Sensory Evaluation Manual 2002

THE HUMAN SENSES IN SENSORY EVALUATIONTHE SENSES - AN INTRODUCTION The sensory properties of foods are related to three major attributes: • Appearance - colour, size, shape; • Flavour -

These notes form the basis of a practical workshop presented for personnel at NaresuanUniversity, Phitsanulok, Thailand in July, 2002 We would like to thank Michael O’Mahonyfor his permission to include copies of the statistical tables from his book “SensoryEvaluation of Food: Statistical Methods and Procedures” and material supplied by the Centrefor Food Technology, DPI, Brisbane

TABLE OF CONTENTS

ACKNOWLEDGEMENTS 2

PROGRAM ERROR! BOOKMARK NOT DEFINED INTRODUCTION 5

THE HUMAN SENSES IN SENSORY EVALUATION 7

THE SENSES - AN INTRODUCTION 7

SENSE OF SIGHT- 9

THE SENSE OF SMELL 13

THE SENSE OF TASTE 15

THE SENSE OF HEARING 19

THE SENSE OF TOUCH 20

SENSORY INTERACTION 21

OPERATIONAL PRINCIPLES OF SENSORY TESTING 23

DESIGN OF A SENSORY TESTING AREA 31

STATISTICAL PRINCIPLES 34

SENSORY EVALUATION METHODS 38

AFFECTIVE TESTS 38

SPECIFIC TEST METHODS 39

PAIRED PREFERENCE TEST 39

RANKING FOR PREFERENCE 41

RATING FOR PREFERENCE 44

SENSORY EVALUATION IN CONSUMER TESTING 46

ANALYTICAL SENSORY TESTS: 53

DIFFERENCE TESTING 53

SIMPLE DIFFERENCE TEST 53

TRIANGLE TEST 53

DUO-TRIO TEST 55

TWO-OUT-OF-FIVE TEST 59

“A” – “NOT A” TEST 59

DIFFERENCE-FROM-CONTROL TEST (DFC) 59

DIRECTIONAL DIFFERENCE TESTS 65

PAIRED COMPARISON TEST 65

RANKING TEST 67

RATING TEST 69

STATISTICS FOR SENSORY: DIFFERENCE TESTING 73

DESCRIPTIVE TESTING 76

STATISTICS FOR SENSORY: DESCRIPTIVE TESTING 81

SELECTION, TRAINING AND MOTIVATION OF A PANEL 86

REPORTING 91

SELECTED BIBLIOGRAPHY 92

JOURNALS 94

STATISTICAL TABLES 95

Sensory evaluation - A scientific discipline used to evoke, measure, analyse and interpretreactions to those characteristics of foods and materials as they are perceived by the senses ofsight, smell, taste, touch and hearing

Sensory evaluation was one of the earliest methods of quality control and it is still widelyused in industry However, the level of application depends on the situation (e.g beer andwine tasting to operators sampling of products from production line)

Four variables affect sensory evaluation:

• Gives real answer regarding consumer quality

• Relatively cheap process (depending on how it is done)

• Rapid

• Many applications

• Objective methods are more reliable, accurate and reproducible However, they must

be correlated to sensory evaluation to indicate a consumer response

• Panel selection / training

• Correlation subjective / objective

Sensory Standards

Aus Standard Year Title

AS 2542.0 1995 Sensory analysis of foods - Introduction and list of methods

AS 2542.1.1 1984 Sensory analysis of foods - General guide to methodology -General requirements

AS 2542.1.2 1984 Sensory analysis of foods General guide to methodology

-Types and choice of test

AS 2542.1.3 1995 Sensory analysis of foods - General guide to methodology -Selection of assessors

AS 2542.2.1 1982 Sensory analysis of foods - Specific methods - Pairedcomparison test

AS 2542.2.2 1983 Sensory analysis of foods - Specific methods - Triangle test

AS 2542.2.3 1988 Sensory analysis of foods - Specific methods - Rating

AS 2542.2.4 1988 Sensory analysis of foods - Specific methods - Duo-trio test

AS 2542.2.5 1991 Sensory analysis of foods - Specific methods - 'A not A' test

AS 2542.2.6 1995 Sensory analysis of foods - Specific methods - Ranking

AS 2542.3 1989 Sensory analysis of foods - Glossary of terms

THE HUMAN SENSES IN SENSORY EVALUATION

THE SENSES - AN INTRODUCTION

The sensory properties of foods are related to three major attributes:

