Determination of the Impact of Panel Size and Presentation Mode on the Variability of Odour Detection Threshold Concentration

Determination of odour detection threshold is performed by olfactometric analysis, which is guided by the EN 13725 international standards. This standard allows the use of the following odorants presentation modes during olfactometric analysis: yes/no mode and forced choice mode with two or more ports. Furthermore, the EN 13725 standards limits the minimum panel size to not less than four although a larger panel size is recommended to improve repeatability and accuracy of results.In the course of this study, three known concentrations of three odorant gas were analyzed by dynamic olfactometry: n-butanol, hydrogen sulfide and dimethylamine. The dynamic olfactometry analyses were performed with a panel of 8 assessors. All panel members were qualified and trained according to the EN 13725 standards.The first objective of this research aims to quantify the variation of the average detection threshold when the panel size varies. In the course of this project, the panel size varied from 4 to 8 assessors. The results were analyzed by sub-grouping the 8 assessors in 70 sub-groups of 4 assessors, 56 sub-groups of 5 assessors, 28 sub-groups of 6 assessors and 8 sub-groups of 7 assessors and by comparing the ratios of the highest and lowest detection thresholds of each panel size.Furthermore, three different presentation modes were used during the course of the present research project. The second objective was to determine if a variation in the average detection threshold exists when the presentation mode varies for a same odorant gas analyzed. The presentation modes used are: yes/no, 2 forced choice and 3 forced choice modes.The results show no significant difference in the detection threshold of a panel of 8 assessors when different presentation modes are used during olfactometric analysis. However, the results seems to show that the 3 forced choice presentation mode may increase the detection threshold concentration of the panel by up to 21% when compared to the yes/no method. Furthermore, this study shows that a significant difference exists in the ratios of highest and lowest average detection thresholds concentration of sub-groups among all panel sizes (from 4 to 7 assessors). These ratios vary from 4.13 (4 assessors) to 1.44 (7 assessors). These results show that the choice of the panel size may have a significant impact on the quality of the results.