E-Book Sensory Evaluation

1
S E N S O R Y EVALUAT I ON
Acknowledgements
Actia and Actia centres thank all those involved for the efficiency of their participation.
Special thanks to Christian Touraille (Inra) and Joseph Hossenlopp (Cemagref) for Guide
their enthusiasm, the quality of their work and contributions and for their generous
help in the three years it has taken to produce this document. of good practice
Grateful acknowledgement also to the French Ministry of Agriculture and of Fishing,
and to the Ministry of National Education, Research and Technology, who gave their
financial support.
2 2001 3
Pre f a c e
I S O 3591
Building on forty years of experience, sensory evaluation is now a decision-making

tool ideal for all who work in the field of food quality and innovation, for manufacturers,
processors and retailers.
These techniques are acknowledged to be of strategic interest to organizations, and
have become increasingly so with time. Sensory analysis may be concerned with
several different aspects of a product. The most fundamental may be the study of
raw materials, and the formulation of new products. Then there is the optimisation
of manufacturing processes or packaging. Once a product is in production, sensory
analysis may be concerned with quality assurance, product comparisons or attributes
of the final consumer. Here, assessment of food acceptability and of the pleasure it
causes are all-important. Sensory analysis is also used in other industrial sectors,
such as pharmaceuticals, cosmetics, cleaning products and the automobile industry.
However, sensory evaluation is still in many ways immature. Methods are generally
tried and trusted, but as yet poorly formalized. Nor have they been approved or
validated by the scientific and professional authorities in the same way as for physical
and chemical analytical methods. In particular, the so-called hedonic approach, invol-
ving the study of consumer preferences, has been relatively little codified. Hence, in
spite of its widespread adoption, sensory evaluation is far from fully understood, and
raises questions which are still unanswered.
Wishing to ensure that the discipline is better recognised, and to establish the basis of
a relationship of trust among all involved, Actia's “sensory evaluation” network has
drawn on the expertise of professionals from 14 technical sectors. As part of a progress
initiative, pooling the technical skills and the experience of specialist engineers and
scientists working in many product fields, the Actia network set itself the goal of defining
and formalizing its own quality baseline.
While respecting the existing standard procedures, and giving full consideration to
current scientific knowledge, further requirements have been developed for analytical
and hedonic testing. This now takes the form of a guide of good practice in the sensory
evaluation field.
This guide, intended both for clients and service providers in the sensory evaluation
field, should not be considered a methodologically prescriptive work, but more as a
methodological compendium. But it also testifies to a focused approach to quality,
which has guided our whole undertaking.
A service provider adopting this guide and its practice as a benchmark should undertake
also to comply with its requirements, and so assure a high level of service quality.
Such a commitment to the client, entered into in a spirit of openness and transparency,
provides an assurance that a rigorous and reliable methodological approach has been
adopted to qualitative market research.
Michel Caugant
President of Actia
4 5
Sensory evaluation . Actia . Guide of good practice . 2001 Sensory evaluation . Actia . Guide of good practice . 2001
As valuable in its own way as a Quality Label, This document testifies
the purpose of this guide is to contribute to and to a commitment
ensure reliable results, and so enhance the value to a service of such quality
of the skilled work undertaken by our organizations.
For too long, the qualitative analysis of food products was limited to bac-
ter-iological and chemical assays. The most important quality for a consumer,
The Carrefour intention stands for customer satisfaction. The high quality of
sensory quality – which includes taste, smell and texture – had been little or
products sold under our brand name is an absolute priority for us.
not at all measured. Of course, such an approach raises difficulties, for “there
is no accounting for taste”, as the saying goes. There used to be no objective
Consumer studies enable us not only to measure quality as perceived by our
measurements available, such as temperature, consistency or chemical com-
customers, but also to better understand their preferences, and so develop
position, when making a quality assessment. Sensory evaluation is therefore
products to meet their expectations. Sensory studies are now one of the
an attempt to overcome these very real difficulties.
baro-meters of quality which Carrefour food products cannot afford to do wit-
hout. As an instrument of measurement, any barometer must be reliable.
Actia's guide of good practice is a major step forward when it comes to upholding
the credibility and reliability of these studies. The techniques developed are of
The drafting of a guide of good practice in the field of sensory evaluation, with
the greatest importance for the introduction of new food products and for their
its definition of precise methodologies, will ensure that consumer studies are
quality control. Quality is the basis of customer satisfaction, and should be the
undertaken in a spirit of rigorousness and attention to detail. This guide will
focus itself of assessment and measurement.
also help us select partner laboratories to work with, particularly now that we are
faced with the growing number of often self-styled specialized organizations.
We therefore congratulate the Actia Centres which adopted this approach, and
pooled their skills to do so. This Actia document testifies to a commitment to a
The product of a real commitment to a Quality Label approach, this guide will
service of such quality, that it is in itself worthy of a Quality Label. Congratulations
provide an assurance that our results are reliable. It will also enhance the value
to all involved in this endeavour - and for the tremendous work that has gone
placed on the contribution of skilled organizations in the food quality field.
into it.
Sensory evaluation has now come of age.

Chantal Jaquet
Carrefour
Jean-Claude Guilloteau
Fromagerie Guilloteau
6 7
C onte nts
General methodology
1. Sensory evaluation: why ? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

2. Basic need to differentiate
analytical tests from hedonic tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3. Implementation of sensory evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
First stage: formulating the question and specifying an objective . . . . . 17
Second stage: choosing one's approach and test procedure . . . . . . . . . . 18
Analytical tests
1. General principles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2. Testing procedure (what type of test for which result?) . . . . . . . . . . . . . . . . . 24
3. Assessors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
4. Products . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
5. Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
6. Specific procedures for discriminative and grading tests . . . . . . . . . . . . . . . 28
Discriminative tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
Grading tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
7. Descriptive tests: the sensory profile . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
Hedonic tests
1. General principles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2. Testing procedure (what type of test for which result?) . . . . . . . . . . . . . . . . 40
3. Consumers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4. Products . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
5. Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
6. Special features of the various hedonic tests . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
8 9
Contents
Service provider quality assurance Hedonic tests

Consumers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83 17
Experimental context . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84 18
1. General guidelines for quality management . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
Questionnaires . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85 19
2. Guidelines for follow-up of assessors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
Preference test by paired comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86 20
3. Guidelines for follow-up of consumers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
Preference test by ranking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87 21
Hedonic scoring test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88 22
Statistical tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89 23
Relationship between service provider and client
Analytical and hedonic tests
Products . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 24
1. Studies performed in line with the recommendations of this guide . . . . . . 57
Proposal and service contract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91 25
2. Service provider contracts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
Study report . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92 26 & 27
3. Study report . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
Appendices
Summary tables N u m b e r s o f t a b l e s
Relationship between analytical and hedonic measurements . . . . . . . . . . . . . . 97

Objectives of sensory evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 1
French and international standards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99

Analytical tests
General standards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
Assessors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 2
Analytical tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
Selection, training and follow-up of panel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69 3 & 4
Hedonic test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
Discriminative tests
Standards for specific product or class of products . . . . . . . . . . . . . . . . . . . . 103
Triangle test for differences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71 5
Statistical methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
Triangle test for similarity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71 5
Duo-trio test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72 6
Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
Two out of five test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73 7
Grading tests
Reference works . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
Paired comparison test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74 8
Ranking test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 9
Scoring test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76 10
Descriptive tests: sensory profile
Research and selection of attributes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77 11 & 12
Development of profiles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79 13 & 14
Statistical tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81 15 & 16
10 11
General
methodology
12 13
G e ne ra l m e th od ol og y
1. Sensory evaluation: why?
A degree of confusion and uncertainty continues to exist in the field of sensory

evaluation. The present guide will therefore seek first to give the reader some
insight into the field by focusing on the definitions and methodological details
which are part of the framework of good practice in this technical field.
The analytical approach includes techniques making it possible specifically to

measure the sensory characteristics of a product. This can be done to track the
evolution of properties over time, to compare them with other products, to fol-
low-up specifications.
The hedonic approach, on the other hand, studies acceptability or preference by
an identified group of consumers, for a food product, by assessing the pleasure
caused by that product when tasted or consumed.
The two approaches are related but should never be confused. Each has its own
constraints and limits, which arise from technical considerations and the scope
of current knowledge.
For analytical testing, no standard is currently available. It is therefore impossible

to undertake “absolute” measurements. Sensory
A ny reaction to a sensory stimulus can be
measurement is in practice a comparison among broken down into three distinct dimen-
products or to an implicit benchmark, itself memo- sions. It is important that these should
ry-dependent. The second special feature of this field not be confused. These are qualitative
perception, quantitative perception and
of study arises from individual variations in percep-
hedonic reaction. A sensation is decribed,
tion, particularly in the perception of flavours. This first, in terms of its intrinsic characteristics
is one reason, why a group of assessors is used and (salty or sweet taste, strawberry or vanilla
not a single expert. The group of assessors should flavour, granular texture). Second, it can
furthermore be trained before assessment can take be quantified in intensity (high or low
degree of sweetness). Finally, the sensation
place. The duration of training will depend on the caused in the person perceiving it may
difficulty of the task. give greater or lesser pleasure. The latter
dimension is the so-called hedonic
dimension, and is never to be interpreted
In the field of hedonic testing, the tests described in
as an attribute of the product itself. The
this guide make it possible to assess the acceptabi- sensation of pleasure is a purely individual
lity of or preference of consumers for the products one and will depend on the characteristics
tested. During these studies, an analysis is under- of the person tasting the product, especially
taken less of the products themselves, than of the since reactions may be culturally deter-
mined. It is therefore both unnecessary
characteristic reactions of consumers to the pro- and impossible to seek to determine
ducts. This highlights the importance to be attached to whether a product is sensorially “good”
the structure of any group of consumers involved in or “bad”. A product is neither good nor
hedonic testing. bad, except for a given consumer at a
given time and in a given context.
14 15
General methodology
2. Basic need to differentiate analytical tests from hedonic tests 3. Implementation of sensory evaluation
When attempting to bring about a better professional approach, it is impossible a] First stage: formulating the question and specifying an objective
to overemphasize the fundamental distinction between the hedonic and analytical
approaches. The clear definition of objectives is an important precondition when it comes to
responding to customer requirements. If the fundamental distinctions outlined
The analytical approach describes food products and compares them, using a above have been taken into account, the most frequently encountered questions
group made up of a limited number of trained assessors, whose performance are summarized below:
was the subject of prior assessment. The reliability of the group is dependent on the
number of assessors and the degree of their training. Analytical tests
Why should not the two types of assess-
ment be undertaken simultaneously ?
Product comparison
Firstly, a trained group becomes, with A distinction should be made between discrimination
(Highlighting differences or similarities)
experience, particularly critical towards analysis, which focuses on overall perceptions (simul-
the products submitted to it for assessment. Are there sensory differences between products?
taneously qualitative and quantitative), and permits
Additionally, the small numbers in the What are the sensory likenesses between products?
trained group and the fact that these are
differentiation, as against descriptive analysis, which
How long can the product be stored
not consumers representative of the relates to the qualitative and quantitative dimen-
without changing the perception of that product?
population at large, makes it impossible sions of the product, using attributes and scales of
to consider the group to be representative assessment.
of actual consumer preferences. Product formulation
On the other hand, consumers, because (Assessment of a prototype, imitation of a product)
they do not have the necessary training, The hedonic approach, on the other hand, assesses
Will the sample achieve the purpose attributed to it?
are unable to describe their own sensa- the degree of pleasure caused by a product, by a
tions. This makes it unavoidable to use a Will the sample be at a closer or more distant remove
comparative determination of the numbers of consu-
group of persons who can describe their from the target product in sensory terms?
mers indicating a preference for one product over
sensations in sensory and evaluative How can the product be made more like the target product?
terms. Thus, the same people cannot be others. For this to occur, surveys must be undertaken
What sample is closest to the target product?
used for both analytical testing and hedonic across a representative sample of consumers, requiring
testing. What are the attributes contributing to significant differentiation?
prior definition of their characteristics, insofar as
However, there is an approach known as
they are the target of the assessment.
the “ideal profile” approach, which invol- Quality control and product follow-up
ves asking consumers to describe the
(In terms of raw materials, compliance with specifications, research into deviations
properties of a product and to enquire In the following pages, which are broken down into
about their explanation of their prefe- from a recipe or from manufacturing parameters, change in the stability of the
two main chapters, details are given of the various
rences. This is the only case in which product, the influence of packaging, and storage conditions)
types of tests used in sensory evaluation, and the
consumers can be considered to be equi- What are the sought-after sensory specifications?
valent to trained assessors. technical procedures involved. If it is intended to
Does the product meet the same sensory specifications
Furthermore, statistical techniques, such embark on a study of the sensory characteristics of
as preference mapping, aim at establi- as the target products?
products, reference should be made to analytical
shing a relationship between a number of What are the tolerated variations in intensity for each attribute?
tests. But for a study relating to consumers, hedonic
sensory characteristics and preferences Are there perceptible differences between the target product
expressed by consumers, and hence iden-
tests has to be realised; the methodology is given in
and the tested product?
tifying the type of perceptions underlying the relevant chapter.
consumer preferences.
If there is change to the recipe, manufacturing process or packaging,
A document attached to this guide details how does sensory quality change?
the principles adopted in this approach. Do these changes cause the product to diverge from any standard?
How do the sensory characteristics of a product change in storage?
16 17
General methodology
Hedonic tests For the evaluation of a single sensory magnitude (for example, sweet
taste). Quantification may be by ranking, grading or scoring.
Study of consumer reactions
The third case would involve the simultaneous quantification of
(Acceptance or rejection, preference, pleasure intensity)
more than one sensory magnitude. This is the profile method describing
Which products are preferred?
the sensory properties of the food product, by defining a magnitude
Which products are rejected?
whose intensity can be assessed on some scale. The method is based
To what degree is a product appreciated?
on constructing a list of attributes, each element of which is quantified
Are consumer preferences homogeneous?
by members of a group. This requires intensive training to achieve a
How far can products change before there is any effect
perfect definition of the attributes themselves and a homogeneous
in terms of their appreciation?
approach to quantification by the assessors.
Can categories of consumers be devised
on the basis of identified preferences?
Hence, depending on the tests implemented, and on the testing approach adopted,
How is the product tested positioned against competing products?
the information elicited on the products may be of the following types:
binary (presence or absence, different or similar) from difference
Combined studies
testing,
Relationship between the sensory characteristics of products and consumer unidimensional (ranking or scoring according to a given sensory
preferences property),
What are the characteristics which explain why a product multidimensional (assessment of multiple product properties), using
is liked or rejected by certain consumers? descriptive testing.
Once objectives are clearly defined, it becomes possible to determine whether In the case of the hedonic approach, the consumers evaluate the degree of
assessment should focus on product characteristics or on consumer reactions, pleasure they feel from the food. The pleasure experienced by the assessors will
or on both at the same time. depend on a number of factors, and more particularly on the perception of the
whole set of sensory characteristics of the products. Assessment usually
b] Second stage: choosing one’s approach and test procedure generates unidimensional results, assessed on a scale of acceptability or pleasure.
Consumers can also express preferences among products by comparative
On the basis of objectives defined with the customer, the project manager preference testing. Their preferences are assessed using paired comparison
should determine the type of tests to be performed on products. tests or preference ranking tests. Preferences are built up from complex notions
of sensory complexity. It is therefore rare for consumers to agree about their
The analytical approach may generate three types of response: preferences.
Binary, unidimensional, multidimensional.
The first case involves the determination of whether two product
samples are similar or different in sensory terms. Difference tests are
the main tool for this purpose. They yield binary (two-level) data whose
interpretation must take account of the acceptable levels of statistical
risk of error of various kinds, in the light of the purposes of the test.
18 19
Analytical
tests
20 21
Ana l y ti c a l te sts
This part details the procedures used when performing different tests for analy-
tical purposes - tests which are conventionally known as analytical. The aim
here is to analyse the sensory characteristics of a product, in order to monitor
and or to control them.
The presentation below gives pride of place to Actia's requirements. These are
specified in such a way as to ensure that testing takes place under the best
possible conditions. The approach adopted is based on the AFNOR standards,
“Control of food product quality and sensory analysis” and on the technical and
professional skills of specialists working in the Actia technical centres.
The methodology adopted must take into consideration the assessors themsel-
ves, who test the products, and finally the procedures used. The different types
of tests are summarized, along with the basic purpose of each test. There then
follows a more detailed description of the special features of each. A more com-
prehensive description is given of approaches to the sensory profile, which is
the area of greatest complexity. A theme-based list of the standards to which
reference is made in the procedures is given at the end of the guide.
1. General principles
To ensure that product research is carried out under optimal conditions, the
analyst will closely examine, with the client, the objectives, the context and the
type of products involved.
Careful initial thought is essential to determine:

the one or more tests to be used,
the number and kinds of assessor participating in the tests,
the nature of the product samples presented for testing,
the approach to be used (gathering of responses,
conditions of tasting, statistical issues, etc…).
22 23
Analytical tests
3. Assessors
The question initially raised by the client may be reformulated to determine the
one or more types of test to be performed in the light of the results expected.
Great attention should be given to the various aspects to be taken into account
Among tests currently in use are those detailed below.
when setting up and managing groups of assessors for analytical purposes.
To determine whether there are perceptible sensory differences between

a] Size of panel
products, or on the contrary similarities, without describing their nature.
The number of assessors will depend on:
Discriminative tests the nature of the products,
the type of test,
- Determining that the difference between two products
the level of qualification chosen for the test,
is significantly perceptible.
the acceptable statistical risk (risk a / risk ß).
Triangle test for differences
Duo-trio test (test more exposed to statistical unreliability,
The minimum number of assessors required to interpret the test, whose responses
but simpler for the assessors to perform)
are independently processed, is specified for each test in the detailed description.
Two out of five test (statistically more robust tests, can
be performed with small numbers, but tiring for assessors)
b] Qualification of assessors
- Determining whether or not there are perceptible differences
between two products. Selection will initially relate to the availability and motivation of assessors, and
Triangle test for similarity then depend on their capacity for sensory expression.
Generally speaking, the training required for assessors depends on the purpose
Positioning different products as a function of their sensory character- of the study, on the type of test and on the product itself. Several levels of skill
istics, once nature of the difference between the products is known. may be distinguished:
Grading tests Trained assessor: a person who has already taken part in one or more
- Assessment of the existence of a perceptible differentiation between . . . . . . . discriminatory sensory tests of the same type as the one now being used.
two products according to a determined sensory characteristic. Trained assessors also participate in discriminative tests.
Paired comparison test
- Assessment of the perceptible differences between several pro- Selected assessor: person chosen as a function of his or her capacity to
ducts, depending on the intensity of that difference undertake sensory testing, and whose performance has been controlled.
Ranking test The selected assessors have sensory aptitudes in regard to the product
- Assessment of one or more products according to the perceived under consideration. These aptitudes are validated at the time of the
intensity of a determined sensory characteristic. selection, developed by training, and then controlled. All selected
Scoring test assessors may participate in all types of analytical tests.
Description and assessment, in terms of both the intensity and quality Expert assessor: selected assessor who has acutely developed sensory
of a set of perceived characteristics, of one or more products (description perception. This assessor is trained in the use of sensory evaluation
tests). methods. The expert assessors have sensory skills in regard to a deter-
mined product. These skills are validated at the time of selection,
Sensory profile approach
developed by specific training, and controlled to ensure that they are
repeatable. Expert assessors participate in all types of analytical tests.
24 25
Analytical tests
These assessors are specifically required when great precision and c] Presentation of products
constancy in performance over time are sought.
Samples of products shall be made anonymous for the assessors, by means of
coding. Samples shall be presented according to a balanced block design, and
At the time of testing, assessors should have a homogeneous level of training
under the same conditions (temperature, quantity, recipients…).
in regard both to the products and to the test techniques adopted.
5. Methods
4. Products
The test instructions drafted by the study manager, and the questionnaire and
a] Choice of products
principles of testing, shall be given to the assessors by the panel leader.
The criterion for the choice of products are determined in the light of the purpose
of the study. Special attention should be given: a] Collecting responses
to the representativeness of the sample of products, taking into account
Questionnaires are used to gather assessments from the group of assessors.
the greater or lesser homogeneity or heterogeneity of the products, or
Special attention should be devoted to the questionnaires to ensure that res-
batches of products. The validity of results will depend on the degree
ponses are precise and avoid errors in terms of scales or definition of terms,
of representativeness of the products in regard to the product space
among others.
under study, and the representativeness of the samples in regard to
The design of the questionnaire should ensure the traceability of the assessor /
the product itself. The sampling plan should take into consideration
product / timetable / date / questionnaire / study, and do so irrespectively of
these elements in order to ensure that variability among the products
the medium (paper, display system, graduated scale…).
or samples is under control, in the course of repeated testing,
to the place of product sampling (factory or point of sale),
b] Questions
to the number of products presented per session and per assessor
(to be taken into consideration in the light of tests and type of products), In analytical tests, questions should make no references to hedonic concepts in
to special constraints at client request. the choice of terms or questions raised.
Responses, if any, which relate to preference or final acceptability should be
b] Product identification excluded from the statistical analysis or interpretation based on measurement
for analytical testing purposes. The seeking of responses in regard to preference
Identification of products, such as will guarantee their traceability and the
or acceptability necessarily requires a hedonic approach. Open questions
traceability of the information relating to them. Required in this respect are:
should be used to glean information on the context or background of the tasting
names such as the trading name, brand name, and or the manufacturer’s
(perceived defects, anomalies…).
name, batch number, DLC* or DLUO* * , packaging code, date of sampling,
or of receipt at the service provider's. All this information should be
c] Scales of response
systematically recorded,
further information may be important for particular products such Numeric, semantic or motor scales of response may be used. The same scale is
as weight, type of packaging, list of ingredients, specified mode of generally applied to all questions. A continuous scale is incremental. A disconti-
preparation or use of product, type of manufacturing process, price, nuous scale should contain a minimum of five graduations.
conditions of storage. This information should be recorded so far as is
necessary and possible.
* Date limite de consommation

** Date limite d’utilisation optimum
26 27
Analytical tests
d] Conditions of tasting the standards adopted for each type of test.

Products should be presented simultaneously, with mandatory order of tasting.
Requirements relating to the conditions of assessment of products aim at mini- Assessors are free to take a second taste of the sample products as they wish,
mization of bias and disturbance factors which may adversely affect tasters but must provide a response (forced choice).
during the assessment of the products. Among the conditions required are:
the tests take place in a laboratory compliant with standards, For all tests, the approach should first define the number of assessors required,
the assessors evaluate products without communicating between in the light of acceptable statistical risks agreed on a prior basis with the client.
themselves, Subsequent provision should be made for the ability to interpret data by
the assessors respect a minimum wait time between two samplings of applying binomial or possible χ2 laws.
products. Assessors should be given access to a food product attenuating
the sensory characteristics of the previously tested product (water being Triangle test for differences
the most frequently used, while green apple, bread crumbs or dried (Determination of a perceptible difference between two products)
toast may be used for products with a lingering taste),
a prior meeting should be held to inform the assessors of the instructions This test involves the presentation of three samples, two of which are from a
to be complied with before and during the tasting: single batch of products. The aim is to determine which of the three sampled
- not to consume highly aromatic products prior to the testing products is perceived as different from the two others.
(coffee, chewing-gum, cigarettes…), The size of panel participating in the test will be determined in the light of the
- not to wear highly aromatic products (perfumes…), values of the desired a, ß and Pd values, with a minimum of 18 assessors, and
- be punctual and give forward notice in the event of absence the intention to minimize the a risk.
or temporary incapacity (cold, dental care…).
Duo-trio test
It is difficult to set down in advance a limit on the number of samples presented (Determination of perceptible difference D iscrimination: the question raised may
per assessor and per session. This will depend on the type of test, the nature of between two products) relate, for example, to the difference
between two samples: “Which of the
the product and the degree of taste remanence. Twelve products (six for a profile)
This test presents two samples of products from two three sample products do you perceive
is often sufficient.
different batches, which the assessors compare to a as different from the two others?”
Variations from one session to another relating to experimental conditions
benchmark product taken from one of the two bat-
(assessor, product or context) should be recorded. S tatistics applicable to discriminative
ches. This determines which of the two samples is
tests: risks a, and possibly ß and Pd, are
6. Specific procedures for discriminative and grading tests perceived as identical to the benchmark.
agreed on a prior basis acceptable to the
The size of panel participating in the test will be client. If the purpose is to highlight diffe-
determined in the light of the desired a, ß and Pd rences, interest will focus on a. But if
Special features of methodology, particularly relating to statistical tests and
values. There must be a minimum of 18 in the group, interest is in similarities, ß risk will be
to the various discriminative or grading tests are presented in this chapter.
of at least initiate level, and the intention to minimize taken into account.
Profile procedures are described in chapter 7.
a risk. Risk a or type-1 error: probability of
concluding that there is a perceptible
a] Discriminative tests difference when there is none in reality.
Risk ß or type-2 error: probability of
General concluding that there is no perceptible
difference when there is one in reality.
The assessors should at minimum have been trained to the level of initiates.
Pd: proportion of entire population of
These tests are performed according to a balanced experimental design in order assessors able to make a distinction
to cancel out the effects of presentation order and if necessary the effects of between two products.
initial sample presentation bias. This plan should reflect with the order shown in
28 29
Analytical tests
This type of test is simpler for the assessors and, from the quality control point b] Grading tests
of view, allows the testing of more batches during a single session. However, its
These are tests whose purpose is to evaluate differences in perception, or to
effectiveness is less than that of the triangle test. Because the probability of
classify or to evaluate products according to the perceived intensity of a given
a correct answer by chance (one in two) is higher than in the triangle test
characteristic or property. The most frequently used tests are described below.
(one in three).
General
2 out of 5 test
(Determination of a perceptible difference between two products) These tests are performed only if the difference is known and relates to a single
dimension, characteristic or property.
This test involves the presentation to the assessors of five (5) sample products,
of which two (2) are from one of two lots and three from the other. The aim is to
Trained or expert assessors should be used. They should all have received the
allocate the samples into groups perceived as identical.
same training in regard to the product and to the test. Their performance should
The size of panel participating in the tests will be determined in the light of
be monitored (for repeatability and sensitivity) using the procedures of the
desired a values, with a minimum of 10 selected assessors, if the objective is
AFNOR standard.
to demonstrate the existence of a difference.
This type of test is applicable when there are few of available assessors and
These tests should have a balanced design to cancel Positioning: the question relating to the
the products have a non-lingering taste (low remanence). This method is
out the effects of presentation and response order. differentiation between samples will be of
statistically more robust than the triangle test, as the aprobability of a correct the following kind: “Which is the swee-
Product presentation should be simultaneous and
answer are one in ten. It is nevertheless very much affected by sensory test?”, “ Rank these products in order of
the order of testing made mandatory. Assessors are
fatigue and memory effects. It is mainly used for visual, auditory or textural increasing intensity of banana flavour”,
free to taste samples a second time if they wish. They “Rank products on the scale shown, from
testing, and less frequently for flavour testing.
are debarred from giving products equal ranking. the most granular to the smoothest”…
Triangle test for similarity

(Determination of absence of perceptible difference between two products)
(Perceptible differentiation between two products
In procedure, this test is identical to the triangle test for differences. However, it by a given characteristic or property)
aims to demonstrate the absence of perceptible difference between two
In this test assessors are asked to compare two products in the light of a given
samples.
characteristic or property. The two samples are ranked by the intensity of the
The size of panel participating in the tests (data being independent) is determined
characteristic.
in the light of the desired a, ß and Pd values, requiring a minimum of 81
The size of panel participating in the test will be determined in the light of the
selected assessors (This number could give ß value of 0.01, a value of 0.10
desired a risk values, with a minimum of 18, if it is sought to minimize risk a.
and Pd of 30 %). When trying to demonstrate similarity, the chief requirement
If the direction of differentiation is specified prior to the test, a directional binomial
is to minimize ß.
test will be performed. If not, a non-directional binomial test will be performed.
The prior information as to the direction of differentiation will be an instruction
given by the panel leader.
30 31
Analytical tests
Ranking test a] Choice of assessors

(Perceptible difference between several products according to the intensity of
A panel of at least 12 should be recruited and trained to ensure that at least
a given characteristic or property)
10 will be available for product evaluation. Initial recruitment should give
In this test, assessors are asked to simultaneously compare more than two considerable importance to assessor availability and motivation, as well as
samples of the product, and rank the samples according to the perceived inten- their pre-existing skills for tests of this nature.
sity of a characteristic.
The minimum size of panel participating in the test is 12. A Friedman or Page test
(if the order of ranking is known in advance) should be performed using the b] Generation of attributes
AFNOR standard recommendations for subsequent data processing, so as to be (the consensus method)
able to process data at a later date. The initial choice of the previously known
All assessors participating in the final assessment should, if possible, take part
ranking is recorded by the panel leader.
in the generation of terms.
Scoring test
The method involves the development of a precise definition of each attribute,
(Numerically rating the intensity of a given characteristic)
and of an operating mode and a benchmark product, based so far as possible,
In this test, assessors are asked to evaluate samples by giving a score to the on a single stimulus (the substance is the model), or arising from memories
intensity of a given characteristic on a specified scale. of existing products. The benchmark product should help initiate a learning
Experimental design The size of panel participating in the test is at least process as regards the nature of the attribute and situate it on the scale of
The presentations of products should 12. assessment.
follow complete or incomplete block The number of product samples presented should Sensory profiles
designs, taking into account the effects Sensory profile: description of sensory
not be greater than six per assessor. If there are The range of products presented at sessions should characteristics of samples of products by
of presentation order and initial sample
presentation bias:
more than six products, a balanced incomplete be sufficiently broad to be representative of the pro- assessment of the intensity of the sensory
balanced experimental design: each block (BIB) design should be used. Statistical duct space to be characterized. attributes characteristic of the products.
product is tested as a whole the same interpretation can use Student’s test, if there are Overall profile: assessment relates to the
number of times by every assessor, two products, or by a analysis of variance test fol- Several sessions are required for the generation of terms. overall sensory characteristics of the pro-
complete block design: all assessors lowed by a multiple comparison test, if there are ducts.
taste all products, Partial profile: assessment relates to a part
more than two. The panel leader, guided by statistical analysis (inclu-
incomplete block design: all assessors do of the sensory characteristics of the product.
not taste all products. ding multivariate methods) should seek to reduce the
number of attributes to 15-20. Terms that are very Possible procedures
similar or have hedonic or quantitative meanings are Technique working with a list of consen-
eliminated after discussion with the panel, by consensus-obtained attributes: descriptions
7. Descriptive tests: the sensory profile sus. based on a list of attributes generated and
(Assessment of a set of perceived characteristics for one or more products) defined by the assessors as a whole.
Technique working with a list of predefined
attributes: description based on a list
The sensory profiling permits the description and assessment of the intensity
proposed to assessors by the panel leader
and quality of a set of sensory characteristics perceived in one or more products. (the list may be proposed by the client or
This test, due its to methodological complexity, requires great rigor and scrupu- be the result of previous studies, of the
lous compliance with the requirements set out below. scientific literature, or of standards…).
Free choice profiling technique: description
is based on a personal list generated by
each assessor. This option requires a diffe-
rent type of statistical analysis from the
others.
32 33
Analytical tests
The list of terms, definitions and operating procedures should be available to d] Sensory profile procedures
the assessors throughout the training period.
These tests are performed according to a balanced experimental design, in
order to cancel out the effects of presentation order and the effects of initial
The definitions or terms may be taken from the AFNOR standard, or established
sample presentation bias. The balanced experimental design will be constructed
by the assessors and the panel leader, or arise from earlier work, including
in such a way as to give 10 independent responses per product. Control products
glossaries. If possible, a benchmark product for each attribute will be sought.
may be introduced during the assessment, if the study's purpose is to position
products against a benchmark product or to check the stability of the responses
c] Training of assessors
of a group of assessors.
A period of training is necessary, whether the list of attributes is jointly generated
or predefined. To validate group results, each assessor will evaluate the same sample of
the product at least twice, with the exception of expert assessors, whose
A list of predefined attributes may be applicable or “adjusted”. performance is checked on a continuous basis.
The training of assessors will be specific to a given product, and this will apply Each attribute will be quantified on a agreed scale.
even to a assessor who is used to profile techniques.
Efficient monitoring of the panel and analysis of the results of a sensory profile
A minimum of 10 training sessions should be set up for assessors with no expe- rely on multiple statistical techniques, including:
rience in the product. The number of sessions may be slightly reduced in the
light of the perceived performance of the group. Assessors take part in at least graphic displays of the distributions of scores for each product and for
one session a week throughout the training period. Assessors taste products each attribute,
together and use all described attributes and references generated by the analysis of variance of product effects and, if possible, of assessor
attributes. After the training period, the repeatability and discrimination effects if the distributions are unimodal (if they are not, the Friedman
capacity of the group is measured, in regard to the relevant product space. test may be used),
multivariate methods such as: principal components analysis, discri-
When a group of expert assessors participate in long-term studies, they minatory factorial analysis, multiple correspondence analysis, and
participate in all sessions and their performance is regularly monitored, to hierarchical clustering…
check the continuing homogeneity of the group and, if possible, to check
stability over time.
Performance monitoring of each assessor basis requires a study of repeatability,