• Appearance - colour, size, shape;

• Flavour - odour, taste; and

• Texture - mouth feel, viscosity and hearing

These attributes are expressed as a continuum and not as finite properties It is impossible torate each one individually unless special precautions are taken, e.g blindfolds, nose clips,coloured lights, purees

Humans possess about 30 different senses However, the sensory properties of foods areperceived through the senses of:

A stimulus is any chemical or physical activator that causes a response in a receptor, e.g eye

is receptor for light, ear is receptor for sound

An effective stimulus produces a sensation, the dimensions of which are:

SENSE OF

SIGHT-The appearance of food

Stimuli = visible light

Receptor= retina of the eye

Perception=sight, vision, appearance

The appearance of foods is a major factor governing its acceptability and can be subdividedinto three main categories:

• Optical properties- colour, gloss and translucency

• Physical form-shape and size

• Mode of presentation-lighting packaging etc

Optical properties

Vision

Vision is a complex phenomena consisting of several basic components A stimulus, light,from an external source interacts with the object and is brought to focus on the retina of theeye The retina is the receptor of vision and contains two types of cells The rods areresponsible for vision in dim light and the cones are responsible for colour vision Lightincident on these cells causes a photochemical reaction that generates an electrical impulsewhich is transmitted to the brain via the optic nerve Colour blindness is caused by loss orlack of colour receptor cells in the cones Approximately 8% of the population have somedefect with relation to colour; mostly males

[NOTE: All electromagnetic radiations are physically the same However, the optical system

of the eye is such that only the visible range of wavelengths is absorbed by the lens.]

Light sources

Incandescent lights consist of a tungsten filament which is heated in an inert gas The higherthe temperature, the more light produced Light from this source tends to be harsh and tends

to highlight the red end of the spectrum

Fluorescent lights operate by electrical excitation of atoms that produces spectral lines atspecific wavelengths which then impinge onto fluorescent materials which convert theincident light into light at a longer wavelength Light produced is softer but can producecolour distortion at particular wavelengths

Natural light is too variable for use in evaluating appearance of foods

Light - Object interactions

Light incident on an object may be:

• Sliced, diced, pieces whole

• Length of frozen French fries

• Cut of beans

• Extrusions

Surface texture can indicate product texture Some examples include:

• Open dry structure of meat

• Wrinkling of peas

• Wilting of lettuce

Visual consistency can indicate product viscosity as in:

• Setting of a jelly

• Syrups of different concentrations

• Pastes and purees

Mode of presentation

This aspect should be considered from a marketing point of view and is important because itinfluences sales Mode of presentation is applicable on the supermarket shelf (at retail level)and also in terms of presentation at the table (home and restaurant)

Factors to be considered are:

• Product description - name, price, ingredient, etc;

• Packaging - shape, design, colour;

• Contrast - phenomena of adjacent colours; and

• Illumination - affects apparent product colour

Appearance is an important aspect of food quality as it is the first subjective evaluation made

of food quality The product has to pass the visual assessment before the consumer can orwill consider the other parameters such as taste and texture

Factors that should be considered in evaluating product appearance include:

• use of standard conditions:

• light source (type, intensity, colour);

• background; and

• style of presentation (unless tested)

• selection of appearance attribute(s) for inclusion on scoresheet;

• using appearance to reduce tasting load;

• should be masked to eliminate unwanted interactions when assessing parametersinvolving other senses; and

• colour charts/standards help rating

THE SENSE OF SMELL

(Odour/olfaction)

Stimuli = volatile chemicals

Receptors= olfactory cells in the nose

Perception=smell, odour, aroma, flavour

Smell is one of our most primate senses Supposedly prehistoric people were moreinfluenced by smell than other senses

The human nose is capable of detecting thousands of different odour substances However,our sensitivity is much less than other animals (Animals use smell - food, mating, territoryetc)

Smell is detected both before and during eating Smell is an important aspect of flavour.There are 20x106 olfactory receptors, but only about 1000 taste receptors

Odour description requires the development of an odour/flavour memory, e.g fishy, flowery,woody This is the basis of flavour/odour memory development by wine judges andmilk/cheese graders Individuals vary a great deal in their sensitivity to differentodours/aromas

Anatomy of olfactory system

From the diagram it can be seen that most of air misses the olfactory area Only 5-10% ofinspired air passes over olfactory receptors However, this amount can be increased bysniffing harder; obviously the more air which passes over the receptors the better the

The large number of olfactory receptors (20x106) enable detection of :

