Journal of Cleaner Production 388 (2023) 135936 Contents lists available at ScienceDirect Journal of Cleaner Production journal homepage: www.elsevier.com/locate/jclepro Environmental behavior patterns across clusters of European Union countries: Uncovering heterogeneity in the attitude-behavior-context relationship Öykü H. Aral *, Jordi López-Sintas Universitat Autònoma de Barcelona (UAB), School of Economics and Business, Department of Business, Av. de l’Eix Central s/n, 08193, Bellaterra, Spain A R T I C L E I N F O A B S T R A C T Handling Editor: Dr. Govindan Kannan This study examines how European Union (EU) citizens systematically differ in their environmental attitude- behavior relationships according to country-level contextual drivers. Using Eurobarometer data and a multi­ Keywords: level latent class regression model, we identified four attitude-behavior relationships that we labelled, according European Union to mean environmental behavior scores, as environmentalist, pre-environmentalist, less-environmentalist, and Sustainability non-environmentalist. Regarding their distribution across the EU, we identified four country clusters reflecting Environmental behaviors pattern similarities that we labelled green, yellow-green, brown, and gray clusters. We found that the attitude- Attitude-behavior-context relationships Multilevel latent class regression model behavior relationship was weaker for more environmentalist patterns than for less environmentalist patterns. Furthermore, more environmentalist patterns were associated with more privileged social positions. As expected, the country clusters reflected socioeconomic development levels (more developed countries were environmen­ tally greener). In the interest of reducing inequality in the distribution of environmentally healthy and vital ecosystems, we suggest that EU environmental policies should mainly address the environmental behaviors of countries classified as brown and gray. 1. Introduction of systematic heterogeneity in samples regarding the A-B-C relationship may inflate variability in model estimators, reducing the model’s sta­ Research into sustainable consumption and environmental behaviors tistical significance and predictive capacity. As far as we are aware, has to date been conducted in single countries (He et al., 2019; Pisano there is no study underpinned by systematic statistical analysis that and Hidalgo, 2013; Sun et al., 2019) or in sets of highly heterogeneous questions the homogeneity of the A-B-C relationship. Assuming homo­ countries (Oreg and Katz-Gerro, 2006; Pisano and Lubell, 2017; Y. geneity in the A-B-C relationship may mask several relationships between Wang, 2017). It has also focused on the mean influence of sociopsy­ environmental drivers and environmental behavior. chological drivers of environmental behaviors (Hines et al., 1987; In this research, we are interested in uncovering systematic hetero­ Klöckner, 2013; Morren and Grinstein, 2016) and on how country-level geneity and its determinants for the A-B-C relationship in Europe. More drivers moderate their influence. In the latter case, the attitude-behavior specifically, we are interested in (1) how environmental behaviors of framework has been further developed, as the Europeans systematically differ according to the A-B-C model, and (2) attitude-behavior-conditions (A-B-C) framework, to include the influ­ the extent to which systematic differences among European countries ence of context on individual-level drivers of sustainable environmental can be reduced to a few distinct clusters, each with a similar mix of behaviors (Black et al., 1985; Guagnano et al., 1995; Pisano and Lubell, environmental behaviors. This fact is of particular importance for policy 2017; T. Wang et al., 2021; Y. Wang, 2017). makers, because, to be able to design better environmental policies, the Few studies have closely examined the existence of individual and European Union (EU) needs to know how Europeans systematically contextual heterogeneity in the A-B-C relationship. The relationship be­ differ in their behaviors and whether heterogeneity can be reduced to a tween attitude and behavior depends not only on the context, but also few sets of countries. According to Aldrich and associates (Aldrich et al., systematically differs among groups of individuals, giving birth to a set 2007), individuals need to be classified according to different behavior of A-B-C relationships, in plural. Failing to take into account the existence patterns and expected responsiveness to government policies. Hence, * Corresponding author. E-mail address: OykuHazal.Aral@autonoma.cat (Ö.H. Aral). https://doi.org/10.1016/j.jclepro.2023.135936 Received 22 March 2022; Received in revised form 31 August 2022; Accepted 4 January 2023 Available online 5 January 2023 0959-6526/© 2023 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC license (http://creativecommons.org/licenses/by- nc/4.0/). Ö.H. Aral and J. López-Sintas Journal of Cleaner Production 388 (2023) 135936 environmental policies need to be well grounded in Europeans’ behav­ Hypothesis H1. Europeans systematically differ in their environ­ iors and expected responses, and their design thus requires systematic mental behaviors. unobserved heterogeneity in sustainability patterns and systematic dif­ While psychological research has focused on attitudes and behaviors, ferences between EU countries to be explored. sociological research has studied how an individual’s social position We understand the EU to be composed of individuals with different (based on social categories and resources) and residence in different attitude-behavior patterns, based in country clusters that differ in terms societies shape the behavior of individuals (Bourdieu, 1979; Giddens, of their environmental practices. Our study thus aims to uncover sys­ 1984). Sociological research acknowledges that individuals have the tematic differences in individual-level drivers of environmental behav­ capacity to act independently of what is expected from their social po­ iors and their distributions across the EU. We also explore how sition, and to make choices that explain behavioral variations – varia­ individual environmental behavior patterns are related to social position tions that are not accounted for by temporal social structures (social categories and resources), and how clusters reflect countries’ (Katz-Gerro et al., 2020). Consequently, while individuals holding socioeconomic development and culture. particular social positions may have similar behaviors, agency differ­ The remainder of this paper is structured as follows. Section 2 in­ ences may account for differing attitudes regarding the environment. troduces the theoretical background, focused mainly on the A-B-C The agency-structure opposition upholds the homology hypothesis environmental model. Section 3 describes our methodology, data, and regarding the environment, i.e., that systematic differences in behaviors analyses. Section 4 reports our results. Section 5 discusses our findings, reflect differing attitudes and social positions of individuals (Gifford and implications, and limitations. Finally, we conclude with a brief Sussman, 2012; Kollmuss and Agyeman, 2002). summary. To explain discrepancies, therefore, we need to account for social differences that explain the social mechanisms underpinning sustainable 1.1. Theoretical framework behaviors (Guagnano et al., 1995). Consequently, researchers have tried to identify the social categories and resources that not only influence 1.1.1. Attitude-behavior-context model behavior, but that also modify the impact of rational and pro-social While early research into the relationship between individuals’ inner motivations on environmental behaviors (Ertz et al., 2016; Geiger sociopsychological processes and environmental behaviors assumed a et al., 2019; He et al., 2019). direct relationship between attitudes and behaviors (Fishbein and Ajzen, Hypothesis H2. Europeans’ environmental behaviors differ system­ 1975; Schwartz, 1970; Kollmuss and Agyeman, 2002), further research atically depending on their social position. has provided evidence that the relationship is more complex (Gupta and Furthermore, the social structures that limit or benefit individuals Ogden, 2006, 2009; Oskamp et al., 1991; Wiederhold and Martinez, may have several social levels. Black and associates (Black et al., 1985) 2018). Since the attitude-behavior relationship would seem to depend examined the indirect effect of market drivers (e.g., fuel price fluctua­ on other individual- and contextual-level drivers, Guagnano and asso­ tions) on the relationship between energy-saving attitudes and behav­ ciates (Guagnano et al., 1995) proposed the iors. Two years later, the first meta-analysis of research into attitude-behavior-conditions (A-B-C) model to explain the social mech­ environmental behaviors proposed that situational drivers (e.g., eco­ anisms that connect sociopsychological drivers with contextual or nomic constraints, social pressures, and opportunities to choose ecological drivers. Note that below we prefer to use the broader concept different actions) may directly influence such behaviors (Hines et al., of ‘context’ rather than ‘conditions’. 