The data is part of the “Values in Crisis” (VIC) project, an international cooperation initiated by members of the World Values Survey (WVS) group. The project aims to measure values and attitudes in times of crisis and was collected in the spring of 2020 within a time frame of a year in a total of 18 countries.
1 Due to the special circumstances of the pandemic, the sampling of each country is an online panel considering a fixed quota.
2 Most of the countries were surveyed during or after its first COVID wave. The period of data collection reflects a time, which was particularly characterized by uncertainty. There was limited information about the virus itself, no vaccination, and different political responses to the pandemic, ranging from strict curfews to various relaxed approaches. The participating countries depict a small representation of European, Asian, and South American countries: Austria, Brazil, Chile, China, Colombia, Georgia, Germany, Greece, Hong Kong, Italy, Japan, Kazakhstan, Maldives, Russia, Poland, South Korea, Sweden, and the UK (AUSSDA,
2022; World Values Survey Association,
2020; Seymer et al.,
2021). To somewhat contextualize the countries in terms of their environmental attitudes, past data from the World Value Survey (2010–2014) was used. Comparing the means of environmental concern for those countries that appear in both samples shows that the countries in the VIC sample are, on average, countries with very high concern for the environment. This suggests that the countries participating in the VIC project show less variation regarding environmental concern on the aggregated level. In the following, the main variables and country-level variables are presented.
Dependent variable
The main variable used in this paper is the question “He/She/They strongly believe/s that people should care for nature. Looking after the environment is important to her/him/them.” Respondents had to give their answer on a 6-point scale from “very much like me” to “not like me at all”. This variable is referred to as environmental concern in the following because it asks about the importance and concern for the environment. It is also part of Schwartz’s Theory of Values (
1999,
2012,
2017) and isassociated with the scale for universalism. Respondents with a high score on universalism tend to have a higher regard for nature and the environment. For this reason, universalism, as well as other social-altruistic and biospheric values, is often highly correlated with general environmental attitudes (Schultz & Zelezny,
1999; Weaver,
2002).
By reducing the main variable to a single item, the interpretation of relationships can be delivered more precisely and there has been past research using this variable as a single dependent main variable within their models (Givens & Jorgenson,
2011; Šimac et al.,
2021). Furthermore, as the variable is part of Schwartz’ Theory of Values, it has been validated in various cross-cultural studies. Limitations regarding the use of a singular variable should be addressed at this point. Environmental concerns cannot simply be reduced to a single item. Research on environmental concern and environmental attitudes in general point to several different ways of measurements which provided lots of insightful descriptions and explanations within the field of environmental sociology and psychology (Catton & Dunlap,
1978; Gifford & Sussman,
2012; Maloney & Ward,
1973; Stern & Dietz,
1994). At the same time, cross-national research on environmental attitudes sometimes shows high differences in variance of these dimensions, especially when comparing vastly different cultural contexts (Marquart-Pyatt,
2012; Chaisty & Whitefield,
2015; Mayerl & Best,
2019).
Country-level variables
The spread of the pandemic differs from one country to another. For instance, various countries have experienced distinct waves of infection, leading to diverse policy measures to control the spread of the virus. Divergent COVID-19 experiences may affect the attitudes, opinions, and actions of individuals. As a result, a methodological approach has been adopted to tackle contextual differences. Multilevel modelling can help assess individual experiences in different contexts and identify the variations (Twisk,
2010; Bringé & Golaz,
2022). It should be noted that lower numbers of higher units may cause issues when fitting the overall model, affecting the residual variance and the accuracy of the final predictions (Raudenbush,
2008). The literature includes various suggestions for the minimum quantity of higher-level units, but there isn’t a definitive specification of a minimum number (Leyland & Groenewegen,
2020). To overcome these limitations in the available data for 18 countries, a robustness check is applied in the form of a cased bootstrapping (Van der Leeden et al.,
2008). The country-level variables like the Human Development Index (HDI), the unemployment rate, the cumulative infection rate per million, and the Stringency Index are selected and presented in the following section.
Sumner et al. (
2020) estimated severe economic consequences of the COVID-19 pandemic, especially for developing countries. The
Human Development Index summarizes the key elements of human development: life expectancy, education, and the standard of living measured by the gross national income (GNI) per capita (United Nations Development Programme,
2022). Considering the research stated above, highly developed countries would be expected to perform better in a state of crisis by being able to secure and provide stability in times of financial or health risks. Necessities of life are potentially less threatened than in developing countries and therefore the pandemic could have a lesser negative effect on environmental attitudes (Awuh et al.,
2021; Scruggs & Benegal,
2012). The majority of the participating countries have a HDI score above 0.8 with only three countries below. The underrepresentation of lesser developed countries has to be considered through the results.
