SENSORY EVALUATION METHODS

The basic questions to tackle in sensory evaluation of foods before choosing sensory
methods include:
a. Which test method is most appropriate?
b. What experimental design is most efficient?
c. What statistical treatment is most applicable?
Sensory tests can be described or classified in several ways. Statisticians classify
tests as parametric or non-parametric, based on the type of data obtained from the
test. Sensory specialists and food scientists classify tests as consumer-oriented or
affective and product-oriented or analytical, based on the purpose of the test.
There are many sensory tests and a multitude of different situations in which they
can be applied. The test employed will depend on the test objective(s). It is
imperative that the specific objective of any sensory test is probed and clarified
before testing begins.
Classification of tests by function (purpose):
A. ANALYTICAL TESTS:
Analytical tests or product-oriented tests are tests used for laboratory evaluation of
products in terms of differences or similarities, and for identification and
quantification of sensory characteristics. The two main types of analytical tests are
discriminative and descriptive tests. Both involve experienced or trained panelists.
1. Discriminative tests:
Discriminative tests are the simplest sensory tests which merely attempt to answer
whether any perceptible difference exists between or among products. There are two
types of discriminative tests; which include difference and sensitivity tests.
a. Difference tests: These tests are designed to determine if two samples can be
distinguished from each other by sensory analysis. Difference tests include the
triangle, duo-trio, simple paired comparisons, Scheffé paired comparisons test,
multiple comparisons, ranking, and scoring tests.
i. Triangle test:
This test is commonly used to determine whether they are perceptible differences
among three samples. In this test, three coded samples are presented to panelists.
Panelists are informed that two out of the three samples are identical (the same) and
one is different (odd); and are asked to identify the odd sample. The samples are
presented simultaneously. None of the sample is identified as a standard. The

