Abstract
The Environmental Lead Proficiency Analytical Testing (ELPAT) Program evaluates over 400 laboratories that perform lead measurements in paints, soils, and dusts. A previous National Institute for Occupational Safety and Health study, based on the ELPAT data over a 3-year period (1992–1995), found no large biases among common hotplate and microwave digestion techniques, but did detect small consistent bias between two common instrumental methods. This study expands on the earlier study by examining the total sample variability and its variation components (interlaboratory and intralaboratory). A correlation model was used to separate the variation components by estimating a variation ratio. The correlation model leads to a more general approach than a sample pairing technique developed by Youden. This study found no significant evidence that the relative contribution of intralaboratory and interlaboratory variability to total variability changes with lead loading levels. There were no significant differences in the relative contribution of variation components among three most commonly used analytical methods (combinations of sample preparation techniques and instrumental methods). The interlaboratory relative standard deviation is about 1.7 times the intralaboratory relative standard deviation. Both variation components are important parts of total variation although the laboratory-to-laboratory (including analyst-to-analyst) difference is greater than the within laboratory (including sample-to-sample) variation.