FST 151 - 3 general types of tests:

Sensory Evaluation of Food Products • Discriminative test/ difference test

1st Semester, SY 2022-2023 • Descriptive test
• Affective test
- different scoring test
MODULE 1: 2. use of hedonic scale to evaluate the sensory
INTRODUCTION TO SENSORY EVALUATION quality attributes of food products
- hedonic scale is only used in Affective test
(food acceptance or degree of liking)
Lesson 1: Introduction 3. use of inappropriate panelists
- each type requires a different level of training for
Sensory evaluation panelists
- a scientific discipline used to evoke, measure, 4. use of inappropriate technique
analyze, and interpret those responses to products - depends on the objective
that are perceived by the senses of sight, smell,
touch, taste, and hearing (Stone and Sidel, 1993) Subjective vs. Objective Evaluation
o definition used by Institute of Food • Subjective evaluation
Technologists (IFT) and the - involves consumer evaluations based on
American Association for Testing consumer preference
and Materials (AAsTM) - only acceptable for affective testing
- compromises a set of techniques for accurate - no panel training
measurement of human responses to foods and ➢ Affective test
minimize potential biasing effects of brand • Objective evaluation
identity and other information influences on - measures intensity response to a stimulus
consumer perception - define physical, chemical, and
- concerned with precision, accuracy, and biological properties of a product
sensitivity, and avoiding false positive results - det. product differences and specify the basis
- similarities and differences among products of difference
- identify product characteristics which are significant
History of Sensory Evaluation to consumer acceptance
• 1930- studies on food chemical compounds ➢ Discriminative/ difference test
affecting its acceptability; done by experts ➢ Descriptive test
• 1940-1950- Sensory Evaluation as a field of study
began after WWII at the US Army Quartermaster Sensory analyst
Food and Container Institute; army rations - measures with senses
• 1955- food quantity to food quality - uses exact methods
• 1960- British Standard Institute set up standards - can be compared to an instrument
for sensory analysis; defining SE - participate in appropriate training courses (calibration)
• 1976- Sensory Science became a distinct, and periodic selection tests
recognized scientific specialty that covers: - works in a panel where results are
o Psychology analyzed statistically
o Physiology - objectively measures sensory differences of classifies
o Chemistry the quality of a product using a well-defined scale
o Statistics
- demand in food variety became the key for Lesson 2: Introduction to Sensory
production of processed food products = market Perception Sensory evaluation
competition (man’s reasons for product selection)
• 1970-1990- Sensory Science is inside Research - use our five senses to identify, evaluate, and
and Development = Marketing measure certain product characteristics
• Today- Sensory works together with R&D - X physical sensations = interpret
and Marketing
Sensory vs. Perception
Applications of Sensory Evaluation • Sensation
1. Setting standards - sensory receptors and nervous system receive
- det. of quality criteria for classification, grading and represent stimulus from environment
and pricing of raw materials, ingredients, and • Perception
finished products - organizing and interpreting sensory information
2. Quality control and quality assurance to help us recognize
- det. of sensory guidelines for handling and storage to - dependent on sensation but not all sensations
meet trade standards and consumer acceptance are perceived
- inspection of ingredients, products, and packaging
3. Product development, reformulation, and optimization The Five Senses
- formulation of new products and modification • Vision (sense of sight)
(improve) of existing ones - perceive the initial quality of food (color,
size, texture)
4. Product positioning and competition monitoring
- comparison of old products vs. new
developments, also with competitors
5. Correlations
- Analytical sensory science = physical,
chemical, and instrumental estimates
- Consumer analysis = consumer behavior
6. Perception
- determine insights and reactions of
consumers towards a product
a. cornea- where light is reflected; refracts the rays
Misuse of Sensory Evaluation b. iris- opens and closes to regulate the amount
1. use of a general scoring system of light passing through the pupil
 c. pupil- where refracted light enters o umami
d. lens- further bend the light rays to focus on
the retina Taste perception
e. retina- contains millions of light sensing - detection of stimuli
nerve cells (rods and cones) (dissolved in water, oil,
o rods- outside the macula; peripheral or saliva) by taste buds
vision; dim light - optimal contact:
o cones- concentrated in the center of small sips in the mouth
retina (macula); colors; bright light for 2 seconds
- higher taste sensitivity
Color perception = more taste buds
- different sensitivities - factors:
- several types of color blindness o age (increasing age = decreasing sensation)
• deuteranopia- green and red o meal (meals before evaluation = reduce)
• achromatopsia- no cone cells (black o hunger (more hungry = more sensitive)
and white only) o smoking, obesity (less sensitive),
- exceptional color sensitivity pregnancy, flu, temperature, viscosity
- taste disorders
• Olfaction (sense of smell) o ageusia- absence of sense of taste
- one of the two that can detect chemical o hypoageusia- reduced sense of taste
stimulants (olfaction and gustation)
- odorant molecules are sensed by millions of • Audition (sense of hearing)
cilia (tiny hair-like)
- flavor: composed of taste and aroma

• Touch & mouthfeel

a. Somesthesis
a. (1) olfactory bulb - tactile sense, skinfeel
b. (2) mitral cells- star-like projections
c. (5) glomeruli- sensory input will start to
interact with parts of the brain (smell
identification, memory, emotion)
d. (3) bonglayer
e. (4) nasal epithelium
f. (6) olfactory receptors- transmits signals to
the glomeruli
g. cilia- sense odorant molecules

