenton laypeople’s perception of nuclear power while us-ing a longitudinal, within-subjects design—that is, in-cluding a survey before the accident and one surveyor several surveys after the accident.(2−4) FollowingMenard,(22) we define a longitudinal study as a studythat measures the same variables in the same respon-dents at two or more time points. All three previouslongitudinal studies were conducted before and afterthe Chernobyl accident.

Verplanken(3) was able to follow 154 Dutch peo-ple before and after Chernobyl (one month after and,again, six months after), measuring their attitudes

How a Nuclear Power Plant Accident Influences Acceptance of Nuclear Power 3

and beliefs regarding nuclear power. A third follow-up survey was done 19 months after the event, whichincluded 103 respondents from the original sam-ple. Although nuclear attitudes had become signifi-cantly more negative directly after the accident com-pared to before the accident, they almost returnedto pre-Chernobyl levels six months after the acci-dent. However, 19 months after the accident, pub-lic opinion again became more negative, perhaps dueto widespread media coverage. Moreover, one-and-a-half years after the event, the accident appeared toincrease respondents’ perceived probability of a nu-clear accident occurring and to decrease their ben-efit perception of nuclear power. The correlationsbetween nuclear attitudes and perceived risks andperceived benefits appeared strong at all times. How-ever, the relation between people’s nuclear power at-titudes and their perceived probability of a seriousnuclear accident occurring had become even strongerby the time the last measurement was completed, 19months after the accident. An increased risk percep-tion may explain the increase in negative attitudes af-ter one-and-a-half years compared to those attitudesheld directly after the nuclear catastrophe. Neverthe-less, Verplanken(3) did not examine the causal rela-tion between risk perception and nuclear attitudes.

The second longitudinal study regarding Cher-nobyl was conducted in England.(4) It consisted ofpre- and postincident measurements of public opin-ion taken around the time of the nuclear accident.The measurements included 135 inhabitants of atown close to a nuclear power plant and residentswho did not live far from an onshore oil installa-tion. Acceptance of nuclear power stations was com-pared to that of oil wells, coal power stations, andother industrial plants. After Chernobyl, respondentsshowed less acceptance of nuclear power, whereasacceptance of the other plants had not changed.Moreover, Eiser et al.(4) related post-Chernobyl atti-tudes toward nuclear power to acceptance, perceivedrisk from a hypothetical nuclear power station at var-ious distances, respondents’ attention to news aboutthe nuclear accident, and their level of fear evokedby this news. Prior negative nuclear attitudes wereassociated with a decrease in acceptance after the ac-cident, more attention being paid to the news, andhigher levels of fear being felt about this news afterthe accident. Respondents’ risk perception was notinfluenced by their distance to the hypothetical nu-clear plant.

Another longitudinal study was done in thenorthwest of the United States in a town near a

nuclear power plant.(2) The area was affected by aplume from Chernobyl that contained small amountsof radioactive material. The study included a surveyfive months before Chernobyl and a second surveyone month after the accident. After the accident, the69 respondents perceived fewer personal health ef-fects, such as cancer, from the local nuclear powerplant, although they were more occupied with thetopic of nuclear power than they had been before theaccident. Neither their perceived likelihood nor theirperceived dread of a nuclear accident had changedsince the incident.

Midden and Verplanken(23) observed that atti-tudes toward nuclear power and beliefs about theconsequences of nuclear power were not stable be-tween several measurement points during the twoyears after the accident at Chernobyl. However, theauthors could not compare these beliefs and attitudesto those before the accident.

In sum, only a few studies investigated the effectsof a nuclear accident using a longitudinal design. Thetwo European full-longitudinal studies noted abovefound that people’s views of nuclear power stationsbecame more negative after the Chernobyl catastro-phe.(3,4) However, the attitude changes were rela-tively small. All three full-longitudinal studies couldonly compare acceptance of nuclear power betweenthe two measures.(2−4) The possible determinants ofacceptance, such as attitudes and beliefs, were onlyassessed after the catastrophe. Moreover, the studieshad small sample sizes, ranging from less than 70 to150 respondents.

1.3. Current Situation

Since the accidents in Chernobyl and on TMI, afew developments have taken place in the areas ofscience, society, politics, and the environment as partof a “nuclear renaissance,” that is, a revival of nuclearenergy production.(24) First, the political climate haschanged. In the 1980s, during the Cold War, nuclearpower came to be associated with atomic weaponsand nuclear war.(10) People’s fears about the lattermay have affected their opinions of nuclear powerstations. A study conducted in 2009 showed thatSwiss people mainly associated nuclear power withenergy.(25) In comparison, its associations with risks,accidents, and military use were much less frequentlyreported. Second, the need to reduce CO2 emissionsto mitigate climate change contributed to a more pos-itive perception of nuclear power.(26) The influence


of nuclear power’s climate benefits on laypeople’s ac-ceptance appeared significant but relatively low.(27)

A third, more important factor that seemed todrive the nuclear renaissance was the perceived eco-nomic benefits of nuclear power.(27) A country withnuclear power plants can be less dependent on othercountries, has a secure energy supply, and producesrelatively cheap and sufficient energy to meet theneeds of increasing energy consumption. Moreover,the total costs (including the environmental burden)of nuclear power seem to be much lower than thoseof fossil fuel power plants in Switzerland, such thatnuclear power has come to be perceived as a cheap,green alternative.(28)

These factors resulted in a more favorable publicopinion about nuclear power in Switzerland and else-where before the accident in Fukushima. Whereasonly 49% of the U.S. population favored the use ofnuclear power for electricity generation in 1983, thispercentage increased to 74% in 2010.(29) Similarly,the number of Europeans who accept nuclear powerseemed to have increased lately. Within the 27 EUmember states, for example, the number of peoplesupporting the expansion of nuclear power has re-cently increased (from 14% in 2006 to 17% in 2009),whereas the percentage of Europeans opposing nu-clear power had decreased (from 39% in 2006 to 34%in 2009).(30)

Since people’s attitudes recently have been morein favor of using nuclear power compared to thosein the 1970s and 1980s, a nuclear catastrophe mayhave less influence on the public’s opinion of nuclearpower compared to 25 years ago. It could be that al-though people do recognize the risks and negativeconsequences, the perceived benefits for the climateand for a secure energy supply are still the main de-terminants of people’s acceptance. The nuclear acci-dent in Fukushima provided an opportunity to exam-ine this question.