1
panelists must determine which of the three samples presented differs from the other
two.
Uses:
- It is used in quality control to ensure that the samples from different
production lots are the same.
- It is useful in selection of panelists.
- To find out if ingredient substitution or some other change in manufacturing
results in a detectable difference in the product.
ii. Duo-trio test:
This test involves three coded samples; two identical samples and one different
sample which is identified as reference, R. Panelists are asked to identify which of
the two coded samples is the different or same as R. It can be used in place of the
simple paired comparisons test if no characteristics is specified. It has the same
applications as the triangle test but is less efficient because the probability of
selecting the correct sample by chance is 50% (i.e one half) as it is with the simple
paired test. In the duo-trio and triangle tests, the tester bases his judgment on any
difference he can detect, while in the simple paired comparisons test, the tester is
asked which sample has more of some specified characteristics.
iii. Simple paired comparisons test:
This test is similar to triangle test but the difference is that, in the simple paired
comparison test, two coded samples are presented for comparison on the basis of
specified characteristics such as sweetness, sourness, etc. They are evaluated
simultaneously or sequentially in a balanced order of presentation. Both orders
should be presented to approximately the same number of panelists. The triangle test
differs from it in that, the probability of picking the correct sample by chance is one
third (1/3), whereas it is on half (1/2) or 50% in the paired test.
iv. Scheffé paired comparisons test:
Scheffé in 1952 modified the simple paired comparisons test. In his modification,
the panelists were asked to indicate the size of the difference detected. The results
are then subjected to analysis of variance (ANOVA). An average value (parameter)
for each treatment is calculated. The values of the parameters are relative and the
sum of the parameter must total zero.
v. Multiple comparisons test:
Multiple comparisons tests are used for examining the effects of replacing or
changing an ingredient, packaging materials, or changing a process or storage. It can
2
be used to effectively evaluate up to 4 or 5 samples at the same time. It also gives
information on the direction and magnitude of the difference. A reference sample, R
is presented along with 4 to 5 coded samples. The panelists are asked to compare
each coded sample with the “R” on the basis of some specified characteristics such
as aroma, sourness, sweetness, etc.
vi. Ranking:
In this test, panelists are required to order samples according to the intensity of the
sensory characteristic designated. For example, three or more samples are presented
to the panelist and is asked to rank them for a specific characteristic, e.g. intensity;
from the most intense to the last intense. The test can show where there are
perceptible differences in intensity of an attribute among samples, but it does not
give information about the size of the difference between two samples.
NB: Since samples are evaluated only in relation to each other, results from one set
of ranks cannot be compared with the results from another set of ranks unless both
contain only the same samples.
This test can also be used to obtain preliminary information on product differences,
or even to find out if panelists can discriminate among samples with known
differences. It is rapid and allows the testing of several samples at the same time.
vii. Scoring:
In this test, panelists are required to score samples using line scales or category scales
to detect perceived intensity of a sensory characteristic. The intervals on the scale
can be labeled with numbers or with descriptive terms.
Features of the scoring test:
- The magnitude and direction of differences between samples are evident from
the test (scores).
- Panelists involve are not typical consumers and their likes and dislikes are
necessary.
However, to use this test effectively, all the panelists must be evaluating the same
characteristics. Standards are included at various points in the scale to help minimize
panel variability. The standards act as anchors in counteracting the tendency of the
scale to drift in meaning with time. The instability of the scale is one major
disadvantage encountered when scoring is used in storage stability studies over an
extended period. Practically, when conducting scoring tests, panelists score the
intensity of the specified characteristics in each coded sample on an interval scale
from low intensity to high or strong intensity. Samples are presented in identical
3
sample-container coded with 3-digit random numbers, in a balanced randomized
order. Panelists are instructed to evaluate each sample independently. The
experimenter may present one at a time to each panelist, removing each sample after
testing and before presenting the next sample in order to minimize inter-comparison
of samples. In either case, the panelists are instructed to evaluate each sample and
indicate the intensity of the specified characteristic by checking an appropriate
category or by making a vertical mark on a line scale.
b. Sensitivity tests: These tests are designed to determine or measure the ability of
individuals to detect sensory characteristic(s) such as tastes, odour and feel the
specific factors like hotness. These tests are also used to select and train panel
members for evaluating the quality of products containing spices, salt and sugar, etc.
Sensitivity tests include threshold and dilution tests.
i. Threshold test:
Threshold tests are a specific level below which a compound is not detectable, and
above which it can be detected by panelists with average sensory acuity. Threshold
tests are also defined as a statistically determined point on the stimulus scale at which
transitions in a series of sensations or judgments occur. Thresholds will differ among
panelists, within panelists and among sensory stimuli.
Thresholds are used for the following reasons:
- to set defect action levels; e.g. “At what level is ethyl acetate detectable in a
wine?”
- to determine acceptable levels for additives.
- for selecting and training of panelists.
There are three types of thresholds:
 detection (absolute) threshold
- the lowest stimulus capable of producing a sensation: “I can taste something”.
- minimum stimulus intensity for example; very dilute solutions of NaCl will
taste sweet, not salty.
 recognition threshold
- the level of a stimulus at which it can be recognized and correctly identified:
“This is salty”
 difference threshold
- the extent of change in the stimulus necessary to produce a reliably noticeable
difference. For example, biscuit A is crisper than biscuit B experimentally, i.e
the amount of stimulus increase (or decrease) that is judged as stronger than
4
 the original stimulus intensity on 75% of the trials (or weaker on 25% of the
trials).
When thresholds are given in sensory literature, they are the average of the threshold
measurements of a given population of panelists. For individuals, the threshold is
your 50/50 point, i.e 50% of the time you can detect or recognize a stimulus (or
change in stimulus intensity), and 50% of the time you cannot.
Factors influencing threshold measurements include the following:
- level of stimuli
- physiological state of panelist
- panelist attentiveness
- panelist motivation
NB: Psychophysics is the study of the relationship between physical stimuli and a
person’s sensory experience, i.e., how changes in stimulus intensity correlate with
changes in a person’s sensory response.
ii. Dilution test:
These tests are designed to establish the smallest amount of an unknown material
developed as a substitute for a standard product. The quality of the test material is
represented by the dilution number. The bigger the dilution number the better is, the
quality of the test material.
2. Descriptive tests:
These tests are designed to measure qualitative and/or quantitative characteristics. It
is used when a detailed specification of sensory attribute of a product is desired. It
is also used in comparing between products when descriptions of difference are
needed. Examples include flavour profile analysis, texture profile analysis, sensory
spectrum, free choice profile, descriptive analysis, attribute scaling, category scaling
and quantitative descriptive analysis. The basic test in descriptive analysis is
“Quantitative Descriptive Analysis” (QDA), which is a descriptive test that indicates
the sensory properties of a product in a reproducible manner.

B. AFFECTIVE TESTS:
Affective tests or consumer-oriented tests are tests which assesses subjective
responses to a product. Researchers can gain an insight into consumer preferences,
attitudes, opinions, behaviours and perceptions concerning products using both
qualitative and quantitative methodologies. Consumer testing is, therefore, a key part
of the product development process and can also be used to develop sensory-related

5
brand positioning, communication and advertising. Following product launch,
consumer testing is also vital to monitor market position and, where necessary, to
find avenues for product improvements or optimization. However, such testing is
successful only when the data are reliable and valid. The so called consumer-
oriented tests include: preference, acceptance and hedonic (degree of liking) tests.
Hedonic tests are designed to measure degree of liking for a product. The 9-point
hedonic is the most commonly used scale for preference testing. The term, hedonic
means having to do with pleasure. Hedonic tests are often used to measure
preference or acceptance indirectly. It is used by panelists to express his degree of
liking or disliking a sample. Category scales ranging from like extremely, through
neither like or dislike (as its midpoint), to dislike extremely with varying numbers
of categories are used.
Preference tests are designed to express a choice between samples; one sample is
preferred and chosen over another or there is no preference. The paired-comparison
preference test is the simplest preference test, but category scales and ranking tests
are also often used to detect preference.
Acceptance tests are tests designed to determine the degree of consumers’
acceptance for a product. These tests include: category scales (scoring), ranking and
paired-comparison tests. In practice, during presentation of samples, three or four
samples are presented in identical containers, which are coded with 3-digit random
numbers. All the samples are simultaneously presented to each panelist in a balanced
or random order. The samples could be retested.

