The Russian War in Ukraine and the Visegrád Group: What Impact on Energy Transformation?

Data

All analyses presented in this paper are based on the Eurobarometer survey (standard EB wave edition 98.2 from 2023) conducted on either pure probability or mixed probability and nonprobability samples taken from the population of fifteen years and over residents in each of the countries. Various fieldwork agencies carried out the national surveys with the implementation of different modes of data collection (including computer-assisted personal interviews and self-administered Web-based questionnaires) for the Directorate-General Communication of the European Commission and the European Parliament between 12 January and 6 February 2023. Note that the EB 98.2 was conducted in all twenty-seven EU member states and twelve countries outside the EU; however, we focused only on the V4 group. The overall sample sizes for V4 countries were as follows: Czechia (n = 1,025), Hungary (n = 1,040), Poland (n = 1,014), and Slovakia (n = 1,004). Note that the EB dataset, national questionnaires, and other documents are publicly available through the GESIS Leibnitz Institute for the Social Sciences. ¹⁰

Outcome Variables and Analytical Procedures

Our study aims to explain how the Russian invasion of Ukraine affects public opinion about EU plans to end its reliance on Russian fossil fuels and invest in renewable energy sources. We used three EB questionnaire items for defining three outcomes (i.e. dependent variables) focusing on the different aspects of public attitudes of EU citizens toward the dependency of European countries on Russian energy sources, in particular:

Note that all three items were measured in an agree–disagree format, with four response options explicitly given to the respondents, namely, “totally agree,” “tend to agree,” “tend to disagree,” “totally disagree,” and with a “do not know now” answer not available among the valid response options but noted when the respondent indicated it (we treated “do not know” answer as a missing response). For the descriptive and regression analysis, we dichotomized response options in all items by contrasting “agree” respondents, that is, those who “totally” and “tend to” agree, with their “disagree” counterparts, that is, with those who “totally” and “tend to” disagree.

As we were particularly interested in how citizens of V4 countries differ in their attitudes toward EU plans to end its reliance on Russian fossil fuels (see the description of outcome variables above), we ran a series of Bayesian multilevel (with respondents nested within countries) binary logistic regressions to predict how covariates (as presented below), as well as sociodemographic characteristics of respondents (included as control variables), impact the likelihood that respondents’ will agree (with disagree as an opposite event) with the statements related with three outcome variables. We used the same covariates/explanatory and control variables for each outcome.

Our motivation for utilizing Bayesian multilevel logistic regressions (using the “brms” package in R) was to address the concerns of the assumptions of the standard logistic regression models and adequately account for the hierarchical structure of the data. ¹¹ Our approach allowed random intercepts and random slopes by the country for all covariates to capture potential geographic-level heterogeneity in the associations between covariates and outcomes; the abovementioned specification of regressions helps ensure the independence of errors across observations within the same country. By modeling the logit of the outcome and including relevant fixed effects, we explicitly checked assumptions such as linearity in the logit (via exploratory analyses and posterior predictive checks). We additionally checked for multicollinearity through variance inflation factors (we did not detect any disturbances with multicollinearity; see the appendix). In addition, the Bayesian framework permitted the regularization of estimates, which is especially beneficial if any variables are moderately correlated. Our final specification used four Markov Chain Monte Carlo sampling for robust inference on model parameters and predictive checks to evaluate the fit and assumptions of the model.

The set of explanatory variables included respondents’ opinions about the Russian invasion of Ukraine as a threat to the security of the (1) EU and (2) their own country, as well as the assessment of the financial consequences of the war in Ukraine on (3) personal life and (4) the national economy. All four items were measured in an agree–disagree format; we dichotomized their values before including them in the regression by coding 1 whenever respondents “totally agree” or “tend to agree” and, respectively, coding 0 when “tend to disagree” or “totally disagree.” Besides, we also incorporated the political orientation of the respondent (i.e., self-placement on a left-right scale) as a covariate. Political orientation was measured by the question requesting respondents to place themselves on a ten-point quasi-continuous scale ranging from 1 = on the left to 10 = on the right; we standardized their values across all the samples before including the variable in the regression.

