The Dual Importance of Political Identity in Environmental Governance: The Case of Oil and Gas Policy in Colorado

Abstract

Over the past decade, the U.S. has had significant growth in oil and gas production due in large part to hydraulic fracturing (“fracking”); and yet, we know little about public preferences for oil and gas regulations. Political identity is a strong predictor of policy support in other arenas. The purpose of this analysis is to clarify the role of political identity in support for oil and gas policy. We hypothesized that political identity has a direct impact on policy support and an indirect impact via risk/benefit perceptions and trust in the oil and gas industry. Using the KHB mediation technique, we show that about half of the impact of political identity is indirect via these mediating variables. We suggest that political identity is a uniquely important factor in environmental governance because it impacts both policy preferences and other variables that, in turn, also explain policy preferences. Political identity appears to be salient, even for local governance controversies.

Keywords:

• Fracking
• governance
• partisanship
• political identity

Acknowledgments

The authors thank the following undergraduate research assistants: Chloe Thome, Andrew Walz, Daniel Callahan, Ruby Castro, Marie Harding, Nolan Case, Neil Griffith, Danny Valdez, Heather Crosby, Rich Fordham, Lauren Perotti, Jose Gomez, David Strait, JD Haley, Ryan Becker, Jessie Miranda, Lauren Hartsough, Taylor Loberg, Alyssa Jansekok, Alexandra Poynter, Allison Brown and Jazmine Gonzalez.

Notes

1 We considered, but ultimately did not include, a predictor for party affiliation (e.g. Republican or Democrat). This variable was not included in part because of Colorado’s unique political culture, in which large portion of the state identifies as “independent”. The research team decided that an indicator of conservative to liberal affiliation would best capture our respondent’s political identity.

2 There are several variables that might have a theoretical link to policy preferences that we considered in a series of unreported analyses. Some areas of Colorado have a long history of resource extraction, implying that length of residence in the community might relate to policy preferences. This variable correlated extremely weakly with our dependent variable (r = −.034, p = .480) and had null effects when entered in the regression models. Distance to the nearest well in miles had little relationship with our dependent variable (r = 0.080, p = .113) and had no detectable effect when entered our regression models. Similar null relationships existed for trust in regulators and place attachment indicators.

3 Bootstrapping involves resampling with replacement from the data and developing an average standard error from repeated draws.

Additional information

Funding

The authors are highly thankful for the funding was provided by the Rural Sociological Society and the Institute for Learning and Teaching at Colorado State University.