What can be learned from the TriElec surveys

7At regular intervals between October 2011 and May 2012, the five waves of TNS Sofres-TriElec surveys [8] measured the level of support among the French population for the production of nuclear energy. They did so by posing the same question as that which appeared in the Eurobarometer survey conducted around the same time. Participants were asked whether they were “totally in favour of, mostly in favour of, mostly opposed to or totally opposed to the production of energy by nuclear power plants?”.

8How should the responses to such a polling question be viewed? In French political science, debate between the champions and detractors of opinion polls is fierce. On the one hand, some political commentators, pollsters and journalists generally deem the answers given in opinion polls to be expressions of valid opinions or, in other words, a measure of political preferences. On the other, Pierre Bourdieu and a plethora of French authors after him have refuted such presuppositions. For Bourdieu, “all opinion polls assume that everyone can have an opinion; or, in other words, that the formulation of an opinion is something which is within everybody’s grasp. At the risk of contradicting a naively democratic sentiment, I call this first assumption into question”. [9] If we agree with Bourdieu, then we should view the analysis of responses given to the opinion polls as devoid of any scholarly validity, on the basis that such opinion polls are nothing more than “a pure and simple artefact”.

9This article stands somewhere between the two schools of thought. According to the theory first articulated by Zaller, peoples’ attention to public affairs and interest in politics are not static dispositions. Very few individuals have stable preferences across all political issues. Consequently, the responses given to questions asked in opinion polls must not be considered as the products of fixed and consistent attitudes towards the issue in question. Rather, they should be viewed as the reflection of viewpoints or of “the thoughts that are most accessible in memory at the moment of response”. [10] Despite drastically altering the meaning attached to the results of opinion polls, [11] Zaller emphasises that neither personal opinion nor the results of opinion polls are entirely lacking significance: these results should be seen as a gauge of the relative balance of the views expressed on a given issue at either individual or group level. This approach, both realistic and constructive, is particularly appropriate as a tool with which to study and understand the changes seen in the responses to opinion polls. It extends beyond the overly-simplistic alternatives of a change in attitudes unlikely in the short term and random responses produced by non-attitudes. [12]

10Such an approach provides an analytical framework which proves highly relevant in understanding the development of opposition to nuclear energy over the course of the presidential election campaign. In his thesis on the media and the nuclear energy agenda in France between 1970 and 2000, Philippe Blanchard characterises the majority of those participating in opinion polls on nuclear power as “ill-informed and involved only minimally in the issue; [people] whose choices are made with little conviction and change with the political climate”. [13] If we accept Blanchard’s view, then we cannot consider opinions expressed on nuclear energy as the product of stable attitudes. Moreover, Blanchard contends that “a traumatic event, such as a disaster in France or abroad, or a new scandal regarding the transparency of the authorities, could see the majority shift towards supporting the complete decommissioning of all nuclear power stations. Conversely, a major economic crisis (such as that heralded by the crash of the banks and the stock exchange in 2007) or a disruption to the hydrocarbon supply appear sufficient to prompt a majority shift towards support for a relaunch of the power plant construction programme’. [14] Within this context, it is particularly interesting to analyse the impact of the Fukushima disaster and the 2012 presidential election campaign, marked by a devastating economic crisis, on the French climate of opinion regarding nuclear power.

11As far as this impact on the climate of opinion is concerned, Figure 1 reveals a change in public opinion in the aftermath of the Fukushima disaster which, in theory, should come as a surprise. Less than a year after the tragic events of 11 March 2011, [15] the proportion of the French population in favour of nuclear power quickly and sharply increased. The first signs of this development appeared in the last quarter of 2011, more specifically between the end of the socialist primaries and early December of that year. The proportion of those surveyed expressing support for nuclear power went from 50% in October 2011 (39% mostly in favour and 11% totally in favour) to 58% in December 2011 (44% and 14% respectively). In February and March 2012, this proportion stood at 61% (46% and 15% respectively), before finally rising to 66% in May 2012 (48% and 18% respectively).

Figure 1

Support for the production of nuclear energy in France (October 2011-May 2012)

Support for the production of nuclear energy in France (October 2011-May 2012)

Confirming an unprecedented trend

12Clearly, measuring a phenomenon and the way in which it develops using only one variable can lay researchers open to the accusation of producing entirely coincidental results. [16] In order to confirm the robustness of the trend uncovered by the TriElec surveys, we adopted two validation strategies.

13First, in order to account for the bias which might have been introduced as a result of the fact that the TriElec surveys are not specifically devoted to the issue of nuclear power, but rather monitor the presidential campaign in general terms, we compared our results to those of other surveys available which included indicators on this issue. All of the survey series available confirmed the observations produced using the TriElec data. The trends described above are consistent with those identified by the survey “Studies of the perceptions and expectations of the general public” [17] and by the EDF’s “Energy Barometer”, the annual survey of the French population on energy-related issues. They also tally with the Barometer of Opinion on Energy and Climate [18] conducted by France’s General Commission for Sustainable Development. All produce results which evolved in a similar fashion between 2011 and 2012. [19]

14Second, we constructed a “longitudinal index of preference” [20] (or “mood”) on nuclear power, following the method developed by Stimson. This measurement synthesises the changes in public opinion identified from the responses to several questions posed at more or less regular intervals and combines them in one unique indicator which can be compared at different points at time. Having access to several surveys carried out at frequent intervals during the period in question enabled us to gauge a month-by-month “mood” using the surveys conducted after the Fukushima nuclear disaster, between March 2011 and June 2012 (Figure 2). This indicator, based on the 24 measurements taken from six series of questions on opinions regarding nuclear power, confirms the trend [21] revealed by the TriElec data and rules out the possibility that this trend was merely coincidental or based on biased measurements. The increase in support for nuclear power between November 2011 and March 2012 is plain to see (+14 points), after a period of relatively stable levels of support between March and November 2011. [22]

Figure 2

Changes in levels of support for nuclear energy in French public opinion (mood) between March 2011 and June 2012