When it comes to control variables, we used several sociodemographic respondent characteristics. Gender was indicated as 0 (men) and 1 (women), while age was grouped in four intervals: fifteen to twenty-four (set up as a reference category), twenty-five to thirty-nine, forty to fifty-four, and fifty-five years and older. We also incorporated a variable that requested respondents to evaluate their financial situation by asking whether, during the last twelve months, they had difficulties in paying bills at the end of the month, with three response options: “most of the time,” “from time to time,” and “almost never or never.” Note that for regression analysis, we merged those who perceived difficulties most of the time or from time to time and contrasted them with those who indicated they almost never or never had any difficulties (reference category). Finally, the question about the domicile type allowed us to characterize the respondents’ places of living: rural areas (reference category), small towns, and large towns.

In the descriptive part of our study, we decided to compare the distribution of “agree” and “disagree” answers (or a fraction of “agree” responses) for all items separately in four V4 countries and contrast them with the overall distribution (or an overall fraction, respectively) estimated for all twenty-seven EU member states. We used the population size weight combined with poststratification weights, which allows us to ensure that each country is represented proportionately and reproduce the population distribution of some sociodemographic characteristics in a sample.

Descriptive Analysis

The interdependence of European economies, including those in CEE, on Russian hydrocarbons is a consequence of the Kremlin’s longstanding policy of building a specific sphere of influence. ¹² However, the war rapidly and drastically reconfigured this situation, as observed by McWilliams, Sgaravatti, Tagliapietra, and Zachmann: “the scenario of a complete cut-off of all Russian fossil fuels to the European Union was unthinkable until early 2022, and this piece therefore contributes to an area of literature where analysis is urgently needed.” ¹³

The example of Germany is frequently cited because it represents the largest economy in the EU, and was until recently a close collaborator with Russia (e.g., Nord Stream 2 gas pipeline) and benefited from preferential prices of resources imported from there. ¹⁴ However, we will focus on the V4 countries, which, since the early 1990s, have pursued various strategies concerning Russia’s use of fossil fuels in international relations. Note that as of 2017, the dependence on natural gas imports in former socialist regime countries was significant (Table 1), even though the largest share of this fossil fuel appeared in the energy mixes of Hungary and Poland.

OPEN IN VIEWER

Table 1

Import Dependence of Natural Gas (2017)

Country	Czechia	Hungary	Poland	Slovakia
Dependence percentage a	102%	96%	78%	106%

a

The numbers present import rate as a share of natural gas consumption so it can rise above 100 percent

Source: European Commission, “Eurobarometer 98.2,” cited in J. Dyduch and A. Skorek, “Go South! Southern Dimension of the V4 States’ Energy Policy Strategies—an Assessment of Viability and Prospects,” Energy Policy 140 (2020): 111372, doi:10.1016/j.enpol.2020.111372.

Poland has no nuclear power plant, whereas the other three countries have such plants. At the same time, the national dependence on imports of Russian fossil fuels is highest in Slovakia and Hungary, while significantly lower in the Czech Republic and Poland. ¹⁵

Considering the above, let us analyze the dependent variables, that is, the questions concerning energy policy and its transformation in the context of Russia’s attack on Ukraine. When asked whether the EU should reduce its dependency on Russian energy sources as quickly as possible, the highest percentage of Poles (92%), Hungarians (84.1%), and Czechs (76.4%) supported this demand. The least positive responses were given in Slovakia (55.6%), with over a third of the population holding opposing views (Figure 2).

