Are Implicit Associations With Nuclear Energy Related to Policy Support? Evidence From the Brief Implicit Association Test

Abstract

Nuclear energy has long been assumed to elicit automatic, negative reactions. However, little research has investigated implicit associations with nuclear energy. To assess implicit and explicit attitudes toward nuclear energy, 704 U.S. consumer panelists completed a multicategory Implicit Association Test (IAT) and an Internet survey. Results showed that participants held negative implicit attitudes toward nuclear energy (vs. wind and natural gas) and positive implicit attitudes toward nuclear energy (vs. coal). Strong opponents of nuclear policy implicitly preferred natural gas over nuclear and implicitly disliked nuclear as much as coal. Strong supporters of nuclear policy implicitly preferred nuclear over coal, and showed no implicit preference for gas over nuclear. Implicit attitudes toward nuclear energy (vs. gas and wind) were related to policy support when controlling for explicit attitudes and demographics. Understanding both implicit and explicit nuclear attitudes is important for decision makers as the United States charts its energy future.

Keywords

nuclear energy implicit association test risk perceptions

Background Energy Issues

Although the rate of growth has slowed since 1950, electricity demand in the United States is expected to grow by 0.8% per year through 2035 (U.S. Energy Information Administration [EIA], 2012a). An all-of-the-above approach to energy production using a mix of energy sources to meet increasing demand has been proposed by the current administration, although public concerns about each energy source have been raised in recent years. Coal provides the largest amount (42%) of the energy currently consumed in the United States (EIA, 2012b), though coal is viewed quite negatively by the public (Ansolabehere, 2007; Ansolabehere & Konisky, 2009; Greenberg, 2009) and has been criticized for its greenhouse gas emissions and the potential environmental and health effects related to coal mining and waste storage. Natural gas, which currently accounts for 25% of electricity generation (EIA, 2012b) has recently seen a major boom from new extraction technologies such as hydraulic fracturing. Yet, fracking has raised concerns about potential water contamination, earthquakes, and adverse health impacts of those living near wells (Finkel & Law, 2011; Fountain, 2011; Klemow, 2012). Wind, accounting for 3% of the nation’s electricity in 2011 (EIA, 2012b), is generally favorably regarded by the public (Ansolabehere, 2007; Ansolabehere & Konisky, 2009; Greenberg, 2009). However, concerns have been raised about the potential adverse effects of wind farms on birds as well as complaints about noise pollution and negative visual impacts (Phadke, 2010).

Though the concerns mentioned above related to coal, natural gas, and wind power have framed much of the energy debate in the United States, perhaps no single event in recent energy history has received as much attention as the nuclear disaster at the Fukushima Diiachi Nuclear Power Plant in Japan following the March 2011 earthquake and tsunami. Nuclear power, which accounted for 19% of electricity generation in the United States in 2011 (EIA, 2012b) was in the midst of a rebirth in the United States with licenses being extended for more than half of America’s nuclear fleet and proposals to construct and operate new nuclear plants under review at the Nuclear Regulatory Commission (NRC). The Fukushima disaster (at least temporarily) led to reduced public support for nuclear power (CNN Opinion Research, 2011), fear of radiation contamination, and plans to phase out nuclear power in Germany, Switzerland, and Japan (though Japan’s new administration has since reversed this course). However, the United States largely continued its pre-Fukushima nuclear plans (albeit with safety inspections at U.S. nuclear plants) with the NRC approving licenses for the first new nuclear reactors in 30 years to be built in the United States (Tracy, 2012; Wald, 2012).

The long-term effects of the Fukushima accident on public support for nuclear policy have not yet been established and it will take several more years to fully appreciate them. Considering recent research has shown that one of the most commonly associated images with nuclear energy in the United States is Chernobyl even 25 years after that disaster (Truelove, 2012), it is clear that the effect of the Fukushima event will linger in the public’s mind for decades to come. As such, the purpose of this article is not to attempt to trace the effect of the Fukushima accident on public opinion related to energy sources. Instead this study aims to uncover people’s implicit perceptions of nuclear energy compared with natural gas, wind, and coal against the backdrop of the issues raised above.

Several previous studies have assessed the public’s beliefs about nuclear energy. Those with more positive beliefs about nuclear power, such as beliefs related to the environmental impacts, cost, and safety of nuclear activities, are more favorable toward policies to expand nuclear reliance and build new nuclear plants (Ansolabehere & Konisky, 2009; Greenberg, 2009; Greenberg & Truelove, 2011). However, “for the majority of citizens, nuclear power is typically out of sight and submerged below conscious awareness” (Rosa & Clark, 1999, p. 54), so knowing people’s beliefs about nuclear energy may not fully reveal their attitudes or support for nuclear policy. New methods are now available that allow for the simultaneous assessment of implicit (automatic) attitudes toward multiple categories (Sriram & Greenwald, 2009). Using these methods along with traditional survey techniques, the present study sought to further examine nuclear energy attitudes (compared with natural gas, coal, and wind) at both an explicit and implicit level. Three research questions guided our efforts:

How do attitudes toward nuclear energy compare with those of coal, natural gas, and wind?

Do people view nuclear energy negatively at an implicit level?

How do both implicit and explicit attitudes toward nuclear energy relate to policy support?

Theoretical Considerations and Previous Research

Dual-process models of information processing outline two parallel methods by which incoming information is processed: a cognitive-centered, deliberate evaluation and an emotion-centered, automatic evaluation (Chaiken, 1980; Epstein, 1994; Petty & Cacioppo, 1986; Zajonc, 1980). Using Epstein’s (1994) terminology, the rational system is slower, analytic, beliefs-centered, and conscious, while the experiential system is faster, holistic, emotion-centered, and preconscious. Researchers studying risk perception have found support for the dual-process framework as applied to risk evaluations (Loewenstein, Weber, Hsee, & Welch, 2001; Meijnders, Midden, & Wilke, 2001; Rundmo, 2002; Slovic, Finucane, Peters, & MacGregor, 2004; Truelove, 2012). When people rely on the experiential system when evaluating risks, they have been said to use the “affect heuristic” (Finucane, Alhakami, Slovic, & Johnson, 2000). Specifically, the affect heuristic has been described as a mental shortcut used to make judgments that relies on the positive and negative affect connected to images people associate with the object to be evaluated (Finucane et al., 2000). The more positive the affect associated with the images, the less risky the object is perceived.

Research exploring the affect heuristic has found that nuclear power is one of the most dreaded risks the public faces (Slovic, 1987) and often elicits negative images (Keller, Visschers, & Siegrist, 2012; Truelove, 2012) and feelings (Sjöberg, 2007) from laypeople (though people who live near nuclear waste facilities show less negative associations; Greenberg, 2012). Nuclear-related negative feelings and images have been found to relate to support for decreasing nuclear energy reliance and siting of local nuclear facilities (Keller et al., 2012), even after accounting for more cognitive-focused beliefs (Truelove, 2012), providing support for dual-process models of risk perception as applied to nuclear energy perceptions.

Although it has been argued that the affect heuristic may be equivalent to the concept of implicit attitudes (Dohle, Keller, & Siegrist, 2010; Siegrist, Keller, & Cousin, 2006; Spence & Townsend, 2008), much less research has investigated the relationship between implicit attitudes toward risks and risk policy support (cf. Siegrist et al., 2006). Implicit attitudes are automatic associations between mental categories and can be thought of as part of the experiential system, as they are immediate and are often not consciously accessible.

