This research project will develop techniques for using national survey data to estimate dynamic measures of public opinion across a variety of types of subnational units such as states, congressional districts, and state legislative districts. These techniques will allow researchers to generate accurate estimates of public opinion over time by fine-grained demographic-geographic-temporal subgroups. National surveys are designed to give good estimates of national public opinion at a particular point in time. They do not, however, necessarily give good estimates of opinion for subnational units. They also do not allow for understanding time trends of within these units. Recent advances have been made on estimating subnational opinion, but this work has yet to meet the special challenges of measuring opinion over time. This research will improve statistical tools, assess the accuracy of these tools, and create a set of guidelines for proper implementation. The researchers will employ these new tools to make substantive contributions to social science research. The newly developed techniques will be made available to a broad range of researchers and poll analysts through the creation and distribution of statistical software packages.
This project will develop a dynamic multilevel regression and poststratification (MRP) technique that will allow researchers to generate time-varying estimates of public opinion. MRP is a statistical technique that uses individual survey responses from national opinion polls coupled advances in Bayesian statistics and multilevel modeling to generate opinion estimates by demographic-geographic subgroups or "types." Dynamic MRP improves on standard MRP by taking advantage of additional information over time. Using this approach, the project will undertake a new investigation of the relationship between public preferences and policymaking at the state level. Specifically, the project will focus on two important issues: the death penalty and same-sex marriage. By tracking and tracing changes in opinion and policy over long periods of time, the investigators will gain causal leverage on the effects of public opinion that is lacking from existing research efforts. The project is supported by the Methodology, Measurement, and Statistics Program and a consortium of federal statistical agencies as part of a joint activity to support research on survey and statistical methodology.
