The Energy Mix: Understanding People’s Diverging Energy Preferences in Belgium

Abstract

To mitigate global climate change, drastic transformations of the energy system are needed. Whereas the public is asked to adapt its energy demand patterns, their perspective is often neglected. In this study, we incorporated a more human-centered dimension into energy research by examining how social characteristics determine the composition of individuals’ preferred energy mix. Previous studies have been mainly limited to the exploration of preferences for one energy system in isolation. Hence, little is known about how various energy sources are combined into various energy mixes. Furthermore, empirical research regarding the heterogeneity of energy preferences often lacks an intersectional approach. Against this background, we used Belgian data from the European Social Survey (N = 1766) to examine the diversity of preferred energy mixes among individuals and how this relates to social characteristics. Specifically, a segmentation analysis was conducted to cluster Belgian respondents into intersectional, meaningful groups related to their preferred energy mixes. The results of the segmentation analysis underpin the existence of vulnerable and privileged groups in the establishment of a green transition. This study highlights the importance of focusing on energy mixes from an intersectional stance, as it provides an excellent tool to uncover the power dynamics underlying an energy transition.

1. Introduction

Climate change is recognized as one of the major challenges of our society, as it is currently affecting every inhabited region across the globe (Arias et al. 2021). The increasing recurrence of extreme weather events, rising temperatures, etc., have led to a growing consensus about severe environmental threats and the anthropogenic nature of climate change. The large-scale exploitation of natural resources due to excessive energy consumption is recognized as one of the main drivers behind climate change (Von Borgstede et al. 2013). Therefore, to mitigate global climate change, drastic transformations of the energy system are needed on both the supply and demand sides. Whereas the public is asked to adapt its energy demand patterns, their perspective is often neglected, both in research and policy. As argued by Sovacool et al. (2015), there is a need to incorporate a more human-centered dimension into energy research. Energy transitions are not only technological or economic phenomena: they are embedded into social structures of power, as they can foster new injustices or worsen existing inequalities by ignoring the realities and interests of vulnerable people (James et al. 2022; Sovacool et al. 2019b). Therefore, the recognition of the diversities and needs of different social groups is of fundamental importance to pursue a fair and successful energy transition (Walker and Day 2012).

In this research, we aim to enhance our understanding of the social aspects behind the energy transition by exploring individuals’ preferred energy mixes. Existing research about public energy preferences is mainly limited to the exploration of preferences for one energy system in isolation (e.g., nuclear energy or renewable energy technologies or fossil fuels). Hence, little is known about how various energy sources are combined into diverging energy mixes, i.e., the preferred combination of the different sources that are used to produce electricity. Furthermore, empirical research regarding the heterogeneity of energy preferences often lacks an intersectional approach, as it merely summarizes the effects of one or two traditional categories of difference (i.e., gender, class, race) (Kácha et al. 2022). While these studies have concluded that levels of climate change concerns, political orientation, and socio-demographic and socio-economic characteristics influence public attitudes toward energy systems, they rarely combine predictors, which reduces the complexity and nuances (Kácha et al. 2022; Perlaviciute and Steg 2014; Poortinga et al. 2006).

Belgium is an interesting case to investigate, as the country has one of the lowest shares of renewables in energy consumption across the EU (Eurostat 2020). At the time when the survey was conducted, Belgium’s energy mix consisted mainly of oil (51.5%), gas (22.6%) and nuclear energy (15.3%) (Ritchie et al. 2020). While Belgium endorses the EU’s target of carbon neutrality by 2050 and has adopted a long-term strategy for energy and the climate, the targets have not been clear, and the IEA recommends that Belgium update its strategy in order to make the commitment and the path toward it much clearer (IEA 2022). In addition, the country faces a set of major challenges, such as rising energy prices, an increasing number of households that live in energy poverty, a high demand for energy, a high dependence on energy imports, controversies about the planned nuclear phase-out, etc. These struggles make the question of how to account for a successful transition even more pressing.

