Ten-year panel data confirm generation gap but climate beliefs increase at similar rates across ages

Accumulating evidence indicates that climate change awareness and concern has increased globally, but commentators suggest a climate change generation gap whereby younger people care more about climate change than older people. Here we use a decade of panel data from 56,513 New Zealanders to test whether belief that “Climate change is real” and “Climate change is caused by humans” increased over the 2009-2018 period; and whether changes are uniform across 12 five-year birth cohorts spanning those born from 1936 to 1995. Results confirm a generation gap in mean (intercept) climate change beliefs but not in over-time increase (slope). The generation gap occurs because older cohorts started from a lower initial belief level (circa 2009), but all age cohorts increased their belief level at a similar rate over the last decade; and these results were not qualified by respondents’ gender. The findings offer hope for collective action that bridges efforts across generations.

shows sample size and gender distribution was comparable for both climate change beliefs and across cohorts.
Results from the age-based trajectory model indicate that agreement with the reality of climate change was higher at all age groups than agreement with anthropogenic climate change for both women and men. This pattern of higher agreement levels for the reality of climate change than for anthropogenic climate change is evident when contrasting the darker lines from Supplementary Figures 1 to 4.
Agreement levels regarding both the reality of climate change and its human cause followed a U-shaped trajectory across the adult lifespan for both women and men (see Supplementary  Figures 1-4). Results in Supplementary Table 7 confirm that the longitudinal changes in climate change beliefs did not follow a linear increase but a quadratic change trajectory for both climate change beliefs, as indicated by the statistically significant quadratic slopes (pvalues < .001) for both women and men. There was one exception to this: women also showed a linear increase in the reality of climate change (p-value = .001). Overall, respondents' belief in the reality of climate change and belief in anthropogenic climate change was comparatively higher in early adulthood, lower in middle adulthood, at which point the pattern became higher in older ages; and this was equally observed for women and men. Table 8 indicate that forcing a normative ageing effect to the data was less fitting than a cohort-based trajectory model allowing variability in each of the 12 five-year birth cohorts. Chi-square results in Supplementary Table 9 comparing the two models clearly indicate that the age-based trajectory model had relatively worse fit (reflected by higher chi-square values) for all birth cohorts compared to the cohort-based trajectory model.

Results in Supplementary
We then examined the multi-group cohort sequential models estimating mean-level change over the ten annual assessments in each of the 12 five-year birth cohorts separately for women and men. Similar to findings for the overall sample, results indicate that agreement levels about the reality of climate change and human causation are increasing for people across all age cohorts at about the same rate, but older people are starting from a lower level of belief. As depicted in Supplementary Figure 1, all female age cohorts showed longitudinal patterns of increase for belief in the reality of climate change, except for the 1945-1941 cohort. For men, respondents' belief in the reality of climate change had longitudinal patterns of increase for just over half the age cohorts, except for the 1985-1981, 1960-1956, 1950-1946, 1945-1941 and 1940-1936 cohorts which shows longitudinal stability in agreement levels (see Supplementary Figure 2). Given the visual similarities in change trajectories, we also fitted a model to these data that formally tested whether each birth cohort's change trajectories were equal over time, while allowing their intercepts to vary. Results in Supplementary Table 8 indicate that this model was commensurate with the cohort-unconstrained model, suggesting that beliefs in the reality of climate change had a similar growth rate for all birth cohorts over ten years amongst women and men.
Respondents' belief in the human causation of climate change had longitudinal patterns of increase for over half the female age cohorts, except for the 1990-1986, 1985-1981, 1950-1946 and 1945-1941 cohorts which shows longitudinal stability in agreement levels (see Supplementary Figure 3). As depicted in Supplementary Figure 4, most male age cohorts showed longitudinal patterns of increase for belief in the reality of climate change, except for the 1985-1981, 1950-1946 and 1940-1936 cohorts. Again, we also fitted a model that formally tested whether each birth cohort's change trajectories were equal over time, while allowing their intercepts to vary. Similar to findings for climate reality, results in Supplementary Table 8 suggest that beliefs in human causation had a similar growth rate for all birth cohorts over ten years amongst women and men. Supplementary Tables 10-13 provide more information about these models.
Supplementary Table 6. Age and sample sizes by birth cohort for "Climate change is real" (reality) and "Climate change is caused by humans" (cause) by gender. The youngest age in each birth cohort was taken as indication of age at Time 1.

Sample Sizes
Birth cohorts Supplementary Table 9. Chi-square (χ 2 ) contributions and differences of each birth cohort for "Climate change is real" (reality) and "Climate change is caused by humans" (cause) by gender Chi-square (χ 2 ) contribution differences (χ 2 Δ) between the key competing models are shown in bold. χ 2 A = Chi-square contribution of the age-based (constrained) model. χ 2 B = Chi-square contribution of the intermediate model where intercepts are freely estimated but slopes are constrained. χ 2 C= Chisquare contribution of the cohort-based (unconstrained) model.