Multivariate Stepwise Regression Analysis of Indoor Radon Data From Ohio, U.S.A.
James A. Harrell and Ashok Kumar
The indoor radon concentration in 125 houses was investigated in six 100 square kilometer areas in Ashtabula, Cuyahoga, Erie-Huron, Franklin, Pike, and Logan counties in the state of Ohio, U.S.A. All six study areas were located within the outcrop of the uraniferous rock formation known as the Ohio “black” Shale. For each house the thickness of the sediment over-burden above the Ohio Shale was determined as were also numerous parameters relating to the design, construction, and maintenance of the house.
Multivariate stepwise regression analyses were performed with indoor radon concentration as the response variable, and the house and geological parameters as the predictor variables. The resulting regression equations are screening models that can be used to predict indoor radon levels. Of the several predictor variables included in the regression analyses, only two consistently made statistically significant contributions to the explanation of the variation in indoor radon. These variables are the penetration factor (a measure of the prevalence of radon entry points in the house substructure; higher values indicate more entry points) and the air exchange factor (a measure of the “tightness” of a house; higher values indicate greater indoor-outdoor air exchange). Increasing indoor radon levels were found to be associated with increasing penetration factor and decreasing air exchange factor.
Radon; stepwise regression; Ohio Shale