Against this background, we used Belgian survey data from the European Social Survey (N = 1766) to examine the diversity of the preferred energy mixes among the Belgian population and the social characteristics associated with these energy mixes. A segmentation analysis was used as the primary analysis technique to investigate whether Belgian respondents could be divided into groups based on their preferred energy mixes. More specifically, a latent class analysis was conducted to cluster the Belgian respondents into intersectional, meaningful groups related to their preferred energy mixes. This allows us to compare the energy preferences of different audiences by looking at the influence of socio-demographic characteristics, such as educational attainment, income, climate change concerns and political orientation, from an intersectional approach. By applying a segmentation analysis, we can better understand the larger patterns of power behind energy transitions, which will provide a more holistic and nuanced framework on the social dimension of energy systems (Metag and Schäfer 2018).

In this article, we wish to contribute to the underexamined role of social sciences in energy research. More concretely, by bridging intersectionality with energy studies, we hope to contribute to the theoretical as well as the methodological debate on investigating energy transitions from a social scientific point of view. In what follows, we explore Belgium as a unique case to investigate energy preferences. Additionally, we use theoretical insights from intersectional studies, such as the white male effect, to highlight the underlying power structures of an energy transition. The methodological section describes the data collection and the analysis. The fourth section summarizes the results, and in the discussion, we apply the results to our conceptual framework and discuss the broader implications.

2. Case: Energy Mixes in Belgium

In this paper, we focus on preferences toward electricity portfolios in Belgium. An energy mix refers to the different sources that are used to produce electricity. In this research, we distinguished three different types of energy systems that can be used to produce electricity: renewable energy, fossil fuels and nuclear energy. Renewable energy is collected from natural sources or processes that are constantly replenished and includes sources such as solar power, wind power, hydroelectricity and biomass. Fossil fuels are finite resources, such as natural gas, coal and oil, and are known for their polluting characteristics. Nuclear energy is described as a separate category because of its complex relationship with climate change and the environment. On the one hand, nuclear energy does not release any greenhouse gasses and can therefore be considered a green energy system. On the other hand, it does have a major impact on the environment due to its waste and uranium mining, as well as the risk of major accidents and the potential use of nuclear energy for weapons or terrorist attacks (Bisconti 2018). Due to this complexity, nuclear energy can be categorized as neither a renewable energy source nor a fossil fuel.

Energy mixes vary greatly among times and countries. For a long time, traditional biomass was the main source of energy. During the industrial revolution, coal became an important energy source, alongside traditional biomass. From 1960 onwards, energy mixes became more diverse as nuclear energy and renewable forms of energy, such as wind and solar power, were added to the mix (Ritchie et al. 2020). While energy transitions were slow in the past, the current transition to a low-carbon society is occurring at a rapid pace and on a large scale.

The energy market in Belgium presents an interesting case due to several reasons. At the time when the survey was conducted, Belgium’s energy mix consisted mainly of oil (51.5%), gas (22.6%) and nuclear energy (15.3%) (Ritchie et al. 2020). Belgium has, in comparison to other European countries, one of the lowest shares of renewable energy consumption (13%) (Eurostat 2020). In the last several years, Belgium has faced major challenges in terms of providing a diversified, reliable energy mix. Belgium is highly dependent upon other countries for a reliable supply of fossil fuels, which often puts the country in a vulnerable position, depending on volatile geopolitical relations. The potential of renewable energy sources is relatively limited in Belgium due to its population density and widespread urban sprawl, which make it difficult to, for instance, integrate large-scale wind energy facilities into the landscape (Gusbin 2015). Additionally, there is an intense political debate regarding the composition of an ideal energy mix. Whereas all dominant political parties are more or less in favor of a renewable energy transition, there is a lot of discussion about the scope, pace and preferred technologies behind a green energy transition. Specifically, there is a great deal of controversy in Belgium about nuclear energy and its role in the current and future energy mixes. Right-wing political parties generally support nuclear energy because they portray it as a green energy source that provides reliable and safe energy. Left-wing parties, in contrast, oppose nuclear energy by highlighting its disadvantages, such as environmental degradation. In addition to nuclear energy, there is also a lot of indecisiveness regarding future investments in coal-fired power plants due to their negative impact on climate change. All these factors make it difficult for Belgium to offer a stable and diversified energy mix. This context makes it particularly interesting to investigate individuals’ preferences toward electricity portfolios (Gusbin 2015).

