Don’t Mention the War! Second World War Remembrance and Support for European Cooperation

At the core of the European project is the idea that through increased state cooperation and dependency, national divisions can be overcome and peace can be secured on the European continent. National and European elites often make reference to past devastations of the Second World War (WWII) in order to convey the added value of European cooperation among the public. Does WWII remembrance enhance public support for European cooperation? By presenting evidence from a set of novel survey experiments conducted in the six largest member states (France, Germany, Italy, Poland, Spain and the United Kingdom) in July 2017, this contribution shows that reminding people of the devastations of WWII, increases support for their country’s cooperation within the European Union (EU) today. Yet, only when it comes to providing assistance for other member states in dire economic need. Reminding people of the devastations of WWII does not make people more willing to extend the rights to EU migrants or contribute to the establishment of an European army. These findings are important as they suggest that WWII remembrance triggers a largely transactional response among the public, a willingness to provide financial support combined with an unwillingness to embrace intra-EU migration or security cooperation. This evidence suggests that securing public support for further deepening of free movement of people and European security cooperation through historical rhetoric might be difficult to achieve.


Introduction
On the 12 th of October 2012, the European Union (EU) received one of the important international recognitions, the Nobel peace prize, for its achievements. The Nobel Peace Prize committee awarded the EU because of its contribution to "the advancement of peace and reconciliation, democracy and human rights in Europe" for over six decades. 3 Europe's founding fathers, like Jean Monnet or Robert Schuman, continuously stressed the importance of European cooperation, and the resulting political dependencies, in overcoming war and animosity. Unity on the European continent was seen as key to securing peace. The goal of overcoming national animosities through European cooperation is at the heart of early writings of federalist thinkers like Altiero Spinelli, the Ventotene Manifesto (Spinelli et al. 1944: 8): "The dividing line between progressive and reactionary parties no longer follows the formal line of greater or lesser democracy, or of more or less socialism to be instituted; rather the division falls along the line, very new and substantial, that separates the party members into two groups. The first is made up of those who conceive the essential purpose and goal of struggle as the ancient one, that is, the conquest of national political power -and who, although involuntarily, play into the hands of reactionary forces, letting the incandescent lava of popular passions set in the old moulds, and thus allowing old absurdities to arise once again. The second are those who see the creation of a solid international State as the main purpose; they will direct popular forces toward this goal, and, having won national power, will use it first and foremost as an instrument for achieving international unit." Historians view the legacy of the Second World War (WWII) to have been key in shaping an adherence to a European ideal and a collective understanding of what it means to be European (Dinan 2004(Dinan , 2006. National and European elites often invoke references to the devastations of the WWII in order to convey the added value of European cooperation among the public, and to remind them of a shared identity and past. Think for example of former German Chancellor Helmut Kohl who said that "the most important rule of the new Europe is: there must never again be violence in Europe". 4 A collective identity 5 , like the European one, needs to juxtapose itself against a common other. In the case of the EU, this common other is the past (Risse 2010).
When it comes to public opinion, a growing body of work demonstrates the importance of collective identities for understanding who supports or opposes transnational policy-making in Europe (for an overview see Hobolt and De Vries 2016). Over the past decade or so, students of European integration have highlighted a variety of aspects that are important for understanding how identity structures public opinion towards Europe.