• More odours than tastes;

• A greater variety of odours; and

• Odours at much lower concentration (10 molecules/mL)

In order for odour to register:

• Substance needs to be volatile enough to get into air in the sensory region

• Substance needs to be partially soluble in mucus covering of receptors

• Minimum number of odorous molecules need to be present

• Need to be in contact with receptors for minimum time

Olfactory intensity

Human nose is about 10-100 times more sensitive to odours than any physico-chemicalanalysis (e.g gas chromatography) It has been demonstrated that human nose is capable ofdetecting ethyl mercaptan at a concentration of 0.01 mg/230m3 of air, which is equivalent toabout 8 molecules/receptor

Olfactory threshold

Detection threshold is the concentration where smell is detected

Recognition threshold is the concentration where the smell is recognised

Olfactory interactions

Nature of the response may change with concentration (e.g perfumes at low concentrationare pleasant but at strong concentration may be unpleasant)

Interaction of odours:

• Additive - increase intensity;

• Suppressive - decrease intensity; and

• Blending - when new odour unrelated to originals

Olfactory adaptation

Initial sensation maybe strong - but weakens and makes identification difficult; this is due toadaptation of olfactory receptors

In testing we therefore need to allow for this by:

• Taking first impression of odour and/or

• Waiting between tests to allow receptors to recover

• Smell is a major component of food flavour

• Human nose is much more sensitive than analytical instruments

• Foods contain numerous compounds of varying volatility that can make analyticalinterpretation difficult (e.g strong peaks may produce weak odour whereas weakpeaks may produce a strong odour)

• Smell measures perception of a mixture; analytical testing does not

THE SENSE OF TASTE

(Gustation)

Stimuli = soluble chemicals or chemicals which are solublised during chewing

Receptors= taste buds in mouth

Taste receptors

The receptors for taste are the taste buds and these are mounted on papillae (folds in the skin

of the tongue) The area of greatest response is the top of the tongue Other areas in themouth and throat where taste buds are situated include: palate, pharynx, larynx, tonsils,epiglottis, lips, cheeks, underside of tongue and floor of mouth

Taste buds are mainly located at the tip, sides and rear of tongue There is very little response

in the centre of the tongue Different areas of the tongue are most responsive to differentsensations

The five basic tastes

A basic taste is one for which specific taste buds have been identified as beingphysiologically responsible for the particular taste sensation

Sourness

This is the simplest taste as only acids (H+) produce sourness and as the (H+) increases thesourness increases

However there are some anomalies to this:

• organic acids are more acidic than expected

• sourness of aliphatic organic acids relates to chain length

• some amino acids are sweet (aspartane)

• picric acid is bitter

• sugar may enhance/depress sourness

• sourness is also affected by pH and acid

• presence of buffers affects sourness

Sweetness

The common substances that produce the sweet taste are the sugars and other hydroxycompounds such as alcohols and glycols Other substances such as lead salts, amino acids,proteins, non-nutritive sweeteners (cyclamates, saccharin and aspartame ) also taste sweet

Saltiness

Many crystalline water-soluble salts yield a salty taste, but only sodium chloride gives a puresalty taste Other substances taste salty but also bitter, alkaline, sweet and salt in variouscombinations

Bitterness

Many chemically different compounds have a bitter taste However, bitterness is mainlyassociated with alkaloids such as caffeine, quinine, strychnine and nicotine Originally it wasthought that bitterness was an indication of danger (poison) However, many alkaloids areused as drugs (e.g codeine) and many other bitter substances are harmless (glycosides, estersand aldehydes and tannins in wines and tea)

Bitterness is generally perceived at very low concentration and a relationship appears to existbetween sweet and bitter as many sweet substances produce a bitter aftertaste (saccharin).Bitterness is the taste which most people have difficulty in detecting and response level variesgreatly from individual to individual

Umami is the taste that has been shown to be associated with substances that containglutamate The most notable example is mono-sodium glutamate (MSG) MSG is wellknown as a flavour enhancer and can cause adverse reactions in some sensitive individuals.However, there are many other compounds which contain glutamate and which are capable ofproducing the savoury, spicy, brothy taste associated with MSG Many foods containnaturally high levels of glutamate

Taste interactions

Having described the 5 basic tastes it is obvious that foods are a very complex system whichcontain many different taste compounds and therefore many different tastes The fact thatthere are only 5 basic tastes and yet we are able to detect hundreds of different tastesensations is due to a series of complex taste interactions that can range from simple 2 wayinteractions to complex 5 way interactions

