Readme file

SERIES C  
Applied Statistics

Correlating point-referenced radon and areal uranium data arising from a common spatial process, by B. J. Smith and M. K. Cowles, Appl. Statist., 56 (2007), 313–326

DATASET DESCRIPTION

The data can be found as R (www.r-project.org) objects stored in the ASCII text file 'Data.R'. Description of the data objects are given below.

List: site
Description:
List of the unique geographic sites at which home radon measurements were taken. In order to comply with the confidentiality agreement between Iowa Study investigators and study participants, a small amount of noise has been added to the longitude/latitude coordinates for the radon measurement sites in this public-use dataset. The exact coordinates were used in the analysis presented in our paper.
Objects:
id - unique identifier for the geographic sites
blockid - a numerical index representing the county block in which the
geographic site is located
lat - latitude coordinate of the geographic site
lon - longitude coordinate of the geographic site

List: home
Description:
Radon measurements and variables for home sites.
Objects:
id - unique identifier for the homes
y - natural-log transformed radon measurements
x - matrix of three indicator variables denoting whether a measurement
was taken in the basement, first floor, or higher floors, respectively.
siteid - a numerical index representing the geographic site at which the home is located.

List: block
Description:
Information available at the county-level.
Objects:
id - unique identifier for the county blocks
area - square-mile area for the county block

List: nure
Description:
NURE county radon estimates and variables.
Objects:
y - natural-log transformed county average uranium
x - a matrix with a single column of ones for the intercept of the NURE model
blockid - a numerical index representing the geographic block corresponding to the county average measurement.

List: grid
Description:
Grid of geographic sites for the numerical integration used to calculate
spatial correlation between county mean uranium measurements.
Objects:
id - unique identifier for the geographic sites
blockid - a numerical index representing the county block in which the
geographic site is located
lat - latitude coordinate of the geographic site
lon - longitude coordinate of the geographic site

PROGRAM DESCRIPTION

The model is implemented as an R program and has been successfully tested in version 2.4.1 of the R software. The program files are as follows:

Files
=====

Utils.R - utility functions used in the MCMC implementations

Model_IA.R - MCMC sampler for the radon only model
Model_NURE.R - MCMC sampler for the NURE only model
Model_IANURE.R - MCMC sampler for the combined radon and NURE model

Init1.R, Init2.R, Init3.R - Initial values to run three parallel chains of the MCMC sampler. These files can be used with any of the models.

License
=======

The programs are free software; you can redistribute them and/or modify them under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or any later version.
Usage Notes
===========

Follow the steps below to use the programs. By way of example, we assume that all files are stored in the "C:/Output" directory.

1) The directory in which the program output files are to be saved should be specified in the filename paths that appear at the beginning of the Init files.

2) Start the R statistical software program

3) At the R command line, set variables 'datafile' and 'initfile' equal to the locations of the Data and Init files respectively; e.g.

> datafile <- "C:/Output/Data.R"
> initfile <- "C:/Output/Init1.R"

4) Source the file of utility functions; e.g.

> source("C:/Output/Utils.R")

5) Source the program file for the desired model to run the MCMC sampler; e.g.

> source("C:/Output/Model_IA.R")

NOTE: The R programs are computationally intensive. To reduce run times, we recommend a customized version of R that is linked to an ATLAS (or other suitably) tuned linear algebra library. Information on the use of different linear algebra libraries in R can be found in the document: http://cran.r-project.org/doc/manuals/R-admin.html
.

Brian J. Smith
Department of Biostatistics
University of Iowa
200 Hawkins Drive, C22 GH
Iowa City
IA 52242-1009
USA

E-mail: brian-j-smith@uiowa.edu

 Journals

SERIES A
Statistics in Society

SERIES B
Statistical Methodology

SERIES C
Applied Statistics

SERIES D
The Statistician