Some authors have stressed the importance of symbols, such a flags or money (Bruter 2005, McNamara 2015, others the role of transnational interactions and contact (Favell 2008, Kuhn 2015, while again others focus on the exclusive or inclusive nature of territorial identities Marks 2005, 2009). Yet, the role that narratives about a shared historical past play in the development of identities or people's perceptions of the European project has received much less attention in the public opinion literature (for notable exceptions see Diez -Medrano 2003, Sternberg 2013. This is perhaps not entirely surprising given the fact that the vast majority of studies of European public opinion to date are quantitative in nature and make use of existing public opinion surveys. The reliance on secondary survey data can be a serious setback due to the fact that they can be limited in their scope, especially when it comes to the measurement of identity or historical narratives (Checkel and Katzenstein 2009 While this contribution by no means claims to fill the gap of our lack of understanding of how historical narratives structure public opinion towards the EU, it aims to shed at least some light on some aspects. Specifically, it aims to better understand how perceptions of a shared European history can help us to make sense of the contours of public opinion towards European integration today in two distinct ways. First, it presents a theoretical lens through which to understand the role of history in public opinion formation towards the European integration based on the notion of benchmarking. Specifically, it suggests that history provides an important benchmark against which a possible alternative to current levels of European cooperation can be judged. By focusing on historical benchmarks as a way to construct the added value of European cooperation today also has the advantage of moving beyond the current state-of-the-art in public opinion research towards Europe in which identity is often juxtaposed to interest as two alternative explanations for attitude formation (see also Kuhn and Nicoli this issue). It might be more useful to explore how identity and interest explanations interact, and a focus on historical benchmarks aims to do exactly that. Second, this study empirically tests the importance of historical benchmarks by experimentally examining how reminding people of the devastations of WWII through a vignette affects their support for European cooperation today in the six largest member states (France, Germany, Italy, Poland, Spain and the United Kingdom) in July 2017.
Exposure to an experimental vignette reminding people of the devastations of the WWII, increases support for their country's cooperation within the EU today, but only in terms of providing financial assistance for other member states in dire economic need. Those being exposed to an experimental vignette highlighting the devastations of the WWII were not more willing to extend the rights to EU migrants or contribute to the establishment of an European army. This evidence suggests that garnering public support for European reforms aimed at deepening the free movement of people and further security cooperation might not be easy.

Historical References and Public Opinion Towards European
Integration "Anyone who believes that the eternal question of war and peace in Europe is no longer there, risks being deeply mistaken." 6 When national and European elites, like Jean-Claude reference the devastations of the WWII, they often intend to highlight the added value of European cooperation and the possible risks of losing it. Essentially, these elites are providing people with an alternative state (war) to the current status quo of EU membership and cooperation (peace). When thinking about this in a more analytical way a review of a benchmark approach to public opinion is helpful. The benchmarking theory suggests that people's attitudes towards the EU ultimately boil down to a comparison, a comparison of the status quo of their country's EU membership and their evaluations of a possible alternative state (De Vries 2018). The alternative state is understood as a situation in which a member state was outside the EU.
The notion of benchmarks builds on existing work highlighting the importance of national context in public opinion formation towards European integration. In its simplest version, the argument is that national performance affects people's support for integration. Sánchez-Cuenca (2000), for example, argues that those citizens who are dissatisfied with the performance at the national level mainly because of corruption are more willing to transfer sovereignty to the EU level. Moreover, Rohrschneider (2002) shows that citizens who perceive their national democratic institutions to be working well, display lower levels of EU support because they view the EU to be democratically deficient. Furthermore, Rohrschneider and Loveless (2010) highlight that this relationship is conditional upon the overall level of economic prosperity in a country. Citizens in less affluent nations evaluate the EU mainly on the basis of economic performance, while in more affluent nations publics rely mostly on political criteria, such as the functioning of their national democracies. What is different in the benchmark approach is that rather than viewing national context as influencing public opinion in a one-directional way flowing from national to European evaluations, it suggests that the flow can be two-directional or that events in one national context can impact people's comparisons of the benefits of the status quo of membership as well as the alternative state in another national context (see also, for example, the benchmark approach in economic voting by Kayser and Peress (2012) European integration and cooperation is higher when people think that they or their country would be worse off under the alternative state scenario. The extent to which people support the status quo of membership thus crucially depends on their beliefs about how well their country would do if they alternative state would materialise and the information they use to form these beliefs. While the alternative state traditionally is understood as a hypothetical state in which one's country is not a member of the EU, it does not seem unconceivable that a similar comparison of benefits between the status quo and alternative state could be made based on history. What is the 7 While people could in principle also benchmark the alternative state by judging how well countries like Norway or Switzerland fare that have never joined the EU, the transaction costs associated with leaving are fundamentally different.