2. RATIONALE AND MODEL

The general aim of our study was to investi-gate the effect of the nuclear accident in Fukushimaon people’s perception and acceptance of nuclearenergy in Switzerland. More specifically, we exam-ined how this event affected the degree to whichtrust, perceived benefits, and perceived risks influ-enced people’s acceptance of nuclear power stations.Thus, our main construct under investigation waspeople’s acceptance of nuclear power stations be-fore and after the accident in Japan. Therefore, ac-

ceptance of nuclear power stations before and afterFukushima were the latent variables to be explainedin the model, which is presented in Fig. 1.

Previous research suggests that both risk andbenefit perceptions determine acceptance of nuclearpower.(27,31) Furthermore, people’s perceived bene-fits of a secure energy supply seem to have a strongerinfluence on acceptance than do perceived risks.(27)

Little is known, however, about the stability of riskand benefit perceptions over time.(2,23) Therefore, itis unknown how a severe accident changes the im-pact of risk and benefit perceptions on acceptanceof nuclear power. After a nuclear accident, perceivedrisks may be more important to acceptance of nuclearpower compared to perceived benefits. In line withthis explanation, Hughey and Sundstrom(21) foundthat after the accident at TMI, laypeople’s accep-tance of a new local nuclear power plant was deter-mined mainly by their hazard expectations and lessby the expected benefits.

We hypothesized that in 2010, before the nuclearaccident, perceived risks of nuclear power wouldhave a small, negative effect on acceptance (path ain Fig. 1), whereas perceived benefits would morestrongly influence acceptance of this energy source(path b in Fig. 1). We further expected, as in thecase of gene technology,(32,33) that there would ex-ist a causal path from perceived benefits to perceivedrisks (path c in Fig. 1). Perceived benefits are thoughtto influence perceived risks, and not the other wayaround, because the benefits of nuclear power, suchas energy production and the lack of CO2, emis-sions are concrete and known. The benefits, there-fore, result in a clear evaluation. The risks, though,are mostly vague and difficult to imagine for laypeo-ple. For example, many people do not know the ex-act impact and probability of radioactive leakage.This unclear idea regarding the risks makes their per-ception easy to modify for cases in which they haveconflicting beliefs about nuclear power.(34) That is, ifpeople perceive that there are many benefits to nu-clear power, they can easily reduce their perceptionof risks so that the conflict is solved.

People seem to rely on social trust to assess risksand benefits if they have little knowledge of the tech-nology or hazard.(35) They then assume that theirlevel of trust in specific actors will provide a goodindication of the risks and benefits of the technol-ogy.(36) Trust in the actors managing a hazard ortechnology has been found to be an important de-terminant of the perceived risks and benefits of vari-ous technologies and hazards,(36,37) as well as that of


Fig. 1. Our hypothesized explanatory model of laypeople’s acceptance of nuclear power stations before and after the Fukushima disaster.

nuclear energy.(17,27,37−40) Laypeople reported hav-ing relatively little knowledge of nuclear power inprevious studies.(35,41) Therefore, we hypothesizedthat in 2010, trust in the actors involved in nuclearpower was negatively related to perceived risks (pathd in Fig. 1) and positively related to perceived bene-fits (path e in Fig. 1).

After the accident in 2011, we expected a simi-lar relation between risk perception, benefit percep-tion, and acceptance of nuclear power stations. Thus,the same model explains laypeople’s acceptance ofnuclear power after the accident (paths f, g, and hin Fig. 1). Trust is a crucial variable that determinespeople’s interpretation such accidents.(35,42) Thus, wehypothesized that in 2011, trust would be importantfor determining risk perception (path i in Fig. 1)and benefit perception (path j in Fig. 1) becausepeople rely on trust to interpret the accident.(43) Inother words, for determining trust, value informationseems to be more important than performance infor-mation.(43) Because values are relatively stable withinindividuals, we expected that trust in 2011 is stronglyrelated to trust in 2010 (path k in Fig. 1), althoughthe performance information may have reduced peo-ple’s trust in the nuclear power industry and regu-latory agencies after the nuclear accident. As shownin Fig. 1, we further presumed that risk and benefitperceptions, as well as acceptance, are not correlatedacross the two measurement points after controllingfor trust.

Besides testing the model presented in Fig. 1,we examined the mean changes of the mainconstructs included in the model. On the basis ofthe studies done after the two previous nuclear ac-cidents,(2−4,10) we expected that people’s acceptanceof nuclear power stations in Switzerland, their ben-efit perceptions, and their trust would decrease afterthe Fukushima nuclear accident, whereas perceivedrisks would increase.

3. METHOD

3.1. Procedure and Sample

Data were collected in two waves. The first sur-vey was conducted more than five months beforethe earthquake and tsunami in Japan, at the end ofSeptember 2010. We sent a paper-and-pencil ques-tionnaire together with a stamped and addressed re-turn envelope to a random sample of households inthe German-speaking part of Switzerland. A com-puter program had randomly selected the samplefrom telephone directories. After four weeks, a re-minder was sent to the addresses from which we hadnot received a completed survey. This was repeatedafter three weeks. In the end, we received 1,232 ques-tionnaires (a response rate of 40%). The same groupof respondents was asked to fill out a shorter ver-sion of the previous questionnaire two weeks afterthe catastrophic event, at the end of March 2011.


Fig. 2. Results of structural equation modeling explaining laypeo-ple’s acceptance of nuclear power stations before (values in reg-ular font) and after (values in bold font) the nuclear accident inJapan, including the standardized regression weights and squaredmultiple correlations (SMCs).Note: All standardized regression weights are significant, p <

0.001. Standardized regression weights in regular font refer to databefore the nuclear accident (2010), whereas values in bold font re-fer to data collected after the nuclear accident (2011).

This second survey was sent to the 1,232 addressesfrom the first survey. We sent one reminder tononrespondents after three weeks. In total, 929 ques-tionnaires were returned (a response rate of 77%).

In the instructions for the second survey, recip-ients were explicitly informed that the person whocompleted the first survey should also fill in the newsurvey. We reminded the respondents that their datafrom the previous survey and those of the new surveywere anonymous and that only anonymized data setswere linked. We asked for each respondent’s birthyear and gender to match the data sets of the two sur-veys. Data sets with different birth years or genderswere deleted. After omitting cases missing more than20% of their data as well as cases with two or moremissing data regarding the dependent variable accep-tance of nuclear power stations, the data set included790 respondents, from which the analyses describedin this article were done.

