Improved Multilevel Regression with Poststratification through Machine Learning (autoMrP)

Broniecki, Philipp; Leemann, Lucas; Wüest, Reto

doi:10.1086/714777

Cited by 11 publications

(13 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Besides using a classical multilevel regression model, we also consider four machine learning classifiers: principle components analysis, lasso regression, gradient tree boosting, and a support vector machine. Recent research has shown that this ensemble method can increase the accuracy of MRP methods by up to 20% [20]. The multi-level modelling and poststratification analysis was executed with the R package autoMrP on a high-performance computer using 128 processor cores.…”

Section: Local Area Estimationmentioning

confidence: 99%

Local support of climate change policies in Germany over time

Levi

Wolf

Sommer

et al. 2023

Environ. Res. Lett.

View full text Add to dashboard Cite

Political conflicts about energy and climate change policies have often local implications, yet little is known about local public opinion towards these policies. Here, we estimate support towards 26 climate change mitigation policies for 402 German counties and for up to four periods of time using multilevel regression and poststratification. We find that local support for climate policies varies by up to 60 percentage points across German counties with large disparities between the rural and urban population. While support for the expansion of wind power plants and solar power plants have converged over the last years, attitudes on the phase out of coal power have polarized across regions. Using a spatial panel analysis, we find that support for the expansion of wind and solar plants correlate with the actual development of solar and wind capacities in these regions. Moreover, the spread of climate policy opinion is strongly determined by spatial diffusion as change in one region positively influences the trajectory of policy opinion among its neighbors.

show abstract

Section: Local Area Estimationmentioning

confidence: 99%

Local support of climate change policies in Germany over time

Levi

Wolf

Sommer

et al. 2023

Environ. Res. Lett.

View full text Add to dashboard Cite

show abstract

“…principle components analysis (PCA), lasso regression, gradient tree boosting, and a support vector machine. Recent research has shown that this ensemble method can increase the accuracy of MrP methods by up to 20% 30 . The multi-level modelling and post-stratification analysis was executed with the R package autoMrP on a cluster using 128 cores.…”

Section: Local Area Estimationmentioning

confidence: 99%

Local support of climate change policies is driven by spatial diffusion

Levi

Wolf²,

Sommer

et al. 2022

Preprint

View full text Add to dashboard Cite

Political conflicts about energy and climate change policies have often local implications, yet little is known about local public opinion towards these policies. Here, we estimate support towards 26 climate change mitigation policies for 402 German counties and for up to four periods of time. We find that local support for climate policies varies by up to 60 percentage points across German counties. While support for the expansion of wind power plants and solar power plants have converged over the last years, attitudes on the phase out of coal power have polarized across regions. Using a spatial panel analysis, we detect that support for the expansion of wind and solar plants correlate with the actual development of solar and wind capacities in these regions. Moreover, the spread of climate policy opinion is strongly determined by spatial diffusion as change in one region positively influences the trajectory of policy opinion among its neighbors.

show abstract

“…This approach allows the estimation of state-level public opinion from national surveys (Gelman and Little 1997;Lax and Phillips 2009;Pacheco 2014). To estimate the MRP models and the uncertainty around the resulting estimates, we use the R-package "autoMrP" (Broniecki, Leemann, and Wüest 2021). 5 Stimson's Dyad Ratios Algorithm can be downloaded here: http://stimson.web.unc.edu/software/.…”

Section: Introductionmentioning

confidence: 99%

“…9 The relationship between our measure and Stimson's mood exceeds comparable correlations between the BRFH measure and Stimson's mood, offering further validation to our approach. 10 To gain a further sense of the new estimates, we return to the analysis of southern states and examine the relationship between annual changes in the new measure and 8 Of course, uncertainty exists around these estimates (As noted above, we use the R-package "autoMrP" [Broniecki, Leemann, and Wüest 2021] to estimate this uncertainty). Our measures are based on the percent offering a liberal response, while Stimson's estimates are based on the percent offering a liberal response out of those who offered a liberal or conservative response (i.e., middle categories are omitted from the denominator).…”

mentioning

confidence: 99%

A Validation and Extension of State-Level Public Policy Mood: 1956–2020

Lagodny

Jones²,

Koch³

et al. 2022

State Politics Policy Q.

View full text Add to dashboard Cite

To fully understand state policy outcomes or elections in the US, we need valid over-time measures of state-level public opinion. We contribute to the research on measuring state public opinion in two ways. First, we respond to Berry, Fording, Hanson, and Crofoot’s (BFHC) critique of Enns and Koch’s measure of state policy mood. We show that when BFHC’s analysis is performed using the same states and examining annual change, it validates the Enns and Koch measure and raises questions about the Berry, Ringquist, Fording, and Hanson measure. Second, we generate a new measure of state policy mood building on Enns and Koch’s approach. The new measure has even better properties than the previous measure and relates to state presidential vote and state policy liberalism in similar ways to Caughey and Warshaw’s measure of state economic liberalism. We conclude with recommendations for using the various direct measures of state public opinion.

show abstract

Improved Multilevel Regression with Poststratification through Machine Learning (autoMrP)

Cited by 11 publications

References 18 publications

Local support of climate change policies in Germany over time

Local support of climate change policies in Germany over time

Local support of climate change policies is driven by spatial diffusion

A Validation and Extension of State-Level Public Policy Mood: 1956–2020

Contact Info

Product

Resources

About