Energy Research & Social Science 11 (2016) 29–39
Contents lists available at ScienceDirect
Energy Research & Social Science
journal homepage: www.elsevier.com/locate/erss
Original research article
Women and nuclear energy: Examining the gender divide in
opposition to nuclear power among swedish citizens and politicians
Aksel Sundström a,∗ , Aaron M. McCright b
a
b
Department of Political Science, The Quality of Government Institute, University of Gothenburg, Box 100, S-405 30 Gothenburg, Sweden
Department of Sociology and Lyman Briggs College, Michigan State University, East Lansing, MI 48824, United States
a r t i c l e
i n f o
Article history:
Received 9 May 2015
Received in revised form 5 August 2015
Accepted 18 August 2015
Keywords:
Nuclear power
Gender
Public opinion
Politicians
a b s t r a c t
Whether or not there will be a r̈enaissanceöf nuclear power in the near future may depend upon the
nature of support for this energy source among citizens and elected officials. Continued examination of
the predictors of opposition to nuclear power therefore remains quite policy relevant. While the existing literature finds modest but consistent gender differences in attitudes towards nuclear power in the
general publics of several Western countries, the robustness of this relationship has seldom been investigated across time or among elected officials. This paper addresses both of these gaps. First, analyzing
nationally representative data from the Swedish general public between 1986 and 2011, we confirm that
the theoretically expected gender divide in opposition to nuclear power—whereby women report greater
opposition than do men—is indeed robust over time. Second, examining data from three recent surveys of
elected officials at the local, regional, and national levels in Sweden, we find that female elected officials
at each polity level report greater opposition to nuclear power than their male counterparts. Our results
are consistent with the health and safety concerns argument, whereby women are less supportive than
are men of technologies with considerable perceived health and safety risks.
© 2015 Elsevier Ltd. All rights reserved.
1. Introduction
Even though public support for nuclear power remains lukewarm at best in many countries (e.g., Organization for Economic
Cooperation and Development, 2010), recent years have witnessed
renewed calls for reinvigorating the nuclear energy programs of
many countries. Indeed, some environmental organizations (e.g.,
[34] and climate scientists (e.g., CNN Staff, 2013) are actively
promoting nuclear power as a way to maintain or increase electricity generation while reducing our fossil fuels consumption that
contributes to anthropogenic climate change. Whether or not a
“renaissance” of nuclear power actually occurs may depend, at least
in part, upon the nature of support for it among citizens and elected
officials. This raises the premium on improving our understanding of the dynamics of opinion toward nuclear power among the
general public and elected officials.
We focus here on one of the most well-known patterns in the
literature: that women are significantly less supportive of nuclear
power than are men (e.g., [1,6,7,9,10,20,26,28,36,50]). The global
nuclear power industry has been so concerned about women’s low
∗ Corresponding author.
E-mail address: [email protected] (A. Sundström).
http://dx.doi.org/10.1016/j.erss.2015.08.008
2214-6296/© 2015 Elsevier Ltd. All rights reserved.
level of support for nuclear power that it formed in 1992 the interest group Women in Nuclear, which has since worked to increase
support for nuclear power among women around the world [54].
While this gender divide occurs in many countries and on
a range of nuclear power attitudes, scholars largely have been
unable—because of lack of data availability—to examine the robustness of this gender divide over time and across levels of a country’s
polity (e.g., local government, state or regional government, and
national government). Addressing both of these gaps is important.
Examining the robustness of the gender divide on nuclear power
attitudes over time will help us better understand the likelihood
that policy changes, nuclear accidents, new scientific reports, and
other noteworthy events may influence this gender divide. Further, examining the robustness of this gender divide across polity
levels will help us better understand the likelihood that female
elected officials at different levels of government are significantly
less supportive of policies to promote nuclear power than their
male counterparts. Finally, with this focus, we address recent calls
for increased social science attention to gender in energy research
[46]. Our study helps us better understand the way that perceptions
of energy technology (specifically, nuclear power) is gendered over
time and polity levels ([47], p. 15).
We use a long time-series of nationally representative survey
data from the Swedish public and novel datasets of Swedish elected
30
A. Sundström, A.M. McCright / Energy Research & Social Science 11 (2016) 29–39
officials that recently have been publicly released to answer two
research questions. First, is the gender difference on nuclear power
attitudes in the general public stable over time? Second, is the gender difference on nuclear power attitudes robust across levels of
the polity? Thus, our focus here is on the consistency of the gender
divide over time and across polity levels. Data limitations prevent
us from comparing theoretical arguments explaining the existence
of the gender divide. Nevertheless, our results are largely consistent with the most prominent theoretical argument that scholars
use to explain gender differences in environmental concern and
risk perceptions, which we briefly describe in Section 2.4.
Sweden offers a theoretically interesting context, since it has a
notably high level of gender equality [41,51] and Europe’s highest
percentage of female elected officials at the local level and one of
the highest percentages of female legislators at the national level
[49,25]. Thus, finding robust gender differences in this context suggests that they are likely to present in countries with much less
gender equality. As such, the results of this study serve as a novel
point of comparison for future research on gendered perceptions
of energy technology in Europe and beyond.
2. Background
2.1. Earlier empirical studies of public opinion on nuclear power
Few studies of public attitudes about nuclear power prior to the
early 1970s analyzed possible gender differences. In their review
of empirical studies from the early 1970s to the mid-1990s that
do examine the relationship between gender and attitudes toward
nuclear power, Davidson and Freudenburg [9] find a striking pattern. In every one of the 17 studies about attitudes toward a certain
nuclear power facility and 21 studies of attitudes toward nuclear
power in general, women report stronger concern about the risks
of nuclear power and/or waste than do men [9].
Research since the 1990s finds that adult women are less supportive of nuclear power than are adult men across a range of
countries, including the United States (e.g., [1,20,21,50]), the United
Kingdom (e.g., [6,7]), Switzerland (e.g., [28]), The Netherlands (e.g.,
[10]), Sweden (e.g., [42]), Turkey (e.g., [14]), Japan (e.g., [33]), China
(e.g., [57]), and 27 European Union countries (e.g., [26,36]). Other
studies also find this gender divide among college students (e.g.,
[12,30]) and high school students (e.g., [13,38]).
Further, this gender divide—where women report less support
for nuclear power than men—is robust across a range of attitudes
about nuclear power. Some studies find that women are less accepting than are men of nuclear power generation in general (e.g.,
[6,7,10,14,20,21,26,36,42,50]). Other studies report that women are
more strongly opposed than are men to the construction of new
nuclear power plants or the expanded operation of existing nuclear
power plants near their community (e.g., [1,28,57]). Further, several studies find that women perceive greater risks than do men
from nuclear power generation (e.g., [10,12,33,38]). Finally, a few
additional studies document that women are more concerned than
are men with the handling, transportation, and disposal of nuclear
waste (e.g., [13,30,42]).
Most of these empirical studies rely upon a general theoretical
explanation emphasizing that gender socialization, and the social
roles that men and women are socialized into, results in women
being more concerned than are men with the safety and concern
of others. This “health and safety concerns” argument (e.g., [9]) has
been the most common argument used to explain the gender divide
on nuclear power attitudes in the general public (e.g., [16,45]) and
among elected officials (e.g., [40]). While not the only argument
explaining gender differences in environmental concern and risk
perceptions, the health and safety concerns argument neverthe-
less has been employed widely and enjoys considerable empirical
support (see, e.g., [3,32,44,55]).
Briefly, the processes of gender socialization in most Western
societies tend to confer to boys a masculine identity that emphasizes competitiveness, detachment, and an affinity for the control
and mastery over other people and things, while they tend to
confer to girls a feminine identity that emphasizes cooperation,
attachment, and an affinity for expressing empathy and caring for
others (e.g., [29]). This feminine “ethic of care” [17] translates into
heightened concern for the health, safety, and welfare of others and
greater sensitivity to risk vulnerabilities. As Bord and O’Connor [3]
point out, women perceive specific technologies as considerably
riskier than do men when these technologies become characterized as having significant health and safety risks. Such is the case
for nuclear power in recent decades. Thus, as applied to the issue
of nuclear power, the health and safety concerns argument holds
that women’s heightened concern for the health and safety risks
of nuclear power lead them to be less supportive of nuclear power
than are men.
While the recently released datasets we use allow us to investigate the robustness of the gender divide on nuclear power attitudes
over time and across polity levels, they—like the datasets used
in most earlier studies—nevertheless do not contain high-quality
measures of perceived health and safety risks from nuclear power.