6
 Problem definition

Test objective

Method selection

Panel selection

Experimental design

Conducting experiment

Statistical analysis

Reporting results

Fig. 1: Sensory evaluation profile flow chart

Experimental designs:
Experimental designs are plans, arrangements, or a sequence of steps for setting up,
carrying out and analyzing the results of an experiment. An appropriate and efficient
experimental design should be chosen to enable the reliability of data and test results.
In practical sensory analysis, experimental design refers to all the primary
components of a sensory experiment; which include objective, test (hypothesis)
objective, testing environment, samples, judges, questionnaire and data analysis.
Selection of the appropriate design and analysis for a sensory experiment will
depend on the information associated with the above components.
In sample terms, a design is selected based on the objectives of the study, the type
of product under study, the testing procedures and conditions, the resources
available, and the type of statistical test to be conducted. There are many types of
experimental designs, ranging from the simple, completely randomized designs to
the more complicated, fractional factorial designs. However, good statistical
textbooks and good statistician or biometrician should be consulted ahead of time,
to recommend the simplest and most efficient design which will meet the specific
7
objective of the study. Some of the common features of experimental designs
include: randomization, blocking and replication.
Randomization is a statistical process in experimental design, which minimizes the
effect of uncontrollable sources of variation or error and also eliminate bias. It is also
a procedure for ordering units or samples such that each unit has an equal chance of
being chosen at each stage of the ordering process.
Blocking is a statistical term referring to a procedure for statistical design employed
in statistical analysis. The essence is to control for unknown sources of variation and
to improve efficiency.
Replication is a statistical procedure that involves repeating an entire experiment
under identical conditions. It estimates of experimental error and improves the
reliability and validity of the test results.

Common experimental designs in sensory analysis include:

1. Completely randomized design
Completely randomized design (CRD) is the simplest design for comparative
experiments as it uses only two basic principles/features of experimental designs;
namely randomization and replication. These are designed for studying the effect of
one primary factor without the need to take other variables into account. For CRDs,
the levels of the primary factor are randomly assigned to the experimental units.
Here, products are assigned randomly to assessors who assess only one product each.
Typically, several results are collected for each product and every product is seen by
a different group of assessors. This design is not ideal as assessors cannot be
identified as separate ‘blocks’ and, therefore, variation in their results cannot be
considered in the data analysis.
2. Randomized complete block design
This is a standard design for agricultural experiments in which similar experimental
units are grouped into blocks or replicates. It is used to control variation in an
experiment; for example in a sensory analysis, too many products up to 10 and above
can be divided into 2 sessions to obtain reliable results. All products are presented
in a randomized balanced order of presentation across the panel for each session.
Assessors may also make replicate judgements for each product over one or more
sessions. Variation from individual assessors and the replication (sessions) can be
considered in the data analysis. In behaviourial sciences, this type of design is also
called a ‘repeated measures’ design.
8
3. Balanced incomplete block
In the balanced incomplete block (BIB) design, a subset of the total number of
products is presented during a session. Ultimately, each assessor may see all samples
over several sessions (although in practice, it has been common to present samples
over several sessions and call this a ‘complete’ design), or only a subset of the total
group. In either case, samples presented within a session are randomly allocated and
their order is balanced such that each assessor receives the same number of samples,
each sample is seen an equal number of times across the session and each sample is
seen in combination with every other sample an equal number of times across the
session. Software programs are used to help design BIBs.
Functions of replication:
Replication help to obtain a valid and more reliable estimate than which is possible
with one observation only. Replication provides an efficient way of increasing the
precision of an experiment. The precision increases with the increase in the number
of observations. Replication provides more observations when the same treatment
is used, so it increases precision.

Importance of experimental design:

1. To obtain valid results from which safe conclusions can be drawn
2. To know how widely these may apply
3. To use resources efficiently
4. To minimize severity
5. To ensure reproducibility

Characteristics of a good experiment (that is one with experimental design)

1. Is unbiased
Has independent repeats (replicates) which are randomly assigned to the different
fixed experimental conditions.
2. Is precise
Has uniform material and/or control of variability or is a large experiment
3. Has a wide range of applicability
Includes many controlled variables (e.g. sex, strain)
Allows interaction between the variables to be assessed
4. Is simple to analyze
Keeps to a formal design

9
Has equal numbers in the sub-groups
5. Allows uncertainty to be calculated
i.e. has independent repeats

10