Changes in levels of support for nuclear energy in French public opinion (mood) between March 2011 and June 2012

15Measuring the “mood” is also useful in order to place recent developments in their historical context. From this perspective, it is particularly interesting to compare such developments in the post-Fukushima period with those seen in the months which followed the Chernobyl disaster. The problem of not having a long series of questions [23] was counterbalanced by the fact that the method used to calculate the “mood” neutralised any bias linked to question headings or the methodologies used to conduct the surveys. [24] It has already been demonstrated that the level of support for nuclear energy varies according to whether survey respondents are asked about their general position on the production of electricity using nuclear energy, on the need to phase out such technology, or on not replacing end-of-life nuclear power stations. These differences in the way questions are formulated explain how two surveys, published in the immediate aftermath of the Fukushima disaster, were able to reach dramatically different conclusions. [25]

16Our indicator of the annual level of support for nuclear energy (Graph 3) was calculated using 302 measurements, taken from across the responses to 24 different questions from a variety of sources (surveys and barometers), posed a minimum of three times between 1975 and 2012. [26] In addition to its methodological advantages, the use of this indicator was appropriate for our theoretical approach: Stimson himself explicitly stated that where individual-level research was concerned, the approach devised by Zaller and Feldman was the theoretical basis of his macroscopic approach to changes in public opinion. [27] However, as has been emphasised by Stimson, Tiberj and Thiébaut, “the longitudinal index makes it possible to adopt a new approach to context and sociopolitical changes, producing an image of democracy different from that traditionally extrapolated from previous results of analyses of public opinion using classic theories. What the public mood shows is not consistent with either the view that voters operate on the basis of ideology, and have stable preferences and value systems, nor with the theory that only a small minority of citizens who are particularly attentive to public affairs change their opinions in response to the debates and events that unfold in the political sphere”. [28]

17In the longer-term perspective, the polarisation of public opinion in support of nuclear energy in 2011-12 appears yet more exceptional still. Contrary to popular belief, French public opinion has remained largely anti-nuclear since 1986 and the Chernobyl disaster. The level of support for nuclear power in 2012, reached following a sudden spike (+7 points), was relatively high and only equaled by the levels of support seen prior to the Chernobyl disaster and during the “nuclear renaissance” at the dawn of the new millennium. The changes in public opinion post-Fukushima are even more astounding when compared with those seen in the months and years following the Chernobyl disaster. This period saw a continued decline in support for nuclear energy (-19 points) – support which had fallen to around 40% by 1990. [29] It would thus appear that the two disasters had markedly different repercussions on the climate of opinion. From 2009 onwards, support for nuclear energy broadly began to decline once more: the peaks in support seen in late 2011 and early 2012 thus sharply reversed this downward trend to bring our indicator to its current level.

18The growing support for nuclear power seen in French public opinion during the 2011-2012 election campaign, a trend confirmed by this first section of the article, is unique from several points of view. It is unique, firstly, from a comparative perspective – anti-nuclear sentiment in other countries broadly continued to grow in the wake of the Fukushima disaster. [30] It is unique, secondly, from a historical point of view, as French support for nuclear power post-Fukushima contrasts starkly with the increasingly anti-nuclear public sentiment in the wake of the Chernobyl disaster. How, then, can we explain this spike in support for nuclear power in the year following a major disaster? The extensive media coverage, unprecedented politicisation, and successful reframing of the issue are key to understanding the paradoxical decline in hostility to nuclear power seen during the 2012 campaign.

Figure 3

Changes in levels of public support (“mood”) in France between 1977 and 2012

Changes in levels of public support (“mood”) in France between 1977 and 2012

Nuclear power hits the headlines like never before

19In addition to a changing climate of opinion, France’s presidential election campaign was marked by the unusually extensive media coverage of issues related to nuclear power. Analysing the pages of French daily Le Monde, researchers have charted the development of this media coverage. The crucial role played by this newspaper of reference in placing these issues on the media agenda has already been made clear by Philippe Blanchard, who noted that other media outlets were quick to adopt the publication’s priorities – these included dailies (Libération), weeklies (Le Point and L’Express) and television news broadcasts (France 2). Blanchard concluded that “the issue of nuclear power gained prominence in the public sphere largely due to […] a short-term domino effect among media outlets” [31] and Le Monde played a pivotal role in this process.

20As shown in Figure 4, broadly speaking nuclear energy received only limited attention from the late 1970s onwards, although did occasionally come under the media spotlight. The years 1977-1981 were marked by vehement anti-nuclear protests in France – protests at least equivalent in scope to those seen in West Germany during the same period [32] – mobilisation which explains the increased prominence of nuclear issues in the media. Whilst the early 1980s saw the anti-nuclear movement lose impetus and the issue depoliticised, [33] the years which followed saw media interest in the issue spike on four separate occasions. The first was in 1986, in the aftermath of the Chernobyl disaster. The second came in 1987, sparked by the controversy surrounding the nuclear waste reprocessing plant in The Hague, the arrival in government of ministers from environmental parties, and debates on the decommissioning of French nuclear power plant Superphoenix. [34] A third followed in 2006-2007, triggered not only by the “nuclear renaissance” and the decision to restart the programme of building EPR reactors in France following a lengthy break in construction, but also by the debates on nuclear proliferation and the potential role of nuclear power in the fight against climate change. [35] Interest in the issue peaked for a fourth and final time in 2011, in the wake of the Fukushima disaster.

