Triangle Testing

Mount Vernon Brew Club

Friday, March 1, 2024

Triangle testing is a simple tool that food scientists use. It's useful for detecting whether is a difference between two similar samples and what those differences are, if they exist.

When triangle testing, you should have some questions that you'd like to answer before doing the test. Let's say you have two beers in front of you, one of which is brewed with your built water (pure water that you added salts and minerals to for taste) and one of which is brewed using reverse osmosis water with no added minerals. Some questions you might have are whether the tap water minerals can be tasted, whether the minerals contribute to a chalky texture, and whether using a built water with minerals is better than using pure water. Assuming all else (malts, hops, fermentation, etc.) is equivalent, triangle testing will help you identify differences in your samples. You are essentially double-blind testing on your own or with the help of your brew club.

Here's how to triangle test.

Split the two samples that you will test into four identical glasses or containers, two of each sample. It's important that you can't identify the difference in the containers by feel or texture of the container itself. Mark the containers with what they contain — you can use a sticky note or a marker on the bottom of the container to identify them after you remove your blindfold.
Put a blindfold on, then scramble the four containers until you have no idea which one is which. Pick out one of the four containers at random and set it aside so you only have three containers in front of you. These are the three samples you will try; there will be two of one sample and one of the other, but you won't know exactly which sample you have two of and which you only have one.
Try each of the three remaining samples, and try to identify which of the three is the odd one out. As you pick out the odd one out, identify what is different about it. What components of taste, aroma, and texture can you detect in each of the samples individually? What are the differences between the taste, aroma, and texture that you can identify between the two samples in comparison? Can you even tell a difference?
Take off your blindfold. Were you right in identifying the odd one out? If you were successful, how easy was it to determine? (Some samples are very easy to differentiate, others not so much!) If you were not successful, do adding the appearance and human factors make it easy to differentiate? What conclusions can you gather from your test? Did the test leave you with any more questions?

Trying something using a triangle test is useful for removing the human element and the appearance as factors in the flavors you identify, allowing you to simply focus on taste, aroma, and texture.

Sometimes, it's really tough to tell a difference between two samples, and that's okay. A difficulty differentiating between examples can mean that the difference is really subtle, and it's not necessarily a deficiency in your tongue or olfactory bulb. If you were unable to identify the odd one out, you can conclude that the factors which were changed are not significant to the outcome of the product. However, you also shouldn't rely on one person's test results.

A single triangle test with one participant gives you anecdotal evidence, but it's not backed by statistics; it's the literal example of "sample size of one." Instead, you should consider the sample sizes needed for a one-sided binomial proportion test (the results of your triangle test can be tested using a binomial proportion test). If you have any statisticians or accountants in your midst, use their knowledge to your advantage!

After triangle testing, you may end up with more questions than you started off with, and that's okay! It means that you've figured something out and want to learn more. You shouldn't be afraid to triangle test between multiple pairs of samples.

Citation
Mount Vernon Brew Club. (2024). Triangle Testing. Mount Vernon Brew Club. https://mountvernonbrewclub.org/site/index.php/ontap/triangle-testing/.

This work is licensed under Creative Commons Attribution 4.0 International.