Europeans rallied around the EU flag after Russia’s invasion of Ukraine

When Vladimir Putin launched his full-scale invasion of Ukraine on 24 February 2022, he based his course of action on a number of assumptions. He believed that the Ukrainian military would not put up effective resistance and instead it would fold in a matter of days. He expected that the Ukrainians would welcome his army as liberators. Finally, he probably hoped that his actions would divide Europe and weaken its unity.

Fortunately – for Ukraine and for Europe alike – he was wrong on all three counts. The rapid and robust response of EU governments, united in the face of a largely unexpected aggression in their immediate neighbourhood, has been especially remarkable. But do the actions of EU governments match the feelings of EU citizens?

A recent Eurobarometer survey answers this question in the affirmative. According to its findings, a majority of European citizens agrees with the EU’s response to the Russian aggression (moreover, a large majority agrees with the sanctions imposed on Russia), and the support for the EU across its member states is at its highest since 2007 (European Parliament 2022). The survey, however, was carried out in April-May 2022. Besides the invasion itself, the respondents’ views may have been influenced by a host of other issues: the ability of Ukrainians to repel the initial onslaught, news about war crimes committed by the Russians, or about successes of Western weapons delivered to Ukraine. Similarly, sentiment analyses, such as the one summarised in a recent Ukrainian-debate column by Nguyen et al. (2022), reflect anything that has happened in the period since the invasion got underway. Assessing the impact of the invasion itself separately from the effects of the subsequent developments therefore is not easy.

That wars may cause large swings in public opinion has been demonstrated by research on ‘rally around the flag’ effects. These studies (e.g. Mueller 1970) show how the public unites behind national leaders and institutions in response to military conflicts, reflecting a desire for national unity in the face of an external threat. A recent article and a column in the Ukrainian debate have demonstrated that the rally effect can extend to the supranational level. Gehring finds that the illegal annexation of Crimea by Russia in 2014 caused an increase in European identity and support for the EU in Estonia and Latvia, two countries that share a border with Russia and that, like Crimea, have significant Russian-speaking populations. Did the even larger assault on Ukraine on 24 February 2022 lead to a rallying behind the EU flag beyond the two Baltic countries?

In our new paper, we rely on a natural experiment to ascertain how one particular group of Europeans – university students – responded to Russia’s invasion of Ukraine in the immediate aftermath of the assault. Unlike most academic research, this paper came about by accident: our original objective was to assess to what extent university students’ worldviews and attitudes change after having spent a semester in another country. To address this question, we designed a survey that was sent to a large number of students at nine European universities who were selected for an Erasmus stay. The first round was administered in May-June 2021, before the students would embark on their semester abroad. The second round was sent to the same students in February-March 2022.

Three days after we launched the second round of our survey, Russia’s invasion of Ukraine started: the launch of our survey was on 21 February, and Russia attacked Ukraine on 24 February. As it happened, approximately half of our respondents answered the survey before the invasion, and the remainder after (Figure 1).

Russia attacked Ukraine after months of gradually increasing tensions and sabre-rattling. Nevertheless, few expected a full-scale invasion, and the eventual timing of the attack was also largely unexpected. This is demonstrated by Google trends for a combination of words ‘Russia Ukraine war’ depicted for the UK and Russia in Figure 2. In both countries, the frequency of such searches rose precipitously on 24 February. Thus, the timing of the invasion should have been a surprise to the respondents. As it is essentially random whether survey respondents answered the questionnaire before or after the start of the invasion, we can isolate the causal effect of this surprising event. This allows us to assess the impact of the news about the invasion on attitudes, utilising a methodological approach labelled ‘unexpected event during survey design’ (e.g. Muñoz et al. 2020). Previously, this method has been used to investigate the impact of terrorist attacks, natural disasters, or the Covid-19 pandemic.

What does the data say? We asked the survey participants a number of questions on how they feel about the EU and European integration: whether they think of themselves as mainly Europeans rather than a citizen of their country, how strongly they feel attached to Europe, how closely they follow politics at the EU level, whether they think that their country, and they, personally, benefited from EU membership, whether EU member states in economic difficulties should receive financial support, and whether they agree that European integration should be pushed further. We rescaled the responses to range between 0 and 1, with higher values corresponding to more pro-European attitudes.

These seven attitudinal questions were regressed, in our most parsimonious model, on a dummy variable taking the value of one if the response was submitted on 24 February or after, and zero otherwise. For an intermediate model, we also added the respondents’ age and gender as well as dummies for their nationalities: this helps account for potential differences driven by demographics or potential country-specific effects. Finally, in the full model, we added a time trend, i.e. a variable that counts the days since the beginning of the survey. Including the time trend is potentially important because it helps control for the effect of post-invasion events: for example, news about atrocities committed in the course of the conflict, battles won or lost, as well as sanctions and other penalties imposed on Russia. In addition, the time trend addresses the possibility that individuals who took longer to respond to the survey might differ systematically from those who responded more swiftly. The full model thus should allow us to isolate the effect of the news about the start of the conflict from other potential factors.

The results of estimating the three versions of our model using ordinary least squares (OLS) are depicted in Figure 3. Most of the estimated effects are positive and statistically significant, indicating that the news of the Russian aggression against Ukraine increased pro-European attitudes and views in our sample of Erasmus students. The results are particularly strong in the full model, which we believe to be the most appropriate one for this setting. It is noteworthy that the effects are significant primarily for societal aspects of European integration, rather than for individual considerations: for example, post-invasion respondents were more likely to believe that their country benefits from European integration but not that they also personally benefit. This suggests that the ominous news of a conflict near EU’s borders made the students appreciate more the collective benefits provided by the EU.

Our results are remarkably robust: they are affected little by shortening the window around the invasion date, dropping responses received on 21 or 24 February, re-estimating the model only with responses from the three most-represented countries (Belgium, France, and Germany), replacing nationality dummies with dummies for the country of study, varying the way how we account for the time trend or using nonlinear probability models. For a sub-sample of observations, furthermore, we were able to account for the views that the respondents held one year earlier, during the first round of our survey (in May-June 2021): these dynamic (difference-in-differences) results yield again the same pattern of a positive impact of the invasion news. We also show that the Russian aggression impacted attitudes about European integration, but not views about globalisation, or retrospective assessments of the students’ Erasmus stay. Full details of these robustness checks are in the Online Appendix of our forthcoming paper.

Can these findings be generalised more broadly? After all, they are observed in a sample of Erasmus students, who are younger and likely to be more pro-European than the general population. We believe our estimates to constitute a lower bound of effects that could be observed in a more representative sample. This is because our respondents already held rather favourable views about Europe, so the potential scope for strengthening such views was correspondingly limited. The majority of them were, furthermore, from Western European countries where, arguably, the salience of the Russian aggression was more limited. If we had more respondents from Poland and the Baltic countries, the effects might well have been even stronger.

Russia’s aggression failed to sow discord and deepen divisions among Europeans. Rather, it backfired: the invasion seems to have brought Europeans closer together, by rallying them around our common European flag and converging more closely on our shared European values. Let us hope that the other objectives of Russia’s dictator will fail too.