Individual action, world society, and environmental change: 1993–2010

The last decades have been characterised by a rapid growth of environmental organisations and environmental protection measures at the global level and within nations (Frank, Hironaka, & Schofer, 2000; Hironaka, 2014; Longhofer & Schofer, 2010). Scholars following the ‘World Society/World Polity’ paradigm (Meyer, Boli, Thomas, & Ramirez, 1997) consider this proliferating global regime an important driver of many national environmental outcomes and have demonstrated its effects in areas such as the creation of state parks (Frank et al., 2000), environmental protection measures (Schofer & Hironaka, 2005), the inclusion of environmentalism in textbooks (Bromley, Meyer, & Ramirez, 2011), and on the use of pesticides (Shorette, 2012). Similarly, comparative cross-sectional survey studies have found positive effects on various environmental concerns and behaviours (Givens & Jorgenson, 2013; Hadler & Haller, 2011; Marquart-Pyatt, 2012).

Yet, time-comparative studies of environmental concerns and behaviours reveal divergent trends over time, with some behaviours and concerns becoming less common, others more common, and others remaining stable (see Dalton & Rohrschneider, 2015; Franzen & Vogl, 2013). The presence of such divergent trends suggests that the link between global cultural models and individual attitudes and behaviours is more complex than the positive effect emphasised in existing studies. Following this lead and drawing on ideas developed within world society (WS) literature, this article sets out to, first, study the question, how does WS shape environmental behaviours at the individual level? This is examined using a novel, longitudinal, cross-sectional design. Second, it aims to improve the theoretical conceptualisation of the link between global models and individual behaviours.

The question of how individuals are influenced by global cultural models has attracted attention only very recently (Hadler, 2012; Hallett, 2010; Pierotti, 2013; Velitchkova, 2015), a lag which is surprising given that world-societal concepts draw heavily on the micro-sociological accounts of Berger and Luckmann, Goffmann, Schütz, and others (see Scott, 2014 for an overview). The present article follows this emphasis and considers findings on and explanations for coupling and decoupling processes that have been brought forward in related organisational studies. Cross-national variation in environmental behaviours as well as their local circumstances thus receives less attention than in research focusing on the topic and is summarised only briefly.¹

In substantive terms, this article contrasts the effects of global cultural models on public and private behaviours (see Hunter, Hatch, & Johnson, 2004; Lavelle, Rau, & Fahy, 2016; Stern, 2000 for a detailed discussion). The former encompasses collective political actions related to environmentalist activism, such as protesting, while the latter focuses on habitual behaviours in the private sphere, such as recycling. As the results section will show, both behaviours exhibit different patterns over time, which is an ideal set-up to study the effects of WS and to question its dominant narrative of positive effects on all behaviours.

Following these main thoughts, this article is structured as follows: the subsequent section describes WS theory and proposes several hypotheses that reflect ideas on how to better grasp the macro–micro-link within this theory, followed by a brief summary of alternative views. Subsequently, the data and analysis strategy are described. The results section starts with a descriptive overview of the changes in public and private behaviours across countries and time, followed by various multilevel models. The concluding section discusses the relevance of these findings for the hypotheses and their significance for WS theory.

WS theory conceptualises the diffusion of Western cultural models such as environmentalism and human rights principles and their influence on nation-states, organisations, and individual actors (Meyer et al., 1997; Ramirez, 2012; Schofer, Hironaka, Frank, & Longhofer, 2012). The history and development of cultural models describing environmental enquiry that have become dominant at a global level are succinctly described by Hironaka (2014). Environmental ideas coalesced around the 1972 Stockholm UN conference in which environmental and developmental goals were linked. This meeting gave rise to global cultural models that offered a framework for previously scattered and localised environmentalist actors and actions. It also resulted in further international treaties such as the UN Convention on Long-range Transboundary Air Pollution of 1979, the Stockholm Convention on Persistent Organic Pollutants of 2001, and international non-governmental organisations (INGO) such as Greenpeace or 350.org.

The presence of such organisations within a country as well as a nation’s membership in such organisations is considered a proxy for how strongly countries are embedded in global cultural models. The dominant WS narrative is that these societal ties have positive effects on the diffusion of global ideas, with the impact of international organisations considered stronger the more ties are present, the longer they exist, and the more frequently contact occurs between them and local actors (Schofer & Hironaka, 2005). Accordingly, research has shown positive effects including the creation of national parks, state memberships in environmental organisations, the establishment of environmental laws and ministries (Frank et al., 2000), environmental protection measures (Schofer & Hironaka, 2005), the inclusion of environmentalism in school books (Bromley et al., 2011), and reduced use of fertilisers (Shorette, 2012).

So far, WS has been extensively applied to organisational outcomes and actions, where phenomena such as isomorphism, decoupling, coupling, and other processes have been discussed in detail (Bromley & Powell, 2012; Cole & Ramirez, 2013; Meyer, 2007). The link between global ideas and individual actions, however, has received far less attention (but see Hallett, 2010; Pierotti, 2013; Velitchkova, 2015). The comparatively scant literature on the influence of global ideas on individual attitudes and behaviours has adopted the view that ‘conformity to evolving external institutions would be most rapid among those actors best linked in to the wider environment, through worlds of associational life (e.g. nongovernmental memberships) and cultural life (e.g. through extended schooling)’ (Meyer, 2010, p. 13). It therefore emphasises exposure by assuming that individual conformity to global models is higher in societies in which more international organisations are present (Boyle, 2005). Confirming this, Givens and Jorgenson (2013), Hadler and Haller (2011), and Marquart-Pyatt (2012) identified positive effects of exposure on environmental concerns and individual behaviour. This leads to a first, basic hypothesis:

WS-related research, however, has also shown that global ideas and organisational outcomes are not always aligned and can be decoupled (Bromley & Powell, 2012). Decoupling processes have been studied extensively in human-rights-related topics (Cole & Ramirez, 2013; Hafner-Burton & Tsutsui, 2005). This research has identified a diverse array of effects such as ratification without implementation, initial negative effects due to increased scrutiny, but also positive effects, particularly when monitoring provisions were attached. Environmental institutions share many of the features of international human rights institutions – considering that both areas are seen by the UN as strongly interrelated and that predominantly environmental INGOs such Green Peace also address human rights² – and decoupling processes thus are also perceivable in addition to the diffusion idea expressed in the basic hypothesis above. The following paragraphs, therefore, provide an overview of possible decoupling factors and how they might affect environmental behaviours.

