*************************************************************************
The Social Statistics Section of the Royal Statistical Society is holding an evening meeting on January 19th, from 5pm - 7:00pm at the RSS, 12 Erroll Street, London EC1 (Barbican and Moorgate are the nearest underground stations).
Tea from 4:30pm.
Speaker: David Steel (University of Wollongong)
Discussant: Phil Brown (University of Kent at Canterbury)
UNRAVELLING ECOLOGICAL ANALYSIS
Abstract: Ecological analysis involves the analysis of aggregate data for groups of individuals with the aim of making inferences about relationships at individual level. Often the results of such an analysis give badly biased estimates. While the problem has been recognised for over 60 years, there has been little effort put into solving it by statisticians. Most of the developments have been done by researchers in fields such as political science, geography and epidemiology who are faced with problems of having to use aggregate data to make individual level inferences. This talk will discuss the various sources of aggregation bias and review methods proposed for removing it. The key role of variation of the individual level relationship between groups and how this can be explained by covariates will be considered. The statistical problems that need to be solved to enable ecological analysis to be a useful statistical approach will be identified.
ALL WELCOME
----------------------------------------------------------------------------------
Celia Macintyre
Honorary Secretary, RSS Social Statistics Section
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|