1. Introduction

Just energy transitions are currently leading the political focus worldwide. The concept, rooted in the claims of the Oil, Chemical and Atomic Workers International Union (OCAW) in the 1970s, has re-emerged to face social and environmental deterioration, particularly boosted by the Sustainable Development Goals (SDGs). In roughly five decades, the idea has transcended context-specific unionism to gain international momentum in the mainstream discourse.

As a normative notion, it admits different approaches, as observed in political programmes, current unionist claims, statements of Non-Governmental Organisations (NGOs), and opinions Just Transition Research Collaborative, 2018. The approaches to the just energy transition are the result of different perceptions and values, hence diverse, complex, and evolving.

In the scholarly literature and technical reports, most studies perform socioeconomic and energy indicators analyses and/or develop models to test the social and environmental effects of policies and scenarios García-García et al., 2020. However, recent insights point to the role of perceptions in the process, i.e., the level of awareness of citizens, their beliefs and opinions, and their interpretation of the information that they receive. This issue is currently increasing its relevance in the general literature about the transition to sustainability in general, being proof of this general interest Macht et al. (2022), Morgunova & Shaton (2022), Panarello & Gatto (2022), and Thomas et al. (2022). However, it reaches a special and meaningful relevance in the range of just energy transitions, as they combine environmental awareness, energy policy and management, and socioeconomic impacts and social policies.

Precisely regarding just energy transitions, hence with more restricted publications, literature warns that perceptions can diverge significantly from the information reflected by indicators, potentiate, or hinder the evolution of the energy transition, and be influenced by stakeholders to determine the results Gawel et al., 2015; Gölz & Wedderhoff, 2018; Groh & Ziegler, 2018; Kuschan et al., 2022. Considering their potential, as observed in fields like Sustainable Welfare, the presence of eco-social perceptions in the population is key to fostering a social and environmental synergy Fritz & Koch, 2019; Koch & Fritz, 2014; Otto & Gugushvili, 2020, and subsequently a just transition.

Theoretically, four major approaches to the normative nature of just energy transitions have been described, based on the preference for individualism or collectivism, and economic growth or post-growth. Notwithstanding, the reality could be more complex and richer.

This paper suggests testing the current perceptions of European citizens facing the just energy transition towards sustainable sources to reach decarbonised economies and determining whether such perceptions match the described approaches or not. Section 2 contextualises the topic under analysis and presents the hypothetical classification of approaches. Section 3 discloses the data sources and methodology that this contribution proposes to test such a classification. The results are presented in Section 4. Finally, Section 5 draws the main conclusions.

2. Contextualisation

The seed of the present idea of just energy transition is the claim for adequate conditions in the American oil refining sector in the context of the oil crises in the 1970s Just Transition Research Collaborative, 2018. Originally, it gathered elements of the social environmentalist movement and the demands for occupational health and safety Schlosberg & Collins, 2014).

In the last decades, the idea was adapted to other unionist sectors, overflowed the American context, and was adopted internationally. Recently, it has been updated in the framework of the SDGs by the International Labour Organisation (ILO) as a response to the social and environmental weakening that characterises the 21st century ILO, 2018; Poschen, 2017.

The leap between unionism and mainstream discourses worldwide has motivated the emergence of a diversity of users of this concept. Its normative nature admits multiple approaches to what a just energy transition should be. The Just Transition Research Collaborative (2018) proposes a classification that has adequately served to categorise the discourses of governments, international organisations, NGOs, and activist movements, inter alia. The classification implicitly employs two dimensions: the preference for individualism versus collectivism, and the preference for economic growth versus post-growth. As a result, four major approaches emerge (Figure 1):

Figure 1.Major approaches to the just energy transition Source: Own elaboration.

This categorisation has served to detect that international organisations adopt managerial or structural approaches, and unionist and activist organisations tend to the transformative paradigm Just Transition Research Collaborative, 2018. Moving from the consensus of a group of individuals under the form of governments, organisations, or movements, to the perceptions and attitudes of individuals, there is a gap. Do individual perceptions match the major approaches? Can additional nuances be detected?

3. Data and methodology

To analyse the perceptions and attitudes of individuals in Europe, this paper takes the most recent data from the European Social Survey (ESS) European Social Survey European Research Infrastructure (ESS ERIC), 2022a. The ESS is a biennial cross-national survey that statistically represents individuals aged 15 and over. The process to build its database ensures comparability and compliance with research ethics International Statistical Institute, 2010. The process begins with a call for question module design teams to conform the source questionnaires that are translated into the national languages of the countries that participate in the edition. The design of the sample grants comparability in the process of data collection and monitors the national situation of the country where the data collection is being conducted. Finally, data are processed, archived, and made publicly available after succeeding in a data quality assessment. Further details about these procedures can be consulted in the methodological section of the official webpage European Social Survey European Research Infrastructure (ESS ERIC), 2022b.

