Mildenberger Parses Partisan Climate Beliefs

By Ashley Serpa – Climate change is perhaps the greatest crisis of our time, yet passing meaningful policy to mitigate it proves exceedingly difficult. On November 29, 2017, in a lecture entitled “The Spatial Distribution of Public and Partisan Climate Beliefs”, UC Santa Barbara political scientist Matto Mildenberger tried to make sense of the political barriers to environmental policy action.

Working at a “granular spatial scale,” Mildenberger parses partisan political opinions on climate change and environmental policy at the local level. Making sense of public opinion data poses problems for researchers who want their work to effect change in policy. This is in part because national-level data on its own does not illuminate the nuances in opinion at the local level.

Researchers use several methods to understand public opinion. Surveys that collect data at the national or state level are often viewed as fact, but this assumes that the data can be generalized through time or space. The “pooling and disaggregation” method requires a large amount of data, using many surveys over time that are then collapsed together. This, however, does not account for the fact that individuals from a smaller state might not be good general indicators of public opinion. For example, research conducted by a Stanford scholar using this method found that West Virginia was a hotbed for pro-carbon regulation, which is on its face not very convincing.

Pursuing a clearer picture

Political scientists have begun using the “multilevel regression with post-stratification” (MRP) approach to uncover more detailed information on public opinions on climate change and renewable energy policies. MRP, like pooling and disaggregation, uses national-level surveys, but it goes a step further by taking into account “individual-level demographics as well as area-level and geography-level covariance.” MRP uses partial pooling in order to fill in data where it does not exist—it does this by looking at similar populations in similar geographies where data does exist. This method allows researchers to “weight the model-projected belief for every subtype by the number of people in that subtype” in a given congressional district, county or other geography, which provides a clearer picture of local-level opinions that national surveys obscure.

“Very few political decision-makers are wildly concerned with what the national average belief in a given place or a given time is,” Mildenberger explained. “They’re much more focused on understanding local perceptions.” In other words, they care more about what their constituents think. This makes information on the opinions of people in smaller geographies, such as the congressional level, particularly significant.

Potential allies, silenced

Using an MRP approach, Mildenberger found some surprising results with regard to partisan opinion on climate change. Taken as a whole, the Republican Party represents a powerful voice against policy to mitigate climate change. However, the fraction of Republicans in many states who believe climate change is happening represents a plurality, and 20-30% in several states believe it is human-caused. In Florida, for example, a sizable number of Republicans support a renewable energy policy, yet this is not reflected in the policies of their government representatives—perhaps because the data available generalizes at the national or state level rather than uncovering local opinions.

In other words, there are Republicans who could be potential allies in the fight against climate change, but whose voices have been largely silenced by data that only accounts for their party’s national averages. In order for them to be heard by their government representatives, the methods with which researchers analyze public opinion on climate change must bring local variations to light.

Learn more about Matto Mildenberger.