Figure 4

The number of articles related to nuclear energy published in Le Monde [36]

The number of articles related to nuclear energy published in Le Monde [36]

21However, the historically high visibility of nuclear issues in the media in 2011-2012 cannot be explained solely by the Fukushima disaster: the extensive media coverage was also the result of the unprecedented politicisation of the issue of nuclear power during the presidential election campaign. A detailed analysis of the various phases of media coverage of nuclear power from January 2006 onwards confirms the crucial role played by these two factors (Figure 5). Whilst the media coverage of the issue of nuclear power in Le Monde never exceeded 44 articles a month between early 2006 and the Fukushima disaster, it did increase sharply in March 2011, peaking at 131 articles, and coverage then remained relatively high until June 2011. After this lengthy period of intensity, media interest in the issue of nuclear power dropped to the average levels recorded during previous summers before increasing once more in September and October 2011 – a result of the debates on nuclear power staged during the French Socialist Party (PS) primaries. These debates saw Martine Aubry, an advocate of phasing out nuclear power, go head-to-head with François Hollande, who favoured abandoning the “all-nuclear solution”. Media interest in the issue visibly peaked again in November 2011, with 102 articles devoted to nuclear power – an average of more than three a day. The battle between the two candidates coincided with the turbulent conclusion of the programmatic and electoral negotiations between the EELV and the PS. Whilst this debate raged, the French president and members of the presidential majority, along with several pro-nuclear interest groups, also broached the issue in order to distinguish their own views from those expressed by the PS and the EELV. Media interest in the issue finally began to wane in December 2011, but remained relatively sustained until spring 2012.

Figure 5

Number of articles including the words “energy” and “nuclear” published in Le Monde (January 2006-April 2012)

Number of articles including the words “energy” and “nuclear” published in Le Monde (January 2006-April 2012)

22A comparison of the media exposure of nuclear power during the 2007 presidential election campaign and that of the same campaign five years later confirms that the latter was exceptional in nature. Between September 2011 and April 2012, the number of articles devoted to nuclear power was almost 80% greater than that recorded during the same period in 2006-2007. From this perspective, the pivotal moment where media coverage of the issue is concerned was November 2011, when there were three times more articles written about the nuclear issue that in November 2006. In the months which followed, media exposure remained higher than it had been in 2006-2007. The tensions between the EELV and the PS in fact opened a window of opportunity for the unprecedented politicisation of the nuclear issue.

Unprecedented politicisation

23In 2012, for the first time since the inauguration of the Fifth Republic, nuclear energy policy featured among the major issues debated during a national election campaign. For a long time, nuclear policy had not been politicised – in other words, it was not an issue on which the major political parties had visible differences of opinion. [37] Indeed, looking back over France’s national election campaigns and analysing the programmes of the country’s political parties clearly demonstrates that despite the vocal and widespread protests in civil society towards the end of the 1970s, the major parties remained steadfast in their pro-nuclear stance. As part of a project comparing the choices of nuclear policy in Western Europe, [38] we systematically coded the positions enshrined in each programme from 1958 onwards on a scale from 0 (no nuclear power and/or the immediate decommissioning of nuclear power plants) to 10 (ambitious programme for the development of nuclear power). The positions occupied on this scale by the main French political parties, shown in Figure 6, reveal the broad consensus which has long reigned in France concerning the issue of nuclear energy.

Figure 6

Position held by France’s political parties concerning nuclear energy (1958-2012)

Position held by France’s political parties concerning nuclear energy (1958-2012)

24Between 1958 and 1988, all the political parties, with a single exception (PS in 1981) maintained a clearly pro-nuclear stance. The consensus among the country’s major parties and the fact nuclear energy policy was largely absent from party programmes [39] are testament to the fact that the issue was not politicised. This lack of politicisation can be explained, first, by the key preferences of the parties, whose political projects meant that, to a greater or lesser extent, they were predisposed to support the production of nuclear energy; and second by strategic considerations. For historical reasons, parties in the Gaullist tradition and the Communist Party (PCF) were the political forces most in favour of nuclear power. It is also worth noting that both parties often made use of the same types of argument to champion the cause, pointing to the advantages of this technology in terms of well-being, employment, independence, and competitiveness. [40] The parties at the centre of the political spectrum, the majority of which were absorbed by the UDF from 1978 onwards, also actively supported the development of nuclear energy, doing so with particular fervour under the presidency of Valéry Giscard d’Estaing, between 1974 and 1981. In more general terms, the parties in power avoided challenging [41] a programme which was crucial for the country’s electricity supply [42] and in which large sums had been invested. [43]

25The PS was not historically or ideologically predisposed to promote nuclear energy in the same way as the communists, the Gaullists or the centrists. On the contrary, its links with the largely anti-nuclear trade union CFDT [44] compelled it to challenge the energy policy decisions made in the 1960s and 1970s. In fact, prior to the foundation of the Green Party, the stance most critical of nuclear power was that of the socialists in 1981: whilst not explicitly calling pro-nuclear energy policy into question, François Mitterrand did table a number of proposals in order to respond to the concerns of the anti-nuclear movement, including shelving plans to build a nuclear power plant in Plogoff. [45] Nonetheless, nuclear power was not a particularly politicised issue during the 1981 election campaign and the left’s arrival in power did not prompt any wholesale challenge to French nuclear energy policy. The Plogoff nuclear power plant was the only one whose construction was cancelled and, following a debate on energy held during the second semester of 1981, the country’s nuclear programme continued to grow. [46] This continuity can undoubtedly be explained by the conservative nature of previous policy decisions and the amount of money which had already been invested in the development of nuclear energy. However, it was also a result of the PS’s need to retain the support of the PCF, staunch advocates of nuclear power.

26This situation continued until 1993: no political party opposed to nuclear power obtained more than 3% of the vote in legislative elections and the PCF remained the favoured coalition partner of the PS. However, the first electoral victories of the pro-environment parties and the continued decline of the PCF radically altered the state of play: the PS was now forced to accommodate the Greens in its proposals. As far as this development is concerned, Figure 6 clearly demonstrates a shift in the PS’s programme in 1997 when, for the first time in its history, it signed an agreement to govern with the Greens. However, despite the inclusion of Green ministers in the government, France’s nuclear policy was far from abandoned: although the Jospin government decided to halt the construction of the Superphoenix plant in 1988 and refused to authorise any future construction projects, it nevertheless confirmed the entry into service of the Civaux reactors in 1997 and their Chooz counterparts in 2002, as well as connecting both to the national grid.

