AOAC ERP MICRO AUGUST 2018

OMAMAN-44 A: Collaborative Study Manuscript Expert Review Panel Use Only August 2018

456

Collaborative Study Statistical Analysis

457

458

Each collaborating laboratory recorded the CFU/g results for the reference methods and the

candidate method on the electronic spreadsheet provided. The data sheets were submitted to the study 459

director at the end of the study for analysis. A logarithmic transformation [CFU/g +0.1f, where f is the 460

reported CFU/g corresponding to the smallest reportable result]. A Youden plot was prepared to identify 461

discrepancies between test replicates. Outliers were identified using the Cochran and Grubbs’ tests. The 462

differences of means, including 95% upper and lower confidence limits, were determined for each 463

contamination level (1, 14). If the difference of means between the two methods was < 0.5 Log 10 , it was 464

considered that no statistical difference existed between the two methods (2, 15). The repeatability (s r ) 465

466

) of the methods were also determined (4, 17).

and reproducibility (s R

467

468

Collaborative Study Powdered Infant Formula with Probiotics Results

469

470

Each collaborating laboratory recorded the CFU/g results for the reference methods and the

candidate method on the electronic spreadsheet provided. The data sheets were submitted to the study 471

director at the end of the study for analysis. The candidate method results at 24 and at 48 h along with 472

the reference method results reported by each laboratory were converted to logarithmic values for 473

statistical analysis and were plotted using a Youden’s plot. The Log 10 ERP Use Only individual laboratory results are presented in Appendix Tables 1-2. Figures 1-2 present the Youden plots of each matrix. The transformed 475 data were analyzed for outliers by the Cochran and Grubb’s tests. No outliers were identified. The 476 difference of means (including 95% confidence intervals), repeatability (s r ) and reproducibility (s R ) were 477 474

determined for each contamination level. The results of the interlaboratory data analyses are presented 478

in Table 9. In addition to the test portions, each participant that performed testing and submitted results 479

20