Pesticide residues in food-based proficiency test materials, spiking values versus consensus assigned values.

We examine the differences among the three estimates of the true value of the measurand derived from routine proficiency testing of laboratories analyzing foodstuffs for pesticide residues. The three values are (i) the spike level (Sp), (ii) the mean result found by the laboratory conducting the test for sufficient homogeneity (Ho), and (iii) the consensus of the participants' results used as the assigned value (AV) in converting results into z scores. Data amounting to 205 examples were collected from successive rounds of three series of proficiency tests from the Food Analysis Performance Assessment Scheme (FAPAS): namely, series 05 (fats, oils, and animal products), series 09 (cereals and their products), and series 19 (fruits, vegetables, and their products). Irrespective of the class of test material, we found that the means of AV and Ho were almost identical, while the value of Sp was systematically higher than AV by a factor of 1.22. The dispersion of the individual values of both ratios, Ho/AV and Sp/AV, was examined by analysis of variance. A small part of the variance was attributed to the series, but a greater part, about 40%, was attributed to individual rounds within series. We discovered no connection between the ratios and the chemistry of the analyte.