The investigation was conducted through an online survey sent to a random sample from a self-selected panel of respondents administered by the Department of Psychology, University of Gothenburg. The panel, which still exists, consists of a combination of community and students, and participants have had the opportunity to sign-in via advertisements on town property and at the university. The survey was slightly revised after a questionnaire had been pretested on a smaller sample of respondents. The final version of the survey was sent out in April 2017 to a random sample of 2892 individuals from the online panel, of whom 986 took the survey. Potential respondents were informed that there would be a prize draw of two gift cards worth 500 SEK each among those who completed the survey. After cleansing the dataset from respondents that did not finish the survey (the majority of these quit the survey after a few initial questions), N = 880 remained. All participants were informed about the general purpose of the study, along with contact information to the researchers. Further, respondents were informed that participation was voluntary, that their answers were anonymous, and that they could withdraw at any time. The analysis was conducted in SPSS using hierarchical regression analysis.
2.2. Pro-Environmental Behavior
We are primarily interested in regressing pro-environmental behaviors on subjective resources. The idea was to cover the most important aspects of individual everyday behaviors with potential environmental impact, thus excluding less frequent behaviors, such as vacation traveling, buying a car, choosing a home, etc. We therefore chose to investigate grocery shopping, the meat/fish consumption, and everyday travel modes. Furthermore, we distinguished between recent choices (e.g., what one ate yesterday) and typical/habitual behavior (e.g., how often one eats each foodstuff).
Grocery shopping last time was measured by asking respondents to think about the last time they bought milk/dairy products, fruits/vegetables, eggs, flour/grain/cereals/muesli, meat/fish, canned food, oil, and coffee/tea, and indicate whether they bought an eco-labeled product or not. Respondents could answer that they never buy the product in question. Buying non-eco-labeled (i.e., conventional) products equaled 1, and buying eco-labeled products equaled 2, indicating that a higher value corresponds to a more environmentally friendly the behavior. It could be debated whether the third option (3), never buying a product (which is a necessary option for respondents), is part of the same scale. For instance, never buying eggs (because you never eat eggs) could in one way be viewed as the most pro-environmental behavior (it implies less environmental impact than buying any kind of eggs), but taking this into account would be to mix different logics: pro-environmental behavior in terms of buying eco-labeled products and pro-environmental behavior more generally. Only the former category was of interest for this study. Therefore, answers of 3 on these items were recoded as the mean value of the responses on the scale 1–2. The eight items were then averaged into an index variable. Hence, an interval scale ranging from 1 to 2 was created (Cronbach’s alpha 0.756).
Grocery shopping habits were measured by asking respondents how often they buy environmentally labeled milk/dairy products, fruits/vegetables, eggs, flour/grain/cereals/muesli, meat/fish, canned food, oil, and coffee/tea. Answers could range from (1) “I never buy environmentally labeled products from this category” to (7) “I always buy environmentally labeled products from this category”, with the middle option (4) being “About half of all my purchases (from this category) are environmentally labeled”. There was also the possibility (8) of indicating that “I never buy products from this category”, which was recoded as the mean value of the responses on the scale 1–7 (similar to how we processed this answer for the variable grocery shopping last time). The eight items were averaged into an index variable ranging from 1 to 7 (Cronbach’s alpha 0.879).
The variable ate yesterday was intended to measure the consumption of meat and fish, which on a general level is known to have negative environmental impacts compared to vegetables and legumes. Respondents were asked if they ate any of eight types of meat, fish, or shellfish, and if the product was eco-labeled. Not having eaten the product at all scored the highest value (3), and eating a non-eco-labeled product scored the lowest (1), while eating an environmentally labeled product scored 2. We tested this variable with both weighted and unweighted meat/fish items. In the weighted analysis, each item was weighted according to its estimated environmental impact by the National Food Agency [
36]. How to calculate this is not self-evident, given that what is bad from a climate change perspective is not necessarily bad from a biological diversity or an eutrophication perspective. We chose to focus on emissions of greenhouse gases, since climate considerations more likely guided participants’ choices than for example biological diversity or eutrophication. This implied that beef scored 10 times higher in environmental impact than poultry, fish and shellfish did, four times higher than pork and game did, and 1.5 times higher than lamb did. The eight items were indexed into a single variable ranging from 1 to 3 (Cronbach’s alpha 0.625).