combined with a study of the discrimination capacity of each assessor, both at
the time of training and during the assessment process itself.
Monitoring the performance of a panel involves the study of its global repeata-
bility and of it's ability to discriminate, as well as its homogeneity. Analysis of
variance can be adapted to achieve any of these objectives.
34 35
Analytical tests
statistical tests
Test Purpose What kind of data must be used?
Analysis Study of the ability to discriminate Table of products / assessors and

Hedonic
of variance between products and of assessor repetitions (when comparison involves
repeatability, by measurement more than two products tests
of the following effects: or more than two sessions)
- products
- assessors
- assessor / product interaction
- sessions
- product / session interaction
Principal Visualising similarities Either the table of raw results,

components between products or the table of mean value:
analysis Explaining the special features product / attribute
(PCA) of product combinations
defined after the event
Discriminatory Explaining the special features of Either the table of raw results,
factorial previously defined product com- or the table of mean value:
analysis binations product / attribute
(DFA)
Multiple Explaining the special features Either the table of raw results,
correspondence of product combinations or the table of mean value:
analysis defined after the event product / attribute
(MCA)
Hierarchical Establishing groups Table of mean score;

cluster of attributes or products assessor / product
analysis
(HCA)
36 37
He d oni c te sts
This part of the document details the guiding principles adopted when performing
consumer research. The end purpose here is measuring the acceptability of and
preferences between products in sensory terms. The measurements under-
taken as part of these studies, and the responses given by consumers in terms of
product likes and dislikes, are conventionally known as “hedonic tests“.
Actia's requirements for hedonic tests have been built up on the basis of ASTM*
standards, and on the basis of the scientific studies and technical and profes-
sional know-how of specialists working in the Actia technical centres. Actia
requirements take into account the standardisation approach used by AFNOR
for hedonic testing, working through the centres participating in the working
party preparing these standards.
Hedonic tests, when performed at the time of a product launch, make it possible to
limit factors contributing to failure in the sales phase. However, these tests are not
totally predictive of the market success of any product, insofar as they relate only a
single factor involved in consumption and purchase decisions, namely the factors
bearing on the product's sensory characteristics. In hedonic testing, assessment
results relate solely to the product universe presented to the consumers. These
tests represent assessment at a given moment in time only, under test conditions,
and are not predictive of any change in the basis of assessment. The choice of a
product space is therefore an essential component of hedonic testing.
Methodological aspects relating to assessors, products and procedures are

developed in more detail below. The various types of hedonic testing are
initially summarised for the purposes of this study, and individually detailed
subsequently, in order to take into account special features. The standards to
which the procedures make reference are listed by theme at the end of this guide.
1. General principles
For optimisation of the research process, the analyst will examine the purpose,
context and nature of the products involved.
The result of this initial reflection will be to determine:

the one or more types of tests to be adopted,
the consumers participating in the test,
the samples of products to be presented in the test phase,
the approach to be implemented (choice of products, gathering reponses,
tasting conditions, statistical tests…).
* American Standard for Testing and Materials
38 39
Hedonic tests
2. Testing procedure (what type of test for which result?) members of the client or service provider teams or product professionals (manu-
facturers or producers) to limit possible effects of prior knowledge of the products
or of sensory techniques. By convention, they are known as consumers. Only
Hedonic tests evaluate the acceptability of the product by consumers.
consumers take part in hedonic tests.
Acceptability includes all the sensory characteristics perceptible at the time of
testing. Acceptability may be defined in terms of the pleasure given by one or
a] Number of responses processed in a single study
more products to the consumers, or in terms of their preference. There is a test
for each of these two aspects. The minimum number of consumers generating data for statistical interpretation
purposes is 60. This limitation also affects studies involving several groups, whether
This guide deals with the more frequently used hedonic tests for sensory they are temporarily or geographically segmented or dispersed (study of chan-
evaluation purposes. ging tastes over time, or study of evolving responses from different regions). In
these cases, there will be 60 respondents, either by time period or by region.
Determining a preference between products presented simultaneously. Also, for a study whose target group extends beyond a single given region, tests
Preference test by paired comparison should be performed in at least two regions, in order to ensure a better repre-
Typical questions: “Which do you prefer?”, sentativeness of the group of consumers.
“Which is the most acceptable?“, “Which do you like most?”
b] Consumer selection
Determining a preference between products and ranking them without The criteria for consumer selection depend on study objectives. These criteria
making any assessment on a scale of sensory differentiation. should include at least the consumption habits for the type of product under
study. In the absence of any clearly defined target, the sampling of consumers
Preference test by ranking
should meet criteria such as balanced gender, age and social professional
Typical question: “Rank these products by preference.” representativeness.
Determining the degree of “pleasure” given by a product or evaluating c] Management of consumer groups
the relative scale of preferences among products.
When setting up a panel for survey purposes, the selection criteria are variables,
Hedonic scoring tests such as age, gender, social and professional category, number of persons in the
Typical question: “Assess the degree of pleasure given to you by the household and availability.
one or more products on the proposed scale.” Significant data on the eating habits of consumers, such as the frequency of
consumption of products, may be further useful items of information. Consumers
should not be asked to respond to such surveys more than 24 times in any one
year.
3. Consumers
In the case of occasional recruitment for research Consumer familiarization with products
The makeup of a consumer panel will to a large extent determine the validity of purposes, consumer characteristics are recorded as Familiarization with the products will
they would be for standard recruitment, using an influence hedonic judgement.
test results. It should be made up in such a way as to be as representative as
identification form which is independent of the It is therefore necessary to monitor and
possible of a given consumer target.
control the number of tests undertaken by
The panel should be formed from untrained assessors who do not meet any product questionnaire presented to the consumers.
each assessor for a given family of products.
particular criterion in regard to sensory evaluation, and who only occasionally The consumer group needs to be gradually
participate in hedonic tests. They are not regular participants in discriminative and regularly refreshed.
tests, and never take part in positioning or descriptive tests. They should not be
40 41
Hedonic tests
4. Products d] Presentation of products
Product samples are made anonymous in the eyes of the consumers by the use
a] Choice of products of codes. In the laboratory, samples are presented according to an organised
experimental design, and under the same conditions (temperature, quantity,
Criteria of product choice are determined as a function of study objectives.
receptacle…). A method of product tasting which is consistent with the standard
Special attention should be given to:
consumption behaviour of consumers is preferred.
the representativeness of the product sample, taking into account the
greater or lesser homogeneity or heterogeneous nature of the product
or batches of product. The degree of validity of results will depend on 5. Methods
how representative the products are in regard to the product space
researched, and the representativeness of the samples in regard to the
Any change in the conditions of testing may be a source of substantial bias
product itself. The sampling plan should take into consideration these
effects, which then influence the interpretation of results. Tests should therefore
elements, so that variability between products, or between samples
take place under constant conditions. The requirements pertaining to conditions
in the course of repetitive research, can be controlled,
for product assessment aim at keeping bias and disturbance to a minimum, as
the origin of product samples (point of manufacture or point of sale),
these can affect consumers during the product assessment phase.
the number of products presented at each session and to each consumer,
special limitations at customer request.
Instructions drafted by the research manager relating to the test, to the ques-
tionnaire and to the principles or protocols of tasting should be given to the
b] Product identification
consumers by the panel leader. Instructions sent to several groups of consumers
Product identification is required, to ensure the traceability of the product and involved in a single study should be homogeneous, whether these are predrafted
of the information relating to it. This will involve: instructions or further information in response to a question raised during testing.
information such as the trading name, the brand name and / or manu- This information should then be shared out among all groups by the panel leader,
facturer, batch number, DLC* or DLUO** , packaging code, date of sampling whose responsibility it is to convey this information.
or receipt by the service provider; all this should be systematically
recorded, When the procedures are not those customarily employed by the laboratory, a
other information which may be important for certain types of product, preliminary test is recommended. This may relate only to a new stage in the
such as the weight, type and shape of packaging, list of ingredients, procedure (larger number of products, new questionnaire styles…).
mode of preparation used for product should be recorded so far as
possible. a] Gathering responses
Questionnaires are used to collect the assessments of the consumer group.

c] Product preparation
Special attention should be devoted to the questionnaire in order to ensure that
At the time of testing, the manner in which products are most frequently the consumers express themselves clearly and avoid errors (errors of interpre-
prepared by consumers should be adopted, if possible. When research is tation in the use of scales or definition of terms, for example). The questionnaire
performed in the home, the preparation procedures should be recorded in should be in such a form as to enable traceability in terms of assessor / pro-
the written instructions, which should include the preparation methods duct/
consistent with habitual practice in the household. time / date / questionnaire / testing, and do so irrespectively of the medium
(paper, visual display unit, assessment scale…).

42 43
Hedonic tests
The use of computer equipment, if it appears to be over complex, may require Variations in regard to experimental conditions from one session to the next
some consumer learning time. If these resources are used, the recommendation (in assessors, products or context) should be recorded. The maximum number
is to provide proper information and training to consumers on methods of of samples presented to each consumer per session is Experimental design
saving and recording their responses, before the tests start. difficult to fix. Any limitation will depend upon the Product presentations follow complete or
type of test, the type of product, the remanence or incomplete balanced block designs:
A consumer identification form, which is independent of the questionnaire used otherwise of their taste and the place of study. balanced experimental design: each pro-
for assessment purposes, should be developed by the research manager. 12 product tests in one hour (and 6 in simultaneous duct is tasted by an identical number of
presentation) is a useful guide for a laboratory test consumers,
Questions and, for a home-base study, 2 products simultaneously complete block design: all consumers
The extent to which the questions and the text are easy to understand should distributed. Tasting sessions, when organized with taste all products,
be checked with the consumers. If there is any lack of clarity, as for example consumers without prior appointment, should last incomplete block design: all consumers do
with children, the responses should be written down by the panel leader. less than a quarter of an hour. not taste all products.
Questions intended to assess consumer pleasure or preference should be raised
first, and drafted in a general manner, so that the questions cannot be dissociated
from the assessment exercise. There are a number of ways this can be done. 6. Special features of the various hedonic tests
Additional questions relating to the product should (when they arise) always
be raised in the form of an extra question, and not be included in the statistical a] Preference test by paired comparison
processing. Additional tests, if any, require a new presentation of the products,
Definition
and another independent questionnaire. Additional questions should not be
Method by which products are simultaneously presented in pairs so that the
interposed into the existing questionnaire.
consumers can say which of the two they prefer.
Response scales
Field of application
Response scales, whether numeric, semantic or motor, may be used. A conti-
Used when differences between products are small. Suitable when consumers
nuous scale provides for an item at each point of measurement. If the scale is
(such as young children) may have difficulties in understanding more complex
discontinuous, it should include at least 5 steps.
protocols. When the question is simple, the response is more spontaneous.
In the laboratory, no more than 6 pairs of product should be presented to each
b] Conditions for tasting and assessment
assessor. At home, a single pair only should be distributed at any one time.
The fundamental precautions to be taken are:
tests should take place at home or in a standard laboratory, The technique of forced choice increases the statistical power of this test. If
in the laboratory, consumers evaluate the product without communi- tests are performed with small numbers (60 to 100 consumers), this approach
cating one with another, is preferable. For tests using larger numbers (more than a 100 consumers), the
tastings take place according to a protocol which provides the best “no preference” response may be accepted, in order to glean information on the
simulation of the context and standard times of consumption of the reasons for or extent of failure to reply.
product in question,
consumers should respect minimum wait times between product tasting, Simple paired comparison test
depending on the type of product. At their disposal, there should be a (Comparison of preferences between two products presented in a definite order)
food capable of overcoming the lingering taste (remanence) of the
Experimental design
sensory characteristics of the product previously tested (water is
Complete balanced block design is used.
most frequently used, while bread or dried toast may be preferred to
counteract more lingering tastes).
44 45
Hedonic tests
Statistical treatment Responses which classify products as of equal ranking are disallowed (forced
When using the forced choice technique, a bilateral binomial test is performed choice protocol).
to determine whether a given sample of respondent consumers has an overall
preference for one or other of the products researched. The test results show The maximum number of products to be presented simultaneously to each
frequencies, and the associated confidence interval. When failure to reply is consumer is 6. It is recommended to present 3 to 5 products per session.
acceptable, the frequency of any such failures to reply is calculated.
Statistical processing
Multiple paired comparison test The statistical processing used is the Friedman test.
(Comparison of preferences for more than one samples presented successively, Multidimensional processing techniques may also be implemented to study
two by two) individual preferences. Consumers are thus categorized into the groups of similar
preferences.
Experimental design
A complete balanced block design is used, while ensuring the best possible
c] Hedonic scoring tests
control over the order of presentation of the pairs.
Definition
Statistical processing Method of assessment of the pleasure given by a product using a response
When using multiple pair tests, statistical models are used to determine whether scale. This method evaluates preferences.
the sample of respondent consumers indicates a preference for one of the pro-
ducts researched, and to make a product ranking. There are segmentation tech- Field of application
niques which make it possible, within a given group, to determine sub-popula- These tests assess the acceptability of the product by consumers. Acceptability
tions with distinct preferences. For more details, the reader may refer to the includes all the perceptible sensory characteristics of the product at the time of
work of Herbert A. David. the test. Acceptability may be defined in terms of the pleasure given to the
consumers by one or more products, or in terms of consumer preference.
b] Preference test by ranking
Response scale
Definition Several scales of response can be used. It is recommended to use scales with
Method under which products are presented simultaneously for consumers to schematic smiling or frowning faces for children aged 6 to 11. The greater the
rank them by order of preference. This method is ordinal, and makes no attempt number of categories, the greater the potential discrimination of consumer res-
at estimating the scale of differences, as it assesses momentary preference, in ponses.
regard to all of the products tasted.
Experimental design
Field of application These tests should have a balanced design to cancel out effects of presentation
This type of test is more cost effective in terms of products used than the multiple and response orders.
paired comparison test. The products are presented either singly and sequentially (monadic) to evaluate
the pleasure given, or by simultaneous comparison, for ranking purposes.
Experimental design It is advisable to begin by evaluating a product which is representative of those
These tests are performed according to a balanced experimental design in that are being studied but which is not one of them in order to limit positioning
order to cancel out the effects of presentation order and the effects of initial effects. Each consumer tastes no more than one serie of products. Consumers
sample presentation bias. do not repeat the test.
Product presentation is simultaneous, and the order of tasting is set. Consumers The maximum number of product samples, not to be exceeded by any one
are free to taste the product samples repeatedly, as they wish. consumer in a single session, is six.
46 47
Hedonic tests
Statistical processing
In this type of test, it is important to identify groups of consumers who exhibit
homogeneous behaviour in terms of preferences, in order to identify any signi- Ser vice provider
ficant differences between groups.
quality assurance
A number of supplementary statistical processing techniques are also available:
distribution histogram of respondents’ ratings,
measure of dispersion (such as the standard deviation) in the ratings
for each group of consumers and / or products,
estimated average for whole of target and for each group of respondents,
for two samples, tests of the significance of the difference between
them may be parametric (Student’s) or nonparametric (Wilcoxon,
Mann-Whitney),
for more than two samples, analysis of variance or, in the case of ranking,
a Friedman test (for paired samples) or a Kruskal-Wallis test (non-
paired samples).
48 49
Service provider quality assurance
Any service provider who intends to comply with Actia standards of good practice
when performing a sensory evaluation study should set up a quality assurance
system.
This system describes measures taken to define responsibilities, by setting-up
written procedures and introducing systematic controls at all stages of the
performance of the tests, to ensure the traceability of results.
The guidelines set out below are not exhaustive, but describe the minimum
requirements of the service provider's quality assurance system.
1. General guidelines for quality management
Staff
The staff should be trained on the basis of the recommendations of the Actia guide
of good practice, and of the provisions of the quality assurance system. The staff
should implement these recommendations. Records of training should be kept.
Equipment
The equipment used should be suited to the test and to the products studied, to
ensure that conditions for preparing product samples can be reproduced.
Measurement instruments should be calibrated or checked using approved
national calibration standards (weigh scales, thermometers…).
Calibrated instruments should be correctly identified and labelled.
Calibration and checking procedures should be set down in writing, and all
methodological operations be recorded.
The instructions relating to instruments or devices should be both appropriate
and immediately available.
Instruments should be checked to demonstrate that their operation is com-
pliant to specifications. The product tasting location should meet the recommen-
dations of the Actia guide of good practice. In particular, the temperature of the
tasting room should be both controlled and recorded if necessary.
Product management
The system of product acceptance should be effective and properly documented,
to enable product identification and tracking. Incoming products should be
inspected for acceptance (product code, temperature, a proper match to
initial product description…). A record should be made, including the mode,
the date and the place where samples were taken. Products should be properly
identified and stored. The preparation, presentation and codification of samples
should be controlled and recorded in a manner compliant with the requirements of
the Actia guide of good practice.
50 51
Service provider quality assurance
2. Guidelines for follow-up of assessors