A first qualifying aspect rests on the observation that WS is not a homogenous entity but made up of different actors, levels, and regional densities (Beckfield, 2010; Vasi, 2007). Two important and sometimes opposed actors are governments and civil society. Hafner-Burton and Tsutsui (2005) showed, for example, that the signing of international treaties by governments is often ineffective in reducing human rights violations, whereas the presence of INGOs was effective. Similarly, Tsutsui (2004) did not find any effects of governmental ties on social movements, whereas non-governmental ties had a positive effect. Applying this view to environmentalism, one can expect that environmental behaviours are shaped differently by governmental and non-governmental organisations. As for the distinction between public and private behaviour, governmental ties are likely to have a lower impact on public behaviour, given that political actions often target governments, whereas governmental regulations are often pivotal in shaping private environmental behaviours (Hadler & Haller, 2011).

As pointed out with reference to the role of governments, the influence of global ideas and the effects of ties via international organisations can be amplified by nation-states and national polities. Environmentalism has been institutionalised within nation-states in forms such as environmental ministries, Green parties, laws, and so on (Frank et al., 2000; Hironaka, 2014), which, in turn, can influence individual actions, including private environmental behaviours (Hadler & Haller, 2011). Following Bromley et al. (2011), who refer to these implementations as the national institutionalisation of environmentalism, one thus can expect that individual environmental behaviours are more pronounced in societies with a strong institutionalisation of environmentalism.

Societies also differ in the location and organisation of political sovereignty and the associated roles of individuals in the political process (Jepperson & Meyer, 1991; Schofer & Fourcade-Gourinchas, 2001). Location differentiates between the state and the civil society as origins of political sovereignty (statism dimension). Organisation distinguishes between the incorporation of social actors in the polity process as either collective groups or as individual actors (corporatism dimension). The United States is an ideal representation of polities with a low statism score and a strong individual focus, whereas Germany represents the opposite poles of strong statism and an emphasis on groups. Scandinavian countries share characteristics of the more liberal countries as far as the location of the political sovereignty and the emphasis on the civil society are concerned.

These differences in the location and organisation of political sovereignty are reflected in the initial experiences of the environmental movements in different countries as described by Dryzek, Downes, Hunold, Schlosberg, and Hernes (2003). Environmental movements in Germany were confronted with a paternalistic approach by the civil servants, whereas those in the United States experienced the most inclusive approach. Considering a broader sample of countries, Schofer and Fourcade-Gourinchas (2001) have shown that activism in new social movements, with their focus on topics such as the environment and human rights,³ is more common in societies with less emphasis on the state, whereas the corporatism dimension was more important for gaining support for older social movements. Following these considerations, we can hypothesise that environmental actions will be more common in societies that emphasise the role of the individual.

Finally, we also need to consider that the network of international organisations itself is geographically dispersed, characterised by growing fragmentation and varying regional densities and levels (Beckfield, 2010; Vasi, 2007). Longhofer and Schofer’s (2010) study of the proliferation of environmental associations showed in this regard that national groups appeared mostly after the arrival of international organisations in peripheral countries. Their findings suggest that domestic political structures and movements more strongly influence environmentalism in Western countries whereas global ties are more important in peripheral countries (see also Hironaka, 2002). Shorette (2012) also finds regional differences between core, semi-peripheral, and peripheral countries concerning the effects of WS integration on the use of pesticides and fertiliser. The effects, however, are the strongest in core countries, a finding that differs from the previous conclusions. Asserting that the role of associations is more important for behaviours than outcomes such as pesticide use, we can expect that the effects of international organisations on behaviours are stronger in more peripheral areas.

The review of these findings on different effects leads to the following four additional hypotheses, in addition to the basic hypothesis on exposure:

International differences in environmental outcomes and behaviours are also discussed with a view on world polity and the position of a country in the core, semi-periphery, or periphery. Andrew Jorgensen and other scholars have combined WS ideas – in relation to Wallerstein’s foundational work – with the effects of economic globalisation and the relational economic position of a country (Givens & Jorgenson, 2013; Shandra, Leckband, McKinney, & London, 2009; Shorette, 2012). This research points to an ecologically unequal exchange between countries due to economic processes, which has independent effects from international organisations. More developed countries are able to shift the environmental burden to less-developed countries, which consequently face negative effects such as biodiversity loss (Shandra et al., 2009), air and water pollution (Jorgenson, Dick, & Mahutga, 2007, 2011), and other damage. The effects of these economic processes, however, are not limited to environmental harm, but also shape individuals’ environmental concerns. A country’s economic background thus needs to be considered as a control when studying the effects of international ties.