Thus, we are not able to definitively test this part of the argument; rather, we can only assess the extent to which our results
are consistent with this explanation.
2.2. Views of elected officials
We should be cautious in assuming that the patterns of public
opinion on any social issues are also found among elected officials.
Elected officials are seldom representative of the general public,
especially in terms of demographic and socio-economic characteristics (e.g., [4]). Indeed, those studies that examine how citizens and
elected officials view the same issues find incongruency between
the policy preferences of citizens and those of their political representatives (e.g., [8,53]). This is the case with the issue of nuclear
power in Sweden, where members of the Swedish national parliament (the Riksdag) express greater opposition to nuclear power
than is found in the Swedish public [23].
Because of the paucity of empirical studies [5,27,40,43], existing research provides no clear answer on the extent to which the
gender divide on nuclear power in the general public is evident
among elected officials. In their study of a 1988 national survey
of state legislators in the United States, Carroll et al. [5] find that
female state legislators are more likely to oppose the construction
of nuclear power plants than their male counterparts. Analyzing
data from a sample of members and deputy members of Environment and Health Protection Boards of 27 Swedish municipalities,
Sjöberg and Drottz-Sjöberg [43] find that females report greater
perceived risk of nuclear waste than their male counterparts. While
Reingold [40] finds that females in the California state legislature
more strongly favor further regulation of nuclear power than their
male counterparts, there is no such pattern among state legislators
in Arizona. Finally, in his study of Norwegian Parliament, Jensen
[27] reports that the gender difference in concern about the risks
of nuclear power disappears when controlling for legislators’ party
affiliation.
2.3. The Swedish context
While nuclear power supplies a notable share of the energy
production in Sweden, it nevertheless has been the focus of much
public debate. The future of nuclear power was intensely debated in
Sweden in the 1970s; the issue became highly politicized, though
A. Sundström, A.M. McCright / Energy Research & Social Science 11 (2016) 29–39
in a manner that cut across the traditional Left-Right dimension in
Swedish politics. The partial nuclear meltdown at the Three Mile
Island facility in the United States at the end of March 1979 led to
mass anti-nuclear protests in Sweden and eventually a nationwide
referendum on the future of nuclear power in 1980. The Swedish
electorate voted for a vaguely defined proposal for a “future phaseout” of nuclear power in Sweden.
The nuclear disaster at the Chernobyl facility in what is now
Ukraine at the end of April 1986 increased both opposition to
nuclear power in Sweden and support for the emerging Green Party
in the late 1980s. Other than the Center party, the Green party
consistently has been the strongest opponent of nuclear power in
Sweden. The coalition of political parties opposing nuclear power
disbanded in the late 1990s, as support for nuclear power gradually
had been increasing among the Right-of-center parties.
While Reactor 1 at the Swedish nuclear power plant in Barsebäck was shut down at the end of November 1999 as part of the
planned phase-out (and Reactor 2 shut down at the end of May
2005), changes in the political landscape in the 2000s—notably,
the entrance of the Right-of-center government in 2006—led to a
halt in the phase-out plans. Currently, a nuclear power phase-out
enjoys mixed support within the general public (see the solid line
in Panel B of Fig. 1). While the meltdown of three reactors at the
Fukushima nuclear power plant in Japan in mid-March 2011 led
to a short-term increase in opposition to nuclear power, this effect
soon disappeared (for a more thorough account of these events, see
[23]).
As briefly stated in the introduction, Sweden offers a theoretically interesting context to examine the robustness of a gender
divide on nuclear power attitudes within the general public over
time and across polity levels in recent years. Because of its exceptionally high level of gender equality, Sweden is part of what often
is called the “Nordic exception” in discussions of politics and gender (e.g., [41]). For instance, for the past four decades Sweden has
been at the top of global rankings for women’s share of labor force
participation and men’s share of unpaid care and domestic work
[2,35]. Indeed, the United Nations Development Program ([51] p. 4)
recently ranked Sweden as the second-most gender-equal nation
in the world. Moreover, Sweden has Europe’s highest percentage
of female elected officials at the local level and one of the highest
percentages of female legislators at the national level [49,25].
Given that gender roles in Sweden are likely less disparate than
in societies where men and women are less equal, Sweden serves as
a “strong test” for finding a robust gender divide on nuclear power
attitudes over time and across polity levels. As mentioned, a few
studies do document such a gender divide on nuclear power in the
Swedish public [12,13,42]. However, they rely upon either small,
non-representative samples [13] or single cross-sections [42].
3. Methods
3.1. Datasets
We utilize a long time-series of nationally representative survey
data from the Swedish public to answer our first research question about the extent to which the gender divide on nuclear power
attitudes in the general public is stable over time. The SOM Institute (Society, Opinion, Media), an academic organization affiliated
with the University of Gothenburg, has administered a nationwide
mail questionnaire annually since 1986. The SOM surveys contain
numerous questions on a wide range of topics related to media
and politics. We focus on repeated cross-sections between 1986
and 2011. We do not include data from 1997, as that year’s survey did not include our dependent variable measure. The yearly
samples are representative of the Swedish population and enjoy
31
respectable response rates. The response rates for the 1986–2011
surveys range from a low of 58% in 2008 to a high of 71% in 1992,
and the samples sizes of the surveys during this time period range
from a low of 1,573 in 1991 to a high of 5,007 in 2010 ([52], p. 583).
We utilize three cross-sectional datasets to answer our second
research question about the extent to which the gender divide on
nuclear power attitudes is robust across levels of the polity. The
analysis of citizens’ attitudes is based on the 2009 SOM survey from
the pooled time-series dataset described in the previous paragraph.
The 2009 survey had a response rate of 59% and an initial sample
size of 3529 citizens. The survey item we use to measure nuclear
power attitudes was administered to a randomly split sub-sample
of one-third of the total respondents. The second dataset is based
on a pioneering 2008 survey that sampled all elected members of
the 290 municipal-level and 21 county-level elected assemblies
in Sweden. This internet-based survey received 9,890 responses,
equaling a 70% response rate (see [19]). The third dataset is based
on a 2010 mail questionnaire sent to all members of the Swedish
national parliament (the Riksdag). In total, 327 responses were
received, for a response rate of 89% (see [15]).
It should be noted that elections for the municipal councils,
county councils, and national parliament are held on the same
day every fourth year. The Swedish parliament has one chamber. Municipalities are governed by parliamentarism, where ruling
coalitions are the executive power in each local assembly. Counties are the intermediary governing level in the Swedish political
system (for a more detailed description, see [18]). Eight parties
are represented in the national parliament. At the municipal and
county levels the numbers of parties are higher in some areas,
though the additional parties have only a marginal impact in the
country as a whole.
3.2. Variables
Table 1 provides the name, coding, mean, and standard deviation for each of the variables we use in our analyses. We first
describe the variables we use in our analyses of the stability of the
gender divide in the general public over time. We then describe
how the measurement of these variables differs slightly across the
datasets of elected officials that we use for our second research
question. Table A1 in the Appendix also describes the slight variation in the measurement of these variables across the four levels of
the Swedish polity.
The dependent variable in our study is opposition to nuclear
power. We operationalized this variable using a single-item indicator. SOM survey respondents were asked: “Below are a number of
proposals that have occurred in the political debate. What is your
opinion on each of them?” From 1986 to 1996, the item was “to
keep nuclear power also after 2010,” with five response categories
ranging from “a very good proposal” to “a very bad proposal.” From
1998 to 2011, the item was “the long-term phase-out of nuclear
power,” with the same five response categories as before. Thus, in
the earlier time period, opposition to nuclear power is operationalized as the belief that a proposal to keep nuclear power also after
2010 is “very good” (1) to “very bad” (5); in the later time period,
opposition to nuclear power is operationalized as the belief that a
proposal for the long-term phase out of nuclear power is “very bad”
(1) to “very good” (5). The surveys of elected officials in the municipal and county assemblies and national parliament included the
item from the later period of the citizen surveys. Thus, in the late
2000s, citizens and elected officials were asked the identical survey item about the long-term phase-out of nuclear power. We do
acknowledge the limitation of relying upon a single-item indicator.