OPEN IN VIEWER

Figure 2

Questions about energy imports from Russia and the future of renewables

The remaining two questions addressed the crucial attitudes of citizens of the V4 countries toward energy transformation. The first asked for preferences for overall energy security in reducing oil and gas imports and investing in renewable energy sources. This question divided the Visegrád group countries into two camps. The first, comprising Poland (89.9%) and Hungary (88.4%), strongly advocated limiting fossil fuel imports and investing in renewable energy sources. The second was more moderate, although a majority of the Czech (68.2%) and Slovak (65%) populations supported this direction of ensuring energy security.

The last question aimed to assess whether respondents supported the EU’s policy of massive investment in renewable energy. Like the previous question, Poles and Hungarians most frequently opted for this choice (91.5% and 89.2%, respectively), while Czechs and Slovaks did so to a lesser extent (75.7% and 77.5%, respectively). Interestingly, the share of primary production by energy sources in the V4 countries in 2021 stood at 33.8 percent in Slovakia, 32.2 percent in Hungary, 23 percent in the Czech Republic, and 21.3 percent in Poland. ¹⁶ Considering Poland’s predominant shares of solid fuels (71.5%) and natural gas (5.6%), these social attitudes favoring energy transformation reflect a push toward decarbonization.

Bayesian Regression Results

Table 2 presents fixed effects (mean from the posterior distribution of the regression parameters) derived from the Bayesian logistic models to predict whether respondents within V4 countries agree with three different statements concerning energy sources (outcome variables as defined in Data and Methods section), namely, (1) reducing EU dependency on Russian sources of energy as soon as possible (Model 1), (2) massively investing by EU in renewable energies (Model 2), and (3) reducing imports of oil and gas and invest in renewable energy is an essential component of country overall security (Model 3) with covariates and control variables introduced previously.

OPEN IN VIEWER

Table 2

Summary of Fixed Effects Estimated by Bayesian Logistic Regressions

Predictors	Model 1		Model 2		Model 3
	Odds ratios	CI (95%)	Odds ratios	CI (95%)	Odds ratios	CI (95%)
Intercept	0.97	[0.19, 3.39]	1.08	[0.27, 3.32]	1.02	[0.21, 3.67]
War has consequences for personal finances [Yes = 1]	0.73	[0.25, 1.70]	1.01	[0.35, 2.24]	0.98	[0.50, 1.74]
War has economic consequences for the country [Yes = 1]	1.84	[0.35, 6.44]	1.95	[0.65, 4.93]	1.90	[0.44, 6.86]
War impacts a country’s security [Yes = 1]	3.37	[1.31, 7.53]	3.96	[0.68, 13.74]	3.64	[1.96, 6.48]
War impacts EU security [Yes = 1]	3.36	[1.88, 5.79]	2.25	[0.60, 6.19]	2.82	[1.29, 5.93]
Left-right political orientation	2.46	[0.03, 396.84]	1.04	[0.11, 13.98	0.81	[0.03, 28.74]
Gender [Woman = 1]	1.00	[0.81, 1.23]	1.00	[0.80, 1.23]	1.10	[0.90, 1.34]
Age: 25–39 [vs. 15–24]	1.06	[0.70, 1.57]	1.05	[0.69, 1.59]	0.59	[0.39, 0.88]
Age: 40–54 [vs. 15–24]	0.77	[0.51, 1.13]	0.99	[0.65, 1.47]	0.52	[0.34, 0.79]
Age: 55+ [vs. 15–24]	0.71	[0.48, 1.03]	0.81	[0.55, 1.19]	0.59	[0.39, 0.85]
Domicile: small town [vs rural areas]	0.95	[0.75, 1.22]	0.89	[0.68, 1.15]	0.77	[0.60, 0.98]
Domicile: large town [vs rural areas]	1.16	[0.87, 1.54]	0.84	[0.62, 1.12]	0.72	[0.54, 0.94]
Difficulties in paying bills: From time to time or most of the time [vs. never]	0.76	[0.61, 0.95]	0.71	[0.57, 0.90]	0.70	[0.56, 0.87]
ICC	0.64		0.59		0.59
Observations	3,556		3,572		3,545
Marginal R2 / Conditional R2	0.206 / 0.292		0.151 / 0.168		0.190 / 0.229