The dominant technique to measure implicit attitudes over the past decade has been the Implicit Association Test (IAT; Greenwald, Mcghee, & Schwartz, 1998). The IAT is a reaction time task that consists of a number of trial blocks in which participants must rapidly classify objects (usually words or images) as belonging to a target category or an attribute (usually good or bad). The relative speed at which objects belonging to the target category can be classified as “good” versus classified as “bad” is a measure of the extent to which the target category is connected to “good” in semantic networks that include affective nodes. Participants who classify the target objects faster when the target category is paired with “good” are said to have a positive implicit association with the object, or a positive implicit attitude toward the object. The traditional IAT, by design, compares attitudes toward two objects and is thus a measure of relative preference or relative attitudes. (For more detailed descriptions of IAT block designs, see Nosek, Greenwald, & Banaji; 2007.)

Implicit attitude assessments are especially useful when the attitude object under investigation is expected to elicit automatic preferences that may differ from self-reported conscious evaluations. Along these lines, implicit attitude assessments have frequently been conducted to measure implicit attitudes toward race and ethnicity, drugs, and gender and sexual preference (Greenwald, Poehlman, Uhlmann, & Banaji, 2009). However, more recent research has recognized the potential for implicit attitudes to inform the study of risk perceptions (Visschers & Siegrist, 2008). Support for energy policy, and particularly nuclear energy policy, represents an area where implicit attitudes may be very revealing as people who oppose nuclear energy are often assumed to have immediate, negative reactions to nuclear power.

Implicit associations with energy sources have been relatively unexamined, with only one research group in Switzerland assessing implicit associations with nuclear and hydroelectric energy. In an initial study, Siegrist et al. (2006) measured Swiss students’ implicit attitudes toward nuclear power relative to hydroelectric power using the traditional IAT. Results showed that participants had an implicit and explicit preference for hydroelectric power over nuclear power and the implicit preference for hydroelectric power was negatively correlated with an explicit attitude toward nuclear. In addition to attitudes, Siegrist et al. assessed opposition to new nuclear power plants. They found that participants who opposed new nuclear power plants had significantly negative implicit attitudes toward nuclear versus hydroelectric power, but participants who explicitly favored more nuclear power plants did not have significantly negative attitudes toward nuclear and instead have similar implicit attitudes toward nuclear and hydroelectric power.

Siegrist et al. (2006) noted that due to the relative nature of the IAT, it was impossible to discern from their results whether people have a negative implicit association with nuclear power or a positive association with hydroelectric power (or both). To address this concern, Dohle et al. (2010) conducted a follow-up study using a single category–IAT (SC-IAT) to assess automatic associations with nuclear energy and hydroelectric energy (and home appliances) among Swiss college students. SC-IATs compare response latencies when the target object is paired with positive and negative words, but no alternative object is compared so the attitudes revealed are not relative (Bluemke & Friese, 2008; Karpinski & Steinman, 2006). Dohle and colleagues’ results showed that nuclear energy was viewed negatively, with SC-IAT scores less than zero, while hydroelectric energy was viewed positively. Results also showed that nuclear SC-IAT scores were related to explicit attitudes toward nuclear energy and judgments of the riskiness of nuclear energy, though policy support was not assessed.

Dohle et al.’s (2010) and Siegrist et al.’s (2006) studies provide much of what we know about implicit attitudes toward energy sources. Implicit nuclear attitudes are negative, they are correlated with explicit attitudes and risk perceptions, and they are related to policy support. However, several additional questions remain. First, how generalizable are their findings beyond Swiss college students—would we get the same results in the United States? Second, how do implicit attitudes about nuclear energy relate to those of other energy sources that are also often viewed negatively, such as coal and natural gas? Third, do implicit attitudes about nuclear energy relate to policy support above and beyond demographic variables and explicit attitudes? The present study was designed with these open questions in mind.

Present Study

The present study extends our current knowledge of nuclear energy risk perceptions in several important ways. First, this study used a new variant of the IAT called the Brief IAT (BIAT; Sriram & Greenwald, 2009), which is a shortened form of the IAT that allows for the comparison of multiple categories, to assess implicit attitudes toward nuclear, coal, natural gas, and wind energy. In addition, this study used a sample of American residents drawn from an Internet consumer panel, which provides a different sample type than previous research. Drawing on the work by Siegrist and colleagues (Dohle et al., 2010; Siegrist et al., 2006) showing that Swiss college students hold negative implicit attitudes toward nuclear energy (compared with hydroelectric and using the SC-IAT), we hypothesized that U.S. panel participants will display negative implicit associations with nuclear energy relative to wind, natural gas, and coal energy (Hypothesis 1 [H1]).

We also extend existing knowledge by further exploring the relationship between nuclear energy implicit associations and support for nuclear energy policy. In line with the findings of Siegrist et al. (2006), we expected that opponents of nuclear energy expansion would have negative implicit associations with nuclear energy, but supporters of nuclear energy would not display significant implicit associations with nuclear energy (Hypothesis 2 [H2]). Specifically, we hypothesized that those who desired decreased reliance on nuclear energy or opposed siting of nuclear plants would display significant negative implicit attitudes toward nuclear energy relative to wind, natural gas, and coal energy (H2a). We also hypothesized that those who desired increased reliance on nuclear energy or supported siting of nuclear plants would not display significant implicit preferences for nuclear energy relative to other energy sources (H2b).

The major contribution of this study is the test of a model of nuclear energy policy support that evaluates whether nuclear implicit associations explain unique variance in policy support after controlling for well-known correlates of nuclear policy support, namely, demographic characteristics and explicit attitudes (Ansolabehere & Konisky, 2009; Greenberg, 2009; Greenberg & Truelove, 2011). Recent models of energy policy support have included both experiential and analytical components (Truelove, 2012; Visschers, Keller, & Siegrist, 2011), though direct comparison of explicit and implicit attitudes in their ability to predict support for energy policy has not yet been evaluated. Based on previous research showing that affective components of risk perception influence nuclear energy policy support (Truelove, 2012; Visschers et al., 2011), we hypothesized that nuclear implicit associations will relate to support for nuclear energy after controlling for demographics and explicit preferences (Hypothesis 3 [H3]).

Method

Participants

Participants were selected from an opt-in panel of adult U.S. consumers. Recruitment emails were sent to 5,835 members of the panel inviting them to participate in the study in exchange for entry into one of several drawings for US$100 cash prizes. In all, 847 panelists completed the survey for a completion rate, defined as the percentage who completed the survey among all eligible panel members who were invited, of 14.52% (Callegaro & Disogra, 2009). Participants completed the study from September 14, 2011, to October 3, 2011.

Participants were removed from the data set if their data did not meet the a priori guidelines for inclusion. Although a total of 1,325 participants clicked on the link to begin the survey, we removed the responses from participants who did not complete the survey (n = 478); completed the survey in less than 6 min, which was deemed to be faster than possible for comprehension (n = 40); or had too many errors (>30%) on the IAT task, described below (n = 104). The final data set consisted of 704 participants.

Overall, most respondents were female (68%) with a median age of 48 years old. In terms of education, most had earned a 4-year college degree (25%), or were a junior college graduate (12%) or had completed some college (23%), while 18% had completed graduate work and 14% had completed high school as their highest level of education. The vast majority was not Latino (94%) and was White (88%). In terms of income, 16% earned less than US$25,000, 25% between US$25,000 and US$49,999, 26% between US$50,000 and US$74,999, 14% between US$75,000 and US$99,999, and 16% more than US$100,000. Politically, respondents categorized themselves as Democrat (34%), Republican (27%), Independent (22%), and other (13%). Although this sample was not selected to be representative of the U.S. population, a comparison with the U.S. demographics was conducted. The sample was similar to the U.S. population in terms of median income, but was more educated and older than the U.S. population and had a higher proportion of women, Whites, and non-Latinos compared with the U.S. population (U.S. Census Bureau, 2010a, 2010b, 2010c).