3.2. Questionnaire

The 2010 survey was about people’s perceptionsof nuclear power stations in Switzerland and includeditems regarding 10 different constructs and demo-graphics. Four constructs are of importance here:acceptance of nuclear power stations, risk percep-tion, benefit perception, and trust. Each constructwas measured with three items (see Table I for theitems, their mean scores, and their standard devi-ations). The items were assessed on 7-point Likertscales, with higher scores indicating higher values onthe construct measured.

The shorter questionnaire, filled out after the ac-cident in Japan, contained the same items to assessacceptance of nuclear power, risk perception, ben-efit perception, and trust. In addition, we includedtwo items that assessed respondents’ perceived sim-ilarity between the situations in Japan and Switzer-land regarding nuclear accidents (e.g., “It is verylikely that a similar nuclear accident could happen inSwitzerland”).

3.3. Data Analysis

To be able to apply structural equation mod-eling (SEM) to our data, we had to replace anymissing data in our data set. Forty-six respondents(5.8%) did not answer all of the survey questionsthat were identical in the two waves (24 items; seeTable I). Of the respondents with missing data, mostof them missed one item (3.4% of the sample) orthree items (1.3% of the sample). Using SPSS 18.0,we applied the expectation-maximization (EM) algo-rithm, based on the maximum likelihood approach,to replace missing values. The values of the items inthe respective construct were used to estimate themissing values.

We first tested the explanatory model of accep-tance of nuclear power in 2010 and the explanatorymodel of acceptance of nuclear power in 2011. Wethen tested the full model presented in Fig. 1. Allmodels were analyzed using SEM. SEM is a statis-tical analysis procedure that tests the fit between ahypothesized model and a data set. It combines con-firmatory factor analysis, used to validate the mea-surements of latent factors, with regression analy-sis, which examines the paths between latent fac-tors. Moreover, SEM can simultaneously examineseveral structural relations between latent constructs.Parameters were estimated using the maximumlikelihood method in the Analysis of Moment Struc-tures program (AMOS) version 18.0.


Table I. Questionnaire Items per Scale in 2010 and 2011, Including Their Mean Values (and Standard Deviations) and StandardizedFactor Loadings in the Final Model

2011 2010

Standardized StandardizedItems per Scalea M (SD) Factor Loadings M (SD) Factor Loadings

Acceptance1. Switzerland needs a lot of electricity;

people should therefore acceptnuclear power stations.

4.05 (2.25) 0.90 3.12 (2.12) 0.92

2. Switzerland can renounce nuclearpower stations without anyproblems.b

4.59 (2.13) 0.74 4.11 (2.11) 0.75

3. We need nuclear power stations inSwitzerland because renewableenergy sources alone do not producesufficient electricity.

4.15 (2.21) 0.84 3.44 (2.14) 0.88

Perceived risks4. I am very concerned about the

dangers of nuclear power stations inSwitzerland.

4.12 (1.98) 0.66 4.80 (2.01) 0.66

5. The nuclear power stations that arebuilt now or in the near future aresafe.b

3.75 (1.89) 0.87 4.76 (2.00) 0.87

6. The catastrophe risks in new nuclearpower stations are very small.b

3.63 (1.88) 0.83 4.37 (2.01) 0.85

Perceived benefits7. New nuclear power stations will

protect Switzerland from anelectricity bottleneck.

4.63 (2.06) 0.85 4.01 (2.11) 0.80

8. The electricity price would becometoo high in Switzerland if theexisting nuclear power stations werenot be replaced.

4.24 (1.94) 0.79 3.91 (1.99) 0.77

9. Industry would still have a securedenergy supply without replacing theexisting nuclear power stations.b

4.24 (1.92) 0.75 3.89 (1.97) 0.80

Trust11. . . . in scientists in the field of

nuclear power.4.87 (1.65) 0.78 4.30 (1.77) 0.77

13. . . . in the Swiss Federal Office ofEnergy.

4.41 (1.58) 0.73 3.86 (1.61) 0.75

14. . . . in the operators of nuclearpower stations.

4.01 (2.00) 0.87 3.22 (1.89) 0.90

aAll items were measured on 7-point Likert scales, with increasing scores indicating higher values on the item.bItem was recoded.

We conducted a confirmatory factor analysis tocheck the quality of the constructs. All factor load-ings (λs) were substantial (λs > 0.67) and significant,given that the critical ratios of all variables (i.e., thevariable estimate divided by its standard error) werelarger than 1.96 (ps < 0.05). The items thus repre-sented their constructs well enough to be included inour model.

In the full model, we correlated the error termscorresponding to the identical latent-factor measure-ments across time to get reliable estimates of the

regression paths.(44) Error terms for the same mea-surements tend to correlate when they are assessedamong the same sample across time, which is, ofcourse, the case in our longitudinal study. For ex-ample, the measurement error of an item that be-longs to the latent factor “perceived risks” in 2010is related to the measurement error of the identicalitem in 2011. The factor loadings and the correlationsbetween the error terms of the indicator variablesare, for the sake of simplicity, not shown in Figs. 2and 3.


Fig. 3. The final model explaining laypeople’s acceptance of nuclear power stations before and after Fukushima, including the standardizedregression weights and squared multiple correlations (SMCs).Note: All standardized regression weights are significant, p < 0.001.

Because our sample was relatively large, signif-icant chi-square tests were insufficient for assessingthe model’s goodness of fit. Therefore, we used thecomparative fit index (CFI), the standardized rootmean square residual (SRMR), and the root meansquare error of approximation (RMSEA) as crite-ria for our model.(45) All three statistics have valueranges between 0 and 1. A large SRMR indicates apoorly fitting model, whereas a SRMR smaller than0.08 signifies a model that fits well to the data. Forthe RMSEA, a value equal to or smaller than 0.06indicates a close fit. A CFI value equal to or over 0.95indicates a good fit (see Ref. 45 for an overview ofthe fit indices and their cut-off values). All these fitindices are descriptive; they are based not on infer-ential statistics but on rules of thumb to determinegoodness of fit. In addition, we looked at the theo-retical background of the model as a criterion for thefit of the model.(46) To see whether the model couldbe improved by including other relations between theconstructs, we compared the theoretical backgroundand the modification indices (MIs) with their respec-tive expected parameter change (EPC). The valuesof the MIs indicate whether relations between cer-tain items, latent constructs, and error terms could beincluded as these are likely to improve the model fit.We used the difference in χ2 to test the improvementof the new model.

