Attitudes as an Expression of Knowledge and “Political Anchoring”: The Case of Nuclear Power in the United Kingdom
European Institute, London School of Economics, London, UK.
Abstract
Attitudes toward the use of nuclear energy pose fundamental issues in the political debate about how to meet future energy needs. Development of new nuclear power facilities faces significant opposition both from knowledgeable individuals who display an understanding of the risks attached to various forms of power generation and those who follow strict politically based ideological dogmas. This article employs data from a 2005 Eurobarometer survey of UK citizens to examine the influence of both political preferences and knowledge in explaining support of nuclear power. Findings reveal that attitudes about nuclear power are highly politically motivated while the influence of knowledge about radioactive waste is dependent upon beliefs about the consequences of nuclear energy use. Perceptions of being informed about radioactive waste and trust in sources providing information about radioactive waste management also predict attitudes toward nuclear power generation.
1. INTRODUCTION
In the face of rising global energy prices along with increasing concerns over future reliance on foreign oil imports and pressures to meet carbon emissions targets, the UK government released a review of its long-term energy policy, including an examination of the feasibility of all potential energy sources, in July 2006 (DTI, 2006a). The UK Energy Review Report suggests that the country meet demand partly by building nuclear generation capacity so that nuclear power supplies more than the current level of 19% of the UK's electricity generation (DTI, 2006a, 2006b).
The UK government's acknowledgment of nuclear energy's importance as part of a diverse energy policy has been met with the expected public outcry. Criticisms range from concerns about radioactive waste management to nuclear power plant siting to the vulnerability of nuclear facilities to terrorist attacks. Even in the face of facts stating the relative safety of nuclear energy production versus other forms of energy generation (coal mining, gas drilling) and against other risk-related contexts, individuals maintain serious concerns about nuclear energy generation risks relative to other risks (Slovic, 1987).
Previous public opinion polls in the United Kingdom about nuclear energy have found substantial unease regarding nuclear power generation. Thirty-seven percent of respondents in a You Gov/Economist (2006) poll opposed the building of new nuclear energy facilities in the United Kingdom while 47% demonstrated opposition in a 2006 ICM/Sunday Telegraph Poll. Survey results have also pointed to the influence of political stance on attitudes toward nuclear energy, with conservative voters being more likely to support nuclear power policy (McGowan & Sauter, 2005; ICM Survey/Sunday Telegraph, 2006; YouGov/Economist, 2006).
Support for nuclear power is likely to vary depending on individuals' political views. The use of nuclear power (compared to alternative energy sources) is a politically sensitive issue such that individuals adjust their attitudes and perceptions to the underlying values held by their political position or party affiliation. The party affiliation then becomes their anchor from which to evaluate risks. Therefore, judgments result from the employment of a heuristic, which acts as an anchor around which to adjust views when confronted with new information (Tversky & Kahneman, 1974). This anchoring heuristic becomes evident in cases where people are known to anchor on a specifically mentioned fact (e.g., how many people were injured from the Chernobyl accident) even when they know it is irrelevant to the decision-making process (Tversky & Kahneman, 1982). In the context of nuclear energy, we hypothesize that due to inadequate knowledge and the employment of heuristics (or political anchors) when thinking about nuclear energy risks, individuals rely on their political affiliation to determine their attitudes about nuclear energy generation
This article draws upon data from the June 2005 Eurobarometer study on individuals' support for nuclear power using multivariate regression (Eurobarometer, 2005). When examining the determinants of support for nuclear power, we use the terms “support for” and “attitudes about” nuclear energy interchangeably as both reflect individuals' opinions about nuclear power resulting from what they know and the beliefs they have about the topic. To the authors' knowledge, no prior studies using a large and representative sample that look at the role of political views and knowledge in determining attitudes about nuclear energy employ statistical controls to take into account the sample selection resulting from respondents answering “don't know” to our response variable. We contribute to the literature by exploring the influence of knowledge about radioactive waste, perceived knowledge about radioactive waste, beliefs about the consequences of nuclear energy, and politically anchoring on attitudes about nuclear power. We argue that the employment of a politically related “anchor” gives rise to an unobservable attitude or belief about energy alternatives. A further objective of this article is to examine whether beliefs about the consequences of nuclear energy use mediate the determinative role of knowledge on support for nuclear energy.
The following section conveys the conceptual background behind the analysis. We describe the data and methods in Section 3. Section 4 reports the results and Section 5 concludes.
2. BACKGROUND
Nuclear-power-related risks can be described as low-probability, high-costs events that impact not only the environment, but also society collectively with repercussions beyond the individual. According to the well-known cognitive bias that individuals tend to overestimate the risks of low-probability, high-cost events (Slovic et al., 1977; Kunreuther & Slovic, 1978), we would expect individuals to overestimate the risks of nuclear power generation.
The nature of nuclear-power-related risks being collective rather than individual constitutes an important characteristic involved in assessing this type of risk. Individuals' perceptions of nuclear power generation may be influenced by the idea that nuclear power causes negative externalities by reducing the welfare of others and possibly that of future generations as well (Frank, 1989). Where individuals feel that risk extends beyond them to include family members and members of their larger social circle, they might exhibit more concern about an event taking place.
Energy sources can become topics of discussion and thus information and opinion pooling among the public due to the potential environmental impact and collective nature of risks attached. However, such collective decisions are highly influenced by limited individual knowledge, experience, and foresight about potential effects. Concurrently, the development of political ideologies such as environmentalism anchored in beliefs and knowledge about the environmental effects of public policies suggests that political affiliation guides attitudes. With the growth of ideologies such as environmentalism having an impact across the political spectrum, individuals become more likely to anchor their attitudes based upon their political bent. Therefore, attitudes become more of a political expression, perhaps of discontent, rather than the result of a reasoning process. According to Katz's (1960) framework, values are expressed as a function of attitudes over other potential explanatory factors such as knowledge. This feature is reinforced by the fact that environmental setting largely influences what political views look like. As a consequence, opinions derived from political views may only be an expression of an environment of content or discontent with a policy rather than specific individual preferences.