1987). However, neither of those studies proposed a clear and detailed Environmental behavior nowadays is understood to reflect a com­ theoretical framework that usefully blended individual and contextual bination of individual-level rational and pro-social motivations. levels. It was Stern and Oskamp (P. Stern and Oskamp, 1987) who Rational behavior is when, under ordinary conditions, individuals act developed the first comprehensive theoretical framework that consid­ according to their own best interests. Pro-social behavior reflects an ered the relationship between individual and contextual drivers and individual’s personal connection with their social and natural contexts individual behaviors; this framework was popularized by Guagnano and and with future generations. Depending on the individual, rational associates as the A-B-C model (Guagnano et al., 1995). According to this motivations may influence behavior more than pro-social motivations, theoretical framework, environmental behavior (B) is the outcome of and vice versa. both personal attitudes (A), and conditions (C) – what we refer to more Environmental psychology models explaining attitudes and behav­ generically as context, i.e., the socioeconomic setting in which con­ iors have been constructed from the theory of planned behavior (TPB) sumers operate (Black et al., 1985), at the micro- (individual and fam­ (Ajzen, 1985), the norm-activation model (NAM) (Schwartz, 1977), the ily), meso- (community), and macro-level (country) (Guagnano et al., new environmental paradigm (NEP) (Dunlap and Liere, 1978) and the 1995; Olli et al., 2001). Consequently, the A-B-C model proposes that the value-belief-norm theory (VBN) (P. C. Stern et al., 1999). Research has attitude-behavior relationship may be moderated by the environmental found that knowledge of environmental problems, and of their conse­ decision-making context. Twenty years on, this model has proved useful quences and possible solutions (actions, skills and strategies), influence in explaining private and public pro-environmental behaviors (Dhir environmental behaviors (He et al., 2019; Jiménez Sánchez and et al., 2021; Ertz et al., 2016; Xu et al., 2017). Lafuente, 2010; Pisano and Hidalgo, 2013). Attitudes to the environ­ ment, which reflect favorable or unfavorable feelings toward particular 1.1.2. Differing A-B-C model context levels aspects or objects related to the environment (Gifford and Sussman, In the first A-B-C study, Guagnano and associates (Guagnano et al., 2012), can be split into two components (Kurisu, 2015): environmental 1995) analyzed the impact of having a recycling bin in the household (a attitudes, and attitudes toward environmental behaviors. Perceived structural driver), finding that bin presence directly and indirectly behavioral control refers to an individual’s perceptions of whether their promoted recycling behaviors, given the same motivations: “This rela­ behavior can bring about environmental change (Ajzen, 1985, 2005). tionship indicates that the effect of providing bins was to remove a major Finally, subjective environmental norms represent environmental barrier to action consistent with pre-existing attitudes” (Guagnano et al., values and attitudes regarding others, i.e., expectations that people will 1995, p. 713). act morally (Morris et al., 2015). Consequently, we can expect in­ Heath and Gifford (Heath and Gifford, 2002), in a study framed dividuals with differing levels of those five theoretical constructs (i.e., within the TPB (Ajzen, 1985), analyzed the direct and indirect impacts sociopsychological predictors) – environmental knowledge, environ­ of a university travel pass aimed at reducing student car use; not only did mental attitudes, attitudes toward environmental behaviors, perceived public transport use increase significantly, attitudes and beliefs con­ behavioral control and subjective environmental norms – to systemati­ cerning public transport also changed for the better. This finding would cally differ in their behaviors. 2 Ö.H. Aral and J. López-Sintas Journal of Cleaner Production 388 (2023) 135936 suggest that contextual drivers may positively moderate the effect of population density (European Commission and European Parliament, attitudes on behaviors. 2018). See Table 1 for country’s sample size. At the macro-level, external drivers are broadly classified as physical, financial, legal, and societal factors (Guagnano et al., 1995). An in­ 2.2. A-B-C model measures dividual’s behavior may thus be affected by a country’s affluence, in­ come inequality, environmental regulations, taxation, infrastructures, 2.2.1. Individual environmental behaviors and culture, to just name a few factors (Welch and Southerton, 2019, p. Individual environmental behaviors – the dependent variable – were 33). For instance, studies have shown that the influence of environ­ measured from two batteries of questions regarding 15 environmental mental values on behavior is different in Japan and in the Netherlands practices (see Supplementary Appendix Table A1), measured as 15 (Aoyagi-Usui et al., 2003), that the influence of environmental concerns dummy indicators, with scores for all the items summed between 0 and on behavior is stronger in the USA than in India (Muralidharan et al., 15. Overall mean score was 3.869 and variance was 0.874 (Cronbach’s 2016), and that the relationship between environmental values, atti­ alpha .68). This variable reflects a wide variety of private-sphere envi­ tudes, and behaviors differs between England, Germany, Portugal, and ronmental behaviors, such as public transport use (e.g., walking instead Spain – countries with different structural conditions in terms of eco­ of using a car), eco-friendly purchases (e.g., buying local products), nomic development, social context, and cultural values (Paço et al., reduced resource consumption (e.g., cutting down on water consump­ 2013). Similar differences have been reported for EU countries tion), and eco-vehicle purchases (e.g., buying an electric car), etc. The (Liobikienė et al., 2016). 15 indicators are described in Table A1. Hypothesis H3. The environmental behavior patterns of Europeans 2.2.2. Individual sociopsychological predictors differ systematically in their distribution among country clusters. The five predictors were measured as follows. Environmental atti­ Hypothesis H4. The environmental behaviors of EU countries differ tudes and attitudes toward environmental behaviors were measured systematically according to country socioeconomic development and from questions regarding individuals’ concerns about environmental culture. problems (Hayward, 1990; Kurisu, 2015) and their attitudes to governmental environmental policies. Perceived behavioral control was 1.1.3. Heterogeneity in the attitude-behavior relationship measured through two indicators: a question regarding role in protect­ Few studies have closely examined individual and contextual het­ ing the environment and a question regarding whether polluters should erogeneity in the A-B-C model. One exception – suggesting the impor­ pay for environmental damage. The latter question reflected the fact tance of studying contextual heterogeneity – is a study by Dolnicar and that, since behaviors can instigate change, individuals should be made Grün (Dolnicar and Grün, 2009), who, in testing the hypothesis that responsible for changing their behavior. Subjective environmental individuals differ systematically in their environmental behaviors norms were measured (in a similar way to Kurisu (2015)) as normative depending on situational contexts, found systematic differences (re­ statements about what should be done to solve environmental problems, flected in six behavioral patterns) in attitude-behavior relationships that with questions reflecting environmental expectations regarding others. changed depending on the context: everyday life or holidays. Farizo and Finally, the Eurobarometer survey records three main sources of associates (Farizo et al., 2014) studied water protection preferences in environmental information, and we assumed that the more sources of England and Wales, finding that heterogeneity could be reduced to five information used by Europeans, the better their knowledge. Responses behavior patterns shaped by the living conditions of individuals, and were therefore summed to create an environmental knowledge scale, suggesting, furthermore, that similarities and differences in behaviors scored from 0 to 3 (more sources indicated better environmental may be geographically distributed. knowledge). Any ‘don’t know’ response was coded as a missing value Consequently, to identify suitable strategies to improve environ­ (see Supplementary Appendix Table A2 for individual-level independent mental sustainability, the extent to which environmental behaviors are variables). context-dependent is not only of theoretical interest, but also of para­ Table 1 reports descriptive statistics for individual environmental mount importance (Dolnicar and Grün, 2009). However, none of the behaviors for the EU27+UK. Compared to countries located in the south studies reviewed so far have analyzed how and why attitude-behavior and southeast of Europe, Scandinavian countries obtained higher scores relationships may systematically differ within and across EU countries for environmental behaviors, perceived behavioral control, and envi­ (Gross and Telesiene, 2017; Schaffrin and Schmidt-Catran, 2017). ronmental knowledge, and lower scores for environmental attitudes, attitudes toward environmental behaviors, and subjective environ­ 2. Methodology mental norms. So, at first glance, our results may represent two groups: southern-eastern and northern-western countries. 