Interactions between the 4 basic tastes were previously described simplistically by the tastetetrahedron

Adaptation and fatigue

During exposure to a stimulus, sensitivity decreases due to adaptation and fatigue This loss

in sensitivity varies considerably with the taste (sweet, sour, salty or bitter) and also with thecompound For example, tasting a series of acids causes the sensitivity to be reduced by thepreceding acids However, recovery is usually rapid because most common organic acids arevery soluble

Taste thresholds and sensitivity

There is great variability between individuals in their levels of sensitivity Sensitivity isaffected by:

• ·Five types of taste receptors - salt, sweet, sour, bitter and umami.

• ·Different areas of the tongue respond to different sensations

• ·Substances must be dissolved for taste buds to detect them

• ·Flavour of the food is a complex interaction of different tastes and odours

• ·Sensitivity to taste varies between individuals and is affected by their physiologicalstate

THE SENSE OF HEARING

(Audition)

Stimuli = physical movement of sound waves in a medium (air)

Receptor= ear drum

• Snap, crackle and pop;

• Fizz of champagne or beer;

• Crispiness of lettuce or celery; and

• Tapping a melon for quality

Negative aspects:

noisy environment may distract tasters or mask product sounds

THE SENSE OF TOUCH

(Texture, Kinesthetics)

Stimuli = physical contact between the food and body tissue

Receptors= muscles and nerves in mouth and fingers

Perception=touch, feel, texture, viscosity

Texture usually relates to solid food while viscosity relates to homogeneous liquid foods andconsistency relates to non-homogeneous liquid foods

Instrumental methods only measure one aspect of "texture" and again cannot relate thecomplex interactions which produce the perception of food texture

Finger feel

Firmness/Softness indicates the eating quality of some food products:

• Ripeness level of fruit such as avocado and mango;

• Crumb texture of bread;

• Firmness of cheese; and

• Spreadability of butter or spread

Juiciness can be used as a subjective quality index (eg the “thumbnail” test for corn)

Mouth feel

Liquids

• Viscosity - thin to viscous, e.g milk, cream

• Consistency - thin to thick, e.g fruit yoghurts

Solids

Classification of textural characteristics - assessed mainly by chewing

Textural Terminology Mechanical Characteristics

Hardness Soft, firm, hard, e.g fruit ripeness, cheese maturity.Brittleness Crumbly, crunchy, brittle, e.g muesli bars and biscuitsChewiness Tender, chewy, tough, e.g meat

Grittiness Gritty, grainy, coarse, e.g stone cells in fruit, "sand" in

ice-cream

Fibrousness Fibrous, cellular, e.g string/fibre in vegetables.

Moistness Dry, moist, wet, e.g cracker biscuit, cheeses, water

Interaction between senses

This is the ability of a response from one modality to influence or affect the response fromanother There are two aspects of this:

Positive - interactions giving clues to possible identity, e.g pink milkshake being strawberryflavoured

Negative - If clues are not correct this may lead to confusion and a wrong judgement, e.g.pink milkshake with pineapple flavour

Types of sensory interactions

Taste - sight

This is a very important aspect because vision is the first sense affected and appearance of aproduct will have a major influence on absolute quality Bright colours indicate strongflavours whereas dull colours indicate mild flavours

Other interactions include:

• Odour - Sight

• Odour - Tactile

• Taste – Hearing

• Odour - Hearing

Multiple interactions

Multiple interactions between more than two modalities are also possible

Example: Tasting food pureed, blindfolded and with nose clips gives a different responsethan when interactions are allowed

Interactions between stimuli

These interactions are more difficult to define and measure but are just as important asinteractions between the senses Some examples include:

• suppression of one flavour by another, e.g sweetness is suppressed by acidity This isthe basis of ensuring brix/acid ratio for fruit juices are constant;

• neutralisation of one flavour by another;

• blending to produce a totally different flavour, e.g garlic flavoured cheeses;

• partial blending producing a new flavour and the original flavours;

• no effect; original flavours are distinct and separate, e.g fruit in cheese;

• intensification resulting in enhancement of flavours, e.g salt and MSG on foodimproves the natural flavours

Similar situations may exist for all other stimuli

Summary

Interaction must be considered when designing sensory panels If only one sense or stimulus

is to be evaluated then all others must be masked However, if interactions are required thenensure this can be achieved by means of sample preparation