27Common ground between the Greens and the socialists was scarce, as demonstrated by their legislative proposals. In 2007, the Greens committed to phasing out nuclear power by 2030, whilst the PS presented nuclear energy as “a high-quality sector” and a crucial weapon in the fight against global warming. However, the party simultaneously promised to “reduce the share of nuclear energy in France’s final energy consumption by increasing the share of renewables in this consumption to 20% by 2020 and to 50% in the longer term”. Similar discrepancies between the two parties’ programmes were evident in 2012, with the Greens promising to orchestrate the “total phaseout of nuclear energy within 20 years” and the PS position similar to that adopted in 2007. [47] The centre-right UMP’s position was unchanged, remaining committed to “consolidating the production of nuclear energy”. [48]

28Against this backdrop, the unprecedented politicisation of nuclear power in 2011-2012 does not appear to have stemmed so much from a change in the parties’ positions on the issue as from the revitalisation of the communication strategies used to present the issue to voters. Until 1988, nuclear policy was an issue on which all parties broadly agreed. Subsequently a matter of increasing controversy, it was avoided for strategic reasons. In 2012, the UMP decided to underscore the differences between its own views on nuclear power and those held by the PS and the EELV. [49] First and foremost, this strategy allowed the party to draw attention to the differences of opinion both within the PS-EELV coalition and the PS itself. Such disagreements led to a public debate between the two parties, before and after the coalition agreement was signed, particularly concerning the issue of Mox fuel. To use a metaphor coined by Hylligus and Shields, all that remained for the UMP to do was drive deeper the wedge which already divided the centre-left coalition. [50] With this strategy, the UMP also sought to defend the decisions made by Jacques Chirac and Nicolas Sarkozy to build new reactors in Flamanville (in 2005) and Penly (in 2009). The strategy saw them defend these decisions by reframing the issue of nuclear power in terms of competitiveness, jobs, purchasing power, and the reduction of greenhouse gases in order to counterbalance the risk-focused arguments put forward by the opposition.

The successful “reframing” of the issue by the pro-nuclear camp

29Understanding the way in which the politicisation and the extensive media coverage of the nuclear issue influenced the climate of opinion requires us to conduct a more detailed analysis of the way in which this issue was presented in the media in the year 2011-2012. Several studies have already suggested that the way in which issues are framed has a significant impact on the climate of opinion, [51] particularly where nuclear power is concerned. [52] According to Entman, “[T]o frame is to select some aspects of a perceived reality and make them more prominent […], in such a way as to promote a particular problem definition, causal interpretation, moral evaluation and/or recommendation”. [53]

30The aspects of nuclear energy given particular prominence when the issue was politicised and placed under the media spotlight thus constitute an important parameter which must be taken into account in order to understand the decline in opposition to nuclear power seen in 2011-2012. The politicisation of the nuclear issue – a strategy implemented by Nicolas Sarkozy, the members of the presidential majority, and champions of the nuclear cause – manifested itself as a change in the framing of the issue: whilst the weeks which immediately followed the Fukushima disaster saw a risk-focused frame prevail, the counter-offensive mounted by the pro-nuclear camp saw this frame ousted in favour of one which focused instead on the economic advantages of nuclear power. [54]

31To highlight this development in the framing of the nuclear issue, we conducted a keyword search in the LexisNexis database. This allowed us to target articles published in Le Monde which framed the issue of nuclear power in terms of risk on the one hand, and in economic and industrial terms on the other. [55] Figure 7 illustrates the way in which the priority given to each frame evolved over time; a process evaluated by looking at the number of published articles per month which mobilised each frame between February 2011 and May 2012, as well as by calculating the difference between the number of published articles adopting each frame.

Graph 7

The number of articles published per month on nuclear energy framing the issue in either economic or risk-based terms

The number of articles published per month on nuclear energy framing the issue in either economic or risk-based terms

32The first period when nuclear energy was placed in the media spotlight was during the Fukushima disaster, focusing on its effects on Japan and the reactions of the authorities in both France and other European countries. The high visibility of the issue of nuclear power in the media was accompanied by the domination of a frame which focused on risk and advocated the phasing out of nuclear power. The weeks which followed the Fukushima disaster were marked in particular by the impressive results obtained by the EELV in the March 2011 cantonal elections [56] and the staging of the Green Party primary. In this context, it was the articles framed in terms of risk which dominated the press, with 43 more of these articles published than those framed in economic terms. However, whilst risk-focused articles were perhaps more common, right from the outset the media coverage of Fukushima and its aftermath saw nuclear power also framed in economic terms; framing which was inherently more pro-nuclear in nature. Politicians and the media rallied in defence of the French nuclear industry, not only underlining the efforts made to guarantee the safety of nuclear power plants, but also repeatedly emphasising the advantages of the technology. In particular, Nathalie Kosciusko-Morizet, Minister of the Environment, Eric Besson, Minister of Industry and Energy, and Gérard Longuet, Minister of Defence, accused the anti-nuclear camp of “politically exploiting” the Fukushima disaster in order to strengthen their own position and defended nuclear power, arguing that it allowed France to reduce its consumption of fossil fuels, become more energy independent and produce low-cost electricity for its citizens. [57] Thus, the frequency with which each frame appeared in the press became equally balanced in April 2011, before the economic frame in turn gained the upper hand between May and June of that year (with 38 more articles choosing to frame the issue in economic terms). Over these three months, the fear sparked by events in Japan gave way to questions regarding the political, economic, and energy-related implications of the Fukushima disaster: the price of electricity, the potential consequences of a transition to a new form of energy for the quality of life and well-being of French citizens, its implications for the energy companies EDF and AREVA, and its financial impact as a whole. Debates on these issues often cited the stumbling blocks encountered in Germany, where a possible phase-out of nuclear power was under discussion at the time. These debates came to a head in June, when Nicolas Sarkozy announced that one billion Euros had been allocated to “the nuclear power of the future”.