4. RESULTS

4.1. Respondents

The mean age of our final sample of 790 respon-dents was 57.15 years (SD = 15.02), and 67.2% (n =531) of the respondents were male. Most respondentshad finished vocational school (n = 324, 41.7%) orcollege or university (n = 242, 31%). Almost 19% ofthe sample had completed up to a higher secondaryschool level (n = 147), and 8.5% had only finishedthe country’s obligatory school levels (n = 66). Oursample was older, included more men, more peoplewith a college or university degree, and fewer peo-ple who had only finished obligatory school than wasrepresentative of the average Swiss population.(47,48)

We compared the respondents who participatedin both surveys to those who had only completedthe survey in 2010 (i.e., the nonrespondents inthe 2011 survey). Of the 304 nonrespondents, 287cases remained after deleting those missing morethan 20% of the 2010 data. The gender distribu-tion was similar among respondents and nonrespon-dents (χ2(1) = 0.12, p = 0.73). Nonrespondents wereslightly younger (M = 53.55, SD = 16.37) than re-spondents (M = 57.15, SD = 15.02), t(1,071) =–3.37, p = 0.001, r = 0.10. Also, respondents hada somewhat higher education level (18.6% with ahigher secondary school education and 30.6% with a


Table II. Means and Standard Deviations of the Four Constructsin 2010 and 2011, and Results of Paired-Samples t-Tests Between

the Measurements in 2010 and 2011

2010 2011

M SD M SD t df N

Acceptance 4.26 1.95 3.56 1.91 13.95∗ 789Perceived risks 3.84 1.66 4.65 1.74 −17.47∗ 789Perceived benefits 4.37 1.73 3.93 1.76 9.15∗ 789Trust 4.43 1.54 3.80 1.56 14.60∗ 778

∗p < 0.001.Note: All items measured on 7-point Likert scales.

college- or university-level education) than nonre-spondents (13.6% with a higher secondary schooleducation and 26.1% with a college- or university-level education), U = 98,079.50, p = 0.01, r =–0.08. Nonrespondents showed more acceptance ofnuclear power (M = 4.55, SD = 1.77) than respon-dents (M = 4.26, SD = 1.95), t(555.49) = 2.25,p = 0.03, r = 0.10. Nonrespondents also perceivedfewer risks from this technology (M = 3.57, SD =1.47) than did respondents (M = 3.84, SD = 1.66),t(539.67) = –2.59, p = 0.01, r = 0.11. The re-spondents and nonrespondents did not differ re-garding perceived benefits and trust, ts < 1.77,ps > 0.08.

4.2. Acceptance of Nuclear Power Stations,Perceived Risks, Perceived Benefits, and TrustBefore and After Fukushima

As can be seen in Table II, respondents’ accep-tance of nuclear power stations in 2010 was aroundthe midpoint of the 7-point rating scale, and it sig-nificantly decreased after the nuclear accident. Bothperceived benefits and trust decreased significantlyafter the events in Japan. Respondents’ level of riskperception significantly increased after the nuclearaccident. However, the sizes of the differences (i.e.,0.79 at maximum for perceived risks, measured on 7-point Likert scales) between the two surveys indicatethat respondents had not completely changed theirviews on nuclear power. Moreover, respondents sawthe situation in Japan regarding nuclear power sta-tions as being relatively similar to the situation inSwitzerland, given that the mean perceived similaritywas slightly above the midpoint of the 7-point scale(M = 4.54, SD = 1.85).

4.3. Separate Models for Explaining Acceptanceof Nuclear Power Stations Before and After theNuclear Accident

The model for explaining acceptance of nuclearpower stations before the nuclear accident fit quitewell to our data (χ2(49) = 153.88, p = 0.0001, CFI =0.984, SRMR = 0.027, RMSEA = 0.052). The stan-dardized regression weights of the paths were allsignificant and in the hypothesized directions (ps <

0.001). However, the modification index (MI) be-tween the two error terms of the items trust in sci-ence and trust in the Federal Office of Energy (FOE)in 2010 indicated that including the correlation be-tween these two error terms in our model would re-sult in a substantial improvement of our model fit.People probably see scientists and the FOE as in-dependent actors with respect to nuclear power sta-tions, and their independence may have a large rolein people’s trust evaluation of both actors; hence, ourrespondents rated them similarly. It therefore madesense to include the correlation between the two er-ror terms in our 2010 model. As a result, the model fitindeed improved (χ2(48) = 112.07, p = 0.0001, CFI =0.990, SRMR = 0.019, RMSEA = 0.041), which wasfound to be statistically significant (�χ2(1) = 41.81,p < 0.001). All paths were significant (ps < 0.001)and had similar directions and sizes of influence com-pared to our initial model, described earlier. Fig. 2presents the final 2010 model, including the standard-ized coefficients (in regular font) but without the fac-tor loadings.

In 2010, perceived risks and perceived bene-fits explained 90% of the variance of acceptance ofnuclear power stations. Perceived benefits had thelargest influence on acceptance, and risk perceptionhad a smaller negative relation to acceptance. Trusthad a strong positive influence on perceived bene-fits in 2010 and a negative influence on perceivedrisks. As well, perceived benefits had a negative in-fluence on perceived risks. The squared multiple cor-relations (SMCs) were substantial for perceived ben-efits (SMC = 0.59) and perceived risks (SMC = 0.79).

The same model also explained well the data col-lected after the nuclear accident in Japan (χ2(49) =268.72, p = 0.0001, CFI = 0.970, SRMR = 0.033,RMSEA = 0.075). The MIs again indicated that thecorrelation between the error terms of the items trustin science and trust in the FOE in 2011 would increasethe model fit. We therefore tested a new model thatincluded the correlation between these two errorterms. This, too, improved the model fit significantly


(χ2(48) = 201.61, p = 0.0001, CFI = 0.979, SRMR =0.026, RMSEA = 0.064, �χ2(1) = 67.11, p < 0.001).In 2011, perceived benefits and perceived risksalso largely contributed to the explained varianceof acceptance of nuclear power stations (SMC =0.94). Perceived benefits again had a strong positiveinfluence on acceptance (see Fig. 2, standardized co-efficients in bold font), whereas perceived risks had asmaller, negative effect on acceptance. Trusthad a positive influence on perceived benefits andhad a negative influence on perceived risks. Per-ceived benefits contributed negatively to perceivedrisks. The explained variances were again substantialfor perceived benefits (SMC = 0.66) and perceivedrisks (SMC = 0.85).