We compared the demographics and level of nuclear support between those who were excluded from the analyses (did not complete survey, completed survey too quickly, or made too many errors on the IAT) and those who were included in the final data set. Of those who started the study, those who were included in the sample did not differ from those excluded on any of the major demographic variables assessed including sex, race, Latino heritage, income, education (χ² tests, all ps > .05) or age (t test, p > .05). In addition, participants who completed the study and were included in the sample did not differ from those who were not included in support of nuclear reliance. However, participants who were included in the sample were significantly less likely to support siting a local nuclear facility, t(211.12) = 3.05, p = .003.Further analyses showed that those who were more supportive of siting a nuclear facility were just as likely to complete the survey but were removed because they were more likely to make errors in the IAT and to complete the survey too quickly, t(182.47) = 2.93, p = .004.

Instruments

Survey Instrument

After participants accepted the email invitation to participate in the study, they clicked a link to begin the survey on a secure website. The survey questions described here were part of a larger survey on attitudes toward nuclear, coal, natural gas, and wind energy. Only those questions relevant to the current study are discussed here.

First, demographics were collected. Next, explicit preferences for nuclear energy were assessed by asking participants to state their preference for nuclear energy relative to natural gas, coal, and wind energy (e.g., which do you prefer, coal or nuclear on a 1 [very strongly prefer coal] to 10 [very strongly prefer nuclear] scale). Higher scores on the wind–nuclear and coal–nuclear questions mean more positive explicit attitudes toward nuclear (vs. wind and coal), while higher scores for nuclear–gas question mean more negative explicit attitudes toward nuclear (vs. gas).

Next participants completed the IAT task (described next). Following the IAT, support for nuclear energy policy was assessed with two questions. Nuclear Reliance Support was measured by asking participants to rate whether they think the United States should increase or decrease reliance on nuclear energy in the future on a scale from 1 (do not use) to 6 (greatly increase). Nuclear Siting Support was measured by asking participants to rate how they would react to a new nuclear power plant being built within 25 miles of their home on a scale from 1 (strongly oppose) to 5 (strongly support).

Multiple Category BIAT

After participants answered survey questions assessing their explicit preferences, they were introduced to the IAT task. The IAT was embedded in the survey so it did not require participants to click any outside links. After completing the IAT, the remaining survey items automatically appeared.

The Multiple Category Energy BIAT used in this project was developed in collaboration with Project Implicit (www.projectimplicit.net), the developers of the original IAT and BIAT tests, using the BIAT framework (Sriram & Greenwald, 2009). The key to the BIAT framework is that participants are instructed during each trial block to have only one category (and “good”) focal, meaning to press the “I” key for the stimuli that fall into the category of interest or the “good” category and to press the “E” key for all other stimuli. This focus allows for a reduced number of trials over the traditional IAT, which requires participants to simultaneously categorize stimuli in both categories of interest and “good” and “bad” words within each trial (Sriram & Greenwald, 2009). The reduced number of trial blocks also lends itself more easily to multiple comparisons without overtaxing participants.

The Multiple Category Energy BIAT used in this study compared concept categories of nuclear, wind, natural gas, and coal energy. For each concept category, a combination of three words and four images were chosen as category exemplars by the first author after extensive pilot testing. An effort was made to be consistent across energy sources for the words used to describe the energy sources. The final words chosen were “nuclear power,” “nuclear energy,” and “nuclear”; “coal power,” “coal energy,” and “coal”; “natural gas power,” “natural gas energy,” and “natural gas”; and “wind power,” “wind energy,” and “wind.”

In addition, we initially aimed to be consistent across energy sources with the images selected for the IAT and to include both an image of the original material that is the source of the electricity and an image associated with electricity production related to that source. This was accomplished for wind energy by using images of windsocks and windmills and for nuclear energy by using images of the symbol of an atom and nuclear cooling towers. However, pilot tests showed that few participants recognized images of coal smokestacks as belonging exclusively to coal power, and even fewer recognized images of natural gas containers as relating to natural gas, indicating that these items would be poor IAT stimuli (Nosek et al., 2007). As such, images included for coal power were a coal-filled train, a pile of coal, and two images of coal pieces. Images for natural gas included images of a natural gas pipeline and natural gas flame. Both general and energy-related words were chosen as descriptors to represent the categories of “good” (best, wonderful, excellent, affordable, safe, and clean) and “bad” (worst, horrible, terrible, expensive, dangerous, and dirty).

The Multiple Category Energy BIAT consisted of 14 blocks, as shown in Table 1. The first two practice blocks and first four practice trials in each block were deleted from all analyses. Following the first two practice trial blocks, all subsequent 12 blocks followed an “interleaf pattern” such that every energy combination was repeated with the focal category reversed two trials after the original presentation. For example, Table 1 shows that for participants who received this order of blocks, Trial Block 3 consisted of coal (focal) and nuclear (nonfocal), so Block 5 consisted of the same combination but reversed with coal (nonfocal) and nuclear (focal). Although the nonfocal trials consisted of images and words related to only one energy source, the name of that energy source was not made explicit in the instructions. For example, directions for Trial Block 3 (Table 1) read “The I key is used for Coal Power or Good words. The E key is used for other images and bad words.” The order of blocks was randomly counterbalanced among participants while maintaining the interleaf pattern.

Table 1.

An Example Sequence of Blocks for the Energy BIAT.

Block	No. of trials	Function	Items assigned to key “I” (focal)	Items assigned to key “E” (nonfocal)	Energy stimuli to be sorted
0	12	Practice	Good words	Bad words	Evaluative descriptors
1	16	Practice	Wind power/good	Other images and words/bad	Natural gas–wind
2	4^a + 24	Test	Wind power/good	Other images and words/bad	Natural gas–wind
3	4 + 24	Test	Coal power/good	Other images and words/bad	Coal–nuclear
4	4 + 24	Test	Natural gas power/good	Other images and words/bad	Natural gas–wind
5	4 + 24	Test	Nuclear power/good	Other images and words/bad	Coal–nuclear
6	4 + 24	Test	Nuclear power/Good	Other images and words/bad	Natural gas–nuclear
7	4 + 24	Test	Wind power/good	Other images and words/bad	Coal–wind
8	4 + 24	Test	Natural gas power/good	Other images and words/bad	Natural gas–nuclear
9	4 + 24	Test	Coal power/good	Other images and words/bad	Coal–wind
10	4 + 24	Test	Coal power/good	Other images and words/bad	Coal–natural gas
11	4 + 24	Test	Nuclear power/good	Other images and words/bad	Nuclear–wind
12	4 + 24	Test	Natural gas power/good	Other images and words/bad	Coal–natural gas
13	4 + 24	Test	Wind power/good	Other images and words/bad	Nuclear–wind

Note: BIAT = Brief Implicit Association Test.

The first four trials of each test block are practice trials.

If a participant made an error in classifying the stimulus, a red “X” appeared in the screen until the participant pressed the correct key to classify the stimulus. (Again, participants who made errors on more than 30% of trials were eliminated from the analysis.) Trial latencies less than 300 ms were truncated to 300 ms and latencies above 3,000 ms were truncated at 3,000 ms.

IAT scores for each nuclear–other source comparison were calculated as the difference between mean response latencies for trials where nuclear energy was focal (and the other energy source was nonfocal) and where nuclear energy was nonfocal (and the other energy source was focal) divided by the standard deviation of latencies across the two trial blocks. This procedure was repeated for the three comparison energy sources so that IAT scores were calculated for wind–nuclear, gas–nuclear, and nuclear–coal. Higher IAT scores represent more positive implicit attitudes toward the first source listed. So, for wind–nuclear and gas–nuclear, higher scores mean more negative attitudes toward nuclear (vs. wind and gas), while higher scores for nuclear–coal mean more positive attitudes toward nuclear (vs. coal).