33The second phase of media coverage played out against a different backdrop; this one marked by the politicisation of nuclear policy choices: first between the EELV and the PS, then between these two parties and the presidential majority. In this phase of media coverage, the positions of the political actors were at the very heart of the debate, particularly the divisions on the left. [58] All candidates and political parties ultimately expressed their position on the nuclear issue – a position sometimes complex in nature. At the same time, the pro-nuclear interest groups pulled out all the stops to promote the French nuclear industry, arguing that phasing out nuclear power would be unrealistic, that it would see many made redundant, and that it would be an extremely costly process. [59] This second phase of media coverage of the nuclear issue manifested itself as a sharp increase in the visibility of the economic frame, beginning in November 2011. In addition to the arguments focusing on the perceived advantages of nuclear energy (the cost of electricity, energy independence, no CO₂ emissions), it was at this point that a new line of argument began to emerge. This time, the champions of nuclear power focused on its social and industrial benefits, the jobs created by the industry, the knowledge and expertise of French power companies, and the financial losses which would be incurred if the construction of the Flamanville EPR were brought to a halt. [60] Some of these arguments were put across by the representatives of people employed in the sector, and by trade union organisations in particular. [61] This economic framing of the issue would remain present in the media throughout the election campaign, and right up until the televised debate between the two rounds of voting, which pitted Nicolas Sarkozy against François Hollande and gave the outgoing president the opportunity to confirm his pro-nuclear stance.

34Thus the mediatisation and politicisation of the nuclear issue were accompanied by a “reframing” of this issue in the media as a whole. From April 2011 the risk frame began to disappear to be replaced by the economic frame, which dominated the election campaign from November 2011.

Competing frames surrounding pro-or anti-nuclear sentiment among the French electorate

36The wave of TriElec surveys conducted in December 2011 incorporated the following questions in order to build up a picture of the key considerations in participants’ minds concerning nuclear energy at the time the survey was conducted: “Are you totally in favour, mostly in favour, mostly against, or totally against the production of energy by nuclear power stations? Totally in favour; Mostly in favour; Mostly against; Totally against; I don’t know. Why?”

37Analysing the responses to this open question makes it possible to highlight the existence of several cognitive frames in relation to the perception of the nuclear issue; their potentially contradictory nature; and their significance in forming individuals’ opinions on whether or not the use of nuclear power stations was justified. In this context, it is in no way surprising that the political strategies that parties adopted to emphasise certain frames over others affected the nature of opinions on nuclear power.

38An analysis of the various registers of argument summarised by Table 1 confirms the coexistence of and competition between several considerations relating to different frames. Among the reasons cited in order to justify the positions expressed on nuclear energy, the most commonly cited consideration (36%) was indisputably negative in tone and focused on the risks attached to the production of nuclear energy. The lack of an alternative (31%) was the second most common reason given. Considerations which were positive in nature (“good for the economy”, “guarantees our independence”, “does not pollute”), cited in 28% of cases, came only just ahead of the argument on the role to be played by other forms of energy.

Table 1

Reasons cited as justification for opinions expressed on nuclear power

Nuclear energy… Totally in favour Mostly in favour Mostly against Totally against Total … is dangerous 5 7 75 78 36 … produces unmanageable waste 2 1 9 12 5 … can or must be replaced or supplemented by other sources of energy 6 15 22 14 16 … has no alternatives 56 56 6 3 31 … is good for the economy 33 26 1 1 16 … guarantees our independence 9 6 0 0 4 … does not pollute 21 13 0 1 8

Reasons cited as justification for opinions expressed on nuclear power

39Unsurprisingly, these viewpoints were not equally prevalent across all groups of survey participants and their distribution was strongly linked to the overall position on nuclear power expressed. Those against the technology highlighted the dangers of atomic energy (over 75% of respondents) and mentioned – with frequent reference to the Fukushima disaster – the safety of nuclear power plants, the risks of a disaster, and the consequences of that disaster

40in terms of the environment, health and pollution. To a much lesser extent, the anti-nuclear camp referred either to viable alternatives or those to be developed (around 15-20%) as well as the problem of nuclear waste (around 10%). Those in favour of nuclear power emphasised first and foremost the lack of alternatives available (“we’re not going back to candles”), its comparative qualities, or the difficulties linked to phasing it out. Economic advantages were then cited as factors justifying support for the industry (around 30%): the low cost of nuclear electricity, the numerous jobs supported by the sector, the export opportunities it allowed, and so on. Nuclear energy was also presented as clean energy, neither polluting nor contributing to climate change. Finally, the energy independence which France enjoys thanks to nuclear power was often cited by those fighting its corner. [62]

41Thus, each frame tends to go hand in hand with a particular stance, and only a minority of respondents expressed arguments which were directly contradictory. For example, only 8% of those in favour of nuclear power stated that it was a safe form of energy and only a minute proportion of this group believed that nuclear waste was in no way problematic. Conversely, very few participants from the anti-nuclear camp referred to the cost of nuclear power in relation to other sources of energy. The lines of argument put forward by the two camps each mobilised different methods of framing the issue rather than directly contradicting one other. However, as is the case with the call for research into alternative forms of energy, certain arguments regarding nuclear power tended to garner the support of both camps, and individuals could deem several divergent arguments valid. Thus, nuclear power could be viewed as both risky and economical, and the overall position expressed in such cases was determined by the weight given to each argument. [63] It is most likely for this reason that survey participants chiefly avoided the extremes of the scale when expressing their opinion on the generation of electricity by nuclear power plants (see Figure 1).

42The pivotal role played by the framing of nuclear issues in the formation of opinion, combined with the ambivalent stance expressed by the majority of individuals – willing to contemplate several frames with contradictory implications – confirms that both political and media discourse, as well as the frames [64] that these arguments mobilise, can significantly influence the views of the electorate. This would seem to be consistent with the mechanism according to which “the debate of the elites forms public opinion by formatting public problems and controlling the intensity of citizens’ exposure to the problems selected for discussion”. [65] The frames mobilised in relation to the generation of nuclear electricity are therefore not necessarily mutually exclusive: it is the hierarchy into which these different frames are organised (or their relative prominence) which ultimately explains opinions and the way they develop. As a result, the priority given by those in favour of nuclear power to both economic framing and the argument emphasising the lack of alternatives can be viewed as a consequence of the frame which dominated the media during the second phase of coverage outlined above. These observations support the notion that the successful reframing of the nuclear issue over the course of the campaign is key to explaining the rapid decline of opposition to nuclear energy.