A nation's economic background and available resources are also relevant for other theories, which need to be considered as alternative explanations. These include (a) Inglehart's postmaterialism thesis with its focus on the impact of values on environmentalism (Inglehart, 1995), (b) the related discussion of the competing effects of subjective values and actual levels of environmental pollution (Best & Mayerl, 2013; Brechin, 1999), (c) the prosperity thesis and its emphasis on individual and societal wealth (Fairbrother, 2013; Franzen & Meyer, 2010); and (d) social movement theories which consider available resources and mobilising structures (Gillham, 2008; Smith & Wiest, 2012). The reference to social movement literature also points to another control variable – different political opportunities that influence the success of movements (Kriesi, Koopmans, Duyvendak, & Giugni, 1995; McAdam, Tarrow, & Tilly, 2001; Shandra, Esparza, & London, 2012).

The previous section has emphasised developments at the macro-level and how different contextual characteristics such as the number of organisations in a country, the types of ties to the global level, and so on, can influence the diffusion of cultural models. This section turns towards the processes at the individual level and suggests several ways in which individual characteristics influence the reception of global ideas. In doing this, it emphasises two distinct elements – firstly, more passive individual characteristics that are a proxy for exposure to global ideas and, secondly, more active individual actions, involving elements of an inhabited institution approach (Hallett, 2010).

As for the reception of global ideas, research has identified several traits that increase the likelihood of adoption. A first characteristic is education. Global scripts have been increasingly incorporated in curricula, as can be seen in the appearance of environmentalism in school textbooks around the world (Bromley et al., 2011). The educational system thus is an important factor that exposes individuals to global ideas. Adoption is also likely to be greater among younger respondents as global environmental models have developed from the 1970s onwards. Finally, individuals in urban areas are more likely to be exposed to global ideas due to the presence of international organisations in cities. Accordingly, education, age, and urban residency are frequently used as indicators of exposure to global models (see Hadler & Haller, 2011; Pierotti, 2013; Velitchkova, 2015; Zhou, 2013).

Alongside these aspects of exposure, one also needs to consider the role of the individual. WS emphasises that dominant cultural models prescribe an agentic role to actors. Meyer (2010) discusses the historical shifts in roles, from peasants who have only little freedom in their choices, to citizens who are allowed to make choices for themselves, to contemporary individuals who are agentic actors and make choices for themselves and also act on the behalf of other individuals and general principles. The implementation of global cultural models is thus not just a passive process of reception, but may also lead to deliberate actions. Hallett’s (2010) ethnography of an elementary school describes how a newly hired director consciously implemented the cultural script of accountability and even imposed it on other employees. Similarly, Tsutsui and Shin (2008) have shown how local human rights activists appeal to international organisations to pressure their governments. As for environmental behaviours, one thus can expect that they are also positively influenced by an individual’s agency.

The discussion of changing roles of an individual is not limited to WS. The evolving emphasis on the individual is discussed in detail by Ulrich Beck (Beck, 2014; Beck & Beck-Gernsheim, 2001), who pointed out that individualisation refers to the understanding that social rights centre increasingly on individuals, to the erosion of traditionally fixed social roles, and to the growing importance of self-interest in individual decisions. Individualisation is thus considered as linked to a diminishing importance of traditional institutions such as religion, political parties, or family, which is reflected in declining political participation rates (Dalton, 2008) and retreat from different forms of civil associations (Putnam, 2000). Equally important for environmental issues is the increasing relevance of self-interests. As discussed in the tragedy of the commons (Hardin, 1968), emphasising self-interest and ignoring collective needs can result in overuse of common goods such as clean air. Beck (2014), however, also points out that individualisation can lead to a privatisation of political activism, in the sense that ‘the world is becoming an everyday duty in the form of sorting household waste or politically correct consumer decisions’ and that individuals who are most affected by these trends are frequently actively engaged in civic organisations and thus display an altruistic individualism (Beck, 2014, p. 94f). This more detailed discussion points to different aspects of individualisation: negative effects on environmental behaviours due to the erosion of embedding in traditional institutions (structural individualisation), negative effects of self-interests, yet positive effects due to an agentic role and individual altruism in terms of engagement in civil society.

The discussion of traditional institutions needs to encompass the fact that individuals can be embedded in different micro-institutional settings and that these settings may have varying demands. Meyer (2010), in fact, considers individuals exposed to so many different demands that some degree of non-compliance is inevitable. This idea was empirically explored by Velitchkova (2015) for the case of political violence in a comparative study of 15 sub-Saharan countries. Her analysis showed that respondents who experienced violence in their own family had very salient religious and ethnic identities, held attitudes and values that oppose global models, and were more likely to use political violence. Velitchkova thus concludes that embeddedness in micro-institutional logics and practices that contradict global ideas can lead to a negative stance towards the global dominant cultural models, in this case non-violence. One thus can expect that individuals who hold views that contradict the global dominant models of environmentalism will also be less likely to support environmental organisations.

In sum, this review of previous findings and literature on the reception of global cultural models at the individual level leads to the following hypotheses:

Alongside these WS-related ideas on the link between global models and the individual, alternative views on the determinants of environmental behaviours must also be considered. WS theory is rather silent when it comes to the link between attitudes and behaviours, whereas the relationship between these two characteristics has attracted considerable attention within environmental sociology (see Cleveland & Kalamas 2014; Kollmuss & Agyeman, 2002; Lavelle et al., 2016; Martinsson & Lundqvist, 2010). Environmental behaviours can occur in line with attitudes, but also ‘accidentally’ without having pro-environmentalist attitudes. Further, as discussed within social movement literature, environmental behaviours and actions are also influenced by biographical availability aspects and are less likely to occur whenever an individual faces constraints in work or family life (McAdam, 1986; Xiao & McCright, 2012). In addition, research has also pointed to the influence of various environmental attitudes and values (Dietz, Fizgerald, & Showm, 2005; Dunlap & Jones, 2002; Inglehart, 1995), socio-economic influences (Franzen & Meyer, 2010), and new social class aspects (Kriesi et al., 1995). However, studying these processes in detail is beyond the scope of this article and the analysis thus includes only controls for the main variables considered in these approaches.