Results

To examine H1, t tests comparing mean IAT scores to zero were conducted. If people hold negative implicit associations with nuclear (relative to coal, natural gas, and wind), then IAT scores should be positive for the wind–nuclear and natural gas–nuclear comparisons and negative for the nuclear–coal comparison. Participants had negative associations with nuclear compared with wind, t(699) = 29.85, p < .001, 95% confidence interval (CI) = [.46, .52], and natural gas, t(700) = 13.39, p < .001, 95% CI = [.19, .26]. However, participants had positive associations with nuclear compared with coal, t(697) = 4.14, p < .001, 95% CI = [.04, .11]. Means and standard deviations for all IAT scores are shown along the diagonal in Table 2. Reliabilities were also assessed for the BIAT scores using split-half techniques: wind–nuclear (α = .62), gas–nuclear (α = .60), nuclear–coal (α = .62).

Table 2.

Correlations (Two-Tailed) Between Nuclear Implicit Associations, Explicit Preferences, and Nuclear Support.

Scales	IAT: wind–nuclear	IAT: gas–nuclear	IAT: nuclear–coal	Explicit: wind–nuclear	Explicit: nuclear–gas	Explicit: Coal–nuclear	Nuclear reliance support	Nuclear siting support
IAT: Wind–nuclear	0.49 (0.44)	0.28***	−0.12**	−0.26***	0.10**	−0.08*	−0.22***	−0.22***
IAT: Gas–nuclear		0.22 (0.44)	−0.17***	−0.10**	0.23***	−0.17***	−0.23***	−0.20***
IAT: Nuclear–coal			0.07 (0.46)	0.15***	−0.20***	0.34***	0.22***	0.20***
Explicit: Wind–nuclear				3.03 (2.69)	−0.36***	0.41***	0.55***	0.56***
Explicit: Nuclear–gas					7.15 (2.51)	−0.54***	−0.54***	−0.47***
Explicit: Coal–nuclear						5.06 (2.71)	0.56***	0.47***
Nuclear Reliance Support							3.59 (1.31)	0.69***
Nuclear Siting Support								1.99 (1.20)

Note: IAT = Implicit Association Test. Means (SDs) along diagonal. ns ranged from 684 to 699.

p < .05. **p < .01. ***p < .001.

Pearson correlations (two-tailed) were run between the implicit association scores for each nuclear comparison, explicit preference measures, and nuclear policy support. All implicit–explicit scores were significantly correlated. As expected, the implicit–explicit correlations were strongest among corresponding pairs (e.g., IAT: wind–nuclear had the strongest correlation with Explicit: wind–nuclear). However, the relatively modest magnitude of the correlations suggests that the implicit association scores are distinct from the explicit preference scores. The implicit association scores also significantly correlated in the expected directions with support for nuclear siting and increased reliance. The explicit preference items (compared with the implicit scores) had stronger correlations with the nuclear support variables.

The relationship between nuclear IAT scores and support for nuclear energy was examined in more detail using one-way ANOVAs testing whether IAT scores differed by level of support for each of the two dependent measures: support for siting and desire for increased reliance (Tables 3 and 4). As displayed in the correlations above, IAT scores were dependent on the level of nuclear support. Specifically, implicit preferences for nuclear over coal increased with increasing support for nuclear reliance, F(5, 685) = 7.79, p < .001, partial η² = .05, and siting, F(4, 687) = 9.12, p < .001, partial η² = .05. Implicit preference for natural gas over nuclear was related to decreasing support for nuclear reliance, F(5, 688) = 7.70, p < .001, partial η² = .05, and siting, F(4, 690) = 7.02, p < .001, partial η² = .04. Similarly, implicit preference for wind over nuclear was associated with decreasing support for nuclear reliance, F(5, 687) = 7.79, p < .001, partial η² = .05, and siting, F(4, 689) = 9.00, p < .001, partial η² = .05. Planned pairwise comparisons, using t tests (p < .05) were conducted. Those who wanted to stop using nuclear power, greatly decrease, or decrease nuclear reliance had more negative implicit attitudes toward nuclear (vs. natural gas, wind, and coal) compared with those who wanted to greatly increase nuclear reliance. In addition, those who strongly opposed siting had significantly higher wind–nuclear and gas–nuclear IAT scores and significantly lower nuclear–coal IAT scores than those who supported or strongly supported siting.

Table 3.

IAT Scores by Nuclear Reliance.

	IAT scores
	IAT: Nuclear–coal				IAT: Natgas–nuclear				IAT: Wind–nuclear
	M	SD	95% CI		M	SD	95% CI		M	SD	95% CI
Nuclear reliance	M	SD	Low	High	M	SD	Low	High	M	SD	Low	High
Do not use	.03^a,b	.42	−.16	.22	.41^a	.46	.23	.60	.66^a	.60	.48	.84
Greatly decrease	−.05^a	.50	−.13	.02	.33^a,b	.47	.26	.40	.60^a,b	.44	.53	.66
Decrease	.00^a,b	.46	−.07	.07	.30^a,b,c	.40	.23	.36	.53^a,b	.40	.47	.60
Stay the same	.10^a,b,c	.45	.03	.17	.18^b,c,d	.43	.12	.25	.47^a,b	.42	.40	.53
Increase	.20^b,c	.43	.12	.28	.09^c,d	.44	.02	.16	.44^b	.41	.37	.51
Greatly increase	.29^c	.37	.17	.41	.07^d	.43	−.04	.19	.22^c	.42	.10	.33

Note: IAT = Implicit Association Test; Natgas = natural gas; CI = confidence interval. Do not use, n = 22, Greatly decrease, n = 148, Decrease, n = 165, Stay the same, n = 174, Increase, n = 134, Greatly increase, n = 54. Column means that do not share superscripts are significantly different from each other per pairwise comparison t test at p < .05. Numbers in bold represent IAT scores that are not significantly different from 0 per t tests at p < .05.

Table 4.

IAT Scores by Nuclear Siting.

	IAT scores
	IAT: Nuclear–coal				IAT: Natgas–nuclear				IAT: Wind–nuclear
			95% CI				95% CI				95% CI
Nuclear siting	M	SD	Low	High	M	SD	Low	High	M	SD	Low	High
Strongly oppose	−.03^a	.48	−.08	.02	.29^a	.45	.25	.34	.57^a	.43	.52	.61
Oppose	.17^a,b	.43	.10	.24	.23^a,b	.42	.16	.30	.51^a	.41	.45	.58
Neither oppose nor support	.14^a,b	.43	.05	.23	.12^a,b,c	.42	.03	.21	.39^a,b	.44	.30	.47
Support	.21^b	.43	.10	.31	.09^b,c	.42	−.01	.19	.30^b	.41	.20	.40
Strongly support	.24^b	.44	.08	.40	.02^c	.48	−.14	.17	.32^b	.43	.17	.47

Note: IAT = Implicit Association Test; Natgas = natural gas; CI = confidence interval. Strongly oppose, n = 344, Oppose, n = 153, Neither oppose nor support, n = 99, Support, n = 71, Strongly support, n = 31. Numbers in bold represent IAT scores that are not significantly different from 0 per t tests at p < .05. Column means that do not share superscripts are significantly different from each other per pairwise comparison t test at p < .05.