3. Heterogeneity of Energy Preferences: An Intersectional Lens

Existing research indicates that there exist widely varying public preferences for energy systems rooted in a diverse composition of conversion factors, such as socio-economic and socio-demographic characteristics, the level of climate change concern, values and political orientation (Perlaviciute and Steg 2014; Poortinga et al. 2006). However, most studies regarding the heterogeneity of energy preferences merely summarize the effects of one or two traditional categories of difference (i.e., gender, class, race), neglecting an intersectional approach. According to Davis (2008), intersectionality is ‘the interaction between gender, race and other categories of difference in individual lives, social practices, institutional arrangements, and cultural ideologies and the outcomes of these interactions in terms of power’ (p. 68). He argues that intersectionality can be used as a tool to study the structures of power. While not a lot of studies have applied an intersectional lens to energy research, it is much needed, as it provides more nuanced findings and creates the possibility of understanding the interrelation between social inequalities and energy consumption, which is necessary for the transition toward a sustainable future (Godfrey 2012; Sturgeon 2009). Moreover, by not recognizing how differences in social context shape people’s response to energy transitions, public policies and professional institutions may overrepresent elite narratives (Ford and Norgaard 2020).

Previous research has emphasized the role of inequality and power relations with regard to sustainable practices (Anantharaman 2018). Environmental values and sustainability practices tend to be associated with middle and higher classes in society (Carfagna et al. 2014). Indeed, studies show that preferences for green energy are most salient in highly educated and affluent groups. However, scholars also argue that a class bias can be found in dominant views on sustainability. Specifically, they are often rooted in high-cost practices such as green consumerism (Carfagna et al. 2014; Kennedy and Givens 2019). A similar argument can be made regarding elitist energy policies such as carbon taxes or feed-in tariffs. This is not surprising given that high-status individuals are often able to define sustainability ideals. Consequently, the capability of ‘being green’ can be dependent on access to certain economic or cultural resources (Kennedy and Givens 2019). In turn, because of a sense of powerlessness and alienation, low-status actors may not feel a sense of self-efficacy or an affinity with the environmental movement. This argument is especially relevant in the case of a green energy transition. In Flanders, many policy instruments are used to promote and subsidize private solar panels, for example. In sum, Kennedy and Givens (2019) argue for a more dialectical relation between sustainability preferences and practices. Low-status actors may adjust their (energy) preferences based on their (in)ability to participate in the ongoing energy transition.

Research regarding the white male effect also highlights how an intersectional analysis creates the ability to visualize underlying power structures. McCright and Dunlap (2011), for instance, developed the white male effect, which argues that white (North American) males often exhibit low environmental risk perceptions, resulting in environmental imaginaries of denial. In a later phase, they expanded the theory to the conservative white male effect, highlighting the intersections between gender, ethnicity and political orientation (McCright and Dunlap 2013). While the white male effect is limited to the risk perception of climate change, we could derive possible hypotheses regarding energy preferences from this theory. Previous studies have found a close link between climate change attitudes and energy preferences (Engels et al. 2013; Greenberg 2009; Verschoor et al. 2020). Renewable energy systems are perceived as one of the main solutions for the mitigation of climate change. Increasing public awareness about the negative impacts of climate change is therefore leading to higher preferences for renewable energy. Accordingly, individuals who are less concerned about climate change (i.e., conservative white men) often belief that cheap fossil fuels will contribute to economic development (Clarke et al. 2016). The link between masculinity and fossil fuels is defined as ‘petro-masculinity’. According to this perspective, fossil fuels are more than just a means to generate profit: they create specific identities through conservative symbols that represent masculinity, autonomy and self-sufficiency. Moreover, these male elites established a hegemony that granted them status and power to establish specific policies and public institutions (Daggett 2022). Based on insights from the conservative white male effect and petro-masculinity, we could hypothesize that white males with a more right-wing political ideology are less likely to prefer a sustainable energy mix.

