ENERGY & ENVIRONMENT

Transcription

1 RISK PERCEPTION BY POLITICIANS AND THE PUBLIC by Lennart Sjöberg and Britt-Marie Drottz-Sjöberg Reprinted from ENERGY & ENVIRONMENT VOLUME 19 No MULTI-SCIENCE PUBLISHING CO. LTD. 5 Wates Way, Brentwood, Essex CM15 9TB, United Kingdom

2 455 RISK PERCEPTION BY POLITICIANS AND THE PUBLIC 1 Lennart Sjöberg Center for Risk Psychology, Environment, and Safety, Department of Psychology, Norwegian University of Science and Technology, Trondheim, Norway and Center for Risk Research, Stockholm School of Economics, Stockholm, Sweden Lennartsjoberg@gmail.com, telephone Britt-Marie Drottz-Sjöberg Center for Risk Psychology, Environment, and Safety, Department of Psychology, Norwegian University of Science and Technology, Trondheim, Norway brittds@svt.ntnu.no, telephone ABSTRACT Risk perception by local politicians responsible for health and the environment is compared to that of the public in two large, representative samples, and to a group of nuclear waste management experts. Data were collected on many aspects of risk perception with an emphasis on ionizing radiation and nuclear waste issues. It was found that politicians and the public had, on the average, quite similar risk perceptions, which were very different from expert opinion. Politicians had difficulties in estimating what risk perception the public had. Demand for risk mitigation was mostly related to the respondents ratings of risk, for both politicians and members of the public. Politicians apparently did not consider the public s views to any large degree in forming their policy attitudes. Politicians had less trust in industry than the public did. Both epistemic and social trust were found to be factors in perceived risk, roughly equal in importance. The risk perception models also included Interfering with Nature as an important factor, whereas traditional Psychometric Model factors contributed very little. Frequency of communication with others about a nuclear policy issue was related to attitude; more strongly so if communication was more frequent. The results were interpreted as the outcome of a process of social validation of policy attitudes, a hypothesis which could also explain why experts had so divergent risk assessments as compared to politicians and members of the public. Attitudes to a local highlevel nuclear waste repository were quite negative in both groups, positive among the experts. 1 Study supported by a grant from ISPN, Fontenay-aux-Roses, France.

3 456 Energy & Environment Vol. 19, No. 3+4, INTRODUCTION Risk is an important factor in political decisions [1-3]. This is probably the case for several reasons. It is a common belief that the public demands drastic risk mitigation policies in some areas, which have become salient for different reasons. Politicians are often assumed to react on these public demands, rather than on their own risk assessments, or that of experts, implying that the political process is irrational [4, 5]. The result is a very uneven allocation of resources for risk mitigation [6-9], often criticized by researchers and policy makers [10, 11]. Furedi emphasized the conflict between the public s concern about small, theoretical hazards and the large and wellestablished risks inherent to our lives [12]. Risk, pain, and loss are salient in people s lives [13]. Politicians are much concerned about risk matters [14]. In a study of Swedish parliamentarians, it was found that the number of risk related private bills they submitted to Parliament had increased very strongly in the time period [3]. Other work has pointed to the political importance of blame and moral issues with regard to risk management [15-17]. Risk perception has often been believed to be a major determinant of policy [11, 18, 19]. This may or may not be true, and the notion gives rise to many interesting questions. How do different groups perceive technological and environmental risk? What is the role played by these groups and their risk perception in society? How do they see each other, e.g. how do politicians view the priorities espoused by the public? Do politicians have a correct view of the public s risk perception? There is extensive research on risk perception [20, 21]. Much of this work has its roots in Starr s observations regarding the complex determinants of acceptable risk [22] and the Psychometric Model of risk perception [23]. Most of it has been carried out with members of the public or various convenience samples of college students. In many cases politicians have the final responsibility for the risk policies that society pursues. Yet, there has been almost no work published on how politicians perceive risks, how they perceive the public s risk perceptions and priorities, and how and if they are affected by what they perceive as the public s priorities. These are the questions we investigate in the present paper, in the context of nuclear waste policy. A few studies of politicians risk perceptions and role in safety related policies have been published. Lyons et al. found that local politicians in the UK had considerable influence on safety policy [24]. On the other hand, in an Australian study, members of the public wanted their own policy preferences to play a role, not so much those of politicians [25]. In a study of politicians and administrators in Uppsala County, Sweden, Carter et al. found politicians and the public to have similar objects of worry [26]. Note, however, that worry and risk perception are not synonymous and, in fact, surprisingly little correlated [27, 28]. A few more studies have been published; Barny et al. [29], a study of Rothman and Lichter where many different types of elite groups were investigated [30], and early IIASA work by Otway and Fishbein on administrators and the public with regard to nuclear power attitudes [31-33]. In a Finnish study, it was found that politicians and experts had different value structures with regard to energy options [34]. Gowda [35] argued that politicians are subject to the same biasing factors as the public in their risk assessments. Rundmo and Moen,