The hypotheses are tested using a longitudinal multilevel regression that combines country-level characteristics with individual-level survey data. The survey data were collected by the International Social Survey Programme (ISSP group 2012) in 1993/94, 2000/01, and 2010/11. These surveys are random samples, collected either face-to-face or via mail, and representative of the adult population of each country. Repeated data for all three waves are available for the following countries: Austria, Bulgaria, Canada, the Czech Republic, Germany (East and West), Great Britain, Israel, Japan, New Zealand, Norway, the Philippines, Russia, Slovenia, Spain, and the United States. The total sample size is 57,874 respondents; the average sample size per country and wave is about 1200 respondents.

The ISSP questionnaires contain several repeated questions that are related to public and private environmental behaviours. Behaviours thus are self-reported actions and may be influenced by social desirability and other factors well known in survey research. Public behaviour comprises the items: being a member of an environmental organisation and having donated money, taken part in a demonstration, and signed a petition within the last five years. The answer possibilities were simple ‘yes’ or ‘no’ options. A respondent who was active in at least one of these behaviours was coded 1. Respondents who were not active in any of these behaviours were coded 0. Private behaviour consists of the two items ‘forgoing car driving’ and ‘recycling of waste’ with ‘always’, ‘often’, ‘sometimes’, and ‘never’ as answer categories and ‘no car or driver’s license’ and ‘no recycling available’ as additional options. The two items were reverse coded (always = 4 and never = 1) and each respondents’ replies were summed up and divided by the number of valid answers.

As discussed in the background section, exposure to world societal institutions is more likely among educated individuals, younger respondents, and urban dwellers. Therefore, the items age (in years), education (six-point scale ranging from no formal education to university degree), and the area of residency (rural to urban, three-point scale) are included.

Individualisation is captured in two different dimensions: structural individualisation – eroding ties to traditional institutions – is captured by not being a member of a union, a political party, a religious denomination, not attending church (based on a question about the frequency of church attendance), and not being married or having a partner. The agency dimension is captured by a question asking respondents if it is an individual’s or the state’s responsibility to implement environmental measures. Respondents who opted for the individual agency are coded as 1.

Subscription to conflicting micro-level logics is captured by considering various items on environmental attitudes and mobilisation. Similar to Haller and Hadler (2008) and reconfirmed by exploratory factor analyses, the ISSP items on attitudes are combined to the indices ‘Willingness to make sacrifices’, ‘Fatalism about environmental issues’, and ‘Assessment of environmental risks’. Not to be willing to do anything for the environment, thinking that it is not necessary at all to do anything about the environment, and considering the environment not at risk are indicators of holding views that are opposed or aligned with global cultural models (see  Appendix for more details on these indices).

The following variables are included as controls. Biographical availability aspects are captured by gender and work status (full-time, part-time, student, retiree, homemaker, and a reference group comprising unemployed and disabled). Income, unfortunately, had to be excluded due to limited valid cases. It was not asked in several countries and – adding the missing responses from the countries where it was fielded – is available for only about 48% of the respondents. Among those respondents, income is correlated positively with education and subjective social class. The effects of class and education thus can be used as indicators of possible effects of income. New social movement aspects are grasped by subjective class (six-point scale ranging from lower to upper class) and party affiliation (five-point scale left-right plus an additional dummy variable for ‘not reported’). Inglehart’s (1995) postmaterialism is captured by a categorical variable (materialistic, mixed, and postmaterialistic) based on the items (1) ‘maintain order in the nation’, (2) ‘give people more say in government decisions’, (3) ‘fight rising prices’, and (4) ‘protect freedom of speech’.

Missing answers among the independent variables are treated as follows. When questions were not asked in a substantial number of countries (income, for example), the variable is omitted. If a variable was not asked in one or two countries (class, for example, in New Zealand in 2010), the values are substituted by the mean value across all countries within the given wave. Finally, missing answers within a country are considered as embedded variables in case more than 2% of the respondents had not answered (see Hardy & Reynolds, 2004). This embedded variable method results in two regression parameters: The first parameter indicates the effect of the variable of interest (e.g. education) on the dependent variable. The second parameter indicates the difference between repliers (e.g. education provided) and non-repliers (e.g. education missing) with regard to the dependent variable.

The core variable of the present analysis, ties to WS, is measured in several ways. First, by the commonly used indicator ‘logged count of International Governmental Organisations (IGO) and International Non-Governmental organisations (INGO)’ defined as the ‘number of organisations of which a country or a territory is a member, whether directly or through the presence of members in that country’ (Union of International Associations [UIA], 2007, p. 41). Organisations of type A–D are considered and data are taken from various yearbooks of the Union of International Organisations. Alternatively, a more specific linkage via Environmental International Non-Governmental Organisations (EINGO) – provided by David Frank – was considered, but dropped due to a strong multicollinearity with the INGO variable and its availability only up to 2005. All INGO and IGO measures are included as the logged number of organisations within a country, but were also tested as density measures that are standardised by the population size as used by Givens and Jorgenson (2013). These density measures resulted in similar outcomes and are thus not reported.

The national institutionalisation of environmental measures is captured by an index similar to Bromley et al. (2011) and also labelled ‘national environmentalism’. It includes the strength of the different national Green Parties (percentage of popular vote in national parliament/congress elections), the existence of an environmental ministry within the period of four years before the survey wave, the number of international environmental treaties a country has signed at the time of the survey (United Nations Environment Program [UNEP], 2014), and the percentage of protected land- and/or maritime areas (UNEP, 2014). Each sub-index is standardised and the mean value across the four items calculated.