The first questionnaire, published in June 2022, was conducted from September 2020 to January 2022 in 10 European countries. It addresses multiple topics. Given the goal of this research, this analysis restricts its focus to the questions that reveal the attitudes and perceptions of citizens regarding the just energy transition, i.e., perceptions and values regarding environmental awareness, institutional adequacy, social justice, openness to receive information, and sociodemographic traits. Table 1 presents the variables, summarised description, and valid ranges. Variables admit that the interviewees do not respond or manifest their lack of knowledge about the issue of the question. In such cases, an exceptional value is indicated in the database. Those special values ought to be consulted in the ESS database.

Once the data source and the variables have been disclosed, the methodological procedures are carried out as follows.

First, a general description is made based on basic statistical calculations. This analysis mainly focuses on the statistical mode and standard deviation of each variable. The mere employment of the mean values for description is not adequate due to the presence of the cited exceptional values.

Second, provided that four major approaches have been proposed to classify the views about the just energy transition, this analysis suggests the execution of a clustering algorithm that fits observations into four groups.

Clustering is a technique that calculates the most internally homogenous groups in a sample of diverse individual observations to provide coherent taxonomies based on direct evidence. Clustering algorithms are wide-ranging, yet four typologies can be distinguished. 1) Centroid-based algorithms classify observations based on a user-defined centre. 2) Density-based algorithms generate groups by connecting parts of the sample that present a high concentration of observations. 3) Distribution-based clustering adjusts observations into groups by assigning a given theoretical distribution. 4) Hierarchical algorithms provide a classification of observations at different distances that is consistent in reiterated executions.

This methodology suggests the employment of a centroid-based algorithm. Nevertheless, instead of defining the centres of the clusters a priori, the centres are calculated by the algorithm after inputting that the desired number of groups is four through a k-means typology. Consequently, the interpretation of the groups serves to identify whether clusters align with the previously described approaches or not and the reasons for a potential divergence, by analysing the centre of each cluster and the distribution of individuals among them.

Third, calculating an optimal panoply of clusters that is strictly based on data to derive an adequate categorisation of perceptions. To do so, Ward’s hierarchical clustering algorithm in squared Euclidean distances is prescriptive, provided that this typology calculates a panoply of clusters at different distances. Given the typology of data, with different scales and some open-range variables in different units (time, age, etc.), inputs are standardised in z-scores.

To determine the number of clusters, the user normally selects a cutting point distance according to the analytical needs. Instead, this methodology suggests the application of Thorndike’s criterion of optimality Thorndike, 1953, which determines the optimal number of clusters according to data, hence replacing the arbitrary selection of the user. According to the criterion, the number of clusters that achieves the greatest reduction in distances between clusters is optimal.

Fourth, observing the relations between variables to derive further insights about the functioning of these perceptions and attitudes in the context of just energy transition. To observe such relations, this paper proposes performing a Multiple Component Analysis (MCA). MCA is a statistical multivariant technique that aims at summarising many variables of a nominal nature into a reduced number of dimensions and factors. Consequently, it condenses data into a more manageable set of information and, ultimately, points to the meaningful dimensions and factors that define the methodological selection, as well as the relations between them.

4. Results

According to the analytical goals, the results are structured into four subsections. Section 4.1 draws the observed perceptions and attitudes in the survey through basic statistics. Section 4.2 tests the suitability of a four-group classification according to that proposed in general terms by the Just Transition Research Collaborative (2018), as introduced in Section 2. Section 4.3 calculates the optimal number of groups in a potential classification based on data through Thorndike’s criterion of optimality. Finally, Section 4.4 introduces the explanatory capacity of the variables under study and their relations based on the MCA technique.

4.4. Relations between the variables

To visualise the results of the MCA, Figure 2 presents the discriminant measures of the reduction in two dimensions. Computation has obtained Cronbach’s Alphas equal to 0.992 in Dimension 1 (inertia 0.223) and 0.989 in Dimension 2 (inertia 0.186).

Figure 2.Discriminant measures of the variables in two dimensions. 

MCA indicates that the variables related to human values have the greatest explanatory capacities, as proven by higher measures in both dimensions and, visually, greater distance between these variables and the coordinate origin in Figure 2. These remarkable variables are, in descendant order based on average measures for both dimensions, caring for nature and looking after the environment (impenv, 0.530), following traditions (imptrad, 0.527), helping the people around and caring for their well-being (iphlppl, 0.524), ensuring governmental-led safety and a strong state to defend citizens (ipstrgv, 0.517), thinking up new ideas and being creative (ipcrtiv, 0.514), listening to people who are different even in disagreement (ipudrst, 0.499), treating every person equally and providing equal opportunities in life (ipeqopt, 0.496), making own decisions and liking freedom and independence (impfree, 0.481), and having fun and pleasure (impfun, 0.480).