4 Risk perception by politicians and public 457 working on transportation risks, found experts to be less concerned about risks than lay people and politicians [36]. Lay people and politicians had similar opinions about risks. It is often believed that societal risk management should involve, in particular, socalled stakeholders. How well do they represent the rest of the population? The answer is that they tend to be more extreme than others, either pro or con an issue such as siting a nuclear waste repository [37]. If their views are allowed to dominate the decision making process it could be seriously distorted. People who are less active may well want to have a say, and given a chance, they will attempt to influence policy decision making. In Sweden, there have been two local referenda about the siting of a nuclear waste repository: in Storuman (1995, 71 percent No, turnout 73 percent) and Malå (1997, 55 percent No, turnout 87 percent). The very high turnout shows that people in general were concerned, although only a few of them had acted as stakeholders. Experts are the third important part in the risk management process, in addition to the public and politicians. Experts provide information to politicians which is crucial in many cases. It is therefore important to discuss their risk attitudes and perceptions. Previous research on group differences in risk perception has been mostly concerned with experts as compared to the public [38], see references and critical discussion in Sjöberg [39] and Rowe and Wright [40]. Typical findings in studies of experts are that they judge the risk as much smaller than the public does. But experts do not judge all risks as small. Sjöberg and Drottz-Sjöberg found, for example, that nuclear waste experts judged non-nuclear risks at the same level as the public did [41]. This was true even of an ionizing radiation hazard such as radon. In an earlier study, Drottz-Sjöberg and Sjöberg found that adolescents risk perception was strongly correlated with their choice of study program in secondary school [42]. Those opting for technology or business programs of study judged risks as smaller than those choosing humanities or social sciences. We pointed out that this finding makes it likely that adult experts with their low judgments of risk probably have retained some of their low risk assessment from early socialization processes, which were apparently at work already long before they acquired their expertise. It is also probable that colleagues, sharing their opinions, will strengthen their initial attitudes once they are adults and employed in professional and expert roles. Sjöberg and Drottz-Sjöberg [41] found that experts, compared to the public, had very different levels of perceived risk within their own area of responsibility, and were much more homogenous. Only about one third of the difference could be explained by demographics and level of education. Attitudes are affected by social influence [43]. One type of influence is due to the tendency to validate one s opinions by comparing them to the opinions of others: social validation. Social validation was a likely factor accounting for group differences and the homogeneity of the experts risk assessments.

5 458 Energy & Environment Vol. 19, No. 3+4, 2008 The present study extended earlier findings in three ways: Locally active politicians specialized in health risk and environment questions were targeted for study. Special emphasis was put on an investigation of how the two groups, politicians and the public, viewed each other s risk perception, and how these beliefs were related to demand for risk mitigation. These risk perception dimensions were related to how priorities were set with regard to different risk management issues. The politicians who were invited to participate in the study were members and deputy members of local Environment and Health Protection Boards (EHBs) 2. The Board Members are appointed by the Municipal Council and represent the various political parties. To give an idea of their responsibilities, some quotations from public information provided by the Stockholm EHB are given: The Stockholm Environment and Health Protection Administration handles issues in the city, addressing questions relating to both the outdoor and indoor environments. The Administration is working to ensure a good environment for the residents of Stockholm, both today and tomorrow. The Environment and Health Protection Administration works to promote an ecologically sound city. We monitor both the external and internal environment in Stockholm, and carry out everything from noise level measurements in residential environments to checks on Stockholm s air quality. The Environment and Health Protection Administration is also responsible for supervising environmentally hazardous enterprises, the use of chemicals, food processing, residential hygiene, animal protection issues etc. We help Stockholmers change their lifestyles. Summing up, the purpose of the present study was to investigate risk perception and related attitudes among politicians and the public, how they viewed each other s beliefs and attitudes, and how priorities for risk mitigation were related to the perception dimensions. In some respects, comparisons will be made with a group of nuclear waste experts. 2. METHOD Samples A random sample, N=1000, of the Swedish population in the ages was approached with a mailed questionnaire. At the same time, all members and their deputies of Environment and Health Boards of 27 municipalities received the same questionnaire. There were persons in each municipality approached in this manner, a total of 550. For response rates, see section on results. 2 In Swedish Miljö- och hälsoskyddsnämnd. In some municipalities, a slightly different terminology is used but we have selected boards, which were as close as possible to the EHB.

