Premium
Evaluation of putative hazard tests under background risk heterogeneity
Author(s) -
Liu Yuan,
Lawson Andrew B.,
Ma Bo
Publication year - 2009
Publication title -
environmetrics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.68
H-Index - 58
eISSN - 1099-095X
pISSN - 1180-4009
DOI - 10.1002/env.926
Subject(s) - statistics , monte carlo method , cluster analysis , uncorrelated , statistical power , econometrics , event (particle physics) , hazard , statistical hypothesis testing , intensity (physics) , computer science , mathematics , physics , chemistry , organic chemistry , quantum mechanics
In this paper, an evaluation of some tests for adverse health effects around fixed locations (putative hazard sites) is made. We address the situation where a heterogeneous background is included both in terms of correlated and uncorrelated heterogeneity (UH). In addition, we examine a set of tests for case event data (residential case addresses) including both distance and directional effects. The tests include distance and directional score tests (DIR), integrated intensity tests, the Besag & Newell (BES) test, and Kulldorff's focus clustering scan test. Monte Carlo power of the tests is evaluated and power curves are finally presented. Interesting results from this work include the lack of power found when correlated heterogeneity (CH) is present. This is not found with UH. The BES and Scan tests have lower power than the integrated intensity tests and score tests in general. The integrated intensity test demonstrates its omnibus ability to detect either distance or directional effects. Copyright © 2008 John Wiley & Sons, Ltd.