2.1. Conceptual Framework of Attitude Formation
Individuals' develop their attitudes and risk perceptions as a result of the way their moral norms, sense of group identity, and other endowments enter into their risk assessment setting process. Fig. 1 demonstrates the process of formulating attitudes from starting with a set level of endowments, including information, preferences, demographics, experience, and perceived control, to the roles of trust in assessing new information to the move from formulating risk perceptions to expressing an opinion about policy.
Rational decisionmakers' formulations of risk perceptions.
Source: Authors' own.
When thinking about how individuals progress along the pathway shown in Fig. 1 and eventually come out in support of or in opposition to nuclear power policy, two options appear as follows.
First, some individuals may provide an emotional response that can be qualified as “rational irrationality” or a single-peak attitude (Caplan, 2001; Costa-Font & Mossialos, 2006). Both positive and negative emotions appear strongly correlated with attitudes about nuclear waste (Sjoberg, 2007). In this article, we suggest this rationally irrational response hinges on political stance because of the clear positions political parties take on environmental issues and political parties' abilities to step in as information sources when individuals do not think they have enough information about a subject to make a decision entirely for themselves. Previous literature points to the important role of ideological and political beliefs in the development of risk perceptions on many subjects (Sjoberg & af Wahlberg, 2002). The irrational decisionmaker thinks about the issue in the absence of a cognitively robust process such that new information would be ignored or interpreted in an alternative manner than that which the new information actually suggests because of the presence of existing information that the decisionmaker has committed to believing. The individual sticks to a certain dogma rather than undertake a careful reasoning process.
(1)
(2)Equations (1) and (2) highlight how an individual's attitude or “planned action” would result from the following process. Risk (benefit) perceptions can be conceptualized as the aggregation of a vector of some preexisting estimate of nuclear power risks β0(α0), the influence of each jth information sources or control variables on the ith individual Xij captured by the parameter, β1(α1), and the influence of treatment variables, namely, political attitudes or knowledge Zi, captured by the parameters β2(αs).
(3)This suggests that attitudes are either emotionally based or the result of a rational reasoning process. Yet the sources of emotional attitudes are not well known and often subject to ample debate. Particular beliefs (e.g., the value of environment preservation) that can result from the adoption of a political attitude provide reasons that influence intentions and behavior (Westaby, 2005). If one assumes that decisions result from a rational process, then information and knowledge will determine precision in subjectively estimating the risks and benefits of nuclear power. Components such as gender, age, and socioeconomic factors play an important role by mediating the credibility an individual attaches to information sources, chiefly trust in government and stakeholders involved in nuclear power generation.
2.2. Previous Research on Nuclear Power
Public fear about nuclear power stands as a significant obstacle in permitting governments to pursue a course of energy strategy that includes nuclear energy (Slovic et al., 1991; Rothman & Lichter, 1987). Altering risk perceptions about nuclear technology remains difficult as perceptions appear to be divorced from technical risk assessments (Slovic, 1993) and more aligned to political beliefs or world views (Bouyer et al., 2001; McGowan & Sauter, 2005; Slovic & Peters, 1998) or feelings of dread (Slovic et al., 1979, 1991).
Within the context of nuclear energy (Greenberg et al., 2007), feelings of worry or dread have been found to be strongly associated with lack of trust in state protection against risks while trust in the environmental management of risks has been found to be linked to previous experience with the risk and sociodemographic and economic factors (Williams et al., 1999). Examining dread factors often includes an understanding of personality traits, which this study is constrained from doing because of a lack of information about personality traits apart from those manifested in a very general sense through sociodemographic variables.
Knowledge also appears to impact attitudes and risk perceptions. Sjoberg and Drottz-Sjoberg (1991) found that power plant workers who had less knowledge about radiation risks perceived higher levels of risk. Experts in the field of nuclear power are more likely to have lower risk perceptions about the topic (Roberts, 1990; Fischhoff et al., 1978). When individuals feel less informed about nuclear energy, their risk perceptions are less likely to translate into policy preferences of the expected leaning (Gerber & Neeley, 2005).
Sandquist (2004) asserts that individuals overestimate risks associated with nuclear power plants in reference to calculated actuarial risks mainly because of their lack of control with regard to nuclear energy safety. As a corollary, if individuals have trust in those actually operating nuclear plants, then control becomes less of an issue. Slovic et al. (2000) analyzed data showing that the French have similarly high risk perceptions regarding nuclear power as Americans; however, the French see a greater need for nuclear energy and trusted those officials in charge of the nuclear energy more than Americans. Poortinga et al. (2006) found that individuals are ambivalent about whether the British government regulates the nuclear power industry adequately.
On the other hand, trust in the government does not necessarily imply perception of government technical competence (Gerber & Neeley, 2005). Slovic (1993) points out that individuals' trust in doctors eliminates high risk perceptions when encountering medical technologies such as X-rays that carry radiation risk. In the same way, lack of trust in government and industry representatives that manage nuclear power plants results in higher perceptions of risks and lower perceptions of benefits regarding nuclear power (Flynn et al., 1992; Pijawka & Mushkatel, 1992).
Individuals who have more experience with nuclear power plants either by working at one or having one nearby their home are more likely to have lower risk perceptions than the general public about plant safety (Kivimaki & Kalimo, 1993; Lima, 2004; Taylor & Brown, 1988, Fischhoff et al., 1978). Sjoberg and Drottz-Sjoberg (1991) found that less-experienced power plant workers such as contract employees had more concerns about occupational safety than regular plant workers. The public's direct experience with nuclear power is small and therefore may lead it to overestimate risks.
Like other risks where women appear more risk averse than men (Savage, 1993; Slovic, 1999; Dohmen et al., 2005), females tend to view nuclear energy risks as higher (Davidson & Freudenberg, 1996) and benefits as lower than men (Sjoberg et al., 1999).
2.3. Research Questions
There are three questions to be assessed. First, do individuals' perceptions about their own level of knowledge regarding radioactive waste impact their support for nuclear energy (“certainty effect”RQ1)? The explanatory variable “not informed” being significant would point to the existence or lack of a certainty effect. Second, how does individuals' knowledge about radioactive waste impact their attitudes toward nuclear policy and is this influence of knowledge mediated by beliefs about the consequences of nuclear energy use (“knowledge effect”RQ2)? Examining explanatory variables capturing knowledge about radioactive waste (“knowledge waste”) as well as the presence of variables capturing beliefs (under the heading “beliefs in the consequences of nuclear energy use”) would answer this question. Third, to what extent do individuals' attitudes about nuclear power align consistently with their political views (“political anchoring”RQ3)? Explanatory variables denoting political preferences (under the heading “political anchoring”) being significant would denote political anchoring.