Note: CI means credible interval for the posterior distribution of the parameter with a 95 percent probability. Model 1 for predicting outcome variable: EU should reduce dependence on Russian energy sources as soon as possible [Yes = 1]. Model 2 for predicting outcome variable: EU should invest massively in renewable energy [Yes = 1]; Model 3 for predicting outcome variable: Reducing oil and gas imports and investing in renewable energy is essential to the country’s overall security [Yes = 1]. EU = European Union.

Model 1, which focuses on the support for proposing that the EU reduce its dependence on Russian energy sources as soon as possible, highlights some notable patterns in the fixed effects. Most prominently, perceptions that “War impacts a country’s security” and “War impacts EU security” exhibit relatively large posterior odds ratios (3.37 and 3.36, respectively), suggesting that individuals who recognize these security dimensions of war tend to be more likely to support rapid reductions in Russian energy dependence. While other covariates, such as the statement “War has consequences for personal finances,” do not attain notable magnitudes or exhibit substantial credible intervals that intersect with 1, it is noteworthy that economic concerns may potentially influence respondents’ attitudes despite their less pronounced effect. The demographic covariates—age, gender, and domicile—reveal modest or negligible contributions, with ages twenty-five to thirty-nine slightly inclined (odds ratio [OR] = 1.06) toward favoring reduced energy dependence, while older groups tend to lean marginally away from this position. Similarly, individuals encountering challenges in meeting their financial obligations exhibited a marginal decrease in their propensity to advocate for prompt reductions in Russian energy imports (OR = 0.76). However, the credible interval encompasses values approaching 1, underscoring the necessity for careful interpretation.

Turning to Model 2, which addresses whether the EU should invest significantly in renewable energy, many of the same predictors recur. However, their relative importance undergoes a shift. The impact of war on security remains positively associated with the endorsement of substantial renewable investments (OR = 3.96), albeit with a 95 percent credible interval (0.68–13.74), indicating uncertainty around the precise effect magnitude. This pattern suggests that individuals who strongly believe that war affects a country’s security are also more inclined to see value in robust clean energy initiatives. The perceived financial and country-level economic consequences of war generally hover around a posterior odds ratio near or slightly above 1, indicating no strong effect on opinions about renewable energy. Of particular interest is the Bayes odds ratio of less than 1 for difficulties in meeting financial obligations and support for renewable energy, which indicates the potential for concerns regarding the perceived short-term costs of transitioning to renewables to influence opinions on this matter. Meanwhile, the left-right political orientation (OR = 1.04) demonstrates a near-neutral relationship. Demographic variables yield relatively modest effects again, suggesting that attitudes toward large-scale renewable initiatives might be shaped more by broader security and economic discourses than basic socio-demographic factors.

In Model 3, which focuses on whether reducing oil and gas imports and investing in renewable energy is an essential element of national security, the patterns highlight the significance of security perceptions even more definitively. The findings demonstrate that both the statements, “War impacts a country’s security” and “War impacts EU security,” yield posterior odds ratios above 2.80, with lower bounds well above 1 (1.96 and 1.29, respectively). This finding reinforces the notion that the recognition of the security implications of war fosters support for energy diversification strategies as part of a broader security framework. Of particular interest is the observation that age emerges as a more discernible factor in this model: the two middle-aged groups (fifteen to thirty-nine and forty to fifty-four) and the senior group (fifty five and above) are all less likely than younger adults (fifteen to twenty-four) to deem these energy measures essential, although the effect is most pronounced for those aged forty to fifty-four (OR = 0.52) and twenty-five to thirty-nine (OR = 0.59). It may be that younger cohorts, potentially facing the long-term implications of energy and climate policies, are more convinced of the strategic imperative of transitioning away from fossil fuels. Furthermore, residing in more urbanized areas, particularly large towns, is associated with a decreased perception of these measures as integral to national security (OR = 0.72 for large towns). Finally, consistent with the other models, encountering difficulties in meeting financial obligations has been found to exert a slight negative influence (OR = 0.70), potentially reflecting apprehension about the immediate financial burdens of an energy transformation. These results emphasize that while demographic characteristics play a limited role, a strong sense of security-related and macro-level economic concerns is key in shaping preferences for a shift in energy policy.