Odor perception b. Kinesthesis

- different sensitivities - perception of body movements
- factors: hunger, satiety, concentration, +/- - sensory axons are in the muscles, tendons
of respiratory infections, pregnancy c. Chemesthesis
- terminal threshold- any concentration higher - trigeminal sensation
than the TT = no change in the perceived intensity - hot, burning, tingling, cooling,
of the stimulus astringent, sensation
- optimal contact: moderate sniffing for 1-2 seconds - exposed in chemicals
(5-20 seconds longer for break)
- odor disorder Sensory attributes
a. anosmia- inability to detect odors
b. specific anosmia- unable to detect specific Order of Perception
odors
c. hyposmia- decreased ability to detect odors Appearance > Odor/ aroma/ fragrance >
Texture > Flavor
• Gustation (sense of taste)
- detect chemical stimulus ➢ Appearance
- taste receptor cells in the taste buds - visual perception of food
- taste buds are contained inside papillae - color, size, shape, transparency, and
o fungiform papillae surface characteristics
o foliate papillae - first attribute on which we base decisions
o circumvallate papillae
o filiform papillae (no taste buds) Color
- food is dissolved in the saliva in the taste bud - perception that results from detection of light after it
pores meet cilia (send signals to the brain) has interacted with an object
- brain interprets signals into sensations: taste
o sweet Factors affecting color perception
o salty 1. Physical factors (eye sensitivity)
o sour 2. Psychological factors (perception,
o bitter learning, thinking)
 Importance of color in sensory evaluation of food
Food expectations (visual signals that affects Mouthfeel properties
person’s choices; indicator of maturity; color of meat • Moistness- degree of wetness/ oiliness;
indicates spoilage; completeness of cooking) dry, moist, wet/ oily, greasy
• Moisture release- amount of water/
Other characteristics in appearance: oil excluded/ juicy, oily
- size and shape
- clarity and transparency ➢ Flavor
- surface characteristics - combined senses of taste, aroma, and mouthfeel
- aroma provides 75% of the impression of flavor
➢ Odor/ aroma/ fragrance
- stimulus is in gaseous state
- important component of flavor
- connected to temperature (hot foods are easier to
smell than cold ones) ➢ Sound
- differences in pitch of some rupturing foods (crispy,
Olfaction crunchy, brittle) provide sensory input = assessment
• Orthonasal olfaction- through the nose of freshness/ staleness
• Retronasal olfaction- after entering the
mouth, towards the back of throat, to Sensory interactions
the nasal cavity - sensation intensities are additive
- some subjects misattribute some volatile sensations
Definitions: in the mouth to “taste”
• Smell- neutral, good or bad - harsh tastes generally suppress, and pleasant
• Odor- bad smell tastes generally enhance, volatile flavor
• Aroma- smell of food - interactions change with various taste/
• Fragrance- perfume or cosmetics flavor combinations
➢ Taste
- gustatory sensations caused by soluble Lesson 3: Factors Influencing Sensory Judgements
substances in the mouth
Importance: to produce results with less or no biases
Basic tastes
• Sourness- H+, perception of A. Psychological factors
H+ concentrations, acids • Expectation error
• Sweetness- sugars and other compounds - subjects try to make the right answer from
of similar chemical composition (alcohols) the knowledge about a product
• Saltiness- salts (NaCl) - having some knowledge (brands)
• Bitterness- alkaloids (caffeine, theobromine)
• Umami- glutamate salts or glutamic acid, PREVENT:
Japanese term for “delicious” ✓ don’t include people with product
knowledge on panel
➢ Texture ✓ don’t give any information regarding
- most complex sensory attribute the test and samples
- rheological (viscous of flowy) and structural ✓ coding and randomization
(geometric) attributes of the product perceptible by • Suggestion effect
means of mechanical, tactile, and where - comments/ noises made out loud
appropriate, visual and auditory receptors influence judgements
- measure of quality
- sensory manifestation of structure or inner PREVENT:
make-up of product ✓ sensory booths
✓ prevent discussions before
Components of texture and after evaluation
• Mechanical- reaction to force/ stress • Distraction effect
• Geometric- perception of particles: - presence of stimuli (environment) or
size, shape, orientation); tactile personal preoccupations (time pressure)
• Mouthfeel- physical sensations in the
mouth; tactile PREVENT:
✓ test area must be quiet
Phases of texture perception ✓ prohibit use of electronic devices
• Initial phase- first bite • Stimulus error
• Second phase- mastication/ chewing - assessors use additional (irrelevant)
• Third phase- after swallowing information (plastic cups vs. glass
containers; amount of sample)
Geometrical properties • Logical error
• Hardness- force needed to attain a given - stimulus is logically associated with
deformation/ resistance to indentation; the characteristics being evaluated
soft, firm, hard
• Cohesiveness- degree to which a substance PREVENT (Stimulus & Logical error):
acts to unite its parts; brittle, gummy ✓ mask irrelevant differences
• Adhesiveness- force required to remove ✓ keep samples uniform
sample from a given surface; sticky, dry • Halo effect
• Denseness- compactness of cross- - rating of one attribute may influence the
section; more compact = more dense; ratings of other attributes
dense, airy, puffy - judging several attributes at once
• Springiness- rate of return to original
shape after deformation; springy, rubbery PREVENT:
• Smooth, gritty, grainy, bumpy, powdery, ✓ evaluate one or limited number of
fibrous attributes at a time
✓ use trained panelists
 • Proximity error ✓ limit the number of
- rating of attributes following on from one samples presented
another tend to be related ✓ appropriate time intervals
- judging several attributes at once ✓ provide appropriate palate
- difference from Halo effect: attributes that cleansers (water, milk)
are closely placed in the scoresheet

