The Five Canadas of Climate Change: Using audience segmentation to inform communication on climate policy

and colleagues [15] describe this segment as individuals who believe in climate change but who tend see it as a distant problem, hence not feeling the urgency to take action. In the spring of 2015, the Pew Research Center reported that, while Canadians and Americans expressed rel- atively high levels of concern about climate change (64% vs. 59%), the latter were considerably less likely than the former to believe that climate change is harming people around the world now (41% vs. 56%) [76]. To some extent, the absence of a Cautious segment in Canada reflects these differences in risk perceptions. These differences notwithstanding, our results are similar to existing studies using climate change audience segmentation techniques, which demonstrate the predictive power of audi- ence segments relative to demographics and other political variables found to be important correlates of climate policy support in the literature [4]. Explaining 26% of the variance in atti- tudes toward carbon pricing, the predictive power of our segmentation is consistent with pre- vious studies [12, 15] where audience membership explained between 14% and 55% of the variance in responses to climate change. In terms of the overall impact of messages on climate policy attitudes, we found that both the “Invest in solutions” and “Relative price” messages had, across the sample as a whole, a positive impact on carbon pricing attitudes when policy stringency was higher. This result ech- oes previous experimental evidence of framing effects found in other studies. For instance, Bei-

ser-McGrath and Bernauer [50] employed a conjoint design and found that information provision about revenue use significantly increased public support for a carbon tax in the United States and Germany, but similar to our study, these framing effects were most pro- nounced at relatively higher levels of policy stringency. This finding may suggest that commu- nication of carbon tax benefits is especially important as carbon prices become more substantial, or as concerns about the cost of living rise. Moreover, it is worth noting that of the three options for revenue use and messaging tested, only the “Equal dividend” message had no overall effect on attitudes. This is at odds with research highlighting the role of equal per capita dividends in building support for carbon taxes [52, 57] though it is in line with evidence sug- gesting that the public generally likes the idea of earmarking revenues to support additional emissions reductions [18, 58–60]. It is also possible that this frame was less effective because of the additional information that was introduced to accompany the treatment (i.e., “The atmo- sphere belongs to everyone, and a carbon price gives a signal to everyone—business and households—to reduce their carbon pollution”). While we included this information in an attempt to make the treatment stronger, we acknowledge it could also have weakened the treatment if respondents reacted negatively to the additional text. Beyond these aggregate effects, we also found that emphasis framing affected the Five Cana-

das in very different–and sometimes unintended–ways. The “Invest in solutions” and the “Rel- ative price” messages had a negative impact on attitudes toward carbon taxes among the Alarmed when policy stringency was low, and a positive impact on attitudes among the Con- cerned when policy stringency was higher. While all messages were effective with the Con- cerned, the “Relative price” was the only message positively affecting both the Concerned and the Doubtful. These findings echo recent research looking at the role of political variables in moderating the effect of carbon tax design on levels of policy support, which found heteroge- nous sub-group effects across individuals with different political ideologies [52] and partisan identities [56, 77]. However, these studies focused specifically on revenue recycling options (e.g., tax rebates, deficit reduction, support for renewable energy technology) and used unidi- mensional segmentation criteria. In contrast, we examined information provision and relative price effects using broader seg- mentation criteria, and showed that communicating around the specific consequences of car- bon taxes for the price of essential goods (like energy) may be more effective than PLOS ONE The Five Canadas of Climate Change PLOS ONE | https://doi.org/10.1371/journal.pone.0273977 November 23, 2022 15 / 21