2.1. Data As the Eurobarometer survey was not developed to collect data on particular theoretical drivers, we conducted exploratory factor analysis Data was provided by the 2017 Special Eurobarometer 468 public (EFA) to obtain evidence for the capacity of indicators to discriminate opinion survey on the environment carried out in the 27 EU member among individual-level drivers (Wood et al., 2015). Factors were states1 plus the UK (EU27+UK), available from the GESIS data center extracted using principal component analysis (PCA) and varimax rota­ (European Commission and European Parliament, 2018). The sample tion, aided by the ‘psycho’ package (Makowski, 2018) implemented in includes data on 27,881 individuals aged ≥15 years surveyed between the R Environment and Language for Data Analysis (R Core Team, 23 September and 2 October 2017. Individuals were sampled according 2020). The scree test was used to determine the number of factors to to a standard multistage probability procedure and interviews were retain (see Supplementary Appendix Figure A1). For the first four conducted using computer-assisted personal interview techniques. In components, the cumulative percentage of variance was 65%. Note that each country, a number of sampling points was drawn with probabilities the portion of explained variance needs to be judged according to the proportional to population size (for total coverage of the country) and to research context, and can be as low as 50%–60% in social sciences and humanities (Williams et al., 2010). We found that the four-factor matrix structure could be interpreted in terms of environmental attitudes, 1 Austria, Belgium, Bulgaria, Cyprus, Croatia, Czech Republic, Denmark, subjective environmental norms, attitudes toward environmental be­ Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Latvia, haviors, and perceived behavioral control. All loadings were higher than Lithuania, Luxembourg, Malta, Netherlands, Poland, Portugal, Romania, 0.5, so we could assume that the solution discriminates among those Slovakia, Slovenia, Spain, and Sweden. four theoretical constructs (see Supplementary Appendix Table A3). 3 Ö.H. Aral and J. López-Sintas Journal of Cleaner Production 388 (2023) 135936 Table 1 Mean individual environmental behavior and theoretical construct (predictor) scores for EU27+UK. Country Environmental Perceived Environmental Attitudes toward Subjective Environmental Sample behaviors behavioral control attitudes environmental behaviors environmental norms knowledge size Austria 4.407 6.778 16.076 8.961 13.460 2.602 1026 Belgium 4.932 6.929 16.408 9.986 13.881 2.476 1000 Bulgaria 2.749 6.991 17.467 10.432 14.797 2.132 1036 Cyprus 3.179 7.320 18.756 11.089 15.211 2.137 501 Croatia 2.661 6.834 16.616 9.740 13.750 2.082 1020 Czech 3.713 6.674 15.995 9.062 13.589 2.496 1007 Republic Denmark 5.089 7.154 15.594 9.531 14.607 2.674 1002 Estonia 3.703 6.815 16.184 9.496 13.874 2.494 1007 Finland 4.941 6.878 15.687 9.270 14.357 2.717 1010 France 4.686 6.967 16.669 9.926 14.166 2.489 1016 Germany 4.580 7.182 15.904 9.895 13.875 2.373 1535 Greece 3.287 6.970 17.732 9.834 14.713 2.368 1008 Hungary 3.693 6.698 16.301 10.022 14.286 2.182 1050 Ireland 4.252 7.314 17.048 10.313 14.991 2.474 1002 Italy 3.614 6.859 17.170 9.768 14.111 2.162 1027 Latvia 4.209 6.898 17.408 9.856 14.126 2.520 1002 Lithuania 3.113 6.778 16.851 9.870 14.133 2.256 1003 Luxembourg 5.165 6.979 16.942 9.827 14.071 2.331 504 Malta 4.796 7.223 17.891 9.872 14.595 2.310 497 Netherlands 5.371 7.366 15.044 9.546 14.046 2.784 1003 Poland 3.094 6.557 16.265 9.823 13.520 2.059 1009 Portugal 2.609 6.875 16.313 10.231 14.677 1.967 1062 Romania 3.079 6.728 16.491 9.965 13.937 2.132 1031 Slovakia 3.559 6.668 16.418 10.122 14.307 2.136 1084 Slovenia 4.679 7.166 17.323 10.322 14.343 2.219 1025 Spain 3.494 6.944 16.854 10.389 14.393 1.959 1009 Sweden 6.256 7.476 15.821 10.073 14.557 2.840 1027 United 4.107 7.213 16.073 9.448 14.819 2.321 1368 Kingdom EU27+UK 3.869 6.933 16.427 9.832 14.187 2.287 27881 overall development, respectively). Income inequality was scored, using the Table 2 2017 Gini Index (Eurostat, 2020), between 0 and 100 (lowest and Model selection (Bayesian information criterion): environmental behavior pat­ highest inequality, respectively). Data on local environmental issues was terns and country clusters. obtained from the Environmental Performance Index, jointly published Patterns Clusters every two years by Yale and Columbia Universities in collaboration with 1 2 3 4 5 the World Economic Forum; scores range from 0 to 100 (Hsu et al., 1a 105448 2016), with higher scores indicating greater environmental sustain­ 2 104314 102849 102466 102349 102355 ability (i.e., fewer environmental issues). For the EU countries, scores for 3 104182 102775 102345 102242 102233 2016 (the year closest to the Eurobarometer survey) were taken for 19 4 104166 102724 102313 102205 102216 indicators organized into several weighted categories (health, air qual­ 5 104185 102734 102325 102223 102248 ity, sanitation and drinking water, water resources, agriculture, forests, a If the number of consumer patterns is 1, by definition the number of country fisheries, biodiversity, habitat, climate, and energy). Finally, as a clusters is restricted to 1 (Bijmolt et al., 2004) in terms of considering the reflection of culture, individualism-collectivism was measured using multilevel property of individual-level observations. Hofstede’s index published in May 2020 (https://geerthofstede.com/ research-and-vsm/dimension-data-matrix/; see also Hofstede, 2001); 2.2.3. Individual social indicators values range from 0 to 100, with higher scores indicating greater indi­ The social indicators used to control for the social composition of the vidualism (see Supplementary Appendix Table A4). sample were age (15–24, 25–39, 40–54, ≥55 years), gender (men, women), education level (≥15, 16–19, ≥20 years, still studying, no full- 2.2.5. Analysis time education), income measured by proxy according to difficulty To avoid biased estimators and inconsistent results (Ortega-Egea paying bills (most of the time, from time to time, never), community et al., 2014), we use a multilevel latent class regression model that type (rural area or village, small/medium-sized town, large town/city), systematically consideres attitude-behavior relationships to differ across and household size (1, 2, 3, ≥4 persons). clusters of individuals, and the mixture of the latter to differ across countries. Latent class models assume that any population is composed 2.2.4. Country drivers of clusters of individuals that differ according to a set of criteria A country’s socioeconomic development was measured as per capita (McCutcheon, 1987). While we do not know to which cluster each in­ gross domestic product (GDP) for 2017 (expressed in terms of pur­ dividual belongs, we do know that, once we identify the right number of chasing power parity in USD), sourced from the World Development clusters, the association among the set of criteria inside each cluster Indicators (World Bank, 2017). Educational development, reflecting vanishes (Lazarsfeld and Henry, 1968). The latent class model has been mean years of education for adults and expected years of education for extended to latent class regression models (Wedel and Kamakura, 1998), children, was scored according to the 2017 Education Index – sourced and subsequently, to latent class multilevel regression models (Vermunt, from the Human Development Report dataset (Human Development 2003). Report, 2017) – between 0 and 1 (low and high educational The multilevel extension