The structural socio-political demands on participation are captured following Schofer and Fourcade-Gourinchas (2001). Canada, Great Britain, New Zealand, Norway, and the United States are classified as ‘low-statism’ regimes that emphasise civil society; Austria, Spain, and West Germany as ‘corporatist/welfare’ regimes that demand less-civil participation; Bulgaria, the Czech Republic, East Germany, Slovenia, and Russia as post-communist regimes that demand the least participation; and Israel, Japan, and the Philippines as the residual category of ‘other’ regimes. Further, countries are also split into core and semi-peripheral countries as the repeated design does not include peripheral countries. Core is defined as being an OECD member before the first survey was conducted in 1993 and excludes all former communist countries of Eastern Europe, Israel, and the Philippines.

As for the social movement aspects, POLITY IV data were considered to capture various aspects of the political opportunity structure. However, the country sample of the present study does not result in sufficient variation among these indicators due to a strong overlap with communist history and the characteristics discussed in the previous paragraph. The analysis thus is limited to the regimes’ variables. As for resources, country-level resources are captured by the level of development (GDP per capita in US$1000) based on UN (2015) data. In addition to the resources and regime variables, environmental threats are also included. For this purpose, two pollution measures were captured for each country: logged CO₂ output per capita and air pollution as particulate matter exposure of the population (World Bank, 2014). Finally, following Jorgenson, Dick, and Shandra (2011), the economic position of a country was also captured in two ways: inward foreign direct domestic stock as percentage of GDP and the importance of the primary and secondary sector as percentage of goods within the total exports. These data were derived from the UN Conference on trade and development (2015).

The data are hierarchically structured in individuals, countries, and survey waves with different respondents in each wave. This design can be analysed using multilevel modelling with country invariant characteristics such as a communist history as level 3, country characteristics that change over time such as the number of INGOs at a given wave as level 2, and respondents’ answers as level 1 (see Deeming & Jones, 2015). This approach allows distinguishing between the cross-sectional effects and the longitudinal effects of a variable such as the overall level of wealth and its changes over time.

A logistic model is used for public behaviour, as this variable is 0/1 coded. Private behaviour, on the other hand, is an interval scale that ranges from 1 to 4 and is analysed using a linear model. The reported B-values (see Tables 2–4) for both behaviours thus cannot be compared directly in terms of their strength. The significance and direction, however, can be used to assess differences in their influence on the two different behaviours. Furthermore, the level one variance is set to 1 in logistic hierarchical regressions with a logit link (see Guo & Zhao, 2000) and thus only the variances at level two (wave effects) and level three (country effects) can be presented for public behaviour.

Finally, the sample size in terms of included countries is at the lower end of the required macro-level units of multilevel models (Stegmueller, 2013). In order to validate the reported results, several alternative specifications were tested: (1) regressions with a single upper level of 48 country*wave units, (2) regressions with fixed country effects and dummy variables for the survey wave, and (3) models in which the sample was also divided into smaller subsamples in order to test for possible distortions resulting from outliers. These alternative models resulted in similar outcomes and thus underscore the reliability of the models presented in this paper.

After considering differences in environmental behaviours, Table 1 provides an overview of the changes in contextual characteristics across all countries. It shows that the presence of IGOs and INGOs and thus the ties to global cultural models have increased since 1993. INGOs, for example, have grown by 36% from 1993 to 2000 and by 56% from 1993 to 2010. The increase in national environmentalism, the institutionalisation of these cultural models within societies, is even more pronounced, with a growth of 80% from 1993 to 2000 and of 94% from 1993 to 2000.

As for the macro-level controls, the share of inward direct foreign stocks has increased substantially by 361% from 1993 to 2013, whereas the share of goods among the total exports has remained stable. A significant increase is also visible in the level of affluence, with a GDP per capita growth of 110% since 1993. Finally, as for environmental threats, a small decline in CO₂ output per capita and a considerable decline in exposure to particulate matter can be observed.

This first descriptive analysis reveals quite diverse trends in environmental behaviours and suggests that a proliferating international regime does not necessarily correlate positively with all types of environmental behaviours – given that public behaviour has declined in most countries. However, it also shows that the increasing presence of world societal institutions is accompanied by changes such as increasing wealth and foreign investments and reductions in environmental pollution. These effects need to be controlled for when assessing the proposed hypotheses. The following section thus presents the results of various hierarchical regressions that consider contextual and individual characteristics in explaining environmental behaviours.

Table 2 shows the effects of the micro-level variables and starts with reporting the effects of the proxies for exposure to global cultural models. Greater education has a significant positive effect on public behaviours, but no effect on private behaviour. Age is negatively associated with public behaviour and positively with the private one. Urban residency has positive effects on both behaviours. The effects of the main indicators of exposure as used by Pierotti (2013) and Zhou (2013) and others thus are in line with hypothesis 6 only as far as public behaviour is concerned, whereas their effects on private behaviour are inconsistent – a finding that will be taken up and discussed in more detail later on.

Considering the effects of different indicators of individualisation as discussed in hypothesis 8, distance to traditional institutions in terms of not having a political party preference, not being a union member, and not attending church has the expected negative effects on both behaviours and thus supports the view of an ongoing general decline of association (Putnam, 2000). On the other hand, distance in terms of no religious denomination and being single affect only one dimension and these effects, in addition, are contrary to this hypothesis. As for subscription to individual agency (H7), it has no effects on private behaviour and negative effects on public behaviour. This finding does not support the hypothesis on individual altruism (Beck, 2014). It is rather in line with the third element of Beck’s individualisation – increasing self-interests and less interest in common goods as discussed in the tragedy of the commons (Hardin, 1968). These findings on individualisation offer only limited support for the initial hypotheses on individualisation. The effects of individualisation, however, will be considered in more detail when different interaction models are discussed.

