- Objective: To determine if a difference exists between two samples.
- Procedure: Initially, assessors are presented with two samples, ‘A’ and ‘not A’, and asked to familiarise themselves with their characteristics. The samples must be labelled appropriately - Experimental design: Usually, 10–50 assessors are trained to identify the ‘A’ and ‘not A’ samples. During the test, assessors receive either: • one sample (either ‘A’ or ‘not A’); • two samples (‘A’ and ‘not A’); • several samples (up to 20 samples, equal numbers of ‘A’ and ‘not A’) Same–different test - Objective: To determine if a difference exists between two samples. - Procedure: Assessors are presented with a pair of samples and asked to determine if the samples are the ‘same’ or ‘different’; they may also be asked to describe any differences. Samples are labelled with three-digit codes (blind coded) and should be assessed in the order provided (left to right). Appropriate palate cleansers should be used after each sample. - Experimental design: There are four possible sample presentations. They are AA BB AB BA Assessors either receive one, two or all four pairs. If the samples are complex, or need comparing in half-head or half-face trials, each assessor will receive only one pair. In this instance, all four possible presentations are used an equal number of times. Alternatively, each assessor may receive one ‘same’ pair and one ‘different’ pair, or all four possible pairs. The same– different test is often used with 30–50 assessors, although this number may be increased to as many as 200, particularly when only one pair is given. Duo-trio test (phép thử 2-3) - Objective: To determine if a difference exists between two samples. - Procedure: Assessors are presented with three samples, two blind coded and one labelled as a ‘reference’. They are asked to assess the reference sample, followed by the two coded samples (in the order provided) and determine which is the most similar (or different) to the reference. Appropriate palate cleansers should be used after each sample. The duo-trio test is particularly useful for samples that are not homogeneous, as the question asked is, which sample is the ‘most similar’ (rather than ‘identical’) or ‘most different’ to the reference