Choice test triangle tests

Triangle Three coded samples (two of which are the same and one is different) are distributed to the assessors in six possible sequences of two products A and B (AAB, BBA, BAB, ABA, BAA, ABB) that should be counterbalanced across all panelists. The assessors must indicate which the odd sample is. A forced choice is required also if the selection is based on a guess. These combinations of samples are distributed at random in groups of six among the assessors. In order to balance the results (each sequence uses an equal number of times), the number of assessors taking part in the test should be preferably in multiples of six. 

Three-alternative forced choice (3-APC) The test allows to determine if two products differ in the specified dimension and which sample is higher in perceived intensity of the specified attribute. Three coded samples (two of which are the same and one is different) are distributed between the panelists in randomized or balanced order in three possible combinations (AAB, ABA, BAA, or BBA, BAB, ABB). As for the triangle the null hypothesis is P = 1/3. 

Difference testing can be useful tools for assessing panelist s skills. Methods such as the triangle and duo-trio tests are what are communally known as forced choice discrimination tests because panelists are presented with multiple samples and must decide which examples are the same and which are different. These tests can also be used in assessing the impact of changes in supply chain and production techniques. Succinct descriptions of these and other discrimination tests can be found in Bamforth (2014) and Simpson (2006). More in-depth discussions of these techniques are covered by Kilcast (2010), Meilgaard et al. (2007) and Stone et al. (2012). 

---