Finally, as for subscription to views that are opposed to or aligned with global models of environmentalism – summarised as environmental mobilisation indicators and in hypothesis 9 – the models show that alignment with the globally dominant views of being willing to make sacrifices, considering the environment at risk, and not being fatalistic about possible interventions results in positive effects on both behaviours. This finding is clearly in line with the hypothesis that embedding in micro-logics can reinforce or hinder the adoption of global cultural models thus confirming Velitchkova’s (2015) findings on political violence.

As for the controls, the individual-level variables indicate some differences between men and women with regard to public behaviour, and various effects of work status characteristics on both behaviours. As for new social movement aspects, negative effects of social class on public behaviour and positive effects on private behaviour can be reported. Furthermore, respondents who have a preference for left parties as well as postmaterialists are also more likely to show public and private environmental behaviours.

These micro-level models (level 1) are now extended by adding various macro-level variables. All macro-variables that vary over time are included in two ways: as a country-level indicator (level 3) that captures differences across countries and a growth indicator (level 2) that captures effects over time within countries. Variables that do not change over time, but characterise different country groups such as political regimes or core versus semi-periphery, are included only as level three variables. In order to cope with the small statistical power, macro-level variables are, initially, included as single variables in addition to the micro-level model. These preliminary macro-level results are reported in Table 3, before combinations of different macro-level variables and controls are presented in Table 4 and discussed in relation to the hypotheses.

The presence of INGOs – the first variable that captures ties to WS – has a negative effect on public behaviour at the wave level and no effects at the cross-national level. Growing numbers of INGOs thus are associated with declining public behaviour. Its effects on private behaviour are significant at both the time and country level. Private behaviour is more common in countries with many ties to WS and increases with growing numbers of INGOs. The second variable that measures ties to WS – the number of IGOs – has only one significant effect: growing numbers of IGOs are associated with increasing private behaviour. The national institutionalisation of environmentalism is significant only at the country level, with public and private behaviour being more common in countries which are characterised by a stronger institutionalisation. The regime variables are significant only for public behaviour and indicate that this behaviour is more common in countries that have a strong civil society focus and thus low statism. Finally, differentiating between core and more peripheral countries shows that both behaviours are more prevalent in core countries.

The macro-level control variables show that public behaviour is more common in affluent societies, but decreases with growing wealth. Private behaviour is also more common in wealthy societies and increases even further with growing affluence. As for the indicators of a country’s position in the global economic system, only foreign investments have a significant effect and indicate that public behaviours decline with increasing investments while private behaviours increase. As for the pollution measures, CO₂ does not have any significant effects. PM10 exposure, on the other hand, is associated with lower levels of public and private behaviour at the country level. Increasing levels of PM10 exposure have different effects on both behaviours and result in more public, but less private behaviour.

Following these models with single macro-level indicators, Table 4 presents models that combine different macro-variables relevant for testing the proposed hypotheses. GDP is the sole control variable, as preliminary regressions combining the different macro-level controls showed that only GDP remains significant and renders the indicators of economic position and pollution insignificant. Furthermore, IGOs are not included in this table, as they did not have any significant effects after controlling for societal affluence.

The models for public behaviour (1–4) show that the negative wave effects of INGOs (Model 1), the country-level effects of national environmentalism (Model 2), as well as political regimes (Model 3) and geopolitical region (Model 4) remain significant when controlling for societal wealth. Public behaviour thus declines with increasing numbers of ties to WS and is higher in countries that have a strong institutionalisation of environmentalism, in polities which support individual participation, and in core countries. Models that combine more than two macro-level variables exceed the statistical power of our data. Yet, a model – not shown in Table 4 – was estimated that combines all factors. Whereas the significance of some indicators changes, this model, at the very least, confirmed that all indicators retain the direction of their effects.

The models for private behaviour (5–8) show that after controlling for societal wealth, only national environmentalism remains significant (Model 6), whereas INGOs, regimes, and the geopolitical location lose significance. Private behaviour thus is more common in countries with a strong institutionalisation of environmentalism and increases with growing societal wealth. Given these few significant effects, no additional combinations of variables were tested for private behaviour.

These results provide limited support for the basic hypothesis (H1) that environmental behaviours are influenced by the presence of international organisations. The number and growth of INGOs have a positive effect on private behaviour, but these effects lose their significance once controlled for societal wealth. INGOs have no significant effect on public behaviour at the country level and their growth effect is even negative in this dimension. The basic exposure model, which was used fruitfully by Givens and Jorgenson (2013), Marquart-Pyatt (2012), Zhou (2013), and others, thus cannot be translated directly to these two environmental behaviours.

Mixed support for H1 was anticipated and several additional explanations were proposed in the background section. As for the hypothesis on the different effects of INGOs and IGOs, based on Tsutsui’s (2004) findings, IGOs do not have significant effects, while INGOs exert several significant influences. H2 thus also finds only some support. Another factor is the national institutionalisation of environmentalism (H3). WS scholars consider this institutionalisation an outcome of global forces, which, in turn, can result in indirect effects of global cultural models on societal outcomes (Longhofer & Schofer, 2010). Table 4 shows that this institutionalisation indeed has positive effects on both behaviours, even when controlled for other societal characteristics – a finding that clearly supports the related hypothesis. At the national level, the degree of statism and the associated focus on civil society inherent in different political regimes matters (H4). The analysis shows that a strong focus on civil society leads to increased public behaviour and supports the findings of Schofer and Fourcade-Gourinchas (2001). However, it does not affect private behaviour and thus calls for a more differentiated view on the effects of this regime characteristic. Finally, the hypothesis (H5) on differences between core and more peripheral countries is supported in the sense that both behaviours are more common in core countries. Further tests – not shown in Table 4 – indicate, however, similar effects of INGOs in core and semi-peripheral countries. Longhofer and Schofer (2010) and Shorette’s (2012) findings on differential effects in core and semi-periphery are thus supported only at a very general level.