6 Risk perception by politicians and public 459 The 27 communities were selected on the following basis: The four largest cities (Stockholm, Göteborg, Malmö, and Uppsala). A few especially interesting communities where siting a nuclear waste repository was contemplated (Malå, Storuman). (These are very small communities with regard to population). A wide sampling covering Sweden geographically and in terms of size of the community. Data from a group of experts had been obtained in an earlier study [41]. The population of nuclear experts was defined as all employees with at least a college degree employed by the regulatory authorities in the fields of radiation protection and nuclear power. Lists of their names were provided by their employers. In addition, we approached industry employees with the same qualifications at one nuclear power plant, at SKB (the Swedish nuclear waste management corporation) and at Vattenfall (a major energy producer, both nuclear and hydro) and members of a professional university network with a special interest in nuclear waste questions. In the group of experts, 237 persons received the questionnaire. See section on results for the response rate. Questionnaire The questionnaire was quite extensive. Only some of the results are reported in the present paper. It was printed in A5 format, and had 31 pages. The following topics were covered: A. 26 hazards, judged on a 0-6 category risk scale. No risk target was specified. Earlier research has shown that such instructions are interpreted as referring to general, as opposed to personal, risk [44]. We refrained from making the issue explicit in the present study, in order to avoid making the questionnaire still more complicated. The list of the hazards is provided in Appendix 1. (1) 3 B. The same 26 hazards, but now judged with regard to what the respondents believed that people living in his or her municipality perceived by way of risk. (2) C. Same as B, but with regard to members of the local EHB board. (3) D. Judgments of 18 technologies on global bipolar scales from very bad (-3) to very good (+3). (4) E. Fishbein-type scaling of 5 radiation technologies: nuclear power in Sweden, irradiated food, local siting of a nuclear waste repository, X-raying, and performing a domestic radon test. For each technology, a number of attributes were judged with regard to how likely they were, and how desirable or undesirable. Bipolar 7-step category scales were used for both aspects. The global attitude towards these 5 technologies had already been judged, in a context with other technologies, in section D. (5-14). The number of attributes 3. Question number in the questionnaire.

7 460 Energy & Environment Vol. 19, No. 3+4, 2008 varied from 6 to 12. They were selected based on an extensive pilot study where people were asked to list advantages and disadvantages of each technology freely. The most salient attributes were then retained. See Ajzen and Fishbein [45] for the procedures recommended for Fishbein scaling. F. Judgments of future (short- and long-range) standard of living and environmental situation in Sweden and the world. (15-20) G. Questions about the present Swedish nuclear power: global attitude, risk and utility (personal and general), future risks of technology and nuclear power, policy attitudes with regard to nuclear power, and social aspects of nuclear power topics. (21-28, 36-37, 40-46) H. Questions about nuclear waste: importance, whether the problems had been solved, risks, and policy questions. (29-35, 47-48, 49-53) I. Confidence in the competence and trustworthiness of agencies and experts with regard to nuclear waste. (54-55) J. 21 psychometric dimensions of nuclear waste accidents, judged on a 0-6 category scale. (56) K. Demand for mitigation through state or local government bodies of the 26 hazards judged in sections A-C, on a 0-6 category scale. (57). L. Demographics, etc.: date, sex, age, education, occupation, a judgment of political decisions versus technology with regard to the development of society, marital status, own children or not, age of children if any, type of residence, size of community, name of municipality, gross household income previous year, zip code, confirmation of respondent being the addressee of the questionnaire, confirmation of having responded alone, confirmation of membership of the local EHB, confirmation of Swedish citizenship, and two questions about whether the respondent would like to participate in further studies. M. Finally, the quality of the questionnaire was evaluated on 9 scales and the time to respond was recorded. Comments were encouraged throughout, and ample space was provided for writing them in the questionnaire. Ratings of the questionnaire assessment scales showed that both the public and the politicians gave the questionnaire and the study a positive evaluation. Methods of analysis Data were analyzed using SPSS version The large sample sizes made it especially important to estimate not only significance of group differences but also the sizes of the differences, using the strategy recommended by Cohen et al. [46]. 3. RESULTS Respondents Respondents were deleted if they stated they were not the person initially addressed or if they were not Swedish citizens. Respondents from the public sample who stated they were members of the local EHB were transferred to the sample of politicians. The

8 Risk perception by politicians and public 461 net response rate in the public was 52.8 % (N=516), among the politicians 59.7 % (N=331). One hundred and thirty-seven members of the groups of experts had replied after two reminders, yielding a response rate of 59 percent. This group will be referred to in the following as the experts. All responses were anonymous. Comparing the respondents from the public sample with national data, we find that they were quite close in terms of most demographic variables. On the other hand, the politicians diverged. They were dominated by highly educated males [65.4% males as compared to 48.6% in the public sample (49.5% in the population), about 40% university education as compared to 19% with tertiary education 4 in the population]. The mean age of the national population was 41.3 years, in the sample of respondents from the public 44.7, politicians Self-rated knowledge about the risk assessment of nuclear waste was higher among politicians than in the public with a mean difference of 0.56 on a scale from 1 to 5, p< according to a t-test. On a standardized scale, the difference was rather large, 0.59 sd-units. Politicians rated the issue of managing nuclear waste as more important than the public did, but the difference was modest, 0.20 in sd-units. Among politicians, 85 % rated the issue as very or extremely important. The corresponding figure for members of the public was 77 %. Hence, the issue was rated as quite important by both groups. The high response rates should be seen in the light of this result. Fig. 1. Mean risk ratings, politicians and the public. 4 The national statistics do not distinguish university education from other forms of tertiary education.

