Decision following discussions with ISO S01 (17 April 2014) and IDF/ISO AW Berlin (May):
1. Before statistics, remove “0.00” values, single value (
i.e.
absence of results
corresponding to double blind), select the list of fatty acids considered in the evaluation
for each samples (
i.e.
considering dairy vs non dairy)
2. Remove “outliers” value only on the basis of Cochran and/or Grubbs tests
3. Decide eventually to remove “outliers” lab on the basis of Mandel’s H and K (
i.e.
when
all results provided are concerned)
Not done
4. Remove “outliers” value when serious reason has been identified (e.g. chromatographic
problem)
Not done
+++
Record and provide the
appropriate justification for
deleted values
7/2 - MLT Management
(Statistical Evaluation Process)