However, given our focus here (examining the consistency of the
effect of gender over time and across polity levels) and the paucity
32
A. Sundström, A.M. McCright / Energy Research & Social Science 11 (2016) 29–39
Fig. 1. Gender divide in opposition to nuclear power in the Swedish general public, 1986–2011; Panel A. Percentage of all respondents (solid line), Men, (dotted line), and
Women (dashed line) who oppose a proposal “to keep nuclear power also after 2010”; Panel B. Percentage of all respondents (solid line), Men (dotted line), and Women
(dashed line) Who support a Proposal for “the long-term phase out of nuclear power.”
of multi-item indicators in existing datasets to accomplish this, we
nevertheless feel this measure is adequate for our purpose.
We measure gender demographically as “male” (0) or “female”
(1). In our analyses we control for a group of social, demographic,
and political variables used in previous studies. To accommodate
both how income was initially recorded on the SOM survey and
increasing income in Sweden over the time period, we measure
income in relative categories: “low” (1), “middle” (2), and “high”
(3). For each year, these three categories represent the three lowest
income categories, the three in the middle, and the three highest
income categories, respectively. Education ranges from “primary
school” (1) to “college or university” (3); age ranges from “18–29”
(1) to “65 or older” (4). On this note it can be mentioned that
Pampel and Hunter [37] show how to disentangle age, period,
and cohort effects when examining environmental concern in the
United States. Because of how age was originally measured on the
SOM surveys and our emphasis on gender, we have chosen not to
investigate age, period, and cohort effects.
We measure religiosity as frequency of church attendance, ranging from “never” (1) to “at least once a month” (4). We do not have
1986 or 1987 data for this variable, because church attendance was
not asked in those two years. Child at home is a dummy variable
distinguishing those respondents who report at least one child at
home (1) from those reporting none at home (0). We do not have
data on this variable prior to 1993, as it was not included in earlier
surveys.
Political ideology is a scale from “far left” (1) to “far right”
(5). Sweden has a number of political parties, so we created
nine dummy variables to measure political party identification.
Voters, as well as experts, generally place the major political
parties in Sweden on the left-right continuum in the following order: Left Party, Social Democrat Party, Green Party, Center
Party, Liberal Party, Christian Democrat Party, Moderate Party,
and Swedish Democrat Party (see, e.g., [48]). Using identification
with the Center Party (Centerpartiet) as the reference category, we
utilize eight dummy variables in our analyses to measure identification with the following parties (ordered on the Left-Right
continuum): Left Party (Vänsterpartiet), Social Democrat Party
(Socialdemokraterna), Green Party (Miljöpartiet de gröna), Liberal
Party (Folkpartiet liberalerna), Christian Democrat Party (Krist-
A. Sundström, A.M. McCright / Energy Research & Social Science 11 (2016) 29–39
33
Table 1
Coding, mean, and standard deviations for the variables used in the study.
Variable
General public (1986–2011 N = 60,104)
Opposition to nuclear power (1986–1996)
Opposition to nuclear power (1998–2011)
Female
Income
Education
Age
Religiosity (attendance)
Child at home
Political ideology
Party identification (8 dummy variables)
Devoted party supporter
Municipal-level officials (2008 N = 7,334)
Opposition to nuclear power
Female
Income
Education
Age
Religiosity (attendance)
Child at home
Political ideology
Party identification (8 dummy variables)
County-level officials (2008 N = 1,060)
Opposition to nuclear power
Female
Income
Education
Age
Religiosity (attendance)
Child at home
Political ideology
Party identification (8 dummy variables)
National Parliamentarians (2010 N = 298)
Opposition to nuclear power
Female
Education
Age
Religiosity (pray)
Child at home
Political ideology
Party identification (7 dummy variables)
Coding
Mean
SD
“to keep nuclear power also after 2010” is 1 (very good proposal) to 5 (very bad proposal)
“the long-term phase-out of nuclear power” is 1 (very bad proposal) to 5 (very good proposal)
0 (male) to 1 (female)
1 (low) to 3 (high)
1 (primary school) to 3 (college or university)
1 (18–29) to 4 (65 or older)
1 (never) to 4 (at least once a month)
0 (have none) to 1 (have at least one)
1 (far left) to 5 (far right)
0 (does not support) to 1 (supports)
1 (no) to 3 (yes, very)
2.95
3.09
0.51
4.97
2.18
2.59
1.85
0.37
3.03
1.47
1.34
0.5
2.24
0.48
1.00
1.22
0.48
1.17
1.73
0.71
“the long-term phase-out of nuclear power” is 1 (very bad proposal) to 5 (very good proposal)
0 (male) to 1 (female)
1 (<100,000 SEK) to 9 (>800,000 SEK)
1 (primary school) to 3 (college or university)
1 (18–29) to 4 (65 or older)
1 (never) to 4 (frequent visits)
0 (have none) to 1 (have at least one)
1 (far left) to 10 (far right)
0 (not affiliated) to 1 (affiliated)
3.04
0.42
5.42
2.37
2.82
2.28
0.36
4.76
1.54
0.49
1.94
0.71
0.82
1.02
0.48
3.04
“the long-term phase-out of nuclear power” is 1 (very bad proposal) to 5 (very good proposal)
0 (male) to 1 (female)
1 (<100 000 SEK) to 9 (>800,000 SEK)
1 (primary school) to 3 (college or university)
1 (18–29) to 4 (65 or older)
1 (never) to 4 (frequent visits)
0 (have none) to 1 (have at least one)
1 (far left) to 10 (far right)
0 (not affiliated) to 1 (affiliated)
3.10
0.47
5.97
2.52
2.86
2.33
0.32
4.83
1.56
0.5
1.91
0.66
0.77
1.02
0.47
2.96
“the long-term phase-out of nuclear power” is 1 (very bad proposal) to 5 (very good proposal)
0 (male) to 1 (female)
1 (primary school) to 3 (college or university)
1 (18–29) to 4 (65 or older)
1 (never) to 4 (at least once a month)
0 (have none) to 1 (have at least one)
1 (far left) to 10 (far right)
0 (not affiliated) to 1 (affiliated)
3.09
0.44
2.63
2.29
2.17
0.45
5.46
1.62
0.5
0.56
0.79
1.01
0.50
2.18
Table 2
X-Standardized odds ratios from ordered logistic regression models for opposition to nuclear power in the Swedish general public, 1986–1996.
Predictors
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
Female
Income
Education
Age
Religiosity
Child at home
Political ideology
Left Party
Social Democrat Party
Green Party
Liberal Party
Christian Democrat Party
Moderate Party
Other party
Devoted party supporter
N
Pseudo R-squared
1.73*
0.87*
1.06
0.97
1.62*
0.93
1.63*
0.82*
1.07
0.73*
1.07
1.63*
0.87*
1.12
0.69*
1.21
1.53*
0.94
1.1
0.83*
1.12
1.56*
0.80*
1.09
0.82*
1.08
1.48*
0.90*
1.04
0.85*
1.11*
0.86
0.93
0.49*
0.94
0.67*
0.72
0.40*
0.82
0.95
1210
0.07
0.74*
0.86
0.48*
0.87
0.71*
1.06
0.41*
0.87
0.91
1328
0.07
0.68*
0.98
0.51*
1.13
0.75*
1
0.49*
0.82*
1.04
1262
0.1
0.65*
0.91
0.72*
1.07
0.82*
0.94
0.57*
0.94
1.04
1224
0.08
0.61*
0.77*
0.56*
0.99
0.73*
0.80*
0.49*
0.77*
0.94
1152
0.08
0.67*
0.78*
0.46*
1.04
0.72*
0.85*
0.42*
0.76*
0.89
1223
0.09
0.69*
0.73*
0.49*
1.31*
0.65*
0.78*
0.45
0.74*
0.95
1532
0.07
1.46*
0.78*
1
0.86*
1.06
1.04
0.63*
0.81*
0.54*
1.14
0.74*
0.83*
0.50*
0.80*
0.91*
1435
0.07
1.50*
0.80*
1.03
0.76*
1.21*
1.22*
0.68*
0.86
0.58*
0.96
0.69*
0.76*
0.47*
0.79*
0.98
1243
0.09
1.56*
0.88*
1.07
0.72*
1.14*
1.04
0.69*
0.83*
0.58*
1.02
0.68*
0.83*
0.55*
1.06
0.99
1145
0.08
1.42*
0.85*
1.03
0.73*
1.11
1.07
0.71*
0.82*
0.64*
0.95
0.75*
0.81*
0.53*
0.89
0.90*
1291
0.07
0.83*
Notes: Opposition to nuclear power is operationalized as whether respondents believed that “to keep nuclear power also after 2010” is a “very good proposal” (1) to a
“very bad proposal” (5). The Swedish Center Party, “Centerpartiet,” is the reference category for the party dummy variables. No data on religiosity was available for these
respondents in 1986 and 1987. Prior to 1993, respondents were not asked if they had a child living at home.
