The Triangle Test

The triangle test is perhaps one of the best known and most used difference tests in the beverage industry. The triangle test, sometimes also called a triangular test, is used to determine if a sample is statistically different from a standard. This can be immensely helpful when determining if a process or ingredient change has had an impact on the final flavor of a product or if a new batch of product has changed in some way from previous batches.

In a triangle test assessors are presented with three samples (see diagram below). Two of the samples are filled with the baseline product while the third is filled with the product being tested. Then testers are asked to identify which of the samples is different. If enough assessors chose the sample being tested it is considered significantly different. This can help a brewer to understand if a new grain or different fermentation temperature has resulted in a change to the final flavor of a sake.

Diagram 1:

When preparing a triangle test it is important that all assessors be familiar with the process. In order to be statistically relevant, it is suggested that there be at least 18 assessors. This can be somewhat difficult for smaller brewers who do not have enough staff to fill the role of assessors. If this is the case the test can be repeated with the same assessors to increase its statistical power; however, it should not be repeated more than once, and the second testing should be completed at a separate event. Otherwise, its statistical power may be degraded.

When presenting samples to assessors the test sample should be placed in a random position for each assessor to lessen bias. If possible, it is also recommended to serve the same amount of each sample in each glass and to use glassware that does not reveal the color or turbidity of the samples being tested. This will help to remove other forms of bias that may creep in.

After the assessors have recorded their choices, it is up to the brewer to interpret the results. First the brewer looks at the number of respondents and then determines the number that guessed correctly which was the odd sample. Then by using the table below the brewer can determine if the tested sample is statistically different or not. If the number of successful respondents is higher than the number listed in the table, it can be determined that the sample was either 95% or 99% statistically different. If however the number of successful answers is below the number in the table the sample can be said to be statistically the same as the baseline.

Although the triangle test can be a powerful test in the brewhouse it is important that brewers understand its limitations. First off, triangle tests are not a valid way of measuring preference. Preference testing, sometimes also known as hedonic testing, requires a different methodology and should be conducted separately from difference testing. Second, triangle tests only tell brewers that there is a difference between a sample and the chosen baseline. It does not tell you what that difference is. Finally, it must be noted that the numbers given above are only valid when testing for differences between products. If you would like to test products to prove that they are similar than you will need considerably more participants. 

Hopefully this brief introduction into triangle testing has inspired you to incorporate more sensory methodologies into your brewhouse. If you are interested in more about the triangle test and sensory science in general check out some of the links below.

References:

https://www.sensorysociety.org/

https://cpe.ucdavis.edu/certificate-program/applied-sensory-and-consumer-science-certificate-program