Following the procedure of Siegrist et al. (2006), additional t tests were conducted to examine whether participants’ implicit associations with nuclear energy are significantly different from zero as their level of support for nuclear power varies (H2). For the nuclear–natural gas comparison, those who strongly supported nuclear siting and those who wanted to greatly increase nuclear reliance did not express implicit associations with nuclear energy relative to natural gas energy that differed from zero, and those who opposed or wanted to decrease reliance on nuclear power had significantly negative associations with nuclear energy compared with natural gas. Interestingly, the results for the nuclear–coal comparison were basically the opposite of that for the nuclear–natural gas comparison. Those who strongly opposed nuclear energy or wanted to decrease nuclear reliance did not have implicit associations with nuclear energy relative to coal that were different from zero. But, those who supported nuclear energy or wanted to increase nuclear reliance did express implicit positive attitudes toward nuclear (compared with coal). For the wind–nuclear comparison, all groups expressed significant negative attitudes toward nuclear versus wind regardless of how supportive they were of nuclear power reliance and siting. In other words, those who strongly opposed nuclear policy had a negative implicit attitude toward nuclear compared with wind and natural gas but not compared with coal. Those who were strongly supportive of nuclear policy had a negative implicit attitude toward nuclear versus wind, a positive implicit attitude toward nuclear versus coal, and a nonsignificant attitude toward nuclear versus gas.

Next we examined whether IAT scores predict support for nuclear energy after other well-known correlates with nuclear support (i.e., explicit preferences and demographics) are accounted for (H3). As shown in Table 5, hierarchical linear regression models were run predicting each of the nuclear energy policy support variables: support for siting a local nuclear power plant (nuclear siting) and desire for increased reliance on nuclear power in the United States (nuclear reliance). In Step 1 of the regressions, explicit preferences for nuclear energy (relative to gas, coal, and wind) and demographics were entered. All demographic variables were dummy-coded except for age, which was entered as a continuous variable. Reference categories for each demographic variable were as follows: education (no college), sex (female), ethnicity (not Latino), income (>US$100,000), political party (Republican), and race (White). In Step 2, implicit associations related to nuclear energy (relative to gas, coal, and wind) were added to the model. Both policy support variables, Nuclear Reliance Support and Nuclear Siting Support, approximated a normal distribution (Skewness < |3|, Kurtosis < |10|, and q–q plots appearing normal).

Table 5.

Multiple Linear Regression Results Predicting Support for Siting Nuclear Facilities and Increased Reliance on Nuclear Energy.

		Predicting nuclear reliance support			Predicting nuclear siting support
Model	Predictors	β	B	SE	β	B	SE
Model 1	Explicit and demographic model
	Explicit preferences
	Explicit: Coal–nuclear	.26***	.12	.02	.16***	.07	.02
	Explicit: Nuclear–gas	−.28***	−.14	.02	−.23***	−.11	.02
	Explicit: Wind–nuclear	.30***	.14	.02	.38***	.17	.02
	Demographics
	Age	.04	.00	.00	.00	.00	.00
	Education: Some college vs. no college	−.02	−.06	.11	.05	.12	.11
	Education: College Grad vs. no College	.03	.07	.11	.07	.17	.11
	Male (vs. female)	.09**	.27	.08	.14***	.36	.08
	Latino (vs. not Latino)	−.01	−.06	.20	−.04	−.26	.20
	Income: <US$50K vs. > US$100K	−.08	−.21	.11	.02	.04	.11
	Income: US$50K-US100K vs. > US$100K	−.05	−.14	.11	−.01	−.02	.11
	Democrat (vs. Republican)	−.06	−.15	.10	−.05	−.12	.09
	Other Party (vs. Republican)	−.05	−.14	.09	−.05	−.13	.09
	Black (vs. White)	−.01	−.05	.17	−.05	−.28	.17
	Other Race (vs. White)	−.01	−.06	.17	.02	.13	.16
	Constant		3.52	.27		1.76	.26
Model 2	Implicit, explicit, and demographic model
	Explicit preferences
	Explicit: Coal–nuclear	.25***	.12	.02	.15***	.06	.02
	Explicit: Nuclear–gas	−.26***	−.14	.02	−.21***	−.10	.02
	Explicit: Wind–nuclear	.27***	.13	.02	.36***	.16	.02
	Demographics
	Age	.06*	.01	.00	.02	.00	.00
	Education: Some college vs. no college	−.02	−.06	.11	.05	.12	.11
	Education: College grad vs. no college	.03	.08	.11	.07	.18	.11
	Male (vs. female)	.09**	.26	.08	.13***	.35	.08
	Latino (vs. not Latino)	−.01	−.10	.20	−.04	−.28	.20
	Income: <US$50K vs. >US$100K	−.08	−.20	.11	.02	.05	.11
	Income: US$50-US100K vs. >US$100K	−.05	−.13	.11	−.01	−.02	.11
	Democrat (vs. Republican)	−.05	−.15	.09	−.05	−.12	.09
	Other Party (vs. Republican)	−.05	−.14	.09	−.06	−.14	.09
	Black (vs. White)	.00	−.02	.17	−.05	−.26	.17
	Other Race (vs. White)	−.01	−.06	.16	.02	.13	.16
	Implicit associations
	IAT: Gas–nuclear	−.08**	−.23	.09	−.07*	−.19	.08
	IAT: Nuclear–coal	.01	.02	.08	.02	.06	.08
	IAT: Wind–nuclear	−.08*	−.23	.09	−.05	−.14	.09
	Constant		3.58	.27		1.79	.26
	R² Model 1	.50***			.45***
	R² Model 2	.51***			.46***
	ΔR²	.01***			.01*
	n	674			675

Note: IAT = Implicit Association Test.

p < .05. **p < .01. ***p < .001.

The explicit and demographics-only model explained large amounts of variance in nuclear reliance and nuclear siting (50% and 45%, respectively). Explicit preferences for nuclear relative to coal, gas, and wind were significantly associated with nuclear support. The only demographic variable that was related to support was sex, with men being more supportive of nuclear siting and reliance than women.

The addition of implicit energy attitudes to the model explained a significant, though small, amount of additional variance. In the full model, implicit attitudes toward nuclear (vs. gas) significantly predicted siting and reliance support and implicit attitudes toward nuclear (vs. wind) significantly predicted reliance support. As implicit attitudes toward nuclear (vs. natural gas and wind) became more positive, participants were more supportive of nuclear-related policies. However, implicit attitudes toward nuclear compared with coal were not significant in predicting policy support in the second model. Explicit energy preferences and sex remained significant predictors of energy support. Age was significant in the full model, with older participants being more supportive of reliance than younger participants. For reliance support, the magnitude of the relationships between the implicit attitudes and nuclear support were similar to that of the demographic variables of age and sex. For siting support, implicit attitudes toward nuclear (vs. gas) was a stronger predictor than age but weaker than sex.

Discussion

Dual-process models of risk perception outline both an analytical and experiential risk evaluation process (Loewenstein et al., 2001; Slovic et al., 2004). Although researchers investigating nuclear risk perceptions have a long history of incorporating affect-related measurements in their studies (Keller et al., 2012; Peters & Slovic, 1996; Sjöberg, 2007; Slovic, 1987; Truelove, 2012), only recently has there been an effort to assess implicit nuclear attitudes (Dohle et al., 2010; Siegrist et al., 2006) and much is still unknown about the relationship between implicit nuclear attitudes and policy support. Building on the work of Siegrist and colleagues (Dohle et al., 2010; Siegrist et al., 2006), we assessed implicit attitudes toward nuclear energy relative to coal, natural gas, and wind among a sample of participants drawn from a U.S. consumer panel. By assessing implicit perceptions alongside explicit perceptions and testing their effects on policy support, this research contributes to a more complete understanding of nuclear risk perceptions (Visschers, Meertens, Passchier, & de Vries, 2007).