9 462 Energy & Environment Vol. 19, No. 3+4, 2008 Level of perceived risk and demand for mitigation Mean risk ratings by politicians and members of the public are reported in Fig. 1. It is seen that they were quite similar. There were some exceptions, however, radiation from a nuclear waste repository being one. On the other hand, politicians were apparently not aware of the similarity between their own risk perceptions and those of the public, see Fig. 2. The figure shows that politicians underestimated the risks as perceived by the public, when it came to some large risks, mostly lifestyle risks. They overestimated the publics perceived risks for some small risks associated with crime. Fig. 2. Mean risk ratings by the public, and politicians mean ratings of the public s risk ratings.

10 Risk perception by politicians and public 463 Fig. 3. Mean demand for risk mitigation, politicians and the public. Demand for risk mitigation was very similar over-all in the two groups, see Fig. 3. Table 1 provides more detailed data on perceived nuclear waste risks. Table 1. Mean ratings of nuclear waste risks, scale range 0-6. Perceived risk, scale Rated risk perception Demand for risk range 0-6 of the public 1, scale mitigation, scale range 0-6 range 0-6 Hazard Public Politicians Public Politicians Public Politicians High level nuclear waste from Swedish nuclear power plants Nuclear waste in transport Radiation from anuclear waste repository Note 1. Refers to ratings made under the instruction to rate how other members of the public probably perceived the risk.

11 464 Energy & Environment Vol. 19, No. 3+4, 2008 The table shows that politicians judged these risks as smaller than the public did. The differences between politicians and members of the public with regard to perceived risk were all statistically significant, according to t-tests. Members of the public and politicians had similar views about the risks of nuclear waste as seen by the public. The groups were also similar in level of demand for mitigation. Interestingly, both groups believed that the public in general regarded these risks as worse than they did themselves. The differences were significant in two of the three cases, both for members of the public and for politicians. The significant differences occurred for nuclear waste in transport and for radiation from a nuclear waste repository. It is noteworthy that politicians saw the nuclear waste risk as smaller than the public did, yet had the same level of demand for mitigation. Further analysis of demand for mitigation is therefore called for. The average rated demand for mitigation for all hazards, except the three nuclear waste items, was used as an explanatory variable, as well as perceived risk and the ratings of the risk that the respondent believed that others perceived, both for the public and for politicians. The rationale for using the average rated demand for mitigation as an explanatory variable was that respondents tended to use the rating scale in different ways [47, 48]. Some rated mitigation demands as high throughout, others as low. The Cronbach alpha coefficient for the index formed by computing the average demand was Regression analyses of demand for mitigation of nuclear waste risks are reported in Tables 2 and 3 for the public and politicians, respectively. Table 2. Results of regression analyses of demand for mitigation of nuclear waste risks against selected explanatory variables, standardized regression coefficients and adjusted multiple correlations. Data from the public (N=516). Dependent variable Demand for Demand for Demand for mitigation of the risk mitigation of the risk mitigation of the of high level nuclear of nuclear waste risk of radiation waste from Swedish in transport from a nuclear nuclear power plants waste repository Explanatory variable Standardized regression coefficients Respondents average *** *** *** level of mitigation ratings for non-nuclear hazards Respondents own *** *** *** perceived risk ratings Respondents ratings of the wider public s perceived risk Respondents ratings of politicians perceived risk 1 R 2 adj Note 1. All respondents had been instructed to rate how they believed that politicians perceived the risks of each of the hazards.

12 Risk perception by politicians and public 465 Table 3. Results of regression analyses of demand for mitigation of nuclear waste risks against selected explanatory variables, standardized regression coefficients and adjusted multiple correlations. Data from the politicians (N=331). Dependent variable Demand for Demand for Demand for mitigation of the risk mitigation of the risk mitigation of the of high level nuclear of nuclear waste risk of radiation waste from Swedish in transport from a nuclear nuclear power plants waste repository Explanatory variable Standardized regression coefficients Respondents average *** *** *** level of mitigation ratings for non-nuclear hazards Respondents own *** *** perceived risk ratings Respondents ratings of the wider public s perceived risk Respondents ratings of politicians perceived risk 1 R 2 adj The tables show that one s own perceived level of risk was a major factor in demand for show. Average rated demand for mitigation was even more important. This was true for both politicians and members of the public. The multiple correlations were high, showing that demand for mitigation of the nuclear waste risks could be well explained by the models. Average rated demand for mitigation was even more important Summing up, this section has shown that: 1. Politicians and the public had, overall, quite similar risk perceptions, with some exceptions, nuclear waste being one. In this case, politicians rated the risk as smaller than the public did. 2. Politicians systematically erred in estimating the public s risk perception. Lifestyle risks were seen as larger by the public than the politicians expected. The opposite was true for, e. g., some crime related risks 3. Both groups erred in overestimating the public s level of perceived risk in the case of nuclear waste 4. Demand for risk mitigation was similar in the two groups 5. Demand for risk mitigation of nuclear waste risks was, for both groups, mainly related to their own perceived risk, not the risk they estimated that others perceived