of latent class regression simultaneously 4 Ö.H. Aral and J. López-Sintas Journal of Cleaner Production 388 (2023) 135936 Fig. 1. Distribution of four environmental behavior patterns across four EU27+UK country clusters. clusters (level 2 units). Table 3 We used LatentGOLD 4.0 statistical software (Vermunt and Magi­ Classification errors for the selected model. dson, 2005) to estimate our model parameters. The dependent variable Classification statistics Europeans’ patterns Country classes (individual environmental behaviors), as a summation index, was Classification errors 0.3155 0.0001 modelled following Poisson regression (Faria and Gonçalves, 2013), Reduction of errors (Lambda) 0.3973 0.9998 whereas the theoretical predictors were assumed to follow a normal Entropy R-squared 0.3802 0.9993 distribution. Individual-level social indicators, treated as inactive Standard R-squared 0.3391 0.9996 covariates, were assumed to follow a multinominal distribution. Clusters of individuals, reflecting environmental behavior patterns, were identifies clusters of individuals, homogenous within individual-level described using social indicators, while clusters of countries were patterns, whose distribution differs across clusters, in our case, of described using country-level drivers. countries. Nesting individuals within countries makes it possible to consider that countries belonging to different clusters will differ in terms 3. Results of the distribution of individual-level patterns. In our case, the multi­ level latent class regression model consisted of mixed behavior patterns 3.1. Model selection at the individual level and mixed distributions of those patterns within countries. In this model, regression parameters are allowed to differ The process of estimating different models is described in Table 2. To across individual patterns and across their distributions in country identify the best model, the 21 models that combined individual-level Table 4 Europeans’ environmental behavior patterns. Pattern 1 Pattern 2 Pattern 3 Pattern 4 Sample size (Pre- (Less- (Environmentalist) (Non- environmentalist) environmentalist) environmentalist) Class size 0.4720 0.3274 0.1545 0.0460 23854 Mean environmental behavior score 4.3463 2.8649 6.0605 2.0945 Statistical tests Theoretical constructs (predictors) Wald (=) p Intercept 1.3804 0.9858 1.7733 0.1844 572.5518 <0.0001 Perceived behavioral control (exp((β)) 0.1330 0.0252 0.0899 0.8396 52.9108 <0.0001 (1.14) (1.03) (1.09) (2.32) Environmental attitudes (exp((β)) 0.1238 0.0915 0.0992 0.7524 16.2952 <0.0001 (1.13) (1.10) (1.10) (2.12) Attitudes toward environmental behaviors (exp 0.0546 0.0641 0.0623 0.0885 0.7227 0.87 ((β)) (1.06) (1.07) (1.06) (1.09) Subjective environmental norms (exp((β)) 0.0935 0.0525 0.1159 0.4975 24.7374 <0.0001 (1.10) (1.05) (1.12) (1.64) Environmental knowledge (exp((β)) 0.2779 0.2686 0.0252 − 0.1407 167.6259 <0.0001 (1.32) (1.31) (1.03) (0.87) R2 0.3033 0.1824 0.1500 0.6014 0.5066 CAIC (based on LL) 102244.4 5 Ö.H. Aral and J. López-Sintas Journal of Cleaner Production 388 (2023) 135936 environmental behavior patterns and different country clusters were distribution, to better interpret the impact of parameters on the expected simultaneously estimated. The best overall model was selected accord­ number of environmental behaviors, we transformed the β into exp(β) ing to the lowest Bayesian information criterion (BIC) value (highlighted parameters, interpreted as the relative impact of an environmental in bold in Table 2) based on log-likelihood (Vermunt and Magidson, driver. For the environmentalists (with the highest behavioral score), all 2005): the BIC value decreased until a model reflecting four individual predictors increased behavioral score from 3% to 12%; for the pre- patterns and four country clusters was obtained, and then increased for environmentalists, while there was only a marginal increase for most more complex models. Missing values were removed using list-wise predictors, environmental knowledge increased the overall score by deletion, leaving 23,854 individuals out of the original sample of 27, 32%; for less-environmentalists, environmental knowledge increased 881 individuals. the behavioral score by 31%, while the other predictors produced a 3%– Table 3 reports the classification errors for the selected model, 10% increase; finally, for non-environmentalists (with the lowest reflecting four individual patterns and four country clusters, and also behavioral score), the predictors with the greatest impact were envi­ contains information on how well we can predict the classification of ronmental attitudes, perceived behavioral control, and subjective envi­ individuals and countries, i.e., on how well the clusters are separated. ronmental norms (112%, 132%, and 64% increases, respectively), while Posterior class membership probabilities were used to estimate the there was only a marginal increase (9%) in attitudes toward environ­ proportional classification errors, proportional reductions in classifica­ mental behaviors, and a decrease (13%) for environmental knowledge. tion errors, a measure based on entropy, and the standard R-squared. We therefore found support for hypothesis H1, which proposed that Note that the statistics suggest that, while the classification of countries Europeans systematically differ in their environmental behaviors. in clusters is accurate, behavioral patterns are not so well separated. To what extent do Europeans systematically differ in their environ­ 3.2. To what extent do Europeans’ environmental behaviors differ mental behaviors? systematically depending on their social position? In seeking systematic differences between Europeans (hypothesis H1), the individual-level latent class regression models linked environ­ Table 5 shows the chi-square of independence tests and the row mental behaviors to the five predictors (environmental attitudes, atti­ profiles of the social indicators. Rejecting the null hypothesis, we found tudes toward environmental behaviors, perceived behavioral control, a significant relationship between behavior patterns and social in­ subjective environmental norms, and environmental knowledge). Pos­ dicators (age, gender, education, income, area of residence, and terior membership probabilities were estimated from the model pa­ household size). Thus, hypothesis H2 was supported. rameters and used to classify individuals in just one pattern of behavior, Table 5 also provides evidence on how social indicators systemati­ i.e., in the cluster for which posterior probability was highest (Vermunt cally differed in the distribution of social categories across environ­ and Magidson, 2005). mental patterns. To determine whether a row profile was Table 4 depicts the strength of the attitude-behavior relationships for overrepresented in an environmental behavior pattern, we compared each of the four environmental behavior patterns. The upper part re­ each social category row profile with each pattern size, with a higher ports pattern size, sample size (without missing values), and mean row profile value meaning that the category level was over-represented environmental score for each behavior pattern, and the lower part – in the pattern (patterns are indicated in bold). which shows the association of predictors with behaviors – reports the According to Table 5, therefore, the four environmental behavior Wald statistic reflecting different attitude-behavior relationship patterns can be broadly profiled as follows: environmentalists are strengths across the EU, the predictive capacity (R-squared) of each mainly men aged ≥55 years, well-educated and with a high income, who pattern, the overall predictive capacity of the model, and the consistent live alone in a small/medium-sized town; pre-environmentalists have a Akaike information criterion (CAIC) value for the complete model. Note similar profile, except that most such individuals did not receive full- that we can reject the null hypothesis for all the predictors except for time education; less-environmentalists are mainly women aged 15–24 attitudes toward environmental behavior; thus, hypothesis H1 is largely years, with a lower education level and income than the environmen­ supported, as Europeans do systematically differ in terms of the strength talists, who live in large towns/cities in households of three people; of the attitude-behavior relationship, except for attitudes toward envi­ finally, non-environmentalists have a similar profile to less- ronmental behaviors. environmentalists, except that they are mainly men, without full-time Overall, the explanatory capacity of the model was satisfactory education and with a low income (they struggle to pay bills), who live (R-squared = 0.51) according to standards (usually around 0.3) in in a rural area/village in households of three people. The social