We hypothesized that participants would have negative implicit associations with nuclear energy compared with wind, natural gas, and coal energy (H1). Our results provide only partial support for this hypothesis. Participants did hold negative implicit attitudes about nuclear energy compared with natural gas and wind, in line with Siegrist et al.’s (2006) finding that Swiss students had negative implicit attitudes toward nuclear energy compared with hydroelectricity. However, participants also displayed positive implicit attitudes about nuclear compared with coal (cf. Blanton & Jaccard, 2006, who argue against a true zero score for IAT scores). This finding is especially interesting as this study was conducted after the nuclear disaster at Fukushima, an event expected to have a strong influence on automatic responses to nuclear energy. However, this result reinforces an emerging trend in research that indicates that coal may have become the “new nuclear.” Compared with nuclear energy, coal energy is perceived as causing more environmental harm (Ansolabehere & Konisky, 2009), has less support among the public for increasing reliance (Greenberg, 2009), and elicits equally negative feelings (Truelove, 2012).

We also hypothesized that the relationship between nuclear BIAT scores and nuclear policy support would follow the pattern obtained by Siegrist et al.’s (2006) comparison of nuclear energy and hydropower (H2). Siegrist et al. (2006) found that those who support replacing nuclear plants with new plants and building new nuclear plants in Switzerland did not have IAT scores that differed from zero, but those opposed and undecided about nuclear energy had significant negative implicit attitudes toward nuclear versus hydropower. We found the same pattern of results as Siegrist et al. for the nuclear–natural gas comparison: Those who strongly supported nuclear policy did not have nuclear–natural gas IAT scores that differed from zero, but those who opposed nuclear policy had significant negative implicit attitudes toward nuclear (vs. natural gas). However, the comparison between nuclear and coal was almost the opposite pattern of results: Those who strongly supported nuclear policy had positive implicit attitudes toward nuclear (vs. coal), while those who opposed nuclear policy did not have nuclear–coal IAT scores that differed from zero. Finally, for the nuclear–wind comparison, all groups displayed a negative implicit attitude toward nuclear versus wind, regardless of their stated level of nuclear policy support. Stated another way, at an implicit level, strong supporters of nuclear policy like nuclear more than coal and just as much as natural gas, but prefer wind to nuclear, while opponents like wind and gas more than nuclear and dislike nuclear just as much as coal.

The pattern of results from this analysis suggests that those who are supportive of nuclear power may have more positive implicit attitudes toward nuclear relative to coal, whereas those who are opposed to nuclear power may have more negative implicit attitudes toward nuclear relative to natural gas. Additional research should further explore this possibility using alternative methodologies such as focus groups or interviews, as these have been used successfully in the United Kingdom to provide a deeper understanding of perceptions of nuclear power (Bickerstaff, Lorenzoni, Pidgeon, Poortinga, & Simmons, 2008; Pidgeon, Lorenzoni, & Poortinga, 2008; Venables, Pidgeon, Simmons, Henwood, & Parkhill, 2009).

Our results go furthest toward extending existing research by testing a model of policy support that includes implicit attitudes alongside explicit attitudes and demographics, which have been shown to relate to support for nuclear energy policy (Ansolabehere & Konisky, 2009; Greenberg, 2009; Greenberg & Truelove, 2011). We hypothesized that implicit nuclear attitudes would be related to nuclear policy support and that they would explain significant additional variance above and beyond demographics and explicit attitudes (H3). We found correlations between implicit attitudes and policy support that ranged from .20 to .23, in line with, but somewhat lower than, an average correlation of .27 derived from a recent meta-analysis of implicit association–criterion correlations in various domains (Greenwald et al., 2009). Our results also showed that implicit attitudes alone explained approximately 10% of variance in both measures of nuclear policy support. In addition, implicit attitudes toward nuclear versus gas predicted support for nuclear siting, and expanded reliance and attitudes toward nuclear versus wind predicted nuclear reliance after controlling for demographics and explicit attitudes.

Although significant in their contribution, implicit attitudes explained only a modest amount of additional variance in nuclear support and siting after controlling for demographics and explicit attitudes. However, the results here may actually be underestimates of the effect of implicit attitudes on energy support. The relatively low reliabilities obtained for the BIAT scores, which were in line with but on the low end of previous research (Sriram & Greenwald, 2009), may have attenuated the relationships between the implicit attitudes and nuclear support. However, the presentation of the explicit measures before the implicit measures in the survey, as has been done in previous work (Sriram & Greenwald, 2009), may have amplified the effects of the implicit attitudes. Due to the fact that the stimuli chosen for the BIAT were not necessarily those that all participants would spontaneously associate with the energy sources, we were reluctant to expose participants to the BIAT before the explicit attitudes so as not to taint their explicit attitudes. However, it is possible, as one reviewer pointed out, that the explicit attitude questions brought subconscious attitude elements to the surface, which then primed participants for the BIAT and resulted in stronger BIAT scores. This study points to the potential value in incorporating implicit measures in risk perception studies but underscores the need for additional methodological guidelines for doing so.

The data obtained in this study were drawn from a population of consumer panelists who opted to complete Internet-based research studies in exchange for compensation. Although this group provided a nice comparison for the results of Siegrist and colleagues’ Swiss student participants, the method used was not meant to provide a representative sample of the U.S. public and the sample was found to differ in many respects from the general U.S. public. Future research should attempt to replicate these findings with a probability-based sample. In addition, future researchers should particularly focus on samples of residents who live near facilities of interest as other research has shown that affective responses to nuclear differ between those who live near sites and others (Greenberg, 2012; Keller et al., 2012).

In response to the question raised in the introduction, what did we learn by assessing implicit nuclear attitudes in comparison with three other energy sources, instead of just using one comparison energy source? Had we chosen wind as our sole comparison, we would not have learned much about how implicit nuclear attitudes relate to policy support as all participants, regardless of level of support for nuclear policy, had negative implicit attitudes toward nuclear versus wind. Wind power is almost uniformly supported by the American public (CNN Opinion Research, 2011; Greenberg, 2009; cf. the Cape Wind project, which has strong opposition by some groups; Kempton, Firestone, Lilley, Rouleau, & Whitaker, 2005; Phadke, 2010), and the present research shows that implicit attitudes toward wind power are also very positive (relative to nuclear power). If we had chosen coal as the sole comparison, we would have shown that implicit nuclear power attitudes vary with the level of policy support but not after controlling for demographics and explicit attitudes. It seems we learned the most about implicit nuclear attitudes in the nuclear–natural gas comparison, which showed that implicit attitudes toward nuclear (vs. natural gas) were related to policy support after controlling for demographics and explicit attitudes. The results reinforce that the choice of reference category for the IAT is extremely important (Nosek et al., 2007). The use of the multicategory IAT in the present study allowed for a comparison of the energy sources that are at the center of many policy discussions in the United States and enriched our understanding of perceptions of nuclear energy relative to these other sources. However, additional research should consider assessing people’s implicit attitudes toward other sources in the United States, specifically hydroelectricity and solar power. In addition, as the energy mix continues to evolve, an understanding of how implicit attitudes toward energy sources change over time is needed. In the short time since these data were collected, media discussions of fracking practices for natural gas have grown exponentially, whereas the media firestorm related to the Fukushima accident has died down. Researchers should continue to monitor implicit and explicit perceptions of energy sources and the influence of these perceptions on policy support against this changing energy landscape.

Of course, methods (such as surveys) to assess the analytical component of people’s energy-related risk perception process benefit from years of refinement and validation and can be administered with relative ease and minimal cost. However, the present results are part of a growing literature illustrating the need to incorporate measures to assess the experiential component of people’s energy-related risk perceptions (Dohle et al., 2010; Peters & Slovic, 2007; Siegrist et al., 2006; Truelove, 2012). Although measures such as the IAT are sometimes more difficult to administer than standard surveys, researchers who incorporate both implicit and explicit measures of attitudes toward risk are able to garner a fuller understanding of risk perceptions (Visschers et al., 2007). The present study was able to incorporate the BIAT into a standard Internet survey so that both explicit and implicit attitudes were gathered from participants in the same session. Future researchers should investigate opportunities such as this to include both types of assessments in studies of risk perception.