13 466 Energy & Environment Vol. 19, No. 3+4, 2008 Trust The respondents were asked to rate their trust in the competence of a number of organizations or groups, and the trustworthiness of their risk assessments. See Figs. 4 and 5. Fig. 4. Mean trust ratings in the risk assessment by various actors with regard to nuclear waste management, politicians and the public. Fig. 5. Mean trust ratings in competence of various actors with regard to nuclear waste management, politicians and the public.

14 Risk perception by politicians and public 467 These figures tell essentially the same story. Politicians had more trust in other politicians, authorities, and university experts. The public had more trust in industry. In t-tests of the group differences, it was found that they were significant for all rating variables, except for dissident experts. The literature on trust usually does not deal with any other types of trust than social trust. However, some people also trust or distrust the science and types of technology, which form the basis of risk management. In several studies, it has been found that trust in science, epistemic trust, is a more potent determinant of risk perception than social trust [49, 50]. Three survey items were used to form an index of epistemic trust: belief in unknown effects of a repository, a judgment that the risk was unknown by science, and the assessment of whether present solutions to the repository problem are sufficiently developed and trustworthy. These three items had an alpha of Social trust indices were formed based on judgments about competence and trustworthiness of risk assessments, deleting items about dissident experts and politicians. The alpha values of the latter two indices were 0.84 and 0.80, respectively. There was a small, but significant (p=0.007) difference between politicians and the public with regard to epistemic trust. The mean difference was 0.18 in standardized scores. Politicians had a higher value with regard to epistemic trust of epistemic trust than the public. Trust is often considered an important determinant of perceived risk. Tables 4 and 5 give the correlations between the pooled ratings of nuclear waste risks and the trust judgments. Since politicians perceived their own knowledge to be better, these correlations should be more negative for them [51]. Table 4. Pearson correlations between perceived nuclear waste risks and ratings of competence. Public Politicians Experts employed by authorities ** ** University experts ** ** Nuclear industry experts ** ** Dissident experts 0.24 ** 0.24 ** National politicians ** Municipal politicians * 0.09 Authorities ** ** Nuclear industry ** ** Nuclear waste management personnel ** ** * p<0.05, ** p<0.01

15 468 Energy & Environment Vol. 19, No. 3+4, 2008 Table 5. Pearson correlations between perceived nuclear waste risks and ratings of the trustworthiness of risk assessment of nuclear waste. Public Politicians Experts employed by authorities ** ** University experts ** ** Nuclear industry experts ** ** Dissident experts 0.26 ** 0.30 ** National politicians ** Municipal politicians * 0.00 Authorities ** ** Nuclear industry ** ** * p<0.05, ** p<0.01 Epistemic trust correlated more strongly than social trust with perceived risk: for the public and for politicians. Both these more than social trust correlations were highly significant, p< There were several interesting findings with regard to trust and perceived risk: In several cases, expected and moderately high negative correlations between social trust and perceived risk were found. Correlations between epistemic trust and perceived risk were higher than correlations between social trust and perceived risk. In some cases, there was only a very low or zero correlation. These were cases where the ratings pertained to local or national politicians. In one case (dissident experts), the correlation was positive. Politicians gave higher correlations than members of the public did, not lower. The literature on social trust and perceived risk usually does not deal with the third and the fourth cases in the list above. It is usually taken for granted that the message is reassuring, and therefore the correlation in the list above between trust and risk should be negative. The category of dissident experts is an example of a group expected to deliver the opposite message. Therefore, the correlation between trust and perceived risk should be positive in that case. It is also taken for granted that that their message is unambiguous. However, politicians as a group are heterogeneous. Trust in politicians was unrelated to perceived risk, as could be expected since different politicians deliver different messages. Finally, the expected finding of higher correlations in the group with the lowest level of self-rated knowledge did not occur. The opposite was the case. Structure of perceived nuclear waste risks There are two different aspects of risk perception: level of perceived risk, treated in the previous section, and structure, i.e. which factors determine the level of perceived risk and how are they weighted. The present section deals with structure. The 21 risk aspects of nuclear waste were factor analyzed. Four factors accounted for 66 percent of the variance. After Oblimin rotation, they were readily interpreted as Dread, New Risk, Disaster Risk, and Interfering with Nature. Indices were formed as follows:

