Improving odour detection by assessors

In a test of theoretical predictions made by MacRae and Falahee (1993), trained assessors evaluated water samples which sometimes contained 10−5 mg/l of trichloroanisole, using three different procedures. Compared with the blue-book method of single-sample assessment, providing a reference sample identified to the assessor as being odour-free increased the proportion of responses reporting an odour. This decreased the number of odorous samples missed. Although this outcome may be valuable in itself, discrimination between odorous and odour-free samples was not really enhanced since ‘false alarms’ increased correspondingly. Incorporating additional, different odours into the sequence of samples and giving knowledge of results when these were judged, dramatically improved performance — the rates both of missing trichloroanisole and of mistakenly attributing odour to odour-free samples were halved.