Several of the proposed hypotheses have an underlying understanding that the effects of individual-level variables are mediated by different societal contexts. For this reason, several interactions were tested in addition to the models presented in Tables 3 and 4. The findings of these additional models are summarised in the following paragraphs⁴ as they provide further insights into the links between global cultural models and individual behaviours beyond the simple associations expressed in the hypotheses.

A first extension revolves around the idea that the exposure proxies – education, age, and urban residency – have different effects depending on the number of ties between a society and global cultural models. Therefore, interactions between these exposure variables and the number and the growth of INGOs were considered. As for public behaviour, the effect of age does not interact significantly with these variables and remains negative regardless of the presence of international organisations. The models, however, show that the effect of urban residency becomes more positive with greater numbers of INGOs and that the positive influence of education increases with greater and growing numbers of INGOs. These two significant interactions thus clearly support an interpretation of a diffusion of global cultural models via residency and education as far as public behaviour is concerned.

As for private behaviour, the effect of residency has no significant interactions and remains positive. The effect of age, on the other hand, becomes more positive with greater numbers of INGOs. Education has positive effects in societies with few and rather stagnant numbers of INGOs, which diminish in societies with many INGOs. Again, these interactions support the influence of global ties via education and also age. As for education, however, a ceiling effect seems to occur in the sense that education becomes less central once a certain number of global ties is reached.

Another aspect is the effect of individualisation. WS theory considers individualisation and increasing individual agency a centrepiece of the Western cultural account (Meyer, 2010). Hence, it seems plausible to test if these traits are reinforced by international institutions. As for public behaviour, these additional analyses were not able to find any significant interactions between the number of INGOs, their growth, and the effects of individual agency. Regarding private behaviour, however, the effect of individual agency interacts significantly with INGOs and becomes more negative with both larger numbers and growing numbers of these organisations. Better ties to global cultural models thus reinforce the negative effects of individual agency in private behaviour.

The effects of individualisation can be also influenced by the national institutionalisation of environmentalism and a polity’s focus on civil society. The interactions with the national institutionalisation of environmentalism are significant for both behaviours – individual agency has a (stronger) negative effect on public and private behaviour with an increasing importance of environmentalism at the national level. As for a polity’s emphasis on civil society, the models were able to identify a significant interaction for public behaviour, where the negative effect of individual agency is even stronger in polities with a focus on civil society. For private behaviour, on the other hand, no significant interactions can be reported. The significant interaction effects thus point in a similar direction – whenever responsibility is transferred to the societal level, individual agency has stronger negative effects. This finding supports the interpretation of individual agency effect as a self-interest/tragedy of the commons issue. However, considering that the national institutionalisation of environmentalism itself has an overall positive effect, this lowering individual-level effect is offset by the societal responsibility.

Finally, as for the numerous indicators of structural individualisation, interactions were tested by running separate models for core countries and semi-peripheral countries and thus comparing the effects of global ties in better and less well-connected countries. The effects of the various indicators of distance and closeness to traditional institutions such as family, political parties, and religion were similar in both country groups – a finding that refutes the idea of differential effects on environmental behaviours of these structural individualisation characteristics in core and semi-peripheral countries. Again, previous findings on different effects in core and peripheral countries are only broadly reflected in the sense of overall difference in levels of behaviours in these two country groups.

The aim of this paper was to study the question of how global cultural models shape individual environmental behaviours and to better conceptualise the macro–micro-link within WS theory. The literature review pointed out that WS-related research on the influence of global cultural models on individuals often subscribes to a simple exposure model which asserts that behaviours and attitudes align with global cultural models when many international organisations are present. Using individual-level survey data from 16 countries collected by the ISSP, the first descriptive results presented in this article, showed divergent trends in public and private environmental behaviours – declining trends in the former and increasing trends in the latter – and thus indicate that these two behaviours are influenced differently by ties to global cultural models.

The effects of these ties were further scrutinised in multilevel models using a novel cross-sectional longitudinal model that combines individual-level data with contextual characteristics such as the presence of international organisations, the national institutionalisation of environmentalism, and polities’ demands on individual participation. This analysis found only limited support for the hypotheses on the direct positive effects of these ties: ties to WS have positive effects on private behaviour at the national level, but lose their significance once controlled for societal wealth. Over time, negative effects on public behaviour can be reported and positive effects on private behaviours. The latter, however, also lose their significance once controlled for affluence.

These findings do not bode well for the WS narrative of positive and direct effects of global ties on individuals, as found in previous research on the impact of INGOs on environmental concerns and behaviours (Hadler & Haller, 2011; Marquart-Pyatt, 2012). Does the absence of direct effects mean that global cultural models do not have any influence on public and private behaviours and that these two behaviours are decoupled from global models? The answer would be yes, if we only considered their direct effects discussed so far and ignored indirect effects and interactions with different individual-level variables.

