Discussion onthe Data Drift Correction Method in Homogeneity Assessment of Environmental Reference Materials

Abstract

BACKGROUND Assessment of homogeneity of Environmental Reference Material (ERM) is one of the most important steps for the development of an ERM. The data drift has a serious negative effect on the assessment of homogeneity and uncertainty of ERM. OBJECTIVES To develop the drift correction methods for homogeneity test of ERM. METHODS The drift correction methods were discussed from the aspects of experimental scheme and statistical methods through practical cases. The drift correction methods, such as random analysis, randomized block design, interpolation correction, and trend analysis correction, were introduced in details. RESULTS The t-test method was proposed to judge the significance of data drift. For insignificant drift trend, the random analysis method was useful for the evaluation of uncertainty of homogeneity. If the drift trend was significant, better-corrected results were obtained by using interpolation correction and trend analysis correction methods. For example, after data correction by interpolation correction and trend analysis, the relative standard deviation (RSD) of homogeneity data of Ni in soil decreased from 3.0% to 1.1% and 0.84%, respectively. The absolute value of slope of line decreased from 0.2003mg/kg to 0.02870mg/kg and 4.709×10-5mg/kg. The uncertainties of between-bottle homogeneity of Ni, Cu, Co and Tl decreased by 78% compared with the random analysis method. CONCLUSIONS Data drift correction methods are not effective for long periods of measurement, therefore the experimental scheme of randomized block design method is recommended for homogeneity study.

Keywords

• environmental reference materials
• homogeneity assessment
• data drift correction
• interpolation correction method
• trend analysis correction method
• randomized block design method