patterns environmental behavior studies (He et al., 2019; Klöckner, 2013). depicted in Table 5 therefore support hypothesis H2, that Europeans’ However, explanatory capacity varied depending on behavior patterns: environmental behaviors differ systematically depending on their social it was highest (R-squared = 0.60) for pattern 4 (with the lowest mean position. behavioral score, 2.09) and lowest (R-squared = 0.15) for pattern 3 To what extent do the environmental behaviors of Europeans differ (with the highest mean environmental score, 6.06). systematically in their distribution among country clusters? According to the upper part of Table 4, environmental behavior Table 6 depicts country cluster sizes and the distribution of envi­ patterns can be interpreted as follows: pattern 1 (47.2% of Europeans) is ronmental behavior patterns in those clusters. Four countries were characterized by the second-highest environmental behavior score classified in cluster 1 (Belgium, Luxembourg, the Netherlands, Sweden), (4.35) and moderate predictive capacity (R-squared = 0.30); pattern 2 seven in cluster 2 (Austria, Denmark, Finland, France, Germany, Malta, (32.7% of Europeans) is characterized by a low environmental behavior Slovenia), ten in cluster 3 (Czech Republic, Estonia, Italy, Ireland, score (2.86) and low predictive capacity (R-squared = 0.18); pattern 3 Hungary, Latvia, Poland, Spain, Slovakia, United Kingdom), and seven (15.4% of Europeans) has the highest environmental behavior score in cluster 4 (Bulgaria, Croatia, Cyprus, Greece, Lithuania, Portugal, (6.06) and the lowest predictive capacity (R-squared = 0.15); and Romania). We therefore found support for hypothesis H3 that proposes finally, pattern 4 (4.6% of Europeans) has the lowest environmental that European patterns of environmental behaviors systematically differ behavior score (2.09) and the highest predictive capacity (R-squared = in their distribution among clusters of countries. Note that, in Table 6, 0.60). We labelled the four patterns according to environmental the environmental behavior patterns over-represented in each country behavior scores (highest to lowest) as ‘environmentalist’ (pattern 3), cluster are indicated in boldface. ‘pre-environmentalist’ (pattern 1); ‘less-environmentalist’ (pattern 2) In accordance with researcher proposals for color-coding countries and ‘non-environmentalist’ (pattern 4), representing 15.4%, 47.2%, along a continuum from green to gray (Geiger, et al., 2021), and bearing 32.7% and 4.6% of Europeans, respectively. in mind that pro-environmental behavior is often known as green Since we modelled environmental behavior following a Poisson behavior (Stern, 2000), we color-coded country clusters according to the 6 Ö.H. Aral and J. López-Sintas Journal of Cleaner Production 388 (2023) 135936 Table 5 Associations between Europeans’ environmental behavior patterns and social position indicators. Behavior patterns Environmentalist Pre- Less-environmentalist Non-environmentalist p environmentalist (0.3274) (0.0460) (class size) (0.1545) (0.4720) Inactive covariates Age 15–24 years 12% 45% 37% 6% <0.001 25–39 years 14% 45% 36% 5% 40–54 years 15% 46% 34% 5% ≥55 years 17% 49% 30% 4% Gender Man 16% 48% 31% 5% <0.001 Woman 15% 47% 34% 4% Education ≤15 years 12% 45% 38% 5% <0.001 (age on terminating full-time 16–19 years 13% 46% 36% 5% education) ≥20 years 20% 50% 26% 4% Still studying 15% 48% 32% 5% No full-time 18% 53% 21% 8% education Income Most of the time 7% 31% 53% 9% <0.001 (difficulties paying bills) From time to time 10% 40% 44% 6% Never 18% 52% 26% 4% Community type Rural area/village 16% 47% 32% 5% <0.001 Small/medium 16% 49% 31% 4% town Large town/city 14% 44% 37% 5% Household size 1 17% 51% 28% 4% <0.001 2 17% 48% 30% 5% 3 13% 44% 38% 5% ≥4 14% 44% 37% 5% Table 6 Table 7 EU27+UK country clusters reflecting the distribution of environmental behavior EU27+UK country socioeconomic development and culture indicators. patterns. Green Country Brown Gray p Country cluster clusters Behavior pattern 1 2 3 4 Yellow- (cluster size %) Green Yellow- Brown Gray green green Economic 61075.91 48236.83 39930.11 29024.67 <0.001 (14%) (25%) (36%) (25%) development: Environmentalist 0.5294 0.1815 0.0761 0.0188 per capita GDP Pre-environmentalist 0.4698 0.7745 0.5050 0.1244 in USD Less-environmentalist 0.0006 0.0010 0.3942 0.7506 Educational 0.89 0.89 0.86 0.81 <0.001 Non-environmentalist 0.0002 0.0430 0.0247 0.1061 development: Country-mean environmental 5.2571 4.5669 3.8421 3.0289 Education Index score Income 27.57 27.22 30.09 34.12 <0.001 Green countries: Belgium, Luxembourg, the Netherlands, SwedenYellow- inequality: Gini green countries: Austria, Denmark, Finland, France, Germany, Malta, Sloven­ Index Environment: 84.45 87.87 85.74 85.24 <0.001 iaBrown countries: Czech Republic, Estonia, Italy, Ireland, Hungary, Latvia, Environmental Poland, Spain, Slovakia, United KingdomGray countries: Bulgaria, Croatia, Performance Cyprus, Greece, Lithuania, Portugal, Romania. Index Individualism 73.27 59.52 67.25 35.54 <0.001 mix of individual patterns, from green, the color of renewal and hope score Country-mean 5.2571 4.5669 3.8421 3.0289 and an announcement of life (Lewis, 1996, p.2) to gray, the color of environmental devastated nature, using yellow-green and brown (Brick et al., 2017) to score reflect variations within that green-gray continuum. The country clus­ ters were composed as follows: cluster 1/green (n = 4), formed of 53% environmentalists and 47% pre-environmentalists; cluster 2/yellow-­ were statistically meaningful for all five country-level drivers: socio­ green (n = 7), formed mainly of pre-environmentalists (77%) and en­ economic development, educational development, income inequality, vironmentalists (18%); cluster 3/brown (n = 10, the largest cluster), environmental performance, and individualism-collectivism. We thus formed mainly of pre-environmentalists (50%) and reject the null hypothesis and find support for hypothesis H4, proposing less-environmentalists (39%); and finally, cluster 4/gray (n = 7), formed that environmental behaviors of EU countries differ systematically ac­ mainly of less-environmentalists (75%), non-environmentalists (11%) cording to country socioeconomic development and culture. and pre-environmentalists (12%). Fig. 1 depicts the color-coded four In terms of evidence of systematic differences in country-level in­ clusters reflecting the four different environmental behavior patterns dicators, the mean values for each country cluster in Table 6 indicate the mapped onto the EU27+UK. following (highest to lowest environmental scores): green countries To what extent do the environmental behaviors of EU countries differ have the highest socioeconomic development score, high income systematically according to country socioeconomic development and equality (the lower the index the more equally distributed the income), culture? high educational scores, and the highest individualism scores, but have Table 7 reports t-test results jointly with mean country-level indi­ the lowest environmental performance score; yellow-green countries cator values. We found that score differences across country clusters have the highest educational and environmental performance scores, 7 Ö.H. Aral and J. López-Sintas Journal of Cleaner Production 388 (2023) 135936 the second-highest socioeconomic development score, the lowest in­ et al., 2016; Milfont, 2012; Morren and Grinstein, 2016; Pisano and come inequality score, and the second-lowest individualism score; Lubell, 2017; Soyez, 2012). Environmentalists and brown countries have low socioeconomic and educational development non-environmentalists were mainly found in the country clusters clas­ scores, greater income inequality, and higher environmental perfor­ sified as green (Belgium, Luxembourg, the Netherlands, Sweden) and mance and individualism scores; and finally, gray countries have the gray (Bulgaria, Croatia, Cyprus, Greece, Lithuania, Portugal, Romania), lowest socioeconomic and educational development scores, the second- respectively – two contrasting country clusters with important differ­ lowest environmental performance score, the highest income inequality ences in terms of socioeconomic development and culture. In a word, score, and the lowest individualism score. countries’ structural drivers were correlated with environmentalism level. Previous studies have reported that western and northern EU 4. Discussion countries are more pro-environmental than southern and central EU countries, which, in turn, are more pro-environmental than eastern EU Research into the A-B-C relationship has typically focused on the countries (Bozonnet, 2017; Butkeviciene and Morkevicius, 2017). The effect of theoretical drivers on individual behaviors and the moderating environmental pattern of countries supports hypothesis H4, which effect of