We hypothesize that individuals feeling more informed about radioactive waste will be more likely to support nuclear power, thus exhibiting a certainty effect. We also expect that knowledge about radioactive waste will have a determinative role in individuals' support of nuclear power (Roberts, 1990; Fischhoff et al., 1978) but its influence will be tempered by beliefs. Finally, we hypothesize that attitudes toward nuclear power generation will be polarized overall and individuals against nuclear power generation will be concentrated in extreme disagreement rather than having a centrist point of view (Sjoberg, 2003b). We also expect to see evidence of political anchoring, with political views of all inclinations significantly determining attitudes.
Regarding other explanatory and control variables included in our models, we also expect that men (Sjoberg et al., 1999) will be more likely to express support for nuclear power. Based on the findings of Flynn et al. (1992) and Pijawka and Mushkatel (1992), we also predict that trust in information sources and regulatory bodies will be positively associated with support for nuclear power. We do not expect education, age, or respondent cooperation to act as significant determinants. We elect to include respondent cooperation in our models, as measured by interviewers assessing respondents as offering excellent, fair, average, or bad cooperation, because it controls for whether the interviewer has been successful at his or her job in achieving quality data.
3. DATA AND METHODS
3.1. Data
This study uses UK (excluding Northern Ireland) data from the Special Eurobarometer 227 report on radioactive waste. Data were collected for the entire EU-25 over the period February–March 2005 and the survey was published in September 2005. One thousand nine UK (ex-Northern Ireland) residents aged 15 and over were asked questions about their support for nuclear energy, their knowledge about radioactive waste, and their beliefs about the consequences of nuclear energy use as well as their political affiliations and trust in sources disseminating information about radioactive waste.
The Eurobarometer survey series is a social research tool of cross-national surveys that has been undertaken by the European Commission since the 1970s and is designed to regularly monitor social and political opinions of the EU public. This survey was performed under the ethical guidelines defined by the Public Opinion Analysis Sector of the Directorate General Press and Communication of the European Commission. Caveats (Schmitt & Holmberg, 1995; Fuchs et al., 1995; Inglehart, 1990) refer to the sampling procedure and the difficulties associated with measuring income and education among EU member states. Because the survey was conducted on a multistage random sampling basis rather than pure random sampling, it is a “representative sample” of the national population over 15 years of age. The UK population samples were taken on a random basis according to the UK's distribution of metropolitan, urban, and rural resident population densities. In the second stage, a cluster of addresses was selected from each primary sampling unit, which for the United Kingdom was a random selection from electoral registers. Addresses were chosen systematically using standard random route procedures, beginning with an initial address selected at random. The survey was conducted using face-to-face interviewers in respondents' homes.3
3.2. Methods
(4)Therefore, the probability of observing an outcome j=i is the probability of the estimated linear function, plus a normally distributed random error (μj) within the range of the cut points estimated. The model estimates the β1……, βi coefficients, together with cut points k1,……, ki assumed to range from −∞ to +∞. When the underlying distribution is normal, then the resulting estimate follows the lines of an ordered probit model.
Individuals' responses to the attitude question (dependent variable) can also be “don't know” (nonanswers). Since deleting nonanswers could bias results, we added a sample selection correction. This sample selection correction was calculated by running a probit model using the same explanatory variables as the main model with only the “don't knows” as the dependent variable and then including the predicted value of this regression back in the main model. We have also run a binomial probit model (presented in Appendix Table A1) to account for determinants of attitudes for those who responded “don't know” to the attitudes question.
The variables included in our model are the following. We distinguish three key groups of explanatory variables. First, we have variables that measure both revealed (“knowledgewaste”) and perceived (“notinformed”) knowledge (Z1) about radioactive waste. Revealed knowledge (“knowledgewaste”) is derived from how many of the seven questions testing knowledge about radioactive waste and storage each respondent answered correctly. These seven questions capture respondents' understandings of which types of facilities produce radioactive waste and the dangers associated with radioactive waste. The value of this variable can take any integer between 0 and 7. These seven questions appear in Fig. 2.
Level of knowledge about radioactive waste and storage.
Source: Special Eurobarometer 227 (2005).
Furthermore, we include political affiliation (Z2) measuring the extent of political anchoring and beliefs about the consequences of nuclear energy (Z3). These beliefs about nuclear energy are whether nuclear energy use enables European countries to diversify energy sources, if nuclear energy use could reduce oil dependence, and whether nuclear energy use produces less greenhouse gas emissions than alternative energy sources such as oil and coal. We also include a set of control variables distinguished as trust in various information sources (X1) and demographically related information determinants such as age and gender that might both capture experience (X2) and risk attitudes (X3). Finally, another set of controls refers to the sample selection coefficient of inverse mills lambda (λi) and other survey-specific controls that measure the quality of the data.
3.3. Descriptive Statistics
Of the sample of UK citizens surveyed, only 6.9% were “totally in favor” of nuclear power generation whereas 16.0% were “totally opposed.” Those in support of nuclear power generation were less certain of their opinions, with 36.7%“fairly in favor” versus 25.1% being “fairly opposed.” Some respondents (15.4%) did not know if they were in favor of or opposed to energy produced by a nuclear power station, thereby suggesting they have no opinion on the topic and are potentially not informed enough to express a view. When those who expressed being “fairly” or “totally opposed” to energy produced by nuclear power stations were asked how their opinion would change if the issues of radioactive waste were solved, strong opposition remained (19.3% were “totally opposed” and 28.0%“fairly opposed”). These figures demonstrate the static nature of opinions about radioactive waste in the face of new information and greater knowledge of risks and point to factors other than knowledge alone playing key roles in determining attitudes.
The theme of respondents feeling misinformed carries throughout the data as when respondents were asked seven general knowledge questions about radioactive waste, which appear in Fig. 2, only 55% got four or more questions correct and 10% got none correct. Additionally, for many of the belief questions about nuclear energy, a large portion of respondents answered “don't know.”