Eating habits were measured using the same eight items as in ate yesterday, and we tested both the unweighted and the weighted variable. Here, respondents were asked to indicate how often they eat each foodstuff—ranging from (1) “Every day” to (7) “I never eat food from this category”. The eight weighted items were indexed into a single variable ranging from 1 to 7 (Cronbach’s alpha 0.855).
Turning to transportation issues, travel mode last week was measured by asking respondents by which mode of transportation they mostly employed to travel to work/school/corresponding daily activity during the last week. Possible answers were (1) by car, by myself, (2) by car, with others, (3) by public transportation (bus, tram, railway etc.), (4) by bicycle, and (5) by foot, indicating that low scores were the least environmentally friendly, and high scores were considered the most environmentally friendly.
Travel mode habits were measured by asking respondents to indicate on a scale ranging from (1) “every day” to (6) “never” how often they travel to work/school/corresponding daily activity by the same five travel modes described in the
travel mode last week variable. We averaged these items into a single variable, testing variants with all five items, leaving out “by car, with others” and leaving out “by car, with others” and “by car, by myself”, but all indices scored low on Cronbach’s alpha (see
Table 1), and this variable was hence not included in the analyses.
2.4. Trust
Trust is assumed to be a measure of social capital [
38,
39,
40], and thus important for the level of cooperation in a society. The more people find fellow citizens trustworthy, the more likely it is that they will have positive attitudes to contributing to the collective good. It is quite intuitive that a person who trusts anonymous others will also be positive in her evaluation of resources in terms of SAS. Therefore, in our models, we control for generalized trust as a social resource that may influence subjective evaluations of other resources, as well as subjective well-being.
Generalized trust was measured by asking: “In your opinion, to what extent can people in general be trusted?” Respondents were requested to indicate their answer on an 11-point scale ranging from 0 (“You cannot trust people in general”) to 10 (“You can trust people in general”).
Trust in political institutions is related to generalized trust, and studies show that a positive evaluation of the trustworthiness and legitimacy of institutions plays an important role for the acceptance of and compliance with government rules [
41,
42,
43]. Evidence is strong for the impact of generalized trust in political institutions on environmental policy attitudes [
44,
45,
46,
47,
48,
49,
50]. However, it has also been argued that trust in specific institutions is more important than general trust in political institutions, i.e., that taking a specific institution into account is more important for the understanding and acceptance of specific policy measures [
43,
51] Given this, we find it significant to include trust in environmental institutions as a control variable, since it could correlate with both the behaviors and the subjective resources we investigate.
Trust in environmental institutions was measured by asking respondents to what extent they trust that different authorities and other actors provide correct and objective information about the environmental consequences of people’s everyday behavior and consumption. The actors included were the Ministry of Environment and Energy, Ministry of Finance, Swedish Environmental Protection Agency, Swedish Chemicals Agency, Swedish Agency for Marine and Water Management, Swedish Society for Nature Conservation, and environmental scientists. Answers could range from very low trust (1) to very high trust (5). Answers on these items were indexed into a single variable (where the Ministry of Finance was omitted, since it lowered the Cronbach’s alpha score; the Ministry of Finance was included in the questionnaire since this institution actually influences environmental policy. However, it is unlikely that respondents are actively aware of this fact) (Cronbach’s alpha 0.883).
2.5. Social Value Orientation
For nearly 50 years, researchers have tried to capture people’s motives in social dilemma situations through the concept of social value orientation (SVO), showing how people evaluate the resource outcomes for themselves and others [
52]. A vast literature has investigated the importance of individuals’ fundamental value orientations in this context, concluding that there are three core social value orientations that differentiate people: people who are prosocial, individualists, and those who are competitive [
53,
54]. Others have added altruists as a variation of prosocials in a fourth category [
55]. Prosocial individuals (especially altruists) are most inclined to cooperate in social dilemma situations, while individualists and competitive persons have variations of the pro-selfish characteristic, and thus do not choose cooperative behaviors to the same extent [
56,
57]. As a consequence, social value orientation is included as a control variable in the analyses, since this factor could potentially correlate with subjective well-being, the pro-environmental behaviors, and the subjective resources we investigate.
Social value orientation was measured using a slider measure where respondents were asked to allocate money to themselves and another anonymous person. Depending on how much was given to themselves and the other person over six items, each respondent was given a score ranging from altruism to competitiveness, with prosocial and individualism in between (see [
55] for how to calculate scores).