Methods and procedures
The tests should comply with the methods set out in the Actia guide of good
In analytical testing, the recruitment, training and monitoring of the assessors
practice, and staff should be given the latest update of that guide, within six
should comply with the requirements of the Actia guide of good practice.
months of publication.
Internal methods should be fully documented and validated, and be based on
Procedures for monitoring should be documented in terms of recruitment,
the recommendations of the Actia guide of good practice.
selection, training and performance measurement. The monitoring procedure
should cover the recruitment, preliminary and final selection of assessors for
Test traceability
specific tests, initiation to tests, training in the general principles and methods
Working documents (questionnaires…) should specify the date of test, the type
adopted, training more generally and recording the performance of individuals.
of test, the identity of the analyst, the references or descriptions of the test
methods, the characteristics of the one or more products, and all unprocessed
or raw data relating to the test. 3. Guidelines for follow-up of consumers
Results
For hedonic tests, recruitment should comply with the requirements of the code
Explicit procedures should be set up internally to control data validity, in terms
of good practice. Records should be kept of all data relating to consumers, such
of data entry, files, transfer and calculations. This should apply to the other
as age, social and professional category, gender, address, number of tests per-
stages where there are identified risks.
formed… These records should be regularly kept up to date, particularly for
consumers who form part of the basic survey panel.
Study report
The study report should comply with the requirements of the code of good practice.
The procedures for its circulation and archiving should be described in a specific
procedure, particularly as regards confidentiality procedures.
Quality procedures
Procedures relating to working methods as a whole should be kept up to date
and available on request. Other quality-related documents should be also be
kept up to date.
Documented procedures should be used to manage requests for explanations,
claims and complaints, and reports of defects or anomalies.
A quality plan, summarizing or restating the content of procedures, might also
be usefully drafted.
52 53
Relationship between
service provider and client
54 55
1. Studies performed in line with the recommendations of this guide

As part of a deliberate approach, procedures for the performance of the various
sensory tests recommended by this guide have as their purpose ensuring better
partnership between the client and service provider. A mutual commitment bin-
ding on each, based on trust and cooperation, may be entered into, and if so will
depend upon:
the reliability of the guideline requirements as developed and validated
by practitioners of sensory evaluation,
the professionalism and responsibility of the service provider, who should
propose and implement at least the requirements specified in the guide
taking into account customer expectations in regard to the definition
and performance of a tailor-made study,
transparency and traceability in regard to the design, performance and
results of the tests under the service provider contract, and in terms of
the study report.
Any service provider intending to undertake a sensory evaluation study compliant

with the requirements of the Actia guide of good practice should undertake to
comply with all of the quality requirements and techniques required for that
study, without exception or waiver.
2. Service provider contracts
The purpose of this document is to describe and formalize the commitment binding
the service provider and client. Without going into detail or the form of the
contractual clauses, which are left to the discretion of the parties, the service
provider contract may well be set out formally under the following headings:
a] Purpose of the study
To indicate the purpose of the study and specify the spirit of the relationship
which is sought after (simple service, or active collaboration).
b] Purpose of contract
Any study commissioned should meet specifications to the contract. The degree of
cooperation between the parties should be specified in the necessary detail, in the
description of how the study is to be performed. A summary description of the
contents and of the scope of services to be provided should also be clearly set out.
56 57
c] Specifications If the study is to be performed under the guidelines of Actia guide of

good practice, the report should specify all the information required
The specifications, if the study is to be performed according to the Actia guide
for the interpretation of results and the results themselves.
of good practice, should include the general rules contained in this document,

as well as the specific rules arising from the nature and the scale of the research
elements subsequent to the handover of the study, for example, the
to be undertaken.
availability of all documents drawn up or collected by the service provider
while performing the study.
Specific elements will relate in particular to:
the conditions applicable to the study and to the tests: the laboratory
f] Financial conditions
providing the services, the service provider,
description of the resources considered necessary for proper compliance This heading will cover, in particular:
by the service provider with the requirements of the Actia guide of the price, and possible adjustments to it,
good practice, the payment schedule, invoicing, and place of payment,
tests to be performed: procedures, numbers, reference documents, failure to pay and consequences,
the characteristics of the assessors or consumers involved, their numbers procedures for additional work.
and method of selection, and the possibility of setting up a panel,
quantity of products used, specifying the way they are to be used: g] Events affecting contract performance
number of products per assessor, number of repetitions per assessor.
The contract may be affected by:
failure by one or other party to fulfil its obligations (non-performance,
d] Conditions for performance of service
poor performance, delay) entailing the suspension or cancellation of
This heading should include: the contract,
details for the performance of the research, namely the timetable and events independent of the will of a party, or change of financial or
possible extensions, taking into account the time for training of asses- other terms under which the services were to be provided.
sors, and, if necessary, the recruitment of assessors, depending on the
size of the panel, and the delivery or sampling of the products. Various h] General provisions
phases of the undertaking may be distinguished, and in this case,
Three fundamental points should be covered:
procedures for moving from one phase to another should be planned,
responsibility (damage caused, insurance…),
procedures for coordination between the service provider and the
ownership of the outcome of results,
service beneficiary,
confidentiality. In addition to the general clauses affecting confidentiality,
conditions required to adjust or change the services provided, as
the use (if any) which the service provider might make of the information
necessary,
contained in the study should be specified (for example, ability to draw the
guarantees offered by the service provider in the performance of the work.
lessons from the study as part of the service provider's own business).
e] Completion of study and follow-up
i] Other
This heading should include:
A number of points may be legal requirements, in regard to modifications of the
deadline for handover of study (mandatory or otherwise), stating the
contract, and to notifications sent by the parties one to another.
form in which the study should be handed over, its contents, and the
conditions under which it will be presented and further commented.
58 59
j] Disputes b] Objectives of study and of the test
The manner in which disputes between the parties are handled or settled These should include:
should be specified, and if there is no arbitration clause, the court whose framework and context of the study,
jurisdiction is accepted. questions to be resolved,
in the case of tests, the target population and characteristics.
NB: if it is intended to make the contract in the form of a simple written commer-
cial proposition, to be returned signed after acceptance, this document should c] Operational conditions
refer to and define the major items mentioned above.
Operational conditions should take into account to the relevant chapters in
the Actia guide of good practice, so that its requirements are complied with.
3. Study report The study report should contain, in particular:
description of the group of persons, specifying:
To ensure that a rigorous and clear response is given to customer requests, the . the number of respondents whose responses have been used,
study report should follow the main requirements of standard NF EN 45001 . specifications and characteristics of persons involved,
(1989) and the COFRAC program No. 133 (1994). It should in particular contain . for hedonic studies: conditions and mode of recruitment, type
the following items, which are part of the guideline requirements: of consumers and sort keys,
. for analytical tests: current performance of the group of assessors,
a] Heading level of qualification,
. for descriptive tests: the training of assessors involved in the
Information required to identify study: generation of attributes, if they are not part of the consumer panel,
title of study,
date of report publication, the products, specifying:
date of the one or more tests, . criteria for choice of products or if not, customer requirements .
name and address of service provider and the place where tests were in regard to sampling methods,
performed, if different from service provider’s address, . sampling mode,
name and address of client, . non-ambiguous identification and description of products,
an exclusive indicator for the report and for each of its pages, and the and their date of receipt,
total number of pages, . product DLC* or DLUO**,
name, title and signature, or equivalent identification, of the person or . storage conditions,
persons with status to authorise report circulation, . mode of preparation,
the fact that the study has been performed according to the Actia . the fact that product anonymousness is technically impossible,
guide of good practice.

60 61
f] Summary
Remanence description of protocol, specifying:
. . . . . . . . the one or more tests used and their protocol, Optionally, a summary of the results and of the major conclusions may be drafted
. number of experimental approaches, and presented on the first page of the study report.
. for descriptive tests: list of attributes For particularly long report, the summary should provide an overview of the
with definition and their generation protocol, research exercise. The summary may also be useful for those who are not
. practical conditions for performance of test: specialists in sensory evaluation (senior management, marketing managers…).
- time of tasting,
- application or otherwise of forced choice technique, g] Appendices
- experimental design,
- method of sample presentation, Appendices may include the detailed information required for an analysis of
- number of products per assessor and per session, results: detail of test results, response forms, data tables, graphic representations…
- number of repetitions per assessor,
H] Comment
- sequence of questions.
A copy of the study report should be kept for five years by the service
d] Results and interpretation provider.
Results of tests and their interpretation should be the subject of a clear, exact
If the study includes several experimental approaches for each test,
and exhaustive presentation, in line with the instructions which are part of the
each stage should be covered by a specific intermediate report, the
requirements of Actia guide of good practice. In addition to test results, the
relevant information being restated in the final report.
study reports should include:
the assumptions and data required for analysis of results,
Material corrections to a report already made available should be issued
the procedures for access to unprocessed data,
only in the form of another document or of a data transcription clearly
clear identification of results from subcontracted service providers,
identified as “Supplement to study report, serial no., date of publication”.
statistical tests used, with reminder of main objectives,
If not, then in the form of any other information or formulation of a
statistical tools used (software),
type equivalent to “This document cancels and replaces study report,
nature of data to which statistical tests have been applied,
serial n°., date of publication”.
statistical interpretation of results,
explanatory paragraph added in the event of any unsatisfied requirement.
e] Conclusions
This chapter of any report should present the conclusions of the study,
taking into account the pre-established objectives. It should contain neither
recommendation, nor advice arising from results. The results are set out in
a separate report, which could be entitled “Supplementary study report,
serial n°., publication date”.
62 63
Summary
tables
Essais utilisés
NF ISO 8586-1 (juin Nombre de personnes Recrutement
1993) recrutées : externe :
NF ISO 8586-2 2 à 3 fois plus que . voie de presse
(septembre 1994) le nombre de sujets . instituts de
nécessaires au final sondage,
Sélection . fichier
préliminaire selon Recrutement interne : consommateurs
les critères : . personnel de de l’entreprise
. motivation, l’entreprise . relations
disponibilité . implication de la personnelles
. comportement Direction Générale et . évaluer la
vis-à-vis des des responsables capacité du sujet
produits hiérarchiques à décrire des
alimentaires Buts de la sélection : sensations
(répulsion) . détecter les Essais utilisés :
. aptitudes incapacités : vision . essais
personnelles : des couleurs , d’appariement
concentration, agueusies et stimulus
communication anosmies . discrimination
. santé entre 2 niveaux
64 65
Sum m a r y ta b l e s
Objectives of sensory evaluation

1 Question Type of test Type of response
Analytical tests
Discriminative Are there differences? Triangle Products

tests Follow-up of production Duo-trio are different:
- difference Storage time 2 out of 5 yes no
- similarity Changes in:
- raw material, Products
- technology, are identical:
- suppliers… yes no
Grading Intensity Paired comparison

tests of characteristic? Ranking
(single Raw material content Scoring
characteristic) Concentration
Z D G K
of flavour Sweet taste
Appearance of
off-flavour
te
tas
y
uc
er
vo
l
Aft
u
f
Sa
u
t
o
e
Descriptive tests Sensory attributes Sensory profile
v
e
a
w
l
S
F
te
(set of of product? (partial or total)

tas
ing
ger
Lin t
Tar
characteristics) Formulate, develop,

position, typify
Fla Bit
vou
rle ter
ss
Track product r
F
M
u
u
it
s
development
y
t
Fre
Dai
in
f
l
e
a
s
sh
v
s
ry
o
u
r
wa
pro
lnu
duc
Hedonic tests
ts
Measurement Which is Paired comparison 100
80
of pleasure the preferred product? Ranking 60
Acceptability Scoring
40
20
of new product 0
0 20 40 60 80 100 120 140 160 180 200
Relationship between analytical and hedonic measurement
Explain Understanding success Sensory profile and
-100
100
preferences or rejection of product preference assessment
-80
-60
-40
-20
20
40
60
80
0
-80
Sweet
in light on same products,
-60 -40
by consumers
of sensory with appropriate
-20
0
magnitudes stastistical techniques
20
40
Bitter
60
80
66 67
Summary tables - Analytical tests Summary tables - Analytical tests
Assessors Selection, training and follow-up of panel

2 Type of assessor AFNOR-ISO definition Actia definition 3 Actia requirements
Standards NF ISO 8586-1 (June 1993)
Trained Person who has already Assessor intended to perform NF ISO 8586-2 (September 1994)
assessor taken part in a sensory test discriminative tests
1. Selection
This person has undergone
selection tests according Recruitment Number of persons recruited:
to the following criteria: 2 or 3 times more than number of assessors finally required
Personal (motivation, Internal recruitment:

availability, concentration, . staff from company
behaviour towards food . involvement of senior and middle management
and health)
External recruitment:
Sensory (ability . press advertising
to detect differences . market survey organisations
in taste and smell) . company consumer files
. personal relations
and has been introduced
to discriminative tests Preliminary selection according to criteria:
. motivation, availability
. behaviour towards food products (repulsion)
Selected Assessor chosen Assessor intended to perform . personal aptitude: ability to concentrate or communicate
assessor for his / her capacity grading tests and / or . health
to perform sensory tests sensory profiles (good long-
term sensory memory)
Selection Aims of selection:

Expert Selected assessor with Hence: . detect inability: colour perception,
assessor excellent acuteness deficiences in organs of taste or smell
of the senses, who is Training of trained . assessment of sensory acuteness
trained in the use of assessors in quantification . assessment of assessor’s ability to describe sensations
methods of sensory techniques and in stimulus
evaluation and who has recognition for a determined Tests used:
the ability reliably product universe . pair testing
to perform sensory . stimulus detection
evaluation on a variety Assessment and follow-up . discrimination between 2 levels of stimulus intensity
of products of performance . test of description of smell, texture and taste
of each assessor
and of the group Presentation of tests used in sensory evaluation
➢ Selection of group of “trained” assessors
68 69
Selection, training and follow-up of panel (continued) Discriminative tests

4 2. Training of homogeneous group of assessors 5 Triangle test for differences - Actia requirements
Assessors Number: from 10 to 20 “trained assessors” Standard NF V09-013 (July 1983)
to give a group of 10-12 trained assessors AFNOR revision pending
Training Number: 10 minimum, close in time Assessors Number: 18 minimum

sessions Definition: trained at least
Implementation of different tests

on specific products:
detection and recognition of smells and tastes: Products Number: 2 (A and B)
. benchmark products Set of 3 samples of which 2 identical

. complex samples (drink, food) Number of sets: 6

quantification training in response to stimulus Coding: . 2 different codes for A
training in product description . 2 different codes for B
specific to the relevant product space
Technique Balanced experimental design

Compulsory stage Forced choice
Simultaneous presentation:
Assessment Assessment of performance of each assessor: . 1st tasting: imposed order
of assessor (intra-individual variance) . possible repeat testing
performance . repeatability of quantitative responses Number of set per assessor: 1 (no repetition)
. discrimination capacity
➢ profiles: 6 products x 3 assessments x 10 assessors
➢ analysis of variance of individual data Questionnaire “Examine the 3 samples from left to right.
Write down the number of the one you think is different.”
Assessment of group performance:
(repeatability, discrimination, consistency within group):
➢ analysis of variance using mean scores Analysis NF V09-013 (July 1983):
➢ PCA for profiles of results . binomial law, p = 1/3
➢ product / assessor interaction . unilateral test
. taking into account type-1 error : a
NF V09-013 revision pending; based on:
Panel follow-up Motivation: . type-1 error : a
. relationship of trust between panel leader and assessors . type-2 error: ß
. information on results of work . proportion of assessors: pd (distinction)
. tokens of appreciation for work undertaken
Triangle test for similarity - Actia requirements
Periodical assessment of performance:
reduction in performance ➢ new training Assessors Number: › 80
Definition: trained at least
Keep up group numbers:
. training of new assessors (2 or 3),
gradual inclusion in group Analysis In this case, based on:
of results . type-1 error : a = 0.1
. type-2 error: ß = 0.01
. proportion of assessors: pd = 30 %
Other points similar to those of triangle test for differences
70 71
Discriminative tests Discriminative tests

6 Duo-trio test - Actia requirements 7 Two out of five test - Actia requirements
Standard NF ISO 10399 (May 1992) Standard V 09-001 (July 1983)
AFNOR revision pending
Assessors Number: 10 minimum

Assessors Number: 18 minimum Definition: trained at least
Definition: trained at least
Products Number: 2 (A and B)

Set of 5 samples, 2 of one type and 3 of the other
Set of 3 samples: benchmark (A or B) + A, B Number of sets: 20
Number of sets: 2 (constant control) or 4
Coding: . 3 different codes for A
Coding: . 2 codes for A (control, test) . 3 different codes for B
. 2 codes for B (control, test)
Technique Balanced experimental design (20 assessors)

Technique Balanced experimental design or
Draw of presentations (less than 20 assessors)
Forced choice
Forced choice
Simultaneous presentation:
. 1st testing: imposed order Simultaneous presentation:
. possible repeat testing . 1st tasting: mandatory order
. possible repeat tasting
Number of set per assessor: 1 (no repetition)
Number of set per assessor: 1 (no repetition)
Questionnaire “Examine the 3 samples from left to right:

. the first is the control Questionnaire “Look at the 5 samples from left to right.
. one of the two others is different from the control. Make up the two groups of samples by circling
Write down the number of the one you think is different.” the 3 products of the same type.”
Analysis NF ISO 10399 (May 1992): Analysis V 09-001 (July 1983):

of results . binomial law, p = 1/2 of results . binomial law, p = 1/10
. unilateral test . unilateral test
. taking into account type-1 error: a . taking into account type-1 error : a
NF ISO 10399 revision pending; based on:

. type-1 error: a
. type-2 error: ß
. proportion of assessors: pd (distinction)
72 73
Grading tests Grading tests

8 Paired comparison test - Actia requirements 9 Ranking test - Actia requirements
Standard NF V09-012 (July 1983) Standard NF ISO 8587 (May 1989)
AFNOR revision pending
Assessors Number: 12 minimum

Assessors Number: 18 minimum Definition: selected at least
Definition: selected at least
Products Number: greater than 2 (series of samples)

Defined criterion of differentiation: organoleptic property on Defined ranking criterion: organoleptic property
set(s) of 2 samples
Number of combinations: 2 Coding: . serial, different code per product
Coding: . 1 pair per assessor: 1 code coupling: 1 for A and 1 for B . several series, different serial codes
. several pairs per assessor: several code couplings

Technique Direction of difference known in advance: unilateral test
Direction of difference not known in advance: bilateral test Forced choice
Balanced experimental design
Forced choice Simultaneous presentation:
Simultaneous presentation: . 1st tasting: mandatory order
. 1st tasting: mandatory order . 2nd assessment in rising order (verification)
. possible repeat testing
Number of pairs per assessor: 1 or more (series of pairs) Number of series per assessor: one or more
Questionnaire “Look at the 2 samples from left to right. Questionnaire “Examine samples from left to right.
Of the 2 samples which is more “…” ? (sweet, savoury)” Write code numbers in increasing order of intensity
for attribute […].”
Analysis NF V09-012 (July 1983):

of results Unilateral test Analysis NF ISO 8587 (May 1989):
. binomial law, p = 1/2 of results Calculation of sums of rankings: per assessor and per sample:
. taking into account type-1 error: a . general case: Friedman test
Bilateral test . pre-determined order of samples: Page test
. binomial law, p = 1/2
. taking into account type-1 error: a
NF V09-012 (July 1983) revision pending, based on:

. type-1 error: a
. type-2 error: ß
. assessor proportion: pd (distinction)
74 75
Grading tests Descriptive tests: sensory profile

10 Scoring test - Actia requirements 11 Research and selection of attributes - Actia requirements
Standard Special case of sensory profile, reduced to single attribute Standard NF ISO 11035 (July 1995)
Assessors Number: 12 minimum Training of Target numbers: 10 minimum

Definition: selected at least assessors
and role Selection, and training of assessors
of panel leader (NF ISO 8586-1 and 8586-2):
Products Number: 2 to 6 . availability, motivation
. ability to recognise, evaluate and memorise stimuli
Defined ranking criterion: organoleptic property suited to type of product in question
. creativity and skills in verbal expression
Coding: different code per product . ability to use a scoring scale
Training on type of product concerned

Check of assessor repeatability
Simultaneous presentation
Panel leader:
. selection of assessors, maintaining assessor motivation
Questionnaire Number of attribute: 1 . taking into account of individual opinions
and harmonisation of all judgements
Nature of scale:
. structured, 5 levels minimum
or Products Range of products representative of product space
. continuous (small differences) to be described: 8 to 16 products
Number per session: 3 to 6

Analysis Single mode distribution of datas (verification)
of results
Calculation of mean, standard deviation and confidence interval Generation of Personal search for maximum number of terms:
attributes . assessment of products in booth
Product comparison . visual, tactile, olfactory or taste sensations
. 2 products: Student test (overall profile)
. more than 2 products: analysis of variance
followed by multiple comparison test Group discussion with panel leader:
. translation of elements of perception
into appropriate attribute
Preliminary sorting of attribute by panel leader:

. elimination of hedonic terms,
of intensity terms and irrelevant terms
76 77
Descriptive tests: sensory profile Descriptive tests: sensory profile

12 Research and selection of attributes (continued) - Actia requirements 13 Development of profiles - Actia requirements
Reduction First reduction in number of attributes: Standard NF ISO 13299 (June 1995)
of list . further assessment of products in booths
of terms . simple quantification of perceived intensity
. sorting by frequency and intensity Assessors Number: 10 minimum
and role Quality: selected at least
Ranking of attributes by magnitude: of panel leader
elimination of terms with low geometrical mean Panel meeting prior to session:
. presentation of test and type of products
Second reduction in number of attributes by PCA by clear and un-biaised explanations
on each group of terms (appearance, smell, taste…) . special recommandations regarding
and validation by analysis of variance: assessment protocol
. combination of synonymous or antonymous attributes . ensuring assessors and motivation
(check belonging to same continuum (reminder of group performances
for replacement by single attribute) obtained during control sessions)
. elimination of terms of low descriptive value
for product space or making small contribution
to identification of differences between products Products Number: 2 to 6 per session
Total number of attributes: around 15 Preparation of samples:

. mandatory anonymity (random three figure code)
. homogeneous presentation (quantity, recipient, temperature…)
Attributes / Precise definition associated with each attribute
reference Presentation during session:
products Association of a reference product with each attribute, random order, balanced experimental design
so far as possible
Definition of assessment procedure for each attribute Tasting Number of attributes: around 15
card
Make-up of list of terms: listed in order
Panel training Training in use of full range of scale of appearance of sensations during assessment
in use of the
reduced list Number of training sessions: 10 minimum Type of scale:
(with generation sessions) . structured, 5 levels minimum
or
Check of repeatability and of each assessor’s ability . continuous (identification of small
to discriminate for each attribute differences between products)
Timescale Record of different stages of tasting

for assessment Monadic profile (possibly comparative,
if number of attributes is low)
Repetition: periodical
. one of the samples is presented twice
during the same session
or
. the session is doubled-up
78 79
Descriptives tests: sensory profile Statistical tests

14 Dervelopment of profiles (continued) - Actia requirements Statistical tests used Conditions
Data Card index or computer system 15 Discriminative tests
entry
Triangle Binomial law p = 1/3 Independence of responses
Unilateral test (no repeated tasting)
Analysis From tables or unprocessed data:
of results
Analysis of panel performance, attribute by attribute: Duo-trio Binomial law p = 1/2 Independence of responses
. check that distribution is unimodal Unilateral test (no repeated tasting)
. mean, standard deviation, confidence interval,
analysis of variance
2 out of 5 Binomial law p = 1/10 Independence of responses
Product comparison, attribute by attribute: Unilateral test (no repeated tasting)
. 2 products: Student test
. more than 2 products: analysis of variance Grading tests
followed by multiple comparison test
Paired Binomial law p = 1/2 Balanced order
Comparative position of all products: comparison Unilateral test of presentation
discretionary multivariate analysis: if the direction of difference of products
. principal components analysis is known in advance
. discriminatory factorial analysis Bilateral test
. multiple correspondence analysis if not
. hierarchical cluster analysis
Ranking Friedman test Tests applied
in general for complete block designs.
Page test These tests are not
when there is a parameter-based and
pre-determined order require no assumptions
of samples in regard to response
Descriptive tests: distribution.
Sensory profile
Comparison Student test Paired or unpaired samples

of 2 products Continuous variable
attribute Standard distributions
by attribute
Comparison Analysis of variance Standard distributions

of several Homogeneous variance
products Multiple comparison test
attribute (sequel to analysis of variance
by attribute if significant):
- smallest significant
difference
or
- mean comparison test
(Newmann and Keuls…)
80 81
Summary tables - Analytical tests Summary tables - Hedonic tests
Statistical tests (continued) Consumers

Statistical tests used Conditions 17 Actia requirements
16 Descriptive tests:
sensory profile
Comparative
position
Multivariate analysis,
one or more
Participation of untrained assessors only,
conventionally known as “consumers”
of all products of the following:
Principal components Factorial analysis

At least 60 independant responses,
when an isolated statistical analysis is undertaken
analysis (PCA) performed on continuous
quantitative variables.
Model used is linear.
For study with national coverage, minimum survey
in two regions not gegraphically touching
Most frequent option
is reduced centred PCA. Recruitment instructions drafted by Project Manager
with targeting questionnaire.
Instructions given to the relevant personnel
and / or Analysis either
Factorial
correspondence
of qualitative variables
(contingency tables)
Selection of consumers by consumption habits for product surveyed
is minimum requirement (note: if undifferentiated target, consumers
analysis or tables requiring should be selected on basis of balanced distribution by gender,
(FCA) line and column typology age range and income and professional status)
(e.g.: products
and attributes).
In the event of a basic survey panel
and / or Generalisation of CFA

Minimum requirement to take into account (and keep updated records of )
address, age, gender, income and professional status, number of persons
Multiple to qualitative variables in household, availability and number of tests performed
correspondence representing more
analysis
(MCA)
than one modality. Maximum participation by any one consumer: 24 sessions / year
and / or Analysis
Hierarchical making it possible
cluster analysis to set up groups
(HCA) of attributes
or of products.
82 83
Summary tables - Hedonic tests Summary tables - Hedonic tests
Experimental context Questionnaires

18 Actia requirements 19 Targeting of consumer sample - Actia requirements
Premises Laboratory compliant to standard V09-105 (February 1987)
(ISO 8589) for room tests The targeting questionnaire should be drafted by the Project Manager
and made available to the recruiters
Conditions
of tasting
Defined testing times matching customary
conditions of consumption
The questionnaire should be independent
of the assessment questionnaire
Drafting of instructions on test by Project Manager,

to include information from panel leaders to consumers
All criteria for assessor selection should be included
in the consumer sampling questionnaire
Experimental design (to be defined)
Gathering consumer responses - Actia requirements

Sample coding (to be defined)
Individual assessment of products by consumers

Marking of questionnaires by means of assessor identification
(code or name) / date / time / session
Record of events in the event of change to procedures
(irrespective of kind) during session
Specification of scale of assessment used
Personnel Conditions of COFRAC programme 133 (1994) Check that questions are properly understood
Training and information to personnel Single question for overall assessment before and / or after consumption

Raising of personnel awareness to effects All other questions to be dealt with in subsidiary questionnaire,
of conditions of tasting bias and not subjected to statistical processing
Products See table 24
Results Unprocessed data (hard copy or diskette)

to be kept for at least 2 years
Study report to be kept for at least 5 years
84 85
Preference tests Preference tests

20 Preference test by paired comparison - Actia requirements 21 Preference test by ranking - Actia requirements
Standard XP V09-500 (August 2000) Standard XP V 09-500 (August 2000)
Consumers Number: 60 minimum Consumers Number: 60 minimum
Products Number: 2 (A and B) Products Number: 2 or more

Set(s) of 2 samples 6 per consumer maximum
Number of combinations: 2
Coding: . 1 pair per assessor: 1 code coupling: 1 for A and 1 for B Coding: different code per product
. several pairs per assessor: several code couplings
Technique Complete block design or incomplete balanced block design