social factors, while other research has focused on how proposes that the environmental behaviors of EU countries differ sys­ contextual drivers directly influence behaviors and moderate the impact tematically according to country socioeconomic development and of individual-level sociopsychological predictors (He et al., 2019; Pisano culture. and Lubell, 2017; Y. Wang, 2017). However, our focus was on to what To date, we have identified no study underpinned by systematic extent environmental behaviors of Europeans systematically differ ac­ statistical analysis that questions this reality that breaks with stereo­ cording to the A-B-C model, and to what extent systematic differences types about EU sustainability as a whole. Our results show, for instance, among European countries can be reduced to a few differentiated clus­ that Malta, a southern European island, is, environmentally speaking, ters, each representing similar environmental behaviors. greener than Ireland, the United Kingdom, and northern European We selected the best model in a single step using multilevel latent countries. Such findings would suggest that further research is needed to class regression analysis (Vermunt, 2003; Vermunt and Magidson, 2005; find additional evidence on the environmental behaviors of Europeans. Bijmolt et al., 2004), grouping Europeans according to their behaviors Another interesting finding was that environmental knowledge had a and sensitivity to five sociopsychological predictors (environmental at­ negative effect on environmental behaviors for non-environmentalists, titudes, attitudes toward environmental behaviors, perceived behavioral refocusing our attention on context drivers in the A-B-C model. Non- control, subjective environmental norms, and environmental knowl­ environmentalists were located mostly in gray countries, which had edge). Clusters of countries were identified according to distributions of the lowest educational scores. Possible explanations for this negative environmental behavior patterns; i.e., countries belonging to the same effect of environmental knowledge on environmental behaviors may be cluster reflected a similar within-country distribution of individual misinterpreted or misunderstood information as a consequence of a environmental behaviors. lower educational level or media misinformation; furthermore, gray In identifying four attitude-behavior relationship patterns for Euro­ countries had the highest collectivism scores, which suggests that more peans we found support for hypothesis H1, that Europeans systemati­ importance may be attached to information transmitted by family, cally differ in that relationship. Patterns were labelled according to friends, neighbors and colleagues. Future studies should therefore behavior scores – from more (higher scoring) to less (lower scoring) consider educational level and culture orientation of individuals in environmental behaviors – as environmentalist, pre-environmentalist, investigating the specific effects of different information sources on less-environmentalist and non-environmentalist. All environmental environmental behaviors, so as to identify patterns that can be focused patterns were related to individuals’ social position, as expected ac­ on by customized policies aimed at fostering sustainability behaviors. cording to hypothesis H2, proposing that Europeans’ environmental The patterns of the attitude-behavior relationship suggest that so­ behaviors differ systematically depending on their social position. ciopsychological predictors have less explanatory power for environ­ However, predictive capacity was not as expected. The predictive ca­ mentalists and more explanatory power for non-environmentalists. The pacity of the attitude-behavior model was lowest for the environmen­ first study that used the A-B-C model identified a similar pattern talists and highest for the non-environmentalists. This surprising (Guagnano et al., 1995), finding that the importance of sociopsycho­ attitude-behavior relationship suggests that sociopsychological drivers logical predictors may be reduced by favorable pro-environmental set­ have a greater impact on the behaviors of less environmentally oriented tings. The NAM (Schwartz, 1977), for instance, was less significant in Europeans than on the behaviors of more environmentally oriented explaining behavior in households with recycling bins (Guagnano et al., Europeans. 1995). We suggest, therefore, in line with other authors (Gupta and We therefore found support for the proposition that individuals Ogden, 2006, 2009; Oskamp et al., 1991; Wiederhold and Martinez, classified in each environmental behavior pattern varied systematically 2018), that the strength of the effect of attitudes on behavior depends on according to their social position (social categories and resources). This contextual drivers and the social position of individuals. finding is consistent with previous studies suggesting that environ­ The shape of attitude-behavior relationships may be grounded in mentalists are mostly men (Mostafa, 2007), older (Finisterra do Paço theories of motivations, intrinsic motivators (attitudes), and extrinsic et al., 2009; Golob and Kronegger, 2019), wealthier (do Paço and motivators (incentives). Similar attitudes may lead people to similar Raposo, 2009; Rowlands et al., 2003), and well educated (Golob and environmental behaviors in the same context – whether the social Kronegger, 2019; Ortega-Egea et al., 2014), and live in small households context (doing as other people do) or the economic context (responding (Poortinga et al., 2004) in small/medium-sized towns (Berenguer et al., to external incentives). However, similar attitudes may also lead people 2005). to behave differently in different country contexts. The motivation In uncovering environmental behavior patterns, we identified four crowding-out theory suggests that external incentives, whether positive clusters of EU countries that varied according to the distribution of those or negative, may undermine intrinsic motivations (Bruno et al., 2017; patterns. We thus found support for the hypothesis H3, that EU countries Peng and Liu, 2020). Research has also found that the influence of in­ systematically differ according to the distribution of individual envi­ ternal motivators diminishes once people become accustomed to ronmental behavior patterns. Furthermore, patterns with higher scores external social or economic motivators (Frey, 1997; Rommel et al., were overrepresented in country clusters with higher socioeconomic, 2015; Heyman and Ariely, 2004; 2009). educational, and individualism scores, and lower income inequality scores, supporting the proposition that culture and country socioeco­ nomic development have an undeniable effect on behaviors (Liobikienė 8 Ö.H. Aral and J. López-Sintas Journal of Cleaner Production 388 (2023) 135936 4.1. Limitations and future research org/10.1016/j.jclepro.2023.135936. Our study has some limitations that open up opportunities for future References research. The first limitation is inherent to using secondary data: we had no power to determine environmental indicators of individual-level Ajzen, I., 1985. From intentions to actions: a theory of planned behavior. In: Kuhl, J., Beckmann, J. (Eds.), Action Control: from Cognition to Behavior. Springer, behaviors and predictors, and so were constrained in how they were pp. 11–39. https://doi.org/10.1007/978-3-642-69746-3_2. measured. A second, and related, limitation, concerns the aggregate Ajzen, I., 2005. Attitudes, Personality, and Behavior. McGraw-Hill Education, (UK). nature of the A-B-C model, which reflects research into private-sphere Aldrich, G.A., Grimsrud, K.M., Thacher, J.A., Kotchen, M.J., 2007. Relating environmental attitudes and contingent values: how robust are methods for but not public-sphere environmental behaviors. identifying preference heterogeneity? Environ. Resour. Econ. 37 (4), 757–775. Attitude-behavior theories such as the NAM (Schwartz, 1977), the https://doi.org/10.1007/s10640-006-9054-7. TPB (Ajzen, 1985) and the VBN (Stern, 2000) could be applied to further Aoyagi-Usui, M., Vinken, H., Kuribayashi, A., 2003. Pro-environmental attitudes and behaviors: an international comparison. Hum. Ecol. Rev. 10 (1), 23–31. explore this topic, broadening studies to include other attitude variables Berenguer, J., Corraliza, J.A., Martín, R., 2005. Rural-urban differences in environmental that depend on specific behaviors. Heterogeneity in theories could also concern, attitudes, and actions. Eur. J. Psychol. Assess. 