However, the power of white elite men supporting fossil fuels has recently been threatened by the increasing public concerns about climate change, which resulted in the emergence of eco-modern masculinity (Daggett 2018, 2022; Dockstader and Bell 2020). Eco-modern masculinity is characterized by a combination of traditional masculine identities and concern about climate change (Dockstader and Bell 2020). This type of masculinity argues that protecting climate change does not conflict with the capitalist logic. Eco-modern masculinity is frequently used in a technocratic framework, highlighting support for eco-friendly technology. Looking at the early adoption rates of clean technologies, middle-aged and highly educated men were more likely to adopt new, green technologies such as electric vehicles (Sovacool et al. 2019a). Moreover, males generally have more knowledge about energy systems and environmental issues than women. This gendered nature of technological knowledge and the high adoption rates of new technologies can be explained by a combination of eco-modern masculinity and ‘techno-masculinity’. Techno-masculinity can be described as a distinct set of imageries evolving around a conception of manhood that centers on technology, technical expertise and an information society (Poster 2008). While at first glance, eco-modern masculinity seems a threat to petro-masculinity, critics argue that it is an effort to maintain hegemony by acknowledging the issue of climate change while ignoring the unequal impact of climate change policies on poor people, women and people of color (Dockstader and Bell 2020).

4. Data and Methods

The statistical analysis was conducted using secondary data from the European Social Survey (ESS8)—public attitudes to climate change (2016). The ESS is a multi-country survey conducted to monitor and interpret changing public attitudes and values (ESS ERIC 2020). The main topics that are covered in this survey are climate change concerns, energy security concerns, energy preferences and efficacy beliefs. The respondents were selected through multistage national probabilistic sampling. The data were collected through face-to-face interviews (ESS ERIC 2020). While the survey was conducted in 23 European countries, our analysis is limited to data about Belgium (N = 1766). Post-stratification sampling weights were used to compensate for non-response and selection bias, using information on the age group, gender, education and region.

4.1. Variables

To identify distinct groups regarding preferred energy mixes, we included seven 5-point Likert-scale items regarding preferences for diverging energy systems. Energy preferences were measured by asking people about their preferences for a certain energy source. More specifically, respondents were asked to answer the following question: ‘how much of the electricity in Belgium should be generated from …?’. The respondents had to reply on a scale from 1 (a very large amount) to 5 (none at all). A distinction is made between preferences for solar power, wind power, biomass energy, hydroelectric power, natural gas, coal and nuclear energy. Response category 55, ‘I have never heard of this energy source before’, was included in the valid scale. Therefore, we recoded this answer category as a missing value. Looking at the descriptive statistics of the different energy preferences in

Table 1. Descriptive statistics.

Variables	Mean	SD	Min	Max
Energy preferences
Biomass	2.68	1.185	1	5
Coal	4.2	2.137	1	5
Hydroelectric	2.21	3.748	1	5
Natural gas	3.12	2.993	1	5
Nuclear power	3.87	3.763	1	5
Solar power	1.85	1.887	1	5
Wind power	1.79	2.04	1	5
Climate change concerns	3.18	0.857	1	5
Political orientation	4.96	1.975	0	10
Level of education
Primary	0.3	0.458	0	1
Secondary	0.39	0.487	0	1
Tertiary	0.31	0.462	0	1
Financial difficulties
Very difficult	0.07	0.254	0	1
Gender (female)	0.49	0.521	0	1
Age	47.02	18.84	15	100

, it is clear that most respondents prefer solar (m = 1.85) and wind power (m = 1.79). Coal (m = 4.20) and nuclear energy (m = 3.87) are the least-preferred energy sources in Belgium. Interestingly, biomass energy is the least-known energy source among Belgian respondents, as 4.8% of the total respondents report that they have never heard of biomass energy. In contrast, natural gas is the most well-known energy source, as only 0.5% indicate that they have never heard of it before.

