Missing Information in Survey Research

点击翻译按钮获取中文摘要

This research develops new methodological and statistical techniques to handle `missing data` for survey and other micro-level data sources. In Political Science most researchers use `listwise deletion` when encountering missing data, but this wastes valuable information and can bias conclusions drawn about the political world. While the statistical literature includes some attractive approaches to the missing data problem, several important theoretical difficulties with these approaches, and the lack of publicly available software, prevent these methods from being used. This project generalizes the different approaches for use with political science data. At the same time this project develops methods that can be applied simultaneously to both aggregate data (e.g., electoral results or census counts) and survey research.