4.4. The Full Model

In our SEM analysis of the full model (Fig. 1),we immediately included the correlations betweenthe error terms of the items trust in the FOE andtrust in science in 2010 and 2011. The hypothesizedmodel fit quite well to our data (χ2(227) = 719.56,p = 0.0001, CFI = 0.968, SRMR = 0.053, RMSEA =0.052). The standardized regression weights of thepaths were all significant and in the hypothesized di-rections (ps < 0.001). The MIs suggested that includ-ing the path between perceived benefits in 2010 andperceived benefits in 2011 would substantially im-prove the model. This inclusion also made theoret-ical sense because the benefits of nuclear power re-mained the same; the nuclear accident had not af-fected them. People therefore relied on the perceivedbenefits before the nuclear accident to estimate thebenefits after such an event. Including the path be-tween the perceived benefits in 2010 and 2011 did re-sult in a model that fit better to the data (χ2(226) =536.96, p = 0.0001, CFI = 0.980, SRMR = 0.031,RMSEA = 0.042) and was found to be statisticallysignificant (�χ2(1) = 182.59, p < 0.0001). The com-bination of the explanatory models for 2010 and 2011explained 95% of the variance regarding acceptanceof nuclear power plants in 2011. The SMCs of accep-tance in 2010, trust in 2011, and perceived risks andperceived benefits in 2010 and 2011 were substantial(0.61 < SMCs < 0.92). The MIs did not suggest in-cluding any other relations that would substantiallyimprove the model fit. Table I presents the standard-ized factor loadings of the individual items on the la-tent constructs in the final model. The correlationsbetween the error terms corresponding to the iden-

tical latent factor measurements in 2010 and in 2011ranged between 0.04 and 0.34.

The direction of the paths described in the sepa-rate models remained the same in the full model (seeFig. 3). In addition, perceived benefits in 2010 had apositive influence on perceived benefits in 2011. Sim-ilarly, trust in 2010 was strongly related to trust in2011. When comparing the regression path betweenperceived risks and acceptance in 2010 to that of2011, it looks as if perceived risks had a stronger in-fluence on acceptance after the nuclear accident com-pared to before. We checked whether these pathswere indeed significantly different by, first, restrain-ing the two paths between perceived risks and accep-tance in 2010 and in 2011 in a new model and thencomparing the quality of the new model to that ofthe unrestrained model.(49) If the influence of riskperception on acceptance in 2011 was indeed largerthan that in 2010, fixing these two paths should re-sult in a significantly worse-fitting model. The newmodel with the two restrained paths had a good fitto the data (χ2(227) = 538.89, p = 0.0001, CFI =0.980, SRMR = 0.031, RMSEA = 0.042), but therestrained model did not result in a statistically sig-nificantly worse-fitting (or better-fitting) model com-pared to the unrestrained model (�χ2(1) = 1.92, p =0.17). This result seems to indicate that the impactof risk perception on acceptance in 2010 was alreadysimilar to that in 2011 in the unrestrained version ofthe full model.

In sum, 95% of the variance in our respondents’acceptance of nuclear power stations after the acci-dent in Fukushima could be explained by their riskperception and benefit perception after the accident.The model suggests that there is both a direct ef-fect of trust in 2011 and an indirect effect of trust in2010 on perceived risks and benefits in 2011. Both be-fore and after the nuclear accident, perceived bene-fits had the strongest influence on acceptance of nu-clear power stations, compared to perceived risks.

5. DISCUSSION

There is a lack of longitudinal studies in the do-main of risk perception research. Therefore, little isknown about how technological accidents influencepeople’s risk perception and acceptance of varioustechnologies. The accident at Fukushima provided anopportunity to conduct a quasi-experimental study.A random sample of laypeople who participatedin a survey about nuclear power in autumn 2010answered the same set of questions about trust,


perceived risks and benefits, and acceptance of nu-clear power after the accident in Japan. In our model,we predicted that trust is a key factor, one that di-rectly influences risk perception and benefit percep-tion and that indirectly influences acceptance of nu-clear power. Furthermore, we hypothesized that trustbefore the accident had a significant effect on trust af-ter the accident. Our model explained the data verywell.

In the following paragraphs, we will first discussthe effect of the nuclear accident on people’s per-ceptions and acceptance of nuclear power. We willthen elaborate further on the role of trust concern-ing the acceptance of nuclear power stations. We willalso suggest a few implications of our findings forrisk communication about nuclear power stations. Fi-nally, we will provide some critical remarks about ourstudy.

To recap, our respondents reported higher lev-els of risk perception and lower levels of trust, per-ceived benefits, and acceptance regarding nuclearpower plants after the nuclear accident compared tobefore. Our findings, thus, seem to indicate that thenuclear accident in Fukushima resulted in a morenegative perception of nuclear power and, hence, hadimportant implications for the image of this energytechnology among the public, at least in the shortterm. However, the influences of the determinantsof acceptance—risk perception, benefit perception,and trust—seemed to remain stable over time. It istherefore very likely that because perceived benefitsand trust decreased and perceived risks similarly in-creased, the acceptance of nuclear power plants alsodecreased.

Past studies have shown that for gene technol-ogy,(32,33) nanotechnology,(50) and nuclear power,(27)

perceived benefits have a stronger impact on accep-tance of the respective technology than do perceivedrisks. This study not only seems to replicate this find-ing but also suggests that after a severe accident, theimportance of perceived risks increases somewhat,although perceived benefits remain a more importantpredictor of acceptance. Thus, people may considernuclear power’s benefit for the energy supply and fora healthy economy to be more important to their ac-ceptance of it than the possible risks that are associ-ated with it.