Several covariates/explanatory variables were added to the model in order to examine the social characteristics associated with energy mix preferences. Educational level was measured based on the variable ‘eisced’, which measures the highest level of education. The variable was recoded into three categories: primary, secondary and tertiary education. Gender was recoded into the dummy variable ‘female’, with male as the reference category (0 = male; 1 = female). Financial difficulties were measured on the basis of the variable ‘hincfel’. This variable measures the feeling individuals have about their household income and was recoded into a dummy variable with ‘very difficult on present income’ as the reference category. Both climate change concerns and political orientation are also included as explanatory variables. The variable climate change concerns, which indicates how worried an individual is about climate change, is measured using a 5-point Likert scale (1 = not at all worried to 5 = extremely worried). Lastly, political orientation measures an individual’s identification with a political orientation on a scale ranging from 1 (left-wing political orientation) to 10 (right-wing political orientation).

4.2. Analysis

As indicated by the literature, different preferences concerning various energy sources exist. In this study, we built on the idea of heterogeneity among energy preferences to define distinct groups. We focused on unobserved heterogeneity to uncover different subgroups. Unobserved heterogeneity refers to differences between groups that emerge from distinctive path coefficients by looking at response-based segmentation (Hair et al. 2017; Ortega-Egea et al. 2014). As argued by Aldrich et al. (2007), looking at unobserved heterogeneity provides a more nuanced explanation of individuals’ preferences. Heterogeneity was explored by applying a three-step latent class analysis (LCA). A latent class analysis is a statistical method that identifies unobserved classes that contain individuals who share particular criteria (i.e., expressed energy preferences) (Rhead et al. 2018). Specifically, we relied on Mplus and the 3-step method for latent class analysis. The three-step method firstly constructs a model based on a set of indicators. Secondly, it assigns subjects to latent classes based on their posterior class membership probabilities (Asparouhov and Muthén 2014; Vermunt 2010). Lastly, multinomial logistic regression is estimated by adding covariates. More concretely, the 3-step method clusters preferences for energy sources into distinct groups (i.e., energy mixes). Next, socio-demographic variables are added as covariates to analyze the characteristics of class members (gender, level of climate change concern, economic capital, cultural capital, etc.). Whereas the one-step approach simultaneously estimates the latent classes with logistic regression, the three-step model uses a stepwise approach (Vermunt 2010). One of the key advantages of the three-step approach is that the measurement of latent classes is not influenced by auxiliary variables and models (Kamata et al. 2018). The model was run 1000 times with different starting values, and the best model is selected to avoid local optima (Nylund-Gibson and Choi 2018; Peel and McLachlan 2000).

5. Results

5.1. Step One: Identification of Latent Classes

In the first step of the analysis, we defined the optimal class model on the basis of a set of indicators. We compared the fit indices of models with 1–6 latent classes (see Figure 1 and

Table 2. Comparison of fit indices.

	AIC	BIC	Adj. BIC
Model
Two-class	29,292.605	29,604.602	29,423.518
Three-class	28,882	29,352.733	29,079.518
Four-class	28,738.735	29,368.203	29,002.858
Five-class	28,683.95	29,472.152	29,014.677
Six-class	28,653.824	29,600.763	29,051.156

). The model fit was assessed on the basis of the most likely latent class membership through the evaluation of several fit indices: Akaike information criterion (AIC), Bayesian information criterion (BIC) and sample-size-adjusted Bayesian information criterion (adjusted BIC). The sample-size-adjusted BIC reached its minimum with the four-class model, while the BIC reached its minimum on the three-class model. While the AIC kept decreasing, an elbow point was detected in the four-class model. Finally, the Pearson chi-square for the four-class model is not significant, indicating local independence. Based on these statistics, the four-class model was chosen. This is further supported by the fact that the four-class model offered the most theoretically sound interpretation.

