When we are measuring flavor and sensory protocols, there’s a few questions we like to ask. Are we trying to find differences between samples? Or are we trying to understand consumers’ preference? Understanding the purpose of the tasting helps to select and apply the appropriate sensory tests.
When we conduct a taste test, there are a few things that we look for:
- Define the purpose of the tasting
- Choose the correct test protocol accordingly
- When comparing new flavors to existing ones, the most useful protocol is Triangle Test
For new products, use Preference and Hedonics protocols
We also have a few different materials we use when we are measuring flavor and running sensory protocols. Surprisingly, many of the things we use are simple items you can find around your house, such as paper towels, labels, crackers, gloves
Samples can NOT be coded with simple one-digit number or letter to avoid subconscious bias (A is better than D). Because of this, a common method is to use random three-digit numbers. For example 142, 852, 296, 370 are numbers that can be used. It’s important to remember NOT to allocate the two identical samples in a triangle test the same three-digit number. Also, it is a standard practice to rotate the order of the samples so each taster has a difference sequence.
When you’re running taste tests, it’s recommended to run three different tests with volunteers:
If you’re running a preference test, the taster is presented with at least two samples. However, the samples do not need to look or taste similar. The taster simply decides which option (sample) he or she likes best. Make sure that you rotate the samples ensuring some testers taste different first samples.
If 100 people participated in a preference test, how many need to choose one product over another for you to feel confident that most people in the public prefer the same product? The answer may be surprising. Scientists are usually satisfied with a conclusion when they are confident that they will get the same result 95 times out of 100.
Another popular test is the triangle test. In this test, the taster is presented with three samples: two are the same and one is different. Often the differences between the two samples are small. The tester is asked to tell which sample is different. As in the preference test, offering different testers different orders of samples to test is important. For the triangle taste test, six different orders of samples are possible: AAB, ABA, BAA, BBA, BAB, and ABB.
In this test, can the tester tell the difference between the products? As in the preference test, the number of correct choices for the results to be statistically significant depends on the sample size. While 95% significance is best, scientists sometimes report results with 80% significance. Since getting correct responses with the triangle taste test is difficult (depending on how similar the samples taste), 80% significance may be the best outcome.
These tests try to answer the question of which product people prefer, or how much the product is liked. The tasters included in these tests are current consumers of the product or potential ones. There is no need for taster training. Optimally, there are more than 30 panelists. There are several types of preference and acceptance tests. A common one is the 9-point hedonic scale.
In comparing the hedonic to the preference taste test, there’s a few notable differences. The hedonic test is set up in the same manner as the preference taste test. At least two samples are compared to determine which product people prefer. A large number of similar responses must be obtained to determine that people prefer one product more than the other. The minimum number of similar responses needed to determine if the preference is significant is based on the total number of responses obtained, and helps determine a 95% significance.
The hedonic scale is used to determine degree of acceptability of one or more products. This scale is a category-type scale with an odd number (five to nine) categories ranging from “dislike extremely” to “like extremely.” A neutral midpoint (neither like nor dislike) is included. Taster rate the product on the scale based on their response.
If more than two samples are evaluated, a preference ranking test is performed. Usually three to five samples are the most that can be efficiently ranked by a taster. This test asks the taster to order the samples based on preference, with a ranking of “1” meaning most preferred.