Figures 3 to 5 summarize the variation in the association (random effects) for every covariate in four countries. We plotted the posterior predicted probabilities on the y-axis and the associations of the specific values of each covariate with predicted probability on the x-axis. The dot in the first four facets represents the posterior mean probability for a particular variable with error bars corresponding to 95 percent credible intervals. Regarding the posterior predicted probabilities for the (quasi-continuous) left-right political orientation, the curve for each country visualizes how the probability changes alongside the respondents’ position on a scale measuring political orientation.

OPEN IN VIEWER

Figure 3

Summary of regression results for predicting the likelihood of agreeing that the EU should reduce dependence on Russian energy sources as soon as possible

OPEN IN VIEWER

Figure 4

Summary of regression results for predicting the likelihood of agreeing that the EU should invest massively in renewable energy

OPEN IN VIEWER

Figure 5

Summary of regression results predicting the likelihood of agreeing that reducing oil and gas imports and investing in renewable energy is an essential component of the country’s overall security

When examining each of the covariates presented in the left panels of Figure 2 (i.e., personal finances, economic consequences of war for the country, national security, and EU security), it becomes evident that clear cross-country patterns emerge in Model 1’s predicted/posterior probabilities of supporting an immediate reduction in Russian energy dependence. Initially, personal finances demonstrate higher levels of support among respondents who believe the war will financially impact them, although the baseline levels vary notably. Poland is distinguished by its high support levels irrespective of personal financial risks perceived by Poles, while Slovaks demonstrate a notable decline in support even among those who anticipate such risks. Czechia and Hungary are somewhere between, with Czechs exhibiting a closer alignment with Polish levels and Hungarians showing a greater proximity to Slovaks. A similar hierarchy is observed under the category of national security: Poles demonstrate consistent support across the board, while Slovaks exhibit the most hesitation. However, in this case, Slovakia’s divergence between “No” and “Yes” is more pronounced and in a different direction, indicating a more substantial decrease in support when they perceive economic impacts. Poland maintains the highest posterior probabilities in national security and EU security domains, Slovakia the lowest, and Czechia and Hungary are positioned in the middle. However, the shift from “No” to “Yes” is generally substantial for all V4 countries, with the perception of security threats exerting an upward influence on each nation.

Turning to left-right political orientation, Poland and Hungary exhibit a broadly similar pattern, with left-leaning individuals tending to be somewhat more enthusiastic about an immediate termination of Russian energy imports. In contrast, right-leaning individuals exhibit slightly lower but still relatively high probabilities. Czechia demonstrates a smaller, slightly declining slope from left to right, yet overall remains at comparatively high levels of support across the ideological spectrum. In contrast, Slovakia exhibits a particularly pronounced reverse slope, with predicted probabilities that are noticeably lower at the left end and rise sharply among respondents with more right-leaning political orientations. These results suggest that perceptions of war-related economic or security consequences can enhance support for reducing Russian energy dependence in all V4 countries. However, the extent to which one’s ideological beliefs interact with these perceptions is notably impacted by the specific national context.