OPERATIONAL PRINCIPLES OF SENSORY TESTING

When evaluating properties of foods using people as measuring instruments it is important tocontrol the methods and conditions of testing as rigidly as possibly This helps to eliminatethe numerous errors or biases that can be caused by psychological and physiological factors.The mental attitude and physical condition of a taster, and the atmosphere of the testingenvironment all influence their judgements There are therefore a number of basic ruleswhich should always be applied, as stringently as circumstances allow, when running tastepanels These relate to:

• Selection of panellists;

• Preparing the testing environment;

• Designing the experiment;

• Preparing samples;

• Serving samples

General principles that should always be followed are:

Never ask anyone to taste food they do not like;

Make sure that the "correct" panellists are selected (see section on panel selection andtraining) and that they know in advance when they will be required

Keep a strict control over all variables except those being tested (e.g sample size andtemperature)

Make sure the environment gives optimum opportunity for concentration Tasting properly is

a difficult job Train panellists to be silent while tasting This prevents panellists frominfluencing one another

Make tasting interesting and desirable Use rewards to motivate taters, vary these and choosefoods that contrast with those being tasted Motivated tasters are more efficient Givefeedback on results whenever possible

Avoid giving any unnecessary information to panellists that may influence their scores.Tasters usually find what they expect to find; e.g in a storage test they expect to find samplesdeteriorating

Plan your experiment in advance Which will be the best test to use? Consider all aspectsincluding how you will get the information required from your results (statistics) Runpreliminary tests, i.e practise and choose the best method for:

Sample size - adequate but not excessive;

Serving temperature - standardise for all samples It must be maintainable, and be anacceptable temperature for the food;

Serving vessels;

Eating utensils

Sample preparation and serving

Serve tasters promptly and make sure they have everything they need.

Run a taste panel as you would expect a good restaurant to be run, i.e give courteousfriendly service, be efficient, and serve good food

Keep accurate records of any cooking or preparation methods used Record temperatures andsize of samples served and any special conditions (e.g coloured lighting)

It is important that panellists do not see the samples being prepared as this may indicatequality difference

Sample preparation should be uniform:

• Temperature

• Cooking

• Thawing

• Size and shape (provided this is not a variable)

Sample should be randomly allocated to:

• Avoid bias

• Overcome any non—uniformity

Sample size should be adequate:

Containers for presentation

Containers for presentation and tasting should be:

• Clean

• Identical for all samples and sessions

• Disposable containers or re—usable

• Coloured to mask product appearance (if required)

• Relevant to product

Serving temperature

• Serve at room temperature where possible

• Preference tests use normal temperature

• Difference tests may alter temperature to accentuate flavours/odours

Dilutions and Carriers

Most foods should be served in the way they are normally eaten However, some productssuch as spices, chillies, alcohol, onions, etc may require dilution before testing If dilutionsare used they must be uniform in terms of diluent and concentration

Carriers are substances that are added to assist tasting of certain products Carriers are aproblem because they can be:

• Expensive

• Time consuming

• Variable quality

• Difficult to control product/carriers ratio uniformity

For example: developing a cake icing individually may not allow for interaction with flavour

or it may be incompatible with the cake (affects texture or falls off)

Number of samples

Samples / Sessions

The number of samples presented at any testing session will depend on:

• Type of product - strong flavour —> less samples

• Type of test

• Rating scale may require fewer samples

• Test dictates sample number eg: triangle test = 3 samples

• Type of panel — trained / experienced -> more

In calculating the number of sessions consider the following:

• Total number of samples for tasting

• Statistical design

• Taster fatigue

• Motivation

• Type of panel (trained/untrained)

Phsiological factors in taste testing

Time of Tests

• Monday and Friday are recognised as being bad days for tasting

• Normally taste 1 hour before meals and 1 - 2 hours after

• Sometimes this becomes difficult in practice due to:

• Unavailability of tasters

• Number of sessions

Smoking / Taste Affecting Substances

As indicated earlier, smoking affects sensitivity to flavours —therefore should either:

• Not use smokers

• Ensure they do not smoke for at least 1 - 2 hours before tasting

• Chewing gum, mints and spices etc may also influence taste

It is a good idea to get panellists to cleanse their palate:

• Before tasting to remove any lingering tastes

• Between samples to reduce adaptation of taste buds

• Warm water, biscuits, bread, apples may be used as a palate cleaning agent

Palate clearing can be optional but whatever is done must be constant

The time between samples should also be kept constant if possible

• Stressing importance of work

• Stimulating company expansion

• Greater profits

• More pay

• Ensuring panellists know what is involved with the trial ie: sessions, products, whenand where tasting will be conducted

