STATISTICS IN MEDICINE Statist. Med. 2000; 19:2521–2538 Two new templates for epidemiology applications: linked micromap plots and conditioned choropleth maps Daniel B. Carr 1;*;† , John F. Wallin 2;‡ and D. Andrew Carr 3;§ 1 Center for Computational Statistics; MS-4A7; George Mason University; Fairfax VA; 22030; U.S.A. 2 Institute for Computational Science and Informatics; George Mason University; Fairfax VA; 22030; U.S.A. 3 Bureau of Labor Statistics SUMMARY This paper describes two interactive templates for representing spatially indexed estimates. Both templates use a matrix layout of small panels. The rst template, called linked micromap plots, can represent multivariate estimates associated with each spatially indexed study unit. The second template, called conditioned choropleth maps, shows the connection between a dependent variable, as represented in a classed choropleth map, and two explanatory variables. The paper describes the cognitive considerations that motivate the layouts and representation details. The discussion also addresses topics of data quality and access, hypothesis generation, and interactive features such as pan and zoom and dynamic conditioning via sliders. The examples show epidemiological (mortality rates) and environmental (toxic concentrations) applications. Copyright ? 2000 John Wiley & Sons, Ltd. 1. INTRODUCTION The paper describes two dierent templates for representing statistical estimates and their spatial indices. The rst template, called linked micromap (LM) plots, provides a method for representing many types of spatially indexed statistical summaries. The LM template serves either as a format for static summary plots or as a format for interactive plots that provide progressive disclosure of information by mousing on items to obtain more detail (drill down). The second template, conditioned choropleth (CC) maps, focuses on spatial displays that involve one dependent variable and two explanatory variables. CC maps promote interactive hypothesis generation. Environmental applications motivated the development of LM plots while epidemiological applications motivated ∗ Correspondence to: Daniel B. Carr, Center for Computational Statistics, MS-4A7, George Mason University, Fairfax VA, 22030, U.S.A. † E-mail: dcarr@galaxy.gmu.edu ‡ Currently on leave at Los Alamos National Laboratory, U.S.A. § Currently a professional dancer at the State Ballet of Missouri, U.S.A. Contract=grant sponsor: U.S. EPA; contract=grant number: CR8280820-01-0 Copyright ? 2000 John Wiley & Sons, Ltd.