Respondents appeared particularly ill-informed when questioned about where radioactive waste comes from, how dangerous it is, and how it is stored. Fig. 2 demonstrates the lack of information respondents appear to have regarding radioactive waste production and storage with about a quarter to one-third of respondents for most questions electing to answer “don't know.”
The UK population surveyed overwhelming agreed that making a decision about dangerous waste is a politically unpopular one (80% totally agree or somewhat agree) and that it is important for there to be a near-term solution on the issue of radioactive waste (90% totally agree or somewhat agree). These figures demonstrate that the electorate stands willing to engage on the topic but lacks sufficient information to construct attitudes and risk perceptions about policy. Respondents were most trustful of independent scientists, NGOs concerned about the environment, and international organizations working on peaceful uses of nuclear technology as sources of information about radioactive waste. Only 10% of respondents trusted the UK media as source of information, 13% the EU government, and 16% the UK government.
In terms of political leanings making an impact on attitudes concerning nuclear power generation, Table I exhibits that politics appear to determine support for nuclear power. Individuals with center or right political leanings appear less likely to be fairly opposed or totally opposed to nuclear energy generation than those with left-leaning political stances.
| Self-Reported Political Affiliation | Political Leanings | ||
|---|---|---|---|
| Left | Center | Right | |
| Attitude about energy produced by nuclear power stations | |||
| Totally in favor | 7.4% | 5.5% | 12.7% |
| (0.018) | (0.011) | (0.025) | |
| Fairly in favor | 33.7% | 41.6% | 46.8% |
| (0.033) | (0.023) | (0.038) | |
| Fairly opposed | 30.2% | 25.2% | 20.2% |
| (0.032) | (0.020) | (0.031) | |
| Totally opposed | 19.8% | 16.6% | 8.1% |
| (0.028) | (0.018) | (0.021) | |
| Don't know | 8.9% | 11.1% | 12.1% |
| (0.020) | (0.015) | (0.025) | |
- Note: Standard errors in parentheses. All values are significant at the 5% level.
- Source: Special Eurobarometer 227 (2005).
The extent to which attitudes are politically anchored requires further analysis to control for other information determinants.
4. RESULTS
Before undertaking multivariate analysis, we have investigated the reliability of the variables examined and found an alpha reliability coefficient of 0.6, which suggests that the data are highly reliable. Variance inflation factors (VIF) were always below 10, suggesting no evidence of multicollinearity.
Table II reports the descriptive statistics and data definitions of the variables examined.
| Variable | Definition | Mean | Std. Err. |
|---|---|---|---|
| Dependent variable | |||
| Attitudes | Attitude toward energy produced by nuclear power stations defined in a Likert scale ranging from “totally opposed” to “totally in favor.” | 2.963 | 0.037 |
| Explanatory variables | |||
| Knowledge and subjective amount of information | |||
| Knowledgewaste | General knowledge about radioactive waste | 3.438 | 0.058 |
| Notinformed | Perception of not being informed about radioactive waste | 0.721 | 0.014 |
| Political anchoring | |||
| Leftpolitics | Left of center with political views | 0.200 | 0.013 |
| Centerpolitics | In center with political views | 0.448 | 0.016 |
| Rightpolitics | Right of center with political views | 0.172 | 0.012 |
| Refusedeclarepolaffiliation | Refuse to declare political views | 0.041 | 0.006 |
| Trust in information sources | |||
| Infoagencies | Trust in information from UK agencies in charge of radioactive waste about radioactive waste | 0.191 | 0.012 |
| InfoUKgov | Trust in UK government information about radioactive waste | 0.155 | 0.011 |
| InfoNGO | Trust in information from NGOs about radioactive waste | 0.379 | 0.015 |
| Infoscientist | Trust in information from independent scientists about radioactive waste | 0.409 | 0.015 |
| Infomedia | Trust in information from media about radioactive waste | 0.102 | 0.010 |
| InfoEU | Trust in information from EU about radioactive waste | 0.118 | 0.010 |
| Infoindustry | Trust in information from nuclear industry about radioactive waste | 0.155 | 0.011 |
| Infointlorg | Trust in information from international organizations about radioactive waste | 0.286 | 0.014 |
| Infonone | Trust in information from no sources about radioactive waste | 0.112 | 0.010 |
| Beliefs in consequences of nuclear energy use | |||
| Believediversify | Believe that use of nuclear energy enables European countries to diversify their energy sources | 2.900 | 0.045 |
| Believedependence | Believe use of nuclear energy could reduce dependence on oil | 2.497 | 0.042 |
| Believegreenhouse | Believe use of nuclear energy produces less greenhouse gas emissions than other energy sources (oil or coal) | 2.713 | 0.049 |
| Sociodemographics | |||
| Gender | Male = 1 | 0.456 | 0.016 |
| Age 15–24 | Age ranges from 15–24 = 1 | 0.104 | 0.010 |
| Age 25–34 | Age ranges from 25–34 = 1 | 0.158 | 0.011 |
| Age 35–44 | Age ranges from 35–44 = 1 | 0.196 | 0.013 |
| Age 45–54 | Age rages from 45–54 = 1 | 0.156 | 0.011 |
| Age 55–64 | Age ranges from 55–64 = 1 | 0.145 | 0.011 |
| Age 65+ | Age ranges from 65+= 1 | 0.241 | 0.013 |
| Educ<15 | Finished education at 15 or below = 1 | 0.292 | 0.014 |
| Educ 16–19 | Finished education at 16–19 = 1 | 0.482 | 0.016 |
| Educ 20+ | Finished education at 20+= 1 | 0.180 | 0.012 |
| Educstudying | Still studying = 1 | 0.038 | 0.006 |
| Educnone | No full-time education = 1 | 0.009 | 0.003 |
| Liverural | Live in a rural area or village = 1 | 0.321 | 0.015 |
| Livesmalltown | Live in a small- or middle-sized town = 1 | 0.320 | 0.015 |
| Liveurban | Live in a large town = 1 | 0.358 | 0.015 |
| Other controls | |||
| CoopExcel | Excellent respondent cooperation = 1 | 0.792 | 0.013 |
| CoopFair | Fair respondent cooperation = 1 | 0.145 | 0.011 |
| CoopAvg | Average respondent cooperation = 1 | 0.051 | 0.007 |
| CoopBad | Bad respondent cooperation = 1 | 0.013 | 0.004 |
- Note: Cronbach's alpha (reliability coefficient) for the explanatory variable = 0.702.
