Odor Thresholds by Forced-Choice Dynamic Triangle Olfactometry: Reproducibility and Methods of Calculation

Andrew Dravnieks,Walther Schmidtsdorff,Morten Meilgaard
DOI: https://doi.org/10.1080/00022470.1986.10466128
1986-08-01
Journal of the Air Pollution Control Association
Abstract:The reproducibility of odor thresholds determined by forced-choice dynamic triangle olfactometry was studied employing panels of 23- 45 subjects who tested 12 odorants, with four replications for one (1-butanol). Depending on the selection of panelists, their number, and between-session differences in olfactometer variables, the standard deviation of log(threshold) values ranged from 0.12 (same session, different panels of nine) to 0.37 (different sessions, different panels of nine). Among the demographic factors, only age was found significant; those below age of 30 were 1.6-3 times more sensitive than the older group. No systematic sensitivity difference was found between females and males, and nonsmokers and smokers. Suspicion has been raised in the literature that the method used here, ASTM E679, yields biased results. It is shown for panels of 6-45 members that such biases are inconsequential if the range of concentration presented to the panelists is large enough, and the group threshold is well within that range. The Hall-Ellis method of calculation may have a slight advantage for panels of less than eight subjects, but such small panels are not advisable because of their high variability. The method based on the detectability index exhibited more scatter and would need an impracticaliy large number of measurements.
What problem does this paper attempt to address?