*
p < 0.05.
demokraterna), Moderate Party (Moderaterna), Swedish Democrat
Party (Sverigedemokraterna), and other, smaller political parties.
Identification with the Swedish Democrat Party was only included
on surveys in 2007 onwards. Finally, we measure how devoted each
respondent in the general public is as a party supporter, from “no”
(1) to “yes, very” (3).
We now detail how the measurement of these variables in the
2008 and 2010 surveys of elected officials varies from their mea-
34
A. Sundström, A.M. McCright / Energy Research & Social Science 11 (2016) 29–39
surement in the 2009 SOM survey of citizens. Our measurement of
opposition to nuclear power, gender, education, and age are identical across these datasets. While income was not included in the
national parliament survey, it was measured with the following
categories on the surveys of municipal and county elected officials: “less than 100,000 Swedish Kronor (SEK)” (1) to “greater than
800,000 SEK” (9). Among municipal and county elected officials,
religiosity was measured by church attendance, but with response
categories that ranged from “never” (1) to “frequent visits” (4). For
members of the national parliament, religiosity was measured by
their frequency of prayer, ranging from “never” (1) to “at least once
a month” (4). Among elected officials, political ideology is measured
on an expanded scale from “far left” (1) to “far right” (10). For the
political party dummy variables, respondents in the general public self-reported whether they were a supporter of a party. Elected
officials self-reported whether they were affiliated with a political
party. Finally, for national parliamentarians we needed no dummy
variable to signify affiliation with any smaller parties.
3.3. Analytical techniques
To examine the relationship between gender and opposition to
nuclear power over time and across polity levels, we ran a series of
multivariate ordered logistic regression models. For each year and
for each polity level, we ran a model with only gender as a predictor;
a model with gender and the group of socio-demographic variables
as predictors; and a model with gender, the socio-demographic
variables, and the political variables as predictors. The complete
collection of models is available from the authors. Here, we present
the results for the full model for each year or polity level.
Since our dependent variable is ordinal, we use ordered logistic regression techniques. We checked for multicollinearity using
the variance inflation factor (VIF) statistic and found no problem.
Across all models, the VIF values were generally close to 1 and in no
model were they larger than 6.4; the mean VIF value in the models varied from 1.79 in 1990 to 2.4 in 2009. We also checked to
see if influential observations influenced our results and found no
problem in this regard. More specifically, following recommendations for diagnostics in ordered logistic regression ([24], p. 305;
[31], p. 200–201), we performed these tests on a version of our
dependent variable that is coded binary (where the two response
categories representing opposition to nuclear power are denoted
as 1 and the other three categories are denoted as 0). To determine
whether our models had a problem with high leverage points, we
created plots to identify observations that were potential outliers
and, in such occasions, then reran the models without these observations (see [39]). In none of our models did the exclusion of such
observations affect the results in a substantive matter or by rendering a statistically significant effect non-significant or vice versa. To
facilitate comparison across years and polity levels and judge the
relative importance of the predictors, we present X-standardized
odds ratios in our tables [31]. An X-standardized odds ratio represents the ratio changes associated with every standard deviation
change in an independent variable. As such, it allows for the direct
comparison of the effects of all predictors in the model.
4. Results and discussion
4.1. The stability of the gender divide on nuclear power attitudes
over time
Given the difference in question wording between the
1986–1996 and 1998–2011 time periods, we treat these two time
periods separately in our analyses. Fig. 1 displays trends in opposition to nuclear power in the Swedish public since the mid-1980s.
Panel A shows the percentages of all respondents (solid line), men
(dotted line), and women (dashed line) who believe a proposal “to
keep nuclear power also after 2010” is “bad” or “very bad.” Panel
B shows similar percentages for those believe a proposal for “the
long-term phase out nuclear power” is “good” or “very good.”
A quick glance at both panels shows that the gender divide on
opposition to nuclear power has been quite stable across the 25
years of this study, even as aggregate public opinion has fluctuated. In each year, a greater percentage of women than men report
opposition to nuclear power; each yearly difference is statistically
significant. The largest difference appears in 1986, when 72.07% of
women but only 46.48% of men oppose nuclear power (a percentage difference of 25.59%). The smallest gender difference appears
in 1999, when 52.07% of women and 42.60% of men oppose nuclear
power (a percentage difference of 9.47%). The average, yearly percentage difference for the earlier time period is, 18.58%, while that
for the later time period is 16.36%.
To see whether this pattern holds up while controlling for other
relevant social, demographic, and political characteristics, we turn
to the results of our multivariate ordered logistic regression models. Table 2 reports the X-standardized odds ratios for the predictors
of opposition to nuclear power in the Swedish public for each
of the years of the time period 1986–1996. Table 3 reports the
same for each of the years of the time period 1998–2011. Again,
X-standardized odds ratios allow for the direct comparison of the
effects of all predictors in the model. Statistically significant effects
are the change in the odds ratios on the scale of opposing nuclear
power from a one standard deviation increase in a predictor variable.
Most importantly, the pattern illustrated in Fig. 1—whereby
women report greater opposition to nuclear power than do men
in each of the 25 years of the study—in confirmed. Even when controlling for the social, demographic, and political predictors, gender
has a statistically significant effect in the theoretically expected
direction. Indeed, while gender, political ideology, and support for
a few political parties are the most powerful predictors across the
time period, gender is the only one to have a statistically significant
effect in each year. Consistent with the health and safety concerns
argument, women report greater opposition than do men to nuclear
power—a technology characterized as having significant health and
safety risks. Notably, this statistically significant effect of gender
remains even when controlling for worry about climate change, an
item that has only been included in the SOM surveys since 2001.
Table A2 in the Appendix reports the results of ordered logistic
regression models for 2001–2011 that control for worry about climate change, coded “not at all worried” (1) to “very worried” (4).
Interestingly, worry about climate change is positively related to
opposition to nuclear power, likely because both are tapping into
broader environmental concern.
The effects of the social, demographic, and political variables are
also notable. Besides gender, political ideology also is a relatively
stable predictor. In all but one year (1986), Left-leaning individuals
report greater opposition to nuclear power than do their Rightleaning counterparts, consistent with some of the existing research
reviewed earlier (e.g., [22]). The magnitude, direction, and consistency of the effects of the dummy variables measuring party
identification are largely expected. Compared to supporters of the
Center Party, supporters of Right-leaning parties (i.e., the Moderate Party and the Liberal Party) report less opposition in most of
the years of the study. Supporters of the Green Party report greater
opposition to nuclear power than do supporters of the Center Party,
but only in a handful of years.
Consistent with some existing research (e.g., [20,36]), younger
adults report greater opposition to nuclear power than do older
adults in all but five years. Only one of the two indicators of
socio-economic status has even a modest influence on nuclear
A. Sundström, A.M. McCright / Energy Research & Social Science 11 (2016) 29–39
35
Table 3
X-standardized odds ratios from ordered logistic regression models for opposition to nuclear power in the Swedish general public, 1998–2011.
