A visual examination of Selikoff’s 20-year rule using correspondence analysis and the Cressie-Read family of divergence statistics
Publication Name
Proceedings of the International Congress on Modelling and Simulation, MODSIM
Abstract
Irving Selikoff’s impact on our understanding of the side effects of exposure to asbestos fibres is now very well understood. In his seminal 1981 paper, Selikoff established a “20-year rule” where he concluded that it took about 20 years for workers to exhibit abnormal chest x-rays that signal the presence of asbestosis in at least one of their lungs; asbestosis is the scaring and inflammation of the lung tissue caused by the strong, but flexible, asbestos fibres. This paper presents an analysis of Selikoff’s asbestos data using a new method of correspondence analysis for two-way contingency tables that uses the Cressie-Read family of divergence statistics as the general measure of association. While such a family incorporates a wide range of different chi-squared statistics, we shall focus our analysis on Pearson’s statistic and the Freeman-Tukey statistic. We also show how one may identify whether each of the categories of the contingency table is a statistically significant contributor to the association structure between the variables. We show that, for Selikoff’s data, all years of occupational exposure to asbestos, and all grades of asbestosis that he considered, are statistically significant contributors to the association.
Open Access Status
This publication is not available as open access
First Page
38
Last Page
44