AOAC Methods in Codex STAN 234 (Preliminary Methods Review)

J. ASSOC. OFF. ANAt.. CHEM. (VO!.. 65, NO. 4, 1982)

983

CAPAR ET AL:

..., M 0 o- N ..., 0 ON ...,

um and lead at the levels under study. Labora– tories 5 and 15 reported reagent blank lead levels that were approximately 4 times as high as those specified in the instructions. Therefore, we ex– cluded their lead results from our statistical evaluation. The Dixon outlier test (2) was applied to the remaining data at the 95% confidence level. It was also reapplied to data purged of outliers found in the first application, The resultant outliers are indicated in Tables I and 2, but they were not included in the statistical evaluation of the collaborative data. There were 6 outliers among the cadmium results and 4 among the lead results. From the results of a one-way analysis of variance, the within- and between-laboratory variabilities (2) were estimated. The statistical results are given in Table 1 for cadmium and in Table 2 for lead. The quantity So is the within– laboratory standard deviation and CVo is the corresponding coefficient of variation. So is a measure of the variability of readings taken by the same analyst in the same laboratory under constant conditions. So is also referred to as the repeatability standard deviation, i.e., a measure of how well a given analyst can match his own results. The quantity SL is the pure, among– laboratory standard deviation, i.e., a measure of the variability of the laboratory averages. CVL is the corresponding coefficient of variation. The quantity Sx is the composite among-labora– tory standard deviation and CV"' is the corre– sponding coefficient of variation. S,., is a mea– sure of the variability of the readings from dif– ferent laboratories. S,., is also referred to as the reproducibility standard deviation, i.e., a mea– sure of how well an analyst in one laboratory can match the results of another analyst in a different laboratory. This variability (Sx) is considered to arise from 2 sources: the within-laboratory variability (So) and the pure, among-laboratory variability (SL). The reference values established by the sponsoring laboratories, as discussed previously. were used to evaluate the accuracy of the col– laborative results. The accuracy was calculated by dividing each individual result by the corre– sponding reference value. The average accuracy values in percentages and the reference values are given for each commodity level in Tables 1 and 2 for cadmium and lead, respectively. The accuracy ranges from 85 to 114% for cadmium and from 78 to 109% for lead. The effect of commodity type on the accuracy was investi– gated by performing an analysis of variance for

00

N 0 0 \0 ... ,._ 0 do

00

,.._ ~ 00

00

8 o ..... -

00

,., 0 d-

N - ON "'

N N 0 0~

-

a 011'> ...

0:,

(JI 0 ci~

og -

..... 8 d~

8 d– N

eview