|Title||Een mathematisch-statistische studie over de nauwkeurigheid van het organoleptisch onderzoek van boter|
|Source||University. Promotor(en): M.J. van Uven; B. van der Burg. - Zutphen : Thieme - 122|
|Publication type||Dissertation, internally prepared|
|Keyword(s)||boter - dierlijke producten - geur en smaak - aroma - smaak - statistiek - waarschijnlijkheidsanalyse - wiskunde - cum laude - butter - animal products - flavour - taste - statistics - probability analysis - mathematics|
|Abstract||The quality of butter, produced by the dairies in the Netherlands, is regularly tested by experienced testers after storage for 2 weeks usually at 14_C. Odour, taste and texture are tested. Usually a score is given, running from 1 (bad) to 5 (excellent). An analysis was made of the accuracy of the evaluations.
Differences in and between testers in score, for tests on many or one sample were assessed by the standard deviation of the score of one, or the mean score of 2 or more judges (subjective accuracy). Deviations were normally distributed. Accuracy was expressed as the correlation coefficient (r) between the scores by one or more judges and the 'real' score (the mean score given by many judges): the objective accuracy.
An extensive mathematical treatment was given to show how r depends on the number of judges. Starting from the fact that for 2 judges r = 0.80, the value for 1 judge was found to be r = 0.686 and for 3 judges 0.853. At least 3 judges were necessary. Methods were given of expressing the capability of 1 judge in terms of r. The subjective and objective accuracy of judges can differ much.