• Having adequate facilities

• Using effective methods and designs

• Publicising results obtained from work

• Rewarding panellists

Sample Coding

Remove possible bias or influence from samples codes Do not use

• Single digit numbers

• Consecutive letters

• Same codes at consecutive sessions

Randomly or statistically generated three digit number codes are best

Order of Presentation

Always use either a random order of presentation or a statistically balanced design to avoid:

• Donkey vote (first is best; last is worst)

• Position bias - in triangle tests middle one is different

• Contrast effect — good after bad appears better, or bad after good appears worse.Devise your own system for remembering orders, e.g 3 digit numbers - put in sequence ofone of digits Keep it a secret!

Always work systematically in coding, labelling, setting up, e.g as in reading a page

(1) Left to Right

(2) Top to Bottom

This provides an automatic check if something goes wrong

Balance presentation of samples whenever possible This avoids contrast effect

ie 2 samples A, B - Half panel taste A first, other half taste B first

- Half panel receive A on the left, other half

receive B on the left

3 samples 6 different orders in which they are tasted Use

every order the same number of times Number

of tasters is a multiple of six

- Position of samples on plate must also be

balanced

4 samples 24 different orders: use them all if possible (see

table on next page)

4 samples Generate random order Write out set of cards

and shuffle them

When you cannot use balance to eliminate bias, use randomisation

Four sample balanced orders

• Do not give detailed information about treatments

• Do not use people on panel who know what the treatments are

• Sample coding and design can prevent expectation error

Logical / Stimulus Error

Tasters look for clues to get the “right” answer eg: a difference in sweetness may beassociated with sample differences such as size, shape and colour This error can beovercome by ensuring sample preparation is uniform or use masking

Halo Effect

When more than one factor in a sample is evaluated at one time the result obtained may bedifferent than if factors evaluated separately This can be overcome by tasting each aspectseparately However, this is time consuming and would only be done if extremely accurateresults were required Testing one aspect at a time in preference does not simulate the “real

situation” ie: consumers do not taste every aspect separately.

Suggestion

Influence of other panellist may bias or influence results This can be prevented by:

• Using booths

• Not allowing talking in tasting area

• Reducing outside distractions

Questionnaire design

Questionnaire design should be simple and easy to follow in terms of design and languageand make sure tasters know how to use it You may need to include some instructions on thescoresheet itself, but it is usually better to give instructions verbally to your panel first Thequestionnaire should generally not be more than one page and include:

DESIGN OF A SENSORY TESTING AREA

The main considerations to keep in mind when preparing an area for sensory testing concernthe requirements for an atmosphere conducive to concentration, where conditions can becontrolled Sensory panellists need somewhere comfortable and free from distractions if theyare to be able to "tune in" to the sensations triggered by the stimuli in the food products theyare tasting Product characteristics can be markedly affected by temperature and humidity,and appearance is affected by lighting intensity

The conditions should be controlled in order to :

• Reduce bias

• Improve accuracy

• Improve sensitivity

• (compare to the conditions used in an analytical laboratory)

International standard (ISO 8589-1988)

The standard looks at the design of the testing area for both new and existing buildings Italso specifies which recommendations are considered essential and which are only desirable.Important points summarised from the standard are listed below If designing an area that is

to be dedicated solely to taste panel work, these should be seriously considered

Total area should include:

• Testing area with individual booths and a group area;

General testing area

• Easily accessible but in quiet position

• Location - close proximity to preparation area, but separate entrance, and withcomplete "close-off" capability

• Temperature and relative humidity - constant, controllable, and comfortable

• Noise - keep to a minimum, soundproof area as much as possible

• Odours - keep area free from odours (air conditioner with carbon filters, slightpositive pressure)

• Use odourless materials in construction and decoration

• Use odourless cleaning agents

• Decoration - use neutral, light colours for walls and furniture (e.g off-white, lightgrey).