5.2. Step Two: Assigning Subjects to Latent Classes

In the second step, the subjects are assigned to latent classes based on their posterior class membership probabilities. The Belgian population is subdivided into four clearly distinct classes as regards energy preferences. Figure 2 represents the item-response probability of energy preferences for the identified classes. While, in general, the most preferred energy sources for all classes are renewables, the results do support the literature that energy preferences vary among individuals. As we look in more detail, differences among the four classes can be observed.

Class A (12%) appears to represent the smallest group. Compared to all other classes, individuals in this class have the lowest preference for renewable energy sources, such as wind energy, solar power, biomass and hydroelectricity. At the same time, they reveal relatively high preferences for non-renewable and nuclear energy. In contrast to class A, individuals in class B (20%) have very high preferences for renewable energy. Their preferred energy mix consists mainly of renewable energy, as they show high preferences for renewable energy sources together with very low preferences for coal, natural gas and nuclear energy. Compared to all other classes, class C (32%) shows the highest preference for coal, natural gas and nuclear energy. However, individuals in class C do not reject renewable energy systems either, rather on the contrary. Therefore, class C proposes a hybrid energy mix that relies on both non-renewable and renewable energy systems. Lastly, most of the Belgian respondents can be situated in class D (36%). While they have very high preferences for renewable energy, they do not fully reject coal, gas and nuclear energy. Just as in class C, individuals in class D prefer a hybrid energy mix.

5.3. Step Three: Adding Group Characteristics

In the next step, covariates were added to the established model to characterize the individuals in the four classes in terms of their socio-demographic profiles, climate change concerns and political orientation. For this regression analysis, the vulnerable energy mix class was used as the reference category, as this is the largest category.

Table 3. Results of multinomial logistic regression (ref. class D).

	Class A		Class B		Class C
	Est.	Sig.	Est.	Sig.	Est.	Sig.
Gender (female)	0.006	(0.977)	−0.396	(0.029)	0.452	(0.004)
Age	0.027	(0.000)	0.024	(0.000)	−0.002	(0.811)
Education
Primary	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.
Secondary	0.295	(0.406)	1.025	(0.024)	0.048	(0.858)
Tertiary	0.345	(0.372)	1.265	(0.006)	−0.512	(0.077)
Financial difficulties	−0.054	(0.834)	−0.919	(0.004)	−0.178	(0.375)
Climate change concerns	−0.767	(0.000)	−0.169	(0.094)	−0.325	(0.000)
Political orientation	0.074	(0.119)	0.053	(0.241)	0.108	(0.007)

presents the results of the latent class regression.

Class A: The business-as-usual energy mix (12%):

Compared to all other classes, class A has the lowest preference for renewable energy systems. We define this class as the ‘business as usual energy mix’. Looking at their characteristics, which are presented in Table 3, we can observe that this group, compared to the reference category, mainly consists of elderly people (est. = 0.027 p ≤ 0.001) who are not really concerned about climate change (est. = −0.767 p ≤ 0.001). These results suggest that individuals who belong to the business-as-usual energy mix class do not prefer an energy transition that completely relies on renewable energy systems. As indicated in the literature, older people are often more skeptical toward energy transitions, as they threaten the existing social order, which requires fundamental changes in traditional values (Diamantopoulos et al. 2003).

Class B: The green energy mix (20%):

Class B is defined as the ‘green energy mix’ because it consists of individuals who mainly prefer a renewable energy mix. The results presented in Table 3 reveal a significant and negative effect of financial difficulties (est. = −0.919, p ≤ 0.010) on the membership of class B compared to the reference class. This suggests that individuals in this group do not experience any financial difficulties. Furthermore, the individuals in this group are highly educated, as there exists a positive and significant impact of education (est. = 1.265 p ≤ 0.001) on class membership compared to class D. The green energy mix, compared to the reference class, mainly consists of males.