Factors behind the decline in anti-nuclear sentiment among the French electorate

43Whatever the case, an analysis of the reasoning which underpinned the positions on nuclear energy expressed by survey participants does not yield any information regarding the changes in these positions seen over the course of the election campaign. And yet, if the decline in anti-nuclear sentiment was a campaign effect resulting from intense media coverage, unprecedented politicisation, and the economic reframing of the nuclear issue, changes in individual opinions must necessarily be explained by these three phenomena. Four separate hypotheses can be formulated in order to test the implications of our overall explanatory framework regarding opposition to nuclear power.

The hypothesis of reframing

48Whilst the framing of an issue is likely to affect the position which individuals adopt on that issue, [76] this effect can vary according to the nature of the framing used: [77] for example, the religious framing of the death penalty will resonate more or less strongly depending on whether the individuals targeted are more or less religious. As far as nuclear power is concerned, the economic framing of the issue should logically have a greater impact upon individuals who closely monitor the state of the economy than upon those focused on the environment or health. In order to test this hypothesis, we used the most significant problem variable and compared individuals whose primary concern was the economy (and its various forms) with the other respondents. More specifically, we assumed that the decline in opposition to nuclear power was more pronounced among the individuals most concerned about the state of the economy.

49These four hypotheses were tested by completing a series of logistic regressions on a data file which merged together the five waves of TriElec surveys during which participants were questioned regarding their views on the production of nuclear energy (Table 2). These analyses aimed to assess the impact of party proximity, media practices, interest in the presidential election, and political focus, whilst at the same time controlling for the effect of key socio-demographic variables (gender, age, level of education and income [78]). Our dependent variable was opposition to nuclear energy. Its value was 1 when respondents declared that they were either mostly or totally against nuclear energy, and 0 when they stated they were either mostly or totally in favour of nuclear energy. In order to model the changes in these opinions over time, we also included linear interactions between the explanatory variables and the number of months which had passed since the first wave of the survey. Four successive models were evaluated: the first tested the hypothesis of politicisation; the second tested the hypothesis of media exposure; the third tested the hypothesis of interest in the presidential election and included socio-demographic variables; finally, the fourth of these models brought together all of the previous hypotheses, the hypothesis of reframing, and the socio-demographic variables. [79]

50Model number 4, the most comprehensive of them all, did not reveal any major differences from what had already been seen in the first three models regarding the effect of the variables testing our four hypotheses. The results obtained were reliable, and consequently the discussion which follows relates chiefly to the results of model 4.

Table 2

Factors behind the changes in opposition to nuclear energy (logistic regression)

Opposition to nuclear energy Model 1 Model 2 Model 3 Model 4 Time Number of months 0.922** 0.927** 0.889* 1.004 Party proximity EXG 2.143* 1.857 PCF 0.786 0.858 PG 0.669 0.747 PS Réf. Réf. EELV 4.653** 4.264** MoDem 0.719 0.754 UDI 0.315** 0.317* UMP 0.299** 0.287** FN 0.671 0.617 No party 0.754 0.647* Other party 1.031 1.030 Proximity/ time EXG 0.929 0.921 PCF 0.981 0.951 PG 1.158** 1.126* PS Ref. Ref. EELV 0.989 0.997 MoDem 0.930 0.920 UDI 0.908 0.928 UMP 0.902** 0.924* FN 0.990 0.990 No party 0.993 0.978 Other party 0.990 0.982 Media practices Regular TF1 0.910 0.949 Regular France 2 0.682* 0.683* Regular news channel 0.612* 0.584* Regular other channel 0.511** 0.448** Occasional viewer 1.088 0.892 Does not watch television Ref. Ref. Practices/Time Regular TF1 0.905** 0.921* Regular France 2 1.026 1.033 Regular news channel 0.984 1.028 Regular other channel 1.105* 1.163**

Occasional viewer 0.951 0.981 Does not watch television Ref. Ref. Gender Female 1.846** 2.200** Male Ref. Ref. Gender/Time Female 1.023 Male Ref. Age 18-30 Ref. Ref. 31-45 0.837 1.117 46-60 0,944 1.190 61-75 0.701 1.067 + 76 0.505* 0.785 Age/Time 18-30 Ref. Ref. 31-45 1.044 46-60 1.034 61-75 1.037 + 76 1.037 Level of education Primary Ref. Ref. Secondary 1.102 1.236* HE short course 1.199 1.388** HE long course 1.003 1.177 Level of education/Time Primary Ref. Ref. Secondary 1.026 HE short course 1.049 HE long course 1.082 Income 0-20 Ref. Ref. 20-50 0.802 0.768** 50-75 0.614** 0.659** 75-90 0.685* 0.674** 90-100 0.583** 0.511** Income/Time 0-20 Ref. Ref. 20-50 0.973 50-75 0.997 75-90 0.968 90-100 0.929 Interest in election Interested 1.002 1.029 Not interested Ref. Ref.