Predictors
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
Female
Income
Education
Age
Religiosity
Child at home
Political ideology
Left party
Social democrat party
Green party
Liberal party
Christian democrat party
Moderate party
Swedish democrat party
Other party
Devoted party supporter
N
Pseudo R-squared
1.34*
0.95
1.12*
0.92
1.08
1.05
0.66*
0.84*
0.66*
1.18*
0.84*
0.87*
0.60*
1.25*
0.91
1.04
0.86*
1.17*
1.04
0.73*
0.81*
0.70*
1.15
0.80*
0.83*
0.56*
1.34*
0.96
1.07
0.91
1.24*
1.07
0.59*
0.68*
0.46*
0.97
0.67*
0.70*
0.45*
1.36*
0.86*
1.03
0.83*
1.11*
1.04
0.66*
0.94
0.73*
1.24*
0.89
0.92
0.67*
1.43*
0.90*
1.05
0.86*
1.06
1.11*
0.67*
0.90*
0.64*
1.14*
0.77*
0.86*
0.60*
1.53*
0.88*
1.02
0.81*
1.04
1.03
0.67*
0.86*
0.71*
1.09
0.83*
0.86*
0.65*
1.36*
0.84*
1.06
0.78*
1.12*
1.11
0.64*
0.97
0.85
1.18
0.87*
1.03
0.70*
1.53*
0.88*
1.03
0.84*
1.14*
1.06
0.65*
0.88
0.78*
1.12
0.82*
0.83*
0.68*
1.54*
0.84*
1.12*
0.91
1
0.94
0.69*
0.98
0.67*
1.18*
0.87
0.88*
0.59*
0.84*
1
1468
0.05
0.76*
1.13*
1245
0.05
0.88
0.95
1325
0.07
0.80*
0.99
2525
0.06
0.85*
1.03
2789
0.06
0.84*
0.99
1414
0.06
0.85*
1.02
1373
0.07
0.84*
0.96
1295
0.07
0.79*
1.1
1272
0.08
1.46*
0.87*
1.03
0.82*
1.14
1.17*
0.65*
0.92
0.70*
1.14
0.79*
0.79*
0.67*
0.87*
0.84*
0.96
1288
0.07
1.53*
0.81*
1.03
0.81*
1
1.28*
0.74*
0.89
0.56*
0.96
0.63*
0.84*
0.49*
0.82*
0.79*
0.93
1261
0.08
1.68*
0.9
0.95
0.83*
1.05
1.13*
0.67*
1.04
0.67*
1.21*
0.78*
0.90
0.60*
0.84*
0.9
0.91
1229
0.10
1.52*
0.97
0.98
0.86*
1.14*
1.01
0.62*
0.97
0.8
1.08
0.91
0.92
0.75*
0.91*
0.88
0.92
1332
0.07
1.37*
0.93
1.04
0.98
1.12*
100
0.58*
0.88
0.70*
1.13
0.78*
0.94
0.74*
0.88*
1.00
1.12*
1233
0.07
Notes: Opposition to nuclear power is operationalized as whether respondents believed that “Sweden should in the long-term phase out nuclear power” is a “very bad
proposal” (1) to a “very good proposal” (5). The Swedish Center Party, “Centerpartiet,” is the reference category for the party dummy variables. Identification with the
Swedish Democrat Party is only measured from 2007 onwards.
*
p < 0.05.
Table 4
X-standardized odds ratios from ordered logistic regression models for opposition to nuclear power across four levels of the polity in sweden.
Predictors
Female
Income
Education
Age
Religiosity
Child at home
Political ideology (left to right)
Left party
Social democrat party
Green party
Liberal party
Christian democrat party
Moderate party
Swedish democrat party
Other party
Devoted party supporter
N
Pseudo R-squared
Generalpublic (2009)
*
1.68
0.9
0.95
0.83*
1.05
1.13*
0.67*
1.04
0.67*
1.21*
0.78*
0.9
0.60*
0.84*
0.90
0.91
1229
0.10
Municipal-levelofficials (2008)
*
County-level officials (2008)
*
National-level officials (2010)
1.31
0.91*
1.11*
0.87*
1.01
1.07*
0.54*
0.90*
0.41*
1.34*
0.49*
0.68*
0.29*
0.72*
0.65
1.46
0.9
1.02
0.83*
1.11
1.07
0.50*
0.86
0.28*
1.24
0.35*
0.52*
0.17*
0.68*
0.57
1.48*
7334
0.17
1060
0.23
298
0.38
1.24
1.1
0.9
0.9
0.59*
1.6
0.82
1.69*
0.23*
0.54*
0.13*
0.40*
Notes: The Swedish center party, “Centerpartiet,” is the reference category for the party dummy variables.
power attitudes. While the effect of education is nearly always
non-significant, income has a negative effect in approximately twothirds of the years. That is, respondents in lower relative income
categories report greater opposition to nuclear power than do
those in higher relative income categories. In ten years of the
study, the frequency of religious service attendance is positively
related to opposition. Having a child at home is positively related
to greater opposition to nuclear power in only about one-third of
the years for which data is available. We also ran models including an interaction term between gender and having a child at
home for the 18 years between 1993 and 2011 for which the
“child at home” item was asked. In only one year (2008) did this
interaction term have a statistically significant effect, whereby having a child at home further increased women’s opposition to nuclear
power (these results are available upon request).
4.2. The robustness of the gender divide on nuclear power
attitudes across polity levels
We now turn to the results for our second research question
about the extent to which the gender divide on opposition to
nuclear power is robust across polity levels. Different surveys of
citizens and elected officials at municipal, county, and national
levels in the late 2000s contained the identical survey item asking respondents their assessment of a proposal for “the long-term
phase-out of nuclear power.” Within the general public in 2009 and
among elected officials at the municipal and county levels in 2008,
the difference in the percentage of women compared to men who
believe such a proposal is “good” or “very good” is statistically significant. Briefly, 47.87% of women but only 28.61% of men in the
Swedish public oppose nuclear power (a percentage difference of
19.26%); 51.22% of women but only 38.27% of men in municipallevel elected office oppose nuclear power (a percentage difference
of 12.95%); and 54.18% of women but only 38.62% of men in countylevel elected office oppose nuclear power (a percentage difference
of 15.56%). Among national parliamentarians in 2010, the gender
divide is smaller and not statistically significant; 54.89% of women
but only 47.93% of men in the Swedish Riksdag oppose nuclear
power (a percentage difference of 6.96%).
To see how the pattern in these bivariate relationships fares
while controlling for other relevant social, demographic, and
36
A. Sundström, A.M. McCright / Energy Research & Social Science 11 (2016) 29–39
Table A1
Description of the variables used in this study and their availability at four levels of the Swedish polity.
Variable
Description
General public
Municipal level
County level
National parliament
Opposition to nuclear power
“Below are a number of proposals that have
occurred in the political debate. What is your
opinion on each of them? . . . Sweden should in
the long-term phase out nuclear power” (very
bad proposal=1 to very good proposal=5)
sex of respondent (male = 0, female = 1)
annual household income in Swedish currency
(<100,000 SEK = 1 to >800,000 SEK = 9)
respondent’s education level (primary
school = 1, high school = 2, college or
university = 3)
age of respondent (18–29 = 1, 30–49 = 2,
50–64 = 3, 65 + = 4)
how many times the respondent has visited a
religious service/practice in the last year
(never = 1 to at least once a month = 4)
respondent has a child living in his/her
household (no = 0, yes = 1)
respondent’s identification on a left-right
political scale (far left = 1 to far right = 10)
respondent is affiliated with Vänsterpartiet
(no = 0, yes = 1)
respondent is affiliated with
Socialdemokraterna (no = 0, yes=1)
respondent is affiliated with Miljöpartiet
(no=0, yes=1)
respondent is affiliated with Folkpartiet (no = 0,
yes = 1)
respondent is affiliated with
Kristdemokraterna (no = 0, yes = 1)
respondent is affiliated with Moderaterna
(no = 0, yes = 1)
respondent is affiliated with
Sverigedemokraterna (no = 0, yes = 1)
respondent is affiliated with one of the
remaining smaller parties (no = 0, yes = 1)
respondent is a devoted supporter of a party
(no = 1; yes, slightly = 2; yes, very = 3)
Xa
X
X
X
X
Xb
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Xc
Xc
Xd
Xe
X
X
X
Xf
X
X
X
Xg
X
X
X
g
X
X
X
X
Xg
X
X
X
Xg
X
X
X
g
X
X
X
X
Xg
X
X
X
Xg
X
X
X
Xg
X
X
Female
Income
Education
Age
Religiosity
Child at home
Political ideology
Left party
Social democrat party
Green party
Liberal party
Christian democrat party
Moderate party
Swedish democrat party
Other party
Devoted party supporter
X
a
In the general public samples from 1986 to 1996, the survey item asked if the proposal “to keep nuclear power also after 2010” was a “very good proposal” (1) to a “a
very bad proposal” (5).
b
In the general public sample over time, income is measured in three relative categories (low = 1 to high = 3)
c
Response categories are: never; more seldom; a couple of times; frequent visits.
d
How many times the respondent has prayed in the last year: never = 1; more seldom = 2; several times a year = 3; at least once a month = 4.
e
In the general public sample, child at home is available from the year 1993 and onwards.
f
In the general public sample, ideology is measured on a 5-point scale (far left = 1 to far right = 5).
g
In the general public sample, respondents are asked if they are a supporter of (and not affiliated with) a party.
Table A2
X-standardized odds ratios from ordered logistic regression models for opposition to nuclear power in the Swedish general public, 2001–2011.