Our primary interest lies in the determinants of attitudes about (support for) nuclear power generation. Therefore, we examine the determinative power of actual and self-perceived levels of knowledge, political affiliation, trust in information sources about radioactive waste, and beliefs about the consequences of nuclear energy use, along with other control variables for demographic characteristics. Correlation analysis reveals that no-answers were highly correlated with knowledge (ρ= 0.24 (p < 0.001)) while not being correlated with political affiliation, thus highlighting the importance of examining the knowledge and political affiliation variables.
4.1. Effects on Support for Nuclear Energy Production
Table III provides the results for models examining attitudes about energy produced by nuclear power stations. We run three regressions: (1) all control and explanatory variables, including political affiliation and beliefs about the consequences of nuclear energy use (“total sample”), (2) all control and explanatory variables except for the political extremes (“restricted sample A”), and (3) all control and explanatory variables except for beliefs about the consequences of nuclear energy use (“restricted sample B”).
| (1) Total Sample | (2) Restricted Sample A(*) | (3) Restricted Sample B(**) | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Coef. | Std. Err. | z | Coef. | Std. Err. | z | Coef. | Std. Err. | z | |
| Knowledge and subjective amount of information (Z1) | |||||||||
| Knowledgewaste | −0.27a | 0.06 | −4.22 | −0.22a | 0.07 | −3.09 | −0.05 | 0.03 | −1.83 |
| Notinformed | 1.32a | 0.30 | 4.45 | 1.21a | 0.35 | 3.48 | 0.25b | 0.10 | 2.54 |
| Political anchoring (Z2) | |||||||||
| Leftpolitics | −3.80a | 0.73 | −5.17 | − | − | − | −1.02a | 0.17 | −6.11 |
| Centerpolitics | −3.26a | 0.67 | −4.86 | −2.92a | 0.77 | −3.81 | −0.72a | 0.15 | −4.88 |
| Rightpolitics | −1.73a | 0.44 | −3.94 | − | − | − | −0.12 | 0.16 | −0.78 |
| Trust in information sources | |||||||||
| Infoagencies | −0.74a | 0.19 | −3.82 | −0.58b | 0.23 | −2.51 | −0.12 | 0.11 | −1.09 |
| InfoUKgov | 0.20 | 0.12 | 1.61 | 0.23 | 0.16 | 1.46 | 0.07 | 0.12 | 0.62 |
| InfoNGO | −1.48a | 0.30 | −4.86 | −1.27a | 0.35 | −3.63 | −0.36 | 0.09 | −3.98 |
| Infoscientist | −0.03 | 0.09 | −0.39 | −0.12 | 0.12 | −1.06 | 0.07 | 0.08 | 0.87 |
| Infomedia | 0.83a | 0.19 | 4.25 | 1.04a | 0.24 | 4.28 | 0.28b | 0.13 | 2.13 |
| InfoEU | −1.55a | 0.34 | −4.51 | −1.19a | 0.39 | −3.02 | −0.32b | 0.13 | −2.49 |
| Infoindustry | 0.86a | 0.16 | 5.36 | 0.84a | 0.20 | 4.16 | 0.43a | 0.12 | 3.68 |
| Infointlorg | 0.56a | 0.11 | 4.84 | 0.47a | 0.14 | 3.25 | 0.29a | 0.09 | 3.18 |
| Infonone | −0.22 | 0.17 | −1.28 | −0.12 | 0.20 | −0.61 | 0.13 | 0.14 | 0.94 |
| Beliefs in consequences of nuclear energy use | |||||||||
| Believediversify | 0.55a | 0.15 | 3.57 | 0.45b | 0.17 | 2.57 | − | − | − |
| Believedependence | 0.54a | 0.15 | 3.60 | 0.44b | 0.17 | 2.54 | − | − | − |
| Believegreenhouse | 0.44a | 0.10 | 4.28 | 0.42a | 0.12 | 3.51 | − | − | − |
| Sociodemographics | |||||||||
| Gender | −1.42a | 0.34 | −4.15 | −1.30a | 0.39 | −3.35 | −0.14 | 0.09 | −1.54 |
| Age 15–24 | 0.19 | 0.18 | 1.08 | −0.05 | 0.23 | −0.23 | 0.22 | 0.18 | 1.20 |
| Age 25–34 | −0.18 | 0.14 | −1.26 | −0.25 | 0.19 | −1.33 | −0.13 | 0.14 | −0.90 |
| Age 35–44 | −0.04 | 0.13 | −0.34 | −0.21 | 0.18 | −1.21 | −0.02 | 0.13 | −0.14 |
| Age 45–54 | −0.05 | 0.13 | −0.38 | 0.00 | 0.18 | 0.02 | −0.06 | 0.13 | −0.42 |
| Age 55–64 | −0.03 | 0.13 | −0.27 | −0.03 | 0.17 | −0.16 | −0.04 | 0.13 | −0.31 |
| Educ<15 | 0.43 | 0.42 | 1.02 | 0.46 | 0.49 | 0.93 | 0.51 | 0.42 | 1.20 |
| Educ 16–19 | 0.41 | 0.42 | 0.99 | 0.47 | 0.48 | 0.97 | 0.47 | 0.41 | 1.13 |
| Educ 20+ | 0.48 | 0.42 | 1.12 | 0.40 | 0.50 | 0.80 | 0.53 | 0.42 | 1.26 |
| Educstudying | −0.08 | 0.45 | −0.18 | 0.14 | 0.51 | 0.28 | −0.04 | 0.44 | −0.09 |
| Liverural | 0.08 | 0.10 | 0.80 | 0.08 | 0.13 | 0.66 | 0.07 | 0.10 | 0.78 |
| Livesmalltown | −0.03 | 0.10 | −0.30 | −0.03 | 0.12 | −0.23 | −0.02 | 0.09 | −0.23 |
| Other controls | |||||||||
| CoopExcel | 0.30 | 0.49 | 0.62 | 0.41 | 0.62 | 0.66 | 0.09 | 0.48 | 0.19 |
| CoopFair | 0.32 | 0.50 | 0.64 | 0.38 | 0.63 | 0.60 | 0.10 | 0.48 | 0.21 |
| CoopAvg | 0.29 | 0.52 | 0.56 | 0.31 | 0.67 | 0.46 | 0.03 | 0.51 | 0.06 |
 |
6.37a | 1.26 | 5.06 | 5.72a | 1.44 | 3.98 | 1.53a | 0.14 | 11.20 |
| Pseudo R2 | 0.132 | 0.143 | 0.128 | ||||||
| LR χ222 | 290.11 | 180.38 | 270.77 | ||||||
| Log likelihood | −913.4 | −542.3 | −923.1 | ||||||
- (*) Excluding political extremes.