added value of the status quo of EU membership vis-à-vis a situation in which the EU would not exist? Indeed, the importance of historical benchmarks features prominently in work of psychologists and sociologists on nostalgia. While psychologists view feelings of nostalgia, commonly quite casually defined as a feeling that the past used to be better, primarily in reference to personal experiences, such as a birth, degree or other personal milestones, (Batcho 1995, Sedikidis et al. 2008, sociologists define it in a much broader sense and relate it to more general views about the state of the world (Davis 1979, Duyvendak 2011. Notwithstanding these differences, both approaches highlight that nostalgia develops in comparison to a benchmark in the past. Understanding public opinion towards Europe as a comparison between the benefits of the current status quo of membership and those associated with an alternative state, historical events and experiences have the potential to be important. Providing people with a historical benchmark about an alternative to European integration may feed into the calculations of the benefits of the status quo versus those of an alternative state that are the essence of the benchmarking theory. Specifically, one would expect support for European cooperation to increase when the benefits of alternative state look less favorable. When the alternative state looks less attractive, support for the status quo of membership or even further integrative steps in Europe should rise as a result. The EU developed as a peace project against the backdrop of the devastations of WWII (Dinan 2004). Reminding people of these devastations, and this historical past, should increase support for their country's cooperation with other member states within the EU. This is because providing people with a negative historical benchmark highlights the risks and possible costs associated with the alternative state. Reducing the attractiveness of the alternative state should increase support for European cooperation today.

Historical Benchmarks and Support for European Cooperation
As I have highlighted in the previous section, based on the benchmarking theory one would expect that reminding people of the devastations of the WWII to increase support for European integration and cooperation. Yet, it seems important to reflect more on how we understand public support for European cooperation. Clearly, different types of cooperation exist. Also, while some areas of European cooperation seem largely uncontroversial in the eyes of voters given the clear added value of cross-border cooperation, think of pollution or protection of privacy for example that do not stop at the border, other areas may be much more controversial. This may especially be the case when they involve core state powers. This raises the question of how core state powers should be defined? Bremer, Genschel and Jachtenfuchs (this issue) highlight two approaches to defining core state powers: one based on policy area, and another based on power resources. When it comes to the policy-based approach, policy areas are perceived to differ on the basis of how important they are for defining a state as a sovereign entity, examples of these include the political notion of "high politics'" (Hoffmann 1966), or the the legal notion of "reserved powers" (Weiler 1991). In the resource-based definition, core state powers refer to the resources deriving from the monopoly of legitimate coercion and taxation of a state: coercive capacity (military, police, border patrol), fiscal capacity (money, taxes, debt), and the administrative capacity needed to manage coercive and fiscal capacity and to implement and enforce public laws and policies (e.g. Weber 1978, Tilly 1990).
While these definitions differ, they also overlap. In both approaches, the core areas of sovereign government usually concern the mobilization and direction of coercive, fiscal or administrative capacities think of defense policy, budgetary policy, social security policy or border control as examples. When we keep these commonalities in mind and think about how they overlap with areas in which further cooperation which are currently salient in European and domestic political discourse, three areas stand out: 1) financial assistance for economically struggling member states, 2) free movement of people, and defense cooperation. These areas of European cooperation involve core state powers, as they touch on issues of territory and border, taxation and spending as well as coercive capacity. The empirical analysis, therefore, sets out to examine how providing people with a negative historical benchmark affects support for more cooperation in financial assistance, free movement of people, and defense cooperation respectively.

Experimental Design
In order to test whether providing people with a negative historical benchmark, by reminding them of the devastations of the WWII, increases support for European cooperation, this study employs a survey experiment.