Technique Number of pairs per assessor:
1 or more (series of pairs) Forced choice
in laboratory: 6 pairs maximum
at home: 1 pair maximum Simultaneous presentation:
. 1st tasting: mandatory order
Complete balanced block design . possible repeat testing
Simultaneous presentation in mandatory order
No repetition
Forced choice (consumer group 60 to 100) Questionnaire “Examine the samples from left to right.
Forced choice not essential (consumer group more than 100) Rank products according to your preference.”
Questionnaire “Examine the 2 samples from left to right. Analysis Friedman test
Which of the 2 samples do you prefer?” of results Supplementary mutidimensional tests
(consumer segmentation)
Analysis Simple paired comparison

of results Unilateral test
Direction of difference known in advance
. binomial law p = 1/2
Bilateral test
Direction of difference not known in advance
. binomial law p = 1/2
Multiple paired comparison

Analysis of variance (Scheffé model…)
86 87
Scoring tests Statistical tests

22 hedonic scoring test - Actia requirements 23 Tests Statistical tests used Conditions
Standard XP V09-500 (August 2000) Preference Binomial law p = 1/2 No repetition
by simple paired by consumers.
comparison Unilateral test,
if the direction of difference Equivalent presentation
Consumers Number: 60 minimum is known in advance order for the 2 products
Bilateral test, if not

Products Number: 2 or more
6 per consumer maximum
Preference Analysis of variance No repetition
Coding: different code per product by mutiple (Scheffé model…) by consumers
paired
comparison Equivalent presentation
Technique Complete balanced block design order for each product
Mandatory order in each pair
No repetition
Sequential monadic presentation to evaluate pleasure
Simultaneous presentation for ranking Preference Friedman test No repetition of series
by ranking
Simultaneous presentation
Questionnaire “Examine samples from left to right. of samples with permutation.
Assess amount of pleasure given by one
or more products using the scale provided.” Defined order of tasting for
first tasting (from left to right).
Several types of response scales
Hedonic Histogrammme No repetition of series

Analysis Several supplementary statistical processing techniques scoring of mark distribution
of results are possible Permutation of products
Mean and standard deviation
See table 23 by group of consumers
Parameter tests
Student test 2 products and

homogeneous variances
Analysis of variance More than 2 products
Non parameter tests
Wilcoxon test 2 products and

or Mann-Witney test all types of distribution
of marks or rankings
Friedmann test or More than 2 products

Kruskal-Wallis test
88 89
Summary tables - Analytical and hedonic tests Summary tables - Analytical and hedonic tests
Products Proposal and service contract

24 Actia requirements 25 Actia requirements
Systematic record of identification on packaging of all products tested: Main constituent articles
- selling name
- trade name and / or manufacturer Purpose of study
- Batch n°
- DLC or DLUO Purpose of contract
- sanitary code or packaging code
Specifications, notably:
Depending on the products, other information should be recorded .
conditions of study and testing
as far as possible: .
description of resources
- weight .
tests to be performed
- form of packaging .
assessors or consumers involved
- list of ingredients .
quantities of products used
- specified mode of preparation or use of product
- manufacturing procedure Conditions of performance of services (deadlines, coordination…)
- price
- storage conditions Completion of study and subsequent commitments (deadline, study report…)
Record of dates of sampling and acceptance on service provider's premises Financial terms (price, invoicing, settlement…)
Record of conditions of storage of laboratory samples Events liable to affect study performance
(temperature / duration / specific conditions…)
General provisions (confidentiality, ownership, liability)
Representative portions of products, as required by study objectives
Presentation of products under identical conditions and anonymously For details, the reader should refer to the corresponding chapter in the guide.
(except if study relates to the influence of non-sensory information,
as in the case of a hedonic test)
Checking and recording of temperatures of samples, NB: If it is intended to make the contract in the form of a simple written commercial
when presented to assessors. Temperatures should be homogeneous. proposition, to be returned signed after acceptance, the proposition document
should at least refer to and define the major items mentioned above.
For home tests, drafting of instructions for product preparation
90 91
Summary tables - Analytical and hedonic tests Summary tables - Analytical and hedonic tests
Study report Study report (continued)

26 Study report - Actia requirements 27 Actia requirements
Heading
Description of protocol specifying:
.
the one or more tests used and their protocol
Title of study .
the number of experiments
Date of publication of report .
for descriptive tests: list of attributes
Date of one or more tests with definition and generation protocol
Name and address of service provider, and place of performance of test .
practical conditions for performance of tests:
Name and address of client - time of tasting
An exclusive indicator for the report and for each - whether or not forced-choice technique is applied
of its pages, and the total number of pages - experimental design
Name, title and signature, or equivalent identification, - mode of sample presentation
of the person or persons with status - number of products per assessor, per session
to authorize report circulation - number of repetitions per assessor
The fact that the study has been performed - sequence of questions
according to the Actia guide of good practice
Results of tests and their interpretation

Objectives of study and test
Detailed, clear and exhaustive presentation of test results
Framework and context of study
Assumptions and data required for result analysis
Questions to be resolved
Clear identification of results obtained by subcontracted service providerss
Target population and characteristics if there are tests
Means of access to untreated data
Statistical tests used with reminder of main objectives
Operational procedures *
Nature of data to which statistical tests have been applied
Description of group of assessors Unambiguous statistical interpretation of results
.
number of respondents whose propositions have been used Explanatory paragraph in the event of unsatisfied requirements
.
details and characteristics of persons
.
for hedonic studies: conditions and mode of recruitment, Conclusions
type of consumers and sort keys
.
for analytical tests: current performance
Conclusions of the study in the light of pre-established objectives
The study report should contain neither recommendation nor advice.
of group of assessors, level of qualification
Recommendations and advice should be set down in a separate report
. for descriptive tests: training of assessors participating

in the generation of attributes, if not part of consumer panel
Products
Summary (optional)
.
criteria for choice of products or if not,
Appendices
customer requirements for sampling methods
Details of information, detail of test protocol, response forms,
.
sampling mode
graphical representations…
.
non-ambiguous identification and description of products
.
product DLC or DLUO, date of receipt
NB:
.
storage conditions
- a copy of the study report should be kept for five years by the service provider,
.
mode of preparation
- if the study includes several experimental approaches for each test,
.
the fact that product anonymousness is technically impossible
the stages should be the subject of a specific intermediate report, for restatement
in the final report,
- material correction to a study report which has already been delivered requires
* The reader should refer to the corresponding chapters
a separate, further document.
of the Actia guide of good practice for detailed requirements.
92 93
Appendices
94 95
Appendices - Relationship between analytical and hedonic measurements
Relationship between analytical and hedonic measurements
Two approaches to sensory evaluation, namely the analytical and hedonic

approaches, have been covered in this guide. The need to make a proper dis-
tinction between the two has been emphasized. However, one of the major
expectations of professionals working in the food industry is to find ways in
which to assess the hedonic response, and to determine the sensory charac-
teristics which underlie product acceptability. This brings us to the methods
which enable a relationship to be established between sensory and hedonic
responses, when using a single set of products.
This field is currently the subject of a considerable amount of exploratory work,
and deserves full scale examination in its own right. This, however, is not the
intention of the present document, the first to be published in this field by Actia.
But the question deserves to be raised, as work along these lines is already
being unertaken in the Actia centres.
Among the most frequently practiced methods are:
PCA (principal components analysis),

Preference mapping,
MFA (multi-factorial analysis).
Principal components analysis
This is the simplest of the methods available, a symmetrical approach in which
consumer responses are taken into account to build up a space of representation.
PCA is performed on two sets of combined data (data table: lines = products;
columns = descriptive and hedonic variables).
Preference mapping
This technique represents consumers and products simultaneously, in a single
space, specifying the individual preferences of this group of consumers. There
are two types of mapping:
Internal mapping: established using hedonic data alone. This determines

which products are preferred by which types of consumers. It relies
on PCA, in which the individuals (lines in a table) are the products tested,
and the variables (columns in a table) are the consumers.
96 97
Appendices - Relationship between analytical and hedonic measurements Appendices - French and international standards
External mapping: this involves the display on a sensorial map estab- French and international standards
lished by a study of the reactions of a group of trained assessors, built
up by overlays, indicating the point of acceptance or direction of the
The technical requirements of the Actia guide of good practice make reference
ideal preference for each consumer. The projection points of consumers
to the following standards:
are the result of a polynomial regressional model using the principal
components and describing the products in terms of descriptive sensory
Contrôle de la qualité des produits alimentaires - Analyse sensorielle, 5e édition,
measurements. Four models may be used:
AFNOR, 1995, 420 pages. (Quality control of food products – sensory analysis)
- vectorial,
- circular,
- elliptic,
General standards
- quadratic.
General directives
The method has the advantage of properly translating the diversity of
consumer expectations. This may be an interesting tool when it comes V 0 9- 0 0 1 (juillet 1983) Analyse sensorielle - Méthodologie
to the choice of a product formulation. However, to date it has a number Directives générales
of limits, related both to the experimental design adopted and the (July 1983) Sensory analysis - Methodology - General directives
analysis of the data. However, recent work seeks to overcome these
difficulties. Vocabulary
N F I S O 54 92 (mai 1992) Analyse sensorielle - Vocabulaire
MFA
(May 1992) Sensory analysis - Vocabulary
Multi-factorial analysis may be used instead of preference mapping, as it
enables an association of groups of variables of different types to a single set
of individuals. These variables may be the descriptive data produced by a Premises
trained group, but also instrumental data measured from products, combined V 0 9- 10 5 (février 1987) (ISO 8589) Analyse sensorielle
with consumer preferences, or other types of data. Directives générales pour l’implantation de locaux destinés
MFA introduces into PCA analysis a number of sets of variables (non-reduced à l’analyse sensorielle
descriptive variables, reduced instrumental variables) as well as hedonic data. (February 1987) (ISO 8589) Sensory analysis - General directives for the
A typology of taste may at a later stage be established using the usual methods installation of premises for sensory analysis
of classification, based on hedonic data alone.
Personnel
For all these methods, the space produced, when analysed in terms of preferences,
is strictly limited to the products chosen for study purposes. I S O 1330 0 - 1
(en projet) Analyse sensorielle - Guide général à l’attention
Du personnel des laboratoires d’analyse sensorielle
Partie 1 : organisation et responsabilités du personnel
(draft) Sensory analysis - General guide for the attention personnel
working in sensory analysis. Part 1: Organization and responsibilities
of the personnel
98 99
Appendices - French and international standards
Analytical tests Magnitude estimation
Qualification and training of assessors NF ISO 11056

(mars 2000) Analyse sensorielle - Méthodologie - Méthode
NF ISO 8586-1
(juin 1993) Analyse sensorielle - Guide général pour la sélection,
d’estimation de la grandeur
(March 2000) Sensory anal.-Methodology-Method of magnitude estimation
l’entraînement et le contrôle des sujets
Partie 1 : sujets qualifiés Scale of response
(June 1993) Sensory analysis - General guide for the selection training .
and control of assessors. Part 1: selected assessors
N F V 0 9- 0 1 5
(mai 1985) Analyse sensorielle - Méthodologie - Classification
des produits alimentaires - Méthodes utilisant des échelles
NF ISO 8586-2
(septembre 1994) Analyse sensorielle - Guide général pour et des catégories
(May 1985) Sensory analysis - Methodology - Classification of food
la sélection, l’entraînement et le contrôle des sujets
products - Methods using scales and categories
Partie 2 : experts
(September 1994) Sensory analysis - General guide for the selection
I S O 4 12 1
(décembre 1987) Analyse sensorielle - Méthodologie
training and control of assessors. Part 2: experts
Évaluation des produits alimentaires par des méthodes
N F V 0 9- 002 (octobre 1989) Analyse sensorielle - Méthodologie - Méthode

utilisant des échelles
(December 1987) Sensory analysis - Methodology - Evaluation of food
d’éveil à la sensibilité gustative
products by methods using scales
(October 1989) Sensory analysis - Methodology - Methods of awakening

of test sensitivity
Personnel
N F V 0 9- 006
(octobre 1989) Analyse sensorielle - Méthodologie - Initiation et I S O 1330 0 - 2 (en projet) Analyse sensorielle - Guide général à l’attention
entraînement à la détection et à la reconnaissance des odeurs du personnel des laboratoires d’analyse sensorielle
(October 1989) Sensory analysis - Methodology - Initiating and training Partie 2 : recrutement et formation des animateurs de jury
in the detection and recognition of odours pour l’analyse descriptive
(draft) Sensory analysis - General guide for the attention of personnel
I SO 13 301 (en projet) Analyse sensorielle - Méthodologie - Directives of sensory analysis laboratories - Part 2: Recruitment and training of
générales pour la définition et le calcul des seuils individuels panel leaders for descriptive analysis
et de groupe à partir de trois ensembles de données obtenues
par choix forcé “A” - “non A” test
(draft) Sensory analysis - Methodology - General directives for . N F V 0 9- 0 17 (juin 1984) Anal. sensorielle - Méthodologie - Essai « A »-« non A »
the definition and calculation of individual and group thresholds based on
(June 1984) Sensory analysis - Methodology - “A” - “non A” tests
three sets of data obtained by forced choice
Triangle test
Preparation of samples
N F V 0 9- 0 13
N F V 0 9- 005 (septembre 1982) (ISO 5497) Analyse sensorielle
(juillet 1983) (ISO 4120) Analyse sensorielle - Méthodologie
Essai triangulaire
Méthodologie - Directives pour la préparation d’échantillons
(July 1983) (ISO 4120) Sensory analysis - Methodology - Triangle test
pour lesquels l’analyse sensorielle directe n’est pas possible
(September 1982) (ISO 5497) Sensory analysis - Methodology - Directives
Duo-trio test
for the preparation of samples where direct sensory analysis is not possible
NF ISO 10399 (mai 1992) Analyse sensorielle - Méthodologie - Essai duo-trio
(May 1992) Sensory analysis - Methodology - Duo-trio test
100 101
Appendices - French and international standards
2 out of 5 test Hedonic tests

V09-001 (juillet 1983) Analyse sensorielle - Méthodologie General
Directives générales
(July 1983) Sensory analysis - Methodology - General directives
XP V 0 9- 50 0 (août 2000) Analyse sensorielle - Méthodologie
Directives générales pour la réalisation d’épreuves hédo-
Paired comparison test niques en laboratoire d’évaluation sensorielle ou en salle
N F V 0 9- 012 (juillet 1983) (ISO 5495) Analyse sensorielle - Méthodologie

en conditions contrôlées impliquant des consommateurs
(August 2000) Sensory analysis - Methodology - General directives for
Essai de comparaison par paires
the development of hedonic tests in sensory evaluation laboratories or
(July 1983) (ISO 5495) Sensory analysis - Methodology
in test halls involving consumers

XP V 0 9- 50 1 (août 2000) Analyse sensorielle - Guide général pour
Ranking test
l’évaluation sensorielle - Description, différenciation et
N F I SO 8 587 (mai 1989) Analyse sensorielle - Méthodologie mesure hédonique
Essai de classements par rangs (August 2000) Sensory analysis - General guide for sensory evaluation
(May 1989) Sensory analysis - Methodology - Ranking test Description, differentiation and hedonic measurement
Sensory profile
N F I SO 1 10 35 (juillet 1995) Analyse sensorielle - Méthodologie - Recherche Standards for specific product or class of products
et sélection de descripteurs pour l’élaboration d’un profil
sensoriel, par approche multidimensionnelle Apples