16 Risk perception by politicians and public 469 Dread: 5 items, alpha 5 = 0.87 New Risk: 3 items, alpha= 0.79 Disaster Risk: 4 items, alpha= 0.93 Interfering with Nature 7 items, alpha= 0.90 The first three risk dimensions are the same as the original dimensions of the Psychometric Model [23]. The fourth dimension has also been called Tampering with Nature [53]. Dread, as defined traditionally, is not a dimension measuring only a certain type of emotional reactions. Only 1 of, as defined traditionally, the 5 items is about emotions [54]. We still chose to retain the traditional name and contents of the factor. Perceived nuclear and contents waste risk was estimated as the mean risk rating given to three of the items in the list of 26 hazards. The alpha value of the risk index was The items were: High level nuclear waste from Swedish nuclear power plants Nuclear waste in transport Radiation from a nuclear waste repository In order to model perceived nuclear waste risk, we also computed the mean risk rating of all other hazards. The Cronbach alpha of this index variable, called Risk Sensitivity, was The rationale is the same as in the previous section where mean rated demand for risk mitigation was used as an exploratory variable. In addition, an index of social trust was formed using 11 items of ratings of trust in competence and risk assessment of various actors. Dissident experts and politicians were excluded from this index. Cronbach s alpha was Table 6 gives the results of regression analyses of perceived nuclear waste risk with the four dimensions as explanatory variables. Table 6. Results of regression analyses of the perceived risk of nuclear waste. Standardized regression coefficients Explanatory variable Public Politicians Gender *** Age Level of education New risk * Disaster risk Dread Interfering with Nature * * Social trust *** *** Epistemic trust *** *** Risk sensitivity *** *** R 2 adj for all explanatory variables R 2 adj for 3 dimensions comprising the Psychometric Model and social trust * p<0.05 *** p< Cronbach s alpha, a measure of the reliability of indices measuring attitudes and similar psychometric variables [52]. Alpha varies between 0 and 1; values at or above 0.6 are usually considered to be acceptable.

17 470 Energy & Environment Vol. 19, No. 3+4, 2008 The table shows that the most important explanatory factor was Risk sensitivity, while trust and Interfering with Nature also were important factors. The traditional dimensions (Dread and New risk) had almost no explanatory power when entered together with Risk sensitivity and Interfering with Nature. The Psychometric Model in its traditional form with social trust added had some explanatory power, as seen by the last row of the table, but only about percent of the power of the models, which included Epistemic trust, Risk sensitivity, and Interfering with Nature. The table shows that politicians and members of the public displayed a very similar structure of their perception of the perceived risk of nuclear waste. Summing up, this section has shown: Members of the public and politicians had very similar structures of perceptions of risks associated with nuclear waste The most important factors in perceived risk were Risk Sensitivity, trust (both social and epistemic) and Interfering with Nature The traditional approach, comprising the Psychometric Model and social trust, only reached percent of the level of the models which comprised the additional variables defined in the present paper Attitude to a nuclear waste repository Attitude scaling Attitude to a siting of a nuclear waste repository in one s municipality was rated on a scale from 3 (very bad) to + 3 (very good). It was also measured by a question about the respondent s intended choice in a future local referendum about siting 6. The results are given in Figs. 6 and 7 7.The figures show that politicians were less negative to a local repository than the public was. The differences were significant in both cases. For attitude,. 2 (6)=21.96, p= For intended vote, χ 2 (5)=13.77, p= There were 12 salient aspects of a nuclear waste repository, as found from preliminary study of such aspects according to standard Fishbein scaling procedures. They were: Improved local economy More jobs locally Solidarity with country Community known & respected Radiation injury to humans Radiation injury to nature People would often be worried Radiation injury to future generations 6 A few respondents stated that they would not participate in such a referendum, as shown in the figure. 7 This figure also includes data from a group of experts, to be discussed in a later section.

18 Risk perception by politicians and public 471 Community would get a bad reputation Lower real estate values Waste could explode like a bomb Effects unknown today Fig. 6. Frequency distribution of attitudes to a local nuclear waste repository, politicians and the public. Fig. 7. Frequency distribution of intentions to vote about a local nuclear waste repository, experts, politicians and the public.

19 472 Energy & Environment Vol. 19, No. 3+4, 2008 The Fishbein attitude model assumes that the attitude is accounted for by the sum of products of ratings of beliefs and values, taken across the aspects under study. In the present case, there were 12 salient aspects, listed above. The model has the form of a traditional expected value model. The sum of products of values and beliefs correlated with the global rating of bad-good (see above): 0.45 for the public and 0.61 for the politicians. These are normal fits for the Fishbein attitude model, maybe even a little better [55]. The difference between the two correlation coefficients is significant (p=0.005, two-tailed). Apparently, the model gives a fuller description of politicians attitudes than the public s attitudes. Fishbein scaling allows for a more penetrating analysis since data pertain to both probabilities and values, see Figs. 8 and 9. Fig. 8. Mean value ratings of aspects of a local nuclear waste repository, politicians and the public. There were significant (but small) differences between groups in both probabilities (7 of 12 cases) and values (9 of 12 cases). Politicians rated probabilities for positive values as larger and probabilities for negative values as smaller than the public did. They tended to rate the negative values as worse and the positive ones as better.