• Lighting - ambient lighting must be uniform, shadow-free and controllable Forconsumer testing - as close to home conditions as possible

Booths

Number - minimum three, normally five to ten - six is a useful number since it fits in wellwith balanced ordering of 3 samples

Space - allow sufficient space for movement of tasters and for serving samples

Set-up - permanent booths recommended Temporary acceptable If adjacent to preparationarea include openings in the wall to pass samples through Size and style specified Considerspace for samples, utensils, spittoons, rinsing agents, scoresheets and pens, computerisedequipment Include comfortable seats

Lighting - uniform, shadow-free, controllable, adequate intensity for assessing appearance.Devices to mask appearance (e.g dimmers, coloured lights or filters)

Group work area

General Necessary for discussion and training purposes Include large table andseveral chairs "Lazy Susan" useful Include board for discussion notes, etc

Lighting As for general area, with coloured lighting options like booths

resources or physical space available.

Very few industries are able to start from scratch, designing new premises solely dedicated tosensory analysis work I therefore would like to abbreviate the list proposed in the standard

to one which I consider includes the bare essentials

Minimum of 2 areas:

Preparation area and office area If possible position these at opposite ends of the room toavoid messy paperwork!

Testing area with entrance separate from preparation area

Preparation area requires

• Adequate storage for utensils and equipment;

• Adequate working surfaces to set out samples;

• Washing up facilities - minimum double sink with hot and cold running water;

• Refrigerator - minimum 2 door with separate freezer, preferably at least auto-defrost;

• Cooking equipment - depending on sample requirements;

• Rubbish bin - large with liner bags;

• Source of boiling water;

• Hand washing facilities

Testing area requires

• Comfortable chairs for panellists;

• Minimum space - 4 panellists;

• Table which can be easily divided into booths if required;

• All equipment likely to be needed while a panellist is tasting, e.g pencils, spittoons,toothpicks, tissues/serviettes;

• Well placed, efficient lighting;

• Waiting are with noticeboard - for tasters to wait for booths to become free and tocollect rewards after tasting

A system using collapsible booths can work quite well if it is not possible to keep an areasolely for sensory work These may be made of painted wood, heavy duty cardboard, or

"corflute" They can be made specifically to fit any available benches or tables and foldedand stored when not in use

The type of facility will depend on:

• Finance

• Available space

• Frequency of use

• Tests conducted

STATISTICAL PRINCIPLES

This section looks at the role of statistics in sensory evaluation and introduces some termsand concepts required to correctly apply statistical methods in evaluating sensory type data

Why do we need statistics in sensory evaluation?

When we measure something (eg salt level in cheese) we find there is variation in what we are measuring This variation is called natural variation or experimental error and implies

that there is some true measurement but because of our limitations we cannot reproduce thecorrect readings every time This is a fact of life and we have limited control over this sort oferror

Because of this variation there is some risk in making decisions about changing formulations

or introducing new products onto the market Using statistics we have rules to estimate and

minimise the risk and enable us to extrapolate our results from an experiment to a more

general situation

What is an experiment?

It is any process that generates raw data

There are many sources of error in sensory data Some of these include

• differences between people, (likes and dislikes)

• differences within a person from time to time, eg adaptation

• differences among samples,

• differences in interpretations of scales

• and many more

How can we describe our data?

Lets say we have collected some data from an experiment and we have 20 scores of flavouracceptability in a mango sample rated on a 9 point hedonic scale If we plot a bar graph(histogram) using the score along the horizontal axis and the count for a particular score onthe vertical axis then we have a frequency distribution An example is shown below

Looking at the graph or distribution we ask what is the best single estimate of the panels score and what is a good measure of their variability? The best or most likely single estimates are called measures of central tendency The three most commonly used are:

mean - or average (sum of all data values divided by number of observations)

median - 50th percentile or middle value

mode - most frequent value, good for categorical data

Measures of variability include the range, standard deviation and variance The range is

simply the difference between the smallest and the largest The standard deviation isprobably the most common and is calculated by using the formula below

where M is the mean or average of X scores and N is the number of scores This formulacalculates the deviation of each score from the mean and squares it to take into accountpositive and negative values and the square root is then taken to bring it back to the originalunits The variance is simply the square of the standard deviation and is used in a number ofstatistical formulas

0 1 2 3 4 5 6

Flavour Acceptability

The normal distribution

Many things we measure about a group of people will be normally distributed This meansthey will form a bell shaped curve described by an equation usually attributed to Gauss

How does the standard deviation relate to the normal distribution?

