The increase in political polarization suggests that there may be weaknesses in the traditional model for science communication ¨C the deficit-model - which assumes that increased communication and awareness about scientific issues will move public consensus towards scientific consensus and reduce political polarization around scientific-based policy. In this study, we draw upon the theories of motivated reasoning, social identity, and persuasion to present an alternative theoretical framework for understanding how certain forms of science communication may amplify political polarization and result in "boomerang effects," effects that generate the opposite impact of what was intended, on policy attitudes held by the general public.
As beliefs about climate change become strongly associated with partisan orientations, individuals are more likely to pay more attention to and interpret information about the issue of climate change in ways that reinforce their political beliefs. In addition, within this polarized environment, structural characteristics of messages about climate change may also serve to amplify partisan differences on the issue depending on what elements of climate change are highlighted in the story. An important dynamic in many science messages is the question of who is affected by an identified problem or issue. For example, some stories may focus on the impact of global climate change on local communities or for the United States as a whole, while others may examine the impact of global climate change on groups of people located in areas such as Zimbabwe or Vietnam. To the best of our knowledge previous research has not investigated how the identity of potential victims in science messages regarding climate change may influence audience polarization. We begin to address this research gap with the current study. Specifically, this study tests the hypothesis that political partisanship and the identity of potential climate change victims interact to affect support for climate mitigation policies.
In order to test this interaction, experimental participants (n=240; mean age = 38.42 years; age range = 18 - 80 years; 54% female) were recruited via mall intercepts in an upstate, rural New York state community and assigned to one of two stimulus conditions or to a control condition. In the two stimulus conditions, participants read a simulated news story about climate change; no story was read in the control condition. The simulated new story was designed to be "non-political" as it did not contain any explicit political partisan cues and focused on the potential health impacts of climate change, an increasingly salient and important aspect of climate change. The story discussed the potential for climate change to increase the likelihood that individuals who spend a lot of time working outdoors, such as farmers, will be infected by diseases such as West Nile virus, and included pictures and names of ten farmers who were potentially at risk. The two stimulus conditions differed by whether the farmers were located in the same region where the experimental participants resided or were from a different region (this was done with changes to the story headline, body text, and exemplar names). The news story was generated explicitly for the experiment, but was based on facts reported on by the associated press.
After reading the story, participants were assessed on their policy support for increasing government regulations and taxes on industries and businesses that produce a great deal of greenhouse emissions. An OLS regression was run comparing the stimulus conditions to the control, and found a significant interaction between political partisanship and victim identity. Exposure to the out-group message condition decreased support (b=-.34, p ≤ .01) for climate mitigation as subject political orientations became more conservative. However, political partisanship was only a marginally significant moderator on identification when comparing the in-group message condition to the control condition. (b=-.19, p ≤ .10).
For Democrats the out-group message, and to a lesser degree the in-group message, significantly increased support for climate mitigation compared to the control group. In contrast, for Republicans, message exposure significantly decreased support for climate mitigation policies compared to control. The result was a significant increase in opinion polarization between Democrats and Republicans.
These findings have implications for science communicators and our understanding of how media coverage of climate change is likely to influence public opinion. As climate change is a global phenomena, news stories often highlight the impact that climate change is having and will likely have in the future on different parts of the world. While media messages are often created with an informational, rather than persuasive intent, our results suggest that broad public exposure to news stories discussing the impacts of climate change on other (out)groups outside the United States is likely to amplify the partisan divide on climate mitigation policies.
Science communicators may be effective by focusing on messages that target specific segments of the public and reduce the likelihood of activating unintended constructs. Audience segmentation analysis may provide useful tools for targeting messages to different population segments. In addition, this study suggests that when creating general messages for the public, science communicators and environmental organizations can lower the risk of creating a boomerang effect amongst conservative segments of the population by focusing on local effects and including implications for local areas when discussing the impact that climate change may be having on distant populations. 2010-->