20 Risk perception by politicians and public 473 Fig. 9. Mean probability ratings of aspects of a local nuclear waste repository, politicians and the public. Scale from 1=very unlikely to 7=very likely. Summing up, the present section has shown: Politicians had a less negative attitude to a local repository than members of the public, but they were still negative Attitudes of both groups could be well modeled with data on values and probabilities of 12 salient aspects Politicians less negative attitudes were accounted for both by their beliefs and values Communication and social validation Trust is presumably of importance because people communicate. The effects of communication about an issue should depend not only on trust, but also on the contents of communication (pro or con nuclear power), and on its extent or frequency. The present data provided information about attitude to nuclear power, the perception of friends attitudes to nuclear power, and frequency of communication about the issue. A two-way ANOVA with attitude to nuclear power as dependent variable, friends attitude and frequency of communication as independent, showed significant main effects and a significant interaction. For friends attitude, F(2,772) 8 =166.84, p< For frequency of communication, F(4, 772)=2.14, p= For the interaction, F(8,772)=1.59, p= The model explained 37 percent of the variance 8 F ratio and degrees of freedom.

21 474 Energy & Environment Vol. 19, No. 3+4, 2008 of nuclear power attitude, which is remarkably high in view of the few explanatory variables. The trends were similar for politicians and members of the public analyzed separately, see Fig. 10. Fig. 10. Mean attitude to nuclear power as a function of friends attitudes and frequency of communication, politicians and the public.

22 Risk perception by politicians and public 475 The findings on friends attitudes show that the contents of communication were crucial, as well as frequency of communication. The findings suggest that friends attitudes are powerful determinants of the nuclear policy attitude. A process of social validation can be assumed to take place, and it shapes attitudes. It is also conceivable that trust interacts with both friends attitudes and frequency of interaction since we usually choose to communicate with people who share our views, and trust them more because of a common worldview and values. Further work on these issues would be interesting, putting trust in a perspective of communication. Experts, politicians and the public Experts provide important information about nuclear waste risks. Yet, there is typically a large gap between the publics risk perception and the risk assessment communicated by the experts. Few studies have, so far, investigated politicians perceived risk and compared to that of experts. See Fig. 11, which gives data from the public, politicians, and experts in response to a question 9 about the proposed solution to long-term management of nuclear waste. Fig. 11. Frequency distribution of answers to a question about whether the problem of nuclear waste management had been solved, experts, politicians and the public. 9 The question was: Is it your opinion that the problems associated with the final storage of nuclear waste have been satisfactorily solved?

23 476 Energy & Environment Vol. 19, No. 3+4, 2008 It is very clear that experts differed strongly both from both the public and politicians. The analyses reported above on mitigation suggested that politicians were guided by their own risk beliefs, and not by an opportunistic tendency to accommodate themselves to what they believed that the public wants. The question therefore remains why the public and the politicians were so strongly different from the experts. Sjöberg and Drottz-Sjöberg [41] found that about one third of the difference was accounted for by demographics, the rest by trust both epistemic and social - and attitude to nuclear power. The same social validation process, which takes place in a group of lay people or politicians, probably happens also to experts. In the cited study, the experts were found to be much more homogenous in their risk perception and attitudes than lay people were. They work in a social environment where a large majority agrees that risks are small and that the problem of nuclear waste management has been solved. The large gap between experts and lay people, found in many studies as well as the present one, probably is a function of social validation, and to some extent of demographics and level of education. Experts are most often men and they come from a relatively high socio-economic stratum. 4. DISCUSSION A methodological point of some importance is raised by the fact that the politicians studied here were mostly highly educated males, clearly different from the distribution of the sample from the public. Both gender and education are known to be correlated with risk perception and attitudes. Yet, many of our results show remarkable similarities between the two samples. In addition, we wanted to describe the attitudes and risk perceptions of this group, not necessarily explain them. That is a question for future work 10. The public sample was reasonably representative, as far as could be determined. There is no reason to suspect that the present data are misleading and the response rate is respectable when compared with much international work where mailed questionnaires, often very short, seem often to give 40 percent or less in response rate. Several published risk perception studies from leading groups in the field have been based on data with less than 30 percent response rate; see the discussion by Sjöberg and Drottz-Sjöberg [41]. Much work, of course, has been based on convenience samples. Others use quota samples. The present group of respondents from the public is well in line with a regular quota sample, since they closely resembled the national population in almost all pertinent respects. As noted above, there was a bias towards a high level of education among the respondents, about 12 percent too many. However, educational level was marginally related to risk perception, a typical result. Perceived risk correlated on the average -0.1 with level of education, to name just one example. The present results therefore possibly show a somewhat too low level of perceived risk, but the bias is probably not important. 10 ANOVA s showed that sex and level of education had some importance in accounting for risk perception, but the distinction between politicians and the public was more important. Thus, the effects of the latter distinction were not absorbed by background data. Interactions were not important.