Standard deviations describe discrete percentages of observations at certain degrees ofdifference from the mean So for a normal distribution about 66% of our data will be withinone standard deviation of the mean and about 95% will be within two standard deviations.For our mango flavour data with a mean of 6.0 and standard deviation of 1.89 then 66% ofour data lie between 4.11 and 7.89 If the standard deviation had been 1.00 then 66% of thevalues would be between 5.0 and 7.0, a smaller range indicating less variability

In addition any score, X can be described in terms of a z-value, which describes how far thescore is from the mean in standard deviation units

Z = X-µ/σSince z-scores are related to percentages under the normal curve they can predict how far ascore is from the mean and how likely or unlikely it is So the z-score can be converted to aprobability value or p – value This p - value is found from the area under the curve outsidethe z score and is the chance with which we would see a score of that size or greater Tablesare often used to convert z - scores to p – values

An important concept

When we do an experiment we are using results from a sample taken from a larger

population of possible results Since we cannot take all possible results from the population

we infer from our sample results what should happen in the rest of the population By

making this generalisation we often express our results in terms of probability or p- values.

This is our safety margin or level of confidence about our result It is often quoted like this the flavour score for naturally ripened mango was significantly higher (P<0.05) than that forartificially ripened mango

-But what does this mean?

We are at least 95% certain that based on our experimental conditions the naturally ripenedmango will have more flavour than artificially ripened mangoes This conclusion will bewrong about five times out of 100 Sometimes a 1 % value or 0.01 is used for greaterprecision

How does all this help?

We need to identify some more concepts before we can be confident in using statistics.Experiments need to be planned and carried out correctly before we can use statistics and two

important principles are replication and randomisation.

Replication is the assessment of each treatment more than once A treatment can be the

addition of sweetener to a product or the storage temperature of a fruit With replication we

can assess the natural variability and separate this from our variability due to treatmentdifferences This is like a signal to noise ratio Is our signal greater than the backgroundnoise (natural variation)?

Random allocation of treatments to samples or products ensures each sample has an equal

opportunity of receiving any treatment, and that this chance is unaffected by the treatmentsassigned to other samples For example if two products are tasted by 24 tasters and they alltaste product A first then this may well bias the results, as the first product tasted may tend to

be preferred regardless of which it is Subjective allocation of treatments in a haphazard way

is not a satisfactory alternative to randomisation

SENSORY EVALUATION METHODS

There are two main types of sensory methods:

Affective :tests which involve consumer preference or acceptance

Analytical : tests which are involved with analyzing specific product attributes in terms of:

SPECIFIC TEST METHODS

PAIRED PREFERENCE TEST

(Reference: AS 2542.2.1; 1982.)

Application: to establish whether there is a preference between two samples.

Principle: a pair of samples (one may be a control) is presented to each assessor The

assessors are asked to choose the sample they prefer

This test is a ‘forced choice’ ie: the assessors must select one sample as being morepreferable Responses indicating no preference are not permitted

Statistically based on null hypothesis that there is no preference between the samples

ie:PA = PB = 50%= 0 5

Bilateral Test - no expectation of preferences

Specimen Answer form for bilateral paired preference test

Conclusions

• no preference

• A preferred to B

• B preferred to A

Question — which of the two samples do you prefer? Count the number of replies citing one

of the two samples the more frequently

Conclude that this sample is significantly preferred to the other if the number obtained isgreater than or equal to that shown in Table 4

Which sample do you prefer?

Please examine code 349 first

Please tick the appropriate box

Place tick

YOU MUST MAKE A CHOICE

of significance (ie: P < 0.05%) It is not known which of the two samples contains moresugar.

Question - Which sample do you prefer?

Replies: 22 prefer ‘A’

8 prefer ‘B’

From Table 4 it can be concluded that Drink ‘A’ is preferred to Drink ‘B’.

Unilateral Test - expect one sample to be preferred.

Specimen Answer form for unilateral paired preference test

Conclusion

• no preference

• the declared sample is preferred

Question — Do you prefer sample ‘A’ to sample ‘B’? Conclude sample A is preferred if

number of positive replies is greater or equal to the number shown in Table 3

Example:

Two drinks, ‘A’ and ‘B’, are offered to a panel of 30 assessors The two samples arepresented under a random number eg: ‘789’ and ‘379’ The test supervisor accepts a 1%level of significance (ie: P < 0.01%) It is known that drink ‘A’ contains more sugar thandrink ‘B’

Question - Do you prefer sample ‘A’ to sample ‘B’?

Replies: 23 Yes and 7 No

From Table 3 it can be concluded that there is preference for drink ‘A’ over drink ‘B’.

PRODUCT……….DATE………… TIME………ASSESSOR……

………

Do you prefer sample 186 to sample 592?

Please examine code 592 first

Please tick the appropriate box

YOU MUST MAKE A CHOICE