Reviews
Part of A plain-English look at how food tasting methods work
How to run a triangle test on two pantry products at home
The triangle test is the one piece of laboratory sensory method a home cook can run properly, and it settles arguments about whether two products differ at all.
What to take away
- A triangle test answers one question: can you tell these two apart at all?
- It deliberately does not tell you which is better, or in what way they differ.
- You need a second person to pour, because the test only works blind.
- Chance alone gets it right one time in three, so a single correct answer proves nothing.
What the test actually is
Three samples, two of them identical, one different. The taster's job is to pick the odd one out. That is the whole design.
The published method is an ASTM standard, and its scope is worth reading before you run one. The triangle test covers a procedure for determining whether a perceptible sensory difference exists between samples of two products. It applies whether the difference is in one sensory attribute or in several, and it is used when the nature of the difference is unknown. It does not determine the size or the direction of the difference, and the attributes responsible are not identified.
The standard also notes that compared with the duo-trio test, the triangle test reaches an equivalent level of statistical significance more efficiently.
Why serving consistency is the whole game
If the three samples differ in temperature, volume or glass, you are testing your pouring rather than the product. ASTM has a separate guide for exactly this. Its serving protocol guide states that consistency must be maintained in all aspects of preparation and serving of samples to ensure reproducible data from sensory respondents, and that the guidelines for consistency may change with the test objective.
In practice that means identical vessels, identical volumes, the same temperature, and codes rather than labels.
The codes matter more than they look. Cups labeled A, B and C invite an assumption that A came first or is the original, so sensory work uses arbitrary three-digit numbers instead. It costs nothing and removes a bias you would never notice yourself making.
Running one at your kitchen table
- Get a helper. They pour, they code the cups, they keep the key.
- Use three identical opaque cups, coded with three digits each.
- Same volume, same temperature, poured within a minute of each other.
- Taste in the order presented, left to right, with water between.
- Record only which cup was the odd one, then repeat with a fresh set.
Reading the result honestly
| Trials | Correct answers that mean something |
|---|---|
| 1 | Nothing. Chance is one in three |
| 6 | Five or six is a real signal |
| 12 | Eight or more is a real signal |
Those are rough working thresholds, not the standard's tables, which set the numbers formally against a chosen risk level. The point is the shape of the thing: you need repeats, and you need clearly more correct answers than chance alone would hand you before the result means anything at all.
What to do with a positive result
If you can reliably tell them apart, then and only then is it worth asking how. That is a different method, described in how food tasting methods work, and the practical setup for it is covered in building a tasting flight.
Common questions
Can I run this alone? Not properly. Knowing which cup is which destroys the test, and you cannot unknow it.
Does it work for strong flavors? Yes, but fatigue sets in fast. Keep portions small and leave time between sets.
What if I get it wrong every time? Getting it wrong consistently is also a signal, usually that you have found a difference and are systematically misreading it. Repeat with fresh samples.
Is this overkill for buying jam? For one jar, yes. For deciding whether a more expensive supplier is actually different, it is the cheapest honest answer available.