21 (2), 128–138. https://doi. be examined in more detail using a hybrid multilevel causal model org/10.1027/1015-5759.21.2.128. Bijmolt, T.H.A., Paas, L.J., Vermunt, J.K., 2004. Country and consumer segmentation: (Lamberti et al., 2017, 2021). A final suggestion is that more detailed multi-level latent class analysis of financial product ownership. Int. J. Res. Market. analyses could delve into the influence of political history, law, taxation, 21 (4), 323–340. https://doi.org/10.1016/j.ijresmar.2004.06.002. and international relationships on the A-B-C model (Bodur and Sar­ Black, J.S., Stern, P.C., Elworth, J.T., 1985. Personal and contextual influences on igöllü, 2005; Dolnicar and Grün, 2009; Ortega-Egea et al., 2014). househould energy adaptations. J. Appl. Psychol. 70 (1), 3–21. https://doi.org/ 10.1037/0021-9010.70.1.3. Brick, C., Sherman, D.K., Kim, H.S., 2017. Green to be seen” and “brown to keep down”: 5. Summary visibility moderates the effect of identity on pro-environmental behavior. J. Environ. Psychol. 51, 226–238. Bodur, M., Sarigöllü, E., 2005. Environmental sensitivity in a developing country: Our study is, as far as we are aware, the first that examines how consumer classification and implications. Environ. Behav. 37 (4), 487–510. https:// Europeans systematically differ, depending on their EU country context, doi.org/10.1177/0013916504269666. in their environmental behaviors and their attitude-behavior relation­ Bourdieu, P., 1979. La Distinction: Critique Sociale Du Jugement. Editions de Minuit. Bozonnet, J.-P., 2017. Explaining environmental activism by national cultures: the ships. We contribute to sustainability knowledge by providing evidence hypothesis of hysteresis. In: Green European Environmental Behaviour and Attitudes of socially distributed heterogenous attitude-behavior relationships, of a in Europe in a Historical and Cross-Cultural Comparative Perspective, first ed. heterogenous mix of behaviors in different EU countries, and of EU Routledge. Butkeviciene, E., Morkevicius, V., 2017. Explaining membership in pro-environmental country clusters with similar country-level drivers and similar distri­ groups: a comparative European perspective. In: Green European Environmental butions of individual-level environmental behavior patterns. We also Behaviour and Attitudes in Europe in a Historical and Cross-Cultural Comparative provide an explanation for heterogenous attitude-behavior relationships Perspective, first ed. Routledge. Dhir, A., Sadiq, M., Talwar, S., Sakashita, M., Kaur, P., 2021. Why do retail consumers based on the crowding-out effect of external motivators. buy green apparel? A knowledge-attitude-behaviour-context perspective. Since environmental behaviors are the outcome of internal and J. Retailing Consum. Serv. 59, 102398 https://doi.org/10.1016/j. external drivers, interventions need to be tailored according to the jretconser.2020.102398. attitudinal and contextual limitations or opportunities for pro- do Paço, A., Raposo, M., 2009. Green” segmentation: an application to the Portuguese consumer market. Market. Intell. Plann. 27 (3), 364–379. https://doi.org/10.1108/ environmental behaviors in individuals. Indirectly or directly incorpo­ 02634500910955245. rating contextual drivers in attitude-behavior relationships is essential Dolnicar, S., Grün, B., 2009. Environmentally friendly behavior: can heterogeneity to the design of more focused environmental policies. Our findings, we among individuals and contexts/environments Be harvested for improved sustainable management? Environ. Behav. 41 (5), 693–714. https://doi.org/ hope, will help policy makers design better environmental action plans 10.1177/0013916508319448. that consider systematic differences in the individual environmental Dunlap, R.E., Liere, K.D.V., 1978. The “new environmental paradigm. J. Environ. Educ. 9 behaviors of Europeans and their distribution across EU countries. (4), 10–19. https://doi.org/10.1080/00958964.1978.10801875. Ertz, M., Karakas, F., Sarigöllü, E., 2016. Exploring pro-environmental behaviors of consumers: an analysis of contextual factors, attitude, and behaviors. J. Bus. Res. 69 Funding (10), 3971–3980. https://doi.org/10.1016/j.jbusres.2016.06.010. European Commission, European Parliament, 2018. Eurobarometer 88.1. European Commission and European Parliament; GESIS Data Archive, Cologne. https://doi. Grant #2017 SGR 1056funded by AGAUR-Generalitat de Catalunya, org/10.4232/1.12959, 2017. and Grant # PID 2021-124089NB-I00 funded by the Spanish Ministry of Eurostat, 2020. Eurostat—Data Explorer. https://appsso.eurostat.ec.europa.eu/nui/s Science and Innovation through the National Scientific, Technical, and how.do?dataset=ilc_di12&lang=en. Faria, S., Gonçalves, F., 2013. Financial data modeling by Poisson mixture regression. Innovation Research Plan. J. Appl. Stat. 40 (10), 2150–2162. https://doi.org/10.1080/ 02664763.2013.807332. CRediT authorship contribution statement Farizo, B.A., Joyce, J., Soliño, M., 2014. Dealing with heterogeneous preferences using multilevel mixed models. Land Econ. 90 (1), 181–198. https://doi.org/10.3368/ le.90.1.181. Öykü H. Aral: Formal analysis. Jordi López-Sintas: Research Finisterra do Paço, A.M., Barata Raposo, M.L., Filho, W.L., 2009. Identifying the green design. consumer: a segmentation study. J. Target Meas. Anal. Market. 17 (1), 17–25. https://doi.org/10.1057/jt.2008.28. Fishbein, M., Ajzen, I., 1975. Belief, Attitude, Intention, and Behavior: an Introduction to Declaration of competing interest Theory and Research. Addison-Wesley, Reading, MA. https://people.umass.edu/aiz en/f&a1975.html. The authors declare that they have no known competing financial Geiger, J.L., Steg, L., van der Werff, E., Ünal, A.B., 2019. A meta-analysis of factors related to recycling. J. Environ. Psychol. 64, 78–97. https://doi.org/10.1016/j. interests or personal relationships that could have appeared to influence jenvp.2019.05.004. the work reported in this paper. Geiger, S.J., Brick, C., Nalborczyk, L., Bosshard, A., Jostmann, N.B., 2021. More green than gray? Toward a sustainable overview of environmental spillover effects: a Bayesian meta-analysis. J. Environ. Psychol. 78, 101694. Data availability Giddens, A., 1984. The Constitution of Society: Outline of the Theory of Structuration. University of California Press. Data will be made available on request. Gifford, R., Sussman, R., 2012. Environmental attitudes. In: The Oxford Handbook of Environmental and Conservation Psychology. https://doi.org/10.1093/oxfordhb/ 9780199733026.013.0004. Appendix A. Supplementary data Golob, U., Kronegger, L., 2019. Environmental consciousness of European consumers: a segmentation-based study. J. Clean. Prod. 221, 1–9. https://doi.org/10.1016/j. jclepro.2019.02.197. Supplementary data to this article can be found online at https://doi. 9 Ö.H. Aral and J. López-Sintas Journal of Cleaner Production 388 (2023) 135936 Gross, M., Telesiene, A., 2017. Introduction. How green are green Europeans?. In: Green Oreg, S., Katz-Gerro, T., 2006. Predicting proenvironmental behavior cross-nationally: European: Environmental Behaviour and Attitudes in Europe in a Historical and values, the theory of planned behavior, and value-belief-norm theory. Environ. Cross-Cultural Comparative Perspective, first ed. Routledge. Behav. 38 (4), 462–483. https://doi.org/10.1177/0013916505286012. Guagnano, G.A., Stern, P.C., Dietz, T., 1995. Influences on attitude-behavior Ortega-Egea, J.M., García-de-Frutos, N., Antolín-López, R., 2014. Why do some people do relationships: a natural experiment with curbside recycling. Environ. Behav. 27 (5), “more” to mitigate climate change than others? Exploring heterogeneity in psycho- 699–718. https://doi.org/10.1177/0013916595275005. social associations. PLoS One 9 (9), e106645. https://doi.org/10.1371/journal. Gupta, S., Ogden, D., 2006. The Attitude-Behavior Gap in Environmental Consumerism, pone.0106645. vol. 3, pp. 199–206, 1. Oskamp, S., Harrington, M.J., Edwards, T.C., Sherwood, D.L., Okuda, S.M., Swanson, D. Gupta, S., Ogden, D.T., 2009. To buy or not to buy? A social dilemma perspective on C., 1991. Factors influencing household recycling behavior. Environ. Behav. 23 (4), green buying. J. Consum. Market. 26 (6), 376–391. https://doi.org/10.1108/ 494–519. https://doi.org/10.1177/0013916591234005. 07363760910988201. Paço, A. do, Alves, H., Shiel, C., Filho, W.L., 2013. Development of a green consumer Hayward, K., 1990. Responsible Environmental Behaviour: A Test of the Hines Model behaviour model. Int. J. Consum. Stud. 37 (4), 414–421. https://doi.org/10.1111/ [Masters’ Thesis. Wilfrid Laurier University. https://libraryguides.vu.edu.au/harva ijcs.12009. rd/theses. Pisano, I., Hidalgo, M.-C., 2013. Environmental concern in Andalusia, Spain: what, how He, Q., Duan, Y., Wang, R., Fu, Z., 2019. Factors affecting consumers’ purchase intention much, who and why? PsyEcology 4 (3), 287–310. https://doi.org/10.1174/ of eco-friendly food in China: the evidence from respondents in Beijing. Int. J. 217119713807749841. Consum. Stud. 43 (5), 457–470. https://doi.org/10.1111/ijcs.12525. Pisano, I., Lubell, M., 2017. Environmental behavior in cross-national perspective: a Heath, Y., Gifford, R., 2002. Extending the theory of planned behavior: predicting the use multilevel analysis of 30 countries. Environ. Behav. 