In Model 2, which predicts support for substantial EU investment in renewables (see Figure 4), Poland demonstrates near-ceiling levels in almost every “war consequence” scenario, with Poles exhibiting overwhelming endorsement irrespective of their perception of personal or national-level harm from the war. Hungary and Slovakia exhibit slightly lower levels but frequently exceed .9 in predicted posterior support, exhibiting only minor fluctuations. Conversely, Czechia demonstrates the lowest baseline whenever respondents do not perceive the war as having economic or security impacts. However, when Czechs perceive such consequences, they rapidly catch up, displaying similarly high posterior probabilities around .9. In summary, for Czechia, concerns about economic or security threats from the war act as a catalyst, aligning their stance with that of the other V4 countries’ strong pro-renewables position.

Turning to the left-right dimension, it is notable that all four countries exhibit substantial majorities in favor of renewable energy investment, irrespective of their ideological positioning. However, some cross-national variations emerged. In the case of Czechia, the posterior probabilities begin at a somewhat lower level on the left but steadily increase among those with right-leaning views, resulting in an evident upward trend. Conversely, Poland and Slovakia converge around the upper end of the scale, displaying only modest declines from left to right, suggesting widespread, near-universal endorsement. Hungary maintains above .9 and exhibits a slight downward trend, yet continues to closely align with the high approval levels observed in Poland and Slovakia. The findings indicate a prevalent inclination among the V4 states to advocate for substantial EU renewables investment, though Czechia’s support is contingent on perceptions of war-related threats.

In Model 3, which predicts whether people agree that “reducing oil and gas imports and investing in renewable energy is essential to their country’s overall security,” notable cross-country patterns emerge when examining perceptions of the war’s consequences. In terms of personal finances, Hungary and Poland demonstrate consistently high posterior probabilities (well above .9) for respondents who answer “No” or “Yes.” At the same time, Czechia remains in the mid-1980s, and Slovakia hovers around .8 for “No.” Notably, Czechia and Slovakia exhibit a substantial upward shift, reaching the .85 to .9 range when they perceive personal financial risks. In terms of economic consequences, Hungary and Poland exhibit elevated support when perceiving an absence of economic impacts, subsequently experiencing a modest decline in responses when acknowledging the presence of such impacts. Conversely, Czechia and Slovakia demonstrate a contrasting trend, with Slovakia demonstrating a substantial increase upon recognizing the economic consequences of the war. A similar pattern is evident in national and EU security domains. Hungary and Poland maintain a probability over .9 when respondents perceive no security threat; however, they exhibit a modest decline when individuals recognize a threat. In contrast, Czechia and Slovakia, initially positioned lower when respondents do not perceive a security threat, ascend to higher levels in the “Yes” scenario.

Regarding left-right political orientation, a slight downward trend is evident in Hungary and Poland, suggesting that individuals on the right are marginally less inclined to perceive reducing fossil fuel imports and investing in renewables as imperative for national security compared with those on the left. However, both countries score well above .85 across the ideological spectrum. In contrast, Czechia and Slovakia demonstrate an upward trend, with predicted probabilities commencing lower on the left but increasing steadily on the right, approaching the levels observed in Hungary and Poland. Overall, Model 3 indicates that perceptions of war-related threats can influence Czech and Slovak respondents, causing them to align with the otherwise high baseline support seen in Hungary and Poland. However, ideology also strongly influences these attitudes across the V4 national contexts.

To summarize the results of all Bayesian logistic regression models, Poland consistently shows the highest support across all three outcomes, with posterior probabilities close to or at the ceiling—particularly among those who perceive the war as having personal, economic, or security consequences. On the contrary, Slovakia tends to be at the lower end of the range. However, the country shows gains once respondents perceive war-related threats, suggesting that perceptions of war can notably increase Slovak support. Czechia also starts from a relatively modest base but increases when respondents believe the war affects finances or security, converging on higher levels of support. Hungary is generally close to the top but shows more muted differences between those who perceive an impact of the war and those who do not. Regarding political ideology, Polish and Hungarian respondents often show a slight downward slope from left to right—indicating slightly stronger support on the left—while Czech and Slovak respondents tend to move up the ideological axis, with those on the right expressing higher support.