بدبینی زمانی و خوشبینی فضایی در ارزیابی های زیست محیطی: مطالعه 18 کشور

Abstract The personal assessments of the current and expected future state of the environment by 3232 community respondents in 18 nations were investigated at the local, national, and global spatial levels. These assessments were compared to a ranking of each country's environmental quality by an expert panel. Temporal pessimism (“things will get worse”) was found in the assessments at all three spatial levels. Spatial optimism bias (“things are better here than there”) was found in the assessments of current environmental conditions in 15 of 18 countries, but not in the assessments of the future. All countries except one exhibited temporal pessimism, but significant differences between them were common. Evaluations of current environmental conditions also differed by country. Citizens' assessments of current conditions, and the degree of comparative optimism, were strongly correlated with the expert panel's assessments of national environmental quality. Aside from the value of understanding global trends in environmental assessments, the results have important implications for environmental policy and risk management strategies.

Results 4.1. Missing data In total, 25 data sets were received and merged into one file. The data were scanned for missing or errant values. Responses were considered missing when respondents (a) apparently misunderstood the scales and consequentially, used incorrect values for their current or future evaluations (e.g., some respondents gave numbers lower than “1” for “current” ratings, or higher than “2” for “future” ratings) or (b) left some parts of the scale blank because they did not know enough about an aspect, or did not believe that it applied to their local and/or national areas (some respondents wrote “N/A” or “don't know” on the scale). A case summary for missing data showed that 971 (or 31%) respondents did not answer, or gave incorrect answers to, at least one of the items. Six hundred and ninety-nine respondents were missing 10% or less of their data. Given the very high internal consistency of the EFS (see below), missing data for these respondents were substituted with their mean responses to that particular subscale. However, those missing more than 10% of their responses (n = 272, or approximately 9% of the total sample) were excluded from further analyses. Research affiliates in Germany elected to omit three items from the EFS (pesticides, fish, and natural disasters), which they deemed inapplicable to their country, and therefore all German respondents necessarily were missing more than 15% of their data. However, rather than excluding German respondents from the analyses, the missing values from these three variables were replaced with respondents' means on the corresponding subscales. Given the very high internal consistency of the entire scale and of each of the six subscales (as described below), the substituted responses probably very closely approximate these respondents' choices, had they answered the questions. After the substitutions, 79 respondents from Germany had no missing data. Of the remaining 32 German respondents, 30 had less than 10% of their data missing, and so mean substitution was used as for the other respondents, leaving two respondents from Germany who were excluded from the analyses. The number of valid cases on each subscale that remained for the analyses, after these substitutions, may be seen in Table 2. Table 2. Descriptive statistics for the EFS subscales. Assessments of N Mean SD Current environmental conditions At the local level 3014 2.93 0.61 At the national level 3014 2.75 0.57 At the global level 2992 2.39 0.54 Expected future environmental change At the local level 2992 −0.28 0.57 At the national level 2994 −0.35 0.61 At the global level 2991 −0.48 0.70 Table options 4.2. EFS internal consistency and descriptive statistics Cronbach's αs for the six subscales on the EFS were as follows: current local conditions α = 0.91), current national conditions (α = 0.92), current global conditions (α = 0.91), future local conditions (α = 0.91), future national conditions (α = 0.92), and future global conditions (α = 0.93). The reliability of the full EFS was extremely high (α = 0.97). Table 2 presents the means, standard deviations, and ranges for the six subscale variables. These means are also displayed in Fig. 1. Means for all current environmental conditions were slightly below the scale midpoint of 3 (“acceptable”), but declined for increasingly distant spatial levels. Mean ratings for expected future conditions were below the scale midpoint of zero (“no different”), and scores were increasingly pessimistic as spatial levels expanded. Specific country means for each subscale are listed in Table 3 and are displayed in Fig. 2. Current local assessments were most positive in Finland (M = 3.59, SD = 0.45), and lowest in Mexico (M = 2.55, SD = 0.52). The future local means were somewhat surprising: For future local means, Romanians were the most optimistic (M = 0.10, SD = 0.60), and Australians were the most pessimistic (M = −0.55, SD = 0.53). Mean ratings of current environmental conditions and expected future change (in ... Fig. 1. Mean ratings of current environmental conditions and expected future change (in 25 years) at the local, national, and global spatial levels summed across countries. Figure options Table 3. EFS subscale means and ESI scores for each country. Country Mean ratings ESI scores Local National Global Current Future Current Future Current Future Australia 3.27 −0.55 2.91 −0.70 2.11 −1.00 61.00 Brazil 3.01 −0.47 2.69 −0.56 2.41 −0.65 62.20 Canada 3.42 −0.42 3.13 −0.49 2.07 −0.82 64.40 England 3.15 −0.32 2.87 −0.35 2.21 −0.58 50.20 Finland 3.59 −0.24 3.62 −0.27 2.43 −0.53 75.10 France 2.95 −0.29 2.65 −0.36 2.03 −0.71 55.20 Germany 3.38 −0.27 3.27 −0.32 2.59 −0.73 56.90 India 2.78 −0.19 2.72 −0.21 2.75 −0.14 45.20 Italy 2.92 −0.25 2.65 −0.35 2.33 −0.49 50.10 Japan 2.81 −0.26 2.61 −0.35 2.34 −0.64 57.30 Mexico 2.55 −0.50 2.26 −0.69 2.30 −0.65 46.20 Netherlands 3.10 −0.30 3.01 −0.35 2.34 −0.62 53.70 Portugal 2.82 −0.18 2.68 −0.23 2.50 −0.28 54.20 The Romania 2.66 0.10 2.62 0.12 2.96 0.32 46.20 Russia 2.51 −0.23 2.56 −0.25 2.63 −0.22 56.10 Spain 2.68 −0.43 2.43 −0.51 2.04 −0.64 48.80 Sweden 3.58 −0.12 3.45 −0.15 2.38 −0.34 71.70 United States 2.91 −0.38 2.69 −0.46 2.26 −0.61 52.90 Table options Mean ratings of current environmental conditions and expected future change at ... Fig. 2. Mean ratings of current environmental conditions and expected future change at the local, national, and global spatial levels for each country. Figure options 4.3. Assessments of current environmental conditions To examine variations across spatial levels and countries among assessments of current environmental conditions, a two-way mixed design ANOVA was conducted, with spatial level as a within-subjects factor and country as a between-subjects factor. Demographic variables (i.e., age, gender, years of education, and years lived in the current area) were entered as covariates. Given the very high internal consistency of the scales, all ANOVAs were conducted on subscale values that were averaged across each respondent's 20 EFS scale items. The means are shown in Table 2. Because Mauchly's sphericity test of spatial level indicated a violation of the sphericity assumption, and given that the Greenhouse–Geisser correction was greater than 0.75, the corrected Huynh–Feldt values were used (Field, 2005). A significant main effect of spatial level (across all countries) was found, F(1.51, 4339.58) = 4703.60, P < 0.001, indicating that respondents assessed the quality of proximal environments more favorably than that of more distant locales. Based on Cohen's (1988) guidelines, this is a medium effect size (f2 = 0.22). Contrasts among the three spatial levels revealed that assessments of local environmental conditions were significantly more positive than those at the national level, F(1, 2859) = 671.02, P < 0.001, an effect size of d = 0.31, and at the global level, F(1, 2859) = 3266.89, P < 0.001, an effect size of d = 0.94. This supports the first hypothesis, that assessments of current environmental quality decrease as spatial level increases (see Table 2). A significant country effect was apparent, F(17, 2859) = 36.74, P < 0.001, which is a medium effect size (f2 = 0.26). This supports the second hypothesis, that when averaged across spatial levels, country membership is related to respondents' assessments of current environmental conditions. The results of Games–Howell multiple comparisons (adjusted α = 0.002) revealed that residents of Finland, Sweden, and Germany made significantly more positive assessments of current environmental conditions than 15, 15, and 14 other countries, respectively. In contrast, residents of Mexico and Spain made significantly more negative assessments than all the countries from which they differed (12 and 14 other countries, respectively). The other 13 countries differed significantly from between three to eight other nations, but these differences were neither as pronounced nor as unidirectional as those for the five countries mentioned above. The complete matrix of national differences in current environmental assessments is displayed in Table 4. Table 4. Significant mean differencesa of current national ratings between countries. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 1. Australia − − + + + + + + − 2. Brazil − − − + − + − 3. Canada + − + + + + + + + + + + − + 4. England − − + + + + + + + − 5. Finland + + + + + + + + + + + + + + + + 6. France − − − + − + − 7. Germany + + + − + + + + + + + + + + + 8. India − − − + − + − 9. Italy − − − − − + − + − 10. Japan − − − − − + − + − 11. Mexico − − − − − − − − − − − − − − − − 12. Netherlands + − + − + + + + + + + + − + 13. Portugal − − − − + − + − 14. Romania − − − − − + − + − 15. Russia − − − − − + − − 16. Spain − − − − − − − − − − − − − − − − 17. Sweden + + + + + + + + + + + + + + + 18. United States − − + − + − a Comparisons are in reference to the country in the left-hand column. Table options 4.4. Does the spatial bias exist everywhere? To examine whether a spatial bias existed in each country, current comparative optimism scores were first computed by subtracting average global from average local EFS scores. Values above zero indicate that local conditions were viewed as superior to global conditions; those below zero indicate that global conditions were viewed as better. Next, one-sample t-tests (Bonferroni adjusted α = 0.002) were conducted for each country to examine whether these scores significantly differed from zero. Fifteen countries manifested significant optimistic spatial biases (i.e., that local conditions are better than global conditions). Interestingly, respondents in Russia and Romania demonstrated significant pessimistic spatial biases: global assessments were significantly more positive than local assessments. Among the 18 nations, only assessments in India exhibited no significant change with spatial level. These trends are illustrated in Fig. 2. 4.5. Assessments of future change in environmental conditions 4.5.1. Temporal trends Next, we examined whether assessments changed from present to future. One-sample t-tests were conducted on each of the future change subscales (at the local, national, and global levels) to evaluate whether or not their means differed significantly from zero, which would suggest the existence of a temporal trend. The means are shown in Table 2. Scores below zero indicate pessimism and those above zero signify optimism. Using a Bonferroni correction, the Type I error rate for each comparison was reduced to α = 0.02. The subscale means reveal significant temporal pessimism at all three spatial levels: the local, t(2882) = −25.63, P < 0.001, d = −0.48, national, t(2883) = −29.59, P < 0.001, d = −0.55, and global, t(2881) = −36.31, P < 0.001, d = −0.68. All three effect sizes are medium to large. Thus, respondents were, on average, pessimistic at all spatial levels in their projections of future environmental conditions. 4.5.2. Temporal trends across countries To test the hypothesis that environmental optimism differs across countries, a one-way ANOVA was conducted on assessments of future environmental change. A significant main effect of country on future ratings supported this hypothesis, F(17, 2838) = 56.50. This effect size (f2 = 0.28), once again, is medium in size. Pairwise comparisons (all Ps <0.002) reveal that, although assessments from every country differed significantly from at least one other country, some countries were more (or less) optimistic than many others (see Table 5 for the full matrix of national differences). For example, respondents from Finland, Germany, and Canada were significantly more temporally pessimistic than respondents from five other countries and, notably, temporal pessimism in Australia exceeded that in 12 other countries. Assessments of the future from Russia and Portugal were less temporally pessimistic than those in seven other countries. Finally, Romania was the only country for which assessments of the future were at all temporally optimistic, and Romanian assessments were significantly more temporally optimistic than those of all other nations. Table 5. Significant mean differencesa of future national ratings between countries. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 1. Australia − − − − − − − − − − − − 2. Brazil − − − − − − 3. Canada − − − − 4. England + + − 5. Finland + + + − + 6. France + + − 7. Germany + + − 8. India + + + + − + 9. Italy + + − − 10. Japan + + − − 11. Mexico − − − − − − − − − − − − 12. Netherlands + + − 13. Portugal + + + + − + 14. Romania + + + + + + + + + + + + + + + + + 15. Russia + + + − + 16. Spain − − − − − − 17. Sweden + + + + + + − + + 18. United States − − a Comparisons are in reference to the country in the left-hand column. Table options 4.6. Environmental assessments by citizens and experts How do these lay assessments relate to those by experts? Mean ratings of current national environmental conditions by citizens was correlated with expert rankings on the ESI, and a strong positive relation was found, r = 0.78, P(one-tailed) < 0.001. In a second correlation, mean spatial optimism scores (average local minus average global) were analyzed in relation to the ESI rankings. Again, a large association was observed, r = 0.68, P(one-tailed) = 0.001. This suggests that, in general, countries with more spatial optimism are also those with better environmental conditions, and supports the final hypothesis, that assessments of environmental conditions by citizens strongly agree with expert assessments of environmental quality.