This survey experiment was embedded in the July 2017 wave of the eupinions survey. Eupinions is a politically independent platform that collects data on European public opinion. The survey is conducted via a mobile phone sample and is representative with regard to age 8 , gender, education and region. The The order in which respondents receive this vignette is randomized. One groups of respondents, the treatment group, received the experimental vignette prior to answering a set of questions eliciting their views about political cooperation in Europe, while another group of respondents, the control group, received the vignette after answering these questions.
In order to make sure that respondents read the vignette and understood it, both the treatment and control group were given the following question after receiving the vignette: How many people lost their lives in the Second World War? They were provided with two possible answer categories, one which was false, 6 million people, and one which was correct, 60 million people. Overall, 83 per cent of respondents answered the follow-up question correctly which suggests that most respondents read and understood the vignette. Also, there were no differences in the degree of correct answers based on the placement of the vignette, before or after the questions asking about people's views about European cooperation. The empirical results presented in the next section are based on analyses including all respondents.
After the vignette and follow-up question, the respondents in the treatment group were presented with three questions aimed at soliciting their views about future cooperation in Europe. The respondents in the control group received these questions without being exposed to the vignette or follow-up question, while the treatment groups received the vignette and follow-up question before. In order to elicit people's views about future cooperation Europe, the questions capture the three most contentious areas of cooperation, namely providing financial assistance to struggling member states, the rights of EU migrants, and cooperation in the area of defense.
Specifically, respondents were asked the following three questions: 1. To what extent do you think that [your country] should provide financial aid to another EU Member State facing severe economic problems? Respondents could place themselves on a 10-point scale ranging from 1 "My country should not give financial aid" and 10 "My country should give financial aid". 2. To what extent do you think that [your country] should try to limit the rights of EU citizens to work and live here? Respondents could place themselves on a 10-point scale ranging from 1 "My country should not allow them to work and live here" and 10 means "My country should allow them to work and live here".

To what extent do you think [your country] should help to
establish an European army? Respondents could place themselves on a 10-point scale ranging from 1 "My country should not help" and 10 means "My country should help".
The order of the three questions was randomized to rule out any ordering effects. In order to examine whether reminding people of the devastations of the WWII increases support for cooperation in Europe, we compare the answers of those respondents who received the vignette before the questions to those who received the vignette after.
Although the vignette was carefully formulated and failed to refer to any perpetrators, it could be possible that exposure to it triggered anti-German sentiment. In order to rule out this possibility, two additional questions were added to the survey after the questions soliciting people's views about European cooperation. Again the treatment group answered these questions after being exposed to the vignette, while the control group answered these questions after. The two questions were the following: 1. How much do you approve of German Chancellor Angela Merkel? Respondents could choose from the following answer categories: 1) completely approve, 2) somewhat approve, 3) somewhat do not approve, 4) completely do not approve.

Germany is often seen as taking a leadership role in the European
Union. Do you think this is …? Respondents could choose from the following answer categories: 1) very good, 2) good, 3) bad, 4) very bad.
When it comes to respondents' answers to the first question, the average response of the treatment group was 2.57 closest to "somewhat do not approve", and that of the control group was 2.59, also closest to "somewhat do not approve". This difference is small (-0.02) and statistically insignificant (p=0.55). When it comes to the second question, the average response of the treatment group was 2.47 closest to "somewhat do not approve", and that of the control group was 2.49, also closest to "somewhat do not approve". This difference is again very small (-0.02) and fail to reach conventional levels of statistical insignificance (p=0.20).
Finally, in order to make sure that any differences between the treatment and control group in the outcome variables tapping into support European for European cooperation, or lack thereof, is not driven by any other characteristics that differ between the two groups, balance tests were conducted. Table 1 below provides the results. The results show that there are no significant differences between the treatment and control groups when it comes to a series of important background variables. This increases our confidence in the experimental results.

