UNDERSTANDING EVANGELICAL PROTESTANT IDENTITY, RELIGIOSITY, EXTREME WEATHER, AND AMERICAN PUBLIC PERCEPTIONS OF GLOBAL WARMING, 2006-2016

abstract

This paper presents a comprehensive effort to analyze the relationship between Evangelical Protestant identity and religiosity along with socioeconomic characteristics, political predisposition, and extreme weather events on the one hand, and opinions toward global warming on the other. Using survey data from the Pew Research Center from 2006–2013 and the 2016 American National Election Study merged with extreme weather data from the Center for Disease and Control from 2005–2016, several major findings stand out. First, Evangelical Protestants are less likely to believe in the existence and seriousness of global warming than others. Second, those who are more religious are less likely to view global warming as human caused and a serious problem than their less-religious counterparts. Lastly, the effects of extreme weather represented by heavy precipitation and extreme heat on one’s perceptions of global warming’s existence, cause, and seriousness are largely absent. We suspect that the public awareness of extreme weather events being linked with climate change has been heightened in recent years. More studies using data after 2016 are warranted to examine how extreme weather determines public perceptions of global warming.

Keywords:

• perceptions of global warming
• Evangelical Protestant
• religiosity
• extreme weather events

Acknowledgments

The authors thank the Pew Research Center for making available its data from 2006 to 2013 containing survey item related to global warming seriousness and American National Election Studies (ANES) for making available its 2016 Time Series Study. The authors also thank the Center for Disease and Control for making available its data on extreme precipitation and heat. The authors acknowledge that the data sources bear no responsibility for the interpretations presented or conclusions that we have reached based on our analyses of these data. The authors thank Dr. James C. Garand for his early comments and his work on recoding some of the variables in the 2016 ANES data. The authors also thank Ryan O’Connor for estimating the 115^th Congressional District Population weighted centroids. Wanyun Shao is supported by the early-career research fellowship of the National Academies of Sciences, Engineering, and Medicine Gulf Research Program (grant 2000008396)

Notes

1 We use the terms “global warming” and “climate change” interchangeably throughout the text. However, when explaining the results of our analyses, we use the former term because the survey data employed in our analyses contains survey questions that ask respondents about their opinion on the seriousness of “global warming.” For discussions on the differences in effects of question wording regarding “climate change” and “global warming” see, Villar and Krosnick (2010) and Schuldt and others (2015).

We use the terms “global warming” and “climate change” interchangeably throughout the text. However, when explaining the results of our analyses, we use the former term because the survey data employed in our analyses contains survey questions that ask respondents about their opinion on the seriousness of “global warming.” For discussions on the differences in effects of question wording regarding “climate change” and “global warming” see, Villar and Krosnick (2010) and Schuldt and others (2015).

2 The Pew Research Center has made its data sets from 2006 – 2013 available for the public. The authors have attempted to download the 2015 dataset but found geographic codes finer than state code not publicly available.

The Pew Research Center has made its data sets from 2006 – 2013 available for the public. The authors have attempted to download the 2015 dataset but found geographic codes finer than state code not publicly available.

3 Scholars have also noted the importance of attitudinal variables such as attitudes toward scientists and the influence of the mass media (Shao et. al 2014). Given the limitation of the data over a long time span, we do not examine the impacts of these two groups of determinants. Future studies should consider including a more comprehensive list of independent variables.

Scholars have also noted the importance of attitudinal variables such as attitudes toward scientists and the influence of the mass media (Shao et. al 2014). Given the limitation of the data over a long time span, we do not examine the impacts of these two groups of determinants. Future studies should consider including a more comprehensive list of independent variables.

4 It is the theological belief associated with one’s destiny relating to death or the afterlife, where one’s soul goes after death.

It is the theological belief associated with one’s destiny relating to death or the afterlife, where one’s soul goes after death.

5 We also estimate a series of models by including an interaction of evangelical Protestant and religious service attendance. This interaction is found to be statistically non-significant across all eight surveys except the 2013 November survey. For interested readers, these results can be made available upon request.

We also estimate a series of models by including an interaction of evangelical Protestant and religious service attendance. This interaction is found to be statistically non-significant across all eight surveys except the 2013 November survey. For interested readers, these results can be made available upon request.

6 In addition to including evangelical Protestant as the religious identity variable, we estimate a series of models by including all religious groups such as evangelical Protestants, black Protestants, mainline Protestants, Catholics, Jewish, other religion, and non-affiliated including “atheist” and “agnostics” with mainline Protestants being the exclusion group. Only evangelical Protestants are consistently different from mainline Protestants. Results will be made available upon request.

In addition to including evangelical Protestant as the religious identity variable, we estimate a series of models by including all religious groups such as evangelical Protestants, black Protestants, mainline Protestants, Catholics, Jewish, other religion, and non-affiliated including “atheist” and “agnostics” with mainline Protestants being the exclusion group. Only evangelical Protestants are consistently different from mainline Protestants. Results will be made available upon request.

7 Similar to what do with each individual model, we also include the interaction of evangelical Protestant and religiosity in the estimation of the mega data. We do not find this interaction is significant.

Similar to what do with each individual model, we also include the interaction of evangelical Protestant and religiosity in the estimation of the mega data. We do not find this interaction is significant.

8 We created an interaction term of Evangelical Protestant and religious service attendance. We did not find any significant effects of this interaction term on either perceptions of global warming existence or perceptions of its causes.

We created an interaction term of Evangelical Protestant and religious service attendance. We did not find any significant effects of this interaction term on either perceptions of global warming existence or perceptions of its causes.

Additional information

Funding

Wanyun Shao was supported by the National Academies of Sciences, Engineering, and Medicine Gulf Research Program (grant 2000008396).