Identification and Empirical Inference

点击翻译按钮获取中文摘要

The investigator plans to expand his recent research on treatment choice with partial knowledge of treatment response, which has developed out of his earlier work on partial identification. Part of the research willaddress general methodological questions and another part will investigate specific substantive problems ofsocial planning. He also will continue his longstanding program of research on partial identification per se.Intellectual Merit: The investigators research on identification is deliberately conservative. The traditionalway to cope with sampling processes that partially identify population parameters has been to combine theavailable data with assumptions strong enough to yield point identification. Such assumptions often are notwell motivated, and empirical researchers often debate their validity. Conservative analysis enablesresearchers to learn from the available data without imposing untenable assumptions. It enables establishmentof a domain of consensus among researchers who may hold disparate beliefs about what assumptions areappropriate. It also makes plain the limitations of the available data.The investigators analysis of treatment choice is similarly conservative. His research shows howsocial planners and other decision makers can cope coherently with difficult problems of choice underambiguity induced by identification problems and the necessity of statistical inference from sample data,without imposing untenable assumptions. This is achieved using well-established principles of statisticaldecision theory, particularly through application of the minimax-regret criterion.Broader Impacts: Many persistent public policy controversies reflect divergent beliefs about the effects ofgovernment policy on society. Such divergent beliefs are often manifest in dueling policy studies that usedifferent analytical approaches or data sources to reach different policy conclusions. Each study may makesense in its own terms, each combining data with conjectures to draw logically valid conclusions. However,there may be no way to determine which study (if either) makes realistic conjectures and which (if either)draws empirically correct conclusions. The investigators conservative approach to empirical inference andtreatment choice can enable the public to better evaluate the credibility of existing policy studies, enhancethe credibility of future policy research, and improve the quality of policymaking