16 
BSH Report No. 56 
Method of data evaluation 
The key issues to be addressed within this miniRR were to find conclusive indi 
cations if two paraffin-wax samples are statistically similar or not. Therefore, 
two peak-based features, “relative standard deviation” (RSD) and the “critical 
difference” (CD) are used for each analytical parameter (= peak) of a sample 
and are used for comparison. Thus, the number of RSD- or CD-features 
depends on the number of peaks on which a comparison Is based on. 
These features have the advantage of being well known by people conducting 
analysis according to CEN2012; moreover, CEN2012 provides statistically 
derived thresholds of 5% and 14% for RSD and CD, respectively. If these val 
ues are exceeded, dissimilarity of the samples could be concluded. 
To generate the data regarding the two desired features, peak heights of the 
target compounds in the chromatograms of the samples are extracted and 
maximum-normalised, allowing for a concentration-independent handling. 
The second step is to calculate the two features of RSD and CD for every Indi 
vidual peak comparison. Each feature with the established threshold can be 
applied directly and might respond positively or negatively, meaning It is below 
(matching) or above (non-matching) this threshold. 
In a third step, the results of every feature-test can be further summarised by 
building another ratio, the so called “matching-ratio”, showing the number of 
positive responses to the total number of features tested. A sample’s set (all 
peak comparisons) of matching-ratios results in the conclusion matrix. Thus, a 
matching-ratio can be based on RSD- or CD-features. The following assump 
tion was made for the final comparison of samples in this study: A match Is 
concluded when at least half of the values in step 2 exceed their thresholds of 
5% and 14%, respectively, resulting In a matching-ratio of >50% (step 3). 
Ratios below that 50%-limlt were believed to be dissimilar, as the majority of 
RSD- or CD-features considered are not fulfilling their criterion In the first place. 
Those are coloured red in the conclusion matrix (Step 3).