Considering this project is one of the first to use the BIAT in the study of energy source perceptions, the field would surely benefit from additional work using variations on the stimuli we used for the BIAT. In the present study, great care was taken to select BIAT stimuli that would be representative of the energy sources, while also being recognizable by the participants. However, some of the stimuli selected may have elicited more positive or negative affect from participants and that affect may have influenced the energy support questions (Truelove, 2012). In addition, our selection of the attribute stimuli was meant to capture risk perceptions (e.g., safe, clean, dangerous, and dirty) and so differed from more traditional IAT negative and positive words (e.g., love, hate, good, bad). However, considering previous research has shown that the selection of stimuli for the IAT can influence IAT scores (Bluemke & Friese, 2006; Gast & Rothermund, 2010; Steffens & Plewe, 2001), future work should consider using alternate attribute stimuli. Overall, we hope our project will be a launching point for additional research and that future researchers will build on the present method to further establish the validity of the findings.

Overall, the pattern of results shows that the relationship between implicit nuclear attitudes and policy support is more complex than revealed in previous research. As electricity demand continues to increase globally, the mix of sources providing energy is constantly fluctuating. As new sources enter the fray and new plants of traditional sources come on-line, an understanding of how people perceive one energy source relative to another is a key tool for gauging public support. We found that support for nuclear energy differed with different levels of implicit attitudes toward nuclear energy, but the comparison energy source was also very important. As utilities make trade-offs about which energy sources to pursue in planning new energy facilities, it would behoove them to use tools such as those used in this study to garner an understanding of the public’s relative energy attitudes at both the explicit and implicit level.

Footnotes

Authors’ Note

This work was undertaken while Truelove was a postdoctoral fellow affiliated with Vanderbilt’s Consortium for Risk Evaluation With Stakeholder Participation and the Vanderbilt Institute for Energy and Environment. This report was prepared as an account of work sponsored by an Agency of the U.S. Government. Neither the U.S. Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied; or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed; or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the U.S. Government or any agency thereof. The opinions, findings, conclusions, or recommendations expressed herein are those of the author and do not necessarily represent the views of the Department of Energy or Vanderbilt University.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This report is partially based on work supported by the U.S. Department of Energy, under Cooperative Agreement Number DE-FC01-06EW07053 titled “The Consortium for Risk Evaluation With Stakeholder Participation III” awarded to Vanderbilt University.

Author Biographies

Heather Barnes Truelove is currently an assistant professor of applied psychology in the psychology department at the University of North Florida. When this work was undertaken, she was a research associate at Vanderbilt University in the Consortium for Risk Evaluation With Stakeholder Participation and the Vanderbilt Institute for Energy and Environment. Her research interests center on the social psychology of proenvironmental behavior with lines of research focusing on perceptions of nuclear energy, individual behaviors that mitigate climate change, and agricultural adaptation.

Michael R. Greenberg is professor and associate dean of the Faculty at the Edward J. Bloustein School of Planning and Public Policy at Rutgers University, the director of the National Center for Neighborhood and Brownfields Redevelopment, and the director of the Center for Transportation Safety, Security, and Risk. His research interests include perceptions of nuclear waste management, risk analysis, environmental health policy, and urban neighborhood redevelopment.

Charles W. Powers is professor of environmental engineering in the civil and environmental engineering department at Vanderbilt University where he is also coprincipal investigator (with David Kosson) of the Consortium for Risk Evaluation With Stakeholder Participation (CRESP III). His research focuses on understanding and improving the technical, social, and regulatory interface for nuclear waste management.

References

Ansolabehere

(2007). Public Attitudes Toward America’s Energy Options: Insights for Nuclear Energy. Report of the 2007 MIT Energy Survey. CEEPR Working Paper 07-002 and CANES working paper.

Ansolabehere

Konisky

D. M.

(2009). Public attitudes toward construction of new power plants. Public Opinion Quarterly, 73, 566-577. doi:10.1093/poq/nfp041

Bickerstaff

Lorenzoni

Pidgeon

N. F.

Poortinga

Simmons

(2008). Reframing nuclear power in the UK energy debate: Nuclear power, climate change mitigation and radioactive waste. Public Understanding of Science, 17, 145-169. doi:10.1177/0963662506066719

Blanton

Jaccard

(2006). Arbitrary metrics in psychology. American Psychologist, 61, 27-41. doi:10.1037/0003-066X.61.1.27

Bluemke

Friese

(2006). Do features of stimuli influence IAT effects? Journal of Experimental Social Psychology, 42, 163-176. doi:10.1016/j.jesp.2005.03.004

Bluemke

Friese

(2008). Reliability and validity of the Single-Target IAT (ST-IAT): Assessing automatic affect towards multiple attitude objects. European Journal of Social Psychology, 38, 977-997. doi:10.1002/ejsp

Callegaro

Disogra

(2009). Computing response metrics for online panels. Public Opinion, 72(5), 1008-1032.

Chaiken

(1980). Heuristic versus systematic information processing and the use of source versus message cues in persuasion. Journal of Personality and Social Psychology, 39, 752-766. doi:10.1037//0022-3514.39.5.752

CNN Opinion Research. (2011). March 18-20 poll. Retrieved from http://i2.cdn.turner.com/cnn/2011/images/03/22/rel5c.pdf

10.

Dohle

Keller

Siegrist

(2010). Examining the relationship between affect and implicit associations: Implications for risk perception. Risk Analysis, 30, 1116-1128. doi:10.1111/j.1539-6924.2010.01404.x

11.

Epstein

(1994). Integration of the cognitive and the psychodynamic unconscious. American Psychologist, 49, 709-724. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/8092614

12.

Finkel

M. L.

Law

(2011). The rush to drill for natural gas: A public health cautionary tale. American Journal of Public Health, 101, 784-785. doi:10.2105/AJPH.2010.300089

13.

Finucane

M. L.

Alhakami

Slovic

Johnson

S. M.

(2000). The affect heuristic in judgments of risks and benefits. Journal of Behavioral Decision Making, 13, 1-17. doi:10.1002/(SICI)1099-0771(200001/03)13:1<1::AID-BDM333>3.0.CO;2-S

14.

Fountain

(2011, December 13). Add quakes to rumblings over gas rush. The New York Times, p. D1.

15.

Gast

Rothermund

(2010). When old and frail is not the same: Dissociating category and stimulus effects in four implicit attitude measurement methods. Quarterly Journal of Experimental Psychology, 63, 479-498. doi:10.1080/17470210903049963

16.

Greenberg

M. R.

(2009). Energy sources, public policy, and public preferences: Analysis of US national and site-specific data. Energy Policy, 37, 3242-3249. doi:10.1016/j.enpol.2009.04.020

17.

Greenberg

M. R.

(2012). The affect heuristic, correspondence analysis, and understanding LULUs. Risk Analysis, 32, 478-480. doi:10.1111/j.1539-6924.2012.01793.x

18.

Greenberg

M. R.

Truelove

H. B.

(2011). Energy choices and risk beliefs: Is it just global warming and fear of a nuclear power plant accident? Risk Analysis, 31, 819-831. doi:10.1111/j.1539-6924.2010.01535.x

19.

Greenwald

A. G.

Mcghee

D. E.

Schwartz

J. L. K.

(1998). Measuring individual differences in implicit cognition. Journal of Personality and Social Psychology, 74, 1464-1480.

