Australian voters’ attitudes to climate action and their social-political determinants

can be explained, once again, by the strong relationship between age and political party prefer- ence, and the multivariate OLR treating all independent variables as equal (i.e. not recognising causal/path relationships between independent variables). On the willingness to incur a personal cost in efforts to address climate change A majority of respondents (72%) reported that they would be willing to incur some personal cost in return for emissions reductions, but only 15% indicated they would be willing to wear ‘significant’ personal cost. Across political party preference, the proportion of voters who are willing to accept a small personal cost is relatively similar (60% of progressive voters, 55% of conservative voters), but there are differences in distributions in those who are willing to incur a ‘significant’ personal cost, and those who are not willing to incur any personal cost at all (Fig 6). While 26% of progressive voters are willing to incur a significant personal cost, only 5% of conservative voters feel similarly. At the other end of the spectrum, 40% of conservative voters are unwilling to incur any personal cost, but only 13% of progressive voters feel the same. Ordinal logistic regression (multivariate) analysis shows that the variation in responses to the question of willingness to incur a personal cost is best explained by political party prefer- ence (Greens & Labor p < 0.0001), educational attainment (more education p < 0.0001), and living in an area with lower socio-economic disadvantage (regardless of personal status) (less

disadvantage p = 0.0148) (Fig 7). We also found some differences between residents of differ- ent states/territories. Residents of Queensland were less inclined than residents of the Austra- lian Capital Territory (p = 0.0051), Western Australia (p = 0.0394), and New South Wales (p = 0.0338) to accept a personal cost. Residents of the Australian Capital Territory were more inclined to accept a personal cost than residents of Victoria (p = 0.0085) and South Australia (p = 0.0369), as well as residents of Queensland. Fig 5. Forest plots illustrating odds ratios and 95% confidence interval of the odds ratio for Q4: “How much did the issue of climate change influence your vote in the 2019 Federal election? For you personally, would you say climate change was. . .?” Left panels are univariate analyses; right panels are multivariate analyses; upper panels are continuous independent variables; lower panels are categorical independent variables. For continuous independent variables, the variables have been modified where necessary so that the OR is measuring more quantity of the variable listed on the y axis (i.e., more income, more education, more disadvantage in the area of residence, more age). For categorical variables, comparisons are to the default baseline category only (baseline category not displayed in plots: Liberal; Male; NSW; Capital city; English speaking background; Language (English)). Variables that are statistically significant at (α = 0.05) do not cross the OR value of 1, indicated by dashed vertical line, with their 95% confidence

interval. Note the differing scales on the x-axes across plots. https://doi.org/10.1371/journal.pone.0248268.g005 PLOS ONE Australian voters’ attitudes to climate action and their social-political determinants PLOS ONE | https://doi.org/10.1371/journal.pone.0248268 March 24, 2021 9 / 18