PREVENT:
✓ randomize the order of attribute • Stimulus interaction (enhancement)
evaluation - presence of one substance increases the
• Habituation error perceived intensity of another (sugar and
- scoring similar products on a regular salt concentration)
basis = develop the habit of assigning • Stimulus interaction (suppression)
similar scores - presence of one substance decreases
the intensity of another (fruit wines)
PREVENT: • Stimulus interaction (synergy)
✓ vary products/ introduce a modified - intensity of mixture is greater than the
sample sum of individual components (effect of
✓ make panelists know that you MSG, herbs and spices)
are tracking their performance
• Order effect C. Cultural factors
- score assigned to a sample can be • product codes
influenced by the sensory character of the • spiritual restrictions
preceding product (very sweet then less
sweet) Applications
- some sample positions are often favored - wine would be too sour after eating a dessert
(Hedonic test: First sample effect) - after eating sour food, wine would taste too sweet
- chocolate- child labor (psychological effect)
PREVENT: - brands
✓ randomize/ balance the order
✓ use dummy sample in position
one (affective test)
• Contrast effect MODULE 2: DESIGNING AND CONDUCTING A SENSORY
- two samples that are strikingly different = TEST
exaggerated difference rating
• Convergence effect
- opposite of Contrast effect Lesson 4: Panel Selection and Training
- products as part of a widely varying
sample set = difference may be rated Panelists are considered as measuring equipment/ tools.
smaller than it actually is
Consumer Panel vs. Trained Panel
PREVENT (Contrast & Convergence effect):
✓ randomize/ balance the order Consumer panel
of presentation - determine preference and acceptability
✓ consider removing outlying - likes and dislikes (no reason why)
samples from the set - untrained, current/ potential users of the product
B. Physiological factors Trained panel
• Error of central tendency - trained and employed
- use of scales, assessors avoid extremes
- determine sensory differences, sensory acuity
and confine ratings at the middle - never used for consumer testing (too sensitive)
PREVENT: Steps in selecting and training Panel members
✓ training of assessors
✓ use scales large enough to 1. Recruitment
differentiate between products - interest and willingness, motivation
• Motivation error - health (allergies)
- motivated panelist will learn better - integrity
and perform more reliably - attitude, stability
- given a questionnaire (record-keeping)
PREVENT:
✓ respect assessors 2. Screening
✓ give regular feedback to them - focus on all parameters that would be
✓ carry out sessions in measured during the actual test
professional manner - common tests:
• Time error/ Positional bias a. Threshold test- identify basic taste
- panelists’ attitude undergoes subtle and common odors
changes over series of tests b. Triangle test- sensitivity/ discriminating
ability (difference/ discriminative test)
PREVENT: - use of food samples that are identical in all
✓ randomize order serving but differs in one flavor/ texture
✓ conduct evaluations at appropriate - high sensitivity = chosen
time
• Adaptation error 3. Training
- continued exposure to stimulus = decrease - give panelists enough information about the
sensitivity test (stimulate interest and know what is
expected of them)
PREVENT: - let testers participate in setting up the score sheet
 - keep score sheets simple (easy to use - adjacent to testing area (people do not have
and understand) to pass)
- group of 10-15 members (top performers from the - well-ventilated (prevent odors to come into the
screening test) testing area)
- training is done half an hour a day for 2-4 times - components: space, sink, cooking equipment,
a week refrigerator, dishes, cleaning and storage facilities
a. difference and ranking test- few • Testing area
sessions - actual evaluation
b. quantitative test- at least 10 sessions - separate but adjacent to preparation area
- calibrating the instruments (panelists) (centrally located)
- supply reference samples (similar to the test) - divided into 5-10 small booths
- allow members to compare results with experienced - temperature controlled: 22-24 degrees C, RH=
panels after the test and keep them informed 45-55%
- proceed when discussed or when panelists - free from odor (positive pressure: air must
give consistent responses be circulating outward)

4. Performance evaluation Testing booth

- determine the progress of training - sufficiently large (0.9m wide and 0.6m deep)
- evaluate individual panel performance and as - color of walls: white/ off-white, even-shadow-
a whole free illumination (lighting system: 300-800)

5. Panel retraining Types of Hatches for passing samples to and from

- maintain familiarity with relevant attributes the panelists:
- reference samples ➢ sliding door type (top to bottom, left to right)
- substitution of panel members should be ➢ bread box type
avoided (reduce variability) ➢ carousel type

Summary: Product control

- way samples are stored, prepared, screened,
numbered/ coded, and served (must be similar)
- handling, preparation, and presentation
- no foreign odor or flavor should contaminate

Product sampling
- determination of how much of product is required
- should come from one source (uniform and
homogenous)

Materials used
- plastic cutlery, storage containers, wraps/ bags
are unsuitable
- wooden materials should not be used as
cutting boards, bowls, mixing utensils (porous)
- recommended: glass materials, chinaware,
stainless steel
Lesson 5: Control of Sensory Variables
Number of samples in a session
Sensory variables that should be controlled: - function of:
➢ Test control o sensory and mental fatigue
➢ Product control (subject) o nature of product
➢ Panel control o complexity of sensory attributes
o experience of judge
Test control o evaluation time
- physical setting that are controlled to: o availability of sample
✓ minimize subjects’ biases
✓ maximize their sensitivity Serving size
✓ eliminate variables which do not come from the - proper and uniform amount
products themselves - 15mL for liquids or 15-20g for solids

Common design of sensory laboratory: Sample presentation

- presented on containers/ plates that are same in
size, shape, color
- should be identical (except when physical
attributes are judged)

Serving temperature
- present samples at same temperature specified in
the protocol
- - at which they are normally consumed

Holding time
- minimum and maximum time after preparation
• Briefing room
- assembly area for evaluators (briefing) Serving matrix
- comfortable seating - carrier- any food that accompany the sample
- should be adjoining to testing booths being tested
- where descriptive tests and trainings are conducted
• Preparation area Coding
- preparation of samples - mask brand identities
Order of serving
- balanced and randomized

Palate cleansing
- room temperature water or plain bread

Expectoration vs. Swallowing

- swallowing is more acceptable

Panel control
- interactions with environment, product, and test
procedure

Panel selection
- equal number of men and women
- age distribution
- recruit right people (committed, know what is
expected, of good health, non-smokers)

Panel training and orientation

- panelists should be familiar with:
✓ test procedure (how to evaluate)
✓ score sheet design (scale)
✓ type of evaluation required

Types of panels
• Trained/ descriptive panels- highly trained (5-
10 members)
• Semi-trained/ discriminative panels- technical
who are familiar with product (25-30 members)
• Untrained/ consumer panels- random, represent
consumer population (large number, more than
100 members)

Panelist and environment

- lights, humidity, and condition in testing are should
be proper

Incentives
- give snacks or small amount of money
(participate voluntarily)

Product/ time of the day

- products should be evaluated at the time of the
day when the product is normally consumed
- 11am and 3pm