- (**) Excluding beliefs about the consequences of nuclear energy use.
- aSignificant at 1%.
- bSignificant at 5%.
In the first model, we find that nuclear power attitudes are driven by knowledge (RQ2). Individuals who have less knowledge about radioactive waste appear more likely to demonstrate support for nuclear power production. However, the extent of the influence of knowledge on attitudes about nuclear power depends upon beliefs about the consequences of nuclear energy use as when beliefs are removed from the restricted sample B in Column 3, knowledge no longer has a determinative influence on attitudes.
Individuals who do not perceive themselves as being informed about radioactive waste are expected to be more likely to support nuclear power policy. Therefore, the rejection of nuclear power could be in part a reflection of an underlying demand for information or, alternatively, they could be satisfied with the level of information they have to the point that they have enough information to reject nuclear power (RQ1).
In the total sample model including all explanatory variables, we find significant political effects indicating that support for nuclear power is more likely to be lower among those with any self-proclaimed political affiliation (RQ3). Beliefs about nuclear energy having a positive role in reducing the UK's dependence on oil, allowing for diversification of energy sources, and producing less greenhouse gas emissions proved significant in leading to positive attitudes toward nuclear energy policy.
Regarding information sources, support for nuclear power depends positively on trust in the media, the nuclear power industry, and international organizations, while trust in UK informational agencies, NGOs, and the EU lead individuals to being less likely to support nuclear power. Age, living in an urban versus rural setting, and respondent cooperation proved insignificant, while gender is significant. Education is not significant, meaning any information effects that education could convey are already captured by the knowledge effect, while the remaining socioeconomic effects related to education are insignificant.
Restricted sample A (Column 2) removes political extremes but knowledge and beliefs remain with the same relationship in determining support for nuclear power. Restricted sample B (Column 3) leaving out beliefs about the consequences of nuclear energy use shows that influence of knowledge is highly dependent on beliefs because knowledge is no longer a significant determinant of attitudes once beliefs are removed from the model. Therefore, the influence of knowledge is dependent on whether beliefs are taken into account, though individuals who see themselves as not informed are more likely to support nuclear power regardless of whether beliefs are included. We also see a change in the impact of political anchoring on support for nuclear power as right-leaning political views are no longer significant when beliefs about the consequences of nuclear energy use are removed from the model. Right-wing political affiliation depends on beliefs to be a meaningful determinant of attitudes about nuclear energy generation whereas left and center political affiliations do not. The fact that left-leaning and centrist political views remain significant even when beliefs are excluded provides additional evidence of political anchoring in driving attitudes about nuclear power.
Table A1 in the Appendix provides the determinants of individuals professing to have no attitude about nuclear energy. We find that “nonattitudes” were mainly explained by beliefs about the consequences of nuclear energy use and political affiliations, suggesting that not only are attitudes determined by anchoring, but also nonattitudes. Only when beliefs are excluded do we find that knowledge explains the probability of providing a response, suggesting evidence of some “rational ignorance” in the formation of attitudes toward nuclear power. Women were more likely than men to fail to express an attitude about the favorability of nuclear energy while those with a self-expressed political stance were less likely to fail to express some kind of attitude about nuclear energy generation than those with no political stance. Those who held beliefs about how using nuclear energy enables energy diversification and how nuclear energy differs in greenhouse gas emission levels than other energy sources were more likely to not express an attitude about nuclear energy. In our descriptive findings we highlighted that individuals largely lack information about nuclear energy and radioactive waste. This finding could point to individuals perhaps having beliefs about nuclear energy but not enough confidence in their level of being informed to actually make a decision about whether nuclear power generation presents a good energy solution. Therefore, beliefs appear to not be enough to lead someone to form an attitude about supporting nuclear power generation. Instead, perceptions of being informed and political affiliation play more significant roles in leading individuals to develop attitudes. All other variables failed to have a significant impact on determining if individuals respond “don't know” to a question about their support for nuclear power generation.
4.2. Marginal Probability Effects
Table IV reports the marginal effects of political affiliation and knowledge about radioactive waste on all possible attitudes about nuclear power generation after controlling for the same set of relevant covariates included in the models presented in Table III. Marginal effects denote the effect of a 1% change in one explanatory variable on the probability that an individual changes from being strongly against to fairly against (or fairly against to fairly in favor or fairly in favor to strongly in favor of) energy produced by nuclear power. Coefficients suggest that higher knowledge (or not being informed) increases (reduces) the probability of one changing his or her attitude from being strongly against to fairly against nuclear power generation by 6% (37%). Greater knowledge (or not being informed) reduces (increases) the probability of an individual moving from being fairly against to fairly in favor of energy produced from nuclear power by 8% (40%). We find that knowledge tends to drive people to move from a favorable to a skeptical position about supporting nuclear power generation. Compared to not revealing a political ideology, if an individual associates himself/herself with a particular political viewpoint, then he/she is more likely to move from being fairly against to strongly against nuclear power generation. The more to the left an individual stands, the less likely he/she is to be fairly or moderately in favor of nuclear power as a source of energy. However, the coefficient is not statistically different between those individuals who identify themselves as right and center (48%), which suggests a less extreme marginal effect toward being in favor of nuclear energy generation as opposed to being opposed to it.
| Attitudes* | Strongly Against (Fairly Against) | Fairly Against (Fairly in Favor) | Fairly in Favor (Strongly in Favor) | |||
|---|---|---|---|---|---|---|
| Coeff | t-value | Coeff | t-value | Coeff | t-value | |
| Knowledge and subjective amount of information (Z1) | ||||||
| Knowledgewaste | 0.06a | 4.16 | −0.08a | −4.11 | −0.08a | −4.11 |
| Notinformed | −0.37a | −3.91 | 0.40a | 5.33 | 0.34a | 5.33 |
| Political anchoring (Z2) | ||||||
| Leftpolitics | 0.94a | 19.31 | −0.63a | −27.62 | −0.63a | −27.62 |
| Centerpolitics | 0.73a | 6.12 | −0.47a | −8.24 | −0.48a | −8.24 |
| Rightpolitics | 0.55a | 3.72 | −0.48as | −6.30 | −0.48a | −6.30 |
- aSignificant at 1%.