Class C: The incremental change energy mix (32%):

We define class C as the ‘incremental change energy mix’. Individuals in this class propose a hybrid energy mix, as they have a high preference for renewable energy sources but do not completely reject fossil fuels and nuclear energy. Looking at the social characteristics, individuals in this class are less concerned about climate change compared to the reference class (est. = −0.325 p ≤ 0.001). Moreover, the results indicate a positive and significant effect of political orientation on the membership of this group (est. = 0.108 p ≤ 0.010) compared to class D. This suggests that individuals in this class mainly identify with a right-wing political ideology compared to the reference class.

Class D: The vulnerable energy mix (36%):

Class D is labeled as the ‘vulnerable energy mix’. While these individuals are very concerned about climate change, they do not completely reject fossil fuels and nuclear energy. Just as in the incremental change energy mix, they propose a hybrid energy mix. The main difference between these two classes was found in the composition of climate change concerns, educational attainment and financial stability. More concretely, individuals in the vulnerable energy mix class are confronted with several trade-offs due to their lack of resources (e.g., lower educational level and financial resources), which influences the composition of their preferred energy mix.

6. Discussion

Exploring the energy preferences of diverse social groups provides useful insights into the perceptions of a carbon-free energy transition and the often-overlooked social dimensions behind this transition. However, existing research regarding energy preferences is limited to the exploration of preferences for one energy system in isolation. In the current study, we investigated the role of social characteristics in individuals’ preferred energy mixes, i.e., the preferred combination of the different sources that are used to produce electricity. An intersectional analysis of the preferred energy mixes provides more nuanced findings and a better understanding of the interconnectedness between energy transitions and power. The results are generally consistent with previous studies: climate change concerns, political orientation, and socio-economic and socio-demographic characteristics have an influence on individuals’ energy preferences (Perlaviciute and Steg 2014; Poortinga et al. 2012). Moreover, we found that there is generally a high acceptance of renewable energy. However, in this study, we uncovered a number of interesting findings ) by looking at the heterogeneity of energy preferences through an intersectional lens focusing on energy mixes, instead of energy preferences in isolation, which reveals a more nuanced understanding of the public’s perception regarding energy transitions. Our analysis revealed that a large number of the Belgian respondents prefer a hybrid energy mix (i.e., a combination of both renewable energy sources and fossil fuels). This highlights the complex decision-making process for the ideal energy mix, especially in a country that faces major challenges in the ongoing energy transition. Moreover, by analyzing the influence of socio-demographic characteristics on the preferences of energy mixes, we revealed the existence of trade-offs within individuals’ energy preferences. More concretely, the results suggest that vulnerable, low-income groups are confronted with a continuous conflict between environmental protection and financial stability. Although they can be deeply concerned about climate change and want to participate in an energy transition toward a low-carbon society, they face several structural barriers that prevent them from denouncing unsustainable energy sources such as fossil fuels. Rather, they may be more inclined to prefer a slower phase-out of fossil fuels, where fossil fuels still play an important role as a ‘transitional energy source’. A possible explanation for these results may be the high upfront costs of renewable energy sources, the lack of information regarding subsidies, no home ownership and the generally high energy burden of low-income households (Durkay 2017). These results highlight the existence of vulnerable groups who experience recognition injustices in the process of achieving a green energy transition. Recognition justice acknowledges the diversity of needs, values and interests among individuals, as it pays attention to those individuals in our society that are underprivileged or ignored (Williams and Doyon 2019) New green energy technologies can be disempowering or alienating for these specific groups (e.g., financially insecure, low-educated, older individuals) (Sovacool et al. 2019a). Hence, energy policies that only focus on subsidizing green technologies such as solar panels or electric cars often misrecognize a specific audience, which could create new inequalities.