Other factors:
✓ avoid using strong smelling cosmetics
✓ avoid eating highly spiced foods
✓ rinse your mouth prior to evaluation
✓ should not hurry
✓ should not ingest any other food for at least 1
hour and should not chew gum
✓ requires intense concentration
✓ men and women

Excused panelists
✓ suffering from colds or fever
✓ emotionally upset or pressured
✓ had a major meal in less than 1 hour
Lesson 6: Designing & Conducting a Sensory - can be analyzed by more non-
Test Human subjects as instruments. parametric statistical tests
- do not provide amount of the differences
- more sensitive - screening three or more products for specific
- instruments can’t measure liking properties

Panels/ judges are: Ex:

- variable over time Please rank the following food items from the most (1)
- variable among themselves- enough number to least (3) eaten food item.
of subjects (representative)
- prone to bias (must be avoided) Food item Rank
- prone to fatigue and attention drift Chicken ala king _____
Chicken salad _____
Types of panels: Tuna noodle casserole _____
- differences:
o number of panelists involved 3. Interval
o method of selection/ level of training - panelists place items into numbered
groups separated by constant interval
1. Consumer panel - easy to understand, use, and analyze
- more than 100 respondents - can provide information about relative judgments
- randomly selected - can be analyzed by non-parametric, and often
- no training by parametric statistical techniques
- first impressions or preference or acceptability only
Ex:
2. Consumer-type panel Hedonic: Intensity:
- 50-100 respondents _ Dislike very much 0 None
- large companies: employees _ Dislike moderately 1 Very slight
- research and academic institutions: students _ Dislike slightly 2 Slight
- less expensive than consumer panels _ Neither like nor dislike 3 Moderate
- first impressions or preference or acceptability only _ Like slightly 4 Strong
- depends on the consumer intentions _ Like moderately 5 Very strong
_ Like very much
3. Laboratory panel
- trained/ semi-trained ➢ each category in the scale responds to
- less respondents a numerical value (subjected to
- objective: comparable to an instrument statistical analysis)
- product development and improvement, shelf-life ➢ Hedonic scale- distance among
studies, and determination of product features and choices are equal
characteristics
➢ Discrimination panel- determines if small 4. Ratio
differences exist among samples - can be analyzed by all parametric and
➢ Descriptive panel- identifies and described non-parametric tests
products based on specific attributes - panelists use numbers which indicate how many
times the stimulus is stronger than the
4. Expert judge reference stimulus presented earlier
- extraordinarily sensitive to one or two odor or - disadvantage: complex task, require training
taste sensations
- wine testers, tea experts, coffee experts, chefs Methods of Measurement
- selection of the best among many formulations I. Classification
II. Grading
Measuring Sensory Responses III. Ranking
IV. Scaling
Types of data: V. Threshold

1. Nominal 1. Classification
- items examines ae placed in two or more groups - items evaluated are sorted into groups which differ
which differ in name but do not follow any order in nominal manner
or quantitative relationship - analysis is limited to tallying the responses
- ex: number assigned to basketball players
- classification tests and gathering Ex:
preliminary information Place a check on the blank corresponding to
the term(s) which best describe(s) the sample.
Ex: _ sweet _ sour _ lemony
Do you have any food restrictions? Please check _ blended _ thick _ refreshing
all that apply. _ pulpy _ natural _ w/ aftertaste
________ food allergies (nuts _ milk _ wheat _ others:
specify _) ➢ How many answered thick-lemony?
________ health or medically related, specify _) ➢ Data are NOMINAL.
________ vegetarian (consume: chicken _, fish _, dairy ➢ Choices should be based on the actual
_, eggs _) product characteristics (must be complete
and best choice of words)
✓ answers in this type of questionnaire are
summarized and the data are NOMINAL 2. Grading
- giving score based on standards; depends
2. Ordinal of expert graders
- observations are arranged and classified - involves a process integration of perceptions by
according to a succession depending on the the grader
presence or more or less a particular attribute - examples of food subjected to grading:
coffee, spices, butter, fish, meat
 - rate the combined effect of the presence of the other samples for crispiness in
positive attributes, blend and balance of attributes, proportion to 25. If the crispiness of
absence of negative attributes, comparison of any sample is half of the first, assign
products with written or physical standards it a crispiness value of 12.5:

Ex: Sample 1 25
Coffee bean grading Sample 2 __
Sample 3 __
Grade 1 Specialty grade
Grade 2 Premium grade • without modulus
Grade 3 Exchange grade
Grade 4 Standard grade Ex:
Grade 5 Off-grade Evaluate the first cookie; assign
any number to the crispness of that
3. Ranking cookie. Rate all other samples in
- samples are arranged in order of intensity proportion to the rating given to the
or degree of some specified attribute first sample.
- type of data: ORDINAL
Sample 1 __
Ex: Sample 2 __
Rank the samples from the sweetest (1) to least sweet Sample 3 __
(5).
Sample 1 _ Scale length
Sample 2 _ - tendency for panelists to ignore both ends of
Sample 3 _ the scale and use only middle categories
Sample 4 _ - too few categories: lack discrimination ability
Sample 5 _ - wide range of categories: good reliability
- longer scale = more discriminating
4. Scaling
- use of numbers or words to express the intensity Choosing a scale
of a perceived attribute or reaction to such attribute ✓ simplicity (easy to use, user-friendly)
- sweetness, hardness, smoothness ✓ unbiased
✓ relevant
Types of scales: ✓ sensitive to sample difference
✓ magnitude of error variance (standard
a. Category scaling deviations; must be small)
- rate the intensity of a particular stimulus ✓ reliability (similar results)
by assigning it to a value (category) on a ✓ allows for appropriate statistical analysis
limited scale
Sequential System Approach in
Ex: Sensory Evaluation
Number category scale Word category - provide valid and reliable information to
0 None R&D, production, and marketing team
1 Threshold - finding most cost-effective and efficient method
2 Very slight
3 Slight Selecting the most appropriate technique:
4 Slight-moderate ✓ What is the sensory objective?
5 Moderate ✓ What are the variables involved in the test?
6 Moderate-strong ✓ What sequence of sensory tests are
7 Strong necessary to attain the objectives?
✓ What is the experimental design? (set of
b. Linear scaling treatments and variables to be considered)
- rating the intensity by making a mark on ✓ How will the results be analyzed?
the horizontal line which corresponds to the ✓ What action should be taken from
amount of the perceived stimulus the results obtained?