Predictors
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
Female
Worry about climate change
Income
Education
Age
Religiosity
Child at home
Political ideology
Left party
Social democrat party
Green party
Liberal party
Christian democrat party
Moderate party
Swedish democrat party
Other party
Devoted party supporter
N
Pseudo R-squared
1.26*
1.20*
0.84*
1.04
0.78*
1.12*
1.08
0.66*
0.87
0.64*
1.12
0.87
0.85*
0.62*
1.39*
1.35*
0.94
1.08
0.88*
1.02
1.19*
0.76*
0.91
0.62*
1.09
0.75*
0.81*
0.60*
1.48*
1.32*
0.90
1.10
0.81*
1.04
1.03
0.67*
0.85*
0.70*
1.06
0.82*
0.85*
0.65*
1.47*
1.20*
0.86*
1.02
0.89*
1.11*
0.67
0.92
0.58*
1.08
0.74*
0.82*
0.50*
1.48*
1.40*
0.89*
1.02
0.84*
1.090
1.05
0.66
0.88
0.79*
1.11
0.83*
0.85*
0.70*
1.48*
1.29*
0.85*
1.10
0.91
0.99
0.93
0.68*
0.94
0.65*
1.14
0.86*
0.89
0.60*
0.76*
1.02
1204
0.07
0.82*
1.10
1370
0.06
0.85*
0.99
1411
0.06
0.79*
1.09
1332
0.08
0.84*
0.96
1291
0.08
0.80*
1.08
1261
0.08
1.40*
1.49*
0.85*
1.02
0.82*
1.11
1.16*
0.66*
0.91
0.71*
1.11
0.79*
0.81*
0.68*
0.88*
0.86
0.94
1286
0.08
1.45*
1.45*
0.81*
1.04
0.79*
0.97
1.23*
0.77*
0.94
0.58*
0.96
0.64*
0.87
0.51*
0.85*
0.79*
0.93
1255
0.09
1.60*
1.57*
0.90
0.94
0.82*
1.03
1.09
0.68*
1.11
0.72*
1.22*
0.81*
0.95
0.63*
0.86*
0.94
0.9
1215
0.12
1.43*
1.51*
0.98
0.93
0.87*
1.12*
0.99
0.66*
0.98
0.83
1.05
0.91
0.94
0.77*
0.93*
0.91
0.93
1301
0.08
1.28*
1.49*
0.92
1.02
0.96
1.09
0.97
0.60*
0.91
0.71*
1.09
0.77*
0.94
0.76*
0.89*
0.99
1.11
1219
0.08
Notes: “Opposition to nuclear power” is operationalized as whether respondents believed that “Sweden should in the long-term phase out nuclear power” is a “very bad
proposal” (1) to a “very good proposal” (5). “Worry about climate change” measures whether respondents are “not at all worried” (1) to “very worried” (4) about climate
change. The Swedish Center Party, “Centerpartiet,” is the reference category for the party dummy variables. Identification with the Swedish democrat party is only measured
from 2007 onwards. In 2004, the item measuring religiosity was not asked in the same split-sample of the SOM survey that included the item measuring worry about climate
change.
A. Sundström, A.M. McCright / Energy Research & Social Science 11 (2016) 29–39
37
Table A3
X-standardized odds ratios from logistic regression models for support of nuclear power among women in the swedish general public, 1986–1996.
Predictors
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
Income
Education
Age
Religiosity
Child at home
Political ideology
Left party
Social democrat party
Green party
Liberal party
Christian democrat party
Moderate party
Other party
Devoted party supporter
N
Pseudo R-squared
1.37*
0.85
1.37*
1.10
1.42*
0.97
1.54*
0.94
1.26*
0.85
1.24*
0.90
1.17
0.84
1.30*
0.86
1.30*
0.81
1.15
0.95
1.11
0.91
1.37*
0.97
1.16
1.57
1.86
1.35
1.60*
1.86*
2.11*
1.28
1.04
1.99*
1.05
1.46*
1.16
2.09*
1.41
1.17
654
0.07
1.59*
1.04
1.78*
0.71
1.10*
1.00
1.72*
1.25
1.12
642
0.14
1.51*
1.28
1.33*
0.76
1.25*
1.14
1.41*
1.01
1.09
606
0.08
1.2
1.23
1.49*
1.05
1.23
1.13
2.10*
1.46*
1.21*
538
0.1
1.27
1.07
1.86*
1.12
1.34
1.26
2.12*
1.20*
1.04
568
0.08
1.27
1.18
1.72*
1.13
1.47*
1.17
2.01*
1.22*
1.08
663
0.07
1.26*
0.89
1.34*
0.95
0.99
1.34*
1.11
2.85*
1.14
1.87*
1.44
3.63*
1.49*
1.12
670
0.11
1.29*
1.08
1.51*
0.80*
0.79*
1.43*
1.04
1.34
0.9
1.2
1.18
1.72*
1.17
1.14
610
0.14
1.02
0.97
1.66*
0.88
0.95
1.13
1.44
1.16
0.88
1.39
1.12
1.67*
0.89
1.18
542
0.12
1.05
1.08
1.54*
0.88
0.87
1.33*
1.03
1.09
1.05
1.11
1.11
1.41*
1.01
1.03
659
0.08
1.27*
1.10
580
0.06
Notes: Support to nuclear power is a dichotomous variable operationalized as 1 being respondents that believed that “to keep nuclear power also after 2010” is a “very good
proposal” and a “pretty good proposal.” Conversely, 0 denotes respondents that believe that this is a “very bad proposal,” a “pretty bad proposal,” or “neither a bad nor a good
proposal.” The Swedish center party, “Centerpartiet,” is the reference category for the party dummy variables. No data on religiosity was available for these respondents in
1986 and 1987. Prior to 1993, respondents were not asked if they had a child living at home.
Table A4
X-standardized odds ratios from logistic regression models for support of nuclear power among women in the swedish general public, 1998–2011.
Predictors
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
Income
Education
Age
Religiosity
Child at home
Political ideology
Left party
Social democrat party
Green party
Liberal party
Christian democrat party
Moderate party
Swedish democrat party
Other party
Devoted party supporter
N
Pseudo R-squared
1.21
0.87
1.17
0.83*
0.92
1.80*
1.49*
1.82*
1.23
1.11
1.38*
1.80*
1.12
0.93
1.29*
0.85*
0.89
1.30*
1.35*
1.48
0.94
1.41*
1.39*
1.95*
1.11
0.9
1.29*
0.82
0.88
1.63*
1.38
1.46
0.9
1.4
1.30
1.50*
1.06
1.07
1.41*
0.91
0.94
1.76*
1.14
1.23
0.87
1.14
1.00
1.41*
1.06
1.02
1.21
0.83
0.92
1.70*
1.31
2.01*
0.76
1.42*
1.18
1.99*
1.26*
1.20
1.62*
0.87
0.86
1.64*
1.07
1.38
0.69
1.25
1.09
1.24
1.32*
0.90
1.34*
0.80*
0.92
1.50*
0.94
1.09
0.68
1.14
1.03
0.36
1.26*
0.90
1.22
0.90
0.84
1.43*
1.30
1.20
1.11
1.34*
1.16
1.69*
1.17
1.05
1.46*
0.97
1.11
1.31
0.67
1.24
0.86
1.06
1.10
1.58*
1.25*
1.12
1431
0.09
1.28*
1.05
1223
0.08
1.30
1.05
639
0.10
1.20*
1.13
1186
0.10
1.29
1.26*
720
0.10
1.24
1.24*
726
0.12
1.12
1.06
700
0.12
1.06
1.25*
646
0.11
1.23*
1.01
642
0.11
1.29*
0.93
1.63*
0.96
1.11
1.29
0.71
0.87
0.60*
1.06
1.12
1.09
0.96
0.95
1.26*
656
0.12
1.10
0.94
1.44*
1.15
0.73*
1.17
0.8
1.07
0.91
1.15
1.00
1.50
1.01
1.24
1.20
647
0.12
1.27*
0.92
1.53*
0.92
0.75*
1.41*
0.97
2.08
1.06
1.89*
1.40
2.40*
1.18*
1.38
1.20
596
0.17
1.01
1.07
1.26*
0.96
0.90
1.39*
0.78
0.85
0.80
1.10
1.00
1.13
1.01
1.00
1.06
671
0.08
1.21
1.11
1.32*
1.07
1.00
1.39
0.76
1.20
0.77
1.18
1.16
1.27
1.03
1.00
1.03
629
0.09
Notes: Support to nuclear power is a dichotomous variable operationalized as 1 being respondents that believed that “Sweden should in the long-term phase out nuclear
power” is a “very bad proposal” and a “pretty bad proposal.” Conversely, 0 denotes respondents that believe that this is a “very good proposal,” a “pretty good proposal,”
or “neither a bad nor a good proposal.” The Swedish center party, “Centerpartiet,” is the reference category for the party dummy variables. Identification with the Swedish
democrat party is only measured from 2007 onwards.
political characteristics, we turn to the results of our multivariate ordered logistic regression models. Table 4 reports the
X-standardized odds ratios for the predictors in these models
explaining opposition to nuclear power across four levels of the
Swedish polity.