The
unemployment rate has been a significant predictor of the decline of environmental attitudes in past crises due to its everyday use and common knowledge for a nation’s economic assessment (Kenny,
2020). Because of restrictions and measures to fight the spread of the virus, nations decided on different strategies with one of them being the (complete or partial) lockdown, a shutdown of most public places, and enforced social distancing, to diminish public activities. This was heavily enforced at the beginning of the pandemic and also led to a short-lived collapse of financial returns in the economy and subsequently to the dismissal of many workers. Being in line with past research on recessions, current research estimates that it will take years for the economy to overcome the impact of the pandemic (Ahmad et al.,
2021; Petrosky-Nadeau & Valletta,
2020). Furthermore, the global economy has been under pressure because of the Ukraine war.
For this study, various measurements for the unemployment rate were considered. The aim was to capture the dynamic process of the pandemic and its economic impact, which is why not only the average unemployment rate at the time of the survey was collected for each country, but also comparative values to the year before to make the changes visible. In addition, the highest unemployment rate at the time of the survey was also selected. In the final model, the highest rate and the average change in unemployment within 1 year (2019–2020 to compare before and within the pandemic) were included.
Infection rates are important metrics for observing pandemic activities. As with unemployment, the dynamic structure of the pandemic was considered by selecting different infection rates, such as average infection rates, highest infection rates, or cumulative infection rates per million inhabitants. For the final model, the cumulative infection rate per million inhabitants was selected for each country, as it illustrates best how high the infection incidence was until the start of the survey.
The
Stringency Index records the “strictness of ‘lockdown style’ policies that primarily restrict people’s behavior” (Blavatnik School of Government,
2020) and is part of the Oxford COVID-19 Government Response Tracker Project (OxCGRT). It contains different dimensions of containment and closure like closings of schools and workplaces, other forms of public restrictions, and public information campaigns regarding health systems. It has a scale from 0 to 100; the higher the score of a country, the stricter the lockdown-style policies are (Hale et al.,
2021). For the study, the stringency score of each country was determined at the time of data collection.
Descriptive overview
Table
1 shows an overview of the selected country-level variables. The original dataset contains 18 countries, with South Korea being represented with two samples.
3 For this analysis, only one of the two South Korean samples is used, and Maldives is excluded because of a lack of data at the macro-level.
Table 1
Overview of country-level variables and sample sizes
Central Europe |
Austria | 2018 | 0.922 | 1782.9 | 81.84 | 5.4 | 0.2 |
Germany | 2009 | 0.947 | 1850 | 76.85 | 4.2 | 0.7 |
Greece | 1540 | 0.888 | 275.9 | 84.26 | 16.4 | −1.6 |
Italyf | 1382 | 0.892 | 4211.5 | 93.52 | 9.3 | 0.3 |
Poland | 1000 | 0.88 | 644.6 | 87.04 | 3.2 | 0 |
Sweden | 2554 | 0.945 | 1497.5 | 64.81 | 8.3 | 0.9 |
UK | 2033 | 0.932 | 2560.2 | 79.63 | 4.5 | 0.1 |
Eastern Europe/Central Asia |
Georgia | 1059 | 0.812 | 185 | 100 | 12.1 | 0.5 |
Kazakhstan | 1035 | 0.825 | 224 | 92.13 | 6.1 | 1.3 |
Russia | 1527 | 0.824 | 3387.9 | 68.98 | 5.7 | 1.1 |
East Asia |
China | 3200 | 0.761 | 57.9 | 81.94 | 5 | 0.4 |
Hongkong | 3061 | 0.949 | 690.28 | 66.67 | 5.8 | 2.8 |
Japan | 3000 | 0.919 | 127.7 | 47.22 | 2.8 | 0.2 |
South Korea | 4000 | 0.916 | 219.7 | 82.41 | 4 | −0.1 |
South America |
Brazil | 3543 | 0.765 | 1201.4 | 81.02 | 13.7 | 0.1 |
Chile | 2269 | 0.851 | 32,432.29 | 79.17 | 11.5 | 4.2 |
Colombia | 1730 | 0.767 | 293.6 | 90.74 | 16.1 | 9.5 |
Despite an overrepresentation of countries with higher affluence, the table highlights the differences in a country’s pandemic management. While most of the countries only experienced a small change in unemployment of less than 2% on average when comparing mean percentages from before and after the pandemic started, the infection numbers and lockdown policies differed vastly. European and South American countries suffered from higher infection incidences although the lockdown policies were quite strong, with Sweden being an exception with their looser COVID policy and Colombia having relatively low numbers of cases. The Asian countries differ in the strictness of their COVID policy, but all had small numbers of infections at the time of data collection.