24 Risk perception by politicians and public 477 In a study of politicians and administrators in Uppsala County, Carter et al. found politicians and the public to have similar objects of worry [26]. Note, however, that worry and risk perception are not synonymous and, in fact, surprisingly little correlated [27]. Worry is frequently considered as affective risk perception. Carter et al. also found that administrators in local EHB offices gave the lowest worry figures. The politicians investigated were not specialized in EHB work but members of municipal or county boards. No data exist, as far as we know, on members of parliament and their risk perception, but the county board members in a sense represent a step towards a higher level of responsibility. Risk perception by the public and politicians was quite similar in the present study, for most hazards. This finding shows that politicians who were members of EHB boards in the present study did not differ much from the public in perceived risk - in that respect they were much different from experts. At the same time, they did differ in having somewhat higher trust in politicians in general, and they saw themselves as more knowledgeable than the public did. The politicians also differed from the public in being more interested in economic aspects of technology, not mostly the risks as the public was. Yet, this is a relative statement - politicians were quite attuned to the risk dimension, just as the public. The politicians did not have a very correct view of the risk concerns of the public. They thought the public was less worried about lifestyle risks than they in fact were, and more worried about crime related risks than they in fact were. So much for hazards in general. For nuclear waste, the picture was different. One reason could be that most of the other hazards were more likely to be under current consideration for management by EHB boards than nuclear waste. The politicians rated the risks as lower than the public did, yet expressed a similar than nuclear waste policy attitude. This was paradoxical but could be due to a higher level of precautionary attitude among the politicians [56]. Both groups overestimated the risk as seen by the public. Politicians showed a larger bias of this kind than did members of the public, and so did those who had a more positive attitude. Part of the reason for these findings may have been a larger salience of negative attitudes in the community. For example, media tend to attend more to opponents than to proponents. When it came to the structure of perceived nuclear waste risk, it was found to be similar for politicians and the public. In both groups, regression analysis showed that the traditional dimensions of the Psychometric Model were less powerful as explanatory factors than trust and Interfering with Nature. The power of the latter dimension has been documented in several previous studies; see e.g. Sjöberg [53]. A distinction was made with regard to social and epistemic trust. The latter type of trust refers to trust in science, as distinct from trust in people and organizations. It was found to be somewhat a more powerful factor in perceived risk than social trust was. This finding is in good agreement with previous work on social and epistemic trust in relation to risk perception [49, 50, 57]. Trust probably interacts with communication: we interact more often with people who share our views, we trust them more, and our own views are strengthened by social validation. This is a process which might explain the remarkable differences

25 478 Energy & Environment Vol. 19, No. 3+4, 2008 between experts on the one hand, and politicians and members of the public on the other. What are the implications of the present study for nuclear waste policy? Several findings should be stressed: The politicians studied tended to base their policy attitudes on their own risk perception, not on the views they believed that the public had. Perceived risk was, for both politicians and the public, mainly based on beliefs about the risk generating process, and on moral notions not on the expected emotional reactions of the public ( dread ), on the novelty of the risk, or the expectations of large disasters. The important aspect of trust was trust in science, which correlated more strongly with perceived risk than social trust did. Risk perception was strongly related to social validation, i. e. the views believed to be held by others. The results of the study suggest that perceived risk of nuclear waste is like a preference: it is a personal matter. People listen, as it were, to their own convictions, whatever their base, and not readily to experts or representatives of authorities or industry. They probably pay more attention to friends and family members. There is a rational basis for listening to non-experts: science does not have all the answers. Politicians are not different from members of the public, and the result is a deep conflict about risk management. This conflict exists in most communities, but there are exceptions. In Sweden, people in two municipalities have preliminarily accepted the siting of a repository. Their attitudes and risk perceptions are remarkably different from the rest of the country [58]. Part of the reason seems to be that many are employed in the nuclear industry, but that is not the whole explanation [59]. A dominant positive view has evolved, and over time it is becoming even more positive [60]. Social validation probably reinforces the positive view and makes it even more dominant. Longitudinal studies of the development of the attitudes in a community are called for in order to verify these interpretations and to understand the dynamics of acceptance/rejection better. REFERENCES 1. Beck, U., Risk-Society. Towards a New Modernity, Sage, London, Giddens, A., Modernity and Self-Identity. Self and Society in Late Modern Age, Polity Press, Cambridge, Sjöberg, L., af Wåhlberg, A., and Kvist, P., The Rise of Risk: Risk Related Bills Submitted to the Swedish Parliament in and , Journal of Risk Research, 1998, 1(3), Ball, D.J., Deliberating over Britain s Nuclear Waste, Journal of Risk Research, 2006, 9(1), Ball, D.J. and Boehmer-Christiansen, S., Societal Concerns and Risk Decisions, Journal of Hazardous Materials, 2007, 144(1-2),