Ex: Designing a Sensory Test

Appearance Dark Light I. Definition of the Problem
Yellow color |---------------------------------- | - define the objectives
II. Test Design
Aroma Unchar. Char. - test must be designed in such a way that the
Characteristic |---------------------------------- | number of testing must be minimized to
aroma of mango produce the desired accuracy of the results
juice III. Instrumentation
- subjects must be selected and trained (if
Flavor Unchar. Char. necessary) to attain reproducible results
Characteristic |---------------------------------- | IV. Interpretation of Results
flavor of mango - sensory analyst must be able to come up with
juice the correct choice of Ho and Ha and draw
conclusions which are warranted by the results
c. Magnitude estimation
- free assignment of numbers by a panelist Analytic vs. Hedonic Scale
to indicate relative intensities of stimuli • Analytic Scale
- includes both discriminative and
• with modulus descriptive tests
- judge is viewed as an analytical instrument
Ex: used to detect and measure changes
The first cookie has an assigned • Hedonic Scale
crispiness rating of 25. Rate all - involves the affective test
 - measures consumer liking and preference 3 (2 identical,
1 different)
Duo-Trio * 1 sample is Binomial distribution
labelled, 2 are
coded
Method Selection 3 (2 identical,
1 different)
Class Question of Interest Panel Characteristic Triangle * all samples Binomial distribution
Screened for are coded
sensory acuity, * no standard
Are the products oriented to test Ranking
Discrimination
different? method, sometimes a. Ranking
trained (semi- Difference Test Rank Analysis
trained) 2-7
b. Ranking in Analysis of Variance
Screened for Acceptance/
How do products
sensory acuity & Affective Test
Descriptive differ in specific
motivation, trained Rating
sensory attribute?
or highly trained Difference from Analysis of Variance
How well are 1-18
Screened for the Control Rank analysis
products liked? or (difference test)
Affective product use,
Which products are Threshold 5-15 Sequential Analysis
untrained
preferred? Dilution 5-15 Sequential Analysis
Attribute Rating
Method Selection- Decision Tree Analysis of Variance
(category scaling
Rank Analysis
and ratio scaling 1-18
Factor Analysis
or magnitude
Graphic Presentation
estimation)
Graphic Presentation
Flavor Profile
Principal Component
Analysis 1
and Multivariate
(descriptive test)
Analysis of Variance
Graphic Presentation
Texture Profile
Principal Component
Analysis 1
and Multivariate
(descriptive test)
Analysis of Variance
Quantitative
Analysis of Variance
Descriptive
1-5 Regression Analysis
Analysis
Graphic Presentation
(descriptive test)
1-18 (larger
number of
Hedonic Scale samples only
Analysis of Variance
Rating if it is mild-
Rank Analysis
(affective test) flavored or
rated for
texture only)
1-18 (larger
number of
Food Action samples only
Analysis of Variance
Rating Scale if it is mild-
Rank Analysis
(affective test) flavored or
rated for
texture only)

5. Threshold

5. Threshold

Some Sensory Evaluation Methods

No. of
Method samples per Data analysis
test
Paired
Difference (or
2 Binomial distribution
Paired
Preference)
Lessons 7 and 8: Statistical Analysis of Sensory Data Alternative Hypothesis, Ha
- a statement contrary to the null hypothesis
Branches of Statistics - e.g. The treatment means are not equal

Descriptive Statistics Steps:

- methods that describe the data collected 1. A sample of elements from the population is collected and
- focuses on collecting, summarizing, and presenting a the measurement of interest is taken on each element of
set of data the sample.
2. The value of the statistic used to estimate the parameter
Inferential Statistics of interest is calculated.
- analyzes sample data to draw conclusions about a 3. Based on the assumed probability distribution of the
population measurements and the Ho, the probability that the statistic
- generalizations can be made about a population (a takes on the value previously obtained is calculated. If this
larger group or subjects) out of the data collected probability is smaller than some predetermined α
from a sample (p=value), Ho is rejected in favor of Ha.
- this method is allowed because random sampling - “Fail to reject Ho” and not “accepting Ho” as
gives us the confidence that the sample represents acceptance implies that the null hypothesis is true
the whole population - “Fail to reject Ho” implies that the data that we have
are not sufficiently persuasive for us to prefer the
Measures of Central Tendency alternative hypothesis, Ha over Ho.
- describes a whole set of data with a single value that
represents the middle or center of a distribution Types of Error in Hypothesis
- for symmetric distribution, the mean, median, and
mode, are equal Testing Type I Error
- Error committed when the null hypothesis is rejected
Mean when in fact it is true (Ha false)
- the arithmetic average - pr [Type I error] = α
- can be used for both continuous and discrete numeric - aka significance level
data
- cannot be calculated for categorical data Type II Error
- greatly influenced by outliers and skewed distribution - error committed when the null hypothesis is not
- extremely high values – mean is pulled upwards rejected when in fact it is false (Ha true)
- extremely low values – mean is pulled downwards - pr [Type II error] = β
- power of the test: 1 – β
Median
- the middle value when the values are arranged
(ascending/descending) Statistical Analysis for Two Sample Tests
- divides the distribution in half (50% of the data are on
the left, the other 50% is on the right) [ for discrete data ]
- less affected by outliers and skewed distribution
- preferred measured of central tendency if the Adjusted Chi-Square Test
distribution is not symmetrical - Compares the observed frequencies with the
- cannot be identified by categorical nominal data as it expected frequencies
cannot be logically ordered - Commonly seen/used in difference tests such as:
triangle test, duo-trio test, simple pair difference test
Mode - Test statistic is calculated using the formula:
- the most frequently occurring value
- can be found in both numerical in categorical data
- may not always reflect the center of the distribution
(symmetric distribution)