(July 1995) Sensory analysis - Methodology - Research and selection of
attributes by profile development using multidimensional approach
NF V20-201 (décembre 1981) Méthode de présomption de la qualité
gustative des pommes « Golden Delicious » (indice de qualité)
N F V 0 9- 01 6 (December 1981) Methods of presumption of taste quality of “Golden
(juillet 1983) Analyse sensorielle - Méthodologie - Méthodes
Delicious” apples (quality index)
d’établissement du profil de la flaveur
(July 1983) Sensory analysis - Methodology - Methods of flavour profile
establishment
Cayenne pepper
N F I S O 3513
N F I SO 1 10 36 (février 1995) Analyse sensorielle - Méthodologie - Profil de
(août 1995) Épices et aromates - Piment enragé (piment dit
de « Cayenne ») - Détermination de l’indice Scoville
la texture (August 1995) Spices and aromatics - Hot pepper (so-called “Cayenne
(February 1995) Sensory analysis - Methodology - Texture profile
pepper”) - Determination of Scoville index
N F I SO 1 10 37 (novembre 1999) Analyse sensorielle - Directives générales Foie gras

pour l’évaluation de la couleur des produits alimentaires
(November 1999) Sensory analysis - General directives for the colour XP V 63- 10 0 (mai 1995) Foie gras et préparation à base de foie gras . .
of food products Recommandation pour la préparation et la réalisation d’essais
par les consommateurs - Essais par évaluation sensorielle
NF ISO 13299 (juin 1995) Analyse sensorielle - Méthodologie (May 1995) Foie gras and foie gras-based preparations
Directives générales pour l’établissement d’un profil sensoriel Recommendations for the preparation and management of consumer
(June 1995) Sensory analysis - Methodology tests - Tests by sensory evaluation
Directives for the establishment of sensory profile
102 103
Appendices - Glossary
Green coffee Water

N F V 0 3- 306 (juillet 1977) Café vert - Examens olfactif et visuel NF EN ISO 7887 (janvier 1995) Qualité de l’eau - Examen et détermination
(July 1977) Green coffee - Olfactive and visual inspection de la couleur
(January 1995) Water quality - Examination and determination of colour
N F I SO 6 6 6 8 (septembre 1991) Café vert - Préparation d’un échantillon
en vue de l’analyse sensorielle N F E N 1 62 2 (décembre 1997) Analyse de l’eau - Détermination des seuils
(September 1991) Green coffee - Preparation of sample with a view d’odeur (TON) et des seuils de flaveur (TFN)
to sensory analysis (December 1997) Water analysis - Determination of thresholds (TON) and
flavour thresholds (TFN)
Packaging
NF EN 1420-1 (en projet) Influence des matériaux sur l’eau destinée à la
X P V 0 9-009
(septembre 1995) Analyse sensorielle - Méthodes d’évaluation
consommation humaine - Matériaux organiques - Tubes,
des composés étrangers transmis par les emballages aux
raccords et leur revêtement utilisés dans les systèmes de
denrées alimentaires
distribution - Évaluation de l’odeur et de la flaveur de l’eau
(September 1995) Sensory analysis - Methods of assessment of foreign
Partie 1 : méthode d’essai
components transmitted by packaging to food
(draft) Influence of materials on water intended for human consumption
Organic materials - Tubes, connections and coatings used in distribution
Potatoes systems - Assessment of odour and flavour of water - Part 1: Test method
N F U 4 3-1 51 (avril 1981) Méthodes d’appréciation par dégustation, de
la qualité des pommes de terre provenant de cultures ayant Wine
subi des traitements pesticides N F V 0 9- 1 10 (juin 1971) Appareillage - Verre à dégustation pour vins
(April 1981) Methods of taste assessment of the quality of potatoes
(June 1971) Equipment - Wine-tasting glass
grown with pesticides
I S O 3591 (juin 1977) Analyse sensorielle - Appareillage - Verre à
Spaghetti
dégustation pour l’analyse sensorielle des vins
N F I SO 7 304 (juin 1989) Semoules de blé dur et pâtes alimentaires (June 1977) Sensory analysis - Equipment - Wine-tasting glass for
Appréciation de la qualité culinaire des spaghettis sensory analysis of wine
par analyse sensorielle
(June 1989) Hard wheat semolina and food pasta - Assessment of I S O 54 94 (juillet 1978) Analyse sensorielle - Appareillage- Verre à
culinary quality of spaghetti by sensory analysis dégustation pour l’analyse sensorielle des produits liquides
(July 1978) Sensory analysis - Equipment - Wine-tasting glass for sensory
Tea analysis of liquid products
N F V 0 3- 355 (décembre 1981) Thé - Préparation d’une infusion en vue

d’examens organoleptiques Statistical methods
(December 1981) Tea - Preparation of an infusion with a view to orga-
noleptic examination Méthodes statistiques, AFNOR, 1996, 5 vol., with:
- Vol. 1: Vocabulary and symbols, 408 p.
Tea (black) - Vol. 2: Estimates and statistical tests, 404 p.
- Vol. 3: Statistical acceptance inspection, 590 p.
N F V 0 0 -1 10 (mars 1983) Thé noir - Vocabulaire - Vol. 4: Statistical process control, 250 p.
(March 1983) Black tea - Vocabulary - Vol. 5: Processing of results, 574 p.
104 105
Glossary Assessor
The definitions in the glossary refer to standard NF ISO 5492.
Any person participating in an analytical test.
Acceptability Astm
Status of a product favourably received by a determined individual or American Standard for Testing and Materials.
population, given its organoleptic properties, at a given time and in a
determined context. Aversion
NB: Standard NF ISO 5492 defines acceptability as the status of a product favourably received
by an individual population as a function of its organoleptic properties).
Attitude of avoidance of a stimulus.
Acceptance Basic survey panel
Action taken by a given individual or determined population, indicating

that a product favourably meets expectations. Large scale sample of households or individuals making up a permanent
pool of persons who can be questioned with greater or lesser regularity.
Alliesthesia
Batch
Modification of hedonic enjoyment, due to physiological change (for exa-
mple, satiation). Subset of a population. In practice, a batch will be a subset of the popula-
tion of products sampled. From the batch studied, samples are taken of the
Analysis of variance (Anova) products set aside for study purposes.
Method enabling the analysis of data as a function of several simul- Bias

tan-eously produced effects, so as to quantify the effects and assess
their importance. In sensory evaluation, Anova makes it possible to study Systematic errors, in a positive or negative sense.
the assessors' ability to discriminate between products, and repeat perfor-
mance, and do so by measurement of various effects (products, assessors,
Blind
sessions, interaction assessor / products, interaction products / sessions).
Products are presented blind when the assessors are unaware of their
Anosmia
provenance, and in particular of their brand mark. They are identified by a
code only. Products are presented in double-blind when neither the asses-
Lack of sensitivity to olfactive stimulus. Anosmia may be total or partial, sors nor the panel leader know the identity of the products. The panel lea-
permanent or temporary. der knows only the code of the products.
Aroma
Consumer

Organoleptic property perceptible by the olfactive organs through the rear
nasal cavity at the time of tasting. In general, any person using a product or likely to use it. In sensory evaluation,
a person episodically participating in hedonic tests, who has never taken
part in discriminatory or in descriptive tests.
106 107
Consumer panel Forced choice (technique of )
Fixed sample of consumers periodically questioned. In common parlance, Technique referring to tests during which assessors must give a response
the term consumer panel may also be used as synonymous of a sample. In to the question raised.
this case there is no connotation of questions being raised successively and
put to the sampling of consumers. Free-choice profiling technique
Control Description based on a personal list generated by each assessor.
Sample of the product subjected to testing, and chosen as a benchmark HCA (Hierarchical cluster analysis)
against which other samples can be compared.
Multi-dimensional method which permit to calculate relative distance
Experimental design between items (such as products or assessors). The formation of clusters
can be showed in a tree structure (dendrogram).
Products presentation in complete or incomplete balanced block designs,
taking into account the effects of presentation order and initial sample Hedonic
presentation bias:
This term qualifies the appreciation of the consumers of a product, insofar
Balanced experimental design:
as it is pleasing or displeasing, when their sense organs are brought into
In total, each product is tasted the same amount of times by each assessor.
play, in a determined context and a given time.
Complete block design:
All assessors taste all products. Initial sample presentation bias
Incomplete block design:
All assessors do not taste all products. Bias induced in the assessment of a product by the one or more products
preceding it in taste. Prior presentation effect relates to the product
Expert assessor
immediately prior.
Qualified expert with excellent sensory acuteness, who has been trained in Kinesthesia
the use of methods of sensory evaluation and who has the ability to reliably
perform all types of test. Set of sensations arising when a force is applied to the sample by movement
(for example, test by pressure of fingers in the case of a cheese, or the
FCA (factorial correspondence analysis)
biting for an apple).
Technique based on the same principle as PCA, but suited to special types Level of significance
of data, of contingency table type (table of personnel employed), a logical
table (0 or 1), or presence-absence type behaviour. Probability of conclusion that there is a perceptible difference when there is
no difference in reality (type-1 error or risk a).
Flavour
Liminal
Complex set of taste, olfactory and trigeminal sensations perceived at the
time of tasting. Qualifies a stimulus above the threshold under consideration.
108 109
MCA (Multiple correspondence analysis) Panel leader
Method of data analysis making it possible to study a table of data in which Person whose main task is to direct the activities of groups of assessors, to
individuals (products or assessors) are set out in horizontal lines, and the recruit them, train them and supervise them. The panel leader may also
variables defining them in vertical columns, the variables being qualitative. design and direct the sensory test, as well as analyse and interpret results.
The panel leader may be assisted by one or more session technicians.
MFA (Multiple factorial analysis)
PCA (Principal components analysis)
Multi-dimensional analytical method making it possible to associate
groups of different kinds of variables with one and the same set of indivi- Method of data analysis graphically representing the maximum information
duals. These variables may be the descriptive data generated by a trained contained in a table of data, using individuals (products or assessors) set
group, but also instrumental data generated by product measurements, out along horizontal lines, and determined by quantitative variables (mea-
and the preferences of consumers or others. Multiple factorial analysis surements) in vertical columns. The “P” initial quantitative variables, which
introduces into PCA analysis several sets of variables (non-reduced are more or less correlated one to another, are transformed into “q” new
descriptive variables, reduced instrumental variables, etc) and hedonic data. variables known as principal components. Individuals are observed in
terms constructed from the most significant principal components (those
Monadic which give rise to the greatest differences between individuals). Additionally,
a sphere of correlation makes it possible to apprehend the organization of
Samples are presented monadically when they are presented individually
variables in the dual variable space.
and one by one.
Pd
Odour
A portion of a whole population of assessors able to make a distinction
Organoleptic property perceptible by the olfactory organ when inhaling
between two products.
certain volatile substances.
Perception
Organoleptic
Perception is the conscious experience of a part of the sensations to which
Adjective qualifying the property of a product as perceptible by the sense
the individual is subject over a given period of time. Perception is a function
organs.
of one or more stimuli, and of the context and sensory past of the individual.
Paired
Preference
Samples are said to be paired when they are assessed by the same asses- Expresses the state or emotional reaction of the assessor, according to
sor, and not paired when they are assessed by different assessors. which the assessor may find a product better than one or more other pro-
ducts at a given time and within a determined context.
Panel
Presentation (directed simultaneous)
A group of assessors chosen for participation in an analytical test.
Presentation at one time of all samples to be evaluated, according to a
mandatory order of tasting. A second tasting is often possible, at asses-
sor’s discretion.
110 111
Presentation (monadic) Profile (partial)
Assessment of products, single product per session. Assessment relating to a part of the sensory characteristics of products.
Presentation order effect Profile (sensory)
Bias caused in the sensory evaluation of a product at the time of testing, Description of the sensory characteristics of samples of products by
due to its position in the order of presentation (for example, it is observable assessment of the intensity of the sensory attributes characteristic of the
that the first product presented in a test is generally undervalued). products.
Presentation (sequential monadic) profile (technique of ideal)
Assessment of products, one by one, during a single session. Technique whose aim is to approximate to the optimum level of sensory
characteristics of a product, as far as the consumer is concerned (desired
Presentation (simultaneous) intensity of attributes such as “sweet, salty, spicy…”). Assessors evaluate
the intensity of a series of characteristics of the products presented and of
Presentation at one time of all samples to be evaluated.
each criterion, indicating the ideal level according to their own preference.
Product A variant is to directly position the products tasted on a scale going from

“really not enough” to “really too much”, the mean position being the ideal
Material to be evaluated or assessed. “just right”.
Product family Project Manager
Group of products of the same commercial name. Synonymous with sensory analyst.
Product space
Reference
All of the products which, for a given range of products, exhaustively include
all the sensory characteristics likely to be encountered, combined with the Substance other than the product tested, and used to define the charac-
full range of the intensities of the perceived properties of those products. ter-istics or specific level of a given characteristic.
Generally, the size of the space is directly related to the level of precision of
Reference element or benchmark
the terms describing the range (e.g. space of orange juice is more extensive
than space of 100% pure fruit juice orange juice, which is in turn more
Chosen value (or one or more characteristics of a product), in relation to
extensive than the space of 100% Florida orange juice).
which samples are assessed.
Remanence
Product universe
Synonymous with product space. Persistence or lingering of sensation, in the absence of the product.
Profile (overall) Sample
Assessment of a set of sensory characteristics of products. Unit of prepared product, presented and assessed during the test.
112 113
Sampel (consumer) Scale (single pole)
Group of consumers from a basic survey panel, or from a wider population, Scale with a single attribute only at one end or pole.
recruited for the purposes of a test, according to criteria suited to the target
population. Selected assessor
Sampling plan A person chosen for his or her ability to perform a sensory test, and whose
performance has been controlled.
Procedure set up to collect samples in the appropriate manner by respecting
certain constraints related to the representativeness of the sample in Sensation
regard to the population.
The transfer of the energy of the stimulus to a receptor sensory cell causes
Savour excitation which, if it is sufficiently great, itself causes a chain-reaction
propagating through the central nervous system. This phenomenon is known
Synonymous with taste. as “sensation”.
Scale (bipolar) Sensitivity
Scale whose descriptions at either end or pole are opposite. General term indicating the performance characteristics of a test. The
sensitivity of a test is defined rigorously, in statistical terms, by the
Scale (interval) values used for a (type-1 error), b (type-2 error) and Pd.
Scale where numbers are chosen in such a way that equal numerical intervals Sensory analysis
correspond to equal differences of sensory perception.
Term frequently confused with sensory evaluation. The technique highlights
Scale of measurement and describes the organoleptic properties of a product by the sense organs.
These properties relate both to the product itself and perception of those
Formal relationship between a characteristic and the numbers used to
properties by the assessor. The method described is the perception of the
represent the values of the characteristic.
presence, or intensity, of a number of perceived properties, or the differen-
Scale of relationship tiation of perception. Perception can never be qualified by terms such as
“good” or “bad”.
Scale where the numbers are chosen so that the equal numerical rela-
tionship match equal sensory perception relationships.
Sensory analyst
Scale of response
Person with overall scientific and technical responsibility for a test, who
Method of recording the magnitude of a characteristic to convert it into a digit. defines the test procedures and subsequent processing. This person
has responsibility for the panel leaders who work on the tests, and also
Scale (ordinal) analyses and interprets results.
Scale in which the marker points indicate a continuous or preestablished

progression.
114 115
Sensory evaluation Taste
Examination of the organoleptic properties of a product by the sense Perceptible sensation of the taste organ when stimulated by certain soluble
organs, including also analytical tests for the demonstration of the properties substances. The term also covers the properties of products causing taste
and description of products, and hedonic tests measuring the pleasing or sensations.
unpleasing properties of the products. NB: the term taste should not be used to mean taste bud, olfactory or trigeminal sensations as
a whole, for which the appropriate term is flavour. Although the term taste is generally used in
Session technician this all-encompassing manner, with this latter meaning, the word taste as such should always
be accompanied by a qualifying term such as taste of mould, taste of raspberry, taste of cork.
Person working as an assistant to the panel leader of the group of assessors,

