Downscaling Climate Survey Data — from Large to Local

Public perceptions about climate change, like climate change impacts and adaptation needs, tend to vary from place to place. The environmental, socioeconomic, health and policy importance of local impacts motivates ongoing efforts by physical scientists to downscale projections from global climate models so they better resolve local conditions. Localized information about public perceptions could be similarly important for adaptation planning, but local-scale social data (beyond isolated case studies) often do not exist. Howe et al. (2015; hereafter H15) recently took steps toward addressing this need with downscaling methods (multilevel regression and poststratification, or MRP) applied to national surveys in order to characterize climate-change perceptions in US states, and also in 435 congressional districts and 3,143 counties. Hamilton et al. (in press) tested the validity of their results at the smallest level through comparison with independent surveys conducted in 30 mostly non-metropolitan US counties. They found the H15 estimates unbiased and moderately correlated with independent measures, encouraging further research but with a caveat that MRP downscaling underestimates the true variance. Here we conduct a broader investigation using the H15 data together with 30,000 interviews from independent surveys to study the behavior of MRP in characterizing public views of anthropogenic climate change at small spatial scales. Although MRP exhibits fair criterion validity, an alternative and simpler regression-based method performs better in tests with two datasets.

References

Howe, P.D., M. Mildenberger, J.R. Marlon and A. Leiserowitz. 2015. “Geographic variation in opinions on climate change at state and local scales in the USA.” Nature Climate Change doi:10.1038/nclimate2583

Hamilton, L.C., J. Hartter and T.G. Safford. In press. “Validity of county-level estimates of climate-change beliefs.” Nature Climate Change