- *“Don't know” responses excluded because sample selection correction was always significant.
- Note: All outcomes are for a model run with the total sample.
5. DISCUSSION
This study has examined the influence of knowledge, self-perceived levels of knowledge (“certainty effect”), and political anchoring on determining attitudes toward nuclear power generation. We have drawn from a representative sample of the United Kingdom containing variables to act as controls along with both self-perceived and revealed knowledge and political attitudes as explanatory variables. Examination of these survey data allow us to test whether there is evidence of knowledge, certainty effects, and political anchoring playing a determinative role in attitudes about nuclear power generation.
We have found that knowledge decreases support for nuclear power generation. This finding negates previous literature promoting the importance of information uptake to reduce perceptions of risks and aid individuals in overcoming fears of risks about which they feel uncertain (Fischhoff et al., 1978; Sjoberg & Drottz-Sjoberg, 1991; Roberts, 1990). We posit that the overreliance of political beliefs to form attitudes about nuclear power supersedes the traditional role of knowledge here in adjusting attitudes.
Perceived knowledge also plays a significant role in nuclear power attitudes as those viewing themselves as lacking sufficient knowledge seem to be more likely to support nuclear power. If those individuals who express a lack of support for nuclear power generation also reveal a lack of information then those individuals could be expressing a demand for information (Costa-Font & Mossialos, 2005).
Our findings support our hypothesis that attitudes are politically anchored and thus exhibit single-peak preferences or support the theory of “rational irrationality” (Caplan, 2001). Except for those with right-leaning points of view in the restricted model removing beliefs, politics was significant in all models regardless of whether beliefs was included or not, meaning that individuals rely heavily on their political affiliation.
Similarly to other studies such as Sjoberg (2003a), we found that while individuals may have emotional reactions to technologies, attitudes about nuclear power are largely found to be “ideologically” grounded or “value-laden.” Finding evidence of political anchoring explains political opposition to nuclear power as influencing an amplification of public risk perceptions about the subject. Coupling this finding with prospect reference theory (Kahneman & Tversky, 1979), one might argue that political figures act to overweight the risks of new technologies, thus explaining the distance between opinions of nuclear energy experts and lay individuals (Slovic et al., 1991).
We also find that trust in information sources is a significant determinant of attitude expression, in line with previous literature that emphasizes trust as the main attitudinal determinant (Slovic, 1993; Horning Priest et al., 2003; James, 2001). Trust in the nuclear power industry, international organizations, and the media appears important for shaping support of nuclear power generation. This finding signals the need of strengthening the communication from these sources as individuals appear to attach some credibility to them by allowing them to play a determinative role in attitude formation.
As for control variables, we have found evidence of the influence of gender on affecting attitudes but not in the direction expected. However, the influence of gender was mediated by beliefs, thus providing further evidence of the importance of political views determining attitudes about nuclear energy generation.
The existence of political anchoring is consistent with previous survey evidence and findings that individuals' political stances and worldviews matter in attitudes regarding nuclear power (Bouyer et al., 2001; McGowan & Sauter, 2005; Slovic & Peters, 1998; ICM/Sunday Telegraph, 2006; You Gov/Economist, 2006). Although previous work has analyzed respondent attitudes and risk perceptions about nuclear energy generation, it differs from this work in that it is solely presenting survey results (ICM/Sunday Telegraph, 2006; You Gov/Economist, 2006), involves descriptive work regarding survey results (Bouyer et al., 2001), or takes a different focus by examining how worldviews impact risk perceptions of nuclear energy as well as many other hazards (McGowan & Sauter, 2005). This work takes previous research on the topic of attitudes and risk perceptions about nuclear energy a step further by testing the importance of political views both by removing extreme political views from the analysis and testing the effect of removing beliefs on the presence of political anchoring. We also control for “don't know” responses, which to the authors' knowledge, makes this the first study on the topic to do so. This is a meaningful point of progress since so many respondents select “don't know” when answering questions about nuclear energy and waste. After controlling for potentially having a biased sample of respondents as a result of individuals deciding not to answer about their support of nuclear energy, we find that individuals who reveal a political stance of any kind (left, center, or right) are less likely to have “nonattitudes” about nuclear power generation. This finding provides further evidence of the importance of political views in attitude expression, and indicates that attitudes about nuclear power rely heavily on an interest in social and political issues.
6. CONCLUSION
This study examines the determinative influence of political views (political anchoring), knowledge, and perceived knowledge in the formation of attitudes about nuclear power. The United Kingdom is a particularly interesting country to examine regarding these issues at the time of the study because of significant media attention currently being granted to the policy-making process surrounding nuclear energy generation. Even with the topic often being covered by media outlets and the government focused on coming up with a policy, respondents seem to have limited knowledge about the topic.
Where knowledge appears to play a role in shaping attitudes about nuclear power, its importance is eclipsed by the role of political beliefs as a determinant. Therefore, despite opportunities for individuals to gain knowledge about the risks of nuclear power generation, attitudes appear driven much more by an instinctive political decision rather than digestion of information. Some respondents are found to form attitudes about nuclear power generation based squarely on their political affiliation and beliefs, which implies that they might be immune to communication policies covering the topic. Findings show that political affiliation determines policy support and associated risk perceptions to a greater extent than knowledge or perceived knowledge, suggesting that individuals formulate their opinions more on instinctive reaction than available facts. In fact, when we exclude beliefs about the consequences of nuclear energy use from our analysis, knowledge loses significant determinative power in predicting support for nuclear power. These findings regarding the importance of political views regarding attitudes especially suggest that individuals in the United Kingdom respond to proposals for greater nuclear power generation based chiefly on ideological stance. As a result of this study, we can see that past information campaigns aimed at increasing individuals' knowledge about nuclear power generation and resulting effects appear to have not adequately shaped attitudes and perhaps in the future should be more sensitive to the instinctive-driven ways in which individuals seem to react to unfamiliar technology like nuclear power generation. If opinions about nuclear energy are as driven by political views and beliefs as this research finds, then meaningful debate involving the public could aid in reducing this polarization of beliefs such that views become less reliant on political stance.