Taste inhibition
or the sensory analyst.
Somesthesia Lack of sensitivity to taste stimulus. Taste inhibition may be total or partial,
permanent or temporary.
All sensations of mechanical or thermal origin, perceived by the skin or
Technique by consensus list of attributes obtained (profile)
mucous membranes (mechanical or thermal properties), by the muscles
(kinesthesia properties) and by the joints (proprioceptive properties).
Description based on a list of defined attributes and generated by all the
Statistical risk a assessors.
Likelihood of concluding that there is a perceptible difference, when there Technique by predefined list of attributes (profile)
is none in reality (also called “type-1 error”).
Description on the basis of a list proposed to assessors by the panel leader.
Statistical risk b
Test
Likelihood of concluding that there is a no perceptible difference, when
there is one in reality (also called “type-2 error”). Technical operation to determine one or more characteristics of a product,
service or process, according to a specified operating mode.
Stimulus
Test (analytical)
External or internal agent able to cause a reaction in an excitable system.
A test whose final purpose is to analyse the sensory characteristics of a product.
Study report
Test (descriptive)
Document comprising all test reports for a study and other ad hoc information.
Test whose purpose is the characterising and quantification of one or more
Subliminal perceived characteristics of one or more products.
Qualifies a stimulus below the threshold under consideration. Test (discriminative)
Supraliminal
Any test method involving a comparison of samples.
Qualifies a stimulus above the threshold under consideration.
116 117
Appendices - Reference works
Test (grading) Trained assessor
Test whose final purpose is either to asses a difference in perception or to rank Person who has already participated in a sensory discriminative test of
or assess products according to the perceived intensity of a characteristic. the same kind as the one in which he or she is participating. Initiation is
the minimum training required for discriminative tests.
test (hedonic)
Triad
Test to measuring the pleasing or unpleasing property of a product.
Three samples presented together to an assessor, in a triangle test, two
Test (paired comparison) samples are the same (from the same product) and the other is different
(comes from another product).
Method under which samples are presented in pairs to be compared on the
basis of a specified attribute. Untrained assessor
Test (preference) Synonym of consumer.
Test for the assessment of a preference between two or more samples.
Test (ranking)
Method in which a series of samples are classified by order of intensity

or degree for a specified property. The resulting measurement scale is
ordinal and makes no attempt to estimate the magnitude of differences.
Test (scoring)
Method of product assessment or of properties of a product using marks

mathematically significant marks.
Test report
Document presenting results of tests and other ad hoc information.
Texture
All of the rheological and structural properties (geometrical and surface) of

a food product, perceptible by the mechano-receptors, by the tactile receptors,
and possibly by the visual and auditory receptors.
Threshold
Minimum value of sensory stimulus required to cause a sensation. It is

possible for the sensation to be perceived but not identified.
118 119
Reference works English
Amerine (M.-A.), Pangborn (R.-M.) et Roessler (E.-B.),

Sensory evaluation
Principles of sensory evaluation of food, New-York, Academic Press Inc, 1965, 602 p.
French Astm,
Afnor, Guideline for the selection and training of sensory panel members, N° STP 758, 1981.
Contrôle de la qualité des produits alimentaires - Analyse sensorielle,
5e édition, Afnor, 1995, 420 p. Astm,
Recueil de normes.
Descriptive analysis testing for sensory evaluation,
ASTM Manual series: MNL 13, 1992.
Cidil,
L’enfant et les aliments, Le goût, Paris, Cidil, 1993, livre et fiches. Birch (G.-G.), Brennan (J.-G.) et Parker, (K.-J.),
Sensory properties of foods, London, Applied Sciences Publishers, 1977, 326 p.
Cidil,
Le goût: décrire et analyser arômes, textures, saveurs, European sensory network,
Paris, Cidil, coll. “Les cahiers de la qualité”, 1995. Sensory and consumer study - A multipanel study in eight european countries
- A case study on coffee,
Cofrac, Chipping Campden, UK, The Publication Officers. CCFRA, 1996, 207 p.
Analyses sensorielles, programme 133, Cofrac, November 1994, 17 p.
Jellinek (G.),
Eal, Sensory evaluation of food - Theory and practice,
Accréditation des laboratoires d’analyses sensorielles, Chichester, England, Ellis Horwood, 1985, 429 p.
1re édition, EAL-G16, mai 1995, 26 p.
Lyon (D.-H.), Francombe (F.-A.), Hasdell (T.-A.) et Lawson (K.),
Sauvageot (F.), Guidelines for sensory analysis in food product developpment and quality
L’évaluation sensorielle des denrées alimentaires - Aspects méthodologiques, control, London, Chapman & Hall, 1992.
Paris, Technique et documentation CDIUPA, 1982, 196 p.
Meilgaard (M.), Civille (G.-V.) et Carr (B.-T.),
Sauvageot (F.), Sensory evaluation techniques, vol. I et II, Boca Raton, Florida, CRC Press, 1987.
Contribution à la caractérisation d’un groupe en évaluation sensorielle de
denrées alimentaires, Université de Dijon, 1984, 341 p. Moskowitz (H.-R.),
Ph.D thesis.
Product testing and sensory evaluation of foods,
Westport, Connecticut, Food and Nutrition Press Inc, 1983, 605 p.
Ssha,
Évaluation sensorielle - Manuel méthodologique, Moskowitz (H.-R.),
2e édition, Paris, Lavoisier, Tec et Doc, 1998, 353 p. New directions for product testing and sensory analysis of foods -
Marketing and R&D approaches,
Westport, Connecticut, Food and Nutrition Press Inc, 1985, 371 p.
120 121
Moskowitz (H.-R.), sensory physiology

Applied sensory analysis of foods, vol. I et II, Boca Raton, Florida, CRC Press, 1988.
Beidler (L.-M.),
Pattee (J.-R.),
Handbook of Sensory Physiology, New-York, Springer Verlag, 1971, vol. IV:
Evaluation of quality of fruits and vegetables,
Chemical Senses: part I: Olfaction, 518 p., part II: Taste, 410 p.
Westport, Connecticut, Avi Publishing Co, 1985.
Bonnet (C.),
Piggott (J.-R.),
Manuel de psychophysique, Paris, Armand Colin, 1986.
Sensory analysis of foods,
2nd edition, New-York, Elsevier Applied Sciences Publishers, 1988, 426 p.
Buser (P.) et Imbert (M.),
Neurophysiologie fonctionnelle,
Stone (H.) et Sidel (J.-L.),
Paris, Hermann, 1975, tome I: 410 p., tome II: 465 p.
Sensory evaluation practices, New-York and London, Academic Press Inc, 1985.
Chauchard (P.),
Williams (A.-A.) et Atkin (R.-K.),
Les messages de nos sens, Paris, PUF, 1981, 127 p.
Sensory quality in foods and beverages: definition, measurement and control,
Chichester, Ellis Horwood, 1983.
Faurion (A.),
Étude des mécanismes de la chimioréception du goût sucré,
Paris, Université de Paris VI, 1982, 385 p., Ph.D Thesis.
Food preferences
Guerrier (Y.) et Uziel (A.),

Giachetti (I.),
Psychologie neurosensorielle en ORL, Paris, Masson, 1983, 201 p., Ph.D Thesis.
Plaisir et préférences alimentaires, Paris, Polytechnica, 1992.
Harper (R.),
Mac Fie (J.) et Thomson (D.-M.-H.),
Human senses in action, Churchill Livingstone, 1972.
Measurement of food preferences, Blackie Academic & Professional, 1994.
Laing (D.-G.) et Cain (W.-S.),

Solms (J.) et Hall (R.-L.),
Perception of complex smells and tastes, Academic Press Inc, 1989, 322 p.
Criteria of food acceptance, Zurich, Forster Verlag, 1981.
Laming (D.),
Thomson (M.-H.),
Sensory analysis, Academic Press Inc, 1986.
Food acceptability, Elsevier Applied Science, 1988.
Mac Leod (P.) et Sauvageot (F.),

Bases neurophysiologiques de l’évaluation sensorielle des produits alimentaires,
“Les Cahiers de l’Ensbana”, n° 5, februar 1986, 165 p.
Tiberghien (G.),
Initiation à la psychophysique, Paris, PUF, 1984, 152 p.
122 123
Statistics and experimental design Lavit (C.),

Analyse conjointe des tableaux explicatifs, Masson, 1988.
french
Lebart (L.), Morineau (A.) et Fenelon (J.-P.),
Afnor, Traitement des données statistiques - Méthodes et programmes,
Méthodes statistiques, 5 tomes, 1996. Paris, Dunod, 1982.
Recueil de normes.
Pagès (J.),
Bertier (P.) et Bourroche (J.-M.), AFMULT - Analyse factorielle multiple,
Analyse des données multidimensionnelles, Paris, PUF, 1981. Paris, Addad, 22, rue Charcot, F-75013 Paris, 1993.
Ceresta, Philippeau (G.),

Aide-mémoire pratique des techniques statistiques, Théorie des plans d’expérience, Paris, ITCF, 1989, 205 p.
2e édition, 1986, Revue de statistique appliquée, vol. XXXIV, n° spécial.
Sanders (D.-H.), Murph (A.-F.) et Eng (R.-J.),
Dagnelie (P.), Les statistiques, une approche nouvelle, Paris, Mc Graw-Hill, 1984.
Théorie et méthodes statistiques - Applications agronomiques, Belgique, Les
Presses agronomiques de Gembloux, vol. I: La statistique descriptive et les Spiegel (M.-R.),
fondements de l’inférence statistique, 378 p., vol. II: Les méthodes de l’infé- Probabilités et statistiques - Cours et problèmes,
rence statistique, 1973, 463 p. + Exerc. 186 p. Paris, Mc Graw-Hill, série Schaum, 1983.
Deroo (M.) et Dussaix (A.-M.), Spiegel (M.-R.),

Pratique et analyse des enquêtes par sondage, Paris, PUF, 1980, 302 p. Théorie et applications de la statistique, Paris, Mc Graw Hill, série Schaum, 1982.
Escoffier (B.) et Pagès (J.), Vessereau (A.),

Analyses factorielles simples et multiples. Objectifs, méthodes, et interprétation, Méthodes statistiques en biologie et en agronomie, Paris, Lavoisier, 1988, 539 p.
Paris, Dunod, 1998.
English
Fenelon (J.-P.),
Qu’est ce que l’analyse de données?, Paris, Lefonen, 1981. Box (G.-E.-P.), Hunter (W.-G.) et Hunter (J.-S.),
Statistics for experimenters, New York, Wiley, 1978.
Huon de Kermadec (F.),
Méthodes statistiques permettant d’expliquer l’appréciation hédonique par les Cochran (W.-G.),
caractéristiques sensorielles, Université de Montpellier II, 1996. Sampling techniques, 3rd edition, New York, Wiley, 1977.
Doctoral thesis.
Cochran (W.-G.) et Cox (G.-M.),
Lagarde (de) (J.), Experimental designs, 2nd edition, New York, John Wiley & Sons, 1957, 616 p.
Initiation à l’analyse de données, Paris, Dunod, 1983.
124 125
Das (M.-D.) et Giri (N.-C.), Piggott (J.-R.),

Design and analysis of experiments, 2nd edition, New York, John Wiley & Sons, Statistical procedures for the food industry,
1986, 488 p. London, Elsevier Applied Sciences Publishers, 1986.
David (H.-A.), Schiffman (S.-S.), Reynolds (M.-L.) et Young (F. W.),

The method of paired comparisons, Charles Griffin & co Ltd, 1988. Introduction to multidimensional scaling - Theory, methods and applications,
New York, Academic Press Inc, 1981.
Dey (A.),
Theory of block design, New York, John Wiley & Sons, 1986, 298 p. Smith (G.-L.),
An introduction to statistics for sensory analysis experiments,
Ducan (A.), Aberdeen, Scotland, Torry Research Station, Ministry of Agriculture,
Quality control and industrial statistics, Fisheries and Food, 1989.
4th edition, Homewood, Illinois, R.D. Irwin Inc, 1974.
Gacula (M.-C.) et Singh (J.),

Statistical methods in food and consumer research,
Orlando, Florida, Academic Press Inc, Food science and technology, 1984.
Guilford (J.),
Fundamental statistics in psychology and education,
4th edition, New-York, Mc Graw-Hill Publishing co, 1969, 509 p.
Lea (P.), Naes (T.) et Rodbotten (M.),

Analysis of variance for sensory data, New York, John Wiley & Sons, 1997.
Martens (H.) et Naes (T.),

Multivariate calibration, New York, John Wiley & Sons, 1989, 419 p.
Munoz (A.-M.), Civille (G.-V.) et Carr (B.-T.),

Sensory evaluation in quality control, New York, Van Nostrand Reinhold, 1992.
Naes (T.) et Risvik (E.),

Multivariate analysis of data in sensory science, Elsevier Amsterdam, 1996, 360 p.
O’mahony (M.),
Sensory evaluation of food - Statistical methods and procedures,
New York, Marcel Dekker Inc, 1985, 487 p.
126 127
Graphic design & layout
Anne-Lise Dermenghem
This document is set in

Meta et Goudy
Photogravure
Atelier André Michel, Paris
Printing
Imprimerie de l’Indre, Paris
Print run completed at Imprimerie
de l’Indre at Argenton-sur-Creuse, July 2001
ISBN
in progress
Registration
July 2001
128
Sensory evaluation . Actia . Guide of good practice . 2001

E-Book Sensory Evaluation

Uploaded by

Copyright:

Available Formats

E-Book Sensory Evaluation

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

E-Book Sensory Evaluation

Uploaded by

Copyright:

Available Formats

1

Building on forty years of experience, sensory evaluation is now a decision-making

Sensory evaluation has now come of age.

1. Sensory evaluation: why ? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

Service provider quality assurance Hedonic tests

Relationship between analytical and hedonic measurements . . . . . . . . . . . . . . 97

French and international standards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99

1. Sensory evaluation: why?

A degree of confusion and uncertainty continues to exist in the field of sensory

The analytical approach includes techniques making it possible specifically to

For analytical testing, no standard is currently available. It is therefore impossible

Careful initial thought is essential to determine:

To determine whether there are perceptible sensory differences between

* Date limite de consommation

d] Conditions of tasting the standards adopted for each type of test.

Triangle test for similarity

Ranking test a] Choice of assessors

Performance monitoring of each assessor basis requires a study of repeatability,

Analysis Study of the ability to discriminate Table of products / assessors and

Principal Visualising similarities Either the table of raw results,

Hierarchical Establishing groups Table of mean score;

Methodological aspects relating to assessors, products and procedures are

The result of this initial reflection will be to determine:

* American Standard for Testing and Materials

4. Products d] Presentation of products

Questionnaires are used to collect the assessments of the consumer group.

* Date limite de consommation

1. General guidelines for quality management

2. Guidelines for follow-up of assessors

1. Studies performed in line with the recommendations of this guide

Any service provider intending to undertake a sensory evaluation study compliant

2. Service provider contracts

a] Purpose of the study

c] Specifications If the study is to be performed under the guidelines of Actia guide of

j] Disputes b] Objectives of study and of the test

* Date limite de consommation

Objectives of sensory evaluation

Discriminative Are there differences? Triangle Products

Grading Intensity Paired comparison

(set of of product? (partial or total)

characteristics) Formulate, develop,

of pleasure the preferred product? Ranking 60

0 20 40 60 80 100 120 140 160 180 200

Relationship between analytical and hedonic measurement

Explain Understanding success Sensory profile and

Assessors Selection, training and follow-up of panel

Personal (motivation, Internal recruitment:

Selection Aims of selection:

➢ Selection of group of “trained” assessors

Selection, training and follow-up of panel (continued) Discriminative tests

Training Number: 10 minimum, close in time Assessors Number: 18 minimum

Technique Balanced experimental design

Discriminative tests Discriminative tests

Assessors Number: 10 minimum

Products Number: 2 (A and B)

Technique Balanced experimental design (20 assessors)

Questionnaire “Examine the 3 samples from left to right:

Analysis NF ISO 10399 (May 1992): Analysis V 09-001 (July 1983):

NF ISO 10399 revision pending; based on: