80 chapter 3 Table 3.3. Descriptive Analyses, Reliability and Bivariate Correlations between the Measures included in Study 2, T2 (n = 332) M SD α ωt 1 2 3 4 5 6 7 8 1. Transilience 5.23 0.94 .92 .94 2. Individual adaptation behaviours 5.90 0.75 .87 .91 .42*** 3. Collective adaptation behaviours 5.36 1.01 .77 .83 .46*** .57*** 4. Well-being 5.53 1.26 .26*** .16** .16** 5. Cognitive coping 6.32 0.98 .43*** .59*** .45*** .26*** 6. Positive personal change 4.36 1.27 .88 .91 .49*** .25*** .34*** .12* .13* 7. CC adaptive capacity 3.92 1.67 .25*** .15** .24*** .10 -.06 .51*** 8. CC adaptation intentions 4.18 1.69 .25*** .27*** .34*** .05 .08 .40*** .60*** 9. Disengagement 2.26 1.83 -.08 -.19*** -.05 .00 -.31 *** .15** .26*** .12* Note. M = mean; SD = standard deviation; α = Cronbach’s alpha; ωt = McDonald’s omega; CC = climate change. *p < .05; **p < .01; ***p < .001 Transilience Predicting Adaptation Behaviours and Well-being across time points Table 3.4 shows the bivariate correlations between all relevant variables for T1 and T2 (based on the merged dataset). Interestingly, people seem to perceive transilience more strongly at T2 compared to T1 (Mdiff = 0.28, t(320) = 5.11; d = .29 p < .001). The positive relationship between transilience and adaptation behaviours and wellbeing, respectively, seemed to be robust across both time points (see Tables 3.2, 3.3, 3.4). To formally test this, we conducted three linear mixed models using the Gamlj module (Gallucci, 2019) in Jamovi. In each model, transilience measured at both time points, time, and their interaction were included as predictors; individual adaptation behaviours, collective adaptation behaviours and well-being from both time points were included as outcome variables, respectively. Subjects were included as random effects to account for the within-subject correlation of the data. Transilience (ICC = .45) was centred at the grand mean. Time was coded according to simple code (T1 = - 0.5; T2 = 0.5) to get the average effect of transilience over the outcome variables across time points. To limit chances of type I error, we applied the Bonferroni correction and adjusted the significance level to p < .016 (i.e., .05/3). As shown in Table 3.5, time had a main effect on all outcome variables, indicating that the average engagement in individual and collective behaviours significantly
RkJQdWJsaXNoZXIy MTk4NDMw