Footnotes
Appendix
| (10) Total Sample | (11) Restricted Sample A(*) | (12) Restricted Sample B(**) | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Coef. | Std. Err. | z | Coef. | Std. Err. | z | Coef. | Std. Err. | z | |
| Knowledge and subjective amount of information (Z1) | |||||||||
| Knowledgewaste | −0.03 | 0.03 | −1.01 | −0.01 | 0.04 | −0.35 | −0.11a | 0.03 | −3.58 |
| Notinformed | 0.27 | 0.15 | 1.80 | 0.10 | 0.18 | 0.54 | 0.30b | 0.14 | 2.19 |
| Political anchoring (Z2) | |||||||||
| Leftpolitics | −0.70a | 0.18 | −3.87 | − | − | − | −0.80a | 0.17 | −4.62 |
| Centerpolitics | −0.66a | 0.14 | −4.65 | −0.71a | 0.15 | −4.83 | −0.76a | 0.14 | −5.58 |
| Rightpolitics | −0.47b | 0.19 | −2.53 | − | − | − | −0.60a | 0.18 | −3.40 |
| Trust in information sources | |||||||||
| Infoagencies | −0.12 | 0.17 | −0.70 | −0.09 | 0.21 | −0.45 | −0.22 | 0.16 | −1.34 |
| InfoUKgov | 0.04 | 0.17 | 0.26 | −0.08 | 0.20 | −0.38 | 0.10 | 0.16 | 0.66 |
| InfoNGO | −0.24 | 0.13 | −1.79 | −0.31 | 0.17 | −1.88 | −0.20 | 0.13 | −1.59 |
| Infoscientist | 0.01 | 0.13 | 0.06 | 0.03 | 0.16 | 0.22 | −0.01 | 0.12 | −0.08 |
| Infomedia | 0.15 | 0.18 | 0.79 | 0.20 | 0.23 | 0.85 | 0.12 | 0.18 | 0.67 |
| InfoEU | −0.28 | 0.23 | −1.24 | −0.39 | 0.28 | −1.39 | −0.21 | 0.21 | −1.00 |
| Infoindustry | 0.14 | 0.18 | 0.79 | 0.12 | 0.22 | 0.57 | 0.02 | 0.17 | 0.10 |
| Infointlorg | 0.10 | 0.14 | 0.70 | 0.24 | 0.17 | 1.44 | 0.07 | 0.14 | 0.50 |
| Infonone | −0.14 | 0.19 | −0.73 | −0.23 | 0.22 | −1.05 | −0.05 | 0.18 | −0.27 |
| Beliefs in consequences of nuclear energy use | |||||||||
| Believediversify | 0.15a | 0.04 | 3.26 | 0.13b | 0.05 | 2.48 | − | − | − |
| Believedependence | 0.15a | 0.05 | 3.10 | 0.12b | 0.06 | 2.14 | − | − | − |
| Believegreenhouse | 0.09b | 0.04 | 2.23 | 0.12b | 0.05 | 2.40 | − | − | − |
| Sociodemographic-dependent sources | |||||||||
| Gender | −0.36a | 0.12 | −2.89 | −0.29b | 0.15 | −1.97 | −0.55a | 0.12 | −4.75 |
| Age 15–24 | 0.13 | 0.24 | 0.54 | 0.26 | 0.29 | 0.89 | 0.27 | 0.23 | 1.17 |
| Age 25–34 | −0.04 | 0.20 | −0.20 | 0.04 | 0.25 | 0.16 | 0.17 | 0.19 | 0.86 |
| Age 35–44 | −0.13 | 0.20 | −0.66 | −0.14 | 0.25 | −0.59 | 0.03 | 0.19 | 0.16 |
| Age 45–54 | 0.12 | 0.19 | 0.61 | 0.18 | 0.24 | 0.77 | 0.14 | 0.18 | 0.77 |
| Age 55–64 | −0.02 | 0.19 | −0.11 | 0.00 | 0.23 | −0.02 | −0.01 | 0.18 | −0.08 |
| Educ <15 | 0.56 | 0.80 | 0.70 | 0.69 | 0.83 | 0.83 | 0.71 | 0.74 | 0.96 |
| Educ 16–19 | 0.55 | 0.79 | 0.69 | 0.60 | 0.82 | 0.73 | 0.57 | 0.74 | 0.77 |
| Educ 20+ | 0.52 | 0.81 | 0.64 | 0.57 | 0.85 | 0.67 | 0.43 | 0.75 | 0.57 |
| Educstudying | 0.37 | 0.84 | 0.43 | 0.24 | 0.88 | 0.27 | 0.38 | 0.78 | 0.48 |
| Liverural | 0.03 | 0.14 | 0.25 | −0.04 | 0.17 | −0.21 | 0.09 | 0.13 | 0.66 |
| Livesmalltown | −0.07 | 0.14 | −0.48 | −0.14 | 0.17 | −0.84 | −0.09 | 0.13 | −0.65 |
| Other controls | |||||||||
| CoopExcel | −0.59 | 0.40 | −1.46 | −0.60 | 0.47 | −1.28 | −0.79b | 0.39 | −1.99 |
| CoopFair | −0.44 | 0.41 | −1.07 | −0.37 | 0.48 | −0.77 | −0.63 | 0.40 | −1.55 |
| CoopAvg | 0.08 | 0.44 | 0.19 | 0.12 | 0.51 | 0.24 | −0.21 | 0.43 | −0.49 |
| Constant | −1.66b | 0.86 | −1.93 | −1.60 | 0.90 | −1.76 | −0.03 | 0.78 | −0.04 |
| Pseudo R2 | 0.248 | 0.236 | 0.182 | ||||||
| LR χ222 | 214.24 | 142.23 | 157.28 | ||||||
| Log likelihood | −325.5 | −230.4 | −354.0 | ||||||
- (*) Excluding political extremes.
- (**) Excluding beliefs about the consequences of nuclear energy use.
- aSignificant at 1%.
- bSignificant at 5%.