As for such indirect effects, WS scholars consider the diffusion of global models a multifactorial process with numerous little effects, comparable to a bee swarm (Hironaka, 2014). One outcome of these processes is an implementation of global models in national polity and institutions (Meyer et al., 1997). With a specific view on environmental topics, such an institutionalisation of environmentalism can be seen in the form of laws, ministries, and other polity characteristics (Frank et al., 2000; Hironaka, 2014; Longhofer & Schofer, 2010). The present analysis shows indeed that the indicator of these national implementations has significant positive effects on both public and private behaviours and thus offers support for this argument of indirect influence – especially given that there is also a positive association between the presence of INGOs and the extent of the national institutionalisation.

Indirect effects of ties to global models are also possible via interactions with the proxies of individual exposure to global models – education, age, and urban residency. Whereas the effects of these exposure variables differ for private and public behaviour, the interactions show that the presence and also the growth of international ties enforce the effects of these exposure variables. Together with the direct effects of these exposure variables, these indirect effects of ties to global models clearly support the views of Pierotti (2013), Velitchkova (2015), and Zhou (2013) that global ideas also find their way to the individual via education and other exposure proxies.

Alongside exposure, the impact of global ties is also paralleled by other forces. The analyses showed that the socio-cultural focus on civil society, reflected in different regime types, matters for public behaviours and thus confirmed Schofer and Fourcade-Gourinchas’ (2001) view that activism in new social movements is particularly high in polities with a low level of statism. The regime types, however, do not have any significant effects on private behaviour. Private behaviour, at the macro-level, was predominantly influenced by the national institutionalisation of environmentalism. Considering Cole and Ramirez’s (2013) summary that institutions have a greater impact, whenever sanctions are attached – and assuming that national implementations such as mandatory recycling programmes are enforced – suggests that private behaviours are more driven by normative aspects of institutionalisation, whereas public behaviours may be rather influenced by cultural models of participation (see Scott, 2014 on the difference between cultural and normative pillars of institutions).

In addition to these macro-level processes, the micro-level findings indicate a strong coupling between behaviours and the different attitudinal dimensions of environmental mobilisation. These results are in line with Velitchkova’s (2015) view that an embedding in micro-institutional logics is an important factor in addition to exposure. Given that the effects of these attitudinal variables on behaviours are very consistent, the present analysis thus does not indicate any gap between environmental attitudes and behaviours. These results differ from Boyle, McMorris, and Gómez’s (2002) finding which showed that individual attitudes are rather in line with global models, whereas behaviours are not. However, we need to bear in mind that public and private behaviours are self-reported and respondents are inclined to present themselves in a favourable light and that the regression model further simplifies real world processes. Thus, keeping Hallett’s (2010) ethnography of institutional work in mind, one can easily envisage conflicting situations in which individuals are exposed to contrarian pressures and views from their immediate context.

A more differentiated view is also necessary in terms of the effects of individualisation processes. Individualisation and the role of individuals were addressed in the background section in three different ways (Beck, 2014; Beck & Beck-Gernsheim, 2001; Meyer, 2010): as a global historical process of prescribing increasing agency to individual actors, a picture of a general atomisation syndrome of detachment from traditional institutions, and an increasing importance of self-interest in individual decisions. Among these effects, some models indicate that ties to global cultural models enforce the impact of individual agency – which is line with the WS narrative. Their effects, however, are negative on both behaviours – a finding which refutes the idea of an altruistic individualism. This puzzle can be solved by considering the third element of individualisation – increasing self-interest – and relating it to the tragedy of the commons (Hardin, 1968). Public behaviour is a type of collective behaviour as it is related to collective actions. Private behaviour – forgoing car driving – partially also addresses a similar issue, as continuous drivers also win when a single individual forgoes car driving. So, while there is an indication that global forces affect both behaviours, their effects may even be counter-productive – especially if a successful influence of international organisations results in less public behaviour – the behavioural dimension which also includes measures of support for these organisations.

Before turning to some final conclusions, consider possible differences between the findings presented here and previous research, due to the nature of the data and the analysis strategy, which are particularly obvious in public behaviour. The empirical analysis splits the effects of international organisations into a time-invariant country-level effect and a wave effect that captures the impact of growing numbers of organisations within countries (see methods section). The multilevel models presented were able to identify a significant negative effect on public behaviour from the proliferation of organisations, but failed to identify a significant effect at the country level. These findings differ from the results of Hadler and Haller (2011), Marquart-Pyatt (2012) and Givens and Jorgenson (2013), who found overall positive effects of WS ties on environmental concerns. The latter studies, however, were cross-sectional only and thus not able to capture the longitudinal effects of growing numbers of organisations. The difference between cross-sectional and longitudinal effects becomes even more striking when comparing the present results to the findings of Hadler and Haller (2011). They used the same data-set and a similar dependent variable, but restricted it to the second wave of the ISSP data. Limiting the current analysis to the second wave of ISSP data also results in a positive effect of INGO on support (B = .816, SE = .317, not shown in the tables), which suggests that a reason for the different findings is the use of longitudinal data.

In sum, this article adds to our knowledge in both methods and theory. As for the methodology, the results underscore possible differences between cross-sectional and longitudinal effects when studying the influence of ties to WS on individual attitudes. In terms of the influence of global cultural models on individual behaviour this article shows that exposure to global ideas is an important factor in the diffusion of global ideas to individuals. The direct effects of ties to WS, however, are limited. The results rather suggest indirect effects via the institutionalisation of environmentalism at the national level and via individual exposure. In addition, both behaviours are also positively influenced by attitudinal alignment with global models and negatively by various individualisation processes. This article thus paints a rich picture of how global cultural models shape individual behaviours. The current empirical analysis, however, is limited to environmental behaviours and a small number of countries. Future research could thus consider similar factors in areas beyond environmentalism, widen the geographical scope by including more peripheral countries, and further scrutinise the conditions and circumstances that foster or hamper the diffusion of global models to the individual level.

No potential conflict of interest was reported by the author.