It may come as a surprise that risk perceptionseemed to have a smaller influence than benefit per-ception on acceptance of nuclear power stations af-ter learning about the nuclear accident. Studies con-ducted after Chernobyl and TMI pointed toward the

importance of risk perception on attitudes towardnuclear power.(20,21) In our model, we were able tocompare the influence of risk perception on accep-tance to that of benefit perception, which showedthat the impact of perceived risks had not changedmuch after the Fukushima accident and that per-ceived benefits remained the strongest predictor ofacceptance. An explanation for these findings couldbe that our respondents had already acknowledgedthe risks of nuclear power before the nuclear acci-dent and they therefore did not differ much in theirrisk perception both before and after the nuclear ac-cident. Alternatively, our respondents varied moreregarding their benefit perception. Perceived benefitscould therefore make a larger contribution towardexplaining acceptance than could perceived risks.This finding is supported by findings from previ-ous studies. Proponents of nuclear power have beenfound to be more ambivalent about this technology,seeing both its benefits and risks, and therefore havea less stable attitude toward nuclear power than op-ponents.(23) Moreover, people have been found ei-ther to have both positive and negative images aboutnuclear power or to have only negative images, buthardly any person reports only positive images.(51)

Even proponents seem to acknowledge some risks ofnuclear power.(25)

Next to perceived benefits, social trust continuedto be an important determinant in our full model. Itinfluenced risk and benefit perceptions directly and,therefore, acceptance of the technology, though in-directly. Also, the high correlation between trust in2010 and trust in 2011 seems to confirm the funda-mental role of trust in risk perception, benefit percep-tion, and acceptance. After the nuclear accident, thedirect influence of trust on perceived benefits seemedsmaller than before the nuclear accident. This wascaused by the additional path between the perceivedbenefits at the two measurement times; thus, trust in2010 not only influenced perceived benefits in 2011through trust in 2011 but also through the connec-tion between perceived benefits in 2010 and per-ceived benefits in 2011. These findings support theidea that trust remains fundamental to risk and ben-efit perceptions, even when new information aboutthe hazard is available. The importance of trust isalso in line with the trust, confidence, and coopera-tion (TCC) model, which distinguishes between trustand confidence.(36,42) According to this model, confi-dence is determined by past performance and com-petence, whereas social trust is determined by valuesimilarity. The disaster in Japan, which can be viewed


as past performance information, reduced people’strust. The reduction was rather small, however, giventhe severity of the accident, which may imply thatit is not performance information but value similar-ity that influenced people’s trust. The second surveytook place right after the accident. Future studies areneeded to examine changes in trust over time. It ispossible that the more time that passes after an ac-cident, the less effect the accident will have on trustand, as a result, on acceptance of nuclear power.

The findings of our longitudinal study seem toimply that even after a nuclear accident, perceivedbenefits remain the most important determinant ofacceptance, compared to perceived risks. An implica-tion of this finding for communication with the publicafter a nuclear accident may be that focusing on boththe risks and benefits of nuclear power will changethe public’s acceptance of nuclear power. However,only concentrating on the risks will not significantlyaffect the public’s opinion. In comparison, discus-sions about the benefits of nuclear power—such asa secure energy supply—or the lack of benefits, forthat matter, are more likely to change people’s ac-ceptance of nuclear power.(27)

Politicians in various countries reacted in differ-ent ways to the accident in Japan. In France andthe United States, politicians did not reconsider nu-clear power, whereas in Germany and Switzerland,politicians quickly abandoned their support of nu-clear power. The public’s change in opinion after theaccident, however, did not seem to be extreme. Asin previous studies,(3,4) our respondents did not seemto change their attitudes about nuclear power dras-tically. Therefore, politicians should not react tooquickly by changing nuclear policy on behalf of thepublic.

Our study was a longitudinal survey, so we hadthe benefit of being able to follow changes in theperception and acceptance of nuclear power stationswithin the same sample of people. Our study, how-ever, also had a few limitations. The first was thatthe second survey was sent out two weeks afterthe earthquake and tsunami in Japan. People maynot have been able to form a firm opinion aboutnuclear power stations so quickly after the event.The situation at the power plants in Fukushima wasnot stable at that time, so laypeople may have hadthe idea that a catastrophic event such as an ex-plosion was still likely to occur or, conversely, thatthe problems in the reactors would soon be solved.This uncertainty may have resulted in attitudes andbeliefs that are more negative or unstable than if

the situation had been stabilized. If we had post-poned the second measurement for a few weeks,our results may have been different.(3) Consequently,it would be interesting to follow our sample overtime to examine these people’s perceptions of nu-clear power after the accident has ceased being in thenews.

It can be speculated whether and how public per-ception changes in the long term. It could be that thepublic’s view of nuclear power remains more nega-tive than it was before the accident, as was foundafter the Chernobyl accident in a Dutch study.(3)

Verplanken suggested that due to this accident, thepublic was better able to imagine the occurrence ofextremely unlikely events, such that the perceivedrisks of nuclear power remained high after Cher-nobyl. Alternatively, the public could return to theview of nuclear power that it had held before theFukushima accident. Media coverage about the ac-cident became less intense: the number of hits forthe term “Fukushima” in the Swiss version of GoogleNews was reduced to 612 hits six months after theevent compared to about 11,500 hits four weeks af-ter the accident. This reduction could have limitedpeople’s access to mental images about nuclear ac-cidents.(3) Also, Swiss people may have readjustedtheir attitudes a few months after the accident by re-alizing that the physical consequences of the accidenthad happened far away and that the local situationwas safe.(52) This latter effect may, of course, only ap-ply to populations not directly affected by the nuclearaccident in Fukushima.

Another critical remark about our study is thatwe may not have included all relevant determinantsof nuclear power acceptance. For example, attitudestoward nuclear power have also been found to berelated to fairness(15) and to concerns about climatechange and the environment.(18) Also, our trust mea-sure may not have comprised all relevant actors. Aswe wanted to investigate the role of social trust in ourmodel, we assessed trust in three specific actors thatare relevant for the management of nuclear powerstations in Switzerland: the operators of the nuclearpower stations, nuclear scientists, and the Swiss Fed-eral Office of Energy (FOE). Nevertheless, the highfactor loadings of all three items both before andafter the nuclear accident seem to indicate that thethree items are highly related and that they provide agood representation of the social trust construct.

Similarly, we cannot explain the underlyingcause of the decrease in public acceptance of nu-clear power and its related concepts. As was the case


regarding the previous nuclear accidents, media cov-erage was high,(53) and the public’s view of thistechnology became more negative directly after theevents in Fukushima.(3,4) The study did not allow foran investigation of the influence of media on the ef-fects we found. More research is therefore needed tofully understand what caused the increase in a nega-tive view of nuclear power after the nuclear accidentat Fukushima.