The lowest explanatory capacities are located closer to the coordinate origin in Figure 2 and correspond to gender (gndr, 0.000), the consideration about energy frugality to reduce climate change (testic37, 0.007), the belief in people taking advantage or being fair (pplfair, 0.015), the presence of children in the house (chldhhe, 0.017), the worry of climate change (wrclmch, 0.019), and the beliefs about the cause of climate change (ccnthum, 0.019).

The remaining variables oscillate from 0.026 (ccrdprs, i.e., feeling a personal responsibility to reduce climate change) to 0.266 (cttresa, i.e., importance of equal treatment in the courts).

5. Conclusions

Just energy transitions are congregating worldwide attention both politically and academically, particularly driven by the SDGs. Notwithstanding, the notion is not new. It can be traced to North American unionism during the oil crises in the 1970s. After some decades, the idea has transcended industry-specific unionism and is applied by multiple agents. The transcendence has motivated the emergence of different conceptions of the just energy transition as a normative goal to tackle social and environmental deterioration.

Four major approaches have been suggested from a theoretical viewpoint. These approaches are the statu quo, managerial, structural, and transformative conceptions. Implicitly the configuration of approaches is bi-dimensional, as it depends on two main ideas. On the one hand, the preference for individualism versus collectivism. On the other hand, the suitability of green growth strategies versus prosperity without growth strategies. The four-group classification has been useful to determine the position of international organisations, NGOs, unions, and activist moments, inter alia.

This paper proposes testing the four-group classification of approaches regarding the individual attitudes and perceptions of European citizens. The idea is to determine if such classification, which is useful at an aggregate corporative level, is valid in the downscaling towards the individual. To do so, the methodological proposal acts in four steps based on the most recent version of the European Social Survey (2020-2022). As a first step, it obtains basic statistics to draw the situation of the sample. Afterwards, it computes the traits of an empirical four-group classification through a k-means clustering algorithm. Subsequently, it determines the optimal number of empirical approaches through Ward’s hierarchical algorithm supported by the criterion of optimality suggested by Thorndike. Finally, it discloses the relations among variables through a Multiple Correspondence Analysis.

Basic statistics determine both clear perceptions and attitudes and a bias for medium values in certain variables. Citizens tend to follow current affairs daily through conventional media, do not attend demonstrations and volunteering activities, support governmental intervention to alleviate inequality, consider themselves notably happy despite the pandemic context, have experience with children and live in country environments. Likewise, they mostly identify as non-religious people, supporters of prevailing the views of ordinary people against elites, and strong supporters of democratic values. They additionally tend to have little experience with unionist movements. In contrast, citizens tend to answer medium values in questions related to trust in institutions, their positioning in the political spectrum, the functioning of climate change, and the identification with human values.

The test of the four-group classification offers mixed results. Even if some clusters resemble the theoretical classification of perceptions, the alignment between empirical results and theoretical postulates is far from adequate. The cause of the mismatch is the difficulty to determine the difference between green growth and post-growth ideas, as citizens do not show conclusive perceptions in this regard. In contrast, the confrontation between individualism and collectivism is clearer in the sample and is responsible for drawing some similarity between the theory and the results in absence of precise environmental positionings. Moreover, the distribution of individuals in these four clusters is significantly concentrated. One of the clusters gathers nearly 69% of individuals, while a second group congregates approximately 29%. A marginal cluster with inconclusive traits closes the classification.

Provided that the four-group classification is not adequate, this paper has calculated the optimal classification via the hierarchical algorithm under optimality and found 23 clusters. As far as this wide panoply of groups is concerned, the most significant insight is, again, the unequal distribution of observations between them. Three main clusters congregate 91.63% of respondents. The remaining clusters oscillate between 0.01% and 3.4% of respondents. Finding such marginal groups under a methodology based on optimality evinces the relevance that marginal clusters could have for social dynamics and political decision-making.

Considering the underlying structure of the database used to form the clusters, MCA indicates that human values have a higher explanatory capacity than sociodemographic and political variables. Caring for nature, following traditions, helping people, supporting governmental safety, creativity, considering other people’s opinions, egalitarianism, independence, and fun/pleasure-seeking tendencies are more relevant than gender, the positioning in the political spectrum, the presence of children in the house or the beliefs about climate change and individual action to tackle it, among others.

These conclusions serve to further the knowledge about the role of perceptions and attitudes over the ongoing process of just energy transition and point to two lines of research and political action. On the one hand, the need to consider the marginal profiles identified in Table 4 and study their interaction with the participation processes and policies. How can marginal groups be included to define the just energy transition? On the other hand, the necessity to focus on the explanatory capacity of human values to determine the dynamics that led to the suggested panoply of groups. How can participation processes and policies be more aware of human values and look beyond sociodemographic factors such as gender and political positionings?

While the social consequences and justice of the energy transition are mainly studied through modelling techniques and quantitative indicators, the subjectivity of individuals, shaped by their perceptions and attitudes, is of utmost importance and offers a niche of research for future works under the two lines here suggested.