24 different whisky bottlings were tasted for the study. Half the volume of each whisky was additionally chill filtered, resulting in 48 different samples. The overall volume of the whiskies tasted was 50.4 litres (72 bottles * 0.7l/bottle). The overall amount of sample bottles was 48*28=1,344, but only 1,331 evaluation forms were returned.
Each sample set contained 12 samples with 6 different whiskies, each whisky in a non chill filtered version and in a chill filtered version. In order to make a direct A/B comparison possible, each participant received two sets containing six samples. The only difference between the sets was the chill filtration.
The samples were labelled according to a fixed pattern - but randomly. Attention was paid that the numbers of non chill filtered and chill filtered samples did not follow each other and also that the bigger or smaller number did not indicate a certain filtration state. Single-digit sample numbers were avoided in order to rule out mental singularities. In order to make possible secret parallel tastings of two or more test persons more difficult, 50% of each sample set were labelled with a different set of numbers. This resulted in 96 sample numbers ranging from 11 to 106.
The participants could choose between four different sample sets. One contained only unpeated samples, one only peated samples. The other two sets contained mixed samples.