Interest/Time Interested 0.943* 0.923* Not interested Ref. Ref. Focus Debt 0.841 Unemployment/employment 0.783 Purchasing power 0.996 Other economic issue 0.820 Other issues Ref. Focus/Time Debt 0.900* Unemployment/employment 0.987 Purchasing power 0.948 Other economic issue 0.978 Other issues Ref. N 4871 4871 4871 4871 Number of surveys 5 5 5 5 Nagelkerke’s R2 0.175 0.037 0.092 0.241

Factors behind the changes in opposition to nuclear energy (logistic regression)

NB: The models included a control for each wave of surveys. The levels of significance retained were * < 0.05 and ** < 0.01

51From the outset, our analyses confirmed the movement of public opinion towards a pronuclear position over the course of the presidential campaign. In the first three models, which tested our hypotheses independently, the likelihood of being anti-nuclear rather than not adopting such a stance decreased by between 7 and 11% each month, all other things equal (odds ratios of between 0.889 and 0.927 for the Time variable). We can thus conclude that the likelihood of being against nuclear energy during the 2012 presidential election campaign had decreased significantly. However, in model 4, the Time variable was no longer statistically significant and no longer had any effect on opinion, with an odds ratio of 1.004. The changing levels of anti-nuclear sentiment seen over time are thus clearly explained by the variables specifically constructed to test our four hypotheses. At the same time, model 4 is undoubtedly that which has the greatest heuristic value (Nagelkerke’s R² of 0.24).

52In accordance with the media exposure hypothesis, the intensity and variety of media exposure at least partly explain the changes in levels of anti-nuclear sentiment seen during the presidential election campaign. Compared to the survey respondents who stated that they did not watch the television news, regular viewers of France 2, 24-hour news channels and other television channels were significantly less opposed to nuclear energy, all other things being equal. The viewers of each of France’s various television channels thus displayed specific characteristics at the beginning of the presidential election campaign. The way in which the opinions held by each group of viewers changed over time also varied significantly depending on the group in question. The regular TF1 viewers were the only survey participants whose opposition to nuclear energy decreased significantly over the course of the election campaign: each month, the likelihood of an individual in this group expressing anti-nuclear sentiment decreased by around 8% (odds ratio of 0.921). Conversely, those who were regular viewers of channels other than TF1, France 2 and the 24-hour news channels developed increasingly anti-nuclear views in comparison with respondents who claimed not to watch the television news: the odds ratio increased by 16% each month. Whether we look at the beginning or the end of the election campaign, the opinions held by the occasional viewers did not differ significantly from those of the respondents who claimed not to watch the television news. In accordance with the hypothesis of media exposure, significant changes in levels of anti-nuclear sentiment were seen solely among the television viewers most exposed to information (the regular viewers). However, these levels of opposition evolved only selectively and in different directions according to the type of media in question – opposition gradually decreased among TF1 viewers and became increasingly pronounced among those who watched the news on channels other than TF1, France 2 and the 24-hour news channels. There is potentially a myriad of reasons which explain these contrasting changes of opinion among the consumers of different media. It is quite plausible that these contrasts could reflect the varying degrees of coverage of nuclear-related news, the framing of the issue, and the tone used to present it across France’s media outlets, but conducting such an analysis falls far outside the scope of this article. Whatever the case, the results presented here validate the hypothesis of media exposure: the changes in levels of opposition to nuclear power varies according to the intensity of an individual’s exposure to news on the issue and the type of media they consume, even if the impact of media practices is far from systematic.

53In accordance with the hypothesis of interest in the election, the extent to which an individual is receptive to attempts to reframe the nuclear issue does not depend solely on their exposure to the media: such receptiveness is also modulated by their interest in the election. Whilst in the early stages of the campaign such interest had no significant effect on levels of antinuclear sentiment, the same cannot be said for the way opinion on the issue changed over the course of the campaign. Indeed, every month, the anti-nuclear/ pro-nuclear ratio among the respondents interested in the 2012 presidential election was almost 8% lower than that among the respondents who claimed to have no interest in the contest. On the day of the second round of voting, the likelihood of being against rather than in favour of nuclear energy was 1.5% lower among respondents interested in the election than among those with no interest in it whatsoever. The conclusive test of the hypothesis of interest in the election clearly validated our view that the changes seen in levels of opposition to nuclear energy were a campaign effect: it was clear that opposition to nuclear power declined most sharply among those who had paid closest attention to the election campaign.

54Model 3 also presented the effect of socio-demographic variables on opposition to nuclear power. Our results confirm the findings of previous studies conducted on the factors which determine the extent of opposition to nuclear power. [80] Women are more opposed to the technology than men, and older citizens are most favourably disposed towards it. As an individual’s income increases, their opposition to nuclear energy declines. The least educated individuals are the ones most fervently opposed to nuclear power. However, none of the socio-demographic variables significantly affected the development of anti-nuclear sentiment over the course of the campaign. [81] The decline of opposition to nuclear power was not the result of the specific disaffection of one social category, understood in sociodemographic terms.

55Consistent with the hypothesis of politicisation, party proximity produced a significant effect where the supporters of several parties were concerned. Unsurprisingly, all things being equal elsewhere, model 4 indicated that, among Green Party supporters, the likelihood of being against nuclear power rather than not was almost 4.5 times higher than it was among supporters of the PS. [82] Conversely, among supporters of the UMP and the centrist UDI, this likelihood was 3.5 and 3 times lower respectively than it was among PS supporters.

56However, and this is our chief concern, this difference between the two main French parties increased still further over the course of the campaign: the decline in opposition to nuclear energy was much more pronounced among UMP sympathisers. The interaction variable recording the effect of party proximity over time was statistically significant for the UMP and indicates that among its sympathisers, the likelihood of holding an anti-nuclear opinion rather than not doing so decreased by around 7% a month compared to the same likelihood among supporters of the PS. This result is in line with the expectations stemming from our hypothesis – that the opinions of UMP followers would gradually fall into line with their party’s position on the nuclear issue. The ever-growing difference between UMP and PS followers during the election campaign saw opinions become acutely polarised on the nuclear issue and resulted in marked differences between the voters of the two parties: the antinuclear/pro-nuclear ratio was three times lower in the UMP than in the PS in October 2011. By May 2012, this ratio was 5.5 times lower.