Measures of Spread

Range
- the difference between the highest value and the
- to determine if the two samples are significantly
lowest value in the data
different, compare the test statistic chi-square to the
tabular value (Appendix Table 7) with df = 1 (df = no.
Mean Deviation
of treatments – 1) at α = 0.05
- the average of the absolute deviations from the mean
- compare with 3.84
- if x is ≥ 3.84, reject Ho – samples are significantly
Variance
different
- measures how far each number in the set is from the
mean
[ for continuous data ]
Standard Deviation
T-test
- the square root of the variance
- a hypothesis test of the mean of one or two normally
distributed populations
- can be used to determine significant differences between
Hypothesis Testing
two means for small sizes (n < 30, s unknown)
Null Hypothesis, Ho
One-sample t-test
- a statement of no difference or equality of two or more
- tests whether the mean of a single population is equal
factors or situations under study
to a target value
- “null” - commonly accepted fact that the researchers
- e.g. is the mean score for acceptability greater than
want to nullify
6? o Ho: µ = 6
- e.g. The treatment means are equal or no difference.
o Ha: µ  6 (two-tailed) or
o µ < 6 or Ha: µ > 6 (one-tailed)
- test statistic is computed as:
Statistical Analysis for ≥3 Samples
Analysis of Variance (ANOVA)
- - used when the test objective is to determine in which
- Where: way a particular sensory attribute varies over a
o x̄ = mean number of t samples, where t may vary from 3 to 6, or
o µ = target value at most 8
o s = standard deviation - it is possible to compare all t samples as one large set
o n = total no. of judges - test procedure: F-test
- test statistic: F = MSTr / MSE
- Decision Rule (one-tailed): Reject Ho when | t | ≥ - decision rule: reject Ho,
tα,n-1, otherwise fail to reject Ho if Fc ≥ F0.05(DFTr,DFE) otherwise fail to reject Ho
- rejecting Ho means at least one of the treatments is
- Decision Rule (two-tailed): Reject Ho when | t | ≥ significantly different from others
tα/2,n-1, otherwise fail to reject Ho
- Ex: rejecting Ho = “Therefore, at 5% significance level, Post Hoc Analysis:
we have sufficient evidence to say that the average
overall liking of the sample was greater than 6.” Duncan’s New Multiple Range Test (DNMRT)
- only done when the ANOVA is significant and
concluded that there is at least one treatment is
Two-sample t-test significantly different
- Tests whether the difference between the means of - pairwise mean comparison test that is used to
two independent populations is equal to a target value determine which groups/treatments are significantly
- Ex: does the mean score given by female panels different
significantly differ from that of the male panel?
o Ho: µ1 = µ2 or µ1 - µ2 = 0 Steps:
o Ha: µ1  µ2 or µ1  µ2 = 0 1. Calculate the sample means.
- The test statistic is computed as: (two-tailed) 2. Arrange the sample means in ascending or descending
order.
3. Calculate the standard error of the mean.

- Where:
o x̄ 1 and x̄ 2 = means of the two independent 4. Calculate the least significant range (LSR)
populations
o s12 and s22 = variance of population 1 and 2
o n1 and n2 = number of observations in * SSR (Significant Studentized Range)
5. Compare the differences between the sample means with
population 1 and 2
the corresponding LSR.
a. If the difference is greater than or equal to the
- decision rule: reject Ho if | t |≥ tα/2, n1+n2-2, otherwise
LSR, treatments are significantly different from
fail to reject Ho
each other
Paired t-test
Measures of Relationship
- Tests whether the mean of the differences between
- sensory evaluation methods are utilized not only to
dependent or paired observations is equal to a target
measure differences or preferences
value
- there are instances where we need to measure the
- Same set of panelists analyze two sets of samples
existence of degree of relationship or association
- Ex: If you measure the off-flavor scores of wrapped
and unwrapped cheese blocks, is the mean score
Correlation Analysis
significant enough to conclude that the packaging is
- a statistical technique used to determine the strength
effective?
or degree of linear association between variables
o Ho: = µ1 - µ2 = 0 - does NOT cause-effect relationship
o Ha: = 0 (one-tailed test) o Simple Linear Correlation
o Ha: < 0 or > 0 (two-tailed test) - a useful measure of relationship for
- Test statistic is computed as: quantifiable values which values which can
be paired, or which come in pairs
- the nature and degree of relationship
between paired factors is measured by the
- coefficient of correlation, r, whose value
ranges from -1 to 1

- Where:
o = mean difference
o 0 = target value

o S = standard error of the mean difference

o n = number of observations

- Decision rule (one-tailed): reject Ho if |

t | ≥ tα, n-1, otherwise fail to reject Ho

- Decision rule (two-tailed): reject Ho, if |

t | ≥ tα/2, n-1, otherwise fail to reject Ho
 - nature of correlation:

- Ho: There is no linear association between X

and Y (r=0)
- Ha: There is a linear association between X
and Y (r)
- Test statistic, rc, is compared with rα,n-2
- decision rule: reject Ho, if | rc | ≥rα,n-2,
otherwise fail to reject Ho
- if r is (-) , inverse linear association

Regression Analysis
- a statistical technique used to study the functional
relationship between two or more variables
o Simple Linear Regression
- establishes a casual relationship between
the dependent and the independent variable
- estimates the value of a variable given another
variable (ex: what is the value of Y when X=1)
- explains the variation of one variable by the
other variable
- the proportion of change in Y can be
explained by the extent of variation in X is
measured by the coefficient determination, r2
- ex: if r = 0.5, r2 = 0.25, 25% of the variation
in Y is attributable to the variation in X
- equation of the best-fit line:
y = mx + b

Spearman Rank Correlation

- measures the association between ranks
- the nature and degree of relationship between paired
ranks is measured by the Spearman rank correlation
coefficient, rs, whose value ranges form -1 to 1
- test statistic computed as:

- where:
- n = no. of ranked products
- d = difference in ranks between the two data
series
- decision rule: Reject Ho if
rs ≥ p(α,n-1), otherwise fail to reject Ho
 - the judge is required to identify the coded sample
Lesson 9: Discriminative Tests which is the same as the standard
- the probability of selecting the correct match is one-
Discrimination Tests half (1/2)
- Used when we want to determine whether two
samples are perceptibly different 4. Two-out-of-Five Test
- The most appropriate mechanism for answering - five samples are served to the judged, consisting of
questions concerning product substitutability two sets: 2 A’s and 3B’s
- Only useful when the differences between the - judges are asked to identify the 2 A’s
samples are subtle - the chances of correctly identifying the two out of five
- usually performed when there are only two samples samples is one-tenth (1/10)
- nature of the difference is usually not specified – it is
up to the judges if they can find a point of difference 5. A-Not A Test
- two samples labeled “A” and “Not A” are served to the
Case 1: ingredient substitution/change in processing judge followed by coded samples composed of equal
method numbers of “A” and “not A”
- the judge will identify which among the coded
Case 2: product reformulation to make a “new” or samples are the same as “A” and “Not A”
“improved” product
6. Dual Standard Test
- typical discrimination tests involve 25-40 judges who - four samples are presented to the judge; one is
have been screened for their sensory acuity to identified as control A, another is identified as control
common product differences as well as familiarity with B, followed by two coded samples
the test procedures - the judge is asked to identify which of the coded
samples is the same as CTRL A and CTRL B
Data analysis:
- the answers are just counted and compared with 7. Tetrad Test
values from a statistical table derived from a binomial - four coded samples are served to the judge
distribution to give a simple statistical distribution composed of two sets: 2 A’s and 2 B’s
- ex: - the judge is requested to sort the samples into two
groups

8. Octad Test
- eight samples are presented to the judge composed of
4 A’s and 4 B’s
- the judge is requested to sort the samples into two
groups consisting of 4 A’s and 4 B’s

9. Multiple Standard Difference Test

- four coded samples are presented to the judge: one is
-
ODD and three ALIKE
- the judge is asked to identify the ODD sample
Categories of Discrimination
10. Ranking Test
Tests Overall Difference Tests - used to make simultaneous comparisons of several
- does the sensory difference exist between samples? samples based on a single characteristic
- samples are presented simultaneously and ranked
Attribute Difference Tests according to intensity of the characteristic in question
- how does attribute X differ between samples? - NO ties are allowed
- used to screen superior samples from inferior
samples during product development
- maximum of six samples is recommended per session
Overall Difference Tests
Advantages:
1. Simple Difference Test o allows testing of multiple samples
o rapid method
- same/different test
- panelists are presented simultaneously with two o excellent agreement in results can be
samples and asked whether they perceive the obtained if the differences among samples
are large
samples to be the same or different
- best used over triangle test or duo-trio test when the
Disadvantages:
samples have lingering effect or in short supply
- the probability of correct identification is one-half (1/2) o disregards the degree of difference between
samples
o forces a relative decision since no two
2. Triangle Test
samples may be ranked identically
- employs three coded samples presented
o rank value from one set of data cannot be
simultaneously, two are identical and one different compared with another set of data
- none of the samples is identified as the standard
- the judge must identify the sample that is different 11. Rating Difference/ Difference from the CRTL
from the two or the two samples that is different from - used when a control sample is available for
the two or the two samples that are the same comparison with one or more experimental samples
- the probability of choosing the different or odd sample - The judge will indicate the degree of difference
by chance alone is one-third (1/3) between the coded samples and the control
- useful in situations in which a difference may be
3. Duo-Trio Test detectable, but the size of the difference affects the
- employs three samples presented simultaneously, decision
two identical and one different
- one sample is marked “reference” and is the same as
one of the two coded samples
 o referred to as attributes, character notes,
descriptive terms, descriptors or terminology
Attribute Difference Test
Intensity-Quantitative Aspect
1. Quality Scoring o expressed the degree to which each of the
- frequently used method in sensory testing because of characteristics is present
its diversity, apparent simplicity and results can be
easily subjected to statistical analysis Order of Appearance – Time Aspect:
- judges are asked to rate the given samples based on o the order in which certain parameters
intensity of the given sensory attributes manifest themselves

2. 2-Attribute Forced Choice (2-AFC) Aftertaste:

- aka directional paired comparison method o one or two flavor impressions that are left on
- used to determine product differences in a specified the palate after swallowing
dimension/attribute o rated 1 minute after the product is allowed
- two samples are presented to the panelist and the
panelist is asked to identify the sample that is higher Overall Impression – Integrated Aspect
in the specified sensory attribute o aka amplitude
o integration of the product’s properties
3. 3-Attribute Forced Choice (3-AFC)
- a multiple directional comparison method Flavor notes:
- involves selecting the one stimulus among a set of o characterized using 0-3 scale (not present,
three that differs in a defined attribute threshold or just recognizable, slight,
moderate, strong)
o results of FPA are presented whether as
descriptive tables or graphically
Lesson 10: Descriptive Tests
2. Texture Profile Analysis
Description Tests - developed by scientists working for General Foods
- the most comprehensive and informative tests used in during the 1960’s and was subsequently modified by
sensory evaluation several sensory specialists
- answer the question “how do products differ in - assessment of all the texture characteristics of a
specific sensory characteristics?” product, from first bite, through mastication, using
- enable researchers to characterize their products engineering principles
through selective, critical scoring of specific attributes - uses a standardized terminology to describe textural
- used to obtain detailed description of aroma, flavor, characteristics
oral texture of foods and beverages - definitions and listing order of the terms are
- usually conducted using a scorecard containing determined through consensus by the TPA panelists
precise word descriptions - the reference scales anchor both the range and the
- each product characteristic is described over a range concept for each term
- the panelist selects the specific description matching - the full range of a specific parameter by reference
each product characteristic on the scoreboard products helps panelists confirm the intensity
- define sensory properties of target products for new increments within each scale
product development - sample preparation, presentation, and evaluation
- define the characteristics / specifications for a control should be strictly controlled
or standard for QA/QC and R and D - panelists should be trained to bite, chew, and swallow
- track product’s sensory changes over time with in a standardized way
respect to understanding of shelf-life, packaging etc.
- involve the detection and the description of both the 3. Quantitative Descriptive Analysis
qualitative and the quantitative sensory aspects of a - developed during the 1970’s to correct some of the
product by trained panels perceived problems associated with FPA
- panelist must be able to detect and describe the - relies heavily on statistical analysis to determine the
perceived qualitative sensory attributes of the sample appropriate terms, procedures, and panelists to be
- panelist must learn to differentiate and rate the used for analysis for specific products
quantitative or intensity aspects and to define to what - unlike FPA, QDA is NOT a consensus discussion and
degree each characteristics or qualitative note is panel leaders are not active participants
present in the sample - uses unstructured line scales (6-in or 15-cm) to
describe the intensity of the rated sensory attributes
Descriptive Sensory Methods