20.

Greenwald

A. G.

Poehlman

T. A.

Uhlmann

E. L.

Banaji

M. R.

(2009). Understanding and using the Implicit Association Test: III. Meta-analysis of predictive validity. Journal of Personality and Social Psychology, 97, 17-41. doi:10.1037/a0015575

21.

Karpinski

Steinman

R. B.

(2006). The single category Implicit Association Test as a measure of implicit social cognition. Journal of Personality and Social Psychology, 91, 16-32. doi:10.1037/0022-3514.91.1.16

22.

Keller

Visschers

Siegrist

(2012). Affective imagery and acceptance of replacing nuclear power plants. Risk Analysis, 32, 464-477. doi:10.1111/j.1539-6924.2011.01691.x

23.

Kempton

Firestone

Lilley

Rouleau

Whitaker

(2005). The offshore wind power debate: Views from Cape Cod. Coastal Management, 33, 119-149. doi:10.1080/08920750590917530

24.

Klemow

K. M.

(2012). Science and energy policy: Leave the cherries alone. Frontiers in Ecology and the Environment, 10, 115. doi:10.1890/1540-9295-10.3.115

25.

Loewenstein

G. F.

Weber

E. U.

Hsee

C. K.

Welch

(2001). Risk as feelings. Psychological Bulletin, 127, 267-286. doi:10.1037//0033-2909.127.2.267

26.

Meijnders

A. L.

Midden

C. J.

Wilke

H. A.

(2001). Role of negative emotion in communication about CO2 risks. Risk Analysis, 21, 955-966. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/11798129

27.

Nosek

B.A.

Greenwald

A.G.

Banaji

M.R.

(2007). The Implicit Association Test at age 7: A methodological and conceptual review. In Bargh

J.A.

(Ed.), Social psychology and the unconscious: The automaticity of higher mental processes. London: Psychology Press; 2007. p. 265-292.

28.

Peters

Slovic

(1996). The role of affect and worldviews as orienting dispositions in the perception and acceptance of nuclear power. Journal of Applied Social Psychology, 26, 1427-1453.

29.

Peters

Slovic

(2007). Affective asynchrony and the measurement of the affective attitude component. Cognition & Emotion, 21, 300-329. doi:10.1080/02699930600911440

30.

Petty

R. E.

Cacioppo

J. T.

(1986). Communication and persuasion: Central and peripheral routes to attitude change. New York, NY: Springer-Verlag.

31.

Phadke

(2010). Steel forests or smoke stacks: The politics of visualisation in the Cape Wind controversy. Environmental Politics, 19, 1-20. doi:10.1080/09644010903396051

32.

Pidgeon

Lorenzoni

Poortinga

(2008). Climate change or nuclear power—No thanks! A quantitative study of public perceptions and risk framing in Britain. Global Environmental Change, 18, 69-85. doi:10.1016/j.gloenvcha.2007.09.005

33.

Rosa

E. A.

Clark

D. L.

Jr (1999). Historical routes to technological gridlock: Nuclear technology as prototypical vehicle. Research in Social Problems and Public Policy, 7, 21-57.

34.

Rundmo

(2002). Associations between affect and risk perception. Journal of Risk Research, 5, 119-135. doi:10.1080/136698702753499597

35.

Siegrist

Keller

Cousin

M.-E.

(2006). Implicit attitudes toward nuclear power and mobile phone base stations: Support for the affect heuristic. Risk Analysis, 26, 1021-1029. doi:10.1111/j.1539-6924.2006.00797.x

36.

Sjöberg

(2007). Emotions and risk perceptions. Risk Management, 9, 223-237.

37.

Slovic

(1987). Perception of risk. Science, 236(4799), 280-285.

38.

Slovic

Finucane

M. L.

Peters

MacGregor

D. G.

(2004). Risk as analysis and risk as feelings: Some thoughts about affect, reason, risk, and rationality. Risk Analysis, 24, 311-322. doi:10.1111/j.0272-4332.2004.00433.x

39.

Spence

Townsend

(2008). Spontaneous evaluations: Similarities and differences between the affect heuristic and implicit attitudes. Cognition & Emotion, 22, 83-93. doi:10.1080/02699930701298432

40.

Sriram

Greenwald

A. G.

(2009). The Brief Implicit Association Test. Experimental Psychology, 56, 283-294. doi:10.1027/1618-3169.56.4.283

41.

Steffens

Plewe

(2001). Items’ cross-category associations as a confounding factor in the Implicit Association Test. Zeitschrift für Experimentelle Psychologie, 48, 123-134.

42.

Tracy

(2012, March 31). U.S. approves nuclear plants in South Carolina. The Wall Street Journal, pp. 2-3. Retrieved from http://online.wsj.com/article/SB10001424052702303816504577313873449843052.html

43.

Truelove

H. B.

(2012). Energy source perceptions and policy support: Image associations, emotional evaluations, and cognitive beliefs. Energy Policy, 45, 478-489. doi:10.1016/j.enpol.2012.02.059

44.

U.S. Census Bureau. (2010a). Table DP02. Selected social characteristics in the United States: 2010 American Community Survey 1-Year Estimates. Retrieved from http://factfinder2.census.gov/faces/tableservices/jsf/pages/productview.xhtml?pid=ACS_10_1YR_DP02&prodType=table

45.

U.S. Census Bureau. (2010b). Table DP-1. Profile of general demographic characteristics: 2010. Retrieved from http://factfinder2.census.gov/faces/tableservices/jsf/pages/productview.xhtml?pid=DEC_10_DP_DPDP1&prodType=table

46.

U.S. Census Bureau. (2010c). Table S1901. Income in the past 12 months (in 2010 inflation-adjusted dollars): 2010 American Community Survey 1-Year Estimates. Retrieved from http://factfinder2.census.gov/faces/tableservices/jsf/pages/productview.xhtml?pid=ACS_10_1YR_S1901&prodType=table

47.

U.S. Energy Information Administration. (2012a). Electricity explained. Retrieved from http://www.eia.gov/energyexplained/index.cfm?page=electricity_use

48.

U.S. Energy Information Administration. (2012b). Electric power monthly: March 2012 (Vol. 0226, pp. 1-171). Retrieved from http://www.eia.gov/cneaf/electricity/epm/epm_sum.html

49.

Venables

Pidgeon

Simmons

Henwood

Parkhill

(2009). Living with nuclear power: A Q-method study of local community perceptions. Risk Analysis, 29, 1089-1104. doi:10.1111/j.1539-6924.2009.01259.x

50.

Visschers

V. H. M.

Keller

Siegrist

(2011). Climate change benefits and energy supply benefits as determinants of acceptance of nuclear power stations: Investigating an explanatory model. Energy Policy, 39, 3621-3629. doi:10.1016/j.enpol.2011.03.064

51.

Visschers

V. H. M.

Meertens

R. M.

Passchier

W. F.

de Vries

N. K.

(2007). An associative approach to risk perception: Measuring the effects of risk communications directly and indirectly. Journal of Risk Research, 10, 371-383. doi:10.1080/13669870701252463

52.

Visschers

V. H. M.

Siegrist

(2008). Exploring the triangular relationship between trust, affect, and risk perception: A review of the literature. Risk Management, 10, 156-167. doi:10.1057/rm.2008.1

53.

Wald

M. L.

(2012, February 10). Federal regulators approve two nuclear reactors in Georgia. The New York Times, p. B3. Retrieved from http://www.nytimes.com/2012/02/10/business/energy-environment/2-new-reactors-approved-in-georgia.html

54.

Zajonc

R. B.

(1980). Feeling and thinking: Preferences need no inferences. American Psychologist, 35, 151-175.