On the other hand, our segmentation analysis also revealed privileged groups within the process of an energy transition. In addition to the occurrences of trade-offs, we found that preferences for a renewable energy mix may be linked to high-class practices. This is illustrated by the ‘green energy mix’ class, which consists of highly educated, financially secure men. These results outline that the capability of ‘being green’ often depends on access to certain economic or cultural resources (Kennedy and Givens 2019). Consequently, privileged, high-status individuals are regularly able to define sustainability ideals by re-articulating renewable energy preferences as a high-class practice and orienting privileged people toward it (Carfagna et al. 2014). Moreover, our analysis indicates that mainly men belong to the green energy mix. This gendered nature of energy preferences is consistent with the rise of eco-modern masculinity, which promotes support for eco-friendly technological fixes. The declining public support for the industrial fossil fuel sector has undermined the traditional hegemony of white male elites. By acknowledging the issue of climate change through the support of eco-friendly energy sources, white elite men create the opportunity to maintain their hegemony within the capitalist status quo (Dockstader and Bell 2020; Foster 2012). The results of our analysis illustrate the importance of conducting intersectional research in energy studies. As argued by Davis (2008), intersectionality can be used as a tool to study power structures. Applying an intersectional lens revealed that energy policies are often (unconsciously) designed in accordance with the more powerful, wealthy social class, which increases inequalities by ignoring poor people, women and people of color (Dockstader and Bell 2020). Moreover, these results acknowledge that policies are often designed solely based on technological or economical knowledge. A more human-centered dimension in both energy research and energy policy could decrease inequalities. While the current democratic strategies of climate mitigation clearly have some flaws, our findings do imply that participation and inclusion are crucial in order to create a fair and inclusive energy transition. The results suggest that it is not the unwillingness to change that drives citizens in their preferences. Rather, preferences are connected with individuals’ socio-economic backgrounds, highlighting the role of structural inequalities. While authoritarian systems may be more efficient in pushing through sustainable transitions, they risk increasing inequalities between groups by not recognizing structural barriers. Accordingly, the freedom of civil society organizations cannot be underestimated in democratic regimes, as they have the power to raise the voices of the unheard and change policies in such a way that they include all different social groups. It is important to bear in mind that the energy landscape has changed significantly since the collection of the data due to major events such as the coronavirus crisis and the war in Ukraine. The war, for instance, resulted in rising energy prices along with threats to halt energy exports. This could be a turning point for an energy transition in Europe, as public support for phasing out fossil fuels could increase significantly (Steffen and Patt 2022). However, our study remains relevant, as we highlight the importance of investigating energy mixes through an intersectional lens to uncover significant trade-offs and power dynamics in energy transitions, which are still relevant today. Moreover, the wider application of our study implies that energy transition policies cannot be successful without addressing social dimensions. The findings of this study provide interesting avenues for future research. Qualitative research could contribute to a more in-depth understanding of the four identified classes and how energy transitions are related to injustices and power dynamics. Since the European Social Survey was conducted in 2016, qualitative research could also account for the temporal stability of this study. Additionally, the analysis was limited to the Belgian population, which raises some concerns about the generalizability of the results. Hence, we recommend cross-national research on energy preferences to compare the results among European countries and to include country-level determinants.

7. Conclusions

In conclusion, our study contributes to a more nuanced understanding of the social dimension of the energy transition by examining the varying preferences for energy mixes through an intersectional lens. While most of the Belgian respondents are positive about a sustainable energy transition, many respondents face structural barriers that prevent them from denouncing fossil fuels and nuclear energy. The results reveal the complex decision-making process for the ideal energy mix, especially in a country that faces major challenges in the process of achieving a successful energy transition. Moreover, the results reveal the existence of both vulnerable and privileged groups in the process of an energy transition. This highlights the importance of focusing on energy mixes from an intersectional angle, as it is an excellent tool to uncover trade-offs and underlying power structures within the energy system. Moreover, it provides a fuller understanding of the interconnectedness of energy transitions and social inequality and could inspire policy makers to design socially just and inclusive energy policies.