SENSORY EVALUATION: ANALYTICAL TESTS
Diagram
Profiling a food or spotting a difference
Descriptive analysis
A trained panel lists every attribute of a food, sweet, sour, crunchy, brown, and rates how strong each one is. The result is a full profile.
↓
Discrimination testing
Panellists are asked only one thing: are these two samples different. The result is a yes or a no.
↓
Same family
Both are analytical tests: they measure the food, not liking. They differ in how much they ask.
Figure 1: Descriptive analysis profiles all attributes; discrimination testing only asks if two products differ. Drawn in our own style from the descriptive analysis chapters and the discrimination testing chapter (p. 116) of Lawless and Heymann, Sensory Evaluation of Food.
Introduction
Descriptive analysis and discrimination testing are both analytical sensory methods, but they ask different questions. Descriptive analysis profiles all attributes of a product, with a trained panel rating each one, while discrimination testing only asks if two products differ. Lawless and Heymann cover descriptive analysis in its own chapters and discrimination testing in chapter 7 (p. 116), and the ASTM Manual on Descriptive Analysis Testing for Sensory Evaluation covers the descriptive method.
The core difference is depth: descriptive analysis says what a product is like in full, and discrimination testing says only whether two products can be told apart.
Quick comparison
Table 1: Descriptive analysis compared with Discrimination testing
| Aspect | Descriptive analysis | Discrimination testing |
|---|---|---|
| Question asked | What is this product like, attribute by attribute | Are these two products different |
| Result | A full profile: every attribute with its intensity | A yes or no: different, or no difference found |
| Who tastes | A trained panel | Trained or untrained panellists, according to the test |
| How many samples | One product at a time, profiled in full | Two or three samples, compared with each other |
| Vocabulary | A shared vocabulary, built by the panel for the product | No vocabulary needed: the panellist only picks the odd one or the match |
| Time per test | Long: each product is rated on many attributes | Short: one choice per panellist |
| What it can show | Which attributes differ, and by how much | Only that a difference exists, not what it is |
| Use | Guiding recipe changes, describing products, setting standards | Checking whether a change has changed the product at all |
| Sensitivity | Can find small differences in single attributes | Can find a difference in any attribute, but cannot name it |
| Cost | High: training and long sessions | Low: quick tests with many panellists |
Scientific explanation
Definition
Descriptive analysis is an analytical method in which a trained panel lists the sensory attributes of a product, agrees on a vocabulary for them, and rates the intensity of each attribute, giving a full profile of the product. Discrimination testing is an analytical method in which panellists are asked only whether samples differ: the triangle test, the duo-trio test and the paired comparison test are all discrimination tests.
Lawless and Heymann cover descriptive analysis in its own chapters and discrimination testing in the discrimination testing chapter (p. 116). The ASTM Manual on Descriptive Analysis Testing for Sensory Evaluation is the standard reference for the descriptive method.
Working principle
Descriptive analysis works by breaking a food into its parts: the panel agrees on words for what it senses, such as sweet, sour, crunchy or brown, and rates each one on a scale. The ratings together make a profile, which can be drawn as a chart and compared with other products.
Discrimination testing works by forcing a choice: the panellist must pick the odd sample or the matching sample, and the count of right answers is checked against guessing. It can prove that a difference exists, but it cannot say what the difference is, because the panellist is never asked.
Apparatus
Descriptive analysis needs a sensory test room with separate booths, reference standards for each attribute, and ballots with intensity scales for every attribute. Discrimination testing needs booths, coded samples, water and crackers, and a simple ballot with one choice. Both need a trained panel leader for descriptive work.
Measurement procedure
For descriptive analysis, the panel first tastes the product together, agrees on the attributes and the words for them, practises rating with reference standards, and then rates each sample on every attribute. For discrimination testing, panellists are served coded samples in a balanced order and make one choice: the odd one out, or the one that matches the reference.
The books used here describe both methods in full; they do not give one single standard procedure for both, because the two methods are run differently.
Factors affecting readings
For descriptive analysis, the training matters most: without a shared vocabulary and practice, the panel cannot agree, and the profile means nothing. For discrimination testing, fatigue matters most: the samples must be small and the order balanced, or the panellist tires and guesses. The size of the real difference matters for both.
Formulas and calculations
The books used for this article give no formula for this comparison, so none is given here. The comparison is qualitative: it rests on definitions, materials and method, not on a calculation.
Industrial applications
- Descriptive analysis is used to profile a new product, so the recipe can be aimed at the right attributes.
- Descriptive analysis is used to set a standard for a product, so later batches can be matched to it.
- Descriptive analysis is used to track how a product changes over its shelf life, attribute by attribute.
- Discrimination testing is used to check whether a recipe change, a new supplier or a new process has changed the product.
- Discrimination testing is used in quality control, to catch a change before the buyer notices it.
Advantages and limitations
- Descriptive analysis, advantage: shows exactly what a product is like and what changed, attribute by attribute.
- Descriptive analysis, limitation: slow and costly, needing a trained panel and long sessions.
- Discrimination testing, advantage: fast and cheap, and it can prove a difference exists with a simple test.
- Discrimination testing, limitation: cannot say what the difference is, only that one exists.
FAQ
Which should I use first?
Discrimination testing first, to find out whether a difference exists at all, and descriptive analysis second, to find out what the difference is. The two are often used in that order.
Can discrimination testing tell me how big the difference is?
No: it only says a difference was found. Descriptive analysis rates each attribute, so it can say how big the difference is and where it sits.
Do I need a trained panel for discrimination testing?
Not always: the triangle and duo-trio tests can be run with untrained panellists, though trained panellists find smaller differences. Descriptive analysis always needs a trained panel.
What is a sensory profile?
A list of a product’s attributes with the intensity of each, as rated by a trained panel. It is the result of descriptive analysis, and it can be drawn as a chart.
Can descriptive analysis tell me if buyers will like a product?
No: it measures the product, not liking. Liking is measured with consumers in hedonic tests.
References and further reading
- Harry T. Lawless and Hildegarde Heymann, Sensory Evaluation of Food: Principles and Practices, Springer, 1999 (Google Books page read)
- ASTM International, Manual on Descriptive Analysis Testing for Sensory Evaluation