49 (1), 31–58. https://doi.org/ of public Transportation1. J. Appl. Soc. Psychol. 32 (10), 2154–2189. https://doi. 10.1177/0013916515600494. org/10.1111/j.1559-1816.2002.tb02068.x. Poortinga, W., Steg, L., Vlek, C., 2004. Values, environmental concern, and Hines, J.M., Hungerford, H.R., Tomera, A.N., 1987. Analysis and synthesis of research on environmental behavior: a study into household energy use. Environ. Behav. 36 (1), responsible environmental behavior: a meta-analysis. J. Environ. Educ. 18 (2), 1–8. 70–93. https://doi.org/10.1177/0013916503251466. https://doi.org/10.1080/00958964.1987.9943482. R Core Team, 2020. R: A Language and Environment for Statistical Computing. https:// Hofstede, G., 2001. Culture’s Consequences: Comparing Values, Behaviors, Institutions www.R-project.org/. and Organizations across Nations. SAGE Publications. Rowlands, I.H., Scott, D., Parker, P., 2003. Consumers and green electricity: profiling Hsu, A.D., Esty, M., Levy, A., de Sherbinin, 2016. The 2016 Environmental Performance potential purchasers. Bus. Strat. Environ. 12 (1), 36–48. https://doi.org/10.1002/ Index Report. Yale Center for Environmental Law and Policy (YCELP) and Center for bse.346. International Earth Science Information Network (CIESIN), Columbia University. Schaffrin, A., Schmidt-Catran, A., 2017. Changing concern about environmental threats https://doi.org/10.13140/RG.2.2.19868.90249. and risks – a longitudinal and multilevel perspective on the relationship between Human Development Report, 2017. Human Development Reports. http://hdr.undp. values and interests. In: Green European: Environmental Behaviour and Attitudes in org/en/indicators/103706. Europe in a Historical and Cross-Cultural Comparative Perspective. Routledge. Jiménez Sánchez, M., Lafuente, R., 2010. Defining and measuring environmental Schwartz, S.H., 1970. Elicitation of moral obligation and self-sacrificing behavior: an consciousness. Rev. Int. Sociol. 68 (3), 731–755. https://doi.org/10.3989/ experimental study of volunteering to be a bone marrow donor. J. Pers. Soc. Psychol. ris.2008.11.03. 15 (4), 283–293. https://doi.org/10.1037/h0029614. Katz-Gerro, T., Greenspan, I., Handy, F., Vered, Y., 2020. Environmental behavior in Schwartz, S.H., 1977. Normative Influences on Altruism11This work was supported by three countries: the role of intergenerational transmission and domains of NSF Grant SOC 72-05417. I am indebted to L. Berkowitz, R. Dienstbier, H. Schuman, socialization. J. Environ. Psychol. 71, 101343 https://doi.org/10.1016/j. R. Simmons, and R. Tessler for their thoughtful comments on an early draft of this jenvp.2019.101343. chapter. In: Berkowitz, L. (Ed.), Advances in Experimental Social Psychology. Klöckner, C.A., 2013. A comprehensive model of the psychology of environmental Academic Press, pp. 221–279. https://doi.org/10.1016/S0065-2601(08)60358-5, behaviour—a meta-analysis. Global Environ. Change 23 (5), 1028–1038. https:// 10. doi.org/10.1016/j.gloenvcha.2013.05.014. Soyez, K., 2012. How national cultural values affect pro-environmental consumer Kollmuss, A., Agyeman, J., 2002. Mind the Gap: why do people act environmentally and behavior. Int. Market. Rev. 29 (6), 623–646. https://doi.org/10.1108/ what are the barriers to pro-environmental behavior? Environ. Educ. Res. 8 (3), 02651331211277973. 239–260. https://doi.org/10.1080/13504620220145401. Stern, P.C., 2000. New environmental theories: toward a coherent theory of Kurisu, K., 2015. Pro-environmental Behaviors, first ed. Springer Japan https://www.sp environmentally significant behavior. J. Soc. Issues 56 (3), 407–424. ringer.com/gp/book/9784431558323. Stern, P.C., Dietz, T., Abel, T., Guagnano, G.A., Kalof, L., 1999. A value-belief-norm Lamberti, G., Aluja, T.B., Sanchez, G., 2017. The Pathmox approach for PLS path theory of support for social movements: the case of environmentalism. Hum. Ecol. modeling. Discov. construct. different. segm. 33 (6), 674–689. Rev. 6 (2), 81–97 (JSTOR). Lamberti, G., Lopez-Sintas, J., Sukphan, J., 2021. The social process of internet Stern, P., Oskamp, S., 1987. Managing Scarce Environmental Resources. Handbook of appropriation: living in a digitally advanced country benefits less well-educated Environmental Psychology. Wiley, pp. 1043–1088. Europeans. Telecommun. Pol. 45 (1), 102055 https://doi.org/10.1016/j. Sun, Y., Liu, N., Zhao, M., 2019. Factors and mechanisms affecting green consumption in telpol.2020.102055. China: a multilevel analysis. J. Clean. Prod. 209, 481–493. https://doi.org/10.1016/ Lazarsfeld, P.F., Henry, N.W., 1968. Latent Structure Analysis. Houghton Mifflin j.jclepro.2018.10.241. Company. Vermunt, J.K., 2003. Multilevel latent class models. Socio. Methodol. 33 (1), 213–239. Lewis, C.A., 1996. Green Nature/human Nature: the Meaning of Plants in Our Lives. https://doi.org/10.1111/j.0081-1750.2003.t01-1-00131.x. University of Illinois press. Vermunt, J.K., Magidson, J., 2005. Technical Guide for Latent GOLD 4.0. In: Basic and Liobikienė, G., Mandravickaitė, J., Bernatonienė, J., 2016. Theory of planned behavior Advanced. Statistical Innovations, Belmont, MA. approach to understand the green purchasing behavior in the EU: a cross-cultural Wang, T., Shen, B., Han Springer, C., Hou, J., 2021. What prevents us from taking low- study. Ecol. Econ. 125, 38–46. https://doi.org/10.1016/j.ecolecon.2016.02.008. carbon actions? A comprehensive review of influencing factors affecting low-carbon Makowski, D., 2018. The psycho package: an efficient and publishing-oriented workflow behaviors. Energy Res. Social Sci. 71, 101844 https://doi.org/10.1016/j. for psychological science. J. Open Source Softw. 3 (22), 470. https://doi.org/ erss.2020.101844. 10.21105/joss.00470. Wang, Y., 2017. Promoting sustainable consumption behaviors: the impacts of McCutcheon, A.L., 1987. Latent Class Analysis. Sage University. In: Papers Series on environmental attitudes and governance in a cross-national context. Environ. Behav. QASS. Sage Publishers, 07-64. 49 (10), 1128–1155. https://doi.org/10.1177/0013916516680264. Milfont, T.L., 2012. Cultural differences in environmental engagement. In: The Oxford Wedel, M., Kamakura, W.A., 1998. Market Segmentation: Conceptual and Handbook of Environmental and Conservation Psychology. Oxford University Press, Methodological Foundations. Kluwer Academic. pp. 181–200. https://doi.org/10.1093/oxfordhb/9780199733026.013.0011. Welch, D., Southerton, D., 2019. After Paris: transitions for sustainable consumption. Morren, M., Grinstein, A., 2016. Explaining environmental behavior across borders: a Sustain. Sci. Pract. Pol. 15 (1), 31–44. https://doi.org/10.1080/ meta-analysis. J. Environ. Psychol. 47, 91–106. https://doi.org/10.1016/j. 15487733.2018.1560861. jenvp.2016.05.003. Wiederhold, M., Martinez, L.F., 2018. Ethical consumer behaviour in Germany: the Morris, M.W., Hong, Y., Chiu, C., Liu, Z., 2015. Normology: integrating insights about attitude-behaviour gap in the green apparel industry. Int. J. Consum. Stud. 42 (4), social norms to understand cultural dynamics. Organ. Behav. Hum. Decis. Process. 419–429. https://doi.org/10.1111/ijcs.12435. 129, 1–13. https://doi.org/10.1016/j.obhdp.2015.03.001. Williams, B., Onsman, A., Brown, T., 2010. Exploratory factor analysis: a five-step guide Mostafa, M.M., 2007. Gender differences in Egyptian consumers’ green purchase for novices. Austr. J. Paramed. 8 (3) https://doi.org/10.33151/ajp.8.3.93. behaviour: the effects of environmental knowledge, concern and attitude. Int. J. Wood, N.D., Gnonhosou, D.C.A., Bowling, J.W., 2015. Combining parallel and Consum. Stud. 31 (3), 220–229. https://doi.org/10.1111/j.1470-6431.2006.00523. exploratory factor analysis in identifying relationship scales in secondary data. x. Marriage Fam. Rev. 51 (5), 385–395. https://doi.org/10.1080/ Muralidharan, S., Rejón-Guardia, F., Xue, F., 2016. Understanding the green buying 01494929.2015.1059785. behavior of younger Millennials from India and the United States: a structural World Bank, 2017. GDP Per Capita, PPP (Current International $) | Data (GDP.PCAP.PP. equation modeling approach. J. Int. Consum. Market. 28 (1), 54–72. https://doi.org/ CD). https://data.worldbank.org/indicator/NY. 10.1080/08961530.2015.1056328. Xu, X., Maki, A., Chen, C., Dong, B., Day, J.K., 2017. Investigating willingness to save Olli, E., Grendstad, G., Wollebaek, D., 2001. Correlates of environmental behaviors: energy and communication about energy use in the American workplace with the bringing back social context. Environ. Behav. 33 (2), 181–208. https://doi.org/ attitude-behavior-context model. Energy Res. Social Sci. 32, 13–22. https://doi.org/ 10.1177/0013916501332002. 10.1016/j.erss.2017.02.011. 10