Empirical Results
Before presenting the results from the survey experiment, let us first get a sense of the average support for financially assisting struggling member states, extending rights to EU migrants and aiding the establishment of an European army. Figure 1A shows the mean level of support for European cooperation in the three areas within the six countries together, while Figure   1B shows the mean level of support for each of the six countries. Figure 1A: Support for European Cooperation, All Countries Figure 1B: Support for European Cooperation, Per Country Figure 1A shows that by and large respondents are more supportive of European cooperation than opposed to it, but not by much. Support for extending rights to EU migrants finds the most support, while support for financial assistance for a member state in severe economic difficulty the least. What happens if respondents are provided with a historical benchmark reminding them of the devastations of the WWII, a time when Europe was hugely polarized and member states were at war with one another. Figure 2 displays the results of a regression analysis in which the respondents from all six countries were pooled to estimate the change in support for European cooperation in the three areas based on being exposed to the WWII vignette.
This analysis includes country fixed effects to deal with potential heterogeneity across countries. The results suggest that being reminded of the devastations of the WWII increases support for providing financial assistance to a struggling member state, but has little to no effect support for extending the rights of EU migrants or aiding the establishment of an European army.
Support for financial assistance is about a quarter of a point higher for those who were exposed to the WWII vignette compared to those who were not.
Given that support for financial assistance was measured on a 10-point scale, this would be equivalent to a 2.5 per cent increase in support. This finding is in line with the benchmarking theory of public opinion towards the EU that suggests that when people are provided with an alternative state current membership which is negative, support for the European cooperation should increase. That said, this only seems to be the case for providing financial assistance to another member state that is struggling economically. This underscores the highly transactional nature of public opinion towards the EU.
Extending rights of EU migrants or aiding to an European army that could send fellow countrymen into battle perhaps infringes much more on a sense of national community and of national control.  Another difference that is borne out by the data is the fact that exposure to the WWII vignette increases support for aiding an European army among British respondents. In the five remaining countries, there is no effect. This difference might in part be due to the fact that support for aiding an European army was lower in Great Britain to begin with. Why exposure to the vignette did not trigger the same response Italian and Polish respondents would be a matter of speculation at this point. Further research is needed to address this difference more in-depth.
A last step examines potential heterogeneous effects based on ideology. 10 We know from previous work that people on the right of the political spectrum are less likely to support financial aid to other member states or the extension of rights to EU citizens (Kuhn et al. 2018, Vasilopoulou andTalvig 2018). In addition, historically the left has been quite wary of the establishment of an European army (e.g. Hofmann 2013). For these reasons, we would expect differences based on respondent's ideological selfplacement. In the survey, respondents were asked in the pre-treatment section of the survey, how they identify themselves politically in left or right terms.
The answer categories differ from 1) very left to 6) very right. While the majority of respondents self-identify as centre-left or -right, there is a considerable proportion, namely 10 per cent that places themselves either on the 'very left' or 'very right' of the political spectrum. Figure 4 shows the effect of being exposed to the WWII vignette by left-right ideological selfplacements based on a regression that includes an interaction between the vignette and ideology.
The results of the analysis suggest that while left-versus right-leaning respondents do not really differ in how they feel about extending the rights of migrants after being exposed to the vignette, but they do when it comes to their views about financial assistance and aid to an European army. When it comes to financial assistance, we see that only leftwing respondents differ based on exposure to the vignette. Specifically, when leftwing respondents are reminded of the devastations of the WWII they are more likely to support their country providing financial assistance to a member state in need. There is no difference for respondents who self-identify as centrist or rightwing.
When it comes to potential aid to an European army, we again find that leftwing respondents respond to the vignette. When respondents who identify as leftwing are exposed to the vignette, they are more likely to support their country providing aid to a European army compared to when they are not exposed. Interestingly, this seems to suggest that respondents who self identify on the left respond more to this historical benchmark than rightwing respondents.   France is the reference category. *** p<0.01, ** p<0.05, * p<0.1