When controlling for other relevant variables, we find that
women report greater opposition to nuclear power than do men
at each level of the polity. Indeed, when controlling for political
ideology, the gender difference among national parliamentarians
becomes statistically significant. Thus, female politicians across all
three levels are more likely to oppose nuclear power than are their
male counterparts. This difference in politicians’ attitudes exists
even when controlling for factors such as party affiliation and
socioeconomic characteristics. As such, these results confirm the
pattern of a gender divide on nuclear power attitudes found in a
few earlier studies among municipal-level officials [43] and statelevel officials [5,40]. This pattern whereby female elected officials
are more opposed to nuclear power than are their male counterparts is consistent with the health and safety concerns argument.
At least for the case of Sweden in the late 2000s, nuclear power—a
technology characterized as having significant health and safety
risks—experiences a similar same gender divide among elected officials as is often found among general publics.
The effects of the social, demographic, and political variables
also deserve attention. As in the general public, Left-leaning
elected officials report greater nuclear power opposition than do
their Right-leaning counterparts. Further, compared to elected
officials affiliated with the Center Party, those affiliated with
Right-leaning parties (i.e., the Moderate Party, the Liberal Party,
Swedish Democrats, and Christian Democrats) report lesser opposition to nuclear power, while those affiliated with the Green Party
report greater opposition. Younger municipal-level and countylevel elected officials report greater opposition to nuclear power
than do their older counterparts. The two indicators of socioeconomic status have conflicting effects among municipal-level
officials, with education positively and income negatively related
to nuclear power opposition. Finally, while religiosity has no influence at any level of the polity, having a child at home does increase
opposition to nuclear power among municipal-level elected officials.
38
A. Sundström, A.M. McCright / Energy Research & Social Science 11 (2016) 29–39
4.3. Exploring the determinants of female supporters of nuclear
power in the general public
Our analyses above find that women are more likely than men
to oppose the use of nuclear power. Yet, a sizable percentage of
women do support the use of nuclear power. To extend our analysis,
we briefly examine the social, demographic, and political characteristics of female supporters of nuclear power in the general public
from 1986 to 2011. Here we drop men from our analyses and create
a dichotomous variable to distinguish nuclear power supporters
from non-supporters. For the years 1986–1996, we operationalized supporters as those who believe that “to keep nuclear power
also after 2010” is a “very good” or a “pretty good proposal.” For
the years 1998–2011, we operationalized supporters as those who
believe that “Sweden should in the long-term phase out nuclear
power” is a “very bad” or a “pretty bad proposal.”
Tables A3 and A4 in the Appendix report the results of a series
of logistic regression models predicting whether or not a woman
supports nuclear power. Over the years, the most consistent predictors are age, party identification, political ideology, and income.
In 21 of 25 years, older women are more likely to support nuclear
power than are younger women. In 19 years, female supporters of
the conservative Moderate Party are more likely to support nuclear
power than are female supporters of the Center Party. In 15 years,
right-leaning women are more likely to support nuclear power than
are their left-leaning counterparts. Finally, in 11 years, women with
higher incomes are more likely to support nuclear power than are
women with lower incomes.
5. Conclusion
Whether or not we see a “renaissance” of nuclear
power—perhaps brought about by greater concern about climate change—may depend, at least in part, upon the nature of
public and political support for it (Organization for Economic
Cooperation and Development, 2010). In this study, we focused
on the seemingly intractable gender divide in nuclear power
attitudes, which has been found across several dozen studies in
several countries over the last few decades. Nearly all of these
studies rely upon data from single cross-sections of general publics
in their respective countries.
This leaves two notable gaps in the literature. First, due to the
lack of available high-quality time-series data, scholars have been
unable to examine the stability of the gender divide on nuclear
power attitudes over time. Second, because the same survey items
on nuclear power attitudes that are asked of the general public are
rarely asked of elected officials, we have been unable to examine
the robustness of the gender divide on nuclear power across levels
of the polity (i.e., from local to regional to national levels). Using
a time-series dataset of the Swedish public and a few datasets of
Swedish elected officials that recently have been publicly released,
we addressed both gaps.
We found that the gender divide in opposition to nuclear power
in the Swedish public was quite stable between 1986 and 2011,
even as aggregate levels of opposition fluctuated annually. This
gender divide remained when controlling for a range of social,
demographic, and political characteristics. Further, we found a similar gender divide in opposition to nuclear power across four levels
of the Swedish polity in the late 2000s. Both the robustness of the
gender divide over time and across polity levels is consistent with
the health and safety concerns argument, whereby women report
greater opposition than do men to nuclear power—a technology
characterized as having significant health and safety risks. We also
extended our analyses to examine the predictors of support for
nuclear power among women in the general public, finding that
older women, supporters of the conservative Moderate Party, rightleaning women, and women with higher incomes are more likely
to support nuclear power than are their respective counterparts.
It remains to be seen whether the efforts of groups such as
Women in Nuclear (2014) will have a noticeable influence on the
nuclear power attitudes of women across many countries. At least
for the case of Sweden, the percentage of women in the general public opposing nuclear power varied little over the last 20 years. In that
country, it seems reasonable to conclude that heightened attempts
to market nuclear power or new declarations of its safety will likely
have little effect on the nuclear power attitudes of Swedish women.
The same may likely be the case for female elected officials at each
level of the polity.
While our results are consistent with the health and safety
concerns argument, future research should aim to more explicitly
test its performance in explaining gender differences in nuclear
power attitudes. Indeed, scholars also should test the performance
of similar arguments from environmental social science and the
risk perception literature: values orientations (e.g., [11] and institutional trust (e.g., [56]. Testing such arguments in sufficient depth
will likely require data from new surveys populated with items that
scholars can use to create high-quality measures of key theoretical
concepts (e.g., perceived risks, values, trust, etc.).
To further enhance our scholarly understanding of the nuclear
power attitudes of voters and their representatives, future research
also should examine the extent to which the relationship between
gender and nuclear power attitudes is consistent over time in other
countries—and not just those in Europe or North America. Further,
such future work should also examine the extent to which such a
gender divide in nuclear power attitudes is present among national
legislators. Perhaps even more important, this work also should
investigate the extent to which there is a gender divide in legislative
behavior (e.g., bill introduction, voting, etc.) with regard to nuclear
power.
References
[1] S. Ansolabehere, D.M. Konisky, Public attitudes toward construction of new
power plants, Public Opin. Q. 73 (2009) 566–577.
[2] E. Bernhardt, T. Noack, T.H. Lyngstad, Shared housework in Norway and
Sweden, J. Eur. Soc. Policy 18 (2008) 275–288.
[3] R.J. Bord, R.E. O’Connor, The gender gap in environmental attitudes, Soc. Sci.
Q. 78 (1997) 830–840.
[4] N. Carnes, Does the numerical underrepresentation of the working class in
Congress matter? Legis. Stud. Q. 37 (1) (2012) 5–34.
[5] S.J. Carroll, D.L. Dodson, R.B. Mandel, The Impact of Women in Public Office,
Center for the American Woman and Politics, New Brunswick, NJ, 1991.
[6] A. Corner, D. Venables, A. Spence, W. Poortinga, C. Demski, N. Pidgeon,
Nuclear, power, climate change and energy security, Energy Policy 39 (2011)
4823–4833.
[7] J. Costa-Font, C. Rudisill, E. Mossialos, Attitudes as an expression of
knowledge and ‘political anchoring’, Risk Anal. 28 (2008) 1273–1287.
[8] R. Costello, J. Thomassen, M. Rosema, European parliament elections and
policy representation, West Eur. Politics 35 (2012) 1226–1248.
[9] D.J. Davidson, W.R. Freudenburg, Gender and environmental risk concerns,
Environ. Behav. 28 (1996) 302–339.
