FOOD9102 Chap.4 Discrimination Tests: 2-AFC, Triangle and Duo-Trio
Discrimination Tests: 2-AFC, Triangle and Duo-Trio
Discrimination tests ask whether a difference between products can be perceived.
This chapter follows the Week 2 lecture through desired and undesired differences, overall versus single-attribute questions and the situations that justify a test, then sets out the directional paired comparison (2-AFC), the non-directional paired comparison, the same-different test, the triangle test, the duo-trio test and the A-not-A test, with their guessing rates and serving orders.
A decision tree and a fully designed example show how to pick the right one from a brief.
The chapter leans heavily on the lecture’s summary table of which test suits which scenario, and on the workshop’s vignettes, which show that more than one test can be defensible if the reasoning is stated.
What this chapter covers
- 01
Wanted and unwanted differences between products
- 02
Overall (unspecified) versus single-attribute (specified) questions
- 03
The 2-AFC and its forced choice
- 04
Non-directional paired comparison and same-different tests
- 05
The triangle test and its six serving orders
- 06
The duo-trio and A-not-A tests for reference matching
- 07
Choosing a test from the business objective
Worked example · free
Choose the discrimination test for three briefs
- 1(a) Duo-trio with the fresh muesli as the reference, because the question is whether a sample matches a standard; P = 1/2.
- 1(b) Triangle test, because several attributes may change and the company wants to know about any overall difference; P = 1/3.
- 1(c) 2-AFC asking which sample is more lemony, because one attribute was changed in a known direction; P = 1/2.
Key terms
- Directional paired comparison
- A directional paired comparison serves two samples and asks which has more of a named attribute, forcing a choice.
- Forced choice
- A forced choice is the rule that panellists must pick an answer even when guessing, so every ballot supplies data.
- Same-different test
- A same-different test asks whether two samples are the same, using matched and mismatched pairs.
- A-not-A test
- An A-not-A test asks panellists who have learned sample A whether a later sample is A or not.
- Specified attribute
- A specified attribute is the single named characteristic, such as saltiness, that a test asks panellists to judge.
- Reference sample
- A reference sample is the labelled standard, such as the original recipe, that the coded samples are compared against.
Discrimination Tests: 2-AFC, Triangle and Duo-Trio FAQ
When is a discrimination test a waste of money?
When the difference is obvious. The lecture is clear that discrimination tests are for subtle differences, where you genuinely do not know whether consumers can tell. Testing a difference everyone can see or taste yields no useful data, although a change that is obvious to the company may still be subtle to the general population.
Why does the triangle test need six serving orders?
Either product can be the odd one out, and the odd sample can sit in any of three positions. Balancing both gives AAB, ABA, BAA, ABB, BAB and BBA, so no position or product is favoured and the guessing rate stays at one in three.
Can a discrimination test tell me which product consumers prefer?
No. It shows only whether a difference is perceived, not its size and not which version people like. If the brief involves liking, for example when a stored product becomes unpleasant, pair the discrimination test with an affective test.
Why must the 2-AFC name the direction of the attribute?
Because the test is directional: it asks which sample is sweeter, not whether sweetness differs. You can do that because you know your own product and which way it was changed. If you do not know the direction, the non-directional paired comparison is the right choice instead.
Assessment move
Draw the decision tree from memory, then test yourself on product briefs from news stories or packaging: for each, state the objective, the test, the guessing rate, the serving orders and two biases. Practise writing the ballot instruction for each test in your own words, since the laboratory hands you ballots and expects you to recognise the test from them.
Keep the guessing rates beside the test names on your summary sheet, because the next chapter's analysis depends on them. For each test, also write one situation in which it would be the wrong choice, since quiz questions often test the boundary between two tests.
Working through Discrimination Tests: 2-AFC, Triangle and Duo-Trio in FOOD9102? Sia is AskSia’s AI Science tutor — ask any FOOD9102 Discrimination Tests: 2-AFC, Triangle and Duo-Trio question and get a clear, step-by-step explanation grounded in how FOOD9102 is taught and assessed. Read this chapter free, then take your hardest questions to Sia.
