Ecological Inference in the Social Sciences精品.pdfVIP

NIH Public Access Author Manuscript Stat Methodol. Author manuscript; available in PMC 2011 May 1. N Published in final edited form as: I H Stat Methodol. 2010 May 1; 7(3): 307–322. doi:10.1016/j.stamet.2009.09.003. - P A A u Ecological Inference in the Social Sciences t h o r 1 2 M Adam Glynn and Jon Wakefield a 1 Department of Government, Harvard University, Boston n u s 2 Departments of Statistics and Biostatistics, University of Washington, Seattle c r i p t Abstract Ecological inference is a problem of partial identification, and therefore reliable precise conclusions are rarely possible without the collection of individual level (identifying) data. Without such data, sensitivity analyses provide the only recourse. In this paper we review and critique approaches to ecological inference in the social sciences, and describe in detail hierarchical models, which allow both sensitivity analysis and the incorporation of individual level N data into an ecological analysis. A crucial element of a sensitivity analysis in such models is prior I H specification, and we detail how this may be carried out. Furthermore, we demonstrate how the - P inclusion of a small amount of individual level data can dramatically improve the properties of A such estimates. A u t h o r 1 Introduction M a In this paper the problems of making individual-level inference from ecological data is n u considered. In