57The significant increase in opposition to nuclear power among Parti de Gauche sympathisers compared to supporters of the PS is also consistent with the hypothesis of politicisation. Indeed, unlike the situation in the Green Party, Parti de Gauche sympathisers were largely unaware of the anti-nuclear stance of their party and its leader until the politicisation of the issue during the election campaign. As a result, for these voters, the presidential campaign allowed them to access the information necessary to bring their stance in line with that of their party, with the paradoxical result that a Parti de Gauche sympathiser was more likely to oppose nuclear energy at the end of the campaign than at the beginning. [83] That UMP and Parti de Gauche sympathisers should undertake journeys in opposite directions is entirely consistent with Zaller’s notion that public opinion becomes polarised in an environment comprised of diverse information, which pits one elite against another. [84]

58The politicisation hypothesis is also confirmed by the absence of any significant changes in opinion among Green Party voters. Given the prior and perennial politicisation of the nuclear issue amongst Green voters, the 2012 campaign in no way altered the state of play. These results were consistent with our expectations and unambiguously validate the hypothesis of politicisation. Moreover, not only does model 1 have the most explanatory power of the first three models discussed (with a Nagelkerke’s R² of 0.175), but this power compares favourably with that of the final model. This result suggests that the party dimension is the key factor in explaining opposition to nuclear power (or a lack thereof), and the way levels of such opposition changed over the course of the campaign.

59Finally, beyond politisation, the final hypothesis tested implied that those most concerned about economic issues should logically be those most affected by the economic reframing of the issue. These concerns were not, in themselves, significant in determining opposition to nuclear power, even if all the odds ratios indicated a negative effect. In terms of changes in opinion, whilst all odds ratios revealed that the likelihood of voters expressing anti-nuclear sentiment continued to decrease, only those concerned by debt and the public deficit were significantly more opposed to nuclear power during the election campaign than the respondents for whom the state of the economy was not a primary concern. This result is consistent with the hypothesis of reframing, but suggests it only applies to a specific group of respondents concerned with the economy. The economic framing of the nuclear issue resonated particularly strongly among voters concerned solely about debt and public deficit, and triggered a more marked decline in opposition to nuclear power among this group. All other things being equal, concern about employment or purchasing power did not constitute particularly fertile territory for economic framing of the nuclear issue, nor for the effects of this on opposition to nuclear power.

60To conclude, the results of the logistic regression analyses conducted support the four hypotheses outlined. In doing so, they corroborate the various elements of our explanation of the decline in opposition to nuclear power witnessed over the course of the presidential election campaign. Far from uniform in nature, this campaign effect was characteristic of an unprecedented process of politicisation, unusually intense media exposure, and the economic reframing of the nuclear issue.

61* * *

62The dramatic events of Fukushima catapulted the issue of nuclear power into the media spotlight, prompting several countries to announce that the production of nuclear energy would be halted. Such decisions were taken against the backdrop of increasing opposition to the technology among public opinion. This state of affairs sparked unprecedented levels of media interest in the issue and triggered its unprecedented politicisation during the 2012 presidential election campaign. However, paradoxically, the situation also prompted a decline in opposition to nuclear power.

63France’s political parties had not previously sought to extensively politicise the issue – broadly preferring to maintain the pro-nuclear consensus – since, for the most part, they were contending with numerous internal differences of opinion and were anxious not to alienate either their voters or their coalition partners. However, the 2012 presidential election campaign saw this situation markedly change: the public disagreements between the EELV and the PS on the nuclear issue provided Nicolas Sarkozy and members of the presidential majority with the opportunity to draw a clear distinction between themselves and their opponents by reframing the issue in economic terms. This resulted in significant (and sometimes diverging) changes of opinion among specific sections of the electorate: a marked decline in anti-nuclear sentiment among those most interested in the campaign, those most concerned about the issue of debt and public finances, regular viewers of the television news on TF1 and UMP sympathisers; a relative increase in opposition to nuclear power among Parti de Gauche supporters and regular viewers of the television news on channels other than TF1, France 2 and 24-hour news channels. In identifying these dynamics, we have supplemented the conclusions of previous studies into opinions on matters of energy [85] by moving beyond their occasionally static approach to the relationships between groups characterised by specific values and opinions.

64However, our study of the nuclear issue during France’s 2012 presidential election campaign has emphasised above all else that this campaign had political effects. The extensive media coverage, the considerable politicisation, and the reframing of the issue by Nicolas Sarkozy and his allies did not see the outgoing president re-elected but did have a significant impact on public opinion. In addition to describing the ways in which opposition to nuclear power evolved and linking these changes in opinion to the characteristics which this issue acquired during the presidential campaign at a macroscopic level, this article has explained how the campaign effect made itself felt at an individual level. From this perspective, revealing these dynamics empirically has a significance which extends beyond the particular issue in question.

65The politicisation of an issue has a decisive influence upon the accessibility of positions taken by the party for party sympathisers and allows them to adjust their opinions concerning the issue to bring them closer to those expressed by their party. This phenomenon is likely to have decisive consequences for future electoral dynamics, in particular in terms of issue competition. [86] The media coverage of an issue also has certain effects which vary according to the intensity of an individual’s exposure to it and the type of media consumed. Unlike those analyses which present the twenty-first century as an era in which the media’s impact is minimal, [87] our results clearly reveal that media effects remain visible, at least where opinions on electoral issues are concerned. Moreover, our article acts as a reminder that the framing of an issue does not affect all sectors of the electorate in the same way, but only selectively, according to the extent to which this framing resonates with these sectors’ concerns. Finally, our analyses confirm that the effect of election campaigns depends on the degree of interest which these campaigns arouse: at an individual level, it is those who follow election campaigns most closely who are most susceptible to the campaign effect. Beyond this, election campaigns have an effect if they alter the relative prominence given to the various, largely ambivalent, viewpoints which individuals have on a particular issue.

66Finally, our analysis has implications for the research agenda which seeks to examine the relationship between public opinion and public action, and which does so most often by studying the extent to which policymakers are swayed by public opinion. [88] Our study underscores the need to consider not only “policy feedbacks”, [89] but also the impact on the climate of opinion of discourses which politicise and legitimise public action. As far as nuclear energy is concerned, and particularly in France, the impact which these discourses have on the climate of opinion appears greater than the impact of the climate of opinion on public policy choices. [90]