1. Flavor Profile Analysis -

- developed by scientists at Arthur D. Little in the late - 10 – 12 panelists begin their training by generating a
1940’s and early 1950’s consensus vocabulary
- Name and technique are trademarked to Arthur D. - Actual product evaluations are performed by each
Little and Co., Cambridge, Massachusetts panelist individually, usually in an isolation booth
- a qualitative descriptive test that involves formal - Results are analyzed statistically, and generally
procedures for describing and assessing the aroma, contains a graphic representation of the data in the
flavor, and aftertaste of a product in a reproducible form of a “spider web” with a branch or spike for each
manner attribute
- a consensus technique (all judges must come to an
agreement)
- considers the overall flavor, and the individual 4. Sensory Spectrum
detectable flavor and flavor notes, their intensities, - created by Gail Civille, a TPA expert at General Foods
and overall impression (amplitude) - unique feature: panelists do not generate a panel-
specific vocabulary to describe sensory attributes of
Qualitative Aspect: products, but rather use a standardized word list
o the perceived sensory parameters which (lexicon)
define the product
 - terminology used to describe a particular product is - based on statements, regarding frequency of
chosen a priori and remains the same for all products consumption and motivationally related statements
within a category over time - ex:

Lesson 11: Affective Tests

Affective Tests
- defined by ASTM International (2008) as any method
to assess acceptance, liking, preference, or emotions
for a stimulus or stimuli
- primary purpose: determine the personal response by -
current or potential customers of a product, a product
idea, or specific product characteristics
3. Just About Right Scale
Reasons for Conducting a Consumer (Affective) Test - measure the appropriateness of the level of a specific
- product maintenance attribute
- product improvement/optimization - used to determine the optimum levels of attributes in
- development of new products a product
- assessment of market potential

Quantitative Affective Tests Qualitative Affective Tests

- used to look for trends in the marketplace, new product
Preference Test opportunities, or determine whether a segment of the
- which sample do you prefer? population is receptive to a product or service idea
- Which product do you like better?
1. Focus Group Discussion
Acceptance Test - a round table gathering of a small number of target
- How much do you like the product? consumers of a product, service, or concept, normally
- How acceptable is the product? between 6-12 to discuss, share, and explore feelings
and attitude towards a product or concept or to
generate ideas
Preference Tests
2. Mail Panels
1. Paired Preference Test - used with the products that are suitable for mail
- two samples are presented to the participant delivery
simultaneously, and the participant is asked to evaluate
the samples and tell which one he/she prefers 3. Telephone Interview
- forced choice or non-forced choice (judge is given the - getting to be popular but it can’t stand on its own
option “ no preference” ) because it deals only with verbal stimuli

2. Ranking Test Respondents/Consumers in Affective Tests:

- participant is asked to rank a number of products in 1. Employees
descending or ascending order according to 2. Local Residents
preference or liking 3. General Population
- used to compare several samples according to one
single attribute (e.g. saltiness) Locations:
1. Laboratory Test
- advantages:
3. Best-Worst Scaling o product preparation and presentation can be
- aka “maximum difference scaling” or “maxdiff” fully controlled
- participant is asked to pick out which product he/she o employees can be contacted on short time
prefers the most and the least notice to participate
- scores are computed by subtracting the number of o color and other visual aspects which may not
times a product is voted the best and the number of be fully under control in prototype can be
times it is voted the worst masked
- disadvantages:
o location suggests that the products originate
Acceptance Tests in the company or specific plant
o lack of normal consumption may influence the
1. Hedonic Scale detection or valuation of positive or negative
- most widely used scale for measuring food attributes
acceptability o product tolerances in preparation or use may
- also used for personal care and household products be different from those of home use

2. Central Location Tests (CLT)

- recruitment is conducted by intercepts in a public place
- once a participant agrees to sample the product,
he/she is ushered into a separate area where the
samples will be evaluated to complete the survey
- advantages:
o respondents evaluate the product in a
standardized location under conditions
controlled by the organizer
- o products are tested by end users which can
assure the validity of the results
o conditions are favorable for a high percentage
2. Food Action Rating (FACT) Scale return of responses from a large population
- disadvantages:
 o products are tested under artificial conditions
o number of questions asked are limited which
in turn limits the amount of information from
the data

3. Home Use Tests (HUT)

- advantages:
o the test products are used under conditions
which approximates the normal

The final judge of a product is the consumer.

LESSON 12: Threshold

Thresholds
- limits of sensory capabilities
- can be used to determine the point at which some
substances begin to reduce acceptability
- may also be used as means of selecting or testing
panelists

Absolute Threshold
- detection threshold
- lowest stimulus capable of producing a sensation
- minimum detectable concentration

Difference Threshold
- is the least amount of change of a given stimulus
necessary to produce a change in sensation
- the internal used is JND (Just Noticeable
Difference) or JNND (Just Not Noticeable
Difference)

Recognition Threshold
- level of stimulus at which the specific stimulus
can be recognized and identified
- usually higher than the absolute threshold

Terminal Threshold
- magnitude of a stimulus above which there is no
increase in perceived intensity of the appropriate
quality of the stimulus