An advantage of our study was that we could in-vestigate changes in the perception of nuclear powerwithin the same subjects. However, a disadvantageof our longitudinal survey was that we lost a numberof respondents in the second wave. Consequently,our sample became older and better educated. More-over, for some points, respondents’ views aboutnuclear power were less positive than those of non-respondents. In short, it could be that our final sam-ple was less representative of the Swiss populationthan our initial sample, which may have affected ouroutcomes. When we compared the 2010 model thatincluded the initial sample to the 2010 model thatincluded only respondents who had completed bothsurveys, the directions and strengths of the paths,as well as the explained variances, were similar inboth models. Furthermore, because our final samplewas more critical of nuclear power before Fukushimathan was the complete 2010 sample, we may have un-derestimated the relations in our model. Particularly,we would expect the influence of perceived bene-fits on acceptance of nuclear power stations to belarger, whereas the influence of perceived risks on ac-ceptance would be smaller in a more representativesample, that is, a sample with more respondents whowere positive about nuclear power before the nuclearaccident.

One could criticize that the study was conductedin Switzerland. Switzerland is, of course, quite farfrom Japan. It may be that people in countries closerto Japan responded differently to the nuclear acci-dent. Moreover, the Swiss energy situation may bedifferent from that of other countries such that thesurvey results cannot be generalized to other coun-tries. Switzerland depends on nuclear power stationsfor about 40% of its power generation. Three ofthe currently running reactors should be replaced by2020 to maintain the same energy supply. By thetime we conducted the second survey, the decisionto postpone all applications for new nuclear powerreactors had just been made (March 14, 2011). How-ever, our finding that benefit perception mainly de-termined people’s acceptance of nuclear power sta-

tions, both before and after the nuclear accident, is inline with the results of previous studies that showedthe overwhelming influence of benefit perception onacceptance of other technologies.(31,33) Therefore, webelieve that our results can be generalized to othercountries.

In sum, our study is one of the few longitudi-nal studies that investigated the public’s acceptanceof nuclear power before and after a nuclear acci-dent. We believe that it represents a major con-tribution to the limited knowledge available aboutthis subject. After the nuclear accident, the pub-lic’s acceptance and perceptions of nuclear powerstations as well as its trust in the stakeholders hadbecome more negative. Nevertheless, our explana-tory model did not show a shift in the importanceof the relations between the constructs that ex-plained people’s acceptance of nuclear power sta-tions. Therefore, a nuclear accident, although catas-trophic for the people living near the plant and forthe environment, does not seem to result in radi-cal changes in the influences of perceived benefits,perceived risks, or trust on the acceptance of thistechnology.

ACKNOWLEDGMENTS

The first survey reported in this article (con-ducted in 2010) was funded by Swissnuclear. Swiss-nuclear comprises representatives of the Swiss elec-tric supply companies (Alpiq, Axpo, BKW, CKW,and EGL), which are committed to the safe andeconomic operation of the nuclear power plants inSwitzerland.

REFERENCES

1. IAEA. Fukushima Nuclear Accident Update Log,2011. Available at: http://www.iaea.org/newscenter/2011/fukushimafull.html, Accessed on April 13, 2011.

2. Lindell MK, Perry RW. Effects of the Chernobyl accident onpublic perceptions of nuclear plant accident risks. Risk Anal-ysis, 1990; 10(3):393–399.

3. Verplanken B. Beliefs, attitudes, and intentions toward nu-clear energy before and after Chernobyl in a longitudi-nal within-subjects design. Environment and Behavior, 1989;21(4):371–392.

4. Eiser JR, Spears R, Webley P. Nuclear attitudes before andafter Chernobyl: Change and judgment. Journal of AppliedSocial Psychology, 1989; 19(8):689–700.

5. van der Pligt J. Public attitudes to nuclear energy: Salienceand anxiety. Journal of Environmental Psychology, 1985;5(1):87–97.

6. McDaniels TL. Chernobyl’s effects on the perceived risksof nuclear power: A small sample test. Risk Analysis, 1988;8(3):457–461.


7. Peters HP, Albrecht G, Hennen L, Stegelmann HU. “Cher-nobyl” and the nuclear power issue in West German pub-lic opinion. Journal of Environmental Psychology, 1990;10(2):121–134.

8. Eiser JR, Hannover B, Mann L, van der Pligt J, Webley P.Nuclear attitudes after Chernobyl: A cross-national study.Journal of Environmental Psychology, 1990; 10(2):101–110.

9. Renn O. Public responses to the Chernobyl accident. Journalof Environmental Psychology, 1990; 10(2):151–167.

10. Rosa EA, Dunlap RE. Nuclear power: Three decades of pub-lic opinion. Public Opinion Quarterly, 1994; 58(2):295–324.

11. de Boer C, Catsburg I. The impact of nuclear accidents onattitudes toward nuclear energy. Public Opinion Quarterly,1988; 52(2):254–261.

12. Bolsen T, Cook FL. Public opinion on energy policy: 1974–2006. Public Opinion Quarterly, 2008; 72(2):364–388.

13. Eiser JR, van der Pligt J. Beliefs and values in the nuclear de-bate. Journal of Applied Social Psychology, 1979; 9:524–536.

14. Pidgeon NF, Lorenzoni I, Poortinga W. Climate change or nu-clear power – No thanks! A quantitative study of public per-ceptions and risk framing in Britain. Global EnvironmentalChange, 2008; 18(1):69–85.

15. Besley JC. Does fairness matter in the context of angerabout nuclear energy decision making? Risk Analysis, 2012;32(1):25–38.

16. Greenberg M, Truelove HB. Energy choices and risk beliefs:Is it just global warming and fear of a nuclear power plantaccident? Risk Analysis, 2011; 31(5):819–831.

17. Greenberg M. Energy sources, public policy, and public pref-erences: Analysis of US national and site-specific data. En-ergy Policy, 2009; 37(8):3242–3249.

18. Spence A, Poortinga W, Pidgeon N, Lorenzoni I. Public per-ceptions of energy choices: The influence of beliefs about cli-mate change and the environment. Energy & Environment,2010; 21(5):385–407.

19. van der Pligt J, Midden CJH. Chernobyl: Four years later: At-titudes, risk management and communication. Journal of En-vironmental Psychology, 1990; 10:91–99.

20. Drottz-Sjoberg B-M, Sjoberg L. Risk perception and worriesafter the Chernobyl accident. Journal of Environmental Psy-chology, 1990; 10(2):135–149.