[10] J.I.M. de Groot, L. Steg, W. Poortinga, Values, perceived risks and benefits, and
acceptability of nuclear energy, Risk Anal. 33 (2013) 307–317.
[11] T. Dietz, L. Kalof, P.C. Stern, Gender, values, and environmentalism, Soc. Sci. Q.
83 (2002) 353–364.
[12] B.-M. Drottz-Sjöberg, L. Sjöberg, Risk perception and worries after the
chernobyl accident, J. Environ. Psychol. 10 (2) (1990) 135–149.
[13] B.-M. Drottz-Sjöberg, L. Sjöberg, Adolescents’ attitudes to nuclear power and
radioactive wastes, J. Appl. Soc. Psychol. 21 (1991) 2007–2036.
[14] P. Ertör-Akyazi, F. Adaman, B. Özkaynak, Ü. Zenginobuz, Citizens’ preferences
on nuclear and renewable energy sources, Energy Policy 47 (2012) 309–320.
[15] P. Esaiasson, S. Holmberg, H. Oscarsson, L. Wängnerud, The Riksdag Survey
2010 Documentation. Department of Political Science, University of
Gothenburg, Gothenburg, Sweden, 2010.
[16] W.R. Freudenburg, D.J. Davidson, Nuclear families and nuclear risks, Rural
Sociol. 72 (2007) 215–243.
[17] C. Gilligan, In a Different Voice, Harvard University Press, Cambridge, MA,
1982.
A. Sundström, A.M. McCright / Energy Research & Social Science 11 (2016) 29–39
[18] M. Gilljam, D. Karlsson, M. Persson, Representatives’ attitudes toward citizen
protests in Sweden, Legis. Stud. Q. 37 (2012) 251–268.
[19] M. Gilljam, D. Karlsson, A. Sundell, Politics on the Home Ground, Stockholm,
SKL Kommentus, 2010.
[20] M. Greenberg, Energy sources, public policy, and public preferences, Energy
Policy 37 (2009) 3242–3249.
[21] M. Greenberg, H.B. Truelove, Energy choices and risk beliefs, Risk Anal. 31
(2011) 819–831.
[22] S. Holmberg, The Nuclear opinion Pre-Fukushima, in: S. Holmberg, L. Weibull
(Eds.), The Island of Bliss, SOM Institute, Gothenburg, Sweden, 2012, pp.
215–222.
[23] S. Holmberg, P. Hedberg, Will of the People?: Swedish Nuclear Power Policy.
Report 2011:6 of the Swedish National Election Studies Program, University
of Gothenburg, Gothenburg, Sweden, 2011.
[24] D.V. Hosmer Jr., S. Lemeshow, Applied logistic regression, Second ed., Wiley
and Sons, New Jersey, 2000.
[25] Interparliamentary Union, The Swedish Parliament. (accessed on 03.24.14.)
(2012) <http://www.ipu.org/parline/reports/2303.htm>.
[26] S. Jäckle, R. Bauschke, Comparing socialization, cultural, and individual level
effects on attitudes towards nuclear energy, Politics Cult. Soc. 2 (4) (2011)
341–366.
[27] T.K. Jensen, Risk Perceptions among Members in Parliament, in: P. Esaiasson,
K. Heidar (Eds.), Beyond Westminster and Congress: The Nordic Experience,
The Ohio State University Press, Columbus, OH, 2000, pp. 210–236.
[28] C. Keller, V. Visschers, M. Siegrist, Affective imagery and acceptance of
replacing nuclear power plants, Risk Anal. 32 (2012) 464–477.
[29] E.F. Keller, Reflections on Gender and Science, Yale University Press, New
Haven, 1985.
[30] K.S. Larsen, Attitudes toward the transportation of nuclear waste, J. Soc.
Psychol. 134 (1) (1994) 27–34.
[31] J.S. Long, J. Freese, Regression Models for Categorical Outcomes Using STATA,
Second ed., STATA Press, College Station, TX, 2005.
[32] A.M. McCright, C. Xiao, Gender and environmental concern: insights from
recent work and for future survey research, Soc. Nat. Resour. 27 (2014)
1109–1113.
[33] R. Morioka, Gender difference in the health risk perception of radiation from
Fukushima in Japan, Soc. Sci. Med. 107 (2014) 105–112.
[34] T. Nordhaus, J. Lovering, M., Shellenberger, How to Make Nuclear Cheap. July
2013 Report. The Breakthrough Institute, Oakland, CA. (accessed on 03.13.14.)
(2013) from <http://thebreakthrough.or/index.php/programs/energy-andclimate/how-to-make-nuclear-cheap>.
[35] A. Nyberg, Gender equality policy in Sweden: 1970–2010s, Nordic J. Work.
Life Stud. 2 (4) (2012) 67–84.
[36] F.C. Pampel, Support for nuclear energy in the context of climate change
evidence from the European union, Organ. Environ. 24 (2011) 249–268.
[37] F.C. Pampel, L.M. Hunter, Cohort change, diffusion, and support for
environmental spending in the United States, Am. J. Sociol. 118 (2012)
420–448.
[38] L.K. Pifer, The development of young American adults’ attitudes about the
risks associated with nuclear power, Public Underst. Sci. 5 (2) (1996) 135–155.
39
[39] D. Pregibon, Logistic regression diagnostics, Ann. Stat. 9 (4) (1981) 705–724.
[40] B. Reingold, Representing Women, University of North Carolina Press, Chapel
Hill, NC, 2000.
[41] F. Rosenbluth, R. Salmond, M.F. Thies, Welfare works, Politics Gend. 2 (2006)
165–192.
[42] L. Sjöberg, B.-M. Drottz-Sjöberg, Attitudes towards nuclear waste and siting
policy, in: A.P. Lattefer (Ed.), Nuclear Waste Research, Nova Science
Publishers, New York, 2008, pp. 47–74.
[43] L. Sjöberg, B.-M. Drottz-Sjöberg, Risk perception by politicians and the public,
Energy Environ. 19 (2008) 455–486.
[44] P. Slovic, Trust, emotion, sex, politics, and science: surveying the
risk-assessment battlefield, Risk Anal. 19 (1999) 689–701.
[45] L.S. Solomon, D. Tomaskovic-Devy, B.J. Risman, The gender gap and nuclear
power, Sex Roles 21 (1989) 401–414.
[46] B.K. Sovacool, Diversity: energy studies need social science, Nature 511
(2014) 529–530.
[47] B.K. Sovacool, What are we doing here? Analyzing fifteen years of energy
scholarship and proposing a social science research agenda, Energy Res. Soc.
Sci. 1 (2014) 1–29.
[48] Statistics Sweden, The Eight Parties Election 2010. General Elections, Election
Study, Statistics Sweden, Stockholm, 2010.
[49] A. Sundström, D. Stockemer, What determines women’s political
representation at the local level? A fine-grained analysis of the European
regions, Int. J. Comp. Sociol. (2015).
[50] H.B. Truelove, M. Greenberg, Who has become more open to nuclear power
because of climate change? Clim. Change 116 (2013) 389–409.
[51] United Nations Development Program, Explanatory Note on 2013 Human
Development Report composite indices, Sweden. (accessed on 03.25.14.)
(2013) from <http://hdr.undp.org/sites/default/files/Country-Profiles/SWE.
pdf>.
[52] F. Vernersdotter, The National SOM Survey 2011, in: L. Weibull, H. Oscarsson,
A. Bergström (Eds.), In the Shadow of the Future, SOM Institute, Gothenburg,
Sweden, 2012, pp. 575–608.
[53] A. Walczack, W. van der Brug, Representation in the European parliament,
Eur. Union Politics 14 (2013) 3–22.
[54] Women in Nuclear, History. (accessed on 03.15.), (2014) from <http://www.
win-global.org/about/history>.
[55] C. Xiao, A.M. McCright, Explaining gender differences in concern about
environmental problems in the United States, Soc. Nat. Resour. 25 (2012)
1067–1084.
[56] C. Xiao, A.M. McCright, Gender differences in environmental concern:
revisiting the institutional trust hypothesis in the USA, Environ. Behav. 47
(2015) 17–37.
[57] N. Yu, Y. Zhang, J. Wang, X. Cao, X. Fan, X. Xu, F. Wang, Knowledge of and
attitude to nuclear power among residents around Tianwan nuclear power
plant in Jiangsu of China, Int. J. Med. Sci. 9 (5) (2012) 361–369.