21. Hughey JB, Sundstrom E. Perceptions of Three Mile Is-land and acceptance of a nuclear power plant in a distantcommunity. Journal of Applied Social Psychology, 1988;18(10):880–890.

22. Menard S. Longitudinal Research, 2nd ed. Thousand Oaks,CA: Sage University Paper, 2002.

23. Midden CJH, Verplanken B. The stability of nuclear attitudesafter Chernobyl. Journal of Environmental Psychology, 1990;10(2):111–119.

24. World Nuclear Association. The Nuclear Renaissance, 2011.Available at: http://www.world-nuclear.org/info/inf104.html,Accessed on May 5, 2011.

25. Keller C, Visschers V, Siegrist M. Affective imagery and ac-ceptance of replacing nuclear power plants. Risk Analysis,2012; 32(3):464–477.

26. European Commission. Attitudes Towards RadioactiveWaste. Brussels: TNS Opinion & Social: Special Eurobarom-eter 297 / Wave 69.1, 2008.

27. Visschers VHM, Keller C, Siegrist M. Climate change benefitsand energy supply benefits as determinants of acceptance ofnuclear power stations: Investigating an explanatory model.Energy Policy, 2011; 39(6):3621–3629.

28. Roth S, Hirschberg S, Bauer C, Burgherr P, Dones R, HeckT, Schenler W. Sustainability of electricity supply technologyportfolio. Annals of Nuclear Energy, 2009; 36(3):409–416.

29. Bisconti Research Inc. Public Support for NuclearEnergy at Record High, March 2010. Available at:http://www.nei.org/filefolder/2010 March MEMO PublicOpinion2.pdf, Accessed on January 11, 2011.

30. European Commission. Europeans and Nuclear Safety. Brus-sels: TNS Opinion & Social: Special Eurobarometer 324 /Wave 72.2, 2010.

31. Tanaka Y. Major psychological factors determining public ac-ceptance of the siting of nuclear facilities. Journal of AppliedSocial Psychology, 2004; 34(6):1147–1165.

32. Siegrist M. The influence of trust and perceptions of risks andbenefits on the acceptance of gene technology. Risk Analysis,2000; 20:195–203.

33. Siegrist M. A causal model explaining the perception and ac-ceptance of gene technology. Journal of Applied Social Psy-chology, 1999; 29:2093–2106.

34. Abelson RP. Modes of resolution of belief dilemmas. Journalof Conflict Resolution, 1959; 3(4):343–352.

35. Siegrist M, Cvetkovich G. Perception of hazards: The role ofsocial trust and knowledge. Risk Analysis, 2000; 20(5):713–720.

36. Earle TC, Siegrist M, Gutscher H. Trust, risk perception,and the TCC model of cooperation. Pp. 1–49 in Siegrist M,Earle TC, Gutscher H (eds). Trust in Cooperative Risk Man-agement: Uncertainty and Scepticism in the Public Mind.London: Earthscan, 2007.

37. Siegrist M, Cvetkovich G, Roth C. Salient value similarity,social trust, and risk/benefit perception. Risk Analysis, 2000;20(3):353–362.

38. Flynn J, Burns W, Mertz CK, Slovic P. Trust as a determinantof opposition to a high-level radioactive waste repository:Analysis of a structural model. Risk Analysis, 1992; 12(3):417–429.

39. Viklund MJ. Trust and risk perception in Western Eu-rope: A cross-national study. Risk Analysis, 2003; 23(4):727–738.

40. Whitfield SC, Rosa EA, Dan A, Dietz T. The future of nuclearpower: Value orientations and risk perception. Risk Analysis,2009; 29(3):425–437.

41. European Commission. Europeans and Nuclear Safety. Brus-sels: TNS Opinion & Social: Special Eurobarometer 271 /Wave 66.2, 2007.

42. Earle TC. Trust in risk management: A model-based re-view of empirical research. Risk Analysis, 2010; 30(4):541–574.

43. Earle TC, Siegrist M. Morality information, performanceinformation, and the distrinction between trust and confi-dence. Journal of Applied Social Psychology, 2006; 36(2):383–416.

44. Joreskog KG. Statistical models and methods for analysis oflongitudinal data. Pp. 129–169 in Joreskog KG, Sorbom D(eds). Advances in Factor Analysis and Structural EquationModels. Cambridge, MA: Abt Books, 1979.

45. Hu LT, Bentler PM. Cutoff criteria for fit indexes in co-variance structure analysis: Conventional criteria versus newalternatives. Structural Equation Modeling: A Multidisci-plinary Journal, 1999; 6(1):1–55.

46. Iacobucci D. Everything you always wanted to know aboutSEM (structural equations modeling) but were afraid toask. Journal of Consumer Psychology, 2009; 19(4):673–680.

47. Swiss Statistics. Bildungsstand der Wohnbevolkerung nachAlter und Geschlecht [Education Level of the Resident Pop-ulation According to Age and Gender], 2010. Available at:http:/www.bfs.admin.ch/bfs/portal/de/index/themen/15/02/key/ind5.indicator.51131.511.html?open=9#9, Accessed onMay 31, 2011.

48. Swiss Statistics. Bevolkerungsstand und -struktur [Popula-tion Size and Composition], 2009. Available at: http:/www.bfs.admin.ch/bfs/portal/de/index/themen/01/02/blank/data/01.html, Accessed on May 31, 2011.

49. Byrne BM. Structural Equation Modeling with AMOS: Ba-sic Concepts, Applications, and Programming. Mahway, NJ:Lawrence Erlbaum Associates, 2001.


50. Siegrist M, Cousin M-E, Kastenholz H , Wiek A. Public ac-ceptance of nanotechnology foods and food packaging: Theinfluence of affect and trust. Appetite, 2007; 49(2):459–466.

51. Peters E, Slovic P. The role of affect and worldviews as ori-enting dispositions in the perception and acceptance of nu-clear power. Journal of Applied Social Psychology, 1996;26(16):1427–1453.

52. Butler C, Parkhill KA, Pidgeon NF. Nuclear power afterJapan: The social dimensions. Environment: Science and Pol-icy for Sustainable Development, 2011; 53(6):3–14.

53. Friedman SM. Three Mile Island, Chernobyl, and Fukushima:An analysis of traditional and new media coverage of nu-clear accidents and radiation. Bulletin of the Atomic Scien-tists, 2011; 67(5):55–65.